|Year : 2019 | Volume
| Issue : 4 | Page : 305-312
Evaluation of key survey components of bio-behavioral surveillance among HIV high-risk subpopulation in Western India
Radhika G Brahme1, Sheela V Godbole2, Raman R Gangakhedkar3, Kuldeep Singh Sachdeva4, Vinita Verma5, Arun R Risbud6
1 Scientist C, Epidemiology and Biostatistics, ICMR-National AIDS Research Institute; PhD Scholar, Faculty of Health and Biological Science, Symbiosis International University, Pune, Maharashtra, India
2 Scientist F, Epidemiology and Biostatistics, ICMR-National AIDS Research Institute, Pune, Maharashtra, India
3 Scientist G, Epidemiology and Biostatistics, ICMR-National AIDS Research Institute, Pune, Maharashtra, India
4 Deputy Director General, Strategic Information Division, National AIDS Control Organization, New Delhi, India
5 Consultant Research, Strategic Information Division, National AIDS Control Organization, New Delhi, India
6 Former Scientist G, Epidemiology and Biostatistics, ICMR-National AIDS Research Institute, Pune, Maharashtra, India
|Date of Web Publication||18-Dec-2019|
Dr. Radhika G Brahme
National AIDS Research Institute, Post Box: 1895, 73-G Block, Midc Bhosari, Pune - 411 026, Maharashtra
Source of Support: None, Conflict of Interest: None
| Abstract|| |
Background: High-quality data are of prime importance in any health survey because survey data are considered as a gold standard for nationally representative data. The quality of data collection largely depends on the design of the questionnaire, training, and skills of the interviewer. Objectives: In the present study, we tried to evaluate three key components, such as questionnaire design, human resource and training of the field staff for Integrated Biological and Behavioural Surveillance carried out among the HIV high-risk subpopulation. Methods: A mixed-methods approach was used. Qualitative and quantitative data collection was carried out in the year 2015 with cross-sectional survey design in western states of India. The in-depth interviews of 10 stakeholders, structured interviews of the survey respondents (n = 560), and field investigators (n = 71) were conducted. Data triangulation was used to find out the concurrence of the qualitative and quantitative data. Results: Comprehensive and standardized survey questionnaire, structured training agenda, and strategic preparation for recruiting human resources were the overall strengths of the survey. However, during the implementation of the survey, there were some difficulties reported in data collection process. Overall, the respondents and investigators felt that the questionnaire was long and exhaustive. Difficulties were faced while collecting data on sexual history. The field staffs were not adequately experienced to work with sensitive population. Conclusions: In order to have accurate, reliable data, especially on sexual behavior; emphasis should be given on simple questionnaire with the use of community-friendly language, skilled and experienced interviewers for data collection, and extensive field training.
Keywords: Data quality, HIV high-risk population, questionnaire design, survey data
|How to cite this article:|
Brahme RG, Godbole SV, Gangakhedkar RR, Sachdeva KS, Verma V, Risbud AR. Evaluation of key survey components of bio-behavioral surveillance among HIV high-risk subpopulation in Western India. Indian J Public Health 2019;63:305-12
|How to cite this URL:|
Brahme RG, Godbole SV, Gangakhedkar RR, Sachdeva KS, Verma V, Risbud AR. Evaluation of key survey components of bio-behavioral surveillance among HIV high-risk subpopulation in Western India. Indian J Public Health [serial online] 2019 [cited 2020 Jul 9];63:305-12. Available from: http://www.ijph.in/text.asp?2019/63/4/305/273352
| Introduction|| |
India has the third-largest burden of HIV in the world  The HIV epidemic in India is concentrated in high-risk subpopulation such as female sex workers (FSWs), men having sex with men (MSM), transgender (TG), and intravenous drug users (IDUs). Hence, it is important to understand the dynamics of the sexual behavior of this key population. The data on sexual behavior are used to do the risk assessment, and the strategies for interventions are planned accordingly.
The data collected in public health activities are used to monitor trends in health patterns and also to determine the development of public health policy. There are three major indicators of data quality, such as representativeness, completeness, and accuracy. Data quality largely depends on the instrument used for data collection along with training and skills of the interviewer. The soundness of public policy developed from evidence using questionnaire-based research also depends on the questionnaire administration.
Sexual behavior is of private and of sensitive nature. Therefore, getting accurate information from the respondent about the sexual history continues to be challenging and requires certain interviewing skills. The accuracy of behavioral data related to risk for HIV and other sexually transmitted infections is prone to misreporting because of social desirability effects. There were reports that the fidelity of sexual behavior data obtained by self-report has been questioned. Self-reports of sexual behavior were unreliable due to multiple sources of bias, including under-reports of stigmatized behaviors and over-reports of normative behaviors.,
The National AIDS Control Organization (NACO) carried out the National Integrated Biological and Behavioral Surveillance (IBBS) in India as a part of the second generation HIV surveillance activity in the high-risk group subpopulation. In the present study, we tried to evaluate three major components of this survey, such as design of the questionnaire, human resources, and training, which may have an impact on data quality.
| Materials and Methods|| |
During 2014–2015, IBBS was carried out by NACO in 31 states and 271 districts in India with a planned sample size of 138,400 (400 per district/domain) to find out the HIV prevalence and the risk behaviors of HIV-high risk subpopulation. IBBS was a community-based unlinked anonymous cross-sectional survey of probability-based sampling. The survey was carried out among HIV-high risk population of FSWs, MSM, IDUs, TGs, migrants, and currently married women.
In this survey, behavioral data were collected using computer-assisted personal interviewing (CAPI) devices. For HIV testing, the blood samples were collected as the dried blood spot.
The questionnaires were designed in 16 Indian languages and had more than 100 structured questions for each typology. The field activity, including data and sample collection, was outsourced to field research agency (FRA). The field staff was appointed by FRA, and the software was developed by a separate technology partner. The training was provided to the field staff by FRA, experts from regional institutes (RI) and NACO.
The present analysis is a part of a substudy, “Data Quality Assessment and Process Evaluation” which was carried out in three states of western India – Maharashtra, Goa, and Gujarat between May and December 2015. Since the regional institute was involved in the monitoring of the survey in these states, the substudy was conducted in these states. The survey was started simultaneously in majority of the domains, and hence, it was not feasible to do visits in all these domains at the same time for data collection. Hence, for this substudy, 27 domains representative of all the typologies were considered. Of 27 domains, 21 were from Maharashtra, 5 were from Goa, and 1 from Gujarat. The data collection was performed during the 2 days of monitoring visits to these domains. During these 2 days, the IBBS participants who were enrolled in the main survey participated in the exit interview of the substudy. The convenient sampling method was used considering the feasibility and exploratory nature of conducting these interviews in the field. The random 5% of respondents of each domain (out of total sample size of 400) were interviewed. Thus, the sample size of minimum 540 was considered for exit interviews of the respondents. However, we were able to take some more interviews, and hence, a total of 560 exit interviews were conducted in 27 domains. The sociodemographic characteristics of the individuals who participated in “exit interviews” were compared with the remaining ones, and it was observed that there was no significant difference in both the populations. Hence, we do not expect the variations even if the sample size would have been more than 5%. In addition, the interviews of the field investigators (FI) were taken who were involved in survey data collection. There were two field teams having four FI at each domain. Of these, 4 FIs at each domain, minimum two interviews of investigators were planned. Hence, the sample size for FI interview was 54; however, few additional interviews were conducted with the total of 71 interviews.
Ten in-depth interviews of the stakeholders, who worked on various expert committees in IBBS for planning and implementation and representatives of FRA, were conducted either personally or telephonically based on the feasibility to approach with them. Criteria-based purposive sampling method was used for in-depth interviews. The sample size for in-depth interviews was ten because it covers all the program elements of the study. The information saturation was started when eight interviews were completed. The in-depth interviews of the key stakeholders were conducted after completion of the survey. The qualitative data of interviews were transcribed and translated, and themes were formulated. Data triangulation method was used to do the synthesis of the qualitative and quantitative data.
The substudy of “Data quality assessment and process evaluation” was approved by the NARI Ethics committee (Ref. protocol no. 2013–2012, dated December 4, 2013). The written informed consent in four languages was approved by NARI-EC on April 28, 2015. The written informed consents were obtained from the IBBS study participants for taking the exit interviews. Verbal informed consent was obtained for in-depth interviews from the stakeholders and FI.
| Results|| |
The evaluation of the three major components of the survey is presented below. Based on the themes emerged from the qualitative and quantitative analysis, the findings are represented with subthemes. The interpretation of the qualitative data is presented in [Table 1].
Structure of the questionnaire
The structure of the questionnaire was standard and time-tested for such surveys, and the same was mentioned in the in-depth interview.
”The existing pattern of having different sections for different types of partner to capture the number of contacts and condom use was correct because it is easier for the respondent to recollect the history of contact with individual partner and will result in more accurate data.”
#KII-1 (Questionnaire design team)
Majority of the stakeholders felt that the questionnaires were well designed and encompassed all aspects of high-risk behavior, however, were very exhaustive. In the quantitative analysis, 16 out of 71 (22%) investigators and 23% of respondents reported that the questionnaire was lengthy. Repetitive nature of questions was reported by 17% of FIs [Table 2].
”In sexual history, there were repeated questions on sexual acts, condom use etc., about regular, commercial, casual, and occasional partners. This elicits the detailed sexual history, however there were time constraints to explain this and may have created confusions in respondent's minds.”
#KII-8, (Project coordinator, RI)
Translation and language of the questionnaire
The translation of the questionnaire was done by the experts in local languages and was field-tested at certain regions. However, concerns were reported about the language by 13% of respondents. In Mumbai, there were many migrated FSWs from Karnataka, Tamil Nadu, and West Bengal, who were unable to understand Marathi or Hindi. Similarly, in Goa, Konkani was the most common language used by FSWs, but the interviews were conducted in Marathi.
”When we are dealing with a subpopulation such as FSWs, who may not be local resident of the area and are in-migrated from other regions, the questionnaire in a local language may not be useful in complete understanding of the questions.”
#KII-6 (Senior Research Officer, RI)
MSM and TG communities use specific terminologies for their sexual orientation. The words used by the investigators were theoretical translated words. Around 6% investigators and 9% respondents reported the problem of translation and language of the questionnaire and 6% investigators suggested the use of simple and community-friendly language.
”Most of the MSM here are of “kothi” typology. Their language is different.”
#KII-4, 9, 10 (Representative from SAC and TI)
The questions related to “Anal sex” were difficult to ask and sensitive as per the FI as well as respondents (13% and 16%) in case of FSWs. There was a problem of understanding the questions on HIV knowledge by the respondents (38% overall and 61% in IDUs). Among IDUs, 36% showed a lack of understanding for questions on sexually transmitted diseases. Overall, 53% of the respondents expressed hesitation in answering the questions on sexual history. Some respondents expressed difficulty to give accurate information due to recall bias (21% overall and 41% in TGs) [Table 2], additional data not shown in table].
”Questions were not difficult but the framing of the questions was not appropriate.”
#KII-4 (SAC Representative)
Expertise and experience
Interviews of the stakeholders reflected that the investigator's younger age, the inadequate experience of working in such communities, cross-gender, and attrition of the staff during the survey posed difficulties in getting accurate and reliable data. One of the FSWs mentioned that the girl taking interview was very young, and hence, she had hesitation to answer the questions on sexual behavior.
Staffing structure and responsibilities
Overlap of roles of higher authorities, inadequate IT experts at the field to resolve the CAPI problems and lack of communication between various agencies were the concerns reported in in-depth interviews.
”Sometime there were problems in the system of reimbursements of the field expenditure to the field staff.”
#KII-5 (Assistant Project Leader-FRA)
Training of the field staff
Structure and timelines
Overall, the structure of the training was satisfactory with respect to expertise, arrangements, and duration of the training.
”There was a good blend of trainers from RI, SACS, NACO, field experts and experts from other agencies; so that way the training was a well-planned agenda.”
#KII-8 (Project Coordinator)
However, the feedback from the interviews reflected that the hands-on training on CAPI and field activity was insufficient. The feedback of the FI about the training was as follows (Training on–Questionnaire, CAPI and Field was rated as Extremely useful [29%, 15%, 35% respectively], Useful [63%, 55%, 53%], Average [11%, 20%, 4%], Not much useful [0%, 8%, 0%] and Not useful at all [0%, 0%, 2%]).
”When the survey started, especially, in the second phase, a lot of people left and new people joined and they were not trained the way we provided training initially.”
#KII-3, (Management and IT expert)
A three-tier training was planned. In the first phase, all the top authorities from participating organizations were trained, this was followed by training of RI, FRA, and supervisors, and then, they became the trainers for the training of the hired field staff. However, the resource material of training was same for all three phases.
”Understanding ability of person attending the TOT training and of grass root worker is different. So if the session on TOT is for one hour, for field training it should be for three hours.”
#KII-7 (Senior Research Officer, RI)
To summarize, a data triangulation matrix was developed. There was concordance in findings of both these methods for majority of the domains derived from different sources of data used in this analysis [Table 3].
| Discussion|| |
Understanding of human sexual behaviors is important to study the characteristics of the high-risk behavior as these are the predictors of getting HIV risk in high-risk subpopulation. Appropriate methods of data collection are crucial to get the valid and reliable data on sexual behavior as the accuracy of self-reporting sexual risk behaviors may be susceptible to misreporting.
The overall strength of IBBS was the involvement of various national and international experts in the planning of the study. This included formulating various expert committees for designing the tools and software, phase-wise training, use of effective technology for easy data capture (CAPI), and role-based information management system. Recruiting large number of staff within a short time was one more challenge, but the FRA managed to do it in a timely manner as per the guidelines provided by NACO. The survey questionnaires were standard and widely used in many countries for HIV surveillance activity. There was a well-planned three-level training program with the involvement of various expertise.
Challenges in implementation of survey
The feedback given by the respondents on questionnaire design, contents, and procedures were different in different typologies as the level of understanding, and investigators were different at different domains.
In the survey questionnaire, the set of the same questions was repeated to get the sexual history for every type of sexual partner which may have been misunderstood by the respondents that those were “repeat questions.” The length of the questionnaire, the use of community-specific terminologies by respondents, and language were other reported concerns. Another problem was “recall biases” when multiple questions of multiple time points were asked about exact numbers of sex acts, condom use, and drug doses. All these problems may reflect in more number of “non-response” (Do not know/remember).
Similar observations were made in different studies in other countries. In Jamaica health survey, the highest level of item nonresponse was on questions related to income and sexual activity. A meta-analysis of 28 studies investigated that shorter periods were found to be more reliable for recall of drug-use behaviors. In a Tanzanian study, some participants changed their responses when asked a sensitive question a second time, such as a number of lifetime sexual partners which reflects either a recall bias or social desirability toward providing the answers.
It would be easier to elicit the sexual history in a tabular format, which helps the interviewer to understand the sexual behavior comprehensively and will be easier for the respondent to recollect and reproduce at one stroke. The participation of community peers in the training process is important. It is difficult to get in-depth knowledge of sexual behavior in such survey-based model having time and privacy constraints; hence, we recommend having additional qualitative studies on smaller groups of individuals subsequent to survey to understand the in-depth dynamics of sexual practices. Questions on “Anal sex” in case of FSWs should be reworded for better understanding. Investigators should be aware of the community's language and terminologies to explain the questions properly if required. We observed more problems when the questions were “how many.” Detailed quantified history about sex acts and condom use and drug use can be reframed. The questions could be designed to get a broad idea about the orientation, practices of safer sex, and recent exposure rather than quantifying in exact numbers.
Irrespective of educational qualifications, it is important to understand the sensitivity while dealing with such population, maturity while asking the questions and importance of the data in research. A study of Poulin suggests that in places where a formal interview with a stranger especially, about sexual behavior, data quality may be enhanced by developing interviewer's rapport skills  The adequate financial remuneration and proper systems to ensure timely availability of funds for field expenses is crucial to retain the field staff.
It was also reported that respondents were hesitant to answer the questions in front of the opposite gender or younger age investigators. The effect of cross-gender was also reported in the other studies that women interviewed by males were considerably less likely to report sexual intercourse compared to females; and moreover, the likelihood of reporting sexual intercourse was greater for women interviewed by a married individual. Similar recommendations were also made in another study of matching respondents and investigators on gender and the use of more supportive wording in sexual behavior questions as per comfort level of the community.
The training of IBBS was structured and well-planned. However, the timelines between the trainings, dropouts of attendees at the time of training, and inadequate time provided for field and CAPI training were the major lacunas reported in this study.
Majority of the recruits were not exposed to working conditions in such communities, and hence, the emphasis should have been given to the field training, mocks, and discussion with community peers to understand the sensitivity, confidentiality, and ethical issues, especially when we are dealing with home-based FSWs. A study among FSWs from Andhra Pradesh indicated that the consent process, staff gender, behavior, study environment, survey content, time requirements, and perceived FSW community support for research were key factors when they participate in the study.
The training agenda should be different for planners and ground-level field staff based on the experience and ability of the trainees. More innovation is required in terms of the structure of field training. Refresher's training should be arranged periodically to ensure that the staff do not deviate from decided procedures, and training should be a prerequisite for new recruits.
| Conclusions|| |
The HIV surveillance methodologies are sound, structured, and time tested in India and also in other developing countries. The questionnaires and trainings are designed accordingly. However, in developing countries, there are growing number of studies that have assessed bias in sexual behavior data., In order to have accurate, reliable data and lesser nonresponses, special emphasis is required to be given on simple and innovative design of the questionnaire, selection of appropriate skilled and experienced field staff, and extensive field-based training.
We would like to acknowledge the support of Dr. Neeraj Dhingra, Dr. Yujwal Raj, and Dr. Pradeep Kumar from NACO for providing permission to conduct the substudy along with IBBS. The primary author is Ph.D. scholar of Symbiosis International University (SIU), Pune, and we acknowledge the support provided by the SIU. We thank Dr. Nikhil Gupte from CTU, BJ medical college, Pune, and Dr. Amit Lokhande, Mrs. Neelam Joglekar, Ms. Sucheta Deshpande, Mr. Rajesh Yadav who were part of NARI IBBS team for providing the inputs for data collection and analysis. We sincerely thank the stakeholders and study respondents for their participation and sharing their feedback.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Mirzaei M, Ahmadi K, Saadat SH, Ramezani MA. Instruments of high risk sexual behavior assessment: A systematic review. Mater Sociomed 2016;28:46-50.
Chen H, Hailey D, Wang N, Yu P. A review of data quality assessment methods for public health information systems. Int J Environ Res Public Health 2014;11:5170-207.
Bowling A. Mode of questionnaire administration can have serious effects on data quality. J Public Health (Oxf) 2005;27:281-91.
Fenton KA, Johnson AM, McManus S, Erens B. Measuring sexual behaviour: Methodological challenges in survey research. Sex Transm Infect 2001;77:84-92.
Kelly CA, Soler-Hampejsek E, Mensch BS, Hewett PC. Social desirability bias in sexual behavior reporting: Evidence from an interview mode experiment in rural Malawi. Int Perspect Sex Reprod Health 2013;39:14-21.
Brener ND, Billy JO, Grady WR. Assessment of factors affecting the validity of self-reported health-risk behavior among adolescents: Evidence from the scientific literature. J Adolesc Health 2003;33:436-57.
Kinsey AC, Pomeroy WR, Martin CE. Sexual behavior in the human male 1948. Am J Public Health 2003;93:894-8.
Ramesh BM, Moses S, Washington R, Isac S, Mohapatra B, Mahagaonkar SB, et al.
Determinants of HIV prevalence among female sex workers in four South Indian states: Analysis of cross-sectional surveys in twenty-three districts. AIDS 2008;22 Suppl 5:S35-44.
Phillips AE, Gomez GB, Boily MC, Garnett GP. A systematic review and meta-analysis of quantitative interviewing tools to investigate self-reported HIV and STI associated behaviours in low- and middle-income countries. Int J Epidemiol 2010;39:1541-55.
Wilks R, Younger N, Mullings J, Zohoori N, Figueroa P, Tulloch-Reid M, et al.
Factors affecting study efficiency and item non-response in health surveys in developing countries: The Jamaica national healthy lifestyle survey. BMC Med Res Methodol 2007;7:13.
Napper LE, Fisher DG, Reynolds GL, Johnson ME. HIV risk behavior self-report reliability at different recall periods. AIDS Behav 2010;14:152-61.
Norris Turner A, Paul P, Norris AH. Limited benefit of repeating a sensitive question in a cross-sectional sexual health study. BMC Med Res Methodol 2013;13:34.
Weinhardt LS, Forsyth AD, Carey MP, Jaworski BC, Durant LE. Reliability and validity of self-report measures of HIV-related sexual behavior: Progress since 1990 and recommendations for research and practice. Arch Sex Behav 1998;27:155-80.
Poulin M. Reporting on first sexual experience: The importance of interviewer-respondent interaction. Demogr Res 2010;22:237-88.
Blanc AK, Croft TN. The Effect of the Sex of the Interviewer on Responses in Fertility Surveys: The Case of Ghana. Annual Meeting of the Population Association of America. Denver; 1992. Available from: https://www.popline.org/node/319563
. [Last accessed on 2018 Oct 28].
Catania JA, Binson D, Canchola J, Pollack LM, Hauck W. Effects of interviewer gender, interviewer choice, and item wording on responses to questions concerning sexual behavior. Public Opin Q 1996;60:345-75.
Reed E, Khoshnood K, Blankenship KM, Fisher CB. Confidentiality, privacy, and respect: Experiences of female sex workers participating in HIV research in Andhra Pradesh, India. J Empir Res Hum Res Ethics 2014;9:19-28.
Gregson S, Zhuwau T, Ndlovu J, Nyamukapa CA. Methods to reduce social desirability bias in sex surveys in low-development settings: Experience in Zimbabwe. Sex Transm Dis 2002;29:568-75.
Jaya, Hindin MJ, Ahmed S. Differences in young people's reports of sexual behaviors according to interview methodology: A randomized trial in India. Am J Public Health 2008;98:169-74.
[Table 1], [Table 2], [Table 3]