The necessary resources for conducting incidence studies are not often available. To monitor the course of the HIV-1 epidemic among women of reproductive age and to evaluate the effect of preventive interventions, the surveillance of HIV-1 prevalence among pregnant women attending antenatal clinics is commonly used. Although the trend of HIV-1 prevalence does not provide a completely accurate description of the dynamics of the infection , it remains an important indicator for planning public health activities.
Differences between pregnant and non-pregnant women in terms of social and behavioural factors related to HIV-1 infection could bias the estimates of the general prevalence based on antenatal clinic data . Studies conducted in Uganda have shown lower fertility rates among HIV-1-positive women in comparison with HIV-1-negative women [3,4], suggesting that HIV-1 prevalence derived from an antenatal clinic surveillance system may underestimate the infection in the general female population of reproductive age.
The objectives of the present study were to evaluate the trend of HIV-1 prevalence among women attending an antenatal clinic in a rural district of North Uganda, an area affected by civil strife and where information on HIV-1 is scarce, and to estimate HIV-1 prevalence among the general female population of reproductive age, using the antenatal clinic data and information on differences in fertility rates between HIV-1-positive and HIV-1-negative women [2,5].
During the period from January 1993 to December 1997, a total of 8627 women aged 15–49 years and living in the Gulu District were anonymously screened for HIV-1 infection at the antenatal clinic of St. Mary's Hospital Lacor in the framework of the Ugandan sero-surveillance system. The hospital has 446 beds and is located in the Gulu District, a rural area of North Uganda bordering on Sudan and inhabited by approximately 340 000 persons, 11.3% of whom live in the main town, known as the Gulu Municipality .
Since there were relatively few antenatal clinic attendees over 39 years of age (n = 72), they were excluded from the analysis, in that the 35–49 years age group would have been skewed toward the lower ages and thus mostly represented by women aged 35–39 years.
Basic socio-demographic information, such as age, tribe, and area of residence, was collected from the antenatal clinic register. The presence of HIV-1 infection was anonymously detected, using an enzyme-linked immunosorbent assay (ELISA), on the sera routinely collected for laboratory examinations during the first visit to the antenatal clinic. Due to commercial availability, different types of ELISA kits were used during the study period and, in some years, the number of tested women was reduced because of their availability.
Since there were no data available on fertility rates by HIV-1 serostatus for the Gulu District, data on age-specific fertility rates among HIV-1-positive and HIV-1-negative women living in a rural subcounty of the Masaka District (South-West Uganda) during the period 1990–1996 were considered when estimating the prevalence of the infection in the general female population aged 15–39 years in the study area .
HIV-1 prevalence rates were estimated and exact binomial 95% confidence intervals (CI) were calculated for antenatal clinic women. The statistical significance of the change in prevalence over time was evaluated using the χ2 for linear trend test; in addition the χ2 test was used for evaluating associations between HIV-1 infection and other categorical variables. Potential risk factors independently associated with HIV-1 infection were evaluated using a log-binomial regression model, which included year tested, age-class, and area of residence (i.e., Gulu Municipality and the remaining areas of the Gulu District) as covariates ; the adjusted prevalence proportions ratios (PPR) with their 95% CI were used to describe the strength of the associations.
For calculating the prevalence of HIV-1 infection among the Gulu District's general female population aged 15–39 years (PGFP), we used the prevalence rates by age and area of residence among women attending the antenatal clinic and living in the Gulu District (PANC), the relative distribution by age and area of residence of the district's female population aged 15–39 years, as reported in the 1991 Ugandan Census (FCEN), and the estimated age-specific fertility rates among HIV-1-positive (FRHIV+) and HIV-1-negative (FRHIV−) women living in a rural subcounty of the Masaka District. The PGFP was calculated by multiplying, for each age-residence stratum, FCEN by the prevalence in the general female population (prevalence estimated adjusting PANC for different fertility rates by HIV-1 serostatus, as described by Boisson et al. ) and then summing these products:EQUATION
The above formula could be simplified to read as follows:EQUATION where FRGFP indicates the age specific fertility rates among the general female population aged 15–39 years.
To increase the precision of the estimated PGFP, the years 1993 and 1994 were combined, as were the years 1996 and 1997, because of the low number of cases observed in some age–residence strata in the single years tested.
All of the analyses were performed using the SPSS  and STATA  statistical packages.
Prevalence trend of HIV-1 infection among antenatal clinic attendees
Among the 8555 screened women aged 15–39 years, 36.0% lived in the Gulu Municipality and 64.0% in other areas of the Gulu District; 95.1% of the screened women belonged to the Acholi tribe, which is characteristic of North Uganda.
During the study period, HIV-1 prevalence showed a significant linear decrease (P < 0.001), from 26.0% (95% CI, 23.2–29.0%) in 1993 to 16.1% (95% CI, 14.8–17.5%) in 1997. This trend appeared to be mainly attributable to a marked decrease in prevalence until 1995. In fact, in 1996, HIV-1 prevalence stabilized at about 14% and then slightly increased in 1997 (Table 1).
When analysing HIV-1 prevalence by age group, a statistically significant linear decrease over time (P < 0.001) was found for the youngest three age groups (women less than 30 years of age), with the only exception observed for women aged 25–29 years living in the Gulu Municipality. No significant change over time was observed for women aged 30–39 years. The age group with the overall highest prevalence was the 20–24-years-old group in 1993 and the 25–29-years-old group in the other years considered, except for 1996, when the 30–34-years-old group had the highest prevalence.
When analysing HIV-1 prevalence by area of residence, a statistically significant linear decrease over time was observed both among women living in the Gulu Municipality (P < 0.001) and among those living in other areas of the Gulu District (P < 0.001). The estimated overall prevalence for the period 1993–1997 significantly differed (P < 0.001) by area of residence: 23.1% (95% CI, 21.6–24.6%) for the Gulu Municipality and 14.9% (95% CI, 14.0–15.9%) for other areas in the Gulu District (Table 2). These differences were also significant when analysing the data separately for each year tested.
To control for potential confounding effects when comparing prevalence rates, a log-binomial regression model was run to identify the risk factors independently associated with HIV-1 infection (Table 2): all the considered variables were found to be significantly associated with the infection. The adjusted PPRs per year tested showed that the risk of being HIV-1 infected significantly decreased from 1993 to 1995, and then slightly increased in 1997, confirming the results obtained in the univariate analysis. Independently of the year tested and the area of residence, nearly all of the age groups had a higher risk than the reference category of women aged 19 years or less. The only exception was the oldest age group (35–39 years), for whom the risk appeared to be not significantly different from that of the reference category. Finally, the adjusted PPRs confirmed that the women living in the Gulu Municipality had a higher risk of being HIV-1 infected than those living in other areas of the Gulu District (PPR = 1.54; 95% CI, 1.40–1.68), independently of the year tested and of the age distribution of the women in the two areas.
Similar results were obtained when running the model separately for each year tested. The interaction between age and area of residence, initially included in the model, was found to be not significantly associated with HIV-1 infection and was thus not included in the final model presented here.
Adjusted HIV-1 prevalence for the general female population aged 15–39 years
The overall HIV-1 prevalence rates in the general female population aged 15–39 years were directly standardized for age and area of residence, assuming the age-residence distribution of females reported in the 1991 Ugandan census as standard (Table 3).
Among the subpopulation of the Masaka District, the relative risk of fertility was 1.2 to 2.3 times higher for HIV-1-negative women, compared with HIV-1-positive women, for all age groups, with the exception of the 15–19 years age group. For the 15–19 years age group, HIV-1-positive women had a higher fertility rate than HIV-1-negative women. As a consequence, when the antenatal clinic data were extrapolated to the general female population, a marked increase in HIV-1 prevalence from the 15–19 years to the 20–24 years age group was observed in our study area. The overall age-residence standardized HIV-1 prevalence was estimated to have decreased from 25.4% in the period 1993–1994 to 17.8% in the period 1996–1997 (from 30.4 to 24.5% in the Gulu Municipality and from 24.6 to 16.7% in other areas of Gulu District).
The overall age–residence standardized HIV-1 prevalence among the general female population was approximately 1.22 and 1.28 times higher in the periods 1993–1994 and 1996–1997, respectively, compared with that among antenatal clinic attendees aged 15–39 years.
Using the trend of HIV-1 prevalence among pregnant women for monitoring the epidemic could lead to inaccurate conclusions. Differences in mortality, fertility, migration, and other HIV-1-related factors between HIV-1-positive and HIV-1-negative women could obscure a high incidence in the presence of a stable or decreasing prevalence [1,10]. However, the temporal trend of HIV-1 prevalence can provide useful indications on the dynamics of the infection when restrictive assumptions on these HIV-1-related factors in the reference population can be made [10–14].
In our study, the trend of decrease in HIV-1 prevalence appears to be consistent with data from other population groups of the study area, such as secondary-school students (HIV-1 prevalence declined from 2.0% in 1994 to 0.8% in 1998), and with data reported in other antenatal clinics of Uganda (HIV-1 prevalence declined from 26.6% in 1993 to 14.6% in 1997 among antenatal clinic attendees at Nsambya Hospital in Uganda's capital, Kampala) [15,16]. Moreover, HIV-1 prevalence was found to be very high, and it was similar to that observed in Kampala . The general decrease seemed to be mainly due to the decreasing prevalence observed in the younger age groups until 1995. Assuming that the differences in mortality, fertility, and migration between HIV-1-positive and HIV-1-negative women were negligible for the younger age groups [18,19], this decrease could reflect a decreasing incidence among these young women [11,20]. HIV-1 prevalence appeared to be higher and more stable over time among women aged 25–34 years with respect to younger women, whereas the lowest prevalence was observed among those aged 35–39 years, when the effect of AIDS-related mortality probably becomes more consistent. This picture is consistent with the beginning of the mature stage of the epidemic in Uganda .
Women living in the Gulu Municipality were found to have a higher risk of HIV-1 than those living in rural areas of the Gulu District, confirming the results of other studies conducted in Uganda [22,23]. However, a significant decrease in HIV-1 prevalence was observed over time for both groups, suggesting that there are no geographic differences in the dynamics of the infection.
It is unlikely that the overall decrease in HIV-1 prevalence among antenatal clinic attendees observed between 1994 and 1995 can be totally explained by changes in incidence and differences in mortality and fertility by HIV-1 serostatus. In fact, assuming a null incidence in the 1994–1995 period and that the differences in mortality and fertility rates between HIV-1-negative and HIV-1-positive women aged 15–39 years are equal to those estimated in the Masaka District (personal communication, ), the observed decrease in prevalence is higher than expected .
Another factor potentially contributing to the observed decrease was the fact that, due to commercial availability, different types of ELISA kits were used to detect HIV-1 during the study period, possibly accounting for an increased test specificity and a consequent biased reduction in prevalence, especially considering the greater variations in the specificity of ELISA kits when used on African sera .
No differences with respect to age structure or area of residence were observed among the antenatal clinic attendees among the years tested, and there were no changes in testing policy during the study period, although the number of women tested was sometimes randomly reduced because of the availability of HIV-1 ELISA kits. Finally, the turnover of the laboratory personnel, supervisors, and data-entry clerks was limited and was unlikely to have caused major biases.
The importance of addressing differences in fertility by HIV-1 serostatus was confirmed by the finding that, in the periods 1993–1994 and 1996–1997, the age-residence standardized HIV-1 prevalence among the general female population aged 15–39 years was, respectively, 1.22 and 1.28 times greater than that observed among pregnant women.
Using data on the level of fertility of HIV-1-positive and HIV-1-negative women in a rural subcounty of the Masaka District for estimating the HIV-1 prevalence among the Gulu District's general female population aged 15–39 years could have introduced a bias in our estimates if these data are not similar to those for women in our study area. With respect to our study population, pregnant women in the reference area, located in south-western Uganda, showed lower HIV-1 prevalence (9.5% in the overall period 1990–1996, compared with 17.8% in the period 1993–1997 for our study population) and lower total fertility rate (TFR; TFR = 6.1 in 1996 compared with TFR = 6.7 in 1997 in our study area; unpublished data).
Despite these differences, we were not able to determine whether the estimated relative risks of fertility of HIV-1-positive women versus HIV-1-negative women living in the Masaka District were similar to those of our study population, although a recent review of different studies on fertility by HIV-1 serostatus suggests that these relative risks tend to be lower in populations with higher HIV-1 prevalence . However, data from other areas of Uganda and Africa have shown similar patterns of differences in fertility by age group, suggesting a substantial geographical homogeneity [19,25].
Another possible bias could be due to the fact that we assumed no differences between rural and urban areas in the relative risks of fertility (HIV-1-positive versus HIV-1-negative women), given that no reference data were available for urban areas. All of the women belonging to the reference population lived in rural areas, compared with 86.1% of the women aged 15–39 years in the Gulu District [4,6]. Thus, the HIV-1 prevalence among the general female population of the Gulu Municipality could represent an underestimate if the relative risk of fertility is lower in urban areas (relatively high prevalence with respect to rural areas) .
Finally, a bias in the time-trend analysis could be due to the implicit assumption that the relative risks of fertility remained constant over time, whereas it could be possible that the overall and age-specific relative risks of fertility changed during the study period, as a consequence of changes in behavioural and biological factors influencing the association between HIV-1 infection and fertility (e.g., increasing age at start of sexual activity, increasing use of condoms and decreasing breast-feeding among HIV-1- positive women, and increasing treatment of sexually transmitted diseases other than HIV-1). The effects of these changes are not the same, and they could affect the relative risks of fertility in different ways, toward an increase or a decrease .
However, the prevalence rates estimated for the general female population aged 15–39 years of the Gulu District were consistent with those of other areas of Uganda with similar HIV-1 prevalence among pregnant women [3,26], partially validating the obtained results and the assumed representativeness of the relative risks of fertility from the Masaka District used in this analysis.
The distribution of the age-specific prevalence rates indicates that young girls, who are just beginning to become sexually active and for whom changes in prevalence more closely reflect incidence, are at a high risk of becoming infected.
The slight increase in HIV-1 prevalence observed in 1997 could be indirectly due to the increased civil strife. In fact, conditions of insecurity have resulted in a dramatic reduction, especially for rural areas, in all activities for health education and prevention. Furthermore, the fact that many people live in protected camps as a result of population displacement may have contributed to increasing the risk of HIV-1 infection .
In conclusion, the overall reduction in HIV-1 prevalence in the period 1993–1997 could be related to many factors, such as incidence, mortality, and migration. However, assuming that the effect of mortality and migration is negligible among young pregnant women, this downward trend could be partially interpreted, as suggested in recent studies , in terms of a reduction in risk behaviour and a consequent reduction in incidence, possibly in part due to prevention and control measures.
Finally, when extrapolating HIV-1 prevalence data from antenatal clinic attendees to the general female population of reproductive age, different fertility rates between HIV-1-positive and HIV-1-negative women should be taken into account to avoid underestimating the HIV-1 prevalence among this sub-group of the population.
This manuscript is dedicated to Dr Matthew Lukwiya, Medical Superintendent of Lacor Hospital, who died of ebola in December 2000 helping his people during this dramatic outbreak. All the staff involved in the Uganda AIDS Project, who have had the pleasure of knowing him, will remember his professionalism, intelligence, gaiety and availability in promoting and improving the health status of the population of the Gulu District.
The authors are grateful to Claudio Blè, Jacob Ouma, Proscovia Akello, and Florence Atim for data collection, blood testing and data entry; to Mark Kanieff for editorial assistance; to Ernesto Costabile for bibliographic research; and to Alessia Ranghiasci, Patrizia Tancredi, Maria Grazia Dente, and Alessandro Di Vincenzo for their continuous support to the project. The authors also thank L. M. Carpenter and A. J. Nunn for providing helpful information in the preparation of this paper.
1. 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: 1023 –1030.
2. Boisson E, Nicoll A, Zaba B, Rodrigues LC. Interpreting HIV seroprevalence data from pregnant women.
J Acquir Immune Defic Synd Hum Retrovirol 1996, 13: 434 –439.
3. Gray RH, Wawer MJ, Serwadda D. et al. Population-based study of fertility in women with HIV-1 infection in Uganda.
Lancet 1998, 351: 98 –103.
4. Carpenter LM, Nakiyingi JS, Ruberantuari A, Malamba SS, Kamali A, Whitwhort JAG. Estimates of the impact of HIV infection on fertility in a rural Ugandan population.
Health Transition Rev 1997, 7 (suppl. 2): 113 –126.
5. Nicoll A, Stephenson J, Griffioen A, Cliffe S, Rogers P, Boisson E. The relationship of HIV prevalence in pregnant women to that in women of reproductive age: a validated method for adjustment.
AIDS 1998, 12: 1861 –1867.
6. Ministry of Finance and Economic Planning (MFEP), Statistics Department. The 1991 Population and Housing Census (District Summary Series) – Gulu District.
Entebbe: MFEP; 1997.
7. Skow T, Deddens J, Petersen M, Endahl L. Prevalence proportion ratios: estimation and hypothesis testing.
Int J Epidemiol 1998, 27: 91 –95.
8. Norosis MJ. SPSS for Windows version 6.1.3.
Chigago: SPSS inc.; 1995.
9. StatCorp. Stata Statistical Software: Release 5.0.
College Station, TX: Stata Corporation; 1997.
10. Batter V, Matela B, Nsuami M. et al. High HIV-1 incidence in young women masked by stable overall seroprevalence among childbearing women in Kinshasa, Zaire: estimating incidence from serial seroprevalence data.
AIDS 1994, 8: 811 –817.
11. Heyward WL, Osmanov S, Saba J. et al. Preparation for phase III HIV vaccine efficacy trials: methods for the determination of HIV incidence.
AIDS 1994, 8: 1285 –1291.
12. Gregson S, Donnelly CA, Parker CG, Anderson RM. Demographic approaches to the estimation of incidence of HIV-1 infection among adults from age-specific prevalence data in stable endemic conditions.
AIDS 1996, 10: 1689 –1697.
13. Podgor MJ, Leske MC. Estimating incidence from age-specific prevalence for irreversible diseases with differential mortality.
Stat Med 1986, 5: 573 –578.
14. Ades AE. Serial HIV seroprevalence surveys: interpretation, design, and role in HIV/AIDS prediction.
J Acquir Immune Defic Synd Hum Retrovirol 1995, 9: 490 –499.
15. Fabiani M, Accorsi S, Corrado B, et al
. HIV prevalence trends among population groups in the Gulu district (Northern Uganda). XIth International Conference on AIDS and STDs in Africa.
Lusaka. September 1999 [abstract 13PT31-14].
16. STD/AIDS Control Programme:HIV/AIDS surveillance report – March 1998.
Entebbe: Ministry of Health; 1998.
17. Asiimwe-Okiror G, Opio AO, 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: 1757 –1763.
18. Boerma JT, Nunn AJ, Whitworth JAG. Mortality impact of the AIDS epidemic: evidence from community studies in less developed countries.
AIDS 1998, 12 (suppl. 1): S3 –S14.
19. Zaba B, Gregson S. Measuring the impact of HIV on fertility in Africa.
AIDS 1998, 12 (suppl. 1): S41 –S50.
20. Mertens T, Carael M, Sato P, Cleland J, Ward H, Smith GD. Prevention indicators for evaluating the progress of national AIDS programmes.
AIDS 1994, 8: 1359 –1369.
21. Stoneburner RL, Low-Beer D, Tembo GS, Mertens TE, Asiimwe-Okiror G. Human Immunodeficiency virus infection dynamics in east Africa deduced from surveillance data.
Am J Epidemiol 1996, 144: 682 –695.
22. Wawer MJ, Serwadda D, Musgrave SD, Konde-Lule JK, Musagara M, Sewankambo NK. Dynamics of spread of HIV-1 infection in a rural district of Uganda.
BMJ 1991, 303: 1303 –1306.
23. Serwadda D, Wawer MJ, Musgrave SD, Sewankambo NK, Kaplan JE, Gray RH. HIV risk factors in three geographic strata of rural Rakai District, Uganda.
AIDS 1992, 6: 983 –989.
24. Thorstensson R, Andersson S, Lindback S. et al. Evaluation of 14 commercial HIV-1/HIV-2 antibody assays using serum panels of different geographical origin and clinical stage including a unique seroconversion panel.
J Virol Methods 1998, 70: 139 –151.
25. Desgrees du Lou A, Msellati P, La Ruche G, Welffens-Ekra C, Ramon R, Dabis F. Estimation of HIV-1 prevalence in the population of Abidjan by adjustment of the prevalence observed in antenatal centres.
AIDS 1999, 13: 526 –527.
26. Kilian AHD, 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: 391 –398.
27. Kayita J, Kyakulaga JB. HIV/AIDS status report, Uganda, July 1997.
Kampala: Uganda AIDS Commission; 1997.
© 2001 Lippincott Williams & Wilkins, Inc.