Interpreting Surveillance Data

WIN Project

(this data is for illustration purposes and cannot be used for publication)

(modified 2/08/99)

Objectives: 

1. To discuss the strengths and limitations of using different indicators to evaluate a program.
2. To conduct simple frequencies.
3. To determine if there are trends in the outcomes using the X ² test of trends
4. To determine if these trends were influenced by any secular events.
5. To  answer the questions and complete the tables included in this exercise.

Description of the Program

The WIN Program (Women's Initiative) is a collaborative grant between a local community based organization and the HRSA foundation. The purpose of the intervention was to improve HIV testing rates among pregnant women in Metropolitan New Orleans. This initiative was a response to the report of the ACTG 076 trial which demonstratd that HIV-infection among infants born to HIV-infected women could be reduced 66% if AZT perinatal prophylaxis was initiated during pregnancy. The intervention consisted of a media campaign (flyers, posters, TV and radio advertisements) to promote testing. In addition, provider education about the protocol was also conducted.

To evaluate the program, three data sources were utilized: 1) street intercept interviews, 2) HIV counseling and testing database, 3) NOAIDS hotline databse. We will use the first data source for this exercise. Street interviews were conducted every six weeks. A quota sample of 15 teenagers (15-18) and 15 adults (19-35) were conducted in the six areas with the highest rate of gonnorrhea (i.e. high risk areas). The program started in 1995 and will be completed in 2000. Interviews were first conducted with women only. Men were included in the interviews after the media message was adapted to target men as well.

For this exercise, we will first examine the study population by summarizing various demographic and behavioral variables. Then we will examine the distributions of demographic variables over the course of the street interviews to look at trends in the demographic makeup of the survey samples. We may adjust the inclusion criteria that we use to subset to the sample that we will include in our analysis. Then we will proceed with an analysis that examines factors associated with the following outcomes: 1) discrepency between self-assessed HIV risk and self-reported HIV risk behaviors, 2) condom use, and 3) self-reported HIV risk behaviors.

Description of the Data 

The data is in SPSS for windows format. The extension SAV is the extension used by SPSS. The following is the coding guide for the database Click here to get the database.

Q1. Conduct frequencies, fill in the table, briefly describe the survey population in a text format.

Table 1. Description of the Study Population (N= )

Q2. What variables have substantial missing data (eg. > 10%)? Do you think this will bias your findings in any way?

Fill in the following table with the percentages and Chi-square test of trends (Chi-square Linear-by-Linear Association p-value).

Table 2. Demographic characteristics of respondent by survey

Q3.  Are there any trends? Do you think this will confound your trends analysis with the outcomes of interest (i.e., % overall tested, % pregnant women tested, % used condoms, % will get tested)?

Let's stop now and decide if we need to change the inclusion criteria 

Q4. Should we eliminate anyone or limit the data set to any time period?

Please fill in the following table using your new inclusion criteria.

Table 3. Factors associated with a discrepant response between self-assessed HIV risk

and HIV risk behavior (N= )

1.00 indicates that this is the reference group.

Q5. Who is at risk for a discrepent response?  Include a discussion of the direction, magnitude, and

significance of the observed effect.

Please use the new criteria and fill in the following table

Table 4. Factors associated with condom use (N= )

Q6.  Who is at risk for not using condoms?

Please use the new criteria and fill in the following table.

Table 5. Factors associated with HIV risk behavior (fill in unadjusted OR with 95% CI)

 Q7.  Who is at risk for each of the above outcomes?

Q8. Summarize your findings from this study.