Bulletin of the World Health Organization

Effect of cotrimoxazole on mortality in HIV-infected adults on antiretroviral therapy: a systematic review and meta-analysis

Amitabh B Suthar a, Reuben Granich b, Jonathan Mermin c & Annelies Van Rie a

a. Department of Epidemiology, University of North Carolina, Chapel Hill, United States of America (USA).
b. Department of HIV/AIDS, World Health Organization, avenue Appia 20, 1211 Geneva 27, Switzerland.
c. Division of HIV/AIDS Prevention, Centers for Disease Control and Prevention, Atlanta, USA.

Correspondence to Reuben Granich (e-mail: granichr@who.int).

(Submitted: 11 July 2011 – Accepted: 15 September 2011 – Published online: 24 October 2011.)

Bulletin of the World Health Organization 2012;90:128-138C. doi: 10.2471/BLT.11.093260


In 2010, there were 1.8 million deaths among the 34 million people infected with the human immunodeficiency virus (HIV).1 Of these deaths, 1.2 million occurred in the 22.9 million HIV-infected Africans.1 Antiretroviral therapy (ART) has transformed HIV infection into a manageable chronic condition and the World Health Organization (WHO) currently recommends ART for individuals with a CD4+ T lymphocyte (CD4 cell) count ≤ 350 cells/µL.2 In 2010, 42% of the 9 million individuals in need of treatment globally were receiving ART.3 Unfortunately, in low-income countries patients are started on ART at lower CD4 cell counts than in high-income countries: in 2006, the median count was reported to be 108 cells/µL and 234 cells/µL in these two types of countries, respectively.4 Moreover, after adjusting for immunodeficiency at baseline, mortality during the first months of ART was higher in low-income countries than in high-income countries.4,5 High mortality rates early in ART have been documented in Africa,6 the Caribbean,7 Latin America7 and south-eastern Asia.8,9 Consequently, additional interventions are needed to decrease early mortality during ART in low- and middle-income countries.

Cotrimoxazole contains two antibiotics: sulfamethoxazole and trimethoprim. Cotrimoxazole provides good coverage against gram-positive bacteria (e.g. Streptococcus pneumoniae), gram-negative bacteria (e.g. Escherichia coli and non-typhoid Salmonella), protozoa (e.g. Isospora belli, Toxoplasma gondii and Plasmodium falciparum) and fungi (e.g. Pneumocystis jirovecii). The patent for cotrimoxazole has expired and it costs around 7 United States dollars for 1 year of daily therapy.10 In high-income countries, cotrimoxazole is used in adults with HIV infection as chemoprophylaxis against P. jirovecii pneumonia and T. gondii infection.11 Randomized trials that included ART-naïve Africans found that cotrimoxazole improved survival while reducing the risk of malaria, pneumonia, sepsis, isosporiasis, T. gondii encephalitis, wasting and Kaposi’s sarcoma.1214 However, a Senegalese trial that used half the recommended adult cotrimoxazole dose reported no mortality benefit in ART-naïve adults.15 A study in Uganda showed that treating HIV-infected adults with cotrimoxazole and ART reduced mortality in their uninfected children and the number of orphans.16

In settings where the health-care infrastructure is limited, WHO recommends cotrimoxazole for adults with WHO clinical stage 2, 3 or 4 HIV infection.17 If the prevalence of HIV infection is high, however, WHO recommends that all infected adults be treated because cotrimoxazole reduces morbidity irrespective of clinical disease stage or CD4 cell count and because it simplifies cotrimoxazole distribution.17

Common causes of mortality in adults receiving ART in low- and middle-income countries include sepsis, tuberculosis, Crytptococcus neoformans meningitis, T. gondii encephalitis, P. jirovecii pneumonia, Kaposi’s sarcoma and chronic diarrhoea.6,1820 Given the results of earlier trials,1214 cotrimoxazole may decrease both mortality and morbidity in adults with HIV infection, regardless of ART status. The aim of this study was to carry out a systematic review of the effect of cotrimoxazole on mortality and morbidity in individuals aged 13 years or more who were receiving ART for an HIV infection.


This systematic review was conducted in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines.21 The PubMed and Embase databases were searched systematically in December 2010 for randomized controlled trials and prospective and retrospective cohort studies on the effect of daily cotrimoxazole in HIV-infected individuals aged 13 years or more who were receiving ART. The search strategies (Table 1, available at: http://www.who.int/bulletin/volumes/90/2/11-093260) were designed with a specialist librarian and there were no language, publication or date restrictions. The primary outcome of interest was death. Secondary outcomes were hospitalization, incident events synonymous with WHO clinical stage 3 or 4 HIV disease, incident malaria events and adverse events leading to hospitalization or cotrimoxazole cessation. The Cochrane Central Register of Controlled Trials and the International Standard Randomized Controlled Trial Number Register were searched using the terms “cotrimoxazole” and “antiretroviral therapy”.

The abstracts of all publications identified were reviewed independently by two of the authors (ABS and AVR). The full texts of all articles selected by one or both reviewers were then matched against the inclusion criteria and articles satisfying these criteria were included in the review (Fig. 1, available at: http://www.who.int/bulletin/volumes/90/2/11-093260). The references of all articles that met the inclusion criteria were also considered for incorporation into the systematic review.

Fig. 1. Flow diagram of article selection for systematic review of cotrimoxazole’s effect on mortality and morbidity in adults with human immunodeficiency virus (HIV) infection receiving antireotroviral therapy (ART)
Fig. 1. Flow diagram of article selection for systematic review of cotrimoxazole’s effect on mortality and morbidity in adults with human immunodeficiency virus (HIV) infection receiving antireotroviral therapy (ART)
WHO, World Health Organization. a One additional study was identified during the review of references for the eight studies meeting eligibility criteria.

Data were extracted from the articles included in the review using a standardized spreadsheet. The information collected included the name of the article’s first author, the year of publication, the study’s methods and design, the study population, the study intervention and control intervention, the duration of follow-up, inclusion and exclusion criteria, primary and secondary outcomes, and losses to follow-up.

In accordance with the recommendations of the Cochrane Collaboration,22 the Newcastle–Ottawa Quality Assessment Scale was used to identify bias in the cohort studies.23 This scale rates studies from 0 to 9 using eight criteria that cover three sources of bias: selection, confounding and measurement bias. Each criterion is worth one point except confounding bias, which is worth two points. Selection bias was assessed using four criteria: (i) the patient cohort receiving cotrimoxazole and ART was representative of adults receiving this treatment in the community from which the cohort was drawn; (ii) the cohort receiving ART alone was representative of the cohort on both cotrimoxazole and ART; (iii) cotrimoxazole use had been ascertained; and (iv) the majority of study participants were known to be asymptomatic (i.e. had WHO clinical stage 1 or 2 disease) at study baseline. A study was regarded as having addressed confounding due to the baseline level of immunodeficiency if the results had been adjusted for the baseline CD4 cell counts. Measurement bias was assessed using three criteria: (i) all deaths were validated; (ii) there was adequate follow-up to detect cotrimoxazole’s preventive effect on death (i.e. a median or mean follow-up period of at least 3 months), and (iii) ≤ 20% of study participants were lost during follow-up. A score of 7 to 9 indicated high methodological quality, a score of 4 to 6 indicated moderate quality, and a score of 0 to 3 indicated low quality.

Publication bias was evaluated using a funnel plot, with the treatment effect measure on the x-axis and the standard error of the log of the effect measure on the y-axis (details are available from the corresponding author on request). Egger’s and Begg’s tests were used to assess the symmetry of the funnel plot. Since the studies were similar enough to combine, a meta-analysis was performed and statistical heterogeneity was assessed. Effect measures were quantified as the natural log of the effect measure and the standard error24 as:

Since both χ2 and tau statistical tests of the heterogeneity of the magnitude of a treatment effect across studies depend on knowledge of the number of events in each study arm and this information was not available for all studies that met the inclusion criteria, an I-squared statistic was used. In practice, the I-squared statistic is calculated by subtracting the number of degrees of freedom from the Q-statistic and then dividing the result by the Q-statistic.25 Values of I-squared less than 25% indicate low heterogeneity, values near 50% indicate moderate heterogeneity and values above 75% indicate high heterogeneity.26 A random-effects statistical model is preferred when I-squared is ≥ 25%.27

Since the value of I-squared obtained for the studies included in this analysis was greater than 25%, a random-effects model was used to derive a summary estimate of the effect of cotrimoxazole on mortality. The possible causes of the observed heterogeneity, including the duration of follow-up and the percentage of participants who had WHO stage 3 or 4 disease at baseline, were explored by a sensitivity analysis. Unfortunately, neither the baseline viral load nor the baseline CD4 cell count could be included in the sensitivity analysis because data on the baseline viral load were not collected in most studies and data on the median or mean baseline CD4 cell count were collected in only four of the seven eligible studies reporting on the primary outcome. All analyses were performed using Stata version 10.0 (StataCorp. LP, College Station, United States of America).


Details of the study participants and study designs described in the nine articles included in the systematic review are shown in Table 2 and Table 3, respectively. Of the nine, seven reported an estimate for the effect of cotrimoxazole on mortality,2834 whereas three reported an estimate for the effect on morbidity.28,35,36 Two studies also reported the preventive effect on malaria.28,36 Assessment of publication bias indicated that three studies2830 were of high methodological quality, while the other six3136 were of moderate methodological quality (Table 4, available at: http://www.who.int/bulletin/volumes/90/2/11-093260). In addition, one on-going trial investigating when to stop cotrimoxazole during ART was identified.37 That trial involves 2000 individuals in Uganda and has a randomized, double-blind, placebo-controlled non-inferiority design. Its aim is to test the hypothesis that stopping cotrimoxazole in adults on ART will not increase morbidity or mortality and will decrease haematological adverse events. The results are due in 2014.

The effect of cotrimoxazole on mortality reported in the seven studies in the meta-analysis is illustrated in Fig. 2. Begg and Egger P-values for publication bias in the studies were 0.29 and 0.49, respectively, which suggests the absence of publication bias. The summary estimate of the effect of cotrimoxazole on the incidence rate of death derived using a random-effects model was 0.42 (95% confidence interval, CI: 0.29–0.61). Survival was increased by either continuing cotrimoxazole at ART initiation,29,3335 initiating cotrimoxazole at the same time as ART29,34 or initiating cotrimoxazole when the patient was stable on ART.28,30

Fig. 2. Forest plot of study and summary effect estimates for cotrimoxazole’s effect on death in adults with human immunodeficiency virus (HIV) infection receiving antiretroviral therapy (ART)
Fig. 2. Forest plot of study and summary effect estimates for cotrimoxazole’s effect on death in adults with human immunodeficiency virus (HIV) infection receiving antiretroviral therapy (ART)
Note: The seven studies were included in the meta-analysis. The summary estimate of the treatment effect was calculated using a random-effects model. The risk is reported either as an odds ratio, a hazard ratio or a relative risk in the different studies. The centre of the squares represents the study point estimates while the horizontal lines represent the study 95% confidence intervals (CIs). The centre of the bottom quadrilateral represents the summary effect estimate, whereas its horizontal corners represent its 95% CI.

The Q-statistic and the I-squared statistic for the seven studies reporting on the primary outcome were 102.3 and 93.2%, respectively, indicating high heterogeneity.26 A sensitivity analysis that evaluated the effect of the median or mean number of years of follow-up indicated that the length of the study did not explain the heterogeneity (P = 0.85). Nor was the percentage of participants with symptomatic HIV infection (i.e. WHO stage 3 or 4 disease) at baseline found to provide an explanation on sensitivity analysis (P = 0.91).

Regarding cotrimoxazole’s effect on morbidity, Walker et al. reported that cotrimoxazole resulted in a reduction in the odds of malaria (odds ratio, OR: 0.74; 95% CI: 0.63–0.88).28 This effect was maintained throughout a follow-up period of 5 years. In addition, Walker et al. also reported that cotrimoxazole reduced the odds of new or recurrent WHO clinical stage 3 or 4 disease events (OR: 0.85; 95% CI: 0.74–0.98),28 whereas Miiro et al. reported a trend towards lower all-cause morbidity (relative risk, RR: 0.66; 95% CI: 0.41–1.06), although all-cause morbidity was undefined.35 A trial identified during the review of references of studies that met the inclusion criteria found that the risk of both malaria (RR: 0.04; 95% CI: 0.01–0.17) and diarrhoea (RR: 0.56; 95% CI: 0.43–0.77) was lower among Ugandans on ART who were randomized to cotrimoxazole than among those who were not.36 The patients had been receiving ART for a mean of 3.7 years, their median CD4 cell count was 489 cells/µL and 94% had a viral load < 400 RNA copies/mL.

Two studies gave details of adverse events associated with cotrimoxazole. Walker et al. reported 22 serious cotrimoxazole-related adverse events during 8128 person–years of treatment: all were either haematological adverse events, rash or hypersensitivity.28 Lowrance et al. reported that 10 of 574 patients on cotrimoxazole stopped treatment during 6 months of follow-up but did not state whether or not the cause was cotrimoxazole-related toxicity.30

Since only two of the eight studies followed participants for more than 13 months on average, it was difficult to gauge whether or not the survival benefit of cotrimoxazole waned after ART-induced immune reconstitution. However, Walker et al. did analyse the effect of the duration of cotrimoxazole combined with ART.28 They found that cotrimoxazole was associated with a substantial reduction in the odds of death between weeks 1 and 12 of the administration of cotrimoxazole combined with ART (OR: 0.41; 95% CI: 0.27–0.65) and that the reduction was sustained between weeks 12 and 72 (OR: 0.56; 95% CI: 0.37–0.86) but ceased to be evident after week 72 (OR: 0.96; 95% CI: 0.63–1.45). There was no evidence of variation in this effect resulting from an updated CD4 cell count.

It was also difficult to determine whether the survival benefit of cotrimoxazole was influenced by the baseline CD4 cell count since in six studies most participants initiated ART with a CD4 cell count of less than 200 cells/µL. However, Hoffmann et al. reported that the reduction in the hazard of death was substantial among individuals with a baseline CD4 cell count less than 200 cells/µL (hazard ratio, HR: 0.64; 95% CI: 0.56–0.72) and those with a baseline count between 200 and 350 cells/µL (HR: 0.62; 95% CI: 0.41–0.94) but was not significant among those with a count greater than 350 cells/µL at baseline (HR: 0.80; 95% CI: 0.38–1.70).29


This systematic review indicates that cotrimoxazole reduced mortality in individuals aged 13 years or more who were receiving ART. The summary estimate of the effect of cotrimoxazole on the incidence rate of death was 0.42 (95% CI: 0.29–0.61). Although there was no evidence of publication bias in the studies included in the review, there was significant heterogeneity in the findings: the I-squared statistic was 93.2%. Unfortunately, only a limited sensitivity analysis could be performed because of differences in the variables recorded in the various studies. Nonetheless, the sensitivity analysis showed that neither symptomatic HIV infection (i.e. WHO stage 3 or 4 disease) nor the duration of follow-up explained the heterogeneity.

Cotrimoxazole is safe, well tolerated, widely available and inexpensive. Although cotrimoxazole can help reduce the high early mortality rate in HIV-infected adults on ART in low- and middle-income countries, it is still not widely used. The slow increase in the uptake of cotrimoxazole has been associated with delays in the dissemination of recommendations on its use (either in stand-alone guidelines or integrated into ART guidelines), problems with drug procurement and supply, poor health-care infrastructure for managing patients before ART and inadequate systems for monitoring and evaluation.30 In addition to resolving these issues, the number of adults receiving cotrimoxazole could also be increased by raising awareness of its benefits and by using indicators to monitor its uptake both globally and at the level of individual treatment programmes.38 Moreover, treatment programmes could also provide forecasts of the future annual demand for cotrimoxazole by using local data on the number of individuals currently receiving or expected to start ART.30

Recently WHO and the Joint United Nations Programme on HIV/AIDS, as part of the Treatment 2.0 initiative,39 prioritized identifying, retaining and caring for people earlier in the course of HIV infection as a way of improving clinical and programmatic outcomes. Indeed, implementing ART earlier within existing guidelines has had a considerable impact: in South Africa, mortality during the first year of ART declined from 8.9 to 5.6% as the median CD4 cell count at ART initiation increased from 68 to 113 cells/µL.40 One way to facilitate earlier access to ART is through expanding HIV testing coverage. For generalized epidemics (i.e. when the prevalence of antenatal HIV infection is over 1%), WHO recommends carrying out provider-initiated HIV testing in all health facilities,41 particularly in patients with tuberculosis.42 Recent evidence suggests that community-based testing can identify individuals earlier in the course of HIV infection43 and increases knowledge of HIV status fourfold relative to health-facility-based testing.44 Reductions in mortality, costs and the rate of new infections will also depend crucially on maintaining good links with personnel monitoring and caring for individuals before they receive ART, on maximizing the retention rate and on initiating ART as early as permitted by national guidelines. These actions will help achieve the Millennium Development Goals on HIV infection.45 Unfortunately, at present over 50% of HIV-infected individuals are lost to care between diagnosis and the start of ART.4649 Giving free cotrimoxazole can help increase the retention rate,50 provide an opportunity to assess an individual’s adherence to treatment before the start of ART51 and improve survival in those not on ART.52 Efforts to reduce early mortality in HIV-infected individuals and to increase retention rates in treatment programmes could be assisted by implementing intensified tuberculosis case-finding at every health-care visit and by providing free isoniazid to those with tuberculosis who do not have a cough, night sweats, weight loss or fever.53

For settings in which the health-care infrastructure is limited, WHO recommends discontinuing cotrimoxazole in adults who show good adherence to ART, have secure access to treatment and who have not had a new WHO stage 2, 3 or 4 disease event for at least 1 year.17 Data from the one study that reported the estimated effect of cotrimoxazole on mortality after 1 year of combined treatment with ART indicate that there was no reduction in mortality after 72 weeks of ART.28 For settings where the health-care infrastructure is good, WHO recommends discontinuing cotrimoxazole in adults who have received ART for at least 6 months and have a CD4 cell count greater than 350 cells/µL.17 The single study that provided an estimate of the effect of cotrimoxazole on mortality in adults on ART with a baseline CD4 cell count greater than 350 cells/µL suggested that cotrimoxazole had no effect.29

Two limitations of this systematic review were that most studies included had a short follow-up and that they did not include estimates of the effect of cotrimoxazole in adults with a high baseline CD4 cell count. Furthermore, only one randomized trial was identified. Although most studies did attempt to control for bias, prospective and retrospective cohort studies are susceptible to unmeasured confounding. Also, none of the studies assessed adherence to cotrimoxazole and ART. Finally, since the cause of the death was not reported in any study, the precise mechanism of cotrimoxazole’s beneficial survival effect is not clear.

Although the data considered in this review were limited and results from an on-going trial are still awaited, our findings support current WHO recommendations that the use of cotrimoxazole should be scaled up in HIV-infected individuals starting or receiving ART. Further research is needed to determine the optimum duration of cotrimoxazole treatment in adults receiving ART for an HIV infection.


Amitabh B Suthar is also affiliated with the Department of HIV/AIDS of the World Health Organization.

Competing interests:

None declared.