|Year : 2020 | Volume
| Issue : 3 | Page : 248-251
Non-participation of female sex workers in HIV sentinel surveillance 2017 in the central zone, and its effect on observed HIV prevalence rate
Shashi Kant1, Ayush Lohiya2, Sanjay Kumar Rai1, Puneet Misra1, S Venkatesh3, Research Team*4, 4
1 Professor, Centre for Community Medicine, All India Institute of Medical Sciences, New Delhi, India
2 Assistant Professor, Department of Public Health, Super Speciality Cancer Institute, Lucknow, Uttar Pradesh, India
3 Directorate General of Health Services, Ministry of Health and Family Welfare, Government of India, Lucknow, Uttar Pradesh, India
|Date of Submission||26-May-2019|
|Date of Decision||24-Aug-2019|
|Date of Acceptance||18-Jun-2020|
|Date of Web Publication||22-Sep-2020|
Super Speciality Cancer Institute, Lucknow, Uttar Pradesh
Source of Support: None, Conflict of Interest: None
| Abstract|| |
Background: During HIV sentinel surveillance (HSS) 2017 round, the sampling strategy to recruit female sex workers (FSWs) was changed from consecutive to random sampling. This may affect the participation and HIV positivity rates among FSWs. Objective: The objective of this study is to estimate the nonparticipation rates among FSWs and its effect on the observed HIV prevalence rate during HSS-2017. Methods: The data were collected from FSW sentinel sites located in the states of Delhi, Jharkhand, Uttar Pradesh, and Uttarakhand (Central Zone). The HIV positivity rate among FSWs who participated in HSS-2017 was compared with the HIV positivity rate of those who did not participate. HIV status of the participants was obtained from HSS-2017 data. The master list of participating targeted intervention sites was accessed to obtain the last known HIV status of the eligible nonparticipants. Results: Nonparticipation rate of FSWs from the central zone during HSS2017 was 10.8%. The HIV positivity rate among nonparticipant FSW was four times and six times higher in Delhi and UP, respectively. Conclusion: Selective nonparticipation of eligible FSWs might have led to the underestimation of the HIV positivity rate in the central zone during the HSS-2017 round.
Keywords: Female sex workers, HIV sentinel surveillance, HIV
|How to cite this article:|
Kant S, Lohiya A, Rai SK, Misra P, Venkatesh S, Research Team*. Non-participation of female sex workers in HIV sentinel surveillance 2017 in the central zone, and its effect on observed HIV prevalence rate. Indian J Public Health 2020;64:248-51
|How to cite this URL:|
Kant S, Lohiya A, Rai SK, Misra P, Venkatesh S, Research Team*. Non-participation of female sex workers in HIV sentinel surveillance 2017 in the central zone, and its effect on observed HIV prevalence rate. Indian J Public Health [serial online] 2020 [cited 2020 Oct 21];64:248-51. Available from: https://www.ijph.in/text.asp?2020/64/3/248/295786
4Shobini Rajan, Assistant Director General (Blood Safety),
5Pradeep Kumar, Programme Officer (Surveillance), National AIDS Control Organisation, Ministry of Health and Family Welfare, Government of India, New Delhi,
6Farhad Ahamed, Assistant Professor, Community and Family Medicine, All India Institute of Medical Sciences, Kalyani, West Bengal, India.
| Introduction|| |
National policymaking requires information on the prevalence and trend of HIV infection. For countries having a low and concentrated epidemic, this information is obtained from key high-risk groups (HRGs) and from the general population. In India, the HIV epidemic is concentrated, i.e., most of the HIV infection and transmission is concentrated among persons with high-risk behavior. To gather information on the prevalence and trend of HIV infection, India started HIV sentinel surveillance (HSS) among HRGs. Female sex workers (FSWs) are one of the HRGs and are an important driver of the HIV epidemic in India.
In the initial years, sentinel surveillance among HRG was facility-based consecutive sampling.,, In consecutive sampling, all individuals who attended sentinel sites after initiation of HSS were assessed consecutively for eligibility and recruited in the order they attended the site. Consecutiveness was maintained throughout the process of enrollment, eligibility, and administration of consent. Consecutive sampling removes the chances of selection or exclusion on the basis of individual preferences. It is convenient and easy to follow as well. However, in the year 2017, the sampling strategy was changed from consecutive to random sampling. Sentinel sites were provided a randomly drawn list of registered FSWs for inclusion in HSS. These randomly selected FSWs were assessed for eligibility and then recruited in HSS-2017 after seeking the consent of the eligible FSWs. This procedure was thought to provide better representativeness, and hence a more accurate estimate of HIV prevalence. The random sampling strategy for selection of FSWs was first tested in 2009 and then extended to many states in 2010–2011.
In consecutive sampling, the FSWs attending sentinel sites only were eligible for inclusion. On the other hand, in random sampling, all registered FSWs were eligible. Therefore, this change in sampling strategy could potentially affect the participation rate in sentinel surveillance, thereby impacting the observed prevalence and trend in HIV infection. Although the sample size for HSS is calculated after adjusting for nonparticipation, we wanted to assess the quantum of nonparticipation and its effect on HIV prevalence, if any. Hence, we undertook this study to assess the quantum of nonparticipation of selected FSWs in sentinel surveillance; and its impact on the observed HIV prevalence rate.
| Materials and Methods|| |
For the supervision of surveillance activities, all the states of India were divided into six zones: Central, East, North, North-East, South, and West. Each zone is supervised by one or more selected Regional Institutes (RI). The FSW sentinel sites located in five states of India (Bihar, Delhi, Jharkhand, Uttar Pradesh, and Uttarakhand) that constituted the central zone of HSS were eligible to be included in the study. Bihar state was excluded due to operational reasons. The sentinel sites were located at targeted intervention (TI) facilities run by nongovernmental organizations.
The study period was from April to July 2017. Each sentinel site was expected to collect blood specimens and fill the interview schedule of 250 FSWs, which contained demographic information and HIV-related risk behavior. TI sites maintained a computerized listing of high-risk individuals (HRIs) ever contacted and registered at the TI project. This list was called as “master list.” From the State AIDS Control Society (SACS), each TI site received a random list of 250 HRIs drawn randomly from the master list. Peer Educator (PE) made an attempt to contact HRI in the list. The PE accompanied the contacted HRI to the sentinel site where details of the HRI were entered in HSS register, and eligibility for HSS was assessed. In case of refusal to participate or inability of the PE to contact HRI even after three attempts, the selected HRI was labeled as nonparticipants.
For this study, data were obtained from two sources. (i) Random list of HRI and data form with HIV test results were obtained from the RI, AIIMS, New Delhi. (ii) Data regarding the participation of selected HRI as recorded in the HSS register and master list were obtained from sentinel sites through SACS. Information-related to HIV status as of April 2017, participation in HSS, and adherence to the random list were extracted from data sources listed above [Figure 1].
Operational definition: (i) Nonparticipants: selected HRIs who could not be contacted even after three separate attempts were labeled as “noncontactable.” Those HRI who were contacted but did not agree to participate in HSS were labeled as “refused.” The details of noncontactable and those who refused (along with the reason for refusal) were recorded in the HSS register. Noncontactable and refused together constituted “nonparticipants.”
For the purpose of quality control, double data entry, and matching of data form was performed. The quality control was performed by cross-checking 10% of data forms with the data entered in the computer and vice versa. All HIV positive and 2% of HIV negative samples were sent by the State Reference Laboratory to the National Reference Laboratory at the National AIDS Research Institute, Pune for cross-checking as a quality control mechanism. All data forms with HIV positive results were cross-checked with data entered in the computer as part of quality check.
The data obtained from the random list of HRI and HSS registers were merged and matched to obtain a list of participants and their HIV status. Merging and statistical analysis were performed using STATA-12 software (StataCorp. 2011. Stata Statistical Software: Release 12. College Station, TX, USA: StataCorp LP). All quantitative variables were expressed in terms of percentage with a 95% confidence interval (CI). Value of P < 0.05 was considered statistically significant.
The written permission to use the data set was obtained from the National AIDS Control Organization, Government of India. Personal identifiers were stripped off the dataset to maintain anonymity.
| Results|| |
The number of FSW sites included in the analysis from the states of Delhi, Jharkhand, UP, and Uttarakhand were 4, 9, 11, and 2, respectively. Nonparticipation rate was highest for the state of Jharkhand (15.5%), followed by Uttarakhand (12.6%), and UP (8.7%). Delhi state had the lowest nonparticipation (4.8%) among the four states [Table 1].
In Delhi state, the HIV positivity rate among nonparticipants (6.0%) was almost four times that of participants (1.6%) (P = 0.02). In Jharkhand state, the HIV positivityrate was similar among participants (0.3%) and nonparticipants (0.3%). The HIV positivity rate among FSWs was six times more among nonparticipants (1.2%) as compared to participants (0.2%) in UP state (P = 0.009). Similarly, in Uttarakhand, the HIV positivity was more among nonparticipants (5.6%) as compared to participants (0.0%). This difference was statistically significant (P < 0.001). Overall, the HIV positivity rate among nonparticipants (1.5%) was significantly higher (P < 0.001) compared to the participants (0.4%) [Table 2].
The HIV prevalence among the original 250 FSWs in the random list was 0.6% which was lower than the actual observed prevalence of 0.4%. This difference was not statistically significant.
| Discussion|| |
This study was done to assess the effect of nonparticipation during the HSS-2017 round on the observed HIV prevalence among FSWs of four states of the central zone.
Nonparticipation rate among FSWs was highest for the state of Jharkhand and it was least for the state of Delhi. There is a need to find out the reasons for the high nonparticipation rate. One possible reason for nonparticipation could be that FSWs who knew about their HIV positive status did not see any value in getting tested again. Selective nonparticipation based on any specific factor could alter the observed HIV prevalence rate. Hence, it is important to know about the reasons for nonparticipation and address it in the next round of HSS to have more robust results.
HIV prevalence was significantly higher among nonparticipants than participants in three of four states included in the study. Overall, the HIV prevalence rate among nonparticipants was four times higher compared to the participants. The probable reason for this could be that individuals who knew their HIV positive status selectively refused to participate. The four-time difference in prevalence rate between participants and nonparticipants is a cause for concern. It directly challenges the main objective of change in sampling methodology i.e., representativeness. A study done in South India had reported that nonparticipation affects HIV prevalence. However, this study was done among the general population. Similarly, a study conducted among antenatal patients in Zambia also reported that HIV prevalence is affected by the participation rates.,
We also tried to see the effect of nonparticipation on the overall HIV prevalence among FSWs. The HIV prevalence rate among total FSWs (participants and nonparticipants) was more than the observed rate, but the difference was statistically not significant. Therefore, the true value of HIV prevalence will be within the 95% of bounds of CI of the observed value. Hence, the findings can be used for monitoring trends in HIV prevalence over time.
However, the same cannot be said for the HIV/AIDS indicator derived from the HSS-2017 data for FSW in the central zone. The lower observed HIV prevalence rate among participants suggests that we are underestimating the absolute number of HIV positive FSW by approximately 33% (0.4% vs. 0.6%) in the central zone.
This is the first step to externally assess the quality of data collected during HSS. The assessment was done quantitatively, which can now be used for comparison in future. Due to logistic reasons, Bihar state was excluded impacting the completeness of assessment for central zone as one unit.
| Conclusion|| |
Differential participation rates based on known HIV status poses a serious challenge to the representativeness of the observed finding. The change in sampling strategy in the HSS-2017 round can lead to under-estimate of absolute numbers of HIV infected FSW in the Central Zone. However, the data are valid for monitoring the trend of HIV prevalence over time.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Diaz T, De Cock K, Brown T, Ghys PD, Boerma JT. New strategies for HIV surveillance in resource-constrained settings: An overview. AIDS 2005;19 Suppl 2:S1-8.
Paranjape RS, Challacombe SJ. HIV/AIDS in India: An overview of the Indian epidemic. Oral Dis 2016;22 Suppl 1:10-4.
Michelo C, Sandøy I, Fylkesnes K. Antenatal clinic HIV data found to underestimate actual prevalence declines: Evidence from Zambia. Trop Med Int Health 2008;13:171-9.
Dandona L, Lakshmi V, Sudha T, Kumar GA, Dandona R. A population-based study of human immunodeficiency virus in south India reveals major differences from sentinel surveillance-based estimates. BMC Med 2006;4:31.
[Table 1], [Table 2]