Software Training Manual - UCLA School of Public Health
Contents
1. 4 What proportion of men during the past month had vaginal and rectal sexual intercourse with either a single partner or two or more partners 5 What is the prevalence of HIV infection based on HIV antibodies in saliva 6 Does sexual behavior and injection practices predict the prevalence of HIV antibodies Frerichs R R Htoon M T Eskes N and Lwin S Comparison of saliva and serum for HIV surveillance in developing countries The Lancet 340 1496 1499 1992 Frerichs R R Eskes N and Htoon M T Validity of three assays for HIV 1 antibodies in saliva Journal of Acquired Immune Deficiency Syndrome 7 5 522 524 1994 Frerichs R R Silarug N Eskes N Pagcharoenpol P Rodklai A Thangsupachai S and Wongba C Saliva based HIV antibody testing in Thailand AIDS 8 885 894 1994 1 4 Epi Info and Stata Introduction Complete Data Set The data file aidsal mdb with information on all 300 men in the 360 households is available at http www ph ucla edu epi rapidsurveys RScourse RSstmanual html This is a realistic data set but does not contain real data Instead it is intended only for teaching purposes Since this is a rapid survey the questionnaire is limited to 24 variables that can be listed on two pages You will soon see that even two pages contain a substantial amount of information which requires time to analyze By understanding how long everything takes you will be more effective at convinci2. should appear Notice that the program is now at the CDC address CD BAFER HEALTHIER FEGFLE fj N What Is Epi Info CDC Home Search Health Topics A Z DISSS Home Contact Us Epi Info Figure 1 3 Enhanced Terronsm TEOT Location SS of Epi Project Info Materials Latest Version Epi Info Version software Downloads 3 4 3 Epi Info National Notifiable Disease Surveillance System Release Date November 26 2007 With Epi Info and a personal computer epidemiologists and other public health and medical professionals can rapidly develop a questionnaire or form customize the data entry process and enter Version 3 4 3 Contents Downloads Epi Info Maps User Support Installation User Forum and analyze data Epidemiologic statistics tables graphs and maps are produced with simple commands such as READ FREQ LIST TABLES GRAPH and MAP Epi Map displays geographic maps with data from Epi Info Data Systems Public Health Surveillance Left click with your mouse on Download then Download again and then either Web Install or Download Setup to transfer the program through your modem to your computer When completed epi the Epi Info icon should appear on your main computer screen Later you will click on the Epi Info icon to start the Epi Info program Csurvey In addition to Epi Info you should obtain the Csurvey 2 0 program This Windows
3. Mantel Haenszel Summary Chi Square 0 02 StatCalc P value 87735274 Su a y Crude RR for all strata 1 04 calculations Mantel Haenszel Weighted Relative Risk of Disease given Exposure 1 04 for both Greenland Robins Confidence Limits 0 80 lt MHRR lt 1 34 strata lt Enter gt for more F18 to quit Fi Help F5 Print F6 Open File F16 Done But there is still more The confidence intervals for the summary odds ratio is an estimate rather than an exact value Sometimes the estimate is very close to the exact value Other times however the two might vary The StatCalc program can calculate the exact value for you To do so press Enter and Figure 1 44 appears Figure 1 44 EpiInfo Version 6 Statale November 1993 Start exact calculations Disease 19 17 Press E for Exact Confidence Limits or lt Enter gt Press E and the program starts to calculate the exact confidence interval This usually takes a few moments so the program tells you to be patient as shown in Figure 1 45 EpiInfo Version 6 Statcalc November 1993 Figure 1 45 s Ruminating Disease Ruminating please be patient Once the calculations are done the screen appears with the answers see Figure 1 46 Epi Info and Stata 1 29 Analysis Epi Info EpiInfo Version 6 Statcalc November 1993 Disease xxExact Confidence Limits x Mehta CR Patel NR Gray R J Am Stat Assoc 1985 78 969 973
4. 20 39 leaving 19 households with eligible men One household cluster 2 household 13 had two eligible men Thus the total number of records is 27 1 e 25 x 1 1 x2 and the total number of records with data for the different variables is 20 1 e 18 x 1 1 x2 m Frequencies Next you will do a frequency distribution of the responses to Question 5 on marital status The program command is Frequencies in the column at left under Statistics When clicking on this program a panel appears that asks which variable is to be included Click on l then move the cursor to Married as presented in Figure 1 28 and click with the left mouse Married should appear in the selected box Figure 1 28 Request for frequency of variable married ILL _ Freq 20 35 37 65 Total 57 100 Weight Output to Table Settings Save Only Clear Help Cancel Press OK and Figure 1 29 appears Notice in the bottom box by the mouse arrow that the Epi Info command for frequencies is FREQ followed by the variable married This is the same command structure as in the DOS version of Epi Info Analysis Epi Info Epi Info and Stata 1 20 Analysis Analysis Commands Data Read Import my w Y g oO Open Bookmark Print Maximize Record Count 27 Deleted records excluded Date 4 3 2004 1 20 19 PM Relate FREQ Married Write Export Merge igea Next Procedure Und
5. MDB viewA LINKNAME TMPLNK_ 1 LIST GRIDTABLE FREQ Married FREQ vaccine m Tables The question arises are single men less knowledgeable about an AIDS vaccine than married men The appropriate analysis to answer this question is a cross tabulation of married and vaccine To create such cross tabulation instance the exposure variable 1s married table select under Statistics the program Tables In this and the outcome variable is vaccine Thatis we want to determine if exposure to marriage has an effect on the outcome of belief that there is a vaccine For the findings see Figure 1 31 Analysis Commands Data Read Import Relate Write Export Merge Variables Define Undefine Assign Recode Display Select If Select Figure 1 31 Cross tabulation of Married and vaccine Cancel Select If Sort Cancel Sort Statistics List Frequencies Match Means Graph Map Advanced Statistics Linear Regression Logistic Regression Kaplan Meier Survival Cox Proportional Hazards Complex Sample Frequencies Complex Sample T ables Complex Sample Means Output Header Type History ai amp Y amp 0 Open Bookmark Print Maximize Previous Procedure Next Procedure Current Dataset Previous TABLES Married vaccine Last Forward 6 AVAILABLE VACCINE 5 Married with wife in HH 1 2 1 il 6 Row 64 7 35 3 Col 84 6 100 0 0 0 85 0 2 A 0 1 3 Row 66 7 0 0 33 3 100 0 Col 15 4
6. by pressing gt and make the necessary changes 1f any When done click on x at the top left of the screen thereby closing the Enter Data program Return to the main menu to proceed to the analysis ANALYSIS WITH EPI INFO The data analysis module in Epi Info is very flexible and allows you to do many things We will explore only a few options here In the main menu click with the left mouse on Analyze Data then in the column at left click on Read import Change the Data Source by clicking with your left mouse on then enter Epi_info 418 aidsex1 mab Finally click in Views on ViewA as shown in Figure 1 26 Current Project CiEpi _Info Sample Mdb Data Formats Epi 2000 Data Source Fi gure 1 26 CrEpi_Info 4 18 aidsex1 MDB Show Read file jf Views C AI with data for analysis jm Change Project Save Onhy OK Clear Help Cancel A screen appears that mentions a temporary link and shows TMPLNK1 Click OK Your screen should now state that you have 27 records in C Epi_Info 41S aidsexI MDB viewA The program editor at the bottom right of the screen should show that your entered the instruction READ followed by details of the command As you proceed with your analysis each step will be recorded in the Program Editor mE List data In the Statistics section the first thing that we will do is list the data to make sure that they have been properly entered To do so click with the left mouse key on L
7. or Prompt Cluster Number Field or ariable Field Name Fi 1 19 Type Number Double click in prompt to change IpuIe 1 Cluster Create Pattern FE entry for Create first variable Peni Grid Related iew R t Last Code Tables Repeat Las Teanga Required Read Only OK Cancel 1 14 Epi Info and Stata Data Entry Continue to enter the information for the seven remaining variables and the second label as presented earlier in Table 1 1 When through your Make View screen should resemble Figure 1 20 iie MakesEdit View A File Edit View Insert Format Tools Help AIDS RISK FACTOR CLUSTER SURVEY 1 Cluster Number Figure 1 20 2 HH Number Created 3 Person Number M data fields 4 Age tin years for data ge r years MI entry 5 Married with wife in HH Do you believe 6 available vaedine a T infected but no disease 6 avallable drug to cure P While all the information is there the entry screen looks somewhat jumbled To arrange in better order hold down the left mouse key and place the ensuing box around the 10 lines of information Release the mouse key and move to the top of screen click format then alignment followed by vertical The Make View screen should now appear as in Figure 1 21 ints Make Edit View A File Edit wiew Insert Format Tools Help AIDS RISK FACTOR CLUSTER SURVEY 1 Cluster Number DR 2 HH Number a rune 1 21 3 Perso
8. person from getting the AIDS virus As before click on Frequencies then in the Frequencies of box select vaccine The results should be as shown in Figure 1 31 This time there are three categories of outcome 1 Yes 2 No and 3 Don t know A fourth category 9 No response was not used by any of the respondents Only 30 percent 1 e 6 of the 20 subjects recognized that a vaccine was not available to protect against AIDS Epi Info and Stata 1 21 Analysis Epi Info Analysis Analysis Commands Data Read Import Relate Write Export Merge Variables Define Undefine Assign Recode Display Select If Select Figure 1 30 Frequency distribution of vaccine Cancel Select If Sort Cancel Sort Statistics List Tables Match Means Graph Map Advanced Statistics Linear Regression Logistic Regression Kaplan Meier Survival Cox Proportional Hazards Complex Sample Frequencies Complex Sample Tables Complex Sample Means Output Header Type i 3 Last History i a i o Open Bookmark Print Maximize Previous Procedure Next Procedure Current Dataset Previous FREQ vaccine Forward 6 available vaccine Frequency Percent Cum Percent 13 65 0 65 0 6 30 0 95 0 1 5 0 100 0 Total 20 100 0 100 0 95 Conf Limits 1 40 8 84 6 2 11 9 54 3 30 1 24 9 rogram Editor New Program New Open Save Print Run Run This Command READ C Epi_Info 418 aidsex1
9. program derives various epidemiological statistics These statistics are revealed when scrolling down the output page as shown in Figure 1 34 Single Table Analysis Warning The expected values of a cell is lt 5 Fisher Exact Test should be used Figure 1 34 e Point 95 Confidence Interval Odds and r isk Estimate Lower Upper ratios for PARAMETERS Odds based aA t Odds Ratio cross product 0 0000 Undefined Undefined iT association Odds Ratio MLE 0 0000 0 0000 7 6742 m between 0 0000 11 8762 F PARAMETERS Risk based married Risk Ratio RR 0 6471 0 4555 0 9192 T and Risk Difference RD 35 2941 58 0113 12 5769 T vaccine T Taylor series C Cornfield M Mid P F Fisher Exact SUATISNGAL TESTS Chi square 1 talled p 2 tailed p Chi square uncorrected LOS 0 3097665756 Chi square Mantel Haenszel 0 9774 0 3228483885 Chi square corrected Yates 0 0448 0 8324138365 Mid p exact 0 2280701754 Fisher exact 0 4561403509 Since one of the cells contained a zero the odds ratio 1s also zero The risk ratio of 0 65 indicates that married men are 35 percent less likely to believe that an AIDS vaccine is available than single men The 95 confidence interval and the various statistical tests are inappropriate with our data set since the information comes from a two stage cluster survey with different variance estimates The statistical tests in this section of Epi Info assume the data were collected in a simple random sa
10. questionnaire continued 24 Code number of interviewer __ in unknown enter 99 This will be our first survey so the Study Number will be 001 The target population is all men aged 20 39 years in Region 234 of the country Based on existing census records we estimated that there are 548 529 people in 510 communities or villages termed clusters potentially accessible to our interviewers These people live in 111 900 households with an average of 4 90 persons per household We further estimated that about 83 percent of the households have at least one man aged 20 39 years At the first stage of our two part sampling process we sampled 30 of the 510 clusters with probability proportionate to the number of households in the cluster This is termed probability proportionate to size PPS sampling and will be further explained in the workshop In each cluster we randomly select 12 households and interview all men aged 20 39 living in these households Included in the sample was 300 men in the 360 selected households Look over the questionnaire All variables to be entered into the computer must have a number and name You also should give thought to how you want to present the findings With Epi Info you will be making an entry screen entering some data and with the complete aidsal mdb data set to be provided doing the initial analysis m Overview of Epi Info Epi Info tends to be self explanatory with many helpful message appea
11. regarding the latter it is more likely that you will want to use a regular word Introduction Epi Info and Stata 1 9 processing program of your own choosing More will come later on the use of a word processor and on StatCalc Finally move the cursor to Help as shown in Figure 1 13 Programs Edit Settings Utilities Ma6 Language EMSLISH Contents Translations How To Edit the Menu i Tutorials Figure 1 13 what s Mer Help About Epi Info menu The Contents tells you all about Epi Info including overviews of the different components of the program In this regard it is like a manual but in your computer rather than in a book While we will be using the English version of Epi Info other languages are either available for planned as explained in the Translations section Besides the example of a cluster sample featured in this manual there are three other tutorials in Epi Info To see them click on Tutorials The first is for an acute outbreak investigation of a food bourne pathogen occurring in Oswego County New York The second is also an outbreak investigation butin a hospital setting following open heart surgery The third tutorial is for a surveillance system showing how case records are computerized and tallied Note that none of the three tutorials deal with cluster surveys which are the subject of this Software Training Manual CREATING THE QUESTIONNAIRE When doing an interview you will need to have seve
12. 0 0 100 0 15 0 TOTAL 13 6 1 20 Row 65 0 30 0 5 0 100 0 Col 100 0 100 0 100 0 100 0 3 TOTAL 0 ig 0 0 100 0 Program Open Save Print Run Run This Command READ C Epi_ Info 418 aidsex1 MDB views LINKNANE TMPLNK_ 1 LIST GRIDTABLE FREQ Married FREQ vaccine TABLES Married vaccine X m If then As seen in Figure 1 31 there was one person who responded I don t know to the vaccine Analysis Epi Info Epi Info and Stata 1 22 question If we want to limit the analysis to those who had a definite opinion 1 e either responded yes or no we need to temporarily remove the code 3 response to vaccine from the data Epi Info lets you do this with various recoding statements one of whichis an if then statement The structure is if vaccine is equal to 3 then vaccine should be recoded as missing To create an if then statement click on Select ifin the Analysis Commands column then click on if Click on available variables and select vaccine Next click onl and end by entering 3 In the box labeled Then enter vaccine as shown in Figure 1 32 f Condition vaccine 3 Figure 1 32 a Create Available Variables if then vaccine Statement to limit vaccine to yes or no Functions Save On OK Clear Cancel Click OK Note that program statement has been added to the Program Editor box With vaccine limited to yes or no responses you will run the tables program again
13. Click on Tables under Statistics in the Analysis Commands column and enter Married and vaccine as before The new table is shown in Figure 1 33 Analysis Ef i poe zu Exit BS a amp O Previous History Open Bookmark Print Maximize Data TABLES Married vaccine Read Import Relate Write Export N Merge Previous Procedure Next Procedure Current Dataset Variables Define Undefine Forward i Assign 6 AVAILABLE VACCINE Figure 1 33 Reco Se Display 5 Married with wife in HH 1 2 TOTAL Select If Knowledge a 270 if ul s 7 Cancel Select Row 64 7 35 3 100 0 of vaccine If Col 84 6 100 0 89 5 hie 2a 2 o0 2 C 1 Sort among Ge Statistics Row 100 0 0 0 100 0 e in Epi Info and Stata 1 23 Frequencies Match Means Graph Map Advanced Statistics Linear Regression Logistic Regression Kaplan Meier Survival Cox Proportional Hazards Complex Sample Frequencies Complex Sample T ables Complex Sample Means Output Header Type RouteOut lann k TOTAL 13 6 19 Row 684 31 6 100 0 Col 100 0 100 0 100 0 Single Table Analvsis Run This Command JREAD C Epi_ Info 418 aidsex1 MDB viewd LINKNAME TMPLNK 3 FREQ Married FREQ vaccine TABLES Married vaccine IF vaccine 3 THEN vaccine END m Odds and Risk ratios Notice that by comparing two dicotomous 1 e two category variables Analysis Epi Info married and vaccine you have created a four fold table and the analysis
14. Directory Set INI Fie Drectory Figure 1 9 Settings menu This menu gives you the option of choosing a Epi Info database version To do so move the cursor to Choose Epi Info Database Version and make sure that the option shown in Figure 1 10 is selected 1 8 Epi Info and Stata Introduction int Database Format Options Select the default format for creating new databases MOE Access 2000 Figure 1 10 New databases created by Epi Info will be created in an Access 2000 compatible format Epi Info 2002 Settings released July 2002 and later can read this format menu Cancel Create a subdirectory in your computer under c Epi_Info named 4 8 This will become your working directory for the course Once the subdirectory has been created click on Settings and then Set Working Directory and move the cursor to 4 8 as shown in Figure 1 11 Click OK when done Please choose a working directory for which you have write privileges Sc N EJE Figure 1 11 EJ Epi_Info Settings menu Cancel The next set of programs in Epi Info are the utilities Move the cursor to Utilities and the screen shown in Figure 1 12 should appear Programs Edit Settings MARIES Language EMGLISH SkatCalc Data Compare Table to view Visualize Data Figure 1 12 ee Utilities Compact Word Processor menu Here are two programs that we will be using in this manual namely StatCalc and possibly Word Processor although
15. Figure 1 46 Pascal program by ELF Franco amp N Campos Filho Ludwig Cancer Institute Sao Paulo Brazil Exact confidence Exact Lower 95 Confidence Limit 0 61 interval Mantel Haenszel Weighted Odds Ratio 1 08 Exact Upper 95 Confidence Limit 1 93 for stratified odds ratio lt Enter gt to continue Fi Help F5 Print F6 Open File F10 Done Press Enter one more time and you return to the calculation screen entry of another set of numbers see Figure 1 47 EpiInfo Version 6 Statcalc November 1993 Disease Figure 1 47 E x Entry screen for new calculations f Fi Help F6 Open File F10 Done The next section features an analysis of two data sets included with the Epi Info software and a rapid survey of 300 men in 360 households described earlier in this chapter Analysis Epi Info Epi Info and Stata 1 30
16. Software Training Manual Windows Ralph R Frerichs D V M Dr P H Professor Department of Epidemiology University of California Los Angeles UCLA Rapid Survey Course UCLA November 2008 Chapter One Epi Info and Stata TABLE OF CONTENTS Obtaining software program eee eee teen eens Introduction Creating the questionnaire Data entry cause ates Analysis with Epi Info Analysis of cluster surveys with Epi Info 1 ens Analysis of cluster surveys with Stata 0 0 ce eee nn ees Concluding remarks Chapter Two Form making Introduction Management forms Concluding remarks Chapter 1 EPI INFO and STATA This training manual was last updated for the Spring Quarter 2008 UCLA course EPI 418 Rapid Epidemiological Surveys in Developing Countries It has been slightly modified for the Rapid Survey Course offered on the web The main software programs for rapid surveys to be presented in this course is Epi Info It is a shareware program free to copy produced by the United States Centers for Disease Control and Prevention CDC and distributed in collaboration with the World Health Organization WHO The program has been used by thousands of epidemiologists around the world including most developing countries The authors of the Epi Info program have included helpful tutorials with their program along with an electronic version of an instructio
17. TIONS 8 10 Do you believe 8 there is a vaccine available that protects a person from getting the AIDS virus 1 Yes 2 No 3 Don t know 9 No response 9 person can be infected with the AIDS virus and not have the disease AIDS 1 Yes 2 No 3 Don t know 9 No response 10 there is a drug available that can cure a person with AIDS disease 1 Yes 2 No 3 Don t know 9 No response Introduction Epi Info and Stata 1 5 AIDS RISK FACTOR CLUSTER SURVEY continued REPEAT FOR QUESTIONS 11 14 How effective do you think is for preventing AIDS disease through sexual activity 11 using a diaphragm 1 Very effective 2 Somewhat effective 3 Not at all effective 4 Don t know how effective 5 Don t know method 9 No response 12 using a condom 1 Very effective 2 Somewhat effective 3 Not at all effective 4 Don t know how effective 5 Don t know method 9 No response 13 having a vasectomy 1 Very effective 2 Somewhat effective 3 Not at all effective 4 Don t know how effective 5 Don t know method 9 No response 14 sexual intercourse only between two people who do not have the AIDS virus 1 Very effective 2 Somewhat effective 3 Not at all effective 4 Don t know how effective 5 Don t know method 9 No response REPEAT FOR QUESTIONS 15 17 During the past year 15 Have you received an injection with a needle in your muscle vein or skin 1 Yes 2 No 3 Don t know 9 No res
18. This has long been one of my favorite components of the program and is useful for analyzing data a wide variety of epidemiologic data Go to the Utilities menu of Epi Info as shown in Figure 1 36 and click with the left mouse on Statcalc Programs Edit Settings Elsi Lanqguage ENMsLISH Stattalc Data Compare Figure 1 36 Table to View Program Visualize Data Epi Lock menu showing Compact StatCalc Word Processor program Assume that you have available the following numbers for an analysis relating the drug question 1 e Do you believe there is a drug available that can cure a person with AIDS disease to the condom question i e How effective do you think is using a condom for preventing AIDS disease through sexual activity stratified by marital status Married Single Believed Effectiveness of Condoms for Preventing AIDS Effect Other Effect Other Drug 156 Yes 19 36 available 54 No 11 18 113 97 210 30 54 Rather than going through the involved steps of entering the data on 264 persons into the computer and doing the analysis as before all you want is a simple calculation of measures of association for the available data As you will see next StatCalc is very useful for this To use the program press Enter and Figure 1 37 appears Analysis Epi Info Epi Info and Stata 1 26 EpilInfo Version 6 Statcalc November 1993 Figure 1 37 StatCalc Sample size amp power i Chi square f
19. and no one will know his identity since his name will not be written on the interview form 2 Region No __ _ 3 Cluster No _ _ sate Maps Household No _ 5 Subject No in HH Age _ _ years 99 if Unk 7 Married with wife in household 1 Yes 2 No 9 Unknown or No response e Reports v For Help press F1 Analyze Data Epi Info Website Abbreviated Data Set Rather than starting with the larger data set we will begin with data on only a few questions and limited to men in the 13 sampled households in Clusters 1 and 2 The abbreviated questionnaire is shown in Figure 1 15 Department of Epidemiology University of California at Los Angeles Los Angeles California AIDS RISK FACTOR CLUSTER SURVEY 1 Cluster No 2 HH No 3 Person No ___ 4 Age _ yrs Figure 1 15 5 Married with wife in household 1 Yes 2 No 9 Unknown or No response Complete text REPEAT FOR QUESTIONS 6 8 Do you believe for short 6 there is a vaccine available that protects a person from getting the AIDS virus questionnaire 1 Yes 2 No 3 Don t know 9 No response 7 a person can be infected with the AIDS virus and not have the disease AIDS 1 Yes 2 No 3 Don t know 9 No response 8 there is a drug available that can cure a person with AIDS disease 1 Yes 2 No 3 Don t know 9 No response The short names of the eight variables and their characteristics for Epi Info s Make View prog
20. dy for data entry Return for a moment to Table 1 2 and note the information on the first sampled household Table 1 2 Data for Make View entry screen CLUSTER HH PN AGE MARRIED VACCINE INFECTED DRUG First household in cluster 1 il 1 1 23 1 1 2 2 Remember that cluster has two digits Thus when you enter 7 it will appear as 01 Enter each of the numbers into the appropriate fields on the screen followed each time by Enter 1 e the Enter key Stop after entering 2 for Drug but before tapping the Enter key Your screen should appear as in Figure 1 23 1 16 Epi Info and Stata Data Entry Vrile Edit Options Help AIDS RISK FACTOR CLUSTER SURVEY 1 Cluster Number 2 HH Number Save data ad dele Figure 1 23 P EE 3 Ferson Number Data for first 4 Age fin years subject 5 Married with wife in HH Do you believe 6 available vaccine f infected but no disease amp available drug to cure Press Enter and the data for the first household are entered into the computer followed by a blank entry screen ready for data for the next subject Notice that some of the households did not have eligible subjects Thus the data fields for them are left blank The first such HH with no eligible subject is number 5 which should be keyed as 1 5 O and then blanks Proceed to enter the remaining data in Table 1 2 until you get to the last field of the last household Table 1 2 Data for Make View entry scree
21. efine i Recode 5 Married with wife in HH Frequency Percent Cum Percent Figure 1 29 e Displ E 8 Bsdii 1 17 85 0 85 0 ihe Saale Select 2 315 0 100 0 distribution pone aren of married If Sort Cancel Sort Statistics Total 20 100 0 95 Conf 100 0 List Limits 1 62 1 96 8 Tables Match 23 2 37 9 Means Graph Map Advanced Statistics Linear Regression Logistic Regression Kaplan Meier Survival Cox Proportional Hazards Complex Sample Frequencies Complex Sample T ables Complex Sample Means Output Header Run This Command READ C Epi_Info 418 aidsex1 MDB viewA LINKNAME TMPLNK_1 LIST GRIDTABLE FREQ Married Eighty five percent of the 20 men in the 26 households were married with a wife present while 15 percent were not None refused or did not answer The frequency distribution includes a 95 confidence interval for both percent married 1 e 62 1 96 8 and not married 1 e 3 2 37 9 Disregard this information The confidence intervals in the FREQ program assume the data were collected in a survey featuring simple random sample rather than two stage cluster sampling For the latter the confidence intervals tend to be much wider as you will learn later The frequency distribution however is applicable for all kinds of sampling Next do a frequency of the variable vaccine to see how the 20 men responded to the question Do you believe there is a vaccine available that protects a
22. ist In the box that appears click M All t Except followed by OK The screen should now show a grid table with all of the data as seen in Figure 1 27 Epi Info and Stata 1 19 Analysis Epi Info tar Figure 1 27 List of 27 records in data file Analysis Analysis Commands Data Read Import Relate Write Export Merge gt Variables Define Undefine Assign Recode Display Select If Select Cancel Select If Sort Cancel Sort Statistics Frequencies Tables Match Means Advanced Statistics Linear Regression Logistic Regression Kaplan Meier Survival Cox Proportional Hazards Complex Sample Frequencies Complex Sample T ables Complex Sample Means Output READ C Epi_Info 418 aidsex1 MDB viewd LINKNAME TMPLNK_1 LIST 23 37 27 23 Missing 25 26 Missing 39 35 Missing 35 27 37 34 Missing 36 Missing 28 26 Missing 28 Missing 26 28 39 20 GRIDTABLE Missing 2 1 Missing 1 1 Missing 1 F 2 Missing 1 Missing 1 1 Missing Missing 1 1 2 Missing 1 1 Missing 2 2 Missing 2 2 1 3 Missing 1 Missing 1 1 Missing 1 Missing 1 1 Missing 2 2 Missing 1 2 Missing 1 1 2 2 Missing 1 Missing 3 1 Missing 2 Missing Missing 1 1 Missing 2 1 Missing 1 1 2 3 Missing 2 Missing 1 2 Missing 2 Missing Notice that the data set contains 26 households 7 of which have no eligible men 1 e aged
23. l first enter an abbreviated version of the questionnaire for data entry The intention here is to have enough words showing to remind person entering the data of the variable field but not so many to clutter up the entry screen First you should enter the ttle and then a short name for the various items or questions with just enough information to remind the person entering the data which field is to be considered To start click on Make View either on the box at the left side of the screen or in Programs at 1 12 Epi Info and Stata Data Entry the top of the screen When the Make Edit View screen appears click at the top on File and then New Create a file name aidsex which should be stored in c Epi_Info 418 as shown in Figure 1 16 This file will hold a database aidsexl mdb once you have entered the data Create or Open PROJECT Look ir Ener c Ez hy Recent Documents Fy 1 Desktop Figure 1 16 Create data entry file My Documents File name aidseni hai My Network Files of type Database Files MDB Cancel Places Open as read only Every page in Make View is termed a view We will be using only one page but it still needs to be named For our example name the view A as seen in Figure 1 17 Click OK to continue C Epi_Infot41 64aidsex1 MDG Name the View Figure 1 17 Create Use only letters and numbers data entry file Do not start View name with a Humber a
24. mple with each subject being independent from others This assumption is not valid in cluster surveys although the risk and odds ratios are valid Means For the final analysis you will determine if those who believe in the availability of a vaccine i e answered yes are different in age from those who responded no Age is a continuous variable Therefore rather than requesting a table as is done for categorical data you should use the means command To do so click on Means in the Statistics section of the Analysis Commands column and enter Means of Age cross tabulated by vaccine The results in the long analysis section are shown in Figure 1 35 Analysis Epi Info Epi Info and Stata 1 24 Figure 1 35 Statistics with means output for Age and vaccine Epi Info and Stata 1 25 Analysis Epi Info Persons who believe in the availability of an AIDS vaccine are 4 3 years younger than men who do not believe that such a vaccine exists 1 e mean of 28 4 years versus mean of 32 7 years If this had been a simple random sample the analysis of variance ANOVA statistics would have been appropriate suggesting the difference is not statistically significant Since the findings come from a cluster survey however the statistical tests in this section of Epi Info should not be used The means however are valid m Statistics Calculator Another analytic feature of the Epi Info program is the Statcalc program
25. n CLUSTER HH PN AGE MARRIED VACCINE INFECTED DRUG Last household in cluster 2 2 13 2 20 2 1 2 2 Notice if your lose track of where you are the record number is shown at the bottom left corner of Record the screen for example here is what it looks like for record 6 pc M lt lt lt gt gt gt Just before entering the last value for the last HH in cluster 2 1 e subject 27 stop again do not press Enter The screen should appear as in Figure 1 24 Data Entry Epi Info and Stata 1 17 me oO i AIDS RISK FACTOR CLUSTER SURVEY 1 Cluster Number 2 HH Number Save data Mark record a deleted 3 Person Number A Age tin years Figure 1 24 Data for first subject 5 Married with wife in HH Do you believe 6 available vaccine T infected but no disease 6 avallable drug to cure Record If your screen shows that you are entering data for the 27 subject and the values are as shown press Enter Save the data by clicking with your left mouse as shown in Figure 1 25 Figure 1 25 Save data on 27 subjects Save data Mark record as dellaed Find 1 18 Epi Info and Stata Data Entry To make sure that you entered the data correctly or want to make changes click on lt lt in the bottom Record left of the screen to return to record 1 as shown here __ FE _ Scroll through the various oP E gt gt gt CTTTEETITET entry screening
26. n Number Aligned data fields for data 4 Age in years entry 9 Married with wife in HH Do you believe 6 available vaccine infected butne disease 3 avallable drug to cure Data Entry Epi Info and Stata 1 15 Notice in Figure 1 21 that four of the variables have space for two digits and four have space for only one digit If this is not so with your Make View screen go back and straighten the variable fields out before continuing When satisfied click on File and then Save to save the Make Screen file aidsexl mdb Abbreviated Data Set Rather than starting with the larger data set we will begin with data on only a few questions and limited to men in the 13 sampled households in Clusters and 2 The abbreviated questionnaire was shown in Figure 1 15 Return to the initial Epi Info menu see Figure 1 6 and click on Enter Data followed by File see the top line of the screen and Open If you had properly set the program so that it opens in C Epi_Info 418 then the screen in Figure 1 22 should appear Look in 418 ey Eg E aidsex1 My Recent Documents Figure 1 22 Open file for data entry My Documents i A 39 My Computer File name a l My Network Files of type Project db Pl aces Open as read only Click with your left mouse on Open and on table A followed by OK The same screen that was presented in Figure 1 21 should now appear rea
27. n manual OBTAINING SOFTWARE PROGRAM The programs for this course can be obtained on the Internet or from a friend mE Internet I assume you are using the Microsoft Internet Explorer Once you have logged on to the world wide web enter http www ph ucla edu epv and the screen shown in Figure 1 1 should U C N Department of Epidemiology School of Public Health University of California Los Angeles UCLA about EPI Department of academics Figure 1 1 courses amp seminars f i centers amp programs Screen for Naai ik So i cA i resources p o Si A IA D AG Epidemiology 1 KITE S A 4 Department AN A 4 b campus map See T 2 k j f School of Public Health Box 951772 Las Angeles CA 90095 1772 USA General Information 310 825 6579 nent se Epidemiology 910 206 6039 appear Click with your mouse on resources in the column at left then when the new page appears scroll down to software and click on it When you do the screen presented in Figure 1 2 should appear showing a list of software programs that can be down loaded from the Department of Epidemiology website You should be at http www ph ucla edu epi software html Only a few of the programs are actually stored at UCLA The web page has instructions however that link you along the electronic highway to another computer where the software is stored Such a computer is termed a file server or simply a server The first softwa
28. nd do not use any Spaces Change Project The first field that you will be entering is not a variable but rather a label which presents the study name The screen should read Right click to create a field Towards the left border of the screen click on the right side of the mouse Enter the title of our survey as shown in Figure 1 18 making the font Arial 12 click on Font for Prompt and the style of the field as Label Title Since we will not be entering information using this line it is considered merely as a label or a title Enter OK when done Move the title with your mouse hold down the left mouse key and move it to the upper left corner as far as it will go Data Entry Epi Info and Stata 1 13 Field Definition Question or Prompt aes RISK FACTOR CLUSTER SURVEY Font for Prompt Fi ure 1 18 Field or ariable Field Name S Create Type Double click in prompt to change AidsRiskFaCTOR first entry as a label Create or title m Grid Related iew Code Tables OK Cancel The first data field that you will be entering is the cluster number which requires two digits The variable is to be named cluster for the data set but identified as 7 Cluster Number for the data entry screen as seen in Figure 1 19 Notice that the number field has two digits signified by The variable name is cluster and the font should be Arial 12 point regular see Table 1 1 Field Definition Question
29. ng those seeking information that less is more That is they will have more useful information readily available for decision making if only they can limit the number of questions being asked In the coming pages I will first present the questionnaire used in the survey see Figure 1 5 You will then use a shortened version of the questionnaire to program the Epi Info software to enter and analyze survey findings Next you will enter data for 20 subjects followed by the analysis of several questions Following that you will use the program s statistics calculator to analyze entered numbers Finally you will analyze data in the aidsal mdb using the cluster and regular analysis features of Epi Info Department of Epidemiology University of California at Los Angeles Los Angeles California AIDS RISK FACTOR CLUSTER SURVEY Complete for all men aged 20 39 now living in the household Tell each 1 that some of the questions are about his personal life so you will want to speak to him in private 2 the information will be used to help plan services for his community and 3 no one will know his identity since his name will not be written on the interview form Figure 1 5 HIV AIDS Study NOs amp 2 Region No _ _ 3 Cluster No _ _ risk factor Household No 5 Subject No in HH questionnaire Age years 99 if Unk Married with wife in household 1 Yes 2 No 9 Unknown or No response REPEAT FOR QUES
30. ogram allows you to convert data that are entered in Epi Info into a file format that 1s compatible with Stata Itis found in Epi Info to Stata Format section of the software linkage at the UCLA website http www ph ucla edu epi csurvey html see Figure 1 4 Stata This program does multivariate analyses well beyond the capacity of the Epi Info program Stata has a set of survey modules that permit the analysis of two stage cluster surveys like those featured in the Rapid Survey Course The program and user manuals can be purchased from Stata Corporation More details are presented on the Rapid Survey Course website http www ph ucla edu epi rapidsurveys RScourse RSstmanual html Obtaining program Epi Info and Stata 1 3 INTRODUCTION This exercise requires both imagination and patience Imagine that a community based survey was done in the rural regions of a developing country to obtain information for an AIDS intervention program With patience proceed through the pages of this teaching exercise and try to learn the strengths and weaknesses of the Epi Info program for entering editing and analyzing the survey findings Assume that a two stage cluster survey was done last September of knowledge of AIDS occurrence of injection practices and various forms of sexual activity and the prevalence of HIV infection as measured by HIV antibodies in saliva Three hundred men aged 20 through 39 years were included in a sample of 360 ho
31. or trend opening menu Fi Help F6 Open File F10 Done Move the cursor to Tables 2 x 2 2 x n and press Enter to start the program Figure 1 38 should appear with an empty table for cross tabulations Notice that the outcome or dependent variable is listed as disease and the risk or independent variable is listed as exposure In our example Condom is the disease variable and drug is the exposure variable EpilInfo Version 6 Statcalc November 1993 Disease Figure 1 38 StatCalc Cross tabulation Fi Help F6 Open File F18 Done First enter the numbers for those who are married 1 e stratum one as shown in Figure 1 39 EpilInfo Version 6 Statcalc November 1993 Disease Figure 1 39 StatCalc entries for Stratum F4 Calc F6 Open File F180 Done Epi Info and Stata 1 27 Analysis Epi Info After the numbers are entered press F4 Calc and Figure 1 40 appears EpilInfo Version 6 Statcalc November 1993 Disease Analysis of Single Table Odds ratio 1 23 0 63 lt OR lt 2 39 Cornfield 95 confidence limits for OR Relative risk 1 10 0 82 lt RR lt 1 49 Taylor Series 95 confidence limits for RR Ignore relative risk if case control study Chi Squares P values Figure 1 40 ee Uncorrected 0 42 0 5147289 StatCalc Mantel Haenszel 0 42 080 5157316 calculations Yates corrected 0 24 0 6219113 for F2 More Strata lt Enter gt No More Strata F10 Quit stratum 1 Fi Help F2 St
32. ponse 16 Have you received a transfusion of blood or blood components platelets or plasma Figure 1 5 1 Y 2 N 3 Don t k 9 N HIV AIDS es O on now O response 17 Not counting injections or transfusions mentioned previously have you had any risk factor part of your body pierced by acupuncture by tatoo or having your ears nose or nipples pierced or something like that questionnaire A 1 Yes 2 No 3 Don t know 9 No response continued REPEAT FOR QUESTIONS 18 21 During the past month 18 Have you had sexual intercourse during which you put your penis in your partner s vagina 2 No 3 Don t know 9 No response 1 Yes 19 If yes have you done this during the past month with more than one partner 1 Yes 2 No 3 Don t know 9 No response 20 Have you had sexual intercourse during which you put your penis in your partner s rectum 1 Yes 2 No 3 Don t know 9 No response 21 If yes have you done this during the past month with more than one partner 1 Yes 2 No 3 Don t know 9 No response 22 Was a saliva specimen collected from this subject 1 Yes 2 No 23 Results of HIV antibody assay laboratory findings 1 Positive 2 Negative 3 Indeterminant 9 No specimen 1 6 Epi Info and Stata Introduction AIDS RISK FACTOR CLUSTER SURVEY continued Figure 1 5 HIV AIDS risk factor This concludes the interview Thank you for taking the time to participate
33. program automates several steps necessary for doing rapid surveys In collaboration with Professor 1 2 Epi Info and Stata Obtaining program Frerichs the program was written by Muhammad N Farid a graduate student in the Department of Epidemiology sponsored by the Fogarty International HIV AIDS Training Program An earlier DOS version also in collaboration with Dr Frerichs was written by Iwan Ariawan M D M P H a former graduate student in Epidemiology now on the faculty of the University of Indonesia When through with getting Epilnfo return to the Epidemiology Department software web site by left clicking with your mouse on lt Back at the top of your screen Move down the screen to Csurvey 2 0 Windows and with your mouse left click on Csurvey The next screen will appear as shown in Figure 1 4 More your cursor down to the Windows Version section at the bottom Down load the program as before by left clicking with the mouse Save the file on your C drive in a subdirectory named download Use zip program if necessary Note that these are DOS programs rather than Windows having been written some while ago To install the program on your computer change directories to C download and enter install The program will automatically create a directory C CSURVEY on your computer and copy the necessary files U C Department of Epidemiology School of Public Health CSURVEY SOFTWARE Search DOS VERSION Ralph R Fre
34. ral pages before you with all of the questions clearly presented along with options for the answers To create such a questionnaire you typically use a word processing program or if you have no favorite program available the Word Processor in Epi Info Once the information is collected you will want to transfer the data to a computer using a data entry screen To this end you will using Make View create a shorter version of the questionnaire appropriate for data entry First if doing a field survey and wanting to use the Epi Info word processor you would return to the Utilities menu and click on Word Processor Then you would enter the questionnaire text shown in Figure 1 5 as presented in Figure 1 14 You would typically print this for the field staff as the survey instrument 1 10 Epi Info and Stata Creating the questionnaire i Epilnfo ooo Lanquage ENGLISH gee Ei Document WordPad File Edit Yiew Insert Format Help Deh 66 aen amp Courier New i 3 Westem Eres ce a ee 5 Figure 1 14 Department of Epidemiology University of California at Los Angeles Create Los Angeles California questionnaire AIDS RISK FACTOR CLUSTER SURVEY for field Complete for all men aged 20 39 now living in the household Tell each that some of the questions are about his personal life so you will want use to speak to him in private the information will be used to help plan services for his community
35. ram are shown in Table 1 1 You will be using the data shown in Table 1 2 First however we need to create the data entry screen using Make View Creating the questionnaire 1 11 Epi Info and Stata Table 1 1 Data labels and characteristics for Make View program No Short description Name DIOILS FONC Size AIDS RISK FACTOR CLUSTER SURVEY Arial 12 Bold 1 Cluster Number Cluster 2 Arial 12 Regular 2 Household Number HH 2 Arial 12 Regular 3 Person Number PN 2 Arial 12 Regular 4 Age Age 2 Arial 12 Regular 5 Married with wife in HH Married i Arial 12 Regular Do you believe Arial 12 Bold 6 available vaccine vaccine 1 Arial 12 Regular 7 infected but no disease infected il Arial 12 Regular 8 available drug to cure drug ji Arial 12 Regular Table 1 2 Data for Make View entry screen CLUSTER HH PN AGE MARRIED VACCINE INFECTED DRUG ae 1 1 23 Ji Ji 2 2 al 2 il oul 1i 2 J 2 1 3 1 PA i al i 1 1 4 1 PAG 1 2 3 1 1 5 0 1 6 1 25 2 L 2 1 i 7 1 26 ii ii 2 l i 8 O 1 9 1 39 il 2 1 2 1 iQ 1 55 ik 2 2 ik 1 11 O db 12 1 35 1 2 1 1 jk L3 1 21 1 2 1 i 2 ik 1 ow 1 ii 2 2 2 2 sl 34 2 3 2 3 2 3 0 2 4 1 36 1 I l 2 2 5 0 2 6 1 28 1 1 3 1 2 7 1 26 i 1 a 2 2 8 1 2 9 1 28 a 1 2 2 2 10 2 ek 1 26 ii ii 1 2 Z 1 2 1 28 i i 1 it 2 cs 1 39 1 1 a gt 2 13 2 20 2 1 2 2 DATA ENTRY To enter the data shown earlier in Table 1 1 you need an entry screen This is created using the Make View program of Epi Info You wil
36. ratum F5 Print F6 Open File F180 Done This is the interim analysis of the stratum one To enter stratum two for the single men press F2 see code line at bottom of screen Enter the next set of numbers as shown in Figure 1 41 EpiInfo Version 6 Statcalc November 1993 Disease Stratified Analysis Table 2 Figure 1 41 StatCalc numeric entries for stratum 2 Fi Help F6 Open File F10 Done When done entering the numbers the program calculates the measures of effect for stratum two see Figure 1 42 EpiInfo Version 6 Statcalc November 1993 Disease Odds ratio 8 71 0 19 lt OR lt X 2 60 Cornfield 95 confidence limits for OR Relative risk 0 86 0 53 lt RR lt 1 4 Taylor Series 95 confidence limits for RR Ignore relative risk if case control study Chi Squares P values Figure 1 42 Uncorrected 0 34 0 5612758 Statcale ao i te calculations F2 More Strata lt Enter gt No More Strata F10 Quit for stratum 2 Fi Help F2 Stratum F5 Print F6 Open File F10 Done Analysis Epi Info Epi Info and Stata 1 28 Since there are no more strata press Enter and the program derives the summary statistical measures as shown in Figure 1 43 EpiInfo Version 6 Statcalc November 1993 Disease xxxxx Stratified Analysis xxx Summary of 2 Tables Crude odds ratio for all strata 1 08 Mantel Haenszel Weighted Odds Ratio 1 08 e Cornfield 95 Confidence Limits Figure 1 43 0 61 lt 1 08 lt 1 94
37. re to be obtained is Epi Info To do so left click with the mouse Epilnfo Windows then click on Downloads and the screen in Figure 1 3 Obtaining program Epi Info and Stata 1 1 Figure 1 2 Search bioterrorism john snow site centers amp programs faculty amp preceptors resources Dept of Epidemiology University of California Los Angeles UCLA School of Public Health U C N Department of Epidemiology School of Public Health EPIDEMIOLOGY SOFTWARE DOS AND WINDOWS VERSIONS ONLY Epilnfo Windows Easier to use version of CDC s popular analysis word processing and database management program for epidemiologists The S creen for UCLA epidemiology program includes Complex Sample modules for the analysis of cluster surveys Epimap a geographic information system and ublic about EPI Nutstat a nutrition anthropometry program Used in EPI 418 and p ee ae featured in the EPI 418 Software Training Manual domain a E a Epi Info Tutorials software Epilnfo 6 DOS Analysis word processing and database management program for epidemiologists It also contains CSample necessary for the analysis of cluster surveys Those who prefer using DOS should consider this version Not used in EPI 418 OpenEpi Web This internet site is the brain child of Andy Dean the father of Epilnfo infrastructure and website and Kevin Sullivan statistics They have created a web based menu driven system which runs or links
38. richs The program is necessary to plan and organize two stage cluster surveys It is taught in EPI 418 Rapid Surveys but is also available for Bioterrorism free to others Contemporary history of bioterrorism i Disease detectives e Installation of Csurvey HIV controversies Information in a PDF file for Windows XP users for downloading parena site extracting and installing the zip file which contains the Csurvey a cluster survey program Figure 1 4 UCLA epidemiology Csurvey and about err Epi2dct exe academics pro gram S courses amp seminars centers amp programs CSurvey Manual PDF files It requires Adobe Acrobat Reader software to view and to print the manual e Csurvey Cluster Survey Program e Manual faculty amp preceptors resources links e Winzip Program to be purchased It requires Zip Program to open the program and the manual Dept of Epidemiology e Epilnfo to Stata format Stata Utility University of California Los Angeles UCLA A utility to convert Epilnfo data to Stata format School of Public Health Box 951772 e How to convert Los Angeles CA 90095 1772 USA Information on how to convert Epilnfo data to Stata format General Information 310 825 8579 WINDOWS VERSION General Fax 310 206 6039 CSurvey 2 0 recently debugged 3 1 08 is now available The program is being used in EPI 418 Rapid Survevs but is also available for free to Epi2dct exe This small pr
39. ring here and there To start epi the program click on the screen icon kig and the screen in Figure 1 6 should appear The top row Epi Info shows the various components of the program We will briefly explore each Figure 1 6 Initial menu Move the cursor with your mouse and click on Programs You should see the menu shown in Figure 1 7 Introduction Epi Info and Stata 1 7 saue Edit Settings Utilities Help Make View Questionnaire Enter Data Analyze Data Create Maps Figure 1 7 Create Reports Programs Mutrition menu Exit In this exercise you will be using Make View Enter Data and Analyze Data but not until after we have looked at some of the other features in this program You will return to this menu showing the main programs many times Next move the cursor to Edit by pressing the right arrow key gt and the menu in Figure 1 8 appears Programs Edit Settings Utilities Help Picture Edit This Menu Buttons on or off Figure 1 8 Move Resize Button lt Shift F2 gt Edit menu This provides editing functions that you will later explore on your own once you become more familiar with the program Now move the cursor to Settings either with your mouse or by pressing the right arrow key gt and the menu in Figure 1 9 appears Programs Edit Settings Utilities Help Manage Translations Choose Epi Info Database Version Choose Epi 6 Import YEAR and SPLITYEAR Set Working
40. using units from a population of 93 250 housing units then interviewed and asked for saliva specimens The investigators who created the present study were interested in learning what people believe about AIDS and AIDS prevention the prevalence of high risk injection practices sexual activity and HIV infection and the association between current infection and various risk factors They reasoned that with this information they would 1 have some idea as to how quickly HIV infection is spreading through the population 2 be able to provide information for planning a health education program and 3 have baseline information to evaluate HIV control measures QUESTIONS TO BE ANSWERED Specifically the investigators were interested in answering the following questions 1 Do young and middle aged men at the village level know that friends and neighbors could be infected with the AIDS virus but not have the AIDS disease that there 1s no vaccine to prevent AIDS infection and there is no drug available to cure a person with AIDS disease 2 How effective do men feel are various devices or methods for preventing AIDS infection Included are the use of a diaphragm or condom having a vasectomy and limiting sexual intercourse to two people who do not have the AIDS virus 3 What percentage of men during the past year were injected with a needle received a blood transfusion or had their skin pierced for some other reason such as acupuncture or a tattoo
Download Pdf Manuals
Related Search