Share this article on:

Estimating HIV Prevalence and the Impact of HIV/AIDS on a Ugandan Hospital by Combining Serosurvey Data and Hospital Discharge Records

Fabiani, Massimo; Accorsi, Sandro; Aleni, Robert; Rizzardini, Giuliano; Nattabi, Barbara; Gabrielli, Antonio; Opira, Cyprian; Declich, Silvia

JAIDS Journal of Acquired Immune Deficiency Syndromes: September 1st, 2003 - Volume 34 - Issue 1 - p 62-66
Clinical Science: Brief Report

Objective To estimate the disease-specific HIV prevalence in a northern Ugandan hospital and to evaluate the impact of HIV/AIDS on hospital services.

Design HIV serosurvey and analysis of routinely compiled hospital records.

Methods The serosurvey was conducted among all 352 patients admitted to the medical ward of the Lacor Hospital in March 1999 (this ward consists of 3 units: general medicine, tuberculosis, and cancer). The impact on hospital services was estimated using the hospital discharge records for all 3447 patients admitted in 1999, in combination with serosurvey data, and was expressed as the percentage of bed-days attributable to HIV-positive patients.

Results The overall HIV prevalence was 42.0% (52.6, 44.6, and 13.2% in the general medicine, tuberculosis, and cancer units, respectively). The disease-specific prevalence ranged from 45–65% for patients with tuberculosis, pneumonia, malaria, and enteritis. HIV-positive patients, compared with HIV-negative patients, had a higher in-hospital mortality (14.6 vs. 3.0%) and a lower average length of stay (41.4 vs. 48.9 days). AIDS cases accounted for 5.0% of hospital admissions, 4.1% of bed-days, and 11.5% of deaths. When considering all HIV-positive patients, these accounted for 37.2% of the bed-days.

Conclusions Knowledge of disease-specific HIV prevalence and of the patterns of HIV-related diseases is crucial for early case management. The impact of HIV-positive patients on hospital services is quite high, accounting for >1/3 of the bed-days in 1999. Providing a continuum of care through inpatient, outpatient, and outreach home care services probably represents the only means of relieving the pressure on overloaded hospitals.

From the Laboratory of Epidemiology and Biostatistics, Istituto Superiore di Sanità, Rome, Italy (Drs Fabiani, Accorsi, Gabrielli, and Declich); St. Mary's Hospital Lacor, Gulu, Uganda (Drs Aleni, Nattabi, and Opira); and Division of Infectious Diseases, Busto Arsizio Hospital, Italy (Dr Rizzardini).

Received for publication January 2, 2003; accepted June 10, 2003.

Supported by Uganda AIDS Project no. 667 (Istituto Superiore di Sanità, Rome, Italy).

Reprints: Silvia Declich, Laboratory of Epidemiology and Biostatistics, Istituto Superiore di Sanità, Viale Regina Elena 299, 00161 Rome, Italy (e-mail:

This manuscript is dedicated to Dr. Matthew Lukwiya, Medical Superintendent of Lacor Hospital, who died of Ebola in December 2000 and who substantially contributed to the conception of this and previous serosurveys, in the continuous effort to find new ways of helping the people of his native northern Uganda.

In Uganda, the HIV/AIDS epidemic has placed an additional burden on the already overwhelmed health care system, and, as the epidemic matures, the gap between health care needs and available resources widens. 1 To prioritize these needs and develop sustainable strategies for providing adequate health care to HIV-positive persons, geographically specific information on the epidemic's dynamics and disease patterns is necessary, 2,3 especially as the epidemic consists of a variety of opportunistic infections, whose patterns of occurrence differ among patients and countries. 4

In the past decade, northern Uganda has suffered greatly from war, population displacement, and the consequent poverty, socioeconomic disruption, and collapse of the health infrastructure; little information is available on the impact of HIV/AIDS on health care services. The objectives of the present hospital-based study were to estimate the prevalence of HIV infection by reason for hospitalization and to evaluate the impact of HIV/AIDS on hospital services, by combining data from an HIV serosurvey and from routinely compiled hospital discharge records.

Back to Top | Article Outline


Study Population

The study population included patients >12 years of age 5 who had been admitted in 1999 to the medical ward of the St. Mary's Hospital Lacor (“Lacor Hospital”), a private, nonprofit, teaching hospital located in northern Uganda's Gulu District; the hospital has 463 beds and there were 17,649 admissions in 1999 (Lacor Hospital, activity report, 1999, unpublished data). The medical ward has 203 beds and consists of 3 units: general medicine (69 beds), tuberculosis (90 beds), and cancer (44 beds).

Back to Top | Article Outline

HIV Serosurvey

To estimate the HIV-1 prevalence by specific reason for hospitalization (i.e., the principal diagnosis at discharge), we conducted a 1-month serosurvey among all 352 patients (>12 years of age) consecutively admitted to the medical ward, or already present, in March 1999. Blood was drawn after informed consent and was tested using ELISA (Wellcozyme HIV Recombinant; Murex Biotech, Ltd., Dartford, UK); positive samples were retested with a second ELISA (Recombigen HIV-1/HIV-2 EIA; Cambridge Biotech Corp., Galway, Ireland). Patients with discordant results were also tested with Western blot (New Lav Blot 1; Sanofi Diagnostics Pasteur, Marnes-la-Coquette, France), and those with an undetermined test result were excluded from the analysis. Pre- and posttest counseling was available. Demographic and clinical data and information on length of hospital stay, underlying reason for hospitalization, and survival outcome were collected from hospital charts. Differences in prevalence and the associations between categorical variables were evaluated using the χ2 test or, when appropriate, the Fisher exact test. A t test was used to evaluate differences with respect to continuous variables.

Back to Top | Article Outline

Hospital Discharge Records

For the 3447 patients >12 years of age who had been admitted to the medical ward in 1999, we identified the underlying reason for hospitalization as the principal diagnosis reported on hospital discharge records. These diagnoses are coded according to the current reporting system of the Ugandan Ministry of Health (based on the 10th revision of the International Classification of Diseases 6) and entered into a computerized database. In accordance with the Ugandan AIDS case definition 7 (a slightly modified version of the “Bangui definition”), clinical AIDS cases were identified using the code for AIDS and that for Kaposi sarcoma (KS). The code for AIDS is based on clinical criteria (i.e., all patients with at least 2 of the 3 major signs and at least 1 of the 7 minor signs in the Ugandan AIDS case definition). Cryptococcal meningitis, the other disease sufficient for diagnosing AIDS, was not considered because in the hospital discharge database no distinction is made among the various types of meningitis.

Back to Top | Article Outline

Hospital Bed-Days

The impact of HIV infection on hospital services was evaluated in terms of the percentage of bed-days, which is based on the frequency of admission and the average length of stay (ALOS), attributable to HIV-positive patients, as follows:MATH



where IHS is the impact on hospital services, BDHIV+ is the disease-specific proportion of bed-days accounted for by HIV-positive patients (data from the HIV serosurvey), and BDTOT is the total proportion of bed-days attributable to the specific underlying reason for hospitalization (data from the hospital discharge records).

Diagnoses with >10 cases in the serosurvey were considered individually, whereas all other diagnoses were combined into the category “others.” Cases of KS were combined with those of AIDS, whereas cryptococcal meningitis was included in the category “others” for the sake of consistency with the hospital database.

A sensitivity analysis was performed to account for possible differences in the case-mix of this category when comparing the serosurvey patients to the 1999 medical ward patients. The serosurvey and the use of hospital data were approved by the hospital's management and ethical committee.

Back to Top | Article Outline


Of the 352 serosurvey patients, 343 had determined test results (169 males, 174 females): 154 patients in the general medicine unit, 121 in the tuberculosis unit, and 68 in the cancer unit. The mean age was 36.2 years overall (SD = 13.8), 34.3 years (SD = 10.3) for HIV-positive patients, and 37.5 years (SD = 15.7) for HIV-negative patients (P = 0.025). Regarding area of residence, 9.9, 51.7, and 38.4% of the patients lived in the Gulu Municipality, the rural Gulu District, and other districts, respectively. No pregnant women were admitted to the medical ward during the serosurvey.

The HIV prevalence was 42.0% (CI: 36.7–47.4%), with no difference between genders. The prevalence significantly differed among the 3 units (P < 0.001): 52.6% (CI: 44.4–60.7%) in the general medicine unit, 44.6% (CI: 35.6–53.9%) in the tuberculosis unit, and 13.2% (CI: 6.2–23.6%) in the cancer unit.

Table 1 shows the HIV prevalence, ALOS, and mortality by reason for hospitalization for the 343 patients. In addition to the 12 patients clinically diagnosed with AIDS, 10 were diagnosed with KS, and 5 with cryptococcal meningitis, all but 1 of whom were HIV positive (i.e., 1 case of KS). For patients with tuberculosis, pneumonia, malaria, and enteritis, the prevalence ranged from 45.4–65.4%. None of the patients admitted for common forms of cancer other than KS (i.e., cervical, breast, and esophageal carcinoma) were HIV positive. Mortality was significantly higher among HIV-positive patients, compared with HIV-negative patients, also when controlling for age and gender (OR = 6.1; CI: 2.3–15.9). This could in part explain the lower ALOS among HIV-positive patients. In fact, when excluding all patients who died, the difference decreased: ALOS of 44.2 (SD = 36.8) for HIV-positive patients and 49.6 days (SD = 35.8) for HIV-negative patients.



Table 2 shows, for the 3447 medical ward patients, the number of admissions, ALOS, bed-days, and mortality for the 10 leading reasons for hospitalization (the distribution of which was similar between these patients and the 343 serosurvey patients). AIDS cases, including KS, accounted for 173 admissions (5.0%), 2965 bed-days (4.1%), and 47 deaths (11.5%).



Regarding the impact on hospital services, HIV-positive patients accounted for 5967/15,700 bed-days (38.0%) in the serosurvey. Using the above-cited formula, for the patients admitted to the medical ward in 1999, we estimated that 37.2% of the bed-days were attributable to HIV-positive patients. This percentage varies from 32.3%, if all patients in the category “others” are assumed to be HIV negative, to 63.9%, if they are assumed to be HIV positive.

Back to Top | Article Outline


In the 1990s, >50% of the patients in the medical wards of Ugandan hospitals were found to be HIV positive, crowding out patients with other illnesses. 1,8–10 In Lacor Hospital's medical ward, 67.7% of the patients were HIV positive in 1994. 11 However, in a 1997 survey, the prevalence had decreased (45.8%; P < 0.001), 12 with a further, though slight, decrease observed in the present survey (42.0%; P = 0.782). This decrease could in part be due to the shift from inpatient to outpatient care and home assistance, which the hospital has provided since 1994. The finding that HIV prevalence was higher among patients in the general medicine unit than among those in the tuberculosis unit suggests that the overall high prevalence is mainly due to the case-mix in the former group (44.9% of all patients included in the survey) rather than to the high proportion of patients admitted for tuberculosis (38.2% of all patients included in the survey), a disease which is strictly related to HIV infection.

Knowledge of disease-specific HIV prevalence and of the patterns of HIV-related diseases is crucial for successful early case management. 13,14 In our serosurvey, all clinical AIDS diagnoses were confirmed by an HIV test result, except for 1 case of KS, which was a non-HIV-related form. In fact, KS was endemic in Uganda before the appearance of AIDS, although its incidence has dramatically increased in past decades. 15

An HIV prevalence rate ranging from 45.0–53.3% was found for illnesses known to be HIV related in Africa (i.e., tuberculosis and enteritis). Moreover, a high prevalence (65.4%) and in-hospital mortality (19.2%) were observed among patients with pneumonia, and, although it is not possible to perform the etiologic diagnosis of Pneumocystis jiroveci pneumonia at Lacor Hospital, this HIV-related condition may have contributed to this finding. The prevalence was also high among malaria patients (52.2%), which is consistent with the 44.4% prevalence among malaria patients in the 1997 survey 12 and with recent evidence of an association between HIV infection and malaria. 16–19

A relatively low prevalence was found among patients with cancer, including cervical carcinoma, which is comparable to the results of the 1997 survey, in which only 1 of the 13 persons with cervical carcinoma was HIV positive. 12 In Africa, the occurrence of HIV-related cancers appears to be lower than in Western countries, 20 and the effect of the AIDS epidemic on Africa's cancer profile seems to be mostly limited to KS, non-Hodgkin lymphoma, and squamous cell tumor of the conjunctiva. 15,21–23

Regarding the impact on hospital services, the contribution of AIDS to morbidity and mortality in the Medical Ward was moderate in 1999, and this contribution has been quite stable since 1992. 24 However, when considering all HIV-positive patients, these accounted for >1/3 (37.2%) of the bed-days in the medical ward in 1999, thus indicating that the hospital plays an important role in delivering health care to all HIV-infected patients and not only to AIDS patients. Although using the number of bed-days as an indicator could result in an overestimation of the impact because it also includes HIV patients hospitalized for non-HIV-related diseases, 25 it could be useful in defining hospital priorities and sustainable health care strategies for all hospital patients, in planning HIV prevention activities, and in designing appropriate health care programs for HIV-infected patients.

In assessing the impact of HIV/AIDS, indirect effects must also be considered. Specifically, as a consequence of the economic crisis, social disruption, and orphanhood produced by the AIDS epidemic, there has been an increase in non-HIV-related conditions, and preventable or treatable childhood diseases were the leading causes of admissions and hospital deaths in the period from 1992–1999. 24

In interpreting this study's results, some limitations should be considered. Specifically, because the results were based on hospital records that are routinely compiled for purposes other than research, the appropriateness and the quality of the data cannot always be guaranteed. Furthermore, the accuracy of the estimates of disease-specific prevalence in the serosurvey may suffer from the small number of cases per disease. For this reason, only diseases for which >10 cases had been diagnosed were considered individually when calculating the number of bed-days attributable to HIV-positive patients.

In conclusion, the high prevalence observed even among patients not clinically diagnosed with AIDS indicates that the hospital is an important point of entry into the health care system for all HIV-infected patients. To relieve the pressure of HIV-related admissions on the overloaded African hospitals, the only means seems to be that of providing a continuum of care to HIV/AIDS patients, through inpatient, outpatient, and outreach home care services. In the general context of widening gaps between needs and resources, combining data that are readily available with data from ad hoc studies could be valuable for the purposes of health management. A major public health challenge in resource-poor countries will be that of developing innovative data collection strategies based on the integration of management information systems and clinical information systems, which are generally separate.

Back to Top | Article Outline


The authors are grateful to Mark Kanieff for editorial assistance and to Maria Grazia Dente, Patrizia Tancredi, Proscovia Akello, Alessia Ranghiasci, and Alessandro Di Vincenzo for their continuous support to the project. The authors would also like to thank Teresa Rosolen, Jacob Ouma, and Rita Caldarelli for the laboratory work and Pietro Di Mattei for the collection of data during the HIV serosurvey.

Back to Top | Article Outline


1. Armstrong J. Uganda's AIDS Crisis: Its Implication for Development. World Bank discussion paper, no. 298. Washington, DC: World Bank; 1995.
2. De Cock KM, Ekpini E, Gnaore E, et al. The public health implications of AIDS research in Africa. JAMA. 1994; 272:481–486.
3. Forsythe S, Rau B. Evolution of socioeconomic impact assessment of HIV/AIDS. AIDS. 1998; 12(suppl 2):S47–S55.
4. Brookmeyer R, Gail MH. AIDS Epidemiology. A Quantitative Approach. New York: Oxford University Press; 1994.
5. World Health Organization. WHO case definitions for AIDS surveillance in adults and adolescents. Wkly Epidemiol Rec. 1994; 69:273–275.
6. World Health Organization. International Statistical Classification of Diseases and Related Health Problems. 10th revision. Geneva: World Health Organization; 1992–1994.
7. STD/AIDS Control Programme. Surveillance of HIV and AIDS in Uganda. Entebbe: Ministry of Health; 1996.
8. World Bank. Uganda: Social Sector Strategy, Report No. 10765-UG (Eastern Africa Department). Washington, DC: World Bank; 1993.
9. Tembo G, Friesan H, Asiimwe-Okiror G, et al. Bed occupancy due to HIV/AIDS in an urban hospital medical ward in Uganda. AIDS. 1994; 8:1169–1171.
10. Kagimu M, Marum E, Natukunda E, et al. Review of AIDS Prevention and Control Activities in Uganda. Kampala: Uganda AIDS Commission; 1996.
11. Lukwiya M, Declich S, Piconi S, et al. Prevalence of HIV infection among hospital patients in northern Uganda. Italian Journal of Infectious Diseases. 1996; 2:23–26.
12. Accorsi S, Fabiani M, Lukwiya M, et al. HIV sero-survey in medical ward. In:Coping with the Impact of AIDS Epidemic in Northern Uganda: Analysis of Six Years of Activities of Lacor Hospital, Gulu (1992–97). Rapporti ISTISAN 98/18. Rome: Istituto Superiore di Sanità; 1998:57–61.
13. Morgan D, Ross A, Mayanja B, et al. Early manifestations (pre-AIDS) of HIV-1 infection in Uganda. AIDS. 1998; 12:591–596.
14. Bortolotti V, Buvé A. Prophylaxis of opportunistic infections in HIV-infected adults in sub-Saharan Africa: opportunities and obstacles. AIDS. 2002; 16:1309–1317.
15. Parkin DM, Wabinga H, Nambooze S, et al. AIDS-related cancers in Africa: maturation of the epidemic in Uganda. AIDS. 1999; 13:2563–2570.
16. Whitworth J, Morgan D, Quigley M, et al. Effect of HIV-1 and increasing immunosuppression on malaria parasitaemia and clinical episodes in adults in rural Uganda: a cohort study. Lancet. 2000; 356:1051–1056.
17. French N, Gilks CF. HIV and malaria: do they interact? Trans R Trop Med Hyg. 2000; 94:233–237.
18. French N, Nakiyingi J, Lugada E, et al. Increasing rates of malarial fever with deteriorating immune status in HIV-1-infected Ugandan adults. AIDS. 2001; 15:899–906.
19. Francesconi P, Fabiani M, Dente MG, et al. HIV, malaria parasites, and acute febrile episodes in Ugandan adults: a case-control study. AIDS. 2001; 15:2445–2450.
20. Serraino D. The spectrum of AIDS-associated cancers in Africa. AIDS. 1999; 13:2589–2590.
21. Newton R, Grulich A, Beral V, et al. Cancer and HIV infection in Rwanda. Lancet. 1995; 345:1378.
22. Sitas F, Bezwoda WR, Levin V, et al. Association between human immunodeficiency virus type 1 infection and cancer in the black population of Johannesburg and Soweto, South Africa. Br J Cancer. 1997; 75:1704–1707.
23. Chokunonga E, Levy LM, Bassett MT, et al. AIDS and cancer in Africa: the evolving epidemic in Zimbabwe. AIDS. 1999; 13:2583–2588.
24. Accorsi S, Fabiani M, Lukwiya M, et al. The increasing burden of infectious diseases on hospital services at St. Mary's Hospital Lacor, Gulu, Uganda. Am J Trop M Hyg. 2001; 64:154–158.
25. Buvè A. AIDS and hospital bed occupancy: an overview. Trop Med Int Health. 1997; 2:136–139.

disease-specific HIV prevalence; burden on hospital services; bed-days; epidemiology; resource-poor countries

Copyright © 2003 Wolters Kluwer Health, Inc. All rights reserved.