a Medical Research Council Programme on AIDS/Uganda Virus Research Institute, PO Box 49, Entebbe, Uganda. E-mail: mrc{at}starcom.co.ug
b Current affiliation: Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD, USA.
c Current affiliation: Unit of Descriptive Epidemiology, International Agency for Research on Cancer, Lyon, France.
Dr SM Mbulaiteye, 6120 Executive Blvd, EPS Rm.8007 MSC 7248, Rockville, MD 20852, USA. E-mail: mbulaits{at}mail.nih.gov
Abstract
Background Population-based studies are thought to provide generalizable epidemiological data on the human immunodeficiency virus type 1 (HIV-1) epidemic. However, longitudinal studies are susceptible to bias from added attention caused by study activities. We compare HIV-1 prevalence in previously and newly surveyed villages in rural southwest Uganda.
Methods The study population resided in 25 neighbouring villages, of which 15 have been surveyed for 10 years. Respondents (13 years) provided socio-demographic and sexual behaviour data and a blood sample for HIV-1 serology in private after informed consent. We tested the independent effect of residency: (1) original versus new villages; (2) proximity to main road; and (3) proximity to trading centre on HIV-1 serostatus of respondents using multivariate logistic regression.
Results There were 8990 adults censused, 68.3% were from the original villages, 48.2% were males and 6111 (68.0%) were interviewed and had definite HIV-1 serostatus. The HIV-1 prevalence was 6.1% overall, 5.7% in the new, and 6.4% in the original villages (P = 0.25). Residency in the new or original villages did not independently predict HIV-1 serostatus of respondents (P = 0.46). Independent predictors of HIV-1 serostatus were education (primary or higher, odds ratio [OR] = 1.7 and 1.4, respectively), being separated or widowed OR = 4.2, reported previous use of a condom OR = 1.8, or reported genital ulceration OR = 3.3, and age group 2534 and 3544 years OR = 5.8 and OR = 4.8 (all P 0.001).
Conclusions In the context of rural Uganda where there has been considerable health education about AIDS, the additional attention to HIV infection caused by this longitudinal study does not appear to have appreciably affected the prevalence of HIV-1 infection.
Keywords Epidemiology, AIDS, Africa, Hawthorne effect, HIV-1, Uganda
Accepted 26 April 2002
Epidemiological studies of the AIDS epidemic in sub-Saharan Africa are conducted mainly on selected populations such as hospitalized patients,1 individuals with sexually transmitted diseases,2 commercial sex workers,3 and women attending antenatal clinics.4,5 Although these studies provide valuable insights into the spread of the epidemic, they are prone to bias. Better estimates are provided by population-based studies,6,7 which can also be used to estimate the magnitude and direction of bias in studies of selected groups.5,8 For example, human immundeficiency virus type 1 (HIV-1) prevalence from antenatal surveillance systematically underestimates general population rates4 and the study of high-risk groups does not provide generalizable data.9,10
However, longitudinal observational population-based studies are themselves susceptible to bias from non-specific effects, so called Hawthorne effects.11 The Hawthorne effect is the tendency for people to change behaviour because they are the target of special interest and attention in a study, regardless of the specific nature of intervention they may be receiving.12 In this context, the significant declines in HIV-1 prevalence and corresponding change in sexual behaviour, i.e. increased use of condoms, fewer teenage pregnancies and later age of sexual debut, reported between 1989 and 1999 in Uganda4,13 may partly be explained by the Hawthorne effect. A significant contribution of the Hawthorne effect would diminish the generalizability of longitudinal population-based cohort studies.14 Thus a better understanding of the comparability of study and non-study populations would aid the interpretation of trends reported from longitudinal population-based studies.
We recently added 10 new survey villages to a longitudinal cohort comprising a cluster of 15 original survey villages. This expansion also provided a rare opportunity to study the generalizability of our epidemiological findings to non-study villages. In this paper we compare HIV-1 prevalence, socio-demographic characteristics and reported sexual behaviour in previously and newly surveyed neighbouring villages in rural southwest Uganda.
Methods
Survey methods
The cohort resided in a cluster of 25 neighbouring villages in southwest Uganda, about 30 km from the district capital of Masaka, and 16 km from an international highway connecting Mombasa and Kigali (Figure 1). Fifteen villages have been surveyed annually for 10 years (19891999). The original study area covered approximately one-third of a sub-county of about 80 km2. A sub-county is an administrative unit under a county and is sub-divisible into parishes and villages. In 1999/2000, 10 new villages to the south and east of the original study area were added to the cohort to improve the precision of estimates of prevalence and incidence trends. After expansion, the study area covered approximately half of the sub-county (25 of 50 villages) about 120 km2. The study area lies 1200 m above sea level, is characterized by low-lying hills interspaced with swamps and is accessible by an all-season road that runs through the centre of the study area but seasonal dirt roads and footpaths connect the villages. Residents live on homesteads (population density about 150/km2) and practise subsistence farming as the main economic activity. Study methods have been described previously.1517 Briefly, annual censuses and serological surveys were used to study the population. Local registrars provided supplemental information to the census on birth and deaths.
|
Trained mappers constructed village maps and lists of household heads for use by enumerators who visited each household to conduct a census and to collect demographic data on all individuals. Using this information census lists were constructed. A team of trained interviewers (nurses, medical assistants and phlebotomists) visited each household about 3 weeks after enumeration and invited household members to participate in the survey. If subjects were absent during the visit, repeat visits were made to the house until the team moved to the next village, on average, after 3 weeks stay in each village, depending on the size of the village. After obtaining written informed consent, interviewers administered a risk factor questionnaire to individuals in private and also took a venous blood sample for HIV-1 serology. The questionnaire elicited information on condom use, sexual behaviour, and sexually transmitted diseases in the past 12 months.
Ethical issues
Survey staff routinely gave health education and provided information about HIV-1 infection and encouraged the practice of abstinence or safe sex. Survey staff also advised participants to go for HIV-1 counselling and to request their HIV-1 test results. The HIV-1 test results were issued to respondents who requested them after pre-test counselling.18,19 The HIV-1 serostatus of participants was masked to survey and clinical staff. Individuals who were found ill during the house visit were given treatment and if necessary, they were referred to a purpose-built study clinic stocked according to the national Essential Drug List where they received free treatment. The Uganda National Council for Science and Technology gave ethical approval for the study.
Laboratory methods
Two independent enzyme immunoassays (EIA) were used to determine HIV-1 serostatus (Wellcozyme HIV-1 recombinant VK 56/57, Murex Biotech Ltd, Dartford, Kent, UK; and Recombigen HIV-1/2, Cambridge Biotech Ltd, Galway, Ireland) following set algorithms. Samples discordant on EIA and all first time positives were tested on Western blot (Cambridge Biotech Corporation, Rockville, USA) to determine final HIV-1 serostatus.20,21
Statistical methods
Data were entered in duplicate in a database FoxPro 2.6 for Windows (Microsoft Corporation, USA) and checked for consistency before statistical analysis was performed using Stata 6.0 (Stata Corporation, College Station, Texas, USA). Distributions of socio-demographic data were compared between the new and original villages using the 2. Age-sex standardized HIV-1 prevalence for all villages were calculated using the 1999/2000 population of the 25 villages as the standard. We looked for an association between HIV-1 prevalence and place of residence to examine whether HIV-1 prevalence varied as a function of proximity to a main road, a large trading centre (
20 shops) or a previously surveyed village. We used multivariate logistic regression restricted to sexually active respondents to look for an association between various demographic characteristics, residency and reported sexual behaviour with HIV-1 serostatus. We took into account the possible correlation of responses between individuals living in the same village using a robust variance-covariance matrix.
Results
A total of 8990 adults (13 years) residing in 3837 households were censused at the 11th annual survey round (10/1999 to 10/2000). In all, 68.3% of those censused lived in the original survey villages, and 48.2% were males. Of the 2852 adults censused in the new survey villages, 64.3% had been resident in their village for
5 years. The distribution of previous village of residence among those who had immigrated to a new village within the last 5 years was 10.1% from one of the original villages, 40.9% from another new village and 49.0% from outside the study area. A total of 6138 (68.3%) adults participated in the serosurvey, compliance ranged from 64.1 to 77.7% (mean 72.1%) in the new villages and 59.8 to 72.9% (mean 66.5%) in the old villages. Definite HIV-1 serostatus was available for 6111 (68%) respondents.
Table 1 shows the distribution of socio-demographic variables between the new and original survey villages. Respondents from new survey villages were more likely to have attained secondary or higher level of education, to be separated or widowed, and to be sexually active than respondents from the original villages. Table 2
shows the crude and age-sex standardized HIV-1 prevalence for new and original survey villages and for villages according to proximity to the main road or trading centre. Overall HIV-1 prevalence was 6.1% (95% CI: 5.56.8). No difference was observed in HIV-1 prevalence between the new and original survey villages, (P = 0.25). However, the range in HIV-1 prevalence was greater in the original (1.110.7%) than in the new villages (3.48.7%). We found no difference in HIV-1 prevalence among respondents according to proximity of village to the main road (P = 0.37) or to a large trading centre (P = 0.30). Figure 2
shows the distribution of average HIV-1 prevalence by village for categories >7%, 57% or <5% in the study area. Villages closer to the centre of the study area tended to have a higher HIV-1 seroprevalence category than those towards the edges of the study area.
|
|
|
|
We compared HIV-1 prevalence in previously and newly surveyed populations in rural southwest Uganda. The HIV-1 prevalence in the newly surveyed villages was similar to that in the original villages despite the original villages experiencing additional focus on HIV-1 infection caused by survey and other activities. The absence of a significant difference in HIV-1 prevalence between original and new villages despite respondents in the new villages being more likely to report behaviour associated with HIV-1 infection suggests similar practise of appropriate safe sexual behaviour in the new and original villages.4,13,22 These results are also compatible with a relatively less important role of high-risk sexual behaviour in the spread of HIV-1 infection; spread being a function of HIV-1 prevalence in a generalized epidemic.23
Declining HIV-1 prevalence, attributed largely to national AIDS education campaigns, has been reported in Uganda, among women attending antenatal clinics,4,13 and from population-based cohorts.22,24,25 We do not know what the HIV-1 prevalence in the new villages was in 1989 when the original cohort was established. If we assume that the HIV-1 prevalence in the new villages in 1989 was similar to that observed in the original villages (8.2%), then our results imply that HIV-1 prevalence has declined concomitantly in the new villages to the current prevalence of 5.8%. Such significant decline in HIV-1 prevalence in non-study villages would occur because of appropriate community response to effective national AIDS education campaigns,13 rather than because of interaction with survey staff as occurred in the original survey villages. Our results therefore provide reassurance that trends in HIV-1 infection reported from the original survey villages are generalizable to non-survey villages.
Strenuous efforts aimed at creating a high awareness of risk factors for HIV infection in populations in Uganda by the Uganda Ministry of Health are widely acknowledged.26,27 Thus, in the context of considerable AIDS education in Uganda, the striking similarity between previously and newly surveyed villages do not support a Hawthorne effect as an explanation for the reported declining HIV prevalence. Our results are relevant to the design of community intervention studies. Since comparison communities participating in a community randomized trial are likely to be exposed to national AIDS campaigns, the potential for dilution of intervention from national AIDS campaigns need to be considered carefully when designing community intervention trials.28 National interventions are likely to diminish differences between study and non-study villages. HIV-1 prevalence in the original villages was similar to that in the new villages, despite the additional focus on HIV-1 infection these villages were exposed to through close interaction with survey staff over the past 10 years.
The strengths of our study are its large size, its population-based nature, and being able to make direct comparison of previously and newly surveyed populations. Several limitations need to be considered. We used cross-sectional data to compare the new and original villages. We do not know whether HIV-1 prevalence was different between the new and original villages or the magnitude of such a difference for the earlier time periods. Secondly, new survey villages are close to the original villages and could have been subject to similar experiences. We are unable to exclude the possibility that cross-diffusion particularly due to migration between new and original villages partly explains these results. Migration of individuals from the original to the new villages could potentially transfer awareness of the HIV epidemic, however, only 10.5% of recent immigrants to the new villages came from the original villages. Migration between new and original and within villages is complex and depends on where a village is located relative to a new or original village. We have no reason to believe that migration was predominantly in one direction. Ideally, villages that are further away from the original study villages would have provided the best comparative data for our study. However, the primary reason for expanding the cohort was to improve on precision of estimates of HIV prevalence and incidence, and also it was logistically difficult to add villages that were distant from the original study villages.
Thirdly, counselling and community development activities were also conducted in the new villages for ethical reasons and may have caused some focus on the HIV-1 epidemic, thereby diminishing differences in exposure to research staff between the new and original villages. Nonetheless, the quality and intensity of direct personal interaction between survey staff and the study population would be expected to be more intense in the original and minimal in the new villages, despite the relative proximity. We believe also that such direct influence in the new villages was small and irregular over time. Study activities were concentrated in participating villages and recruitment and household health education occurred in participants homes.
We note that this is an observational study, and that our comparisons do not have the rigorous quality of a randomized controlled study. Our study is unable to quantify and take into account the extent of person-to-person interaction between individuals from the new and the original villages. Finally, slightly more than a quarter of residents did not participate in the serological survey, raising the potential for selection bias. However, it is usually difficult to obtain higher compliance rates in studies that involve the taking of blood and of HIV testing.
To summarize, HIV-1 prevalence was similar between survey and non-survey villages. These results provide reassurance that the previously surveyed populations provide data that is generalizable to non-study villages. In the context of rural Uganda where there has been considerable health education about AIDS, the additional attention to HIV infection caused by repeated person-to-person interaction with research survey staff over 10 years does not appear to have appreciably affected the prevalence of HIV-1 infection in the original villages. This study provides reassurance that previously reported longitudinal HIV trends from this cohort are generalizable to non-survey villages.
KEY MESSAGES
|
Acknowledgments
We are grateful to the study population for their participation in the survey and to the Uganda Ministry of Health for allowing this work to be published. The Medical Research Council and the Department for International Development (DFID) of the United Kingdom funded this study.
References
1 Berkley S, Okware S, Naamara W. Surveillance for AIDS in Uganda. AIDS 1989;3:7985.[ISI][Medline]
2 Nsubuga P, Mugerwa R, Nsibambi J, Sewankambo N, Katabira E, Berkley S. The association of genital ulcer disease and HIV infection at a dermatology-STD clinic in Uganda. J Acquir Immune Defic Syndr 1990; 3:100205.[Medline]
3 Mann JM, Nzilambi N, Piot P et al. HIV infection and associated risk factors in female prostitutes in Kinshasa, Zaire. AIDS 1988;2:24954.[ISI][Medline]
4 Asiimwe-Okiror G, Opio AA, Musinguzi J, Madraa E, Tembo G, Carael M. Change in sexual behaviour and decline in HIV infection among young pregnant women in urban Uganda. AIDS 1997;11:175763.[CrossRef][ISI][Medline]
5 Zaba B, Boerma T, White R. Monitoring the AIDS epidemic using HIV prevalence data among young women attending antenatal clinics: prospects and problems. AIDS 2000;14:163345.[CrossRef][ISI][Medline]
6 Mulder DW, Nunn AJ, Wagner HU, Kamali A, Kengeya-Kayondo JF. HIV-1 incidence and HIV-1-associated mortality in a rural Ugandan population cohort. AIDS 1994;8:8792.[ISI][Medline]
7 Wawer MJ, Sewankambo NK, Berkley S et al. Incidence of HIV-1 infection in a rural region of Uganda. BMJ 1994;308:17173.
8 Konings E, Bantebya G, Carael M, Bagenda D, Mertens T. Validating population surveys for the measurement of HIV/STD prevention indicators. AIDS 1995;9:37582.[ISI][Medline]
9 Fylkesnes K, Ndhlovu Z, Kasumba K, Mubanga Musonda R, Sichone M. Studying dynamics of the HIV epidemic: population-based data compared with sentinel surveillance in Zambia. AIDS 1998; 12:122734.[CrossRef][ISI][Medline]
10 Low-Beer D, Stoneburner RL. An age- and sex-structured HIV epidemiological model: features and applications. Bull World Health Organ 1997;75:21321.[ISI][Medline]
11 Murray M, Swan AV, Kiryluk S, Clarke GC. The Hawthorne effect in the measurement of adolescent smoking. J Epidemiol Community Health 1988;42:30406.[Abstract]
12 Bouchet C, Guillemin F, Briancon S. Nonspecific effects in longitudinal studies: impact on quality of life measures. J Clin Epidemiol 1996;49:1520.[CrossRef][ISI][Medline]
13 Kilian AH, Gregson S, Ndyanabangi B et al. Reductions in risk behaviour provide the most consistent explanation for declining HIV-1 prevalence in Uganda. AIDS 1999;13:39198.[CrossRef][ISI][Medline]
14 Campbell JP, Maxey VA, Watson WA. Hawthorne effect: implications for prehospital research. Ann Emerg Med 1995;26:59094.[ISI][Medline]
15 Nunn AJ, Wagner HU, Kamali A, Kengeya-Kayondo JF, Mulder DW. Migration and HIV-1 seroprevalence in a rural Ugandan population. AIDS 1995;9:50306.[ISI][Medline]
16 Mulder DW, Nunn AJ, Kamali A, Nakiyingi J, Wagner HU, Kengeya-Kayondo JF. Two-year HIV-1-associated mortality in a Ugandan rural population. Lancet 1994;343:102123.[ISI][Medline]
17 Nunn AJ, Kengeya-Kayondo JF, Malamba SS, Seeley JA, Mulder DW. Risk factors for HIV-1 infection in adults in a rural Ugandan community: a population study. AIDS 1994;8:8186.[ISI][Medline]
18 Seeley J, Wagner U, Mulemwa J, Kengeya-Kayondo J, Mulder D. The development of a community-based HIV/AIDS counselling service in a rural area in Uganda. AIDS Care 1991;3:20717.[Medline]
19 Gray RH, Quinn TC, Serwadda D, Sewankambo NK, Wabwire-Mangen F, Wawer MJ. The ethics of research in developing countries. N Engl J Med 2000;343:36163.
20 Nunn AJ, Biryahwaho B, Downing RG, van der Groen G, Ojwiya A, Mulder DW. Algorithms for detecting antibodies to HIV-1: results from a rural Ugandan cohort. AIDS 1993;7:105761.[ISI][Medline]
21 Nunn AJ, Biryahwaho B, Downing RG, Ojwiya A, Mulder DW. Computer-assisted quality assurance in an HIV serology laboratory. Methods Inf Med 1994;33:17073.[ISI][Medline]
22 Mulder D, Nunn A, Kamali A, Kengeya-Kayondo J. Decreasing HIV-1 seroprevalence in young adults in a rural Ugandan cohort. BMJ 1995; 311:83336.
23 Hudson CP. Community-based trials of sexually transmitted disease treatment: repercussions for epidemiology and HIV prevention. Bull World Health Organ 2001;79:4858.[ISI][Medline]
24 Kamali A, Carpenter LM, Whitworth JA, Pool R, Ruberantwari A, Ojwiya A. Seven-year trends in HIV-1 infection rates, and changes in sexual behaviour, among adults in rural Uganda. AIDS 2000;14:42734.[CrossRef][ISI][Medline]
25 Wawer MJ, Serwadda D, Gray RH et al. Trends in HIV-1 prevalence may not reflect trends in incidence in mature epidemics: data from the Rakai population-based cohort, Uganda. AIDS 1997;11:102330.[CrossRef][ISI][Medline]
26 Konde-Lule JK, Berkley SF, Downing R. Knowledge, attitudes and practices concerning AIDS in Ugandans. AIDS 1989;3:51318.[ISI][Medline]
27 Asera R, Bagarukayo H, Shuey D, Barton T. An epidemic of apprehension: questions about HIV/AIDS to an east African newspaper health advice column. AIDS Care 1997;9:512.[CrossRef][ISI][Medline]
28 Wawer MJ, Sewankambo NK, Serwadda D et al. Control of sexually transmitted diseases for AIDS prevention in Uganda: a randomised community trial. Rakai Project Study Group. Lancet 1999;353:52535.[CrossRef][ISI][Medline]