In this study, we compared HIV prevalence estimates from routine PMTCT programme and antenatal HIV surveillance in Addis Ababa with the intention to come up with evidence based recommendations on the potential of PMTCT programme data for antenatal HIV surveillance. The proportion of women testing in PMTCT programme increased markedly, particularly following the introduction of the opt-out HIV testing approach. According to the summary reports, the HIV prevalence among pregnant women in Addis Ababa had shown steady decline. Four years trend data from the five health centres that conducted regular antenatal HIV surveillance also confirmed the decline. The HIV prevalence estimates obtained from the PMTCT programmes and the antenatal surveillance showed consistencies. In conclusion, the routine PMTCT programme report in Addis Ababa can provide comparable estimate with that of antenatal HIV surveillance report and has potential to monitor HIV prevalence trends.
Our findings showed that the proportion of women testing in PMTCT programme increased markedly especially following the introduction of the opt-out HIV testing approach. Consistent with our findings, several studies from resource poor settings including Ethiopia have reported dramatic increases in the proportions of women testing in PMTCT programmes following the WHO policy statement on routine opt-out HIV testing approach in 2004 [10, 17–19]. The data obtained from the different sources reported that the HIV prevalence among pregnant women in Addis Ababa had shown steady decline. This is in agreement with a paper of Mirkuzie et al., where a declining HIV prevalence among antenatal care attendees in Addis Ababa was documented . In the 2011 UNAIDS reports, the HIV epidemic in Ethiopia among 15–24 years old pregnant women show 82% decline in 2010 compared to 2001 .
In our study, the HIV prevalence estimates from the PMTCT programmes and from the antenatal HIV surveillance showed high consistency in 2008 and 2009, where the number of women testing in PMTCT programme increased following the introduction of opt-out testing approach. Similar reports were documented in Cameroon with 69% of women testing in PMTCT programme, comparable HIV prevalence estimates were reported i.e. 7.8% from PMTCT programme and 7.3% from antenatal surveillance. Also, in Botswana when HIV testing in PMTCT programme was >95% between 2005 and 2007, the HIV prevalence estimates from PMTCT programme and antenatal surveillance were similar . In India with 68% rate of HIV testing in PMTCT programme no statistically significant difference was observed between the two estimates . In our study, when the rate of HIV testing in PMTCT programme was lowest; 33% in 2005 there appeared to be disparities between the two estimates, 8.8% in the PMTCT programme and 12.1% in the antenatal surveillance (Figure 1). Using the 2002 and 2003 PMTCT reports in Uganda, Mpairwe and colleagues also reported higher HIV prevalence from PMTCT programme than antenatal surveillance with less than 70% of the antenatal attendees testing for HIV in the PMTCT programme but no differences with 70% or more of the antenatal attendees testing .
Marsh et al. have analysed papers from seven sub-Sahara African countries; when the rate of HIV testing is low the possibility of participation bias in PMTCT programmes is high due to perception of risks and differences in demographic characteristics. Participation bias can also happen if the PMTCT programme is newly established as high risk cases are more likely to take advantage of the HIV testing . In general, participation bias is less likely where routine opt-out HIV approach is properly implemented. Routine antenatal HIV testing has been implemented in Addis Ababa since early 2008 and accompanied by significant improvement in the rate of HIV testing in PMTCT programmes, and in other resource poor settings [10, 17, 18]. Marsh and colleagues also assessed the comparability of the two estimates; papers from Cameroon and Botswana asserted the comparability of the two estimates and suggesting its potential utility for HIV monitoring. In contrast, although the overall HIV prevalence estimates appear to be similar in Kenya, Uganda and Zimbabwe there are large differences at clinic level and are discouraging the replacement of PMTCT programme estimate for the antenatal surveillance estimate . Consistent with the latter observations, a report from 43 dual PMTCT and surveillance sites in Ethiopia showed comparable overall HIV prevalence estimates but large differences at clinic level .
Although all these studies acknowledged the potentials of the PMTCT programme data for antenatal HIV surveillance purposes, they all agreed that currently its quality is not good enough to be used for surveillance. The major concerns are lack of uniform registers, reporting formats and reporting date and missing or inconsistent training among those who are doing the routine reporting. Issues related to HIV testing include lack of standard testing algorithm, unreliable supply of test kits and other logistics, variable training among professionals who are doing the testing and lack of quality control system [12, 21]. The lack of international consensus on how to make routine PMTCT reports suitable for antenatal surveillance is another concern .
As the HIV epidemic is maturing and is taking a chronic course, it is imperative to revisit existing strategies to maximize efficiency of the control programme and to direct resources where it is most needed. Larson and colleagues in their paper highlighted the need for strategic responses, including integration of AIDS intervention with other relevant health programmes, and the need for regular systematic programme monitoring and evaluation . Currently, the Global Health Initiative (GHI) is investing on health system strengthening for optimal health gain and is supporting the Ethiopian government . Being one of the building blocks of the health system, improving monitoring and evaluation of programme activities are emphasized.
For optimal monitoring and evaluation, Health Management Information System (HMIS) has been implemented in Addis Ababa for routine programme reporting . The HMIS use standard computerized registers and reporting system and it can be an asset for standardizing the PMTCT reports. Even before the scaling up of the HMIS across the city, the quality of the PMTCT reports seem to be good; over 90% the reports from the sub-cities were complete. Moreover, standard antenatal HIV testing algorithm recommended in the 2007 Ethiopian PMTCT guidelines has been used in Addis Ababa, and could help to minimize potential inconsistencies at the level of sample collection and processing .
In line with these arguments, Addis Ababa seems to quality to use routine PMTCT programme reports for surveillance purpose and can be a pilot site for Ethiopia. Nonetheless, continuing the regular antenatal HIV surveillance in parallel for some time could help to validate the quality of PMTCT programme data until it gets perfected. Meanwhile, issues related to supply of test kits and logistics, particularly the need for continuous in-service training to minimize gaps due to staff turnover need to be emphasized.
Use of summary reports and the lack of cost effectiveness analysis would be some major limitations of the study. However, to ensure quality, the PMTCT reports obtained from the sub-cities were validated with reports from the City Council Health Bureau and from Intra-Health and they were consistent. For ensuring data completeness, missing monthly reports were collected from service outlets. The study could have benefited from cost effectiveness analysis to show how much saving the health system can make when antenatal surveillance is phased out and investment is diverted for generating quality PMTCT programme reports. We used summary reports from PMTCT reporting formats aggregated at the sub-city level and summarized HIV prevalence reports, and hence we lack individual level and clinic based data to assess differences in HIV prevalence by socio-demographic variables and to assess differences among clinics.