Cotrimoxazole prophylaxis has been recommended as part of the essential care and support package for symptomatic HIV-infected individuals in sub-Saharan Africa since 2000.1 This is based on 2 studies of daily cotrimoxazole prophylaxis in Abidjan, Côte d'Ivoire, showing a 43% reduction in deaths or hospitalization from any cause among persons infected with HIV 2 and a 43% reduction in all hospital admissions as well as a 46% reduction in mortality among HIV-1-seropositive adults coinfected with tuberculosis (TB).3 The greatest effects were on bacterial pneumonia, malaria, isosporiasis, acute unexplained fever, and non-typhi salmonella septicemia and enteritis. Subsequently, regular cotrimoxazole prophylaxis in different HIV-positive populations has been shown to reduce mortality by 44% among the general population in South Africa,4 19% in Malawi,5 46% in rural Uganda,6 and 29% among TB patients in South Africa,7 with corresponding reductions in morbidity, although an underpowered study in Senegal failed to demonstrate any beneficial effect of cotrimoxazole.8 Two studies in TB patients in Malawi showed reduced case fatality rates among HIV-positive patients given cotrimoxazole prophylaxis.9,10
These findings were observed in areas of sub-Saharan Africa with different patterns of bacterial resistance to cotrimoxazole and background disease prevalence. Cotrimoxazole resistance was reported in 0% of Streptococcus pneumoniae, 14% of non-typhi salmonella, and 53% of Shigella isolates in Abidjan2,3 but 46% for non-typhi salmonella isolates in Kenya.11 There were concerns that widespread use of cotrimoxazole chemoprophylaxis might increase resistance of Plasmodium falciparum to sulfadoxine and pyrimethamine (SP),12,13 but a recent study in Mali showed that cotrimoxazole prophylaxis in children did not impair the efficacy of SP in the treatment of malaria. 14
Although an initial review was cautious about cotrimoxazole prophylaxis in low-income settings,15 subsequent WHO recommendations have advocated its use, and this is reflected in the current Ugandan policy.16
PATIENTS AND METHODS
An open cohort of HIV-seropositive adults (aged >15 years) attending 2 clinics in a semiurban area of Entebbe, Uganda, was established in 1995.17 HIV-1-infected individuals, willing to attend study clinics whenever ill, residing within 15 km of the study clinics, and giving written consent, are assessed routinely at 6 monthly visits using a standard protocol for clinical, hematological, and immunological evaluation. Disease classification is assigned according to the World Health Organization (WHO) clinical staging system.18 Cohort members have access to free outpatient care and subsidized hospital care whenever ill but have to pay their own transport costs.
In August 2000, although data suggested that cotrimoxazole would be effective, a study was considered necessary to confirm efficacy in this setting of high bacterial resistance. For ethical reasons, a randomized, placebo-controlled trial of cotrimoxazole prophylaxis was not used, but mortality and morbidity rates in the cohort after its introduction were compared with historical rates in the same cohort.
All cohort members were given information on cotrimoxazole and interviewed. Those who gave verbal consent were enrolled on cotrimoxazole prophylaxis at the first routine visit after August 1, 2000. Cohort members were not given cotrimoxazole prophylaxis if the enrolling physician felt that they were unable to adhere to the appointment schedule. Enrolled participants received labeled containers with 60 cotrimoxazole tablets (sulfamethoxazole 400 mg/trimethoprim 80 mg), were told to take 2 tablets daily, and were given monthly refill appointments. Drug-related side effects were monitored, and after the first month, 48% of participants were given 2 monthly appointments to reduce their monthly transport costs.
At each refill appointment, participants were asked about medical complaints. If side effects or other abnormalities were suspected, the physician assessed whether to stop cotrimoxazole. Attendance at scheduled visits was used to measure compliance, and field workers visited participants who did not attend clinic within 2 weeks of the appointment date to establish the reason for defaulting. Every 6 months, full blood counts were made, including white cell, CD4, and granulocyte counts.
To assess the impact of cotrimoxazole prophylaxis on mortality rates, an open cohort was used (Fig. 1) comparing 12 months before the intervention (August 1999 to July 2000) with 12 months when most participants were receiving cotrimoxazole prophylaxis (March 2001 to February 2002). A second analysis using 18-month periods (February 1999 to July 2000 and March 2001 to August 2002) was conducted to check the robustness of the mortality estimates. An intention-to-treat approach was used, excluding cohort members who had not attended the clinic for any regular 6-month check-up during the study periods. Incidence rate ratios (IRRs), 95% confidence intervals (95% CIs), and 2-sided P values were used to compare the 2 periods. Poisson regression was used to adjust for age, sex, and CD4 counts in a multivariate analysis. For the postintervention period, participants were grouped into 3 categories according to attendance at monthly scheduled visits for cotrimoxazole refills: good attenders (at least 75% attendance), attenders (any attendance, including good attenders), and nonattenders (no recorded attendance). For subjects who died or joined the cohort during the observation periods, the available scheduled revisits were for the period when they were in the cohort only. Each group of attenders was compared with the whole cohort during the preintervention period.
The morbidity analysis was confined to the closed subset of cohort members who started cotrimoxazole prophylaxis between August 2000 and February 2001 and who had morbidity data for the 12 months before and 12 months after starting cotrimoxazole prophylaxis. During this time, participants who visited the clinic with a fever (temperature, ≥37.5°C) were examined by a physician who made a diagnosis (primary outcomes) based on reported symptoms, observed signs, and the results of laboratory investigations. Less well-defined morbid events (secondary outcomes) were recorded when participants attended The AIDS Support Organisation (TASO) clinic, where reported symptoms and simple signs were recorded but there was no laboratory confirmation of the diagnosis. Both primary and secondary outcomes were summed over the 12-month periods to obtain all-cause and specific morbidity rates. Each participant was considered their own control, and a paired analysis was performed comparing the 2 periods using generalized estimating equations (GEEs) with a Poisson distribution and adjusting for age and CD4 counts. CD4 counts and body weight changes were compared using a paired t test, between the 2 periods.
During 1995 to 1998, antibiotic sensitivities of pneumococci and non-typhi salmonellae were assessed according to the British Society of Antimicrobial Chemotherapy guidelines. From August 2000, cotrimoxazole sensitivity was tested on clinical specimen isolates from all cohort participants, using standardized discs (Oxoid, Basingstoke, UK) according to the method of the British Society of Antimicrobial Chemotherapy. Test performance was monitored by the use of control organisms Escherichia coli NCTC 10418, Haemophilus influenzae ATCC 49247, and S. pneumoniae ATCC 49619. Plasmodium resistance to cotrimoxazole was not measured.
Data were collected by manual recording of information in a standardized format. Double data entry and validation were performed using databases created in FoxPro (Microsoft, Redmond, WA). Analyses were carried out using STATA statistical software, version 8.0 (Stata Corp, College Station, TX).
Between August 1, 2000 and February 28, 2002, there were 1268 registered cohort participants, 1031 (81.3%) attended at least 1 routine clinic visit, 859 (83.3%) were assessed for cotrimoxazole prophylaxis, and 811 (94.4%) enrolled. Of the 48 not enrolled, 28 were excluded because of high travel expense, 8 for adherence concerns, and 7 for possible allergy to cotrimoxazole; 3 refused; and 2 were too busy.
By the end of February 2002, 5099 (60.9%) of 8371 scheduled revisits had been attended. Only 158 (20.8%) of participants attended all available scheduled revisits, 325 (42.9%) attended at least 75%, and 489 (64.5%) attended at least half. The main reasons for default were logistical, involving lack of money or transport, mobility of participants, and sickness (Table 1). Of those not interviewed after defaulting, 85 were absent from their homes, 38 had died, and 22 could not be found at the recorded address.
Fifty-three participants reported adverse events, of which 29 (3.8% of participants) were regarded as cotrimoxazole related. Seventeen were dermatological (itching or rash), 6 were constitutional (weakness, palpitations, sweating, or headache), 4 were gastrointestinal, 1 had recurrent oral sores, and 1 bruised easily after minor injury. Drug use was discontinued in 22 cases.
Cotrimoxazole suppresses bone marrow function, so we also monitored granulocyte and total white cell counts as measures of toxicity in the morbidity cohort. The mean granulocyte count fell significantly during the intervention period compared with the preintervention period (3.15 to 3.13 × 109/L in the preintervention period [mean change, −0.02 × 109/L] compared with 3.05 to 2.43 × 109/L in the intervention period [mean change, −0.62 × 109/L]; P < 0.001). The mean total white cell count fell significantly during the intervention period compared with the preintervention period (mean change, +0.08 × 109/L in the preintervention period compared with −0.88 × 109/L in the intervention period; P < 0.001). Five participants (1.4%) had dangerously low levels of granulocytes (<1 × 109/L) or total white cells (<2 × 109/L) at the end of the period before cotrimoxazole administration compared with 10 (2.8%) at the end of the cotrimoxazole intervention period. This difference was not statistically significant (P = 0.2).
The characteristics of the participants in the 2 groups for the mortality analysis showed no significant differences in age, sex, WHO clinical stage, and CD4 counts (Table 2). The crude mortality rate was 22.5 per 100 person-years for the period before cotrimoxazole and 18.2 per 100 person-years for the intervention period, giving a crude IRR of 0.81 (95% CI, 0.64-1.02; P = 0.07). After adjusting for age and CD4 count (as a continuous variable), the IRR was 0.76 (95% CI, 0.60-0.96; P = 0.020). Results of analysis comparing 18-month periods were similar with an adjusted mortality IRR of 0.74 (95% CI, 0.61-0.89; P = 0.001). The mortality in excluded subjects was compared to see if background levels of mortality had changed because of causes external to the trial. The mortality rates were similar: 14.4 per 100 person-years among the 219 excluded in the comparison period and 16.8 per 100 person-years among the 281 excluded in the intervention period, giving a mortality IRR of 1.17 (95% CI, 0.69-1.98).
Analysis by attendance group gave an adjusted mortality IRR of 0.44 (95% CI, 0.29-0.67) for 301 good attenders, 0.52 (95% CI, 0.39-0.69) for 654 who attended any scheduled refill appointment, and 1.54 (95% CI, 1.15-2.05) for the 282 attending no scheduled appointments. Mortality IRR for participants with a CD4 count at enrolment of less than 200 cells/μL was 0.80 (95% CI, 0.61-1.04; P = 0.09) and 0.76 (95% CI, 0.44-1.32; P = 0.3) for participants with a CD4 count at enrollment of 200 cells/μL or more.
Morbidity data are available for primary outcomes in 353 participants, all of whom were 1 year older for the second period of observation (Table 3). In the first period, there were 115 participants with CD4 more than 500 cells/μL, 160 between 200 and 499 cells/μL, and 78 with less than 200 cells/μL, and by the second period, 101, 161, and 91 for each of the 3 categories. Overall, there was no difference in primary outcomes (febrile events), but there was a 69% reduction in the incidence of malaria and a 109% increase in diarrhea associated with fever in the intervention period. There was a large difference in the effect of the intervention by CD4 count. In participants with a CD4 count of less than 200 cells/μL, there was an increase of 1.33 (95% CI, 0.97-1.82; P = 0.07) in the overall primary morbidity rate, whereas in subjects with CD4 count between 200 and 499 cells/μL, the IRR was 0.80 (95% CI, 0.53-1.20; P = 0.3), and in subjects with CD4 count 500 cells/μL or more, the IRR was 0.27 (95% CI, 0.07-0.94; P = 0.04).
Morbidity data for 266 participants seen at the TASO clinic showed no difference in secondary outcomes before and after the introduction of cotrimoxazole prophylaxis (Table 4). After adjustment for age and CD4 count, there was a 63% reduction in boils and increases in reported fever (31%), herpes zoster (129%), and oral thrush (191%). There was no significant difference in the effect of the intervention on secondary outcomes by CD4 count (data not shown).
In study participants, the mean weight change was +0.79 kg in the preintervention period compared with +0.24 kg after the introduction of cotrimoxazole (t = 1.39, P = 0.16). Twelve participants (3.4%) lost 10% or more body weight during the period before cotrimoxazole administration compared with 15 (4.4%) after its introduction. Mean CD4 cell declines were more pronounced after the introduction of cotrimoxazole than before (before, 440-414 cells/μL [change, −26 cells/μL]; after, 411-327 cells/μL [change, −84 cells/μL]; t = 6.11, P < 0.001).
Before this study began, 52% of isolates of S. pneumoniae and 44% of non-typhi salmonellae were fully resistant to cotrimoxazole. From isolates collected from August 2000, the prevalence of bacterial resistance to cotrimoxazole was 89% for S. pneumoniae, 60% for non-typhi salmonella, 98% for H. influenzae, and 90% for Shigella. There was no difference in resistance patterns in samples taken from patients enrolled or not enrolled on cotrimoxazole prophylaxis.
This study successfully enrolled more than 90% of potential participants, and more than 60% of scheduled appointments to receive supplies of cotrimoxazole were attended. This shows that a program for cotrimoxazole prophylaxis is feasible in a semiurban setting in Uganda. The main reason for not enrolling or defaulting from clinic visits was lack of affordable or available transport to collect supplies of the drug. Although many participants attended clinic to refill their drug supply every 2 months, a wider availability of cotrimoxazole would reduce the need to attend clinic and may help improve adherence rates. Few participants experienced drug-related adverse events, and serious leukopenia or granulocytopenia was rare among participants on cotrimoxazole.
Overall mortality was assessed on all eligible patients during a 1-year period before and after the cotrimoxazole was introduced. The 24% decrease in mortality is similar to previous reports,2-7 with a 56% reduction in mortality among those attending 75% of scheduled visits and a 48% reduction in those attending any scheduled visits. The results of this study reflect the expected effectiveness of a large-scale program of cotrimoxazole prophylaxis for HIV-positive patients. For every 23.3 patients given prophylaxis for a year, one life would be saved, and with 1-year supply of cotrimoxazole costing US$6, excluding all operational costs, the cost per life-year saved was US$139.5.
Cotrimoxazole was associated with a 69% reduction in incidence of malaria, which agrees with other studies in Abidjan2 and Uganda,6 where malaria is endemic. Although resistance to SP in Uganda is high, there is no evidence that cotrimoxazole prophylaxis would accelerate this trend of increasing resistance to SP,13 although we did not measure this in our study. If wide-scale cotrimoxazole prophylaxis is implemented, then there is a need to monitor SP resistance. Prevention of malaria may slow the decline ofCD4 cells in HIV positive patients, and some suggest that this is a route by which cotrimoxazole reduces mortality.6,19 However, in this study, CD4 cell decline was greater in the period when patients were receiving cotrimoxazole prophylaxis than in the period before its introduction. This suggests that the benefit may be through a more direct action on infections.
There was a significant reduction in boils during the intervention period, which may be due to the effect of cotrimoxazole on staphylococcal skin infections. However, there was an increased incidence of oral thrush, herpes zoster, reported fever, and diarrhea with fever, so that, overall, there was no change in the rate of acute morbid events. The increase in oral thrush could be due to cotrimoxazole use, as could the increase in reported symptoms of fever and diarrhea with fever. Some of the observed increase in morbid events may be due to the progression of disease in HIV-positive patients observed during the 2-year period, although the increase in herpes zoster may be due to chance.
For ethical reasons, this study was nonrandomized and relied on historical controls to assess the effect of cotrimoxazole. To reduce the bias of this study design, we tried to ensure the comparability of the 2 periods for each analysis. For the mortality analysis, we used an intention-to-treat analysis, including all eligible participants from an open cohort for both periods. In the period when cotrimoxazole prophylaxis was available, this included those who received the drugs, as well as those who did not attend, thus replicating the situation that is likely to be seen in regular clinics in sub-Saharan Africa. The morbidity analysis looked at morbid events among those who survived for at least 12 months after enrolling on cotrimoxazole, thus excluding morbid events that immediately preceded death. The primary morbidity analysis was dependent on the clinical and laboratory diagnosis of febrile events, but we also analyzed other secondary morbid events from the TASO clinic. The morbidity analysis assumes that patients attended clinic for all morbid events, and it is unlikely that this changed between periods. For all the morbidity analyses, the same individuals were used for both periods (12 months before and 12 months after the introduction of cotrimoxazole), which reduces bias from different treatment-seeking behavior by different individuals or sicker individuals who may contribute more morbid events. For the analysis, GEE models were used to take into account the correlations between the nonindependent observations.
Resistance levels to cotrimoxazole in isolates collected before and during the study were high, although systematic information about bacteriologic resistance to cotrimoxazole in the full range of pathogens was limited. This may suggest that the effect of cotrimoxazole was not primarily because of the prevention of bacterial infections. However, the sensitivity of pathogens to cotrimoxazole in vitro may not represent the efficacy of cotrimoxazole in preventing infections in vivo.
The effect of cotrimoxazole on mortality among persons with HIV in Africa is now well documented.2-7 However, the mechanism of the effect is uncertain and is not explained by a reduced decline in CD4 counts, as CD4 decline was significantly increased in those receiving cotrimoxazole in this study. In contrast to other studies,3,8 in our study, the effect of cotrimoxazole on mortality was similar, regardless of CD4 count. However, the power of our study to detect such a difference was low, and so these results should be judged with caution. We did find greater reductions in rates of febrile morbid events among participants with higher CD4 counts, which suggests that initiation of prophylactic cotrimoxazole should not be delayed in subjects with symptomatic HIV infection and is beneficial for all persons with HIV, regardless of CD4 cell count or clinical stage. We speculate that the main effect of cotrimoxazole might have been on infections caused by sensitive bacterial pathogens, on preventing deaths associated, directly or indirectly, with malaria, or on unrecognized atypical presentations of Pneumocystis jiroveci pneumonia.
The results from this study should encourage health planners and policy makers to proceed with large-scale care programs including the distribution of cotrimoxazole. Most participants could be enrolled onto cotrimoxazole prophylaxis, and the drug was well tolerated. The major barrier to compliance was the cost of transport, and this could be helped by ensuring that cotrimoxazole was widely available in the community or that enough drug was provided to patients to cover a longer period of time. Cotrimoxazole prophylaxis is inexpensive, safe, and effective and reduces mortality among persons with HIV in Africa, and it should be a standard component of treatment programs throughout the continent.
The authors thank all the patients who participated in the study. We also appreciate the contribution made by various staff of the Medical Research Council Programme on AIDS and TASO to this study. We acknowledge the contribution of the late Professor Daan Mulder and Professor Charles Gilks who were instrumental in setting up the Pneumococcal Vaccine Trial cohort upon which the Entebbe cohort is based.
2. Anglaret X, Chene G, Attia A, et al. Early chemoprophylaxis with trimethoprim-sulphamethoxazole for HIV-1-infected adults in Abidjan, Cote d'Ivoire. A randomised trial. Lancet
3. Wiktor SZ, Sassan-Morokro M, Grant AD, et al. Efficacy of trimethoprim-sulphamethoxazole prophylaxis to decrease morbidity and mortality in HIV-1-infected patients with tuberculosis in Abidjan, Cote d'Ivoire: a randomised controlled trial. Lancet
4. Badri M, Ehrlich R, Wood R, et al. Initiating cotrimoxazole prophylaxis in HIV-infected patients in Africa: an evaluation of the provisional WHO/UNAIDS recommendations. AIDS
5. Zachariah R, Spielmann M-P, Chinji C, et al. Voluntary counselling, HIV testing and adjunctive cotrimoxazole reduces mortality in tuberculosis patients in Thyolo, Malawi. AIDS
6. Mermin J, Lule J, Ekwaru JP, et al. Effect of cotrimoxazole prophylaxis on morbidity, mortality, CD4 cell count, and viral load in HIV infection in rural Uganda. Lancet
7. Grimwade K, Sturm AW, Nunn AJ, et al. Effectiveness of cotrimoxazole prophylaxis on mortality in adults with tuberculosis in rural South Africa. AIDS
8. Maynart M, Lievre L, Sow PS, et al. Primary prevention with cotrimoxazole for HIV-1-infected adults: results of the pilot study in Dakar, Senegal. J Acquir Immune Defic Syndr
9. Mwaungulu FBD, Floyd S, Crampin AC, et al. Cotrimoxazole prophylaxis reduces mortality in human immunodeficiency virus-positive tuberculosis patients in Karonga district, Malawi. Bull World Health Organ
10. Boeree MJ, Sauvageot D, Banda HT, et al. Efficacy and safety of two dosages of cotrimoxazole as preventive treatment for HIV-infected Malawian adults with new smear-positive tuberculosis. Trop Med Int Health
11. Kariuki S, Gilks C, Corkhill J, et al. Multi-drug resistant non-typhi Salmonella in Kenya. J Antimicrob Chemother
12. Iyer JK, Milhous WK, Cortese JF, et al. Plasmodium falciparum
cross resistance between trimethoprim and pyrimethamine. Lancet
13. Kilian AHD, Jelinek T, Prislin I, et al. Resistance in vivo of Plasmodium falciparum
to cotrimoxazole in western Uganda. Trans R Soc Trop Med Hyg
14. Thera MA, Sehdev PS, Coulibaly D, et al. Impact of trimethoprim-sulfamethoxazole prophylaxis on falciparum
malaria infection and disease. JID
15. Grimwade K, Gilks C. Cotrimoxazole prophylaxis in adults infected with HIV in low-income countries. Curr Opin Infect Dis
16. Ministry of Health, Republic of Uganda. National policy guidelines for cotrimoxazole prophylaxis for people with HIV/AIDS. April 2005.
17. French N, Nakiyingi J, Carpenter LM, et al. 23-Valent pneumococcal polysaccharide vaccine in HIV-1-infected Ugandan adults: double-blind, randomised and placebo controlled trial. Lancet
18. The WHO International Collaborating Group for the Study of the WHO Staging System. Proposed "World Health Organisation staging system for HIV infection and disease": preliminary testing by an international collaborative cross-sectional study. AIDS
19. Mermin J, Lule JR, Ekwaru JP. Association between malaria and CD4 cell count decline among persons with HIV. JAIDS