Issues of interpretation

Cross-country comparisons and assessment of trends over time are possible if the same to O

methodology is consistently adopted for data collection.

Common pitfalls Representativeness

Conclusions on syphilis prevalence in the general population based on sample surveys of pregnant women attending antenatal clinics should be made with caution. First, the numerator is not representative of all pregnant and non-pregnant women. Second, the sample is not necessarily representative of all pregnant women, only of women who choose to attend for antenatal care. Third, syphilis serology may not correlate directly with STI prevalence. STIs are a major cause of infertility in developing countries, and infertile women are not effectively accessible through antenatal care. This may lead to an underestimation of the STI prevalence in all women. Conversely, since non-pregnant women include those who are not sexually active and therefore are not at risk of STI, the prevalence among pregnant women may be an overestimation of the prevalence in all women. Moreover, it is representative of this group of women only where all pregnant women are screened and not just those judged by medical professionals to be at high risk.

Another limitation to the representativeness of the indicator could be the contribution of private, semi-private or non-health-ministry public sector services (e.g. armed forces health services, university hospitals, social security hospitals) to overall antenatal care. Where their contribution to antenatal care is large, access to their data would be important as they may represent different population groups.


Disaggregation of data will ensure that important differentials between geographical areas and social groups are acknowledged and the issues of equity addressed. Nevertheless, this may result in inadequate sample sizes and differentials across groups may not be apparent. The main objectives of syphilis surveillance, such as to monitor trends in specific age groups and geographical areas, should therefore be clearly defined in advance and the sample size calculated accordingly.

Was this article helpful?

0 0

Post a comment