Developing a System for Integrated Automated Control of Multiple
Contents
1. 13 SA 13 A o Pe 14 SS Tr O 19 4 5 USABILITY EVALUATION cidad dios 21 DO HIGEEVAILSATION es o 21 7 TASICAN ALVES nr 22 5 THE MULTIPLEX INFUSION SYSTEM cccccccccsccccsscccsccccsccccsccccsccccsscccccsccscscccccccccccccccccccccccssccscccccccccccccccccccscscoscccoscsces 23 5 1 GENERAL DESCRIPTION OF THE MULTIPLEX INFUSION SYSTEM ccccccscssoscosccsscccecccscccssccescessccsscecesccssecsscsecccssccsscecesecscecscesscesscesstees 23 5 0 M A 23 DS BEASIBIEEEY ANALYSIS 24 5 4 KEY ADVANTAGES OVER THE CURRENT IV SYSTEM ccccsccoscccsccessccssccsccssscesscecsccsssccecccesccssccssccsscecscecssesscecscecsccseccescecscesscecasesscecacesazers 25 5o USER INTERFACES enten te Mile a 25 5 6 QUANTITATIVE PHYSICAL AND CHEMICAL UNDERSTANDING AND MODELING OF FLOW AND MIXING IN THE MULTIPLEX INFUSION SYSTEM id de E 28 O EMPIRICAL STUDY asian 31 AMINEL DIU ON 31 6 2 EXPERIMENTAL EVENTS cu e 32 6 3 EXPERIMENTAL CONDITIONS NUR VER _2. asn xajdwoo aatu 5400 Ajquapiya Buiop uoyng sdwnd 0 Asea Apjuessaoauun wajskssiyj pue pesn siwazsAs eo3s3uojaq uoisnjul S Ui 915AS 51 uia3sAs Stt s uaams SiussJ0s ayy ay YM uoun uaamjaq 5141 ay eds au ysayew aye uaaJos UO JealD 51 y 5 uoneuiJojut 0 Ase 51 3 SIE 5 8 8 Isle S 8 S 8 S 8 e 2 8 5 4 5 4 5 2 5 48 5 48 sla sla v v v v v v v v 8 SIE Aye 215 214 121 v v amp u u 5 v a a a a 5 5 5 5 5 suorpe Wwaysds 510 paau 510 waysdAs uia3sAs ajese wajsAs uo uo paJjo suonon Jo u 5143 GUAM SISIYIAUIYI 235101 5143 YW sieAjap oy Asea ayipuy Ajyuapijja au3isn pe aBuisAs 915 5 sayeysiAy Ajisea ued JOM URI Ajisea ue e ejd 4 Ajise a Figure 25 Boxplots of responses to statements 9 16 Error bars indicate standard deviations Boxplots of responses to the first 8 statements Error bars indicate standard deviations Figure 24 x e eM N e 8 45 o c x gt E red a o 2 o x amp o n oO UE 515 2 8 5 8 5 8 2 2 518 al E al E al E E E a E ela a 2 2 2 3 2 3 2 3 3 Z 3 3 gt
3. 2 5 E 2 s 8 a 8 8 Mu Yi 5 Yi st a3uij s Aydwa 15 5 5143 uja3sAs 5101 wuopad ellInjmouyaas s adundse ue jo ose 3uppoM yy m YOM Asea B Ajisea jun saxe y ur op pjnous Aolua 07 99WN asam a15A5 0 aas 943149 032 siya 03 Jeal5 SI 3I y SSe3 auU o LL APPENDIX C HIERARCHICAL TASK ANALYSES The following tree diagrams depict parts of the IV therapy pro cess The colors in the HTA trees indicate who executes that part of the process Figure 27 features a legend for the HTA trees Legend Nurse 1 B Nurse 2 Figure 27 Color clarification for HTA trees ce Uy ip uoneo gua 10 asinu Jeu10 1sv Juud 32112 Z 1 Z uo ebuu s uo eunjeubis ind e z 27 222 TTA abuu s 7 TT uay TZ z 1 ebuu s out JBUIEJUOD osien 6 esoon 9 0 02 9 abuu s pajas ot 429000 MEJO 2727 G esoon o ebuu s uu op jw 06 uljnsu zzZT TZT 9 OS 14 TT ueu Z veld ebuu s ui nsul ajedald 0
4. Medication error Any error in the medica tion process Moyen et al 2008 Slip Failing to execute an in tended action Reason 1990 Lapse Failing to execute an in tended action due to a lapse in memory Reason 1990 Mistake Performing the wrong ac tion for the desired out come Reason 1990 Near miss An error that does not result in any harm Moyen et al 2008 Table 2 Types of errors and their definition An example of an adverse event is when a patient who is not aware of any allergies suffers from an allergic reaction to a drug In this case there is no error causing the adverse event In the case of a wound infection that is caused by a physician who ignored standard hy giene regulations we speak of a preventable adverse event PAE When an error occurs that does not re sult in any harm we speak of a near miss Moyen et al 2008 4 3 3 MEDICATION ERRORS According to the estimates of the Institute of Medi cine between 44 000 and 98 000 Americans die each year as a result of medical errors making it the 8th leading cause of death even by the lower estimate Kohn et al 2000 Mortality prolonged hospitaliza tion permanent health damage psychological im pact on patients family and caregivers and high costs are major consequences of medical errors Moyen et al 2008 Medication errors are more common in the ICU than in any other hospital department Kalisch amp Aebersold 2010
5. lt 7 PURSTIPAK 50 VOELDE mT 58 En Du 46 16s ES rijksuniversiteit groningen oningen Developing a System for Integrated Automated Control of Multiple Infusion Pumps The Multiplex Infusion System Groningen oktober 2013 Auteur Studentnummer Afstudeerscriptie in het kader van Opdrachtgever Begeleider onderwijsinstelling Begeleider UMCG Frank Doesburg 1697803 Master Human Machine Communication Wiskunde en natuurwetenschappen Rijksuniversiteit Groningen dr W Dieperink Intensive Care Volwassenen UMCG mw dr F Cnossen Artificial Intelligence Rijksuniversiteit Groningen dr M Nijsten Intensive Care Volwassenen UMCG 2013 Studentenbureau UMCG Publicaties Groningen Nederland Alle rechten voorbehouden Niets uit deze uitgave mag worden verveelvoudigd opgeslagen in een geautomatiseerd gegevens bestand of openbaar gemaakt in enige vorm of op enige wijze hetzij elektronisch mechanisch door fotokopie n opnamen of enige andere manier zonder voorafgaande toestemming van de uitgever Voor zover het maken van kopie n uit deze uitgave is toegestaan op grond van artikel 16B Auteurswet 1912 j het Besluit van 20 juni 1974 St b 351 zoals gewijzigd in Besluit van 23 augustus 1985 St b 471 en artikel 17 Auteurswet 1912 dient men de daar voor wettelijk verschuldigde vergoedingen te voldoen aan de Stichting Reprorecht Voor het overnemen van gedeelte n uit
6. zou de directie van het UMCG aanraden om met dit systeem te gaan werken vond de opdrachten met dit systeem makkelijk om uit te voeren Usability questions Question in English This system is easy to use This system is unnecessarily complex This system looks nice The main screen is clear and orderly The space on the screen is used efficiently The information on screen makes it clear what the system is doing It is clear which function belongs to a button It is easy to discriminate between infusion pumps The system delivers clear feedback on my actions Mistakes are easy to correct on this system can easily find the functions need feel like am in control of this system can work efficiently with this system think this is a safe system can easily adjust the administration rate can easily replace a syringe with this system can easily see how full a syringe is can easily see how long it takes until a syringe is empty It is clear to me what should do in case of an alarm enjoy working with this system would recommend the CEO of the UMCG to work with this system The tasks with this system were easy to perform x x slal 8 8 al 8 s la E e e e s e s 5 a 5 a 5 4 v v v v v v v v z gt S z z E z S z s z v u v 8 a amp a 8 8 8
7. Concurrent multitasking is performing two or more tasks at the same time Salvucci amp Taatgen 2009 Some tasks like walking and talking simultaneously can be performed effortlessly without any interfer ence Other tasks such as having a phone conversa tion while driving or talking to someone while writing a note are almost impossible to do simultaneously The reason is that these tasks require similar cognitive resources such as memory vision or manual opera tions The more overlap there is in the required cogni tive resources between the two tasks the more interference there will be when trying to execute both tasks simultaneously Wickens 2008 The computa tional model of threaded cognition Salvucci amp Taatgen 2008 offers a way to predict and explain multitasking performance by modeling the use of cognitive resources which are required for performing a task In the threaded cognition model a central ex ecutive control calls upon these resources when a task is performed Although multiple resources may be called upon at once a single resource can be assigned to one task at a time 4 4 2 SEQUENTIAL MULTITASKING In sequential multitasking there is more time to switch between tasks Salvucci amp Taatgen 2008 Of ten a task may be performed for several minutes or even hours before a secondary task is introduced Ex amples of sequential multitasking are writing a paper 19 and reading a letter or cooking a
8. emoq ui nsul uljnsul WOT OT mesg yajuIsig ETT CUI OT 0 e peeu TTT uinsul abuu s jw TT TTT Sueld OT ITT uonnjos UIJNSUI yy ebuu s 06 Il zesinn B Tesunn oale ul pajas TTZ Juud AMD ZZ uueday and 2 2 aoeld zz 272 TU WOS G ue d I emoq uuedeu 3 ejuirsiq abuu s e peeu uo ebuuks uo eunjeubis 5 TI uoneoyueA uuedeu abuu s abuu s ez uay 2 ueui E TZ E TT USUI 2 pu OS 477 10 asinu ce TE ip ue d abuu s uueday 7 ajedald 70 APPENDIX D FEASIBILITY ANALYSIS yses Table 8 lists the abbreviations used in the analysis Figure 29 displays the template that was used for drawing out the IV Tubings The following tubings are digitalized versions of the Below you will find the results of the feasibility analysis Figure IV tubings that were drawn out 28 displays the legend corresponding to the IV tubing anal Triple lumen Catheter CVC VSC Left Proximal Distal Medial CVD set Volumetric pump with 5 glocuse solution at 200ml h Syringe pump administering insulun at 0 4 ml h An three way valve Two interconnected three way valv
9. 8 2 3 EVENT EXECUTION TIMES The results showed that there was no difference be tween the two conditions in the time that it took to perform the tasks related to the five experimental events This indicates that tasks with neither of the two user interfaces are more time consuming than with the other user interface The lack of a difference in execution times indicates that novice users of the Multiplex user interface are equally as fast as expert users of the current infusion system 8 2 4 ERRORS One of the goals of this study was to reduce the likeli hood of errors in IV therapy Results showed that sig nificantly less errors occurred using the Multiplex user interface Errors were generally less severe and in two out of the three errors with this interface the error did not have any consequences as the participant noticed the error or the system made sure the error was han dled correctly In one of these three cases the partici pant tried advancing to the next event before the current task was finished The experimental model forced the participant to finish the task before the next event could start In a real situation it is possible that the nurse would have walked away at that point Although the alarm would keep repeating which in creases the likelihood that another nurse will notice the alarm it is preferable that the alarm is dealt with immediately In the separate pumps case alarms were ignored four times and an ord
10. Although the definition of usability was originally in tended for software engineering the term could be applied any system where is interaction between a human user with a system like a ticket dispenser or DVD player 4 2 1 USABILITY OF INFUSION PUMPS Although nurses in the ICU work with increasingly complex medical devices under hectic circumstances relatively few studies have been carried out that aim at identifying ways to improve the usability of these de vices It is a well known fact that many mistakes are made with intravenous medication Among the possi ble errors with IV therapy are setting up a different administration rate than is prescribed administering the wrong medication and not carrying out an order to change the medication or administration rate Husch et al 2005 In order to reduce the number of errors manufacturers have tried to come up with user friendly solutions Attempts have been made to im plement smart infusion pumps which provide deci sion support and are able to warn medical staff when a certain dosage endangers a patient However this did not succeed in reducing the number of medication er rors Carayon et al 2005 Rothschild Keohane et al 2005 Wetterneck et al 2006 These smart pumps where programmed to alert their users when an ad ministration rate exceeded ten times the suggested administration rate According to a study by Husch et al 2005 smart pumps were unlikely to p
11. of dealing with these events could also be pre vented or reduced Depending on the severity of harm additional costs can range from 382 up to 56 670 euro per case B Braun Melsungen 2011 It is currently not possible to assess how many adverse events can be prevented By anticipating on future actions the Multiplex Infusion System will have the potential reduce both the number and duration of sessions a nurse has to spend on physically configuring the infusion pumps It is currently not possible to quantify this cost reduction Approximately 5 of IV fluids is currently spilled at the ICV because of the conventional method of replacing almost empty syringes The Multi plex infusion system will have the potential to reduce the amount of unused infusion fluid by optimizing the use of IV fluids by automatically switching to a new pump when the old syringe is completely empty With an annual spending of 1 8 million on IV medication automated switch ing could save 90 000 per year 5 5 USER INTERFACE The Multiplex infusion system will have a touchscreen interface that will be placed at the bedside proximal to the patient and the infusion pumps Utilizing the currently available RS232 ports or through an infrared connection it is possible to control the current set of syringe pumps In this project developed a user in terface for the Multiplex infusion system The usability heuristics from M
12. s care which requires an additional 1 5 year educational program Besides basic tasks like washing and grooming a patient the ICU nurse monitors all medical devices surrounding the patient and has to ad just the medication accordingly The ICU nurse is also in charge of the preparation and administration of medication and maintains a record of the patient s progress In a daily deliberation with an intensivist a physician specialized in ICU patients interns and sometimes other specialists such as a surgeon or a ra diologist the patient s progress is reviewed and treatment is adjusted accordingly 3 2 INTRAVENOUS THERAPY Intravenous therapy or IV therapy is the administra tion fluids through a vein Infusion fluids can have sev eral purposes for example for restoring a fluid balance in the patient or as a carrier fluid for the administration of medication IV therapy is the most common form of therapy in the ICU of this study IV fluids are delivered by an infusion pump that is fitted with a syringe or in fusion bag containing the fluid The administration rate which is mostly based on the type of IV fluid and the patient s weight is programmed on the infusion pump When a pump is started IV fluid gradually flows through an IV line into the bloodstream of the patient 3 2 1 CATHETERS In order for IV fluids to enter the bloodstream pa tients are fitted with an IV catheter An IV catheter is a small flexible tube that i
13. It is important to note that this analysis featured a relatively small number of IV tubings As approximately 3000 patients are admitted in the ICU each year there are many different types of patients that could not be included in this feasibility analysis Therefore extending this analysis to a larger group of patients may yield in different results 8 2 EMPIRICAL STUDY In this section will discuss the results of the empirical study divided the discussion into four topics The number of clicks total execution times first response times errors and questionnaire responses 8 2 1 NUMBER OF CLICKS In this study an important goal was to create a user in terface with a better usability and a lower error rate than the current system where the nurse controls multiple pumps separately The Multiplex user inter face was designed to assist the nurse in difficult tasks and to take over stages of the IV process which are prone to error Based on HTAs of the experimental tasks with both user interfaces golden standard were obtained Re sults showed that when the golden standard was higher for one user interface then the actual number of clicks was also higher with that user interface The average number of clicks was consistently higher than the golden standard which was to be expected How ever in some cases the number of clicks was up to twice as high as the golden standard In the Multiplex condition this may be explained b
14. Moyen et al 2008 As intravenous therapy is the most common way to administer medi cation most medication errors are related to this form of therapy Moyen et al 2008 Tissot et al 1999 Ta ble 3 lists the errors that may occur during any stage of the IV process Figure 4 displays the most common IV medication errors as they are reported by the Brit ish National Patient Safety Agency 2007 According to this report 73 1 of all errors in the medication process occur during the administration and prepara tion stages Consequences of PAEs are often more severe with ICU patients than patients in other hospital depart ments as they are often critically ill and therefore more vulnerable when errors occur Giraud et al 1993 According to a study in a French ICU 1976 of medication errors were considered to be life threatening Tissot et al 1999 In a hectic and com plex environment such as the ICU the likelihood of preventable adverse events is twice as high compared to any other hospital department Rothschild Landrigan et al 2005 The most common cause of preventable adverse events in Dutch ICUs are admin istration rate related errors of which 5376 are directly related to the operation of infusion pumps Van Soest Segers et al 2009 The UMCG encourages its staff to report incidents which may be done anonymously Incidents may not always be reported due to time constraints underes timation of the severity o
15. Planned changes in the IV treatment will be imported automatically into the Multiplex infusion system via a link with the PDMS In case of an incoming order the entire panel corresponding to the relevant pump turns green Figure 9 and an audible beep is played A short summary of the order is displayed on the right side of the pump panel Clicking the pump panel starts a dia log asking for permission to change some parameter of the IV therapy such as the administration rate When a change in an administration rate is ordered the nurse only needs to confirm this order at the bed side instead of changing the administration rate man ually By automatically entering the desired administration rate administration rate errors can be prevented When the task is to replace an empty sy 127 ringe a checklist can be used to verify the execution of important steps The checklist is activated by default but the user can choose to deactivate it A user will be able to add a new infusion pump with a new drug via the upcoming PDMS In case of an alarm the entire panel corresponding to the relevant pump turns red and an audible beep is played as is displayed in Figure 9 A short summary of the alarm is displayed on the right side of the pump panel Clicking the pump panel starts a dialog with a description of the problem and suggestions how to solve the problem In order to change an administration rate manually the relevant pump needs to be cli
16. can be retrieved from memory Figure 5 When the system tries to retrieve a chunk from memory it will retrieve the most active chunk Due to noise it is pos sible that a distractor is retrieved instead of the target goal chunk If the activation of the target chunk is above the threshold value it is more likely that it is re trieved successfully If its activation is below the threshold it is less likely to be retrieved When a chunk is retrieved its activation is increased The more a chunk is rehearsed the more likely it is that it can be retrieved from memory Activation Time sec task 1 is interrupted by task 2 the chunk representing goal 2 is activated while the one representing goal 1 decays This process is illustrated in Figure 6 When goal 2 is achieved the goal activation model attempts to retrieve goal 1 which increases the activation cor responding to goal 1 making it more likely that the goal will be retrieved from memory The threaded cognition model incorporates much of the goal activation model although modeling memory is only a part of the threaded cognition mod el When an interruption is announced during a prima ry task a problem representation of this task is rehearsed before starting the interruption task in creasing its likelihood to be retrieved from memory af ter the interrupting task is completed The more time there is for this rehearsal the more likely it is that the problem represent
17. field of psychology only few studies focused on how multitasking affects the workflow and the frequency of errors in an ICU setting In an observational study in the ICUs of two hospitals 46 hours of concurrent mul titasking 1354 interruptions and 200 errors were doc umented Kalisch amp Aebersold 2010 In 46 of cases where nurses were administering medication they were interrupted Although this study did not find a significant effect of interruptions on the error rate it does illustrate the discontinuity in the workflow of the ICU nurse 4 5 USER BASED USABILITY EVALUATION User based usability evaluation is used to collect data from users as they interact with a system Dumas 2003 This data may be performance data or measures of user satisfaction Performance data such as execution times or the number of errors may be acquired by letting users perform several tasks with a system This allows for an objective way to compare the performance and efficiency of multiple systems to each other When combined with questionnaire data rating scales on user satisfaction this type of usabil ity evaluation is very useful when comparing multiple systems If the tasks closely represent real world tasks results of this evaluation also allow to be generalized to the real world A limitation of user based usability evaluation is that performance data is less useful when only a single system is evaluated As the aim of this study is to de
18. move on from step 1 to step 2 step 1 needs to be achieved by completing steps 1 1 and 1 2 HTAs are very easy to construct and interpret They are very useful for gaining insight in the structure of complex tasks In this study HTAs will be used to de compose and compare several tasks related to the in travenous medication process These HTAs will be useful when comparing how various tasks are per formed with both the current infusion system and the proposed control system Plan 1 do 1 then 2 then 3 3 Turn key clockwise Plan 2 do 1 1 then 1 2 2 Put key in the lock 1 1 Check pocket 1 2 Check bag Figure 7 example hierarchical task analysis tree of opening a door 5 THE MULTIPLEX INFUSION SYSTEM In this chapter will provide a theoretical description the proposed Multiplex Infusion System which will act as an automated controlling device for multiple infusion pumps First will provide a general descrip tion of the proposed system and its key features will then discuss the possible advantages of the Multiplex Infusion System over the current IV system Next will discuss the design considerations and functionali ties of the user interface in this chapter Finally will describe the physics and chemical aspects of the pro posed Multiplex Infusion System 5 1 GENERAL DESCRIPTION OF THE MULTIPLEX INFUSION SYSTEM The idea for this system started out from the need to reduce t
19. this difference was significant F 1 15 43 636 p lt 0 000 In the NOR syringe empty event the number of clicks was significantly higher in the separate pumps condition F 1 15 9 201 p 0 05 The same was the case in the change administra tion event F 1 15 39 894 p 0 000 In the interrup tion event there were more clicks in the Multiplex condition F 1 13 10 67 p lt 0 05 There was no sig nificant difference between the two conditions in the obstruction event F 1 15 0 539 p 0 475 In the Multiplex user interface the possibility to close a dialog window using the standard x button in the top right corner was disabled forcing the participant to make a choice during a dialog Unfortunately it is not possible to remove this button from the dialog window as this is a default Java feature As the partici pant clicked the x button a message appeared asking the participant to choose between two options for example accept an order or decline There were two cases during the interruption event where this mes sage was misread leading to multiple attempts to close the dialog window and additional clicks First response times per event Multiplex Syringe empty Separate pumps Syringe empty Multiplex NOR syringe empty Separate pumps NOR syringe empty Multiplex Change administration rate Separate pumps Change administration rate Separate pumps Interruption event e O ae Multipl
20. ICU nurses The UMCG has an ad vanced simulation center which is used for training of medical staff using realistic simulations An experi ment in the simulation center which allows for a real istic and controllable simulation of an ICU could test the effects of different interruptions and cognitive workload on the performance of IV therapy related tasks This experimental setup would also eliminate the need for simplifications of tasks such as replacing an empty syringe and solving an obstruction in an IV line allowing for a better generalizability to real tasks in the ICU 9 FUTURE WORK In this thesis the first steps were taken in the devel opment of the Multiplex infusion system In order to finalize this project there are several theoretical and engineering challenges that need to be faced and goals that need to be achieved Gain control over all infusion pumps for a single patient from a control system proximal to the pa tient This includes controlling both the syringe pumps and volumetric pumps Create an interactive system that guides the user in how to connect the tubing for a patient This will be a subsystem of the Multiplex Infusion Sys tem The compatibility matrix Appendix E is incom plete and knowledge about currently unknown interactions between medication should be add ed Ideally the Multiplex Infusion System knows all compatibilities of all combinations of medica tion in the ICU However it ma
21. Patel V L 2004 Heuristic evaluation of in fusion pumps implications for patient safety in Intensive Care Units International journal of medical informatics 73 11 12 771 9 Hilton E Haslett T M Borenstein M T Tucci V Isen berg H D amp Singer C 1988 Central catheter infec tions single versus triple lumen catheters Influence of guide wires on infection rates when used for replacement of catheters The American journal of medicine 84 4 667 72 Husch M Sullivan C Rooney D Barnard C Fotis M Clarke J amp Noskin G 2005 Insights from the sharp end of intravenous medication errors implications for in fusion pump technology Quality amp safety in health care 14 2 80 6 Kalisch B J amp Aebersold M 2010 Interruptions and multitasking in nursing care Joint Commission journal on quality and patient safety Joint Commission Resources 36 3 126 32 Kaye R amp Crowley J 2000 Medical Device Use Safety Incorporating Human Factors Engineering into Risk Management Guidance for Industry and FDA Premarked and Design Control Reviewers pp 1 33 Kohn L Corrigan amp Donaldson M 2000 To err is human building a safer health system Washington DC The National Academies Press Larsen G Y Parker H B Cash J O Connell M amp Grant M C 2005 Standard drug concentrations and smart pump technology reduce contin
22. THE PHYSICAL AND CHEMICAL CHARACTERISTICS OF THE MULTIPLEX INFUSION SYSTEM Integrating knowledge of the physical and chemical characteristics into the Multiplex Infusion System will be necessary to maximally understand what is happen ing within the system at any point in time Based ona number of analytical equations as well as numerical simulation the system ought to predict where in the infusion tree each individual infusion packet is at any time what the degree of dispersion is For this purpose knowledge of circulation concentra tion and half life of IV drugs is important In many cas es two substances S and S do not interact even in high concentrations Such a situation corresponds with a green cell in the compatibility matrix Appendix E When it is known that substances S and 5 do inter act the nature of this interaction can vary from inacti vation of one or both substances increased effectiveness of a drug toxicity or precipitation reac tions When only inactivation occurs the system could be set to allow the inactivation of a small per centage of the administered IV fluids 6 EMPIRICAL STUDY One of the goals of this study was to develop a user in terface with a better usability than that of the current ly used setup of multiple infusion pumps In the previous chapter presented a new user interface for the proposed Multiplex infusion system In this chap ter will discuss a user based usab
23. This model will also be used to schedule when for how long and at which administration rate each IV fluid should be adminis tered In order to describe and predict the physical and chemical aspects of Multiplex infusion it is important to provide a complete model of the physical behavior of the Multiplex infusion system Therefore this sec tion will decompose the physical and chemical aspects of the Multiplex infusion system so that they can be modeled in a later stage of this project 5 6 1 INFUSION TREES Currently most critically ill patients are treated with infusion systems that use one or more multi lumen in travenous catheters Each lumen is part of a separate infusion tree which will be defined as follows An intusion tree is defined as one or more infusion pumps connected to a single lumen that is directly placed with its tip in a vein ofa patient through a fun neling arrangement of one on one connecting tubings and valves Figure 12 illustrates the principle of an infusion three As a single tree is physically isolated from other infu sion trees we only focus on the physical and chemical aspects of a single tree Furthermore a central con cept of the proposed Multiplex infusion system is ad vanced control and the use of hardware that is commonly used in the ICU Potentially useful devices such as electronically controlled valves or advanced in jector systems are currently not part of the model as they are rarely u
24. be easy to distin guish from each other in order to prevent the wrong pump being selected 5 The user interface only needed to display information that was relevant to the current state of the system 1 which could contribute to a clear and minimalistic design 8 oradrenaline 0 2 mg ml 6 ml h NaCl 0 9 Infunderen 3 mi h spuit heeg over 3 wat Volume 32 mi over Spuit leeg over 10 wur en 39 minuten Pompprofiel Continue dosering Bolus snelheid 300 mi h kaliumchloride Infunderen H Spuit leeg Over 18 uar ex 59 minuten Figure 9 A screenshot of the main screen of the Multiplex user interface Although automated control over multiple infusion pumps would be one of the main features of the Mul tiplex infusion system the user interface still needed to allow its user to have basic control over the system as suggested by the third usability heuristic Therefore the user should be able to start and stop the pump to administer a bolus of IV drugs and to change the ad ministration rate provided a detailed description of these features in chapter 3 The buttons that facilitat ed these features needed to by easily recognizable and preferably matched well known real world symbols 2 The play icon which is universally used on televi sion remotes and radios to indicate starting the play back of a movie or song was chosen to indicate the start functionality on the user interface Likewise
25. core idea behind multiplex infusion Incompatible drugs A B and C are admin istered through a single lumen and sepa rated by a neutral buffer fluid 5 2 KEY FEATURES Since the Multiplex infusion system will be developed from the ground up it allows for the incorporation of various additional features that could partially take over several tasks of the ICU nurse and help to prevent errors The Multiplex Infusion System will have the following key features Control over all pumps from a single user inter face This bed side platform will manage all the infusions for a patient This includes control over volumetric and syringe pumps Multiple incompatible infusion fluids will be administered through a single lumen separated by a neutral buffer fluid Multiple infusion fluids would be administered pseudo simultaneously by rapidly alternating between pumps Incorporation of advanced administration pro files At the ICV many IV fluids are administered continuously Once a syringe is empty it needs to be replaced in order to maintain continuity Some types of medication like antibiotics are administered in multiple sessions a day Others may require a gradual increase or decrease in administration rate These are all examples of profiles A profile describes how a drug is admin istered in terms of time and administration rate Only a few types of infusion pumps provide ad ministration profiles for example pumps
26. deze uitgave in bloemlezingen readers en andere compilatiewerken artikel 16 Auteurswet 1912 dient men zich tot de uitgever te wenden Trefw Critical care multi infusion cognitive engineering ACKNOWLEDGEMENTS would like to thank my external supervisor Maarten Nijsten for his support ideas and enthusiasm during the course of this project would also like to thank my internal supervisor Fokie Cnossen for thinking along since the start of the project and providing useful feedback and ideas during every stage of the project My thanks go out to Wim Dieperink for providing accom modation in the UMCG and making sure had everything needed for this project Thanks to Koos van Ringelensteijn for bringing me into con tact with everyone within the UMCG who contributed to this project am grateful to Johan de Jong for believing in this project and allowing me to continue the project within the UMCG would like to thank Ellen Swiers and Marisa Onrust for showing me around in the ICU and answering all my ques tions Finally would like to thank my family and friends for their support during this project CONTENTS ABSTRACT 1 1 INTRODUCTION a fs o o a o 3 2 SOBIECHNVES PES 5 3 IPRAGCTICAL BACKGROUND uti 7 S T TBHEINTENSIVECARE UNIT o wer haat 7 3 2 INTRAVENOUS THERABY 8 SS d e 10 THEORETICAL BACKGROUND 555 eti a e de 13 SO SER INU Rs CE
27. en try will be expected to equal the CVP central venous pressure which is normally quite low compared to the other pres P venous entry sures lumen 5 6 5 DYNAMIC ASPECTS AN INFUSION TREE UNDER NON STEADY STATE The most important parameters of an infusion tree consisting of N pathways under dynamic steady state conditions are Infusion packet Small amount of a single infu sion fluid the minimal and cen tral element of the Multiplex in fusion concept IP vol ml Volume of an infusion packet Packet ID A unique identifier key of an infusion packet in order for the control system to keep track of this packet on its course from the infusion syringe or bag into the patient Pa Pressure variation resulting from pump itself AFlowy mleh 1 Variation in flow re sulting from the pump itself Dispersion Diffusion of the contents of a packet into neighboring packet 5 6 6 CHEMICAL ASPECTS For all substances used the Multiplex infusion system must know the maximally allowed concentration product C C where and are known throughout the infusion tree The assumption is that the concentration product within the infusion tree Cu Cy is lower than the concentration product within the patient Cy patient Cn patient This combina tion is already acceptable in conventional IV therapy 5 6 7 INTEGRATING KNOWLEDGE OF
28. for the ICU nurse is replacing an empty syringe or infusion bag Infusion bags are pre filled an do not require dissolving or mixing before the bag can be replaced Replacing a syringe requires a 50 ml sy ringe to be prepared before it can be attached to a sy ringe pump The process consists of three general stages preparation filling a syringe and verification by an additional nurse The preparation stage consists of gathering al necessary equipment for filling the sy ringe such as a diluent a 50 ml syringe and gloves Labeling the syringe so that the drug name and con centration are visible on the syringe is also part of the preparation stage How to fill a syringe depends on the way a drug is packaged For example insulin comes in 10 ml bottles and is dissolved in 40 ml of a glucose so lution Heparin is stored in 50 ml bottles which does not require an additional diluent An additional nurse compares the label on the original drug container with the label on the syringe in order to verify the contents of the syringe ADMINISTRATION If the task is to replace a syringe the current IV line needs to be traced from the pump up to the point where it is connected to the rest of the tubing Most often the IV line is connected to a valve which is used to allow IV fluid from multiple IV lines to pass through or to block the flow from one or more IV lines The pump is stopped and the valve is closed so that the old syringe can be disc
29. meal and watching xS television Sequential multitasking sometimes involves one task interrupting another while maintaining a rep resentation of the previous task in order to increase the likelihood that the first task will be completed Af ter completing the interrupting task the primary task may be continued An example of an interruption in the cooking and watching television task may be an alarm indicating that the oven is pre heated The inter ruption task may require a dish to be placed in the ov en and setting a timer When this task is finished one may continue watching television until the next inter ruption for example when the dish is ready How we are capable of returning to a previous goal can be ex plained by the goal activation model Altmann amp Trafton 2002 4 4 3 THE GOAL ACTIVATION MODEL Miller 1956 proposed the term chunk to describe how information is stored in our working memory A chunk can be a single digit a word of one or more syl lables a goal or some other type of grouped infor mation Miller found that humans are generally capable of storing between five and nine chunks in short term working memory In the goal activation model Altmann amp Trafton 2002 a chunk containing a goal is associated with an activation value which de cays over time A noisy threshold value which consists of background noise from other goal chunks distrac tors determines whether or not the target chunk
30. on the design considerations in the previous section a user interface was designed A screenshot of the resulting user interface is displayed in Figure 9 Screenshots of all parts of the user interface can be found under Appendix F Figure 10 A screenshot of the second tab of the Mul tiplex user interface The user interface consists of three tabs a main tab which displays all infusion pumps and pump controls a tab where general patient and treatment information can be consulted and a tab where the IV tubing is dis played so kaliumchloride 2 ml h Volume 49 7 ml over Spuit leeg over meer dan 1 Pompprofiel Continue do Bolus snelheid 300 ml h kaliumchloride infuussnelheid aanpassen 0 ml h insuline 1 IE 3 4 ml h propofol 20 me 10 mi h Figure 11 The dialog window for manually changing an administration rate heid Bolus Stop pomp The main tab contains panels which represent the cur rently used set of infusion pumps the names of the administered IV drugs administration rates an image representing the remaining volume in the syringe and an estimate of the time until the syringe is empty Ad ditional information and features are available when a pump is clicked on When clicked the panel expands displaying additional information and buttons for manually adjusting the administration rate adminis tering a bolus and stopping or starting the pump
31. or a dialog window in the separate pump case with a new order appeared With each user interface 4 of 5 possible events oc curred If the interruption event occurred with the one interface then the obstruction event would occur in the other The occurrence of these events was bal anced over all participants so that both events oc curred equally often After finishing the experiment with the second user interface a digital questionnaire was administered Data analysis For the syringe empty NOR syringe empty and the change administration rate events a within subjects design was used as these tasks oc curred in both user interfaces For the analysis of the interruption and obstruction event a between subjects design was used as these tasks only occurred once per experimental session 8 0 AANPASSEN 73 M NORADRENALINE 0 2 MG ML Varages Nicuwe spant 0 0 AANPASSEN J VOLUME TIV INFUNDEREN 3 0 RED 6 C cw C C 2 NIMOTOP 0 2 MGIML INFUNDEREN BD PLASTIPAK 50 1 8 a AND ASSEN TIV CO ACD O O O INSULINE Novaporid 1 IEIML IAN one Nieuwe spart INFUNDEREN 10 0 AANPASSEN aDC O O PROPOFOL 20 MG ML Figure 13 Screenshot of the interface for the Sepa rate pumps interface The pump adminis tering nimotop center displays an occlusion alarm due to an obstruction Duri
32. the Multiplex condition A paired t test was performed to see if there was a significant dif ference in the number of errors per condition There was a significant difference in the number of errors be tween the separate pump condition M 0 50 SD 0 730 and with the Multiplex Infusion System M 0 06 SD 0 250 t 15 2 150 p lt 0 05 7 2 5 QUESTIONNAIRE RESPONSES Figure 23 displays boxplots of grouped ratings for the statements in de questionnaire Boxplots of ratings per statement can be found under Appendix B The statements were grouped as follows Overall average all 22 statements System appearance statement 2 3 4 and 5 Intuitive design statements 6 7 8 9 11 19 Ease of use statements 10 15 16 17 18 General impression statements 1 12 14 20 21 41 22 m Group differences were found in ratings for system appearance and intuitive design For system ap pearance there was a significant difference between the Multiplex condition M 6 28 SD 0 84 and the separate pumps condition M 5 85 SD 1 33 t 59 2 001 p lt 0 05 There was also a significant differ ence in rating for intuitive design between the Multi plex condition M 6 0 SD 0 88 and the separate pumps condition M 5 65 SD 1 16 t 71 1 99 p 0 05 Significant differences were found in ratings on two individual statements This system looks nice was rat ed significantly higher for the Multiplex user interf
33. which featured both the Multiplex user interface and the manual operation of multiple simulated infusion pumps ICU nurses per formed multiple IV therapy related tasks during two simulated days in the ICU one simulated day for each user interface During a single simulated ICU day multiple different events occurred that required a cer tain task to be performed For example An alarm goes off indicating that an IV syringe is almost empty the event Therefore the participant needs to replace the syringe the task During the experiment clicks re sponse times task execution times and errors were recorded in order to be able to measure participants performance with each user interface After complet ing two simulated days at the ICU a digital usability questionnaire was administered Participants Sixteen ICU nurses at the University Medical Center of Groningen UMCG participated in this experiment Ages ranged between 29 and 58 the mean age was 42 3 Thirteen of the participants worked at the Thorax Intensive Care THIC and three worked at the Surgical Intensive Care CHIC Among the participants were 7 women and 9 men The partic ipants work experience in the ICU ranged from 2 months until 31 years with an average of 12 3 years All participants signed an informed consent form be fore participating in the experiment Apparatus A laptop running Windows Vista was used to run the experimental environment A model was creat
34. with target controlled infusion Davey 2005 The multiplex infusion system would be able to ad minister drugs using various complex profiles including profiles that require multiple pumps Incorporation of domain knowledge By incorpo rating knowledge on compatibilities of drugs the Multiplex infusion system would be able to guide the ICU nurse in arranging the IV tubing optimal ly By connecting the Multiplex infusion system with the patient data management system PDMS orders for changes in administration rates could be fed to the infusion pumps auto matically after a nurse confirmed the order at the bedside Integrated planning and control of alarms Sever al planned tasks could be scheduled to be exe cuted automatically Instead of having multiple sources of alarms alarm messages will be dis played on a single user interface 5 3 FEASIBILITY ANALYSIS In order to assess whether or not multiplexing IV med ication would reduce the number of required lumens in a clinical setting a feasibility analysis was per formed The IV tubing arrangements of 12 randomly selected ICU beds were completely drawn out by two ICU nurses Included in these schemes were all volu metric and syringe pumps the types of medication and administration rate the types and placement of all catheters and connectors The nurses received a tem plate to draw on a set of instructions and a set of ab breviations to use when drawing the scheme
35. Diagnosis The patient s underlying condition is diagnosed based on the patient s symptoms and history Although rare mistakes can occur in this stage For example If a patient suffers from a condition that the physician has never encoun tered before it is possible that the condition is not diagnosed correctly 2 Prescription Based on the diagnosis a therapy is prescribed Prescription errors may occur when the wrong type of medication or administration rate is ordered Dispensing A drug is prepared by the pharmacy or a nurse based on the order by a physician Preparation errors may occur when the order is misread or when the wrong concentration or the wrong drug is prepared The drug may also be dissolved in the wrong type of solution fluid or may be dispensed too late 3 Administration A syringe or bag containing infu sion fluid is attached to a syringe pump or volu metric pump respectively A nurse checks the compatibility of the added solution with other administered infusion fluids and decides which lumen will be used for administration This rou tine step is vulnerable to slips The compatibility can be misread or misinterpreted or the wrong lumen can be selected As a consequence in compatible drugs may be administered through the same lumen The next step in the administration stage is to program the desired administration rate and starting the pump Here it is possible to program the wrong administration It i
36. M 8 TO 10 ML H 2 ERRORS Table 6 The occurrence of errors and their description using separate pumps 7 2 6 OTHER RESULTS In the separate pump condition there were three dif ferent ways to perform the tasks corresponding to event 2 Noradrenalin empty For this task there was a second pump available which could be used One participant changed the administration rate from 8 ml h to 4 ml h on the almost empty pump The addi tional pump was also set up at 4 ml h after which the participant increased the administration rate stepwise on the new pump and decreased the administration rate of the first until the first pump stopped and the new pump was set at 8 ml h Another participant set the administration rate of the new pump to 8 ml h started the pump and immediately stopped the pump with the almost empty syringe Four participants did not use the additional available pump These partici pants stopped the pump replaced the almost empty syringe and started the pump again The other ten par ticipants gradually decreased the administration rate of the first pump and increased the administration rate of the second pump maintaining the total administra tion rate at 8 ml h 1 2 5 4 3 6 7 Overall average Multiplex 5 96 Overall average Separate pumps 5 83 System appearance Multiplex 6 28 System appearance Separate pumps 5 85 Intuitive design Multiplex 6 0 Intuitive design Separate pumps 5 65 44 E
37. Participants also com pleted tasks equally as fast with the Multiplex user interface as with a simulation of the currently used IV pumps This indicates that working with the Multiplex user interface is easy to learn and this system is more effective in preventing errors Having more experience with the Multiplex user interface may emphasize the difference between both user interfaces even more Previous studies focused on evaluating the usability of individual infusion pumps and suggesting changes to their design This study is unique in the sense that it presents an entirely new user interface for the control of multiple infusion pumps and it compares the usabil ity of this user interface to that of the conventional operation of multiple infusion pumps using a user based usability evaluation A strength of this study is that it demonstrated how a user friendly user inter face can prevent errors in the operation of multiple in fusion pumps Previous observational studies in ICUs were not able to demonstrate the influence of interruptions on the occurrence of errors Interruptions are difficult to pre dict and it is difficult to assess whether or not an inter ruption is the cause of an error This study included a task with a controlled interruption A total of 11 errors was recorded of which 7 occurred during the interrup tion task The goal activation model Altmann amp Trafton 2002 and the model of threaded cognition Salvucci a
38. _ _ 33 7 RESULTS RM HM 39 39 B DISCUSSION A EN idet i ei eii n E 45 81 ANA 45 SZ e Ee bs SEDI DA RR e O 45 8 3 LIMITATIONS AND STRENGTHS OF THIS STUDY une 48 8 2 SUGGES TIONS FOR FURTHER RESEARCH e bed 48 9 FUTURE WORK o eben EE b e o 51 10 CONCLUSION SE 53 El REFERENCES id 55 ______ _ 6_6__0__ 606_ 6_6__ ______ __ 59 AAPPENDIXB USABILITY OUESTIONNAIRE 61 APPENDIX GC FIIERARGHIGCAL TASK ANA SES 59 65 APPENDIX D FEASIBIHTY ANALY S 69 APPENDIX E COMPA TB _ ____ ____ 83 APPENDDOEEMTUETIPEEXEUUSER INTERFACE snee ss ees des voe Ed CE 84 ABSTRACT Most errors in ICUs are related to intravenous IV therapy Previous studies suggested that hard to operate infusion pumps and the high cognitive workload for ICU nurses con tribute to these errors Conventional IV therapy requires separate lumens for incompatible IV drugs This often re quires the placement of additional catheters which in creases infection risk and physical discomfort for the patient In this thesis a control syste
39. ace M 5 92 SD 0 996 than for the separate pumps M 4 83 SD 1 467 t 11 2 315 p 0 05 The statement enjoy working with this system was rated significantly higher for the separate pumps M 5 58 SD 1 165 t 11 2 262 p lt 0 05 Responses to all other questions were not found to be significantly dif ferent between conditions Event Condition Description of the error SYRINGE EMPTY SEPARATE PUMPS NOR SYRINGE EMPTY MULTIPLEX INTERRUPTION SEPARATE PUMPS 42 MULTIPLEX OBSTRUCTION SEPARATE PUMPS PARTICIPANT IGNORED EMPTY SYRINGE ALARM PARTICIPANT STOPS THE NORADRENALIN PUMP THEN PROCEEDS TO THE CHECKLIST AFTER WHICH THE SYSTEM AUTOMATICALLY CORRECTS THE ERROR BY STARTING THE PUMP PARTICIPANT DID NOT ADJUST THE ADMINISTRATION RATE OF THE MORPHINE PUMP PARTICIPANT IGNORED OBSTRUCTION ALARM PARTICIPANT IGNORED OBSTRUCTION ALARM PARTICIPANT IGNORED OBSTRUCTION ALARM AND FORGOT TO ADJUST THE ADMINISTRATION RATE OF THE MORPHINE PUMP 2 ERRORS PARTICIPANT PRESSED THE SPACE BAR INDICATING THAT THE TASK WAS FINISHED WHILE THE ADMINISTRATION RATE OF THE INSULIN PUMP HAD TO BE ADJUSTED FIRST PARTICIPANT CLICKED ON THE POTASSIUM CHLORIDE PUMP INSTEAD OF THE PROPOFOL PUMP PARTICIPANT NOTICES THE ERROR AND RECOVERS FROM IT THE TASK WAS TO RESOLVE AN OBSTRUCTION IN THE NIMOTOP PUMP PARTICIPANT STARTS WITH REPLACING THE NIMTOP SYRINGE THEN THE ADMINISTRATION RATE OF NORADRENALINE IS ADJUSTED FRO
40. adm rate 1 Click pump Figure 16 Hierarchical task analyses for the change administration rate task using the Multiplex user interface left and the separate pumps interface right Multiplex Plan 1 1 then 2 interruption event Plan 2 37 1 1 then 1 2 1 Change gt administration 3 1 then 3 2 then 3 3 rate 2 Interruption 3 Solve obstruction 3 2 Click 3 3 Click Check if Done obstruction is solved task 3 1 Click propofol pump 1 1 Click 1 2 Click yes insulin pump Seperate pumps Plan 1 1 then 2 interruption event Plan 2 1 1 then 1 2 1 Change 2 Interruption 3 Solve Plan 3 administration 3 1 then 3 2 task obstruction rate 1 1 Click 1 2 Confirm 3 2 Click start pump 3 1 Click increase adm knotted IV line rate 2x adm rate Figure 17 Hierarchical task analyses of the tasks related to the interruption event using the Multiplex user interface top and the separate pumps interface bottom Multiplex Solve obstruction Plan 1 First 1 then 2 then 3 2 Click Check if obstruction is solved 1 Click 3 Click propofol pump Done Seperate pumps Solve obstruction Plan 2 1 then 2 1 Click knotted IV line 3 2 Click start pump Figure 18 Hierarchical task analyses of the tasks related
41. adrenalin NaCl nimotop insulin propofol and morphine 6 2 EXPERIMENTAL EVENTS Five different experimental events were created that needed to be dealt with by performing tasks that were common for the ICU nurse The five events and a global task description corresponding to the event are listed in Table 5 Figure 14 to 18 display hierarchical task analyses HTAs of all five experimental tasks as they are performed using the Multiplex user interface and the separate pumps user interface in the experi ment The syringe empty event is one of the most com mon events in the ICU and was therefore relevant to include in the experiment When a syringe is almost empty the pump where the syringe is connected to sounds an alarm The ICU nurse then needs to stop the pump replace the syringe and start the pump again in order to deal with this event The NOR syringe empty event may seem similar to the syringe empty event but the task corresponding to this event can be quite different Noradrenalin is a drug that influences the blood pressure of the patient and interrupting the administration of this drug could cause an instant drop in blood pressure When a sy ringe of noradrenalin is empty it is common for the ICU nurse to use a second infusion pump with a full syringe of noradrenalin Some nurses gradually de crease the administration rate of the first pump while increasing the administration rate of the second pump until the desired admini
42. artment may be easier to achieve A study in a pediatric intensive care unit showed that a combination of standardizing drug concentrations smart pumps and human engineered medication la bels reduced the number of reported errors by 73 However it remained unclear if and by how much the smart infusion system contributed to the reduction of errors Larsen Parker Cash O Connell amp Grant 2005 Melles Freudenthal de Ridder amp Snijders 2004 proposed the integration of information sources to build a support system for ICU nurses As many elec tronic medical devices are essentially computers it should be possible to extract information from them and build a system that collaborates with the nurse by giving personalized feedback reminders and support trough relevant checklists that correspond with the current situation Laxmisan et al 2006 also suggest such a support system for ICU nurses in order to re duce the memory load during multitasking Although such a system sounds promising there have been no reports of a practical implementation or experimental testing of such a system 4 4 MULTITASKING Multitasking is generally regarded as performing mul tiple tasks simultaneously However there is more than one type of multitasking In this section will dis criminate between two types of multitasking concur rent multitasking and sequential multitasking Salvucci amp Taatgen 2009 4 4 1 CONCURRENT MULTITASKING
43. as fast as with the familiar user inter face This indicates that working with the Multiplex user interface is easy to learn Based on the analysis of individual statements in the questionnaire it cannot be said which user interface was more satisfying for the user However results from the grouped questionnaire indicate a higher preference for the Multiplex user interface based on system appearance and intuitive design aspects The Multiplex user interface demonstrated a higher efficiency and a lower error rate than the separate pumps user interface The Multiplex user interface was also easy to learn but its learnability cannot be compared to that of the separate pumps user interface as participants were already familiar with the separate pumps condition Users of the Multiplex user inter face were generally more satisfied with its appearance and intuitive design aspects Overall the Multiplex us er interface has a better usability than the separate pumps user interface 8 3 LIMITATIONS AND STRENGTHS OF THIS STUDY The feasibility analysis featured only a small number of IV tubings Although the results were statistically sig nificant the type of patients in an ICU may vary from day to day which may not be representative for the total population of ICU patients suggest that this analysis is repeated using a larger sample size and dur ing a longer period of time in order to be able to gen eralize the outcome to a larger populati
44. ase of use Multiplex 6 12 Ease of use Separate pumps 62 General impression Multiplex EE 5 45 a oo pm Figure 23 Boxplots of grouped questionnaire responses Group averages are displayed on the right of this figure Error bars denote standard deviations There was a significant difference in system appearance and in intuitive design be tween the Multiplex and Separate pumps user interface 8 DISCUSSION In this chapter will start with discussing the results of the feasibility analysis on IV tubing in section 8 1 and of the empirical study in section 8 2 will then discuss the limitations and strengths of this study in section 8 3 Ideas for future research will be discussed in sec tion 8 4 8 1 FEASIBILITY ANALYSIS One of the main goals of this study was to decrease the number of lumens that were necessary for IV ther apy The analysis of IV tubing in chapter 5 indicated that with the Multiplex infusion system less lumens would be necessary than with the current IV therapy methods As most patients in the ICU have a triple lumen catheter the results suggest that in most cases a double lumen catheter would be enough Half of the patients in the analysis had a peripheral catheter ad ditional to the standard triple lumen catheter which would not be required using the Multiplex infusion system This could prevent physical discomfort re duce infection risk in a large group of patients save time and reduce costs
45. asks were performed on a computer real world actions such as replacing a syringe con necting or untangling an IV line were also simulated Questionnaire The questionnaire consisted of 22 us 33 ability related statements per user interface The Eu statements can be found under Appendix B The statements were grouped in five categories The overall category considered all statements The system appearance category considered state ments regarding the participant s visual impression on the user interface Intuitive design considered statements on whether or not it was clear how the user interface should by operated Ease of use considered whether or not functionali ties on the user interface were easy to use General impression considered the overall impres sion and opinion about the user interface A rating using a seven point Likert scale was used to measure the amount of agreement with each state ment The rating levels for the Likert scale were as fol lows completely disagree strongly disagree disagree neutral agree strongly agree and completely agree Design and procedure Before commencing with the experiment patients received an informed consent form that they were asked to read and sign Partici pants received a description of a patient case on pa per Anonymous participant data such as age gender ICU department and years of working experience were gathered before commencing with the experi m
46. ation can be retrieved later on Salvucci amp Taatgen 2008 If the interruption is not announced there is no time to rehearse which makes it is less likely that the problem representation of the primary task is retrieved System attends to retrieval cue for Goal 1 154 1 0 4 Goal Goal 2 Activation 0 0 SE 0 3 3 3 345 9 TER Time sec Figure5 From Altmann amp Trafton 2002 The acti vation of a chunk over time The dotted line represents a threshold above which the chunk can be retrieved from memory In sequential multitasking there are multiple chunks in memory which represent the goals of the relevant tasks Take for example task 1 and task 2 and their goals goal 1 and goal 2 As task 1 is being performed the chunk representing goal 1 gains in activation As Figure6 From Altmann amp Trafton 2002 The acti vation levels of two chunks representing the goals of two tasks As one task is ac tive the activation of the corresponding chunk increases while the other decays 4 4 4 MULTITASKING AND INTERRUPTIONS IN THE ICU Both concurrent and serial multitasking are common parts of the job of the ICU nurse For example when starting or adjusting an infusion pump with noradrena line which affects blood pressure a nurse needs to monitor a screen displaying the patient s blood pres sure and set up the infusion pump at the same time Nurses also need to remember upc
47. cked on first which then expands and enables the user to access the menu where the administration rate can be changed Figure 11 displays this menu After selecting and confirming the desired administration rate the menu closes and the administration rate is changed The second tab which is depicted in Figure 10 dis plays general information on the patient and which IV drugs are administered This data will be imported from the future PDMS The third tab Appendix F displays the tubing for the IV therapy This tab will be displayed when a new syringe or pump is attached The system will guide the nurse in connecting the tub ing efficiently based on knowledge on the compatibil ity of the administered IV drugs However as the current knowledge base on drug compatibilities is in complete this feature is not yet implemented in the user interface 5 6 QUANTITATIVE PHYSICAL AND CHEMICAL UNDERSTANDING AND MODELING OF FLOW AND MIXING IN THE MULTIPLEX INFUSION SYSTEM Briefly the essence of the Multiplex concept is fully integrated and centralized control of multiple infusion pumps These pumps deliver infusion fluids through a tree like network of IV lines to a single lumen that di rectly delivers the infusion fluid intravenously The complete Multiplex concept involves full control of all deliveries of IV fluids to all lumens The Multiplex infu sion system will include an internal model that keeps track of all administered infusions
48. dentify usability problems with heuristic analysis Molich amp Nielsen 1990 Another limitation of heuristic evaluation is that it does not provide a solution to the usability problems it identifies The aim of this study was not to improve usability of the infusion pumps that are cur rently used Instead the aim is to develop a new user interface which takes over the control of the current system of multiple infusion pumps A user based usa bility evaluation of a prototype is more appropriate for this goal 4 7 HIERARCHICAL TASK ANALYSIS Hierarchical task analysis HTA is a way to decompose a task into smaller subtasks using a hierarchical tree like structure Stanton 2006 An advantage of a HTA is that it allows analysis and comparison of the struc tures of complex tasks In a HTA tasks are decom posed into a main goal and one or more sub goals that have to be performed in order to achieve that goal These sub goals may also consist of one or more sub goals depending of the amount of detail that is used the HTA A plan describes the order in which the sub goals are performed for example Perform action A and then action B or perform action C Figure 7 shows an example of a hierarchical task analy sis tree of opening a door Plan 1 describes the steps to achieve the goal which are completing steps 1 2 and 3 Note that the hierarchy demands you to follow the tree structure in a depth first fashion In order to
49. e Lumen Dual Lumen Triple Lumen Figure 1 Cross sections of a single dual and a tri ple lumen catheter 3 2 2 CONNECTORS IV therapy requires various disposable components which are connectable to each other with Luer locks A Luer lock is a standardized type of fitting that allows for a leak fee connection between a set of connectable components will refer to these components as con nectors Although catheters are part of the set of dis posable connectors they are generally not replaced unless there are signs that the catheter is not function ing properly or when it is believed that it has caused an infection Other connectors which are used for IV therapy are syringes IV lines and valves Syringes are replaced after 24 hours or when they are empty IV lines and valves are generally replaced after 4 days 3 2 3 TUBING In order to provide a clear description of all aspects of IV therapy it is necessary to provide a definition which encompasses the collection of connectors that are used in IV therapy and the way in which they are in ter connected will refer to this as tubing and it is de fined as follows Tubing refers to the complete configuration of con nectors between all infusion pumps and all catheters in a patient receiving IV treatment For every patient receiving IV treatment a tubing is ar Figure 2 where the tubing consists of multiple IV lines ranged A patient s tubing may be described in words a t
50. e of 1 7 errors occur per patient per day Because of the high number of infusion lines that run from each infusion pump to the patient s catheter often end up in a spaghetti like tangle which is unfortunately unavoidable Raymer amp Smith 2007 Untangling these lines can be a very time consuming business Reducing this problem demands a reduction in the number of IV lines This may be achieved by interconnecting multiple IV lines allowing multiple drugs to flow through a single IV line Due to the incompatibility of sev eral drug pairs nurses sometimes avoid this option although they are able to check which drugs are compatible and could be combined Because of an expected further increase in the number of infusion pumps per patient an increase of the cur rent problems with IV therapy is also expected In this exploratory study we propose the Multiplex infusion system This system acts as a control device for multiple infu sion pumps at the same time Instead of operating multiple devices separately a nurse controls multiple pumps from a single user interface A smart control algorithm allows for mul tiple incompatible drugs which currently would be adminis tered separately to be administered sequentially by separating them with a neutral fluid This can be achieved by automatically switching between multiple infusion pumps se quentially The control system optimizes the number of drugs that flow through a single IV line As a
51. e of patients they typically accommodate there is much overlap and collaboration between them The THIC typically focuses on patients with conditions re lated to the thorax area for example lung or heart transplant patients The CHIC typically houses trauma patients patients who have had surgery or patients with multi organ failure Both departments collabo rate by exchanging nurses when understaffed or by taking over patients when one of the ICUs tends to get overcrowded Both ICUs consist of a ward with multi ple beds which can be separated by a curtain There are also separate rooms in order to isolate a patient when needed At the left side of a typical ICU bed there is a docking station that can hold multiple sy ringe and volumetric pumps and a screen that will be used for the future patient data management system PDMS At the other side of the bed there is a moni tor that displays the heart rate oxygen levels and blood pressure At the head of the bed there are often 7 multiple feeding pumps For some patients there aa additional machines used for respiratory support dial ysis or EEG measurements 3 1 2 THE TASKS OF THE ICU NURSE The ICU nurse is responsible for the immediate care of one or two patients in the ICU depending on the amount of care the patient requires The tasks of the ICU nurse differ from those of a regular nurse as ICU nurses are more involved in the medical aspects of a patient
52. e often used to provide input to the system and graphical output is commonly displayed on an LCD screen 4 1 2 TOUCHSCREEN USER INTERFACE A touchscreen user interface is a specific type of GUI On a touchscreen interface the user provides input by touching the same screen on which the system pro vides its output Because the displayed content can be changed dynamically there are various ways in which users can provide input e g button presses tapping pinching or sliding their fingers on the screen The touchscreen user interface is a very versatile interface and it is used increasingly in various mobile devices as well as medical devices 4 2 USABILITY Usability refers to the extend in which a user can effi ciently and enjoyably interact with a computer system Nielsen 1994a suggested that a system with good usability should meet the following five criteria 1 Learnability It is easy to learn how to work with the system 2 Efficiency When a user has learned to work with the system a high level of productivity is achieved 3 Memorability It is easy to return to work with a system after a user has not used the system for a while He or she should not need to learn to work with the system all over again 4 Lowerror rate The system has a low error rate If an error occurs it is easy to recover from it 5 Satisfaction The system is pleasant to use Users like to work with the system and are satisfied when using it
53. eady state Ina steady state we have one or more non zero flows that are all constant Therefore non changing flows and pressures are expected theoretically This assumes that infusion pumps are able to generate constant flows A non steady state is a situation where one or more pumps are changing their infusion rates Since the Multiplex system may sometimes change infusion rates every few minutes a significant part of the total time the system is running it may not be in steady state Below will try to define the distinction between a steady state and non steady state more precisely in quantitative terms 5 6 4 DYNAMIC ASPECTS OF AN INFUSION TREE UNDER A STEADY STATE The most important parameters of an infusion tree consisting of N pathways under dynamic steady state conditions are Cross section mm2 Surface of a relevant complete cross section in the infusion tree Flow through Flow mleh 1 Flow though each pathway Note that this flow is time dependent and may vary in the Multiplex infusion system Flow ml eh 1 Where Flowtot Flow1 N mleh 1 Flow through any the oretically or practically relevant cross section in the infusion tree Vol through Vol ml 29 Vol ml Where Voltot lt ZVol1 N TOR since the individual volumes all contain the volume of the common final part the lumen of the venous catheter Pa Pressure at the point where the infusion fluid enters the blood stream Thus Pvenous
54. ed in Java that simulates the behavior of mul tiple running infusion pumps over time keeping track of syringe volumes at certain administration rates On top of this model two user interfaces were created us ing the Java Swing library One represented 7 separate syringe pumps and the other represented the display of the Multiplex Infusion System The separate pumps user interface was modeled after the Alaris Asena sy ringe pump and featured the same functionalities and menu structures In order to display the user interfac es an external monitor resolution 1920 x 1080 60Hz was used in portrait mode A keyboard and mouse were used to record input from the partici pants All key presses and clicks were recorded so that they could be analyzed after the experiment Patient cases For each user interface a patient case and a set of 6 relevant drugs was created in consulta tion with an experienced ICU nurse Two patient cases were created carefully so that the difference between the used sets of drugs did not affect the execution of either of the task sets nor created an unfair advantage when working with either user interface The Multiplex patient case was a 43 year old patient who had suffered a heart attack The patient received noradrenalin NaCl potassium chloride insulin propofol and morphine The separate pump case was an 18 year old patient who suffered from a subdural hematoma after a bike accident This patient received nor
55. enalin syringe of nor syringe is almost adrenalin empty Change ad Incoming order to Change the ministration change an admin administration rate istration rate rate of a pump Interruption Order to change Read order event an administration Answer ques rate tion regarding one of the bstruction in pumps during ps phone call Change the administration rate as ordered Solve the ob struction Obstruction Obstruction in Solve the ob pump struction Table5 The 5 experimental events and a global task description corresponding to the event In order to prevent the participant from anticipating on an interruption the obstruction event was de signed as an alternative to the interruption task Both the interruption task and the obstruction task oc curred only once in the experiment while the other tasks occurred twice once using each user interface 6 3 EXPERIMENTAL CONDITIONS The experimental tasks were performed using two simulated user interfaces which were displayed on a monitor in portrait mode There were two conditions in the experiment the Multiplex condition involved the use of the Multiplex user interface that discussed in chapter 5 The separate pumps condition involved controlling multiple separate pumps and its design and menu structure was based on the currently used Alaris Asena syringe pumps A screenshot of the sepa rate pumps interface is displayed in Figure 13 As all experimental t
56. encies may be re evaluated and infusion pumps may be re programmed accordingly 3 3 INFUSION PUMPS The UMCG owns and maintains about 2500 infusion pumps in total There are two main types of infusion pumps volumetric pumps and syringe pumps Volumetric pumps are used to deliver high volumes of IV fluids with moderate up to high administration rates e g 5 to 999 ml hour IV fluids for volumetric pumps are contained in bags which are hung above the pump An IV line runs from the IV bag through the pump which uses a peristaltic mechanism in order to control the administration rate Syringe pumps are mostly used for small up to mod erately high administration rates 0 1 200 ml h Typi cally a 50 ml syringe with IV fluids is loaded onto the pump The pump gradually pushes the plunger of the syringe thereby pushing the IV fluid outwards 3 3 1 GENERAL FUNCTIONALITY OF AN INFUSION PUMP There are several manufacturers who produce infusion pumps Although the designs may vary the general functionality of infusion pumps is comparable The most common operations with IV pumps are listed in Table 1 Function Description Start stop Starting or stopping the administration of an IV flu id Bolus Rapidly administering large volume of IV fluid Purge Completely filling an IV line with IV fluid Also known as flushing Set up administration Setting the rate of admin rate istration of IV fluid ml h Titrate Adjusti
57. ent The experiment consisted of two blocks of training and experimental tasks Figure 19 Half of the partici pants started with the Multiplex user interface the other half started with the separate pumps During the training stage the execution of all relevant tasks was explained and practiced Participants were told that they would encounter several events which they had to deal with How to deal with an event was practiced during the training stage Participants were instructed to perform the tasks in the same way as they would with a real patient A clock in the top right screen was used to indicate the simulated time of day At the start of the experi ment the time was 8 30 AM and the simulated day ended at 4 00 PM Events occurred at random mo ments in a randomized order during this day After an event participants could speed up time until the next event The participant pressed the space bar in order to indicate that he or she was done dealing with the event which would also speed up the time If an alarm was still active the time would not speed up allowing the participant to finish the task If a participant indi cated that he or she was done with a task by pressing the space bar while the task was not finished this was marked as an error During an event time passed equally as fast as in real life normal speed As soon as an event started a vis ual and audible alarm was presented on the pump were the alarm occurred
58. er A catheter containing two lumens therefore allowing for two separated flows of infusion fluids Infusion packet Small amount of a single infusion fluid the minimal and central element of the Multiplex infusion con cept Infusion tree One or more infusion pumps connected to a single lumen that is directly placed with its tip in a vein of a patient through a funneling arrangement of one on one con necting tubings and valves IV catheter catheter that is placed in a vein of the patient Contains one or more lumens IV configuration The way that all IV lines are in ter connected for a single patient Also referred to as tubing IV treatment NL intraveneuze behandeling Intravenous treatment or intravenous therapy is the infusion of fluids into the vein of a patient Lumen A hollow tube that allows for the insertion of a fluid into the bloodstream of a patient Luer lock Standardized system of fluid fittings which are used to make leak free connections between the male and female part of medical equipment Multiplex infusion system An automated control system that controls multiple infusion pumps simultaneously in order for multiple different fluids to be administered sequentially through a single lumen Peripheral catheter IV Catheter that is placed in a peripheral vein which allows the administration of fluids into the blood stream of a patient Peripheral vein Any vein that is not inside the chest or ab d
59. er to change an administration rate was neglected twice Five out of eight errors with this system occurred during the interruption event The in terruption event consisted of 3 tasks an order to change an administration rate an assignment to look up information on a pump and an obstruction Ac cording to the goal activation model Altmann amp Trafton 2002 the nurse would need to maintain a representation of three goals corresponding to these tasks during the interruption event The first two tasks in this event followed each other up rapidly and the third was introduced during the execution of the se cond task This increased the cognitive workload which increases vulnerability to errors This also could have affected the rehearsal time making it less likely that a goal node is retrieved successfully Salvucci amp Taatgen 2008 As working memory should be able to hold between 5 and 9 chunks of information Miller 1956 the three goal representations are unlikely to use up all working memory capacity It is possible that participants were also maintaining other goals which were not directly related to the experiment As most participants returned to work in the ICU after the ex periment they may have been maintaining other goals that were related their work or personal situation Therefore it is likely that the working memory capaci ty of some participants was fully used during the inter ruption task which resulted in error
60. ere caused by an incorrect setup of an infusion pump lead ing to an administration rate that is either too high or too low 10 of all medication errors occurred in the preparation of the IV therapy leading to events where the wrong drug or the wrong concentration of a drug was administered The incompatibility of infusion fluids is currently dealt with by administering incompatible fluids separately In the hospital of the current study patients in the ICU typically have a central venous catheter CVC which allows for three separated flows of infusion fluids to enter the bloodstream known as a triple lumen catheter Often the number of available lumens is too low for the number of incompatible infusion fluids Therefore there is often the need to place additional peripheral cathe ters which causes physical discomfort for the patient and in troduces additional infection risks Evans et al 2012 Hilton et al 1988 In the ICU in the current study the multitude of infusion pumps increases the difficulty of the ICU nurse s job Com plaints have been made about the high number of mainte nance switching and monitoring actions that these pumps require The ICU nurse also needs to be able to discriminate between up to twelve very similar pumps often in a hectic environment According to a study by Donchin et al 1995 in 13 ICU around 178 activities take place at the bedside of a pa a tient per day and an averag
61. es Atriple lumen catheter An extension set measuring the central venous pressure Figure 28 Legend for the IV tubing forms Abbreviation Meaning Abbreviation Meaning CVC Central venous catheter INS Insulin VJ Vena jugularis NOR Noradrenalin VSC Vena subclavia DOPA Dopamine VF Vena femoralis MID Midazolam PIV Peripheral IV catheter MORF Morphine A Arm KAL Kalium potassium B Been Leg MAG Magnesium H Hand MIL Milrinone V Voet foot VASO Vasopressin L Left G5 Glucose 5 R Right NaCl Natrium Sodium Chloride 0 9 TACRO Tacrolimus AMIO Amiodarone HYDO Hydrocortison PROP Propofol Table 8 Abbreviations used in the IV tubing schemes Type catheter locatie kant Figure 29 The infusion form which was used as a template for drawing out an IV tubing Triple lumen Catheter CVC VJ Right GL NaCl Distal 16g 5 0 ml h CVD set Proximal 18g Medial 18q KAL 2 0 ml h MIL 4 0 ml h PIV Hand Right PROP 4 0 ml h PIV Hand Left MAG 0 8 ml h G5 h NaCL 3 0 ml h KAL 0 0 ml h AMIO 2 0 ml h INS 4 6 ml h NOR 6 0 ml h PIV Hand Right Triple Iumen Catheter CVC VF Right Proximal Distal Medial GI 200 ml h NOR 3 0 ml h INS 0 4 ml h 2 0 m thin Triple lumen Catheter CVC VSC Left Proximal Distal Medial HEP 3 2 ml h NOR 1 0 ml h MAG 0 8 ml h VANCO 2 6 ml h NACL 2 0 mish Triple lumen Catheter CVC VF Left Medial Distal Pr
62. esearch on the usability of infusion pumps has been limited to the use of single pumps In order to prevent errors with IV medication and improve the usability of infusion pumps effective ly pump manufacturers and researchers should study the interaction between the nurse and multiple infu sion pumps in a clinical setting 4 3 HUMAN ERROR The best people can make the worst mistakes error is not the monopoly of an unfortunate few Reason 2000 When errors occur poor motivation negligence inat tention repetition forgetfulness or moral weakness of an individual is often seen as the principle cause of the error This person approach Reason 2000 the ten dency to blame an individual for an error remains a widespread tradition in the medical field and else where Disciplinary measures poster campaigns add ing procedures op top of existing ones are some of the methods that are used to reduce unwanted human behavior However this approach does not succeed in effectively reducing the likelihood of errors A system approach Reason 2000 assumes that one should ex pect human errors even with the best people in the best organization By assuming that we cannot change human behavior but can change the conditions under which humans work errors could be prevented more effectively Reason 1990 defined errors as follows Error will be taken a generic term to encompass all those occasions in which a planned sequence
63. ex Interruption event SS ___ Multiplex Obstruction Zz Separate pumps Obstruction Figure 21 Mean values of first response times per event Error bars represent standard devi ations 7 2 2 FIRST RESPONSE TIMES In order to measure how quickly participants re sponded to the onset of an event looked at the first response times The average first response times and the corresponding standard deviations are depicted in Figure 21 The first response times were significantly faster in the Multiplex condition in the syringe empty NOR syringe empty change administration rate and the interruption event The results of statistical tests for these 4 events were as follows Syringe empty F 1 15 8 595 p lt 0 05 For the NOR syringe empty event F 1 15 4 904 p lt 0 05 For the change admin istration rate event F 1 15 7 642 p lt 0 05 And for the interruption event F 1 13 9 799 p lt 0 05 There was no significant difference in first response times for the obstruction event F 1 15 0 019 p 0 893 7 2 3 TOTAL EXECUTION TIMES PER EVENT In order to measure participants performance during the execution of tasks looked at the total execution times per event The total execution times per event are displayed in Figure 22 There was no significant dif ference in execution times between the conditions in any event The results of statistical tests for the 5 events were as follows For the s
64. f the incident embarrass ment or due the fact that some errors occur unno ticed In the ICU in the UMCG 32 of the reported incidents in 2012 were related to the administration of medication 6076 of these incidents were attributed to human error Among the most frequent errors in the IV medication process are programming the wrong administration rate and not administering the medica tion at all The actual error rates and types of errors that occur may be very different than the decentralized error re ports suggest as some errors can be identified easier than others For example A faulty pump setup can be identified visually by comparing the administration rate on the infusion pump s display with the pre scribed administration rate Errors with drug concen trations are almost impossible to identify visually and are more likely to occur unnoticed Types of IV medication errors 0 Wrong dose strength or frequency Omitted medicine Wrong drug Wrong quantity Wrong route Wrong medicine label Wrong formulation 10 15 20 25 30 Figure 4 The division of frequent medication errors in percentages according to The National Patient Safety Agency 2007 from a total of 14 228 IV medication incidents 4 3 4 ERRORS RELATED TO THE IV MEDICATION PROCESS There are several stages in de IV medication process and errors may occur in any of these stages will brief ly summarize the general IV medication process 1
65. gested that a system with good usability met the following criteria learnability effi ciency memorability low error rate and satisfaction discussed these criteria in chapter 4 will now sum marize the results of the user based usability analysis and link these results to Nielsen s criteria for usability As participants encountered each user interface only once memorability could not be tested in the experi ment Although there were differences in numbers of clicks during multiple events there were no differences in the total time that was needed for dealing with an event This result indicates that the additional steps due to the use of a checklist in the Multiplex user in terface did not influence the speed in which tasks were performed Based on the recorded number of clicks and time to deal with an event a difference in ef ficiency could not be found First responses were generally faster with the Multi plex user interface This may indicate that it is easier to discriminate between infusion pumps on this user in terface than in the conventional setup The fact that participants could start dealing with an event earlier with Multiplex user interface indicates a higher effi ciency Participants were experts in the separate pumps con dition but were novices when it came to using the Multiplex user interface Despite this fact participants 47 made fewer errors with the Multiplex user interface bo and were equally
66. he number of lumens that are required for administering incompatible drugs The core idea be hind Multiplex infusion is illustrated in Figure 8 By administering multiple incompatible drugs sequential ly through a single lumen and separating these drugs by a neutral buffer fluid the number of required lu mens could be reduced In other fields of research this technique is called multiplexing In order to achieve multiplexing behavior using multiple infusion pumps a controller human or computer needs to switch be tween multiple pumps by starting and stopping them sequentially As this would require many timed switching actions it is best to automate this process using a computerized control system Not all drugs are allowed to be interrupted during administration Noradrenaline for example would still be administered through a separate lumen as an inter ruption in administration would cause an immediate decrease in blood pressure The Multiplex infusion system will require a user inter face which provides a representation of the current set of infusion pumps together with the same control op tions as the current infusion pumps As the ICV an nounces plans for the construction of a new intensive care unit and the upcoming integration of a new pa tient data management system PDMS this offers the possibility to radically revise the current IV system T Y drugA NaCl drug B drug C Figure8 The
67. hecklist Start pum Start pum 6 clicks E mda Seperate Pumps Plan 1 1 then 2 then 3 Replace syringe lan 2 2 1 then 2 2 2 Replace P 4 1 Stop pump syringe 3 Start pump 2 1 Remove 2 2 Attach syringe new syringe Figure 14 Hierarchical task analyses of replacing an empty syringe using the Multiplex user interface top and the separate pump interface bottom Plan 1 1 then 2 then 4 OR 1 then 3 then 4 Multiplex Replace syringe Plan 2 3 1 then 3 2 then 3 3 3 Do not 1 Click pump 2 Use checklist use checklist 4 Close dialog 2 1 Click each step in checklist 6 clicks 3 1 Click disable checklist 3 2 Click Start pump 3 3 Click Start pump Seperate Pumps Plan 1 1 then 2 then 3 Replace syringe 2 Set Plan 2 2 1 then 2 2 administration rate then 2 3 on second pump 1 Attach syringe 3 Stop to second pump 36 first pump 2 1 Increase administration rate 8 clicks 2 2 Confirm administration rate 2 3 Start pump Figure 15 Hierarchical task analyses of replacing an empty syringe of noradrenalin using the Multiplex user interface top and the separate pump interface bottom Multiplex Plan 1 1 then 2 seperate pumps Plan 1 1 then 2 Change Change administration administration rate rate 1 Click decrease adm rate 5x 2 Confirm 2 Click yes
68. her D Olin M Badihi Y Biesky M Sprung C L Pizov R et al 1995 A look into the na ture and causes of human errors in the intensive care unit Quality amp safety in health care 23 2 294 300 Dumas J S 2003 User based evaluations In A Sears amp J Jacko Eds The Human Computer Interaction Hand book pp 1093 1117 Hillsdale NJ L Erlbaum Associ ates Inc Evans R S Sharp J H Linford L H Lloyd J F Woller 5 C Stevens S M Elliott C G et al 2012 Reduction of Peripherally Inserted Central Catheter Associated Deep Venous Thrombosis Chest Gagnon R Laberge J Lamsdale a Histon J Hudson C Davies J amp Caird J 2004 A User Centered Evalua tion of three Intravenous Infusion Pumps Proceedings of the Human Factors and Ergonomics Society Annual Meeting 48 15 1773 1777 Garmer K Liljegren E Osvalder a L amp Dahlman 5 2000 Usability Evaluation of a New User Interface for an Infusion Pump Developed with a Human Factors Ap proach Proceedings of the Human Factors and Ergonom ics Society Annual Meeting 44 1 128 131 Giraud T Dhainaut J F Vaxelaire J F Joseph T Jour nois D Bleichner G Sollet J P et al 1993 latrogenic complications in adult intensive care units a prospective two center study Critical Care Medicine 21 40 51 Graham M J Kubose T K Jordan D Zhang J John son T R amp
69. hree way valve and several connections Although but it gets more complex to describe the connections not depicted the connection between the syringe on of connectors as the number of infusion pumps in a syringe pump and an IV line is also part of the tubing creases An example of a patient s tubing is depicted in GIA 200 ml h Penpheral venous catheter Legend Volumetric pump administering a 5 glucose solution at 200 mih Syringe pump administering insulin at 0 4 mih An M line Athree way valve lt A connection between 2 connectors Figure 2 Example of the tubing of a patient receiving IV therapy The catheter on the right side of this figure provides an en try point for intravenous medication into the bloodstream of the patient 3 2 4 IV THERAPY RELATED TASKS In order to see how nurses dispense IV medication a hi erarchical task analysis HTA was performed of dis pensing multiple IV fluids The resulting HTA trees are depicted in Appendix C and the process will be discussed in the next section The IV therapy process consists of 5 main stages diag nosis prescription dispensing administration and monitoring In the current study dispensing admin istration and monitoring are the most relevant stages as these actions are all performed by the ICU nurse who is the end user of the proposed control system DISPENSING As many IV fluids are administered continuously a common task
70. ility analysis where the usability of the new user interface was compared to that of the current method of manually operating multiple infusion pumps The user based usability analysis was performed using a computer simulation of multiple infusion pumps and a simulation of the Multiplex user interface 6 1 INTRODUCTION Previous studies indicated that usability problems are common with infusion pumps and they increase the likelihood of errors in intravenous therapy Gagnon et al 2004 Garmer et al 2000 Graham et al 2004 Studies on the usability of infusion pumps all focused on the user interface of a single pumps although it is common for a patient in an ICU to receive treatment using multiple infusion pumps In a hectic work envi ronment such as the ICU the likelihood of errors is higher than in any other hospital department Kalisch amp Aebersold 2010 Moyen et al 2008 Other studies suggested that a high cognitive workload and a high number of interruptions may contribute to these er rors Adamczyk amp Bailey 2004 Kalisch amp Aebersold 2010 Laxmisan et al 2006 In this study a new user interface was designed for the Multiplex infusion system with the goal to have a bet ter usability than the current method of manually op erating multiple infusion pumps In order to compare the usability of both user interfaces a user based usa bility analysis was performed using a computerized simulation In an experiment
71. imultaneously over a single lu thus should be administered over separate lumens The white cells indicate that it is unknown whether or not the fluids are compatible out of precaution these infusion fluids are general ly but not always administered separately men A red cell means that the fluids are incompatible and AMIODARON AMIODARON BENZYLPENICILLINE CEFTAZIDIM CICLOSPORINE CLINDAMYCINE CLONIDINE EN DIAZEPAM DOBUTAMINE DOPAMINE ESOMEPRAZOL FENTANYL FLUCLOXACILLINE FLUMAZENIL FUROSEMIDE HEPARINE HYDROCORTISON INSULINE ISOPRENALINE KALIUMCHLORIDE KETANSERINE LABETALOL MGSO4 A MIDAZOLAM MILRINON MORFINE NAPO4 NICARDIPINE A NIMODIPINE NITROGLYCERINE NITROPRUSSIDE NOREPINEFRINE OCTREOTIDE PROPOFOL RANITIDINE ROCURONIUM TACROLIMUS VANCOMYCINE VASOPRESSINE VERAPAMIL Noavaomy BENZYLPENICILLINE CEFTAZIDIM WIOQIHZWIJ3O 5071212 CLINDAMY CINE ANISAW WONITS mann LL DIAZEPAM WWd37w10 3JNIW ITO I DOBUTAMINE DOPAMINE ANIW dod ES OMEPRAZOL FENTANYL TO ZWAAN O 3 FLUCLOXACILLINE KALIUMCHLORIDE FLUMAZENIL HYDRCORTIS ON FUROS EMIDE INS ULINE HEPARINE ISOPRENALINE MIDAZOLAM MILRINON MORFINE NICARDIPINE LABETALOL B i EHE EEE ________ m S HEN E TAN WINA ANIA WAH NOSUJODOADAH ANNA SMI NON TIJ 3JNITIIOVXOTOTH Jawiasouns 3ano1H2wWnn
72. ing Systems Vol 6 pp 271 278 Altmann E amp Trafton J 2002 Memory for goals An activation based model Cognitive science 26 1 39 83 B Braun Melsungen 2011 Medication Error Risk Pre vention in Infusion Therapy pp 1 23 Back J Cox A L amp Brumby D P 2012 Choosing to Interleave Human Error and Information Access Cost Proceedings of the 2012 ACM annual conference on Human Factors in Computing Systems 1651 1654 Borst J P Taatgen N a amp van Rijn H 2010 The prob lem state a cognitive bottleneck in multitasking Journal of experimental psychology Learning memory and cog nition 36 2 363 82 American Psychological Associa tion Burris J Brennan T Leape L amp Laird N 1991 Inci dence of adverse events and negligence in hospitalized patients N Engl J Med 324 145 152 Carayon P Wetterneck T B Hundt A S Ozkaynak M Ram P Desilvey J Hicks B et al 2005 Observing Nurse Interaction with Infusion Pump Technologies Ad vances in patient safety from research to implementa tion 4 349 364 Coiera E W Jayasuriya R a Hardy J Bannan A Thorpe M E C 2002 Communication loads on clinical staff in the emergency department The Medical journal of Australia 176 9 415 8 Davey C 2005 Target Controlled Infusion Systems Part Two Alaris Asena PK Device Evaluation Service August Donchin Y Gop
73. ion system was proposed to reduce and solve current problems with usability patient safety and comfort A feasibility study indicated that this sys tem could be able to reduce the number of required lumens and therefore could reduce infection risks and discomfort In this study have shown how the usability of infu sion pumps can affect the occurrence of errors related to intravenous therapy A user friendly user interface for the Multiplex infusion system was developed and its usability was compared to that of the current method of operating multiple infusion pumps in a us er based usability analysis Results suggest that the Multiplex infusion system has an overall better usabil ity and could reduce the number of errors by partially automating tasks and supporting the ICU nurse during the error prone stages of IV therapy In this thesis used theoretical models to explain how interruptions can cause the decay of goals corre sponding to tasks which can lead to errors Experi mental results indicated that errors occurred more during an event that featured such an interruption Re sults of this study indicate the implementation of the proposed Multiplex infusion system could benefit the safety of patients in an ICU 11 REFERENCES Adamczyk P D amp Bailey B P 2004 If Not Now When 72 The Effects of Interruption at Different Mo ments Within Task Execution Proceedings of the SIGCHI Conference on Human Factors in Comput
74. lines and connectors were arranged for multiple ICU patients The second goal was developing a user interface for the Mul tiplex infusion system and comparing its usability with the use of multiple separate infusion pumps The design of the new user interface was the result of an iterative process which in volved the usability principles which will be discussed in chap ter 4 and feedback from various nurses and physicians The usability of the system was compared with that of the current system by measuring the time and button presses during the execution of several prototypical tasks hypothesized that the Multiplex infusion system would have a lower error rate than the current infusion system also expected that the differ ences in the numbers of clicks between both systems could be predicted by the number of clicks by an expert user although I expected the actual number of clicks to be higher than this golden standard as a result of the variation between partici pants did not have a hypothesis on a difference in execution times A questionnaire was administered in order to measure a subjective preference 3 PRACTICAL BACKGROUND As one of the goals of this study was to develop a us er friendly control system for multiple infusion pumps in an intensive care unit ICU it is necessary to know more about the end users their tasks and work envi ronment In section 3 1 will provide a general de scription of the ICU and the task
75. lysis are displayed in Table 4 There was a significant difference in the number of lumens between the current situation M 3 17 SD 1 27 and with the Multiplex Infusion System M 1 67 SD 0 65 t 11 4 18 p 0 002 This re sult suggests that the number of required lumens can be reduced using the Multiplex infusion system 5 4 KEY ADVANTAGES OVER THE CURRENT IV SYSTEM Multiplex infusion system is expected to have multi ple advantages over the current IV system will pro vide an overview of the most important possible advantages Reduction of lumens As indicated in section 5 3 the Multiplex infusion system has the potential to decrease the number of required lumens by administering multiple incompatible drugs through a single lumen Reduction of catheter related infections As the number of required lumens can be reduced less catheter insertions will be needed thus reducing the likelihood of catheter related infections Reduction of patient s discomfort Catheter in sertions are often painful and result in physical discomfort for the patient By optimizing the number of required lumens and catheters less catheter insertions would be needed Reduction of errors By automating tasks which would normally involve manual actions from the ICU nurse errors and preventable adverse events could be prevented Cost reduction By reducing the likelihood of preventable adverse events financial implications
76. m for multiple infusion pumps is proposed to reduce the problems with conventional IV therapy The core idea behind this Multiplex infusion sys tem is reducing the number of required lumens by optimiz ing the number of IV drugs that are administered through a single lumen A feasibility analysis showed that the Multi plex infusion system could significantly reduce the number of required lumens A user interface for this system was designed with the goal of reducing the likelihood of errors by partially automating several tasks In order to compare the usability of the new user interface with that of the con ventional method of manually controlling multiple infusion pumps a user based usability analysis was performed Re sults indicated that the new user interface had an overall better usability and a significantly lower error rate 1 INTRODUCTION There are various types of errors that can occur while operat ing medical devices Among the most frequently occurring errors are device malfunction malfunction of disposable parts and device setup errors The causes of the first two types of error may result from poor design faulty production or maintenance Kohn Corrigan amp Donaldson 2000 The latter is often the result of human error which may be caused by inadequate training high workload and hard to operate devic es Kaye amp Crowley 2000 Among the most frequently used medical devices are infusion pumps These pumps are
77. mp Taatgen 2008 were used to explain how an interruption could cause the decay of a goal in working memory The fact that these errors occurred while performing simplified tasks in a controlled and simulated environment suggests that errors may be more frequent and severe in a real ICU 8 4 SUGGESTIONS FOR FURTHER RESEARCH Based on the results of the current study have the following suggestions for future research The results of the empirical study showed that nurses use different strategies for completing the same task It would be interesting to perform a hierarchical task analysis on the execution of several IV therapy related tasks as they are being performed by multiple differ ent ICU nurses This may help to identify how often certain strategies are used by ICU nurses Some strat egies may involve more manual actions than others which may be inefficient or could have a higher risk of error The results of this analysis could be used to train nurses to become more efficient and may reduce er rors The tasks during the interruption event could be modeled using the threaded cognition model Salvucci amp Taatgen 2008 In section 8 2 4 suggest ed that the errors caused by the interruption in this event could be explained by this model The model could also be used to predict errors during interrup tions Predictions by the model could be compared to data from an experiment which features a simulation involving multiple
78. naire provided in Dutch and a translation into English is provided All statements were rated for the Multiplex Infusion System below A rating was given using a 7 point Likert Scale Figure and the current setup of infusion pumps Statements were 24 25 and 26 display boxplots of the responses to the state ments in the questionnaire Number 10 11 12 13 14 15 16 17 18 19 20 21 22 Table 7 Question in Dutch Dit systeem is gemakkelijk te gebruiken Dit systeem is nodeloos complex Dit systeem ziet er mooi uit Het hoofdscherm is overzichtelijk Er wordt effici nt gebruik gemaakt van de ruimte op het scherm De informatie op het scherm maakt goed duidelijk wat het systeem aan het doen is Het is duidelijk welke functie elke knop heeft Het is makkelijk om de verschillende infuuspompen van elkaar te onderscheiden Dit systeem geeft duidelijke feedback op mijn handelin gen Fouten zijn eenvoudig te herstellen met dit systeem Functies die ik nodig heb kan ik makkelijk vinden Ik heb het gevoel dat ik controle heb over het systeem Ik kan effici nt werken met dit systeem Ik denk dat dit systeem veilig is Ik kan gemakkelijk een dosering aan te passen Ik kan gemakkelijk een spuit verwisselen Ik kan goed zien hoe vol een spuit nog is Ik kan makkelijk zien hoe lang het duurt tot een spuit leeg is Als er een alarm afgaat is het mij duidelijk wat er aan de hand is Ik werk graag met dit systeem
79. ne P H et al 2005 The Critical Care Safety Study The incidence and nature of adverse events and serious medical errors in intensive Critical Care Medicine 33 8 1694 1700 Salvucci D D amp Bogunovich P 2010 Multitasking and monotasking The effects of mental workload on deferred task interruptions Proceedings of the SIGCHI conference on multitasking pp 85 88 Salvucci D D amp Taatgen N a 2008 Threaded cogni tion an integrated theory of concurrent multitasking Psychological review 115 1 101 30 Salvucci D D amp Taatgen N A 2009 Toward a Unified Theory of the Multitasking Continuum L From Concur rent Performance to Task Switching Interruption and Resumption CHI 2009 Conference Proceedings pp 1819 1828 Shneidermann B 1998 Designing the user interface 3rd ed Reading MA Addison Wesley Van Soest Segers B Cheung K amp Hunfeld N 2009 Op IC vooral doseerfout bij infuus Pharmaceutisch Weekblad 04 16 18 Stanton N a 2006 Hierarchical task analysis develop ments applications and extensions Applied ergonomics 37 1 55 79 Tissot E Cornette C Demoly P Jacquet M Barale F amp Capellier G 1999 Medication errors at the admin istration stage in an intensive care unit Intensive care medicine 25 4 353 9 Tyler 2009 Alaris GH Syringe Pump User Manual pp 1 34 Rolle Switzerland Ve
80. ned by the amount of visual search that was required to find the pump corre sponding to the event followed by the time it took to read and interpret the relevant task before starting the execution of the task During the interruption event the first response times were roughly between 40 and 60 seconds This can be explained by the fact that the participant was required to read the task instructions and had to look up how long it took before a certain syringe was empty before the first click could take place Therefore the first response time is a valid measure to use in this experiment In the NOR syringe empty change administration rate and interruption event participants responded significantly faster to the onset of an event in the Mul tiplex condition than in the condition with separate pumps Only in the obstruction event there was no difference in first response times The difference in the first four events could indicate that it is easier to dis criminate between pumps using the Multiplex infu sion system allowing participants to start performing tasks relevant to the event sooner On the Multiplex interface the panel representing the pump entirely changes color to red or green which may be more sa lient than the alarm light in the top right corner of the simulated infusion pumps In the real ICU this could make an important difference in emergency situations when a quick and accurate response to an alarm is re quired
81. nents of the proposed system will also be discussed in this thesis A new graphical user interface GUI will be presented and its usability was tested and com pared with that of the user interface of the current infusion system 2 OBJECTIVES In the previous chapter described various problems and dis advantages related to the current method of IV therapy In this thesis the Multiplex infusion system is proposed It is a con trol system for multiple infusion pumps that could potentially reduce multiple problems related to IV therapy The overall goal of this study was to take the first steps in building a system which improves patient safety and has a bet ter usability than the current IV system The properties and limitations of this system also needed to be identified In order to achieve this two sub goals were set The first goal of this study was to investigate whether or not the Multiplex infusion system has advantages over the current IV system identifying these advantages and setting the de mands for the system on a medical technical level In order to quantify such an advantage analyzed how much the number of lumens per patient could be reduced in a feasibility study Decreasing the number of required lumens could improve patient safety and comfort The expectation was that using Multiplex Infusion System would reduce the number of nec essary lumens In order to provide a good estimation of this number analyzed how IV
82. ng the administra tion rate without stopping the infusion Volume to be infused The user specifies the vol VTBI ume and time span in which an IV fluid needs to be administered The pump stops when the pro grammed volume is admi nistered Table 1 Common functionalities of infusion pumps and their description 3 3 2 ALARIS ASENA GH SYRINGE PUMP At the ICV the Alaris Asena GH Syringe Pump is the standard infusion pump Figure 3 displays the layout of this pump An average patient at this ICU receives IV therapy using seven of these pumps simultaneously often combined with one or two volumetric pumps The syringe pumps are placed in a stacked position us ing a docking station which also provides power to the pumps In case of transportation or a power fail ure the pump contains a battery that can last about 5 5 hours ON OFF Release lever High visibility Display for MDI Alarm Indicator PURGE BOLUS MUTE PRESSURE IV Infusion HOLD Shelf for Syringe Clamp Positive Plunger set hook chevron keys Grippers 1 1 1 and softkeys Figure 3 The layout of an Alaris Asena GW infusion pump Image source Tyler 2009 4 THEORETICAL BACKGROUND Because low usability of infusion pumps is the major problem with the current IV pumps this chapter will discuss different aspects related to usability Usability is a common term in the field of human computer in teraction HCI HCl involves the study and design
83. ng the experiment all clicks response times task execu tion times and errors were recorded The first response time was defined as the time between the start of an event and the first action from the participant to start performing a task in order to deal with the event For each event there were four variables the number of clicks during the event the first re sponse time the total time of dealing with an event and the number of errors A repeated measures analysis of variance was performed in order to compare the differences between the variables corre sponding to the syringe empty NOR syringe empty and the change administration rate events An analysis of variance of the same variables but for the interruption and obstruction event was performed for the same goal Questionnaire responses were logged as numbers ranging from 1 to 7 These values represented the levels of the Likert scale as follows 1 completely disagree 2 strongly disagree 3 disagree 4 neutral 5 agree 6 strongly agree and 7 completely agree The difference in responses was analyzed using a two tailed t test Plan 1 1 then 2 then 4 OR 1 then 3 then 4 Multiplex Replace syringe Plan 2 3 1 then 3 2 then 3 3 3 Do not 1 Click pump 2 Use checklist ne RETTEN 4 Close dialog 2 1 Click each step 3 1 Click disable 3 2 Click 3 3 Click in checklist c
84. of the interaction between humans and computers A well designed user interface can provide an enjoyable and efficient interaction will start this chapter with discussing the types of user interfaces that are com monly used on infusion pumps will then introduce the term usability and how it can be analyzed The types of human error that may result from poor usa bility will be discussed Factors that may increase the likelihood of errors such as multitasking and task in terruptions will also be addressed in this chapter 4 1 USER INTERFACE A user interface UI is the space where there is inter action between a system and its user Any component that is required by the user to provide input to the sys tem and components which are used by the system to provide output to the user are part of the user inter face The user interface consists of all hardware and software components that are involved in the interac tion between a human and a machine Although there are many different types of user interfaces will only highlight two types of user interfaces that are relevant for this thesis 4 1 1 GRAPHICAL USER INTERFACE A graphical user interface GUI is one of the most common types of user interfaces This type of user in terface accepts input via devices such as a keyboard or a mouse and provides graphical output to a screen or monitor In the case of a fully integrated system such as infusion pump physical buttons ar
85. of men tal of physical activities fails to achieve its intended outcome and when these failures cannot be attribut ed to the intervention of some change agency Rea son 1990 For clarification purposes Reason s definition of hu man error will be discomposed into two parts The first part speaks of a planned activity with some in tended outcome For example Sending an e mail to your neighbor while you intended to send it to your mother is considered to be an error The second part of the definition states that we do not speak of an er ror when some change in the environment is to blame for the adverse event For example a plane crash that is caused by a sudden wind shear is not an error Errors that occur due to negligence poor maintenance or design flaws are called latent errors Reason 1990 Latent errors do not instantly lead to an adverse event but they do increase the risk of an adverse event hap pening later on For example if a plane crashes be cause the maintenance crew has installed the wrong parts we speak of a latent error 4 3 1 TYPES OF HUMAN ERRORS Reason 1990 distinguishes three types or human er rors slips lapses and mistakes Slips and lapses occur when the action that is performed is not the action that was intended to be performed The difference be tween a slip and a lapse is that the occurrence of a slip is observable and that of a lapse is not For example pressing the wrong button on s
86. olich amp Nielsen 1990 and Nielsen 1994b were applied during the development of the user interface together with consultation with multi ple ICU nurses and physicians Note that the user in terface has not yet been implemented into a physical system will refer to several of the the following usa bility heuristics from Nielsen 1994a as discuss the design considerations of the user interface in the next section Visibility of system status 1 Match between system and the real world 2 User control and freedom 3 Consistency and standards 4 Error prevention 5 Recognition rather than recall 6 Flexibility and efficiency of use 7 Aesthetic and minimalistic design 8 Help users recognize diagnose and recover from errors 9 Help and documentation 10 5 5 1 DESIGN CONSIDERATIONS The user interface needed to display all infusion pumps that were controlled by the system For each pump all relevant information regarding its current status needed to be visible immediately which corre sponds the first usability heuristic from Nielsen 1994a ICU nurses and physicians indicated that in formation on the administration rate the name and concentration of the currently administered drug the remaining time before the syringe or infusion bag was empty and whether or not the pump was actually ad ministering IV fluid were important to be displayed visually Infusion pumps needed to
87. ome device is a slip The action and the result of this action is observable Not being able to recall something from your memory is a lapse When a mistake occurs an action proceeds as planned but the action itself is the wrong action to achieve the desired outcome Mistakes can occur when a situation is not assessed correctly possibly due to the lack of expertise This often happens in un familiar circumstances Slips and lapses tend to occur during the performance of routine actions Fatigue stress and performing multiple activities are known to increase the likelihood of slips and lapses and mistakes Moyen Camir amp Stelfox 2008 The definitions of different kinds of errors are displayed in Table 2 4 3 2 ADVERSE EVENTS Although the consequences of many errors are not severe and errors often happen unnoticed errors may lead to other errors with more severe consequences and should be prevented if possible The term adverse event is used to describe an injury that is caused by a medical management There is a distinction between adverse events and preventable adverse events In To Err is Human Building a Safer Health System 1999 this distinction is defined as follows An adverse event is an injury caused by medical man agement rather than the underlying condition of the patient An adverse event attributable to error is a preventable adverse event Burris Brennan Leape amp Laird 1991 Kohn et al 2000
88. omen Pump profile alternatively administration profile Settings for administering infusion fluids in a specific manner E g con tinues infusion gradual increase or decrease in flow rate Single lumen catheter A catheter containing one lumen Syringe NL spuit A syringe consists of a plunger that fits into a cylindrical tube The plunger can be pushed or pulled inside the tube which allows for the syringe to expel or take in fluids through the open end of the tube Often the open end is fitted with a male luer lock tip Syringe pump NL spuiten pomp A mechanical device used for the administration of a predetermined volume of in fusion fluid to a patient by gradually pushing the plunger of a syringe Triple lumen catheter A catheter containing three lumens therefore allowing for three separated flows of infusion fluids Tubing The configuration of connectors that connect from any infusion pump to any catheter in a patient receiving IV treatment Usability The user friendliness or ease of use of a device User interface NL gebruikersinterface The system by which a person the user interacts with a machine It allows the user to send input to the machine and receive feedback from it Volumetric pump volumetrische pomp infusion pump designed to deliver moderate to large flows i e 5 to 999 ml hour APPENDIX B USABILITY QUESTIONNAIRE In Table 7 you will find the statements from the question
89. oming appoint ments and other planned tasks such as changing ad ministration rates Multitasking increases the cognitive workload of the clinician and nurses which may result in a higher number of errors Back et al 2012 Borst et al 2010 Coiera Jayasuriya Hardy Bannan amp Thorpe 2002 In an ICU setting interruptions can lead to errors For example when a nurse is interrupted by another nurse a pager or phone call the nurse may forget the task he or she was initially working on This event may be explained by the models of goal activation and threaded cognition Altmann amp Trafton 2002 Salvucci amp Taatgen 2008 A nurse often has to re member multiple tasks at the same time for example an upcoming appointment monitoring a patient or personal errands In a hectic work environment it is possible that the maximal amount of the chunks that represent these tasks is reached As an interruption occurs the activation of a chunk corresponding to one of these tasks may decrease so much that it cannot be retrieved anymore According to the goal activation model it is also possible that other earlier goals dis tractors are retrieved instead of the target chunk Altmann amp Trafton 2002 Interruptions have proven to be more disruptive as the mental workload re quired processing capacity of the brain increases Salvucci amp Bogunovich 2010 Although multitasking and interruptions are common research topics in the
90. on of ICU patients The empirical study featured a simplification of the tasks of an ICU nurse Some aspects of this study may not be generalizable to a real environment as the tasks in the experiment do not feature all aspects of their real world counterparts For example attaching a new syringe in the experiment was a matter of a single click while the real world task involves several sub tasks The real world task involves more complexity while the experimental task featured a simplification of the ICU nurse s tasks related to IV therapy On the other hand the fact that errors already occurred using a simplified version of real world tasks could mean that errors are even more likely in the real world task The audible alarms of a real infusion pump sound from a speaker inside the pump which could make localiz ing the alarming pump easier However this factor was not included in the experiment as all alarms sounded from the same source This may have affect ed the first response times A factor that could have influenced the performance in this experiment and may have influenced the results of the questionnaire is that most participants have years of experience with the current infusion system Even an ICU nurse with only one month of experience would have been likely to have operated an infusion pump over a hundred times Despite the influence of previous experience participants made less errors us ing a completely new system
91. onnected The new syringe is con nected the valve is opened and the pump is started again Some infusions should not be interrupted For exam ple an interruption in the administration of nora drenalin may cause the blood pressure the patient to decrease Therefore when replacing an almost empty syringe of noradrenalin a second pump is often used The ICU nurse decreases the administration rate of the almost empty syringe stepwise while stepwise in creasing the administration rate of the new syringe During this process the nurse needs to monitor the patient s blood pressure and adjust the administration rate accordingly When a new IV fluid is prescribed a compatibility ma trix needs to be checked in order to determine how to arrange the tubing The compatibility matrix that is used in the ICV can be found in Appendix E If the new IV fluid is incompatible with the current IV fluids it needs to be administered through a separate lumen which may require placing an additional catheter If the new IV fluid is compatible with one of the current IV fluids they can be administered through the same lumen After connecting the new syringe with IV fluid the ICU nurse programs the prescribed the admin istration rate on the infusion pump and starts the infu sion MONITORING Monitoring is required in order to review the patients recovery rate and response to the administered IV flu ids In this stage administration rates and frequ
92. oximal VF Right Triple lumen Catheter CVC VJ Right 5 0 ml h Proximal 18g Medial 18g CLINDA 50 0 ml h PROP 8 0 ml h PIV Hand Right NOR 4 0 ml h FOSF 10 0 ml h INS 0 6 ml h C a ED ED G5 PIV Hand Right 42 ml h riple lumen Catheter VC VJ Right TACRO CVD set 0 5 ml h Medial 180 NaCL 8 0 ml h INS 4 7 ml h PIV Hand Left 77 PROP 2 0 ml h NaCL 1 0 ml h NOR 8 0 Wh GI 84 0 ml h NOR 4 0 ml h PROP 12 0 ml h INS 1 0 ml h MAG 0 8 ml h VANCO 2 0 ml h PIV Arm Right Triple lumen Catheter CVC VSC Right Distal 16g Medial 18g Proximal 18g CVD set 25 84 0 ml h Triple lumen Catheter CVC VF Left Distal 16g NOR 4 0 ml h CVD set Medial 18g Proximal 18g KAL 4 0 ml h DEX MED 2 0 ml h NaCL 5 0 ml h MaCL 2 0 ml h HYDRO 2 0 ml h KCL 4 0 ml h MIDA 15 ml h PROP 16 ml h Triple lumen Catheter CVC VJ Left Medial 18g CVD set Distal 16g Proximal 18q PIV Arm Left G5 NaCl 21 0 ml h 612 5 NaCL 0 45 5 0 ml h MAG 0 8 ml h PIV Arm Right riple lumen Catheter VC VJ Left Medial 18g 81 Distal 16g Proximal 18g APPENDIX E COMPATIBILITY MATRIX The compatibility matrix as it is used in the ICU of this study This matrix is updated every 6 months A green cell means that the two intersecting infusion fluids are compatible thus allowed to be administered s
93. plex NOR syringe empty Golden standard Separate pumps NOR syringe empty Multiplex Change administration rate Separate pumps Change administration rate Multiplex Interruption event Separate pumps Interruption event Multiplex Obstruction Separate pumps Obstruction Figure 20 The actual number of clicks and golden standards per event Error bars denote the standard deviations of the ac tual number of clicks 7 2 1 NUMBER OF CLICKS In order to measure participants performance during the execution of tasks looked at the number of clicks A golden standard which is the least number of clicks needed to perform the task correctly was de termined for each event using the HTAs in the previ ous chapter For the syringe empty and NOR syringe empty events in the Multiplex condition there were 2 golden standards one using the checklist and one when the checklist was disabled by the user The gold en standards together with the averages and standard deviations are displayed in Figure 20 A repeated measures analysis of variance was used to test for dif ferences between both user interfaces in the syringe empty NOR syringe empty and the change admin istration rate events A between subjects analysis of variance was performed for the interruption and ob struction events In the syringe empty event the number of clicks was higher in the Multiplex condition than in the separate pumps condition
94. r vice versa An order to change the administration rate was carried out one hour late or not at all A drug is administered that was not ordered An inappropriate drug or administration rate was prescribed The patient s condition is incorrectly diagnosed Failing to detect and respond to and adverse drug event Table3 Types of intravenous medication errors the stage where they may occur and their description 4 3 5 PREVENTING MEDICATION ERRORS According to a report by the Dutch Healthcare Inspec tion Inspectie voor de Gezondheidszorg the likeli hood of intravenous medication errors increases due to the use of multiple different infusion pumps within the same hospital Loekemeijer et al 1997 Hospital wide standardization of the types of infusion pumps to use could prevent these errors However as multiple hospital departments may have their own sets of de mands regarding to functionalities on the pump it The previous section illustrated that errors with intra venous medication are common in ICUs worldwide In order to prevent medication errors multiple studies have identified possible improvements to IV therapy and infusion pumps As many errors occur during the setup of infusion pumps Gagnon et al 2004 evalu ated the usability of multiple infusion pumps and sug gested improvements for the user interface may be inevitable for a hospital to own and use multi ple types of pumps Standardizing infusion pumps per dep
95. result the number of infusion lines can be reduced and with it the spaghetti problem Such a system also allows for the partial automa tion of several tasks reducing the number of human actions and therefore the number of errors Tasks that can be auto mated include setting up the infusion rate or flow rate starting or stopping pumps and gradually increasing or decreasing the infusion rate The plan to build such a new infusion system was commis sioned by the Intensive Care for Adults ICV of the University Medical Center Groningen UMCG Staff members of the ICV both doctors and nurses have reported problems with the current way in which intravenous treatment is adminis tered These problems lie in the usability of the current sys tem the complexity of the work environment and in the physical discomfort that patients experience as a result of the number of different catheters and IV lines Plans for the con struction of a new intensive care unit within three to five years and the integration of a new patient data management system PDMS offer the opportunity to radically revise the current IV system The goals of this study were to determine the demands for the Multiplex infusion system and to assess which practical and technical challenges lie ahead before the system can be de ployed safely This thesis will focus on the medical technical demands as well as on the usability of the system The physics mechanics and all compo
96. revent devi ation errors in 97 3 of all cases where there was a de viation error It is likely that most user inflicted deviations from the prescribed administration rate stay within the system s boundaries while other devi ation errors could not have been detected by the sys tem because they were not related to the programming of the infusion pump but may have been caused by preparing a solution of IV fluid in the wrong concentration There have been a few studies which compared the usability of different types of infusion pumps Gagnon et al 2004 performed a usability study on two types of infusion pumps They found that there was a lack of feedback on the user s input and that menu structures were hard to navigate through In a questionnaire among fifteen users of the Alaris Asena PK infusion pump users indicated that the lack of feedback and hard to press buttons increased the likelihood of un der and overdosing Davey 2005 Heuristic analysis Molich amp Nielsen 1990 on the usability of the user interface of an infusion pump in an intensive care unit identified 231 violations of the usability heuristics Graham et al 2004 Inconsistency in the design and the use of hard to understand language were the most common violations Only a few publications actually propose a redesigned user interface for an existing pump Garmer Liljegren Osvalder amp Dahlman 2000 Liljegren Osvalder amp Dahlman 2000 Current r
97. rkerke G J Ten Cate Hoedemaker H O Drost R Pot J W A Lansbergen M D I Bouwman C A H M Hermens J et al 2011 Medische Technologie at risk Onderzoeksrapportage in opdracht ven het Minis terie voor Volksgezondheid Welzijn en Sport April 1 100 Wetterneck T B Skibinski a Roberts T L Kleppin S M Schroeder M E Enloe M Rough 5 S et al 2006 Using failure mode and effects analysis to plan imple mentation of smart i v pump technology American jour nal of health system pharmacy 63 16 1528 38 Wickens C D 2008 Multiple Resources and Mental Workload Human Factors The Journal of the Human Factors and Ergonomics Society 50 3 449 455 Zhang J Johnson T R Patel V L Paige D L amp Kubose T 2003 Using usability heuristics to evaluate patient safety of medical devices Journal of Biomedical Informat ics 36 1 2 23 30 APPENDIX A DEFINITIONS Algorithm A set of computational rules for solving a prob lem in a finite number of steps CIV catheter IV catheter that is placed in a central vein and thus allows the administration of concentrated or otherwise potentially damaging fluids if given via a peripheral vein Connector Any Luer lock connectable object that is used to connect the syringe of an infusion pump to an IV catheter This includes infusion lines stopcocks syringes and catheters Double lumen cathet
98. s The template used abbreviations and a legend can be found under Appendix D Drawing the schemes on the template was a stepwise procedure The two nurses were instructed to start with denoting the used catheter types in which vein and on which side of the body they were placed The next step was to draw all infusion pumps in the order as they appeared from top to bottom The final step was to draw all IV lines and connectors which were part of the IV tubing Bed number Current num Number of lu ber of used mens with Mul lumens tiplex System 1 4 3 2 2 2 3 3 2 4 4 2 5 4 2 6 4 2 7 3 2 8 4 1 9 5 1 10 1 1 11 1 1 12 3 1 Average 3 17 1 67 Table 4 Number of lumens currently used versus the number of lumens used with Multi plex infusion system All drawn schemes were analyzed in order to deter mine the number of currently used lumens for each patient Digitalized versions of these schemes are at tached in Appendix D Based on the theoretical de scription of the Multiplex infusion system the number of required lumens using the Multiplex infu sion system was determined In practical terms this meant that all IV fluids that were currently adminis tered through a separate lumen could be adminis tered through the same lumen with the Multiplex infusion system Exceptions were IV fluids that were not allowed to be interrupted such as noradrenalin adrenalin dopamine and dobutamin The results of this ana
99. s The models of goal activation and threaded cognition could explain how the interrupting task causes these errors during the interruption event 8 2 5 QUESTIONNAIRE The grouped questionnaire results showed that the Multiplex infusion system had significantly higher rat ings for system appearance and intuitive design Re sults on other categories showed that there was no overall subjective preference nor a difference in ease of use or general impression Analysis of the individual statements indicated that participants liked the appearance of the Multiplex user interface more than that of the separate pumps How ever they enjoyed working with the separate pumps more When asked after the experiment several par ticipants indicated that due to the many years of working with the current infusion pumps they knew the system very well and therefore enjoyed working with the system Their answers on the questionnaire were influenced by their previous experience with the system Multiple participants mentioned that they re quired more experience with the Multiplex Infusion System ina real work environment in order to rate its usability properly Based on the analysis of the individual statements it is not possible to conclude which system had the high est subjective preference However the grouped questionnaire results indicate a higher preference for the Multiplex user interface 8 2 6 INTERPRETATION OF RESULTS Nielsen 1994a sug
100. s also possible that the new administration rate is not confirmed by the used or that the pump is not started after confirming Monitoring Monitoring is required in order to review the patient s recovery rate and response to the administered drugs In this stage admin istration rates and frequencies may be re evaluated The occurrence of adverse drug events may also be noticed and appropriate ac tions can be taken in case of an ADE Medication error Stage of occurrence Description Administration rate diffe rence Incorrect concentration of IV fluid Incorrect IV fluid Combining incompatible IV fluids IV line routing error Delay in administration rate change Unauthorized drug ad ministration Prescription error Diagnosis error ADE response error Administration Preparation Preparation or infusion pump setup Administration Administration Administration Administration Prescription Diagnosis Monitoring The administration rate that was programmed on the infusion pump differs from the administration rate that was prescribed The concentration of the infusion fluid differs from the pre scribed concentration The administered IV fluid is different from the one that was prescribed Two or more incompatible infusion fluids are administered through the same lumen An infusion fluid is administered through a peripheral line while administration through a central line is preferred o
101. s of the ICU nurse Section 3 2 describes the concept of intravenous ther apy and will discuss the currently used infusion pumps in section 3 3 The definitions of the terms which are introduced in this chapter can be found in Appendix A 3 1 THE INTENSIVE CARE UNIT The ICU is a hospital department where care is given to patients with severe and life threatening condi tions These are often vulnerable patients that require continues monitoring by specially trained doctors and ICU nurses The ICU is sometimes referred to as the critical care unit CCU or the intensive treatment unit ITU 3 1 1 THE ICU AT THE UMCG Within the University Medical Center Groningen UMCG there are several types of intensive care units each with their own specialization The current study was conducted at the intensive care for adults Dutch ICV The ICV is the largest ICU in the Nether lands with 330 employees and a total of 53 beds which are divided over four separate units These are the Thorax Intensive Care THIC Surgical Intensive Care CHIC Neurosurgical Intensive Care NCIC and the Respiratory Intensive Care Unit ICB The ICU also has a mobile intensive care unit MICU which is used for the transportation of ICU patients The UMCG also houses a pediatric ICU PICU and a neonatal ICU NICU The current study was mainly conducted on the THIC and the CHIC of the UMCG Although there are dif ferences between the two units in the typ
102. s placed into a vein which al lows an IV line to be connected There are two main types of catheters A peripheral catheter is placed in a peripheral vein is a single lumen catheter which means that it allows for a single stream of IV fluids A central venous catheter CVC is placed in a central vein which allows for the administration of IV fluids which are potentially damaging if they were adminis tered peripherally In the ICU in this study almost eve ry patient has a triple lumen CVC which means that three separated streams of IV fluid can enter the bloodstream simultaneously through the three sepa rate passages in the tip of the catheter This is very useful as there are several types of medication which are not compatible with each other mixture could cause a precipitation reaction in the IV line or it could neutralize the effects of the IV fluids Figure 1 shows the cross sections of three types of catheters Note that the diameter of the catheter generally increases with the number of lumens As an average patient at this ICU receives seven different types of IV medica tion there are often more incompatible infusion fluids than there are available lumens Often additional cath eters are required which causes physical discomfort for the patient adds to the number of IV lines that are required and increases the risks of catheter related in fections Evans et al 2012 Hilton et al 1988 Mermel et al 2001 Singl
103. sed in the ICU Therefore the poten tial components of an infusion tree are Asyringe or volumetric infusion pump A filled syringe filled infusion bag IV line A3 way valve or more way valve Alumen of a single multi lumen infusion line Characteristics of the components that are relevant for modeling are Parameter dimension internal volume ml internal diameter mm length cm resistance Paeml 1eh elasticity mlePa 1 Secondary characteristics of components are weight g cost outer dimensions cm3 material s opacity and valve positions infusion tree A infusion tree B Multi lumen catheter Figure 12 An example of two infusion trees 5 6 2 STATIC DESCRIPTION OF AN INFUSION TREE The static situation is described by the topological lay out of these components as well as their individual characteristics From each entry point in the tree di rectly after a syringe pump or a volumetric pump a minimal pathway to the exit point where the lumen of the intravenous catheter touches the bloodstream can be defined Thus when 8 syringes are connected through a network to a single lumen 8 such pathways are present For each pathway a pathway length in cm volume in ml resistance Paeml 1 eh and elas ticity mle Pa 1 can be defined 5 6 3 DYNAMIC ASPECTS OF AN INFUSION TREE Dynamically speaking the Multiplex infusion system may be ina steady state or a non st
104. ser interface after which the nurse only has to confirm the suggested change The interruption event was designed to test whether or not the participant is able to complete a set of sub sequent tasks while being interrupted by a phone call The event starts with an order to change an admin istration rate Shortly after the incoming order a phone rings which has to be answered The experi menter then asks the participant to find out how long it takes until a certain syringe is empty During this question an obstruction alarm goes off After answer ing the question the participant can continue com pleting the tasks The goal activation model that discussed in chapter 4 would predict that an interrup tion would increase the likelihood that a part of the task would be neglected due to the decay of the cor responding goal chunk in memory Altmann amp Trafton 2002 expected that as orders in the Multi plex user interface remain visible on screen this would remind the ICU nurse to complete the task af ter the interruption Using the separate IV pumps the nurse will not be reminded that a task is not complet ed therefore expected more errors will occur using the separate pumps than with the Multiplex user in terface in this task Event name Event descripti Task on Syringe emp Alarm indicates Replace empty ty that syringe is syringe almost empty NOR syringe Alarm indicates Replace empty empty that noradr
105. stration rate is achieved Oth ers simply start the second pump at the desired ad ministration rate and stop the first pump simultane ously There are also nurses who rapidly replace the almost empty syringe without using a second pump This is sometimes done when the administration rate is low enough to be interrupted without a noticeable change in blood pressure The NOR syringe empty event is included in the experiment as the required task can be complex and may include many steps de pending on the strategy that the participant chooses When noradrenalin is administered the Multiplex in fusion system will guide its user into using a second pump after which the system decreases the admin istration rate of the first pump stepwise while increas ing the administration rate of the new pump As this task execution is very different from the conventional operation of multiple pumps it is important to include this task in the experiment The change in administration rate event is also a task that illustrates the difference between the Multi plex user interface and that of manually operating multiple infusion pumps In the real world version of this task a physician orders a change in administration rate during the daily multidisciplinary consultation af ter which the ICU nurse changes the administration rate on the pump Using the Multiplex infusion sys tem an order to change the administration rate will automatically show up on the u
106. the stop icon was chosen to indicate stopping an infusion and the fast forward icon was chosen to indicate the bolus functionality As a user either needed the start or the stop functionality but never both at the same time a single button was used that could represent ei ther functionality stop if the pump is running start if the pump is not running The system needed to support the ICU nurse during error prone tasks which require completing multiple subtasks or manual actions which cannot be complet ed from a user interface The replacement of an empty syringe for example takes multiple steps which in volve physically connecting IV lines and opening valves By featuring checklists during such tasks a nurse will be able to verify each step of a real world task on the user interface which could prevent errors 5 In order to prevent errors related to programming a wrong administration rate the future Multiplex infu sion system will be connected to a patient data man agement system from which changes in administration rates can automatically be imported in to the Multiplex system An order to change an ad ministration rate needs to be confirmed by a nurse on the user interface An incoming order should be pre sented both visually and audible in order to be per ceived by the nurse The description of the order should be formulated clearly in order to be under stood correctly 5 5 2 DESCRIPTION OF THE USER INTERFACE Based
107. tion Psychological review 101 2 343 352 Molich R amp Nielsen J 1990 Heuristic Evaluation of User Interfaces CHI 90 Proceedings of the SIGCHI con ference on Human factors in computing systems Em powering people April 249 256 Moyen E Camir E amp Stelfox H T 2008 Clinical re view medication errors in critical care Critical care Lon don England 12 2 208 National Patient Safety Agency 2007 Patient Safety Alert Promoting safer use of injectable medicines 20 Nielsen J 1994a Usability Engineering pp 1 362 Elsevier Nielsen J 1994b Usability Inspection Methods CHI 94 Conference Companion on Human Factors in Com puting Systems pp 413 414 Raymer D M amp Smith D E 2007 Spontaneous knot ting of an agitated string Proceedings of the National Academy of Sciences of the United States of America 104 42 16432 7 Reason J 1990 Human Error Cambridge United King dom Cambridge University Press Reason J 2000 Human error models and manage ment The Western journal of medicine 172 6 393 6 Rothschild J M Keohane C a Cook E F Orav E J Burdick E Thompson S Hayes J et al 2005 A con trolled trial of smart infusion pumps to improve medica tion safety in critically ill patients Critical Care Medicine 33 3 533 540 Rothschild J M Landrigan C P Cronin J W Kaushal R Lockley S W Burdick E Sto
108. to the obstruction event using the Multiplex user interface left and the separate pumps interface right Training User interface 1 Experiment Training User interface 1 User interface 2 4 events random order Syringe empty NOR syringe empty Change adm rate Interruption event or obstruction Expenment User interface 2 Questionnaire User interface 1 Questionnaire User interface 2 4 events random order Syringe empty NOR syringe empty Change adm rate Interruption event or obstruction gt Time Figure 19 The structure of the experiment over time If the interruption event occurred using one user interface the obstruc tion event would occur using the other user interface 7 RESULTS One of the main goals of this study was to create user 7 2 EMPIRICAL STUDY interface with a better usability than the current IV system Therefore a user based usability analysis was performed In the following sections will discuss the results of this analysis In the analysis of the interruption event the data of 2 participants was could not be analyzed as due to technical reasons the obstruction failed to start In the questionnaire data from 4 participants was could not be analyzed due to a technical issue Number of clicks per event 10 20 30 Multiplex Syringe empty Bl Actual number Separate pumps Syringe empty Golden standard using checklist Multi
109. uit eae 16 2 minuten Infunderen Spuit leeg over meer dan 1 dag Infuussnelheid Bolus Infunderen Spuit leeg over 14 uur 29 minuten Infunderen Spuit mS minuten Infunderen GU H Spuit leeg over meer dan 1 dag s IV Therapie monitor Therapie Instellingen Configuratie van Infuuslijnen noradrenaline 0 2 mg ml mm kaliumchloride I Nac 0 9 insuline 1 IE ml 1 propofol 20 mg ml morfine 2 mg ml stappen om door te gaan met infusie Klik om pomp te stoppen Infuuslijn gekoppeld aan nieuwe spuit Draai kraan open Klik om pomp te starten insuline 1 IE ml Infunderen 3 4 ml h CH Spuit leeg over 10 uur en 11 minuten propofol 20 mg ml 10 ml h Bevestigen verhogen dosering Order afkomstig uit het MDO Er is toestemming van de arts voor deze handeling 1 ml h Volume 45 5 ml over Spuit leeg over meer dan 1 dag Pompprofiel Continue dosering Bolus snelheid 300 ml h Wilt u de morfine dosering verhogen van 1 0 ml h naar 2 0 ml h 2 Infuussnelheid Bolus Stop pomp
110. uous medication infusion errors in pediatric patients Pediatrics 116 1 e21 5 Laxmisan A Hakimzada F Sayan O R Green R a Zhang J amp Patel V L 2006 The multitasking clinician decision making and cognitive demand during and after team handoffs in emergency care International journal of medical informatics 76 11 12 801 11 Liljegren E Osvalder a L amp Dahlman S 2000 Setting the Requirements for a User Friendly Infusion Pump Pro ceedings of the Human Factors and Ergonomics Society Annual Meeting 44 1 132 135 Loekemeijer P A Doornbos S K Hansen J M M Ko ning de J Meinders L W Moleveld J Mulder H R et al 1997 Infuusapparatuur in Nederlandse ziekenhuizen pp 1 59 Rijswijk The Netherlands Melles M Freudenthal A de Ridder H amp Snijders C J 2004 Designing for enhanced interpretation anticipa tion and reflection in the intensive care unit 2004 IEEE In ternational Conference on Systems Man and Cybernetics 1 809 814 leee Mermel L a Farr B M Sherertz R J Raad I I O Grady N Harris S amp Craven D E 2001 Guidelines for the management of intravascular catheter related infections Journal of intravenous nursing the official publication of the Intravenous Nurses Society 24 3 180 205 Miller G 1956 The magical number seven plus or mi nus two some limits on our capacity for processing in forma
111. used to deliver fluids into a patient s bloodstream in a controlled man ner with a predetermined volume or rate of administration The errors that occur when operating an infusion pump may lead to a wrong dose of the fluid that is to be infused Verkerke et al 2011 In an intensive care unit ICU patients typically receive intra venous IV therapy using multiple infusion pumps simultane ously which are all controlled and monitored by a single ICU nurse These nurses also need to continuously monitor the patients and monitor other equipment such as heart rate mon itors dialysis machines and feeding pumps Obviously the job of a nurse involves multitasking which further increases the likelihood of human error Back Cox amp Brumby 2012 Borst Taatgen amp van Rijn 2010 In such a hectic work environment with already vulnerable patients this poses a serious safety threat for patients Because many of these patients are heavily dependent on their medication for maintaining blood pres sure for example interruptions or dosage errors can have se vere consequences There are several problems with the current way in which in travenous therapy is administered According to recent medi cation error reports which have been gathered from multiple Dutch hospitals 53 of all medication errors in the ICU are caused by errors in drug administration Van Soest Segers Cheung amp Hunfeld 2009 40 of these administration errors w
112. velop a new user interface which takes over the control of the current infusion system a user based usability evaluation is appropriate to evaluate the differences in usability of both systems 4 6 HEURISTIC EVALUATION A heuristic evaluation is an informal way to analyze the usability of user interfaces using a set of guidelines heuristics Zhang Johnson Patel Paige amp Kubose 2003 added the eight golden rules from Shneidermann 1998 to the heuristics from Molich and Nielsen 1990 in order to get a total of 14 usabil ity heuristics for evaluating medical devices A few ex amples of these heuristics are providing good error messages and giving informative feedback on the in put of the user In heuristic evaluation an evaluator is asked to point out usability problems in an interface as accurate as possible using the heuristics as guidelines Although heuristic evaluation can be a valuable tool for usability analysis preferably four or five evaluators should perform the same evaluation in order to be ef fective Molich amp Nielsen 1990 For the heuristic evaluation of an infusion pump all evaluators are re quired to have profound domain knowledge Heuristic evaluation is also limited because it focuses on the ex ecution of tasks in a controlled environment Usability violations that are found are not necessarily a prob lem for real users As the complexity of a task increas es it tends to be more difficult to i
113. w WYTlOIVOM 3NIdIQ3V2IN JNIJIQOWIN JNI33OATO 08 1N NIMODIPINE NITROGLYCERINE NITROPRUS SIDE 3JNIXJJNId33ON EN O A LIN NOREPINEFRINE OCTREOTIDE Jdio3820 oou _ PROPOFOL RANITIDINE ROCURONIUM TACROLIMUS VANCOMYCINE VAS SINE VERAPAMIL AMIODARON BE IEA CEFTAZIDIM BE eeen CLINDAMYCINE mW DIAZEPAM DOBUTAMINE DOPAMINE ESOMEPRAZOL FENTANYL FLUCLOXACILLINE FLUMAZENIL FUROSEMIDE a HEPARINE HYDROCORTISON INSULINE LE ereen KALIUMCHLORIDE P a MEL rome Opee Nome NIMODIPINE MM ED 0 OCTREOTIDE PROPOFOL RANTIONE ROCURONIUM VANCOMYORE INGGUINYA RI WININOANDOA A INIDAWOONYA ANIS HJO SWA TW Wd Was APPENDIX F MULTIPLEX USER INTERFACE Below you will find screenshots of the user interface of the Multiplex Infusion System IV Therapie monitor Therapie Instelingen Configuratie van Infuuskjwen noradrenaline 0 2 mg ml Infunderen 6 ml h NaCl 0 9 3 ml h kaliumchloride 2 ml h Volume 49 6 ml over Spuit leeg over meer dan 1 dag Pompprofiel Continue dosering Bolus snelheid 300 ml h insuline 1 IE ml 3 4 ml h propofol 20 mg ml 10 ml h morfine 2 mg ml 1 ml h UL Spuit leeg over 8 uur en 7 minuten Infunderen sp
114. y be a costly oper ation to test all interactions Therefore suggest to perform an empirical study in order to obtain frequency distributions of combinations of medi cation in the ICU The most frequently used combinations should be tested and knowledge about their compatibility could then be imple mented in the system Test the mixture of multiple substances in a sin gle IV line in order to find the required amount of buffer fluid This is required to alternate between medication through a single lumen safely Multiplex Infusion System should be to be connected to the upcoming patient data man agement system PDMS so that these systems can exchange information Multiplex Infusion System should be able to overrule all features of the current infusion pumps like alarms display messages and admin istration rates In order to be able to fully control these pumps collaboration with infusion pump manufacturers is required As it is important to know how accurate syringe pumps and volumetric are at high and low ad ministration rates the intrinsic variation of the in fusion pumps needs to be tested Contact the Dutch Health Inspection Dutch Inspectie voor Volksgezondheid to assess whether or not the Multiplex Infusion System complies with current healthcare regulations 10 CONCLUSIONS In this thesis illustrated that there are many prob lems with the current IV therapy system The Multi plex infus
115. y inexperience with the system resulting in additional clicks In the Multi plex condition the possibility to close a dialog window using the standard x button in the top right corner 45 was disabled As the participant clicked the x button a aaa message appeared asking the participant to choose between two options for example accept an order or decline As this was sometimes misread this led to multiple attempts and clicks to close the dialog win dow which affected the number of clicks In the separate pumps condition the participants worked with a simulation of the infusion pumps they operate on a daily basis As the results showed there are multiple strategies to perform the same task in the separate pump condition this has affected the average number of clicks and the variance between the partic ipants Although differences in the number of clicks were found in some events neither user interface had an overall lower number of clicks The Multiplex user in terface generally required more clicks than the sepa rate pumps interface when a checklist was used As this checklist may reduce errors during the execution of physical tasks attaching IV lines opening valves the additional clicks are considered to be acceptable 8 2 2 FIRST RESPONSE TIMES The first response times were generally quite high The fastest average response time was 4 seconds in the Multiplex condition during the syringe empty event This may be explai
116. yringe empty event F 1 15 2 925 p 0 108 NOR syringe empty F 1 15 1 635 p 0 221 Change administration rate F 1 15 0 50 p 0 826 Interruption event F 1 13 2 37 p 0 149 And the obstruction event F 1 15 0 257 p 0 620 Total execution times per event 0 20 40 60 80 Multiplex Syringe empty __ Separate pumps Syringe empty HEEE Multiplex NOR syringe empty Separate pumps NOR syringe empty Ooo iS 2 Multiplex Change administration WE Separate pumps Change administration _ __ rate Multiplex Interruption event Separate pumps Interruptionevent B Multiplex Obstruction Separate pumps Obstruction B Figure 22 The average total time for completing the tasks corresponding to each event Error bars denote the standard deviations 7 2 4 ERRORS Errors during the performance of tasks were identified for the Multiplex and separate pump conditions Table 6 lists all errors during the entire experiment An ignored alarm was defined as an event where it took more than 1 reminding alarm after a 10 second delay after the initial alarm in order for the participant to start performing the relevant actions Ignored alarms were eventually dealt with by the participant as the experimental setup did not allow moving on to the next event if an alarm was still active 8 errors occurred in total in the separate pump condi tion and three in
Download Pdf Manuals
Related Search