Behavioural Surveillance Surveys - The Wisdom of Whores
Behavioural Surveillance Surveys - The Wisdom of Whores
Behavioural Surveillance Surveys - The Wisdom of Whores
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Figure 5 : sample size calculations to measure indicators with two dimensions<br />
Example 1<br />
Consider the indicator “proportion having used a condom during the last sexual encounter<br />
with a non-regular partner, <strong>of</strong> those who have had sex with a non-spousal, non-cohabiting<br />
partner in the last 12 months. Using the formula presented above, it might have been<br />
calculated that a sample size <strong>of</strong> n=320 respondents was needed to register a change <strong>of</strong> a<br />
specified size in condom use in irregular sex at the desired levels <strong>of</strong> power and significance.<br />
Data from a recent survey suggest that about 20 percent <strong>of</strong> men and 5 percent <strong>of</strong> women<br />
engaged in such encounters in an earlier 12-month period. <strong>The</strong> number <strong>of</strong> males and females<br />
that would have to be contacted in order to obtain n=320 respondents would thus be<br />
estimated as:<br />
Males: n = 320 /.20 = 1,600<br />
Females: n = 320 /.05 = 6,400<br />
Example 2<br />
For the indicator “proportion using a condom during their last sexual encounter with a sex<br />
worker” among truck drivers, it is thought that most truck drivers in a given setting (75 percent)<br />
have sex with sex workers during any given 12-month period. Thus, if the required sample<br />
size (again calculated as described above) were to be n=275, the number <strong>of</strong> truck drivers that<br />
would have to be contacted would be n = 275 /.75 = 367 per survey round.<br />
In deciding what level <strong>of</strong> change to measure,<br />
surveillance managers should ask themselves<br />
the question: what level <strong>of</strong> change is worth<br />
measuring? In other words, what level <strong>of</strong><br />
change might be deemed to reflect success in<br />
achieving the aims <strong>of</strong> prevention efforts? And<br />
what level <strong>of</strong> change might have an impact on<br />
the risk <strong>of</strong> the spread <strong>of</strong> HIV? <strong>The</strong> doubling <strong>of</strong><br />
condom use during sex with sex workers might<br />
make headlines, but if usage has doubled from<br />
only five percent to 10 percent, it will have<br />
virtually no impact on stemming the potential<br />
spread <strong>of</strong> HIV. Given the very large sample<br />
sizes required to measure small changes with<br />
any degree <strong>of</strong> confidence, it may well not be<br />
deemed worthwhile to register this change.<br />
Some National AIDS Programs have stated<br />
targets for various indicators. For example,<br />
a country might, in its strategic plan, set the<br />
goal <strong>of</strong> increasing condom use in extra-marital<br />
sex by 25 percentage points over five years.<br />
In cases where large changes in indicators are<br />
expected, there is a temptation to set the<br />
magnitude <strong>of</strong> change parameter (P 2<br />
- P 1<br />
) in the<br />
sample size calculations at quite a high level,<br />
thereby decreasing the sample size needed.<br />
It should be recognized, however, that this will<br />
jeopardize the ability to detect smaller changes<br />
that may in fact be programmatically significant<br />
in their own right. By the same token, the cost<br />
<strong>of</strong> measuring smaller changes annually may be<br />
prohibitive, because <strong>of</strong> the large sample sizes<br />
B EHAV I OR A L S U R V EI L L A NC E SURV EY S CHAPTER 4<br />
51