The Lancet 2006; 367:926-937
DOI:10.1016/S0140-6736(06)68383-9
Tuberculosis in sub-Saharan Africa: opportunities, challenges, and change in the era of antiretroviral treatment
Dr Elizabeth L CorbettPhD
a b
, Barbara MarstonMD c, Prof Gavin J ChurchyardPhD a d and Kevin M De CockMD a c
Summary
HIV and epidemiology of tuberculosis in Africa
Tuberculosis control in Africa before antiretroviral treatment
Effect of antiretroviral treatment on tuberculosis epidemiology
Towards a coordinated public-health response to tuberculosis and HIV/AIDS
Monitoring and assessment
Challenges, constraints, and future prospects
Search strategy and selection criteria
References
Summary
Rapid scale-up of antiretroviral treatment programmes is happening in Africa, driven by international advocacy and policy directives and supported by unprecedented donor funding and technical assistance. This welcome development offers hope to millions of HIV-infected Africans, among whom tuberculosis is the major cause of serious illness and death. Little in the way of HIV diagnosis or care was previously offered to patients with tuberculosis, by either national tuberculosis or AIDS control programmes, with tuberculosis services focused exclusively on diagnosis and treatment of rising numbers of patients. Tuberculosis control in Africa has yet to adapt to the new climate of antiretroviral availability. Many barriers exist, from drug interactions to historic differences in the way that tuberculosis and HIV are perceived, but failure to successfully integrate HIV and tuberculosis control will threaten the viability of both programmes. Here, we review tuberculosis epidemiology in Africa and policy implications of HIV/AIDS treatment scale-up.
Back to top
<!--start ce:displayed-quote=-->
?Nothing endures but change?
Heraclites, c535?c475 BC
<!--end ce:displayed-quote-->Although we view tuberculosis as one disease, and Mycobacterium tuberculosis alone as its cause, four epidemiological patterns can be usefully distinguished. So-called traditional tuberculosis in developing countries with low rates of HIV infection responds to well-organised control programmes.1 Tuberculosis in the industrialised world is increasingly attributable to immigration, whereas high rates of multidrug resistance threaten control in parts of eastern Europe.2 Finally, HIV-related disease has emerged as the dominant challenge in sub-Saharan Africa (hereafter referred to as Africa),3,4 with the size of the epidemic calling for a response beyond the traditional boundaries of tuberculosis control.
Increasing access to antiretroviral treatment in Africa, an unimaginable aspiration a few years ago, is now an international priority. The global importance of tuberculosis and its association with the HIV/AIDS pandemic are acknowledged by the Millennium Development Goals, and unprecedented funds are being provided by the Global Fund to Fight AIDS, Tuberculosis, and Malaria, the US President's Emergency Plan for AIDS Relief, and the World Bank's multicountry HIV/AIDS programme (figure 1).

Click to enlarge image
Figure 1. Average donor support per person with HIV/AIDS for the 15 African countries with the highest adult HIV prevalence in 2004Donations calculated from the US President's Emergency Plan for AIDS Relief, the World Bank multicountry HIV/AIDS programme for Africa, and the Global Fund for AIDS, Tuberculosis, and Malaria (excludes support for malaria). For information about donations, see:
http://www.worldbank.org/afr/aids/map_docs.htm
http://www.state.gov/s/gac/
http://www.theglobalfund.org/en/
In this Review, we discuss current understanding of tuberculosis epidemiology and control policies in Africa. The new commitment to antiretroviral treatment will greatly affect tuberculosis control, hopefully for the better but possibly for the worse.5 Patients with HIV-related tuberculosis must make up a substantial proportion of those reached by antiretroviral treatment. Important philosophical and practical changes to address HIV and tuberculosis in a coordinated manner will be needed if the increased tuberculosis incidence that has afflicted Africa during the past decade is to be reversed.
HIV and epidemiology of tuberculosis in Africa
Burden of HIV/AIDS and tuberculosis
Figure 23 summarises the disproportionate burden of HIV and tuberculosis infection and disease in Africa at the start of the new millennium.3 In 2003, an estimated 8?8 million new cases of tuberculosis resulted in 1?7 million deaths.4 27% of these cases and 31% of these deaths arose in Africa, home to only 11% of the world's population.4 HIV prevalence in tuberculosis patients is less than 1% in the Western Pacific region but 38% in Africa.3 In countries with the highest HIV prevalence, more than 75% of cases of tuberculosis are HIV-associated.4

Click to enlarge image
Figure 2. Disproportionate burden of HIV, HIV-related tuberculosis, and M tuberculosis coinfections in Africa, for 2000Every person represents 5% of the global total, with African people shown in red and the rest of the world in blue.3
In Africa, tuberculosis is often the first manifestation of HIV infection, and it is the leading cause of death among HIV-infected patients.6?11 In hospital-based series, 40?65% of HIV-infected African patients with respiratory disease had tuberculosis.6,8 In primary health and chest clinic settings, tuberculosis was confirmed in 43?70% of adults with cough for 3 weeks or longer (chronic cough) in Zimbabwe, Kenya, and Malawi.12,13 Patients with tuberculosis now commonly present with atypical symptoms: M tuberculosis was isolated from 9% of adults with acute pneumonia in Kenya,14 35% of people with cough for less than 3 weeks in Malawi,15 23% of febrile HIV-infected inpatients in Tanzania,16 and 13% of HIV-infected patients with chronic diarrhoea in Kenya.17 In Cote d'Ivoire, the Democratic Republic of Congo, and Kenya, 38?47% of autopsies in HIV-positive adults indicated tuberculosis as the cause of death,9?11 although tuberculosis had been diagnosed during life in only about half of those with autopsy-proven disease.
Increased risk for tuberculosis from HIV infection in Africa
Comparison of HIV prevalence in general populations and tuberculosis patients shows that tuberculosis incidence was 8?3 times higher in HIV-positive than HIV-negative African people in 2003 (figure 3).4,18 In 2000, similar methods led to an estimated relative rate of 5?9, whereas estimates from individual cohort studies range from less than 5 to more than 20.3 Tuberculosis incidence increases with worsening immunosuppression, so that relative rates rise during the course of an HIV epidemic.19?21

Click to enlarge image
Figure 3. Estimated prevalence of HIV in tuberculosis patientsPrevalence measured in national surveys (blue dots) and subnational surveys (red dots; data reported to WHO),4 plotted against prevalence of HIV in adults (data from UNAIDS).18 Country abbreviations available from reference 4. Figure kindly provided by Chris Dye.
Tuberculosis incidence in African countries with high HIV prevalence
Reported tuberculosis case rates rose by 6?4% per year in the WHO African region in the late 1990s,4 but with up to five-fold increases since 1990 in some countries.4 Incidence might have peaked in some countries (figure 4)18 but at very high rates. The high case rate in Africa contributed to a global rise in tuberculosis incidence of 1% in 2003, despite stable or declining rates in the rest of the world.4

Click to enlarge image
Figure 4. Time trends in tuberculosis case-notification rates in southern and east AfricaAs reported to WHO.18
Southern Africa has the highest prevalence of HIV infection and had the highest incidence of tuberculosis before the HIV/AIDS era. In the six southern African countries with adult HIV prevalence of more than 20%, tuberculosis case-notification rates are 461?719 per 100 000 per year; by comparison, the notification rate in the USA was 5 per 100 000 per year.4 True yearly rates in Africa are likely to be even higher because of underdiagnosis and under-reporting.4
HIV and infectiousness and transmission of tuberculosis
The critical period with respect to infectiousness is before diagnosis, because most patients become non-infectious soon after starting treatment, even if they are HIV-infected.22 Both intensity and duration of infectiousness are highly variable, with some individuals remaining very infectious for prolonged periods, sometimes with apparently minor symptoms.23?26 HIV-positive individuals with tuberculosis are less infectious than HIV-negative patients since they are less often and less intensely smear-positive27?30 and because they remain infectious for a much shorter average duration.23,24
The average duration of smear positivity for HIV-negative individuals in resource-poor settings is estimated to be between 1 and 3 years.3,23,24,31 In two African studies in high HIV settings,23,24 mean durations of smear-positivity of only 6 and 8 weeks were estimated for HIV-positive patients with smear-positive tuberculosis, indicating fast progression to symptomatic disease.32 In both studies, most infectious individuals at any given point in time?the driving force for tuberculosis transmission in the community?were HIV-negative, because of their fairly long duration of infectiousness (figure 5).23,24 Thus, HIV-associated tuberculosis contributes greatly to incidence of tuberculosis and deaths, but it might contribute much less to disease transmission because of early diagnosis or death.33 This combination of exquisite vulnerability to disease by HIV-positive individuals and prolonged transmission from HIV-negative patients with tuberculosis together fuel escalating tuberculosis incidence in areas of high HIV prevalence. This fundamental observation has important implications for tuberculosis control in Africa.

Click to enlarge image
Figure 5. HIV prevalence for workers and tuberculosis patients, and for all person-days with undiagnosed smear-positive tuberculosisData from South Africa23 and Zimbabwe.24 *Incident tuberculosis patients self-presenting with symptoms. ?Patients with smear-positive tuberculosis diagnosed on systematic prevalence screening.
Back to top
Tuberculosis control in Africa before antiretroviral treatment
Tuberculosis control has been based on the WHO-promoted DOTS strategy, whose philosophical basis is prompt diagnosis and effective treatment of individuals with smear-positive tuberculosis to interrupt continuing transmission.1 In the face of rising tuberculosis incidence in Africa, international guidance during the 1990s emphasised the need for tuberculosis programmes to focus on diagnosis and treatment of self-presenting patients and to improve adherence and cure rates. HIV services, such as HIV-testing, were only deemed appropriate when priority tuberculosis objectives had been met.34?36 While understandable under the circumstances and funding of the time, this approach did not serve the overall medical needs of HIV-infected patients with tuberculosis.
The HIV epidemic has challenged DOTS as a sole tuberculosis control strategy for Africa, because even rigorous programmes cannot adequately compensate for the rising susceptibility to tuberculosis at the population level that occurs as HIV prevalence increases.37?41 As an example, despite well-funded control programmes in the South African gold mining industry that adhere to all elements of WHO's DOTS strategy and screen miners every year, tuberculosis case rates have risen four-fold since 1990, driven by an increase in population HIV prevalence from less than 1% to almost 30%.42
Containing tuberculosis transmission and preventing drug resistance might be realistic and important goals for conventional DOTS programmes, but they are not routinely evaluated.43?49 Evidence that strong tuberculosis control programmes can control disease transmission is provided by analysis of incidence data stratified by HIV status.46?49Figure 650 shows two possible scenarios of an epidemic of HIV.

Click to enlarge image
Figure 6. Simulated time trends in tuberculosis incidence during the course of an HIV epidemicDeterministic compartmental model of HIV (uninfected, WHO stages 1?4) and tuberculosis infection and disease (susceptible, latent, diseased, on-treatment, post-treatment), as previously described.50 Scenarios shown of tuberculosis trends in the whole population and HIV subpopulations under either falling (A) or rising (B) tuberculosis transmission rates. (C) Time course of simulated HIV epidemic in (A) and (B). HIV prevalence refers to adults in the general population (not tuberculosis patients).
In part A, an epidemic of HIV is shown, during which slowly falling tuberculosis transmission rates (from 1% in 1985 to 0?7% per year in 2010) are maintained despite a rising burden of HIV-related tuberculosis. Disease incidence continues to decline in the HIV-negative subpopulation, but rises in the HIV-positive subpopulation as the proportion of patients with moderate-to-severe immunosuppression increases (maturation of the HIV epidemic). Overall tuberculosis incidence rises to a plateau soon after peak HIV prevalence in 2000. This scenario might describe the course of HIV and tuberculosis in countries with fairly strong tuberculosis control programmes such as Malawi and Tanzania, where peak tuberculosis incidence might already have been reached. In part B, an identical HIV epidemic to that in part A is shown, with the same assumptions about the effect of HIV on susceptibility to tuberculosis disease. However, in this scenario, the potential for HIV-related tuberculosis to increase disease transmission rates is not contained. Tuberculosis incidence rises to very high levels among HIV-positive individuals and increases substantially in the HIV-negative subpopulation as well, because of a rise in yearly risk of infection from 1% in 1985 to 2?7% in 2010. Tuberculosis incidence continues to rise after HIV prevalence peaks. Several countries, including Kenya, Uganda, and Swaziland, might be following a course intermediate between these two scenarios.
Uganda and Kenya have reported continuing rises in tuberculosis case rates despite falling HIV prevalence, associated with a noted rise in yearly risk of M tuberculosis infection in Kenya.4,18,51 Data for Cote d'Ivoire, South Africa, Malawi, and Thailand, however, show stable or declining tuberculosis incidence among HIV-negative individuals despite increasing burdens and incidence of HIV-related tuberculosis.46?49 Similarly, in Tanzania, tuberculosis transmission rates have continued to decline while case-notifications have quadrupled.45 Within the same environment, therefore, strong orthodox tuberculosis control approaches can limit or reduce HIV-negative tuberculosis but not that associated with HIV.52
Back to top
Effect of antiretroviral treatment on tuberculosis epidemiology
Mortality in HIV-infected tuberculosis patients
Antiretroviral treatment scale-up mainly aims to reduce HIV-associated morbidity and mortality. Tuberculosis case-fatality rates (proportion of patients dying while on antituberculous treatment) in Africa are 16?35% in HIV-positive individuals not receiving antiretroviral treatment and 4?9% in HIV-negative patients.53 Increased mortality in the first month of treatment seems largely attributable to tuberculosis itself,53 whereas other HIV-associated pathologies predominate thereafter.54 The highest death rates are present for people with the lowest CD4+ T-lymphocyte counts (CD4 count).55 In a 7-year follow-up in Malawi before HIV/AIDS treatment initiatives, only 11% of individuals who were HIV-infected at tuberculosis diagnosis were known to be still living.56 About a quarter of known deaths arose within 1 month of tuberculosis diagnosis.
Case-fatality rates can be reduced by diagnosis of HIV infection linked to co-trimoxazole prophylaxis57?61 and antiretroviral treatment.62?64 Mortality in tuberculosis patients in London (UK) fell by 72% after introduction of highly active antiretroviral drugs.62 Mortality in African people with HIV-associated tuberculosis is similar to that in patients with tuberculosis before effective antituberculous treatment,53 and provision of antiretroviral drugs could have as revolutionary an effect as antituberculous drugs themselves did when first introduced.
Tuberculosis incidence and recurrence in HIV-infected people
Findings of several studies from different countries show that antiretroviral drugs reduce the incidence of tuberculosis in HIV-infected people by 80% or more,65?68 with the greatest effect at the lowest CD4 counts.67 However, clinically important immune dysfunction persists even during successful antiretroviral treatment,5 and tuberculosis incidence remains well above HIV-negative rates, even at high CD4 counts.67 Rates of recurrent disease in patients with previous HIV-related tuberculosis are also high, suggesting a need for secondary preventive treatment.69,70
Tuberculosis is an aggressive opportunistic infection that arises at higher median CD4 counts than do most other AIDS-defining disorders.6 For example, median CD4 counts were 257 per μL for smear-positive patients in Cote d'Ivoire.55 Current guidelines for resource-poor settings recommend treatment for patients with symptomatic HIV or a CD4 count of 200 per μL or less.71 The potential effect of antiretroviral treatment on tuberculosis incidence, therefore, is lessened because many HIV-infected patients with tuberculosis present before antiretroviral drugs are prescribed.64
Theoretically, even well-functioning antiretroviral programmes could worsen the HIV-associated tuberculosis epidemic if an expanding cohort of patients remains highly susceptible and capable of transmitting tuberculosis for long periods.5,50 Mathematical and statistical modelling suggests antiretroviral drug coverage would have to be high, start early, and be combined with tuberculosis preventive treatment to contain disease incidence and reduce mortality.72,73
Back to top
Towards a coordinated public-health response to tuberculosis and HIV/AIDS
One strategy alone is unlikely to succeed: different approaches based on serostatus need to address the vulnerability to tuberculosis disease of HIV-infected individuals and reduce disease transmission from all affected people, including those who are HIV-negative.74 The need for a coordinated approach towards tuberculosis and HIV control is now stressed at the highest levels.74 There is understandable reluctance to relinquish the traditional disease model for tuberculosis control: past experience, notably in Zambia, shows that well-intentioned reform can disrupt essential tuberculosis control activities.75 However, different programme models are now emerging that retain DOTS as the essential but insufficient minimum,41 while additional elements discussed below are implemented or investigated in collaboration with HIV/AIDS control programmes. Essential outcomes are reduced transmission, disease, and death for both HIV and tuberculosis (panel). Seamless collaboration between tuberculosis and HIV/AIDS treatment programmes is needed, along with a unified public-health vision towards the prevention and treatment of these interacting infectious diseases.
Panel: Key interventions for improving tuberculosis and HIV control in Africa
1)Better implementation of existing policies
?Universal HIV testing of patients with confirmed or suspected tuberculosis
?Universal access to high-quality sputum microscopy for individuals with suspected tuberculosis
?Universal directly observed treatment while taking rifampicin-containing tuberculosis regimen
?Antiretroviral treatment according to national guidelines for all HIV-infected individuals, including tuberculosis patients
?Ready access to voluntary counselling and testing for HIV and condoms to prevent HIV transmission
2)More widespread implementation of additional interventions known to be effective
?Use of the most effective short-course chemotherapy for all tuberculosis patients
?Co-trimoxazole prophylaxis for all HIV-infected tuberculosis patients not taking antiretroviral drugs
?Screening of all HIV-infected individuals for tuberculosis infection and disease
?Primary isoniazid preventive treatment for all HIV-infected individuals
?Environmental measures to prevent nosocomial transmission of M tuberculosis
3)Rapid assessment of promising new approaches
?More sensitive sputum smear microscopy
?More rapid and sensitive diagnostic algorithms for smear-negative individuals with suspected tuberculosis
?Expanded access to tuberculosis culture with rapid liquid culture systems
?Secondary preventive treatment for HIV-infected individuals after successful treatment of active tuberculosis
?Early initiation of antiretroviral treatment in newly diagnosed HIV-infected tuberculosis patients
?Co-trimoxazole prophylaxis for HIV-infected tuberculosis patients taking antiretroviral drugs
?Active case-finding for tuberculosis in the community
?Promotion of universal knowledge of HIV serostatus, with emphasis on prevention of transmission from HIV-infected individuals
?Novel HIV prevention interventions
4)Increased resources and support
?Training and retention of health-care workers in joint HIV and tuberculosis management
?Increased funding for integrated tuberculosis and HIV activities
?Increased funding for tuberculosis control programmes to support HIV diagnosis and initiation of HIV care, and more rapid diagnosis and effective treatment of both HIV-positive and HIV-negative patients with tuberculosis
One barrier to closer collaboration is the philosophical difference historically in how HIV/AIDS and tuberculosis surveillance, diagnosis, and treatment have been approached.76?78 Tuberculosis control programmes have epitomised the public-health approach of case finding, name-based case notification, and, when possible, screening of contacts. Control of tuberculosis transmission and prevention of drug resistance have been paramount aims, with less emphasis on patient-centred goals such as reduction of deaths.79 By contrast, HIV/AIDS programmes have focused on an individual approach to HIV testing that is private, confidential, and voluntary, but which has little emphasis on interrupting chains of transmission.78
Currently, fewer than 10% of African patients with tuberculosis are tested for HIV,4,80 although HIV testing is acceptable to most people when provided in a convenient and confidential way:42,58,81 the major difficulty is that testing is still not routinely offered in most tuberculosis clinics. WHO and UNAIDS guidelines now lend support to diagnostic HIV testing of individuals with HIV-associated disorders, including known and suspected tuberculosis patients, using an opt-out approach,74,82 and these organisations have requested routine reporting of the uptake of HIV testing along with the numbers of notified tuberculosis cases.4
There are three goals of coordinated tuberculosis and HIV interventions: (1) to optimise diagnosis and treatment to improve outcome for all tuberculosis patients; (2) to reduce HIV-associated tuberculosis incidence and recurrence; and (3) to improve HIV and tuberculosis control overall.
Optimisation of tuberculosis diagnosis and treatment
Tuberculosis diagnosis in Africa relies on sputum microscopy followed by broad-spectrum antibiotics and chest radiography if smears are negative. Although specificity is high,12,83,84 major concerns include low sensitivity13,84 and delayed diagnosis of smear-negative disease.13,85 The accuracy of both microscopy and radiography is reduced by HIV, and so assessment of diagnostic approaches with existing methods and continuing research into new diagnostics are necessary.85?88
DOTS programmes have focused on smear-positive disease because it is the most infectious type, but much of the increased tuberculosis caseload in Africa is reported as smear-negative.4 HIV-related tuberculosis is more usually smear-negative, extrapulmonary, or disseminated than tuberculosis among HIV-negative individuals. Worsening immunosuppression correlates with increased mycobacterial load and atypical radiological findings: smear-negative tuberculosis has a worse prognosis than smear-positive disease in HIV-positive patients.27?30,55,89?91 In studies based on blood culture, disseminated tuberculosis typically presents as a non-specific febrile disorder that progresses rapidly to death,16 and in autopsy studies, up to 50% of HIV-related tuberculosis deaths go undiagnosed; these findings indicate the diagnostic challenge.8?11 In programmes, a commitment must be made to prioritise smear-negative disease and lower the threshold for starting antituberculous treatment, with appropriate follow-up and outcome assessment.
Major initiatives to increase culture facilities in Africa are underway but without any consensus about their probable effect. Culture outperforms other investigations for early HIV-related tuberculosis92,93 and could be ideal for screening at HIV diagnosis and before starting tuberculosis preventive treatment. For routine investigation of ambulant individuals with suspected tuberculosis, however, the potential gain over sensitive microscopy and radiology is not clear.7,86,87,94?96 Logistic constraints to decentralisation of culture to primary health-care level are considerable,94 and culture might be too slow to contribute much to clinical decision-making.7,16 Expert groups have prioritised sensitive microscopy over expanded access to culture.94
Active case finding for tuberculosis and HIV
Up to 10% of HIV-infected individuals have active tuberculosis when first seeking knowledge of their HIV status.74 Symptom screening detects most, but not all, active cases,24,92 with culture but not radiology seeming to add substantially to sensitivity.92,97 Every opportunity should be taken to screen HIV-infected African people for active tuberculosis, just as every patient with tuberculosis should be screened for HIV. To increase access to life-prolonging interventions, active case finding for HIV will have to be developed in a way that is acceptable to communities.
Optimisation of antituberculous chemotherapy
The most frequently used treatment for newly diagnosed tuberculosis in Africa is an 8-month regimen introduced in the 1990s to replace thioacetazone-based strategies that were poorly tolerated in patients with HIV infection.98 The 8-month regimen includes rifampicin for 2 months only, but it is inferior to an alternative 6-month regimen containing rifampicin throughout.99 Using rifampicin for 6 months rather than 2 months extends direct supervision of treatment, which is recommended to prevent rifampicin resistance developing, and it makes treatment choices difficult because of drug interactions between rifamycins and antiretroviral drugs.100 However, to strive for antiretroviral drug access while tolerating suboptimum tuberculosis treatment is inconsistent, and the 8-month regimen should be phased out as soon as possible.
Prevention of the emergence of multidrug-resistant tuberculosis
Multidrug-resistant strains of M tuberculosis (resistant to at least isoniazid and rifampicin) arise from inadequate treatment of active tuberculosis, and can then be further transmitted. Treatment outcomes are poor for both HIV-positive and HIV-negative patients, with high case-fatality and treatment failure rates.101?104 Treatment for multidrug-resistant tuberculosis is expensive, toxic, difficult to combine with antiretroviral drugs, and unavailable in most of Africa.62 HIV-care settings are prone to outbreaks of nosocomial multidrug-resistant tuberculosis, which can persist for years without intervention:102,105 lack of diagnostic capacity would make early recognition difficult in most of Africa. Data indicate a growing problem, with primary multidrug resistance in more than 2% of patients in parts of South Africa, and a rise in Botswana from 0?2% to 0?8%.106,107 Continued commitment, more comprehensive surveillance, better access to drug-sensitivity testing, implementation of fixed-drug combination tablets, and policies for management of multidrug-resistant tuberculosis in HIV care settings are needed. The DOTS strategy focuses on standard treatment regimens and direct observation and has contained and even reduced primary rates of multidrug-resistant tuberculosis in other regions of the world.1,106
Optimisation of antiretroviral therapy in HIV-infected patients with tuberculosis
The high death rate in the first 2 months of tuberculosis treatment provides an argument for antiretroviral drugs to be started as soon as possible. However, challenges favouring a delayed start include drug interactions, combined toxic effects, and non-adherence to treatment.71,108,109 Clear definition of the best time to initiate antiretroviral treatment in patients with tuberculosis awaits results from controlled trials.
Detailed discussions of antiretroviral treatment for tuberculosis patients are available elsewhere.71,108,109 In brief, enzyme induction by rifampicin causes many interactions, with additional concerns of combined toxic effects, especially for nevirapine. Nevirapine-containing antiretroviral regimens are first-line in all African countries apart from South Africa, since the drug is cheap, effective, available in various fixed-drug combinations, and safe in pregnancy.71 Rifampicin reduces nevirapine concentrations by about a third,110 and both drugs can cause severe hepatitis.110 Women with a CD4 count greater than 250 per μL?a substantial subgroup of African patients with tuberculosis?are at highest risk of nevirapine-associated hepatitis.71,109,111?113 Clinical experience of concurrent use of nevirapine and antituberculous treatment is accruing, but at the time of writing the risks remain unclear.71
Patients who start antiretroviral drugs early in their tuberculosis treatment can be predisposed to immune reconstitution inflammatory syndrome, which is frequent, has symptoms overlapping with worsening tuberculosis and drug reactions, and can be life-threatening.114,115 WHO guidelines suggest starting antiretroviral drugs within 2 months of tuberculosis treatment at a CD4 count of 200 per μL or less and for extrapulmonary tuberculosis or other manifestations of severe immunosuppression;71 for patients with CD4 counts less than 50 per μL, treatment initiation is advised within 2 weeks. In such cases, efavirenz-containing regimens are recommended unless contraindicated by pregnancy or the potential to conceive.71 Use of protease inhibitors other than full-dose ritonavir is not recommended. Nevirapine can be used ?in the absence of other options?.71
Rifabutin is a less potent enzyme inducer than rifampicin; it can effectively treat tuberculosis and is compatible with antiretroviral drugs, but it is prohibitively expensive at present.109 Triple nucleoside or nucleotide regimens, such as zidovudine, lamivudine, and tenofovir, have potential as tuberculosis-compatible antiretroviral regimens if shown to be sufficiently potent.116
Co-trimoxazole prophylaxis in HIV-infected people with tuberculosis
Results from a placebo-controlled trial of co-trimoxazole in Cote d'Ivoire, showing a 46% reduction in mortality in HIV-infected tuberculosis patients, have been lent support by research in other parts of Africa.57?61 Since 1999, WHO and UNAIDS have recommended co-trimoxazole prophylaxis for all individuals with symptomatic HIV disease or CD4 counts less than 500 per μL, but uptake was estimated as only 3% of HIV-infected adults in 2003. The current drive towards roll-out of antiretroviral treatment might greatly enhance co-trimoxazole uptake as systems are put into place for delivery of chronic HIV care. An important question is whether co-trimoxazole benefits patients with tuberculosis who are taking antiretroviral treatment. In the industrialised world, co-trimoxazole can safely be withdrawn when CD4 count is greater than 200 per μL, but in the African environment, benefit could extend to higher CD4 cell counts.117,118
Back to top
Reduction of tuberculosis incidence and recurrence in HIV-infected individuals
Risk of new tuberculosis disease in HIV-infected individuals can be lowered, but not eliminated, by isoniazid preventive treatment,38 antiretroviral drugs,65?68 and reducing exposure to M tuberculosis. The benefits of antiretroviral drugs69 and secondary isoniazid preventive treatment70 for recurrent tuberculosis disease have not yet been clearly defined, but the rate of recurrence remains very high even for patients on antiretroviral drugs.69
Isoniazid preventive treatment for 6?9 months reduces tuberculosis prevalence by about 60% in HIV-infected individuals with a positive tuberculin skin test, and by about 40% when used irrespective of skin-test results.38,119 However, this low-cost intervention has been little used in Africa, with only Botswana attempting widespread implementation.80 In part, this low use relates to concerns about possible promotion of drug resistance and past absence of additional funding, but it also exemplifies limited commitment to date to joint tuberculosis and HIV interventions. Definition of screening procedures needed in operational settings remains an important research question.92,97 Other unresolved issues include the best duration of preventive treatment, since protection wanes with time in HIV-infected individuals not receiving antiretroviral drugs,120,121 and the role and safety of primary and secondary preventive treatment and antiretroviral regimens.
Epidemics of nosocomial tuberculosis have been well documented in industrialised countries, where they stand out against low background rates.122 In Africa, the potential for nosocomial transmission affecting patients and staff is much higher, and facilities are ill-equipped. In two cohort studies in a goldmining workforce, a rise was noted in the incidence of HIV-related recurrent tuberculosis, from 8?2 to 19?1 per 100 person-years, coinciding with the introduction of HIV clinics.70,123 Increased nosocomial transmission could have contributed to at least some of this trend. WHO has published guidelines for prevention of M tuberculosis transmission in health-care facilities,124 and updated guidance is underway, stressing the need for routine identification and separation of patients with cough from others in waiting areas.
Models for delivery of coordinated tuberculosis and HIV treatment services
Rapid scale-up of antiretroviral programmes dominates public-health interventions in Africa, but with only limited attention to coordination with tuberculosis programmes.125 Coordination can mean referral between services, some provision of joint services, or complete integration of tuberculosis and HIV/AIDS clinics. Important experience has been gained in Malawi, where the national tuberculosis control programme provides a model for scaling-up delivery of antiretroviral drugs.126 First-line treatment is with stavudine, lamivudine, and nevirapine (provided free of charge); clinical staging is used to define eligibility after a positive HIV test. Uptake of routine (opt-out) HIV testing among tuberculosis patients has reached 70%, with high uptake and adherence to co-trimoxazole but with success rates for starting antiretroviral treatment of 20% or less.126 The main constraints seem to be the 8-week delay between starting tuberculosis treatment and becoming eligible for nevirapine-containing antiretroviral regimens (during which time patients are discharged) and logistic difficulties for patients in accessing centralised antiretroviral services when they also have to attend local health clinics for continued management of tuberculosis.126 These difficulties indicate a major limitation of tuberculosis and antiretroviral services that are linked only by cross-referrals, and they show a need for antiretroviral and tuberculosis management to be decentralised and integrated as far as possible into the local health-care system.
Good uptake rates for antiretroviral treatment have been reported from programmes that offer tuberculosis and antiretroviral drugs from the same clinic, with transfer of care once tuberculosis treatment has been completed (partial integration) or continued treatment as a fully integrated HIV and tuberculosis service.76,77,127 Timely investigation and improved management of patients who develop tuberculosis while on antiretroviral drugs might be a further benefit of completely integrated clinics.77 Integrated care needs planning, retraining, and considerable expansion of tuberculosis programme personnel, but it is more patient-orientated and efficient than current systems.77
Back to top
Monitoring and assessment
Traditional outcome measures for tuberculosis programmes need cohort analysis of treated patients to establish successful and adverse outcomes. WHO defines targets of 70% case detection (estimated indirectly) and 85% cure rates.1,80 Collaborative HIV and tuberculosis programmes also need to monitor uptake of HIV testing and antiretroviral treatment, with subsequent measures of adherence, default, and survival. Because antiretroviral regimens, unlike antituberculous treatment, are lifelong, complexity of programmes will be greatly enhanced.128
From a surveillance perspective, universal opt-out HIV testing of tuberculosis patients will provide useful data for HIV prevalence and allow estimation of tuberculosis incidence in HIV-positive and HIV-negative subpopulations (figure 6). HIV-positive tuberculosis trends could usefully be viewed as a surrogate for trends in a country's overall AIDS epidemic,129 and HIV-negative trends would offer assessment of programme performance in controlling tuberculosis transmission. Demonstration that the incidence of tuberculosis was declining in HIV-negative subpopulations could do much to enhance morale of tuberculosis programmes, whose confidence has been shaken by escalating incidence, as could data showing reduced HIV-positive mortality rates.
Back to top
Challenges, constraints, and future prospects
Whatever we aim for, limited human resources, weak management and health systems, inadequate clinical and laboratory infrastructure, and absent training programmes for combined tuberculosis and HIV/AIDS care are all formidable barriers to supporting large numbers of patients in long-term care.125,130?132 Despite potential for increased funding, African tuberculosis control programmes remain poor in resources?in absolute terms and relative to those available for HIV/AIDS control.80,133 Few areas of the health sector have seen demand escalate to the degree experienced by tuberculosis programmes, yet support has not greatly increased and sometimes has declined in important areas such as personnel.132 The incidence of AIDS and HIV-associated tuberculosis in Africa is hundreds of times more than that in the industrialised world, but the ratio of health-care workers to population is a tenth of European levels.132 Programmes cannot succeed without increased funding for basic public-health and clinical infrastructure and appropriate personnel.134
Since the tuberculosis epidemic in Africa has been driven by HIV, activities directed towards HIV-associated disease should logically be the most effective response. Paradoxically, results from mathematical modelling suggest that improvements in tuberculosis case-finding and treatment, and reductions in HIV incidence, have a greater potential than interventions targeted to known HIV-positive individuals.72 Tuberculosis control in Africa remains weak, and large gains can still be made from strengthening basic disease control. Care must be taken to avoid collapse of previously well-functioning tuberculosis control programmes after competition for scarce human resources and to ensure that integration of HIV and tuberculosis services does not compromise core tuberculosis programme functions, such as maintenance of drugs and supplies, prevention of drug resistance, assurance of quality diagnostic microscopy, and cohort analysis of treated patients.77
In the same way that HIV fundamentally changed tuberculosis and its epidemiology in Africa, so the introduction of antiretroviral treatment poses important challenges to how tuberculosis control should be approached. Rapid scale-up of antiretroviral programmes in Africa in the past 2 years has not adequately taken patients with tuberculosis into account.125 Without sufficient coverage of preventive and therapeutic interventions for both diseases, tuberculosis could yet be the limiting factor to the long-term success of antiretroviral programmes because of uncontrolled incidence, recurrence, and institutional transmission. The advent of antiretroviral treatment in Africa is the most important event for tuberculosis patients since the introduction of antituberculous drugs, and HIV/AIDS and tuberculosis programmes will both have to change greatly to benefit as much as possible from this development.
Back to top
Search strategy and selection criteria
Publications related to tuberculosis or antiretroviral treatment in Africa were identified by systematically searching PubMed and Google Scholar with terms including, but not restricted to, the following combinations: ?tuberculosis Africa?, ?tuberculosis HIV?, ?tuberculosis antiretroviral?, ?mortality/survival tuberculosis?. Further publications were identified from references cited in relevant articles, reports, and workshop and conference proceedings. The search was restricted to publications in English, but not restricted by date.
Back to top
Conflict of interest statement
We declare that we have no conflict of interest.
Acknowledgments
We thank Charles Wells for helpful comments. ELC is funded by the Wellcome Trust. GJC received grants from CREATE (Bill and Melinda Gates Foundation), the Centre for the AIDS Program of Research in South Africa, and National Institute of Allergy and Infectious Diseases, USA (grant #1U19AI). None of the funding sources had any role in design or writing of this Review, nor in the decision to submit for publication.
Back to top
DOI:10.1016/S0140-6736(06)68383-9
Tuberculosis in sub-Saharan Africa: opportunities, challenges, and change in the era of antiretroviral treatment
Dr Elizabeth L CorbettPhD


Summary
HIV and epidemiology of tuberculosis in Africa
Tuberculosis control in Africa before antiretroviral treatment
Effect of antiretroviral treatment on tuberculosis epidemiology
Towards a coordinated public-health response to tuberculosis and HIV/AIDS
Monitoring and assessment
Challenges, constraints, and future prospects
Search strategy and selection criteria
References
Summary
Rapid scale-up of antiretroviral treatment programmes is happening in Africa, driven by international advocacy and policy directives and supported by unprecedented donor funding and technical assistance. This welcome development offers hope to millions of HIV-infected Africans, among whom tuberculosis is the major cause of serious illness and death. Little in the way of HIV diagnosis or care was previously offered to patients with tuberculosis, by either national tuberculosis or AIDS control programmes, with tuberculosis services focused exclusively on diagnosis and treatment of rising numbers of patients. Tuberculosis control in Africa has yet to adapt to the new climate of antiretroviral availability. Many barriers exist, from drug interactions to historic differences in the way that tuberculosis and HIV are perceived, but failure to successfully integrate HIV and tuberculosis control will threaten the viability of both programmes. Here, we review tuberculosis epidemiology in Africa and policy implications of HIV/AIDS treatment scale-up.
Back to top
<!--start ce:displayed-quote=-->
?Nothing endures but change?
Heraclites, c535?c475 BC
<!--end ce:displayed-quote-->Although we view tuberculosis as one disease, and Mycobacterium tuberculosis alone as its cause, four epidemiological patterns can be usefully distinguished. So-called traditional tuberculosis in developing countries with low rates of HIV infection responds to well-organised control programmes.1 Tuberculosis in the industrialised world is increasingly attributable to immigration, whereas high rates of multidrug resistance threaten control in parts of eastern Europe.2 Finally, HIV-related disease has emerged as the dominant challenge in sub-Saharan Africa (hereafter referred to as Africa),3,4 with the size of the epidemic calling for a response beyond the traditional boundaries of tuberculosis control.
Increasing access to antiretroviral treatment in Africa, an unimaginable aspiration a few years ago, is now an international priority. The global importance of tuberculosis and its association with the HIV/AIDS pandemic are acknowledged by the Millennium Development Goals, and unprecedented funds are being provided by the Global Fund to Fight AIDS, Tuberculosis, and Malaria, the US President's Emergency Plan for AIDS Relief, and the World Bank's multicountry HIV/AIDS programme (figure 1).

Click to enlarge image
Figure 1. Average donor support per person with HIV/AIDS for the 15 African countries with the highest adult HIV prevalence in 2004Donations calculated from the US President's Emergency Plan for AIDS Relief, the World Bank multicountry HIV/AIDS programme for Africa, and the Global Fund for AIDS, Tuberculosis, and Malaria (excludes support for malaria). For information about donations, see:
http://www.worldbank.org/afr/aids/map_docs.htm
http://www.state.gov/s/gac/
http://www.theglobalfund.org/en/
In this Review, we discuss current understanding of tuberculosis epidemiology and control policies in Africa. The new commitment to antiretroviral treatment will greatly affect tuberculosis control, hopefully for the better but possibly for the worse.5 Patients with HIV-related tuberculosis must make up a substantial proportion of those reached by antiretroviral treatment. Important philosophical and practical changes to address HIV and tuberculosis in a coordinated manner will be needed if the increased tuberculosis incidence that has afflicted Africa during the past decade is to be reversed.
HIV and epidemiology of tuberculosis in Africa
Burden of HIV/AIDS and tuberculosis
Figure 23 summarises the disproportionate burden of HIV and tuberculosis infection and disease in Africa at the start of the new millennium.3 In 2003, an estimated 8?8 million new cases of tuberculosis resulted in 1?7 million deaths.4 27% of these cases and 31% of these deaths arose in Africa, home to only 11% of the world's population.4 HIV prevalence in tuberculosis patients is less than 1% in the Western Pacific region but 38% in Africa.3 In countries with the highest HIV prevalence, more than 75% of cases of tuberculosis are HIV-associated.4

Click to enlarge image
Figure 2. Disproportionate burden of HIV, HIV-related tuberculosis, and M tuberculosis coinfections in Africa, for 2000Every person represents 5% of the global total, with African people shown in red and the rest of the world in blue.3
In Africa, tuberculosis is often the first manifestation of HIV infection, and it is the leading cause of death among HIV-infected patients.6?11 In hospital-based series, 40?65% of HIV-infected African patients with respiratory disease had tuberculosis.6,8 In primary health and chest clinic settings, tuberculosis was confirmed in 43?70% of adults with cough for 3 weeks or longer (chronic cough) in Zimbabwe, Kenya, and Malawi.12,13 Patients with tuberculosis now commonly present with atypical symptoms: M tuberculosis was isolated from 9% of adults with acute pneumonia in Kenya,14 35% of people with cough for less than 3 weeks in Malawi,15 23% of febrile HIV-infected inpatients in Tanzania,16 and 13% of HIV-infected patients with chronic diarrhoea in Kenya.17 In Cote d'Ivoire, the Democratic Republic of Congo, and Kenya, 38?47% of autopsies in HIV-positive adults indicated tuberculosis as the cause of death,9?11 although tuberculosis had been diagnosed during life in only about half of those with autopsy-proven disease.
Increased risk for tuberculosis from HIV infection in Africa
Comparison of HIV prevalence in general populations and tuberculosis patients shows that tuberculosis incidence was 8?3 times higher in HIV-positive than HIV-negative African people in 2003 (figure 3).4,18 In 2000, similar methods led to an estimated relative rate of 5?9, whereas estimates from individual cohort studies range from less than 5 to more than 20.3 Tuberculosis incidence increases with worsening immunosuppression, so that relative rates rise during the course of an HIV epidemic.19?21

Click to enlarge image
Figure 3. Estimated prevalence of HIV in tuberculosis patientsPrevalence measured in national surveys (blue dots) and subnational surveys (red dots; data reported to WHO),4 plotted against prevalence of HIV in adults (data from UNAIDS).18 Country abbreviations available from reference 4. Figure kindly provided by Chris Dye.
Tuberculosis incidence in African countries with high HIV prevalence
Reported tuberculosis case rates rose by 6?4% per year in the WHO African region in the late 1990s,4 but with up to five-fold increases since 1990 in some countries.4 Incidence might have peaked in some countries (figure 4)18 but at very high rates. The high case rate in Africa contributed to a global rise in tuberculosis incidence of 1% in 2003, despite stable or declining rates in the rest of the world.4

Click to enlarge image
Figure 4. Time trends in tuberculosis case-notification rates in southern and east AfricaAs reported to WHO.18
Southern Africa has the highest prevalence of HIV infection and had the highest incidence of tuberculosis before the HIV/AIDS era. In the six southern African countries with adult HIV prevalence of more than 20%, tuberculosis case-notification rates are 461?719 per 100 000 per year; by comparison, the notification rate in the USA was 5 per 100 000 per year.4 True yearly rates in Africa are likely to be even higher because of underdiagnosis and under-reporting.4
HIV and infectiousness and transmission of tuberculosis
The critical period with respect to infectiousness is before diagnosis, because most patients become non-infectious soon after starting treatment, even if they are HIV-infected.22 Both intensity and duration of infectiousness are highly variable, with some individuals remaining very infectious for prolonged periods, sometimes with apparently minor symptoms.23?26 HIV-positive individuals with tuberculosis are less infectious than HIV-negative patients since they are less often and less intensely smear-positive27?30 and because they remain infectious for a much shorter average duration.23,24
The average duration of smear positivity for HIV-negative individuals in resource-poor settings is estimated to be between 1 and 3 years.3,23,24,31 In two African studies in high HIV settings,23,24 mean durations of smear-positivity of only 6 and 8 weeks were estimated for HIV-positive patients with smear-positive tuberculosis, indicating fast progression to symptomatic disease.32 In both studies, most infectious individuals at any given point in time?the driving force for tuberculosis transmission in the community?were HIV-negative, because of their fairly long duration of infectiousness (figure 5).23,24 Thus, HIV-associated tuberculosis contributes greatly to incidence of tuberculosis and deaths, but it might contribute much less to disease transmission because of early diagnosis or death.33 This combination of exquisite vulnerability to disease by HIV-positive individuals and prolonged transmission from HIV-negative patients with tuberculosis together fuel escalating tuberculosis incidence in areas of high HIV prevalence. This fundamental observation has important implications for tuberculosis control in Africa.

Click to enlarge image
Figure 5. HIV prevalence for workers and tuberculosis patients, and for all person-days with undiagnosed smear-positive tuberculosisData from South Africa23 and Zimbabwe.24 *Incident tuberculosis patients self-presenting with symptoms. ?Patients with smear-positive tuberculosis diagnosed on systematic prevalence screening.
Back to top
Tuberculosis control in Africa before antiretroviral treatment
Tuberculosis control has been based on the WHO-promoted DOTS strategy, whose philosophical basis is prompt diagnosis and effective treatment of individuals with smear-positive tuberculosis to interrupt continuing transmission.1 In the face of rising tuberculosis incidence in Africa, international guidance during the 1990s emphasised the need for tuberculosis programmes to focus on diagnosis and treatment of self-presenting patients and to improve adherence and cure rates. HIV services, such as HIV-testing, were only deemed appropriate when priority tuberculosis objectives had been met.34?36 While understandable under the circumstances and funding of the time, this approach did not serve the overall medical needs of HIV-infected patients with tuberculosis.
The HIV epidemic has challenged DOTS as a sole tuberculosis control strategy for Africa, because even rigorous programmes cannot adequately compensate for the rising susceptibility to tuberculosis at the population level that occurs as HIV prevalence increases.37?41 As an example, despite well-funded control programmes in the South African gold mining industry that adhere to all elements of WHO's DOTS strategy and screen miners every year, tuberculosis case rates have risen four-fold since 1990, driven by an increase in population HIV prevalence from less than 1% to almost 30%.42
Containing tuberculosis transmission and preventing drug resistance might be realistic and important goals for conventional DOTS programmes, but they are not routinely evaluated.43?49 Evidence that strong tuberculosis control programmes can control disease transmission is provided by analysis of incidence data stratified by HIV status.46?49Figure 650 shows two possible scenarios of an epidemic of HIV.

Click to enlarge image
Figure 6. Simulated time trends in tuberculosis incidence during the course of an HIV epidemicDeterministic compartmental model of HIV (uninfected, WHO stages 1?4) and tuberculosis infection and disease (susceptible, latent, diseased, on-treatment, post-treatment), as previously described.50 Scenarios shown of tuberculosis trends in the whole population and HIV subpopulations under either falling (A) or rising (B) tuberculosis transmission rates. (C) Time course of simulated HIV epidemic in (A) and (B). HIV prevalence refers to adults in the general population (not tuberculosis patients).
In part A, an epidemic of HIV is shown, during which slowly falling tuberculosis transmission rates (from 1% in 1985 to 0?7% per year in 2010) are maintained despite a rising burden of HIV-related tuberculosis. Disease incidence continues to decline in the HIV-negative subpopulation, but rises in the HIV-positive subpopulation as the proportion of patients with moderate-to-severe immunosuppression increases (maturation of the HIV epidemic). Overall tuberculosis incidence rises to a plateau soon after peak HIV prevalence in 2000. This scenario might describe the course of HIV and tuberculosis in countries with fairly strong tuberculosis control programmes such as Malawi and Tanzania, where peak tuberculosis incidence might already have been reached. In part B, an identical HIV epidemic to that in part A is shown, with the same assumptions about the effect of HIV on susceptibility to tuberculosis disease. However, in this scenario, the potential for HIV-related tuberculosis to increase disease transmission rates is not contained. Tuberculosis incidence rises to very high levels among HIV-positive individuals and increases substantially in the HIV-negative subpopulation as well, because of a rise in yearly risk of infection from 1% in 1985 to 2?7% in 2010. Tuberculosis incidence continues to rise after HIV prevalence peaks. Several countries, including Kenya, Uganda, and Swaziland, might be following a course intermediate between these two scenarios.
Uganda and Kenya have reported continuing rises in tuberculosis case rates despite falling HIV prevalence, associated with a noted rise in yearly risk of M tuberculosis infection in Kenya.4,18,51 Data for Cote d'Ivoire, South Africa, Malawi, and Thailand, however, show stable or declining tuberculosis incidence among HIV-negative individuals despite increasing burdens and incidence of HIV-related tuberculosis.46?49 Similarly, in Tanzania, tuberculosis transmission rates have continued to decline while case-notifications have quadrupled.45 Within the same environment, therefore, strong orthodox tuberculosis control approaches can limit or reduce HIV-negative tuberculosis but not that associated with HIV.52
Back to top
Effect of antiretroviral treatment on tuberculosis epidemiology
Mortality in HIV-infected tuberculosis patients
Antiretroviral treatment scale-up mainly aims to reduce HIV-associated morbidity and mortality. Tuberculosis case-fatality rates (proportion of patients dying while on antituberculous treatment) in Africa are 16?35% in HIV-positive individuals not receiving antiretroviral treatment and 4?9% in HIV-negative patients.53 Increased mortality in the first month of treatment seems largely attributable to tuberculosis itself,53 whereas other HIV-associated pathologies predominate thereafter.54 The highest death rates are present for people with the lowest CD4+ T-lymphocyte counts (CD4 count).55 In a 7-year follow-up in Malawi before HIV/AIDS treatment initiatives, only 11% of individuals who were HIV-infected at tuberculosis diagnosis were known to be still living.56 About a quarter of known deaths arose within 1 month of tuberculosis diagnosis.
Case-fatality rates can be reduced by diagnosis of HIV infection linked to co-trimoxazole prophylaxis57?61 and antiretroviral treatment.62?64 Mortality in tuberculosis patients in London (UK) fell by 72% after introduction of highly active antiretroviral drugs.62 Mortality in African people with HIV-associated tuberculosis is similar to that in patients with tuberculosis before effective antituberculous treatment,53 and provision of antiretroviral drugs could have as revolutionary an effect as antituberculous drugs themselves did when first introduced.
Tuberculosis incidence and recurrence in HIV-infected people
Findings of several studies from different countries show that antiretroviral drugs reduce the incidence of tuberculosis in HIV-infected people by 80% or more,65?68 with the greatest effect at the lowest CD4 counts.67 However, clinically important immune dysfunction persists even during successful antiretroviral treatment,5 and tuberculosis incidence remains well above HIV-negative rates, even at high CD4 counts.67 Rates of recurrent disease in patients with previous HIV-related tuberculosis are also high, suggesting a need for secondary preventive treatment.69,70
Tuberculosis is an aggressive opportunistic infection that arises at higher median CD4 counts than do most other AIDS-defining disorders.6 For example, median CD4 counts were 257 per μL for smear-positive patients in Cote d'Ivoire.55 Current guidelines for resource-poor settings recommend treatment for patients with symptomatic HIV or a CD4 count of 200 per μL or less.71 The potential effect of antiretroviral treatment on tuberculosis incidence, therefore, is lessened because many HIV-infected patients with tuberculosis present before antiretroviral drugs are prescribed.64
Theoretically, even well-functioning antiretroviral programmes could worsen the HIV-associated tuberculosis epidemic if an expanding cohort of patients remains highly susceptible and capable of transmitting tuberculosis for long periods.5,50 Mathematical and statistical modelling suggests antiretroviral drug coverage would have to be high, start early, and be combined with tuberculosis preventive treatment to contain disease incidence and reduce mortality.72,73
Back to top
Towards a coordinated public-health response to tuberculosis and HIV/AIDS
One strategy alone is unlikely to succeed: different approaches based on serostatus need to address the vulnerability to tuberculosis disease of HIV-infected individuals and reduce disease transmission from all affected people, including those who are HIV-negative.74 The need for a coordinated approach towards tuberculosis and HIV control is now stressed at the highest levels.74 There is understandable reluctance to relinquish the traditional disease model for tuberculosis control: past experience, notably in Zambia, shows that well-intentioned reform can disrupt essential tuberculosis control activities.75 However, different programme models are now emerging that retain DOTS as the essential but insufficient minimum,41 while additional elements discussed below are implemented or investigated in collaboration with HIV/AIDS control programmes. Essential outcomes are reduced transmission, disease, and death for both HIV and tuberculosis (panel). Seamless collaboration between tuberculosis and HIV/AIDS treatment programmes is needed, along with a unified public-health vision towards the prevention and treatment of these interacting infectious diseases.
Panel: Key interventions for improving tuberculosis and HIV control in Africa
1)Better implementation of existing policies
?Universal HIV testing of patients with confirmed or suspected tuberculosis
?Universal access to high-quality sputum microscopy for individuals with suspected tuberculosis
?Universal directly observed treatment while taking rifampicin-containing tuberculosis regimen
?Antiretroviral treatment according to national guidelines for all HIV-infected individuals, including tuberculosis patients
?Ready access to voluntary counselling and testing for HIV and condoms to prevent HIV transmission
2)More widespread implementation of additional interventions known to be effective
?Use of the most effective short-course chemotherapy for all tuberculosis patients
?Co-trimoxazole prophylaxis for all HIV-infected tuberculosis patients not taking antiretroviral drugs
?Screening of all HIV-infected individuals for tuberculosis infection and disease
?Primary isoniazid preventive treatment for all HIV-infected individuals
?Environmental measures to prevent nosocomial transmission of M tuberculosis
3)Rapid assessment of promising new approaches
?More sensitive sputum smear microscopy
?More rapid and sensitive diagnostic algorithms for smear-negative individuals with suspected tuberculosis
?Expanded access to tuberculosis culture with rapid liquid culture systems
?Secondary preventive treatment for HIV-infected individuals after successful treatment of active tuberculosis
?Early initiation of antiretroviral treatment in newly diagnosed HIV-infected tuberculosis patients
?Co-trimoxazole prophylaxis for HIV-infected tuberculosis patients taking antiretroviral drugs
?Active case-finding for tuberculosis in the community
?Promotion of universal knowledge of HIV serostatus, with emphasis on prevention of transmission from HIV-infected individuals
?Novel HIV prevention interventions
4)Increased resources and support
?Training and retention of health-care workers in joint HIV and tuberculosis management
?Increased funding for integrated tuberculosis and HIV activities
?Increased funding for tuberculosis control programmes to support HIV diagnosis and initiation of HIV care, and more rapid diagnosis and effective treatment of both HIV-positive and HIV-negative patients with tuberculosis
One barrier to closer collaboration is the philosophical difference historically in how HIV/AIDS and tuberculosis surveillance, diagnosis, and treatment have been approached.76?78 Tuberculosis control programmes have epitomised the public-health approach of case finding, name-based case notification, and, when possible, screening of contacts. Control of tuberculosis transmission and prevention of drug resistance have been paramount aims, with less emphasis on patient-centred goals such as reduction of deaths.79 By contrast, HIV/AIDS programmes have focused on an individual approach to HIV testing that is private, confidential, and voluntary, but which has little emphasis on interrupting chains of transmission.78
Currently, fewer than 10% of African patients with tuberculosis are tested for HIV,4,80 although HIV testing is acceptable to most people when provided in a convenient and confidential way:42,58,81 the major difficulty is that testing is still not routinely offered in most tuberculosis clinics. WHO and UNAIDS guidelines now lend support to diagnostic HIV testing of individuals with HIV-associated disorders, including known and suspected tuberculosis patients, using an opt-out approach,74,82 and these organisations have requested routine reporting of the uptake of HIV testing along with the numbers of notified tuberculosis cases.4
There are three goals of coordinated tuberculosis and HIV interventions: (1) to optimise diagnosis and treatment to improve outcome for all tuberculosis patients; (2) to reduce HIV-associated tuberculosis incidence and recurrence; and (3) to improve HIV and tuberculosis control overall.
Optimisation of tuberculosis diagnosis and treatment
Tuberculosis diagnosis in Africa relies on sputum microscopy followed by broad-spectrum antibiotics and chest radiography if smears are negative. Although specificity is high,12,83,84 major concerns include low sensitivity13,84 and delayed diagnosis of smear-negative disease.13,85 The accuracy of both microscopy and radiography is reduced by HIV, and so assessment of diagnostic approaches with existing methods and continuing research into new diagnostics are necessary.85?88
DOTS programmes have focused on smear-positive disease because it is the most infectious type, but much of the increased tuberculosis caseload in Africa is reported as smear-negative.4 HIV-related tuberculosis is more usually smear-negative, extrapulmonary, or disseminated than tuberculosis among HIV-negative individuals. Worsening immunosuppression correlates with increased mycobacterial load and atypical radiological findings: smear-negative tuberculosis has a worse prognosis than smear-positive disease in HIV-positive patients.27?30,55,89?91 In studies based on blood culture, disseminated tuberculosis typically presents as a non-specific febrile disorder that progresses rapidly to death,16 and in autopsy studies, up to 50% of HIV-related tuberculosis deaths go undiagnosed; these findings indicate the diagnostic challenge.8?11 In programmes, a commitment must be made to prioritise smear-negative disease and lower the threshold for starting antituberculous treatment, with appropriate follow-up and outcome assessment.
Major initiatives to increase culture facilities in Africa are underway but without any consensus about their probable effect. Culture outperforms other investigations for early HIV-related tuberculosis92,93 and could be ideal for screening at HIV diagnosis and before starting tuberculosis preventive treatment. For routine investigation of ambulant individuals with suspected tuberculosis, however, the potential gain over sensitive microscopy and radiology is not clear.7,86,87,94?96 Logistic constraints to decentralisation of culture to primary health-care level are considerable,94 and culture might be too slow to contribute much to clinical decision-making.7,16 Expert groups have prioritised sensitive microscopy over expanded access to culture.94
Active case finding for tuberculosis and HIV
Up to 10% of HIV-infected individuals have active tuberculosis when first seeking knowledge of their HIV status.74 Symptom screening detects most, but not all, active cases,24,92 with culture but not radiology seeming to add substantially to sensitivity.92,97 Every opportunity should be taken to screen HIV-infected African people for active tuberculosis, just as every patient with tuberculosis should be screened for HIV. To increase access to life-prolonging interventions, active case finding for HIV will have to be developed in a way that is acceptable to communities.
Optimisation of antituberculous chemotherapy
The most frequently used treatment for newly diagnosed tuberculosis in Africa is an 8-month regimen introduced in the 1990s to replace thioacetazone-based strategies that were poorly tolerated in patients with HIV infection.98 The 8-month regimen includes rifampicin for 2 months only, but it is inferior to an alternative 6-month regimen containing rifampicin throughout.99 Using rifampicin for 6 months rather than 2 months extends direct supervision of treatment, which is recommended to prevent rifampicin resistance developing, and it makes treatment choices difficult because of drug interactions between rifamycins and antiretroviral drugs.100 However, to strive for antiretroviral drug access while tolerating suboptimum tuberculosis treatment is inconsistent, and the 8-month regimen should be phased out as soon as possible.
Prevention of the emergence of multidrug-resistant tuberculosis
Multidrug-resistant strains of M tuberculosis (resistant to at least isoniazid and rifampicin) arise from inadequate treatment of active tuberculosis, and can then be further transmitted. Treatment outcomes are poor for both HIV-positive and HIV-negative patients, with high case-fatality and treatment failure rates.101?104 Treatment for multidrug-resistant tuberculosis is expensive, toxic, difficult to combine with antiretroviral drugs, and unavailable in most of Africa.62 HIV-care settings are prone to outbreaks of nosocomial multidrug-resistant tuberculosis, which can persist for years without intervention:102,105 lack of diagnostic capacity would make early recognition difficult in most of Africa. Data indicate a growing problem, with primary multidrug resistance in more than 2% of patients in parts of South Africa, and a rise in Botswana from 0?2% to 0?8%.106,107 Continued commitment, more comprehensive surveillance, better access to drug-sensitivity testing, implementation of fixed-drug combination tablets, and policies for management of multidrug-resistant tuberculosis in HIV care settings are needed. The DOTS strategy focuses on standard treatment regimens and direct observation and has contained and even reduced primary rates of multidrug-resistant tuberculosis in other regions of the world.1,106
Optimisation of antiretroviral therapy in HIV-infected patients with tuberculosis
The high death rate in the first 2 months of tuberculosis treatment provides an argument for antiretroviral drugs to be started as soon as possible. However, challenges favouring a delayed start include drug interactions, combined toxic effects, and non-adherence to treatment.71,108,109 Clear definition of the best time to initiate antiretroviral treatment in patients with tuberculosis awaits results from controlled trials.
Detailed discussions of antiretroviral treatment for tuberculosis patients are available elsewhere.71,108,109 In brief, enzyme induction by rifampicin causes many interactions, with additional concerns of combined toxic effects, especially for nevirapine. Nevirapine-containing antiretroviral regimens are first-line in all African countries apart from South Africa, since the drug is cheap, effective, available in various fixed-drug combinations, and safe in pregnancy.71 Rifampicin reduces nevirapine concentrations by about a third,110 and both drugs can cause severe hepatitis.110 Women with a CD4 count greater than 250 per μL?a substantial subgroup of African patients with tuberculosis?are at highest risk of nevirapine-associated hepatitis.71,109,111?113 Clinical experience of concurrent use of nevirapine and antituberculous treatment is accruing, but at the time of writing the risks remain unclear.71
Patients who start antiretroviral drugs early in their tuberculosis treatment can be predisposed to immune reconstitution inflammatory syndrome, which is frequent, has symptoms overlapping with worsening tuberculosis and drug reactions, and can be life-threatening.114,115 WHO guidelines suggest starting antiretroviral drugs within 2 months of tuberculosis treatment at a CD4 count of 200 per μL or less and for extrapulmonary tuberculosis or other manifestations of severe immunosuppression;71 for patients with CD4 counts less than 50 per μL, treatment initiation is advised within 2 weeks. In such cases, efavirenz-containing regimens are recommended unless contraindicated by pregnancy or the potential to conceive.71 Use of protease inhibitors other than full-dose ritonavir is not recommended. Nevirapine can be used ?in the absence of other options?.71
Rifabutin is a less potent enzyme inducer than rifampicin; it can effectively treat tuberculosis and is compatible with antiretroviral drugs, but it is prohibitively expensive at present.109 Triple nucleoside or nucleotide regimens, such as zidovudine, lamivudine, and tenofovir, have potential as tuberculosis-compatible antiretroviral regimens if shown to be sufficiently potent.116
Co-trimoxazole prophylaxis in HIV-infected people with tuberculosis
Results from a placebo-controlled trial of co-trimoxazole in Cote d'Ivoire, showing a 46% reduction in mortality in HIV-infected tuberculosis patients, have been lent support by research in other parts of Africa.57?61 Since 1999, WHO and UNAIDS have recommended co-trimoxazole prophylaxis for all individuals with symptomatic HIV disease or CD4 counts less than 500 per μL, but uptake was estimated as only 3% of HIV-infected adults in 2003. The current drive towards roll-out of antiretroviral treatment might greatly enhance co-trimoxazole uptake as systems are put into place for delivery of chronic HIV care. An important question is whether co-trimoxazole benefits patients with tuberculosis who are taking antiretroviral treatment. In the industrialised world, co-trimoxazole can safely be withdrawn when CD4 count is greater than 200 per μL, but in the African environment, benefit could extend to higher CD4 cell counts.117,118
Back to top
Reduction of tuberculosis incidence and recurrence in HIV-infected individuals
Risk of new tuberculosis disease in HIV-infected individuals can be lowered, but not eliminated, by isoniazid preventive treatment,38 antiretroviral drugs,65?68 and reducing exposure to M tuberculosis. The benefits of antiretroviral drugs69 and secondary isoniazid preventive treatment70 for recurrent tuberculosis disease have not yet been clearly defined, but the rate of recurrence remains very high even for patients on antiretroviral drugs.69
Isoniazid preventive treatment for 6?9 months reduces tuberculosis prevalence by about 60% in HIV-infected individuals with a positive tuberculin skin test, and by about 40% when used irrespective of skin-test results.38,119 However, this low-cost intervention has been little used in Africa, with only Botswana attempting widespread implementation.80 In part, this low use relates to concerns about possible promotion of drug resistance and past absence of additional funding, but it also exemplifies limited commitment to date to joint tuberculosis and HIV interventions. Definition of screening procedures needed in operational settings remains an important research question.92,97 Other unresolved issues include the best duration of preventive treatment, since protection wanes with time in HIV-infected individuals not receiving antiretroviral drugs,120,121 and the role and safety of primary and secondary preventive treatment and antiretroviral regimens.
Epidemics of nosocomial tuberculosis have been well documented in industrialised countries, where they stand out against low background rates.122 In Africa, the potential for nosocomial transmission affecting patients and staff is much higher, and facilities are ill-equipped. In two cohort studies in a goldmining workforce, a rise was noted in the incidence of HIV-related recurrent tuberculosis, from 8?2 to 19?1 per 100 person-years, coinciding with the introduction of HIV clinics.70,123 Increased nosocomial transmission could have contributed to at least some of this trend. WHO has published guidelines for prevention of M tuberculosis transmission in health-care facilities,124 and updated guidance is underway, stressing the need for routine identification and separation of patients with cough from others in waiting areas.
Models for delivery of coordinated tuberculosis and HIV treatment services
Rapid scale-up of antiretroviral programmes dominates public-health interventions in Africa, but with only limited attention to coordination with tuberculosis programmes.125 Coordination can mean referral between services, some provision of joint services, or complete integration of tuberculosis and HIV/AIDS clinics. Important experience has been gained in Malawi, where the national tuberculosis control programme provides a model for scaling-up delivery of antiretroviral drugs.126 First-line treatment is with stavudine, lamivudine, and nevirapine (provided free of charge); clinical staging is used to define eligibility after a positive HIV test. Uptake of routine (opt-out) HIV testing among tuberculosis patients has reached 70%, with high uptake and adherence to co-trimoxazole but with success rates for starting antiretroviral treatment of 20% or less.126 The main constraints seem to be the 8-week delay between starting tuberculosis treatment and becoming eligible for nevirapine-containing antiretroviral regimens (during which time patients are discharged) and logistic difficulties for patients in accessing centralised antiretroviral services when they also have to attend local health clinics for continued management of tuberculosis.126 These difficulties indicate a major limitation of tuberculosis and antiretroviral services that are linked only by cross-referrals, and they show a need for antiretroviral and tuberculosis management to be decentralised and integrated as far as possible into the local health-care system.
Good uptake rates for antiretroviral treatment have been reported from programmes that offer tuberculosis and antiretroviral drugs from the same clinic, with transfer of care once tuberculosis treatment has been completed (partial integration) or continued treatment as a fully integrated HIV and tuberculosis service.76,77,127 Timely investigation and improved management of patients who develop tuberculosis while on antiretroviral drugs might be a further benefit of completely integrated clinics.77 Integrated care needs planning, retraining, and considerable expansion of tuberculosis programme personnel, but it is more patient-orientated and efficient than current systems.77
Back to top
Monitoring and assessment
Traditional outcome measures for tuberculosis programmes need cohort analysis of treated patients to establish successful and adverse outcomes. WHO defines targets of 70% case detection (estimated indirectly) and 85% cure rates.1,80 Collaborative HIV and tuberculosis programmes also need to monitor uptake of HIV testing and antiretroviral treatment, with subsequent measures of adherence, default, and survival. Because antiretroviral regimens, unlike antituberculous treatment, are lifelong, complexity of programmes will be greatly enhanced.128
From a surveillance perspective, universal opt-out HIV testing of tuberculosis patients will provide useful data for HIV prevalence and allow estimation of tuberculosis incidence in HIV-positive and HIV-negative subpopulations (figure 6). HIV-positive tuberculosis trends could usefully be viewed as a surrogate for trends in a country's overall AIDS epidemic,129 and HIV-negative trends would offer assessment of programme performance in controlling tuberculosis transmission. Demonstration that the incidence of tuberculosis was declining in HIV-negative subpopulations could do much to enhance morale of tuberculosis programmes, whose confidence has been shaken by escalating incidence, as could data showing reduced HIV-positive mortality rates.
Back to top
Challenges, constraints, and future prospects
Whatever we aim for, limited human resources, weak management and health systems, inadequate clinical and laboratory infrastructure, and absent training programmes for combined tuberculosis and HIV/AIDS care are all formidable barriers to supporting large numbers of patients in long-term care.125,130?132 Despite potential for increased funding, African tuberculosis control programmes remain poor in resources?in absolute terms and relative to those available for HIV/AIDS control.80,133 Few areas of the health sector have seen demand escalate to the degree experienced by tuberculosis programmes, yet support has not greatly increased and sometimes has declined in important areas such as personnel.132 The incidence of AIDS and HIV-associated tuberculosis in Africa is hundreds of times more than that in the industrialised world, but the ratio of health-care workers to population is a tenth of European levels.132 Programmes cannot succeed without increased funding for basic public-health and clinical infrastructure and appropriate personnel.134
Since the tuberculosis epidemic in Africa has been driven by HIV, activities directed towards HIV-associated disease should logically be the most effective response. Paradoxically, results from mathematical modelling suggest that improvements in tuberculosis case-finding and treatment, and reductions in HIV incidence, have a greater potential than interventions targeted to known HIV-positive individuals.72 Tuberculosis control in Africa remains weak, and large gains can still be made from strengthening basic disease control. Care must be taken to avoid collapse of previously well-functioning tuberculosis control programmes after competition for scarce human resources and to ensure that integration of HIV and tuberculosis services does not compromise core tuberculosis programme functions, such as maintenance of drugs and supplies, prevention of drug resistance, assurance of quality diagnostic microscopy, and cohort analysis of treated patients.77
In the same way that HIV fundamentally changed tuberculosis and its epidemiology in Africa, so the introduction of antiretroviral treatment poses important challenges to how tuberculosis control should be approached. Rapid scale-up of antiretroviral programmes in Africa in the past 2 years has not adequately taken patients with tuberculosis into account.125 Without sufficient coverage of preventive and therapeutic interventions for both diseases, tuberculosis could yet be the limiting factor to the long-term success of antiretroviral programmes because of uncontrolled incidence, recurrence, and institutional transmission. The advent of antiretroviral treatment in Africa is the most important event for tuberculosis patients since the introduction of antituberculous drugs, and HIV/AIDS and tuberculosis programmes will both have to change greatly to benefit as much as possible from this development.
Back to top
Search strategy and selection criteria
Publications related to tuberculosis or antiretroviral treatment in Africa were identified by systematically searching PubMed and Google Scholar with terms including, but not restricted to, the following combinations: ?tuberculosis Africa?, ?tuberculosis HIV?, ?tuberculosis antiretroviral?, ?mortality/survival tuberculosis?. Further publications were identified from references cited in relevant articles, reports, and workshop and conference proceedings. The search was restricted to publications in English, but not restricted by date.
Back to top
Conflict of interest statement
We declare that we have no conflict of interest.
Acknowledgments
We thank Charles Wells for helpful comments. ELC is funded by the Wellcome Trust. GJC received grants from CREATE (Bill and Melinda Gates Foundation), the Centre for the AIDS Program of Research in South Africa, and National Institute of Allergy and Infectious Diseases, USA (grant #1U19AI). None of the funding sources had any role in design or writing of this Review, nor in the decision to submit for publication.
Back to top
Comment