Improving the prevention, diagnosis and treatment of TB among people living with HIV: the role of operational research

Operational research is necessary to improve the access to and delivery of tuberculosis prevention, diagnosis and treatment interventions for people living with HIV. We conducted an extensive review of the literature and reports from recent expert consultations and research-related meetings organized by the World Health Organization and the Stop TB Partnership to identify a TB/HIV operational research agenda. We present critical operational research questions in a series of key areas: optimizing TB prevention by enhancing the uptake of isoniazid preventive therapy and the implementation of infection control measures; assessing the effectiveness of existing diagnostic tools and scaling up new technologies; improving service delivery models; and reducing risk factors for mortality among TB patients living with HIV. We discuss the potential impact that addressing the operational research questions may have on improving programmes’ performance, assessing new strategies or interventions for TB control, or informing global or national policy formulation. Financial resources to implement these operational research questions should be mobilized from existing and new funding mechanisms. National TB and HIV/AIDS programmes should develop their operational research agendas based on these questions, and conduct the research that they consider crucial for improving TB and HIV control in their settings in collaboration with research stakeholders.


Background
From 1995 to 2004, the estimated incidence of tuberculosis (TB) increased steadily around the world, with an excess in sub-Saharan Africa, in relation with the emer gence and expansion of the HIV epidemic [1]. Th e estimated TB incidence has been slowly declining since then, at a rate of 1% per year, making it impossible to reach the 2050 TB elimination target, defi ned as less than one TB case per million population [2]. Th ere were an estimated 9.4 million incident TB cases in the world in 2009, of which an estimated 1.1 million (12%) were living with HIV; 1.7 million deaths were also estimated, including 380,000 among HIV individuals [3]. TB con tinues to be a leading cause of deaths among people living with HIV, accounting for one-fi fth of the 1.8 million AIDSrelated deaths in 2009. While HIV testing among TB patients has rapidly scaled up since 2002, the uptake of antiretroviral therapy (ART) among TB patients living with HIV is unsatisfactory, with only 140,000 TB patients with HIV co-infection placed on ART worldwide in 2009 [3]. Prevention and treatment of TB among people living with HIV remains limited: only 5% of the estimated 33.3 million people living with HIV (about 1.7 million) were screened for TB in 2009, and less than 1% (88,000) of those found without active TB were enrolled on isoniazid preventive therapy (IPT) [3]. Th ese percentages are still very far from the 100% and 50% targets set for 2015 in the Global Plan to Stop TB 2011-2015 [4]. To boost TB control activities in relation with HIV, and reach these targets, important operational research questions need to be urgently addressed.
Th e aim of this paper is to review and present the priority operational research questions to improve the prevention, diagnosis and treatment of TB among people living with HIV, including innovative models of service delivery for collaborative TB/HIV activities. Several defi nitions for operational research have been proposed [5][6][7][8]. For the purpose of this paper, we defi ned operational research according to its ultimate objectives, that are: (i) to improve programme performance and outcomes; (ii) to assess the feasibility, eff ectiveness or impact of new strategies or interventions for TB control; or (iii)

Abstract
Operational research is necessary to improve the access to and delivery of tuberculosis prevention, diagnosis and treatment interventions for people living with HIV. We conducted an extensive review of the literature and reports from recent expert consultations and researchrelated meetings organized by the World Health Organization and the Stop TB Partnership to identify a TB/HIV operational research agenda. We present critical operational research questions in a series of key areas: optimizing TB prevention by enhancing the uptake of isoniazid preventive therapy and the implementation of infection control measures; assessing the eff ectiveness of existing diagnostic tools and scaling up new technologies; improving service delivery models; and reducing risk factors for mortality among TB patients living with HIV. We discuss the potential impact that addressing the operational research questions may have on improving programmes' performance, assessing new strategies or interventions for TB control, or informing global or national policy formulation. Financial resources to implement these operational research questions should be mobilized from existing and new funding mechanisms. National TB and HIV/AIDS programmes should develop their operational research agendas based on these questions, and conduct the research that they consider crucial for improving TB and HIV control in their settings in collaboration with research stakeholders.
Improving the prevention, diagnosis and treatment of TB among people living with HIV: the role of operational research Delphine Sculier 1 , Haileyesus Getahun 1 and Christian Lienhardt* 1,2 to collect data to guide policy recommendations on specifi c interventions [9].

Methods
We used a variety of sources of information. We reviewed PubMed and Web of Science databases using a combination of "tuberculosis" and "HIV" with the following key words: drug-resistant tuberculosis, anti-tuberculosis treatment, treatment of latent infection, antiretroviral therapy, infection control, laboratory services, childhood tuberculosis, pregnant women, community, programme evaluation, and opera tional research.
We also reviewed reports of expert consultations and other research-related meetings with a focus on operational research aspects organized by the World Health Organization (WHO) and the Stop TB Partnership in 2009-2010 [10][11][12][13][14][15][16]. Th e objectives and proceedings of these meetings included questions addres sing the translation gaps in implementing existing and new tools and technologies within TB services, as well as questions on how operational research could best inform policy decision making at national, regional and global levels. Th e meetings and documents also defi ned key areas of TB control for which various research questions were to be addressed as priorities, taking into account the various elements of time, place and person, and issues of relevance and feasibility [16]. In one document, the TB/ HIV research questions were identifi ed by various experts and prioritized using a scoring system that used a set criteria of eff ectiveness, deliverability, answerability and equity [17].
We have re-examined the documents for the aspects of operational research and present our fi ndings in the following key areas: (i) prevention of TB among people living with HIV; (ii) implementation of infection control measures; (iii) early diagnosis and treatment of TB in people living with HIV; (iv) service delivery of collaborative TB/HIV activities; (v) reducing mortality among TB patients living with HIV; and (vi) childhood TB and HIV.

Key areas of operational research
Th e WHO policy on collaborative TB/HIV activities recommends a combination of measures to reduce the burden of TB among people living with HIV [18]. Th ese measures include intensifi ed TB case fi nding, prevention of TB with IPT and ART, and effi cient infection control, which are branded as the Th ree I's for HIV/TB. Th eir implementation requires sound policy and a programme environment that gives due consideration to the local context, the epidemiology of TB and HIV, and the status of health systems and infrastructure that will determine the service delivery models. In addition to the operational diffi culties con fronted in providing quality and eff ective interventions, the cultural and system-wide diff erences between TB and HIV programmes, care providers and partner organi za tions may have contributed to the limited implementation of collaborative TB/HIV activities.
Operational research plays a key role in addressing these challenges and optimizing and scaling up prevention, diagnosis and treatment of TB among people living with HIV through eff ective service delivery models, including community-based interventions. Table 1 summar izes the operational research priorities to improve the implementation of collaborative TB/HIV programme activities [16,17,19]. Th e ultimate objective of each operational research question is presented, as well as how it infl uences the design and methodology of the research, the setting where it is conducted, the degree of generalizability of the results, and the ultimate relevance for TB control from local to global level [9].

Prevention of TB among people living with HIV
Meta-analyses of randomized controlled trials have shown that IPT reduces the risk of active TB by 33% in people living with HIV, irrespective of skin test result, and by up to 64% in those with a positive tuberculin skin test [20,21]. Despite the substantial amount of clinical research that showed the benefi ts of IPT among people living with HIV, several barriers and operational challenges have hampered its roll out [22]. One of these is the diffi culty of reliably excluding active TB before preventive therapy, with subsequent concern about the possibility of selecting for isoniazid resistant M. tuberculosis strains and eventually multi-drug-resistant strains [23]. WHO has recently developed simplifi ed TB screening algorithms for adults and children living with HIV that will promote the delivery of IPT for people living with HIV [24].
Although the adult algorithm was based on an indi vidual participant data meta-analysis of more than 9000 patients [25], the children algorithm mostly relied on expert opinion, due to lack of evidence. Concomitant to intensifying the search for a simple and rapid TB diagnostic tool, the implementation of these algorithms need to be scaled up and their feasibility, eff ectiveness and impact investigated by national AIDS and TB programmes.
Clinicians are faced with challenges on how to maintain adherence to IPT: studies from Malawi, South Africa and Zambia have shown that adherence varies from 24% to 59% [26]. Concerns also exist on the need and feasibility of monitoring IPT in terms of toxicity and occurrence of TB disease [27]. Recent data from more than 24,000 South African patients showed, however, very low rates of IPT-related adverse events (particularly clinical hepatotoxicity: 0.07%), and suggested that monitoring based on clinical symptoms was suffi cient [28]. Th ese encouraging results need to be confi rmed in other populations. Another important question is the duration of preventive therapy. Results from two recent clinical trials from Botswana [29] and South Africa [30] suggest an increased benefi t of IPT when given for 36 months or longer, compared with six months, particularly in people who are tuberculin skin test positive. However, a clinical trial reported no diff erence between 36 and six months of extended use of isoniazid and ethambutol in India [31]. Th e feasibility of life-long IPT under routine conditions and the factors related to it, as well as the timing of initiation in relation to ART, which also greatly reduces TB incidence [32], need to be further investigated. From a programme perspective, it is important to determine which programmes or services (HIV, TB, mother and child health, or others) will best coordinate nationwide implementation of IPT and ensure that it is included in the comprehensive package of care for people living with HIV [24].

Implementation of infection control measures
In addition to preventive chemotherapy, effi cient infection control measures at facility level are crucial to prevent the transmission of M. tuberculosis among people living with HIV [33]. Th ese include: administrative control measures, such as triage; separation of infectious cases; control of pathogens' spread and reduced hospital stay; environmental control with natural or mechanical ventilation systems and the use of ultraviolet germicidal irradiation; and personal protective protection for healthcare workers. Th e identifi cation of the best mix of infection control measures to be implemented and of the best means to monitor these in healthcare facilities (including suitable indicators and methods for detecting the concentration of viable M. tuberculosis droplet nuclei in room air) is needed. It will assist TB and AIDS programmes in improving their performance and evaluat ing their contribution to reducing TB transmission.

Early diagnosis and treatment of TB among people living with HIV
Delay in TB diagnosis may in part explain the high mortality rates among people living with HIV [34]. Th e eff ectiveness, impact and modalities of scaling up existing tools and strategies, such as specimen culturing and clinical algorithms [35], need to be assessed. Similarly, the newly developed, fully automated, cartridge-based nucleic amplifi cation assay, Xpert MTB/RIF, which can detect drug susceptible and rifampicin resistant TB in less than two hours, needs to be rapidly introduced and adapted on a large scale in health services. Th e sensitivity of a single Xpert MTB/RIF test was 98.2% and 72.5% in smear-positive and smear-negative disease, respectively, and 99.1% in rifampicin resistant TB, and was not aff ected by the patient's HIV status [36]. WHO has now recommended Xpert MTB/RIF as the initial diagnostic test for all people living with HIV, as well as for those suspected of having multi-drug-resistant TB [37].
Th e algorithms for the diagnosis of HIV-associated TB, both for ambulatory and seriously ill patients, were revised to include Xpert MTB/RIF and to assist in its implementation in HIV prevalent settings (Figures 1 and  2) [38]. Operational research is crucially needed to assess the feasibility and validity of these revised algorithms. Th ere is an urgent need for collecting further evidence for scaling up Xpert MTB/RIF under programmatic conditions at the various levels of health services, including HIV care facilities, under diff erent epidemiological and resource conditions. Real-time cost-eff ectiveness data are also needed to inform policy.
Th e use of ART has been associated with up to 95% reduction in mortality risk among TB patients living with HIV [39], including those with drug-resistant TB [40,41]. Combined ART and TB treatment is more protective than delaying ART after completion of TB treatment [42]. Th e CAMELIA randomized control trial conducted in Cambodia among 661 HIV-infected TB patients with a median CD4 count of 25 cells/mm 3 found a 34% reduction in mortality associated with the early initiation of ART two weeks after onset of TB treatment, compared with eight weeks after onset [43]. Similarly the STRIDE and SAPIT trials showed that combined and earlier ART and TB treatment (two to four weeks after TB treatment initiation) reduced deaths and AIDS-related events by 42% and 68%, respectively, among those with CD4 counts of less than 50 cells/mm 3 when compared with ART initiation within eight to 12 weeks of TB treatment [44,45]. Th ese results confi rm current WHO recom mendations to start ART as soon as possible in the course of TB treatment [46], especially in people with profound immunosuppression. Operational research to inform national AIDS and TB programmes on how best to implement these recommendations and scale up the uptake of ART in TB patients living with HIV is urgently needed and should include the assessment of various service delivery models for collaborative TB/HIV activities.
A prerequisite for improved early diagnosis and treatment of TB among people living with HIV is knowledge of HIV status. HIV testing for those with confi rmed and presumptive TB and their contacts yields a high number of new diagnoses of HIV [47][48][49][50], and should be encouraged. Any innovative measure to increase the uptake of HIV testing, such as the home-based rapid test [51], needs to be assessed and standardized for TB patients, TB suspects and their relatives.

Service delivery of collaborative TB/HIV activities
Integration of TB and HIV services to deliver collaborative TB/HIV activities is important in settings where most TB patients also have HIV infection and therefore are in need of ART. A recent systematic review has shown that diff erent models of integration of services are being implemented: (i) TB services refer patients for HIV testing and treatment; (ii) HIV services refer people living with HIV for tuberculosis screening and treatment; (iii) TB services test patients for HIV and refer them for treatment; (iv) HIV services screen for tuberculosis but refer patients for treatment; and (v) TB and HIV services are provided at a single facility [52]. However, the best delivery model is unknown and it is unlikely that a "onesize-fi ts-all" approach will work. Ways to decentralize TB and HIV services into primary healthcare, as well as mother and child healthcare facilities, should also be investigated. Operational research is needed to provide evidence on best service delivery models and on their eff ectiveness, so as to enhance the uptake of TB and HIV prevention, diagnosis and treatment for people aff ected by, and at risk for, both diseases. Th e impact of the models on patients' outcomes also needs to be investigated [52]. Specifi c attention should be paid to evaluating these questions in special populations, such as children, women, prisoners and injecting drug users. Particular emphasis is needed for the integrated delivery of TB and HIV services with harm-reduction services in those settings where drug use is fuelling the HIV epidemic.
Community participation for enhanced TB case fi nding and early HIV detection to reduce delay in initiation of TB and HIV care also needs to be further investigated by research stakeholders, including national TB and AIDS programmes. TB control in settings with high HIV Figure 1. Algorithm for the diagnosis of tuberculosis in ambulatory HIV-positive patients using Xpert MTB/RIF. 1 Among adults and adolescents living with HIV, a TB suspect is defi ned as a person who reports any one of current cough, fever, weight loss or night sweats. Among children living with HIV, a TB suspect is defi ned as a person who reports one of poor weight gain, fever, current cough, or history of contact with a TB case. 2 In all persons with unknown HIV status, HIV testing should be performed according to national guidelines. In patients who are HIV negative or remain HIV unknown (e.g., refusal), a TB suspect is defi ned according to national case defi nitions. A person with unknown HIV status can still be classifi ed as HIV-positive if there is strong clinical evidence of HIV infection. 3 The danger signs include any one of: respiratory rate >30/min, temperature >39°C, heart rate >120/min and unable to walk unaided. 4 CPT = cotrimoxazole preventive therapy. 5 ART = antiretroviral therapy. All TB patients living with HIV are eligible for ART irrespective of CD4 count. Start TB treatment fi rst, followed by ART as soon as possible within the fi rst 8 weeks of TB treatment. See ART guidelines. 6 In low MDR-TB prevalence settings, a confi rmatory test for rifampicin resistance should be performed. See MDR-TB Xpert MTB/RIF algorithm. 7 A chest x-ray can assist with the diagnosis of extra-pulmonary TB (e.g., pleural, pericardial) and help assess for other etiologies of respiratory illness. It should only be performed in those settings where the quality of the fi lm and its interpretation are assured. 8 Antibiotics (except fl uoroquinolones) to cover both typical and atypical bacteria should be considered. 9 An HIV treatment assessment includes WHO clinical staging and/or CD4 count to assess eligibility for antiretroviral therapy. See ART guidelines. 10  prevalence will not succeed without reaching out into the community [53], and innovative ways of doing so, such as using mobile vans [54] or health extension workers [55], need to be tested under routine conditions. Identifi cation of the best ways that TB and HIV/AIDS programmes could bring TB prevention, diagnosis and treatment services for both drug-susceptible and drug-resistant TB closer to the community through innovation and decentralization of service is crucial. Finally, cost-eff ectiveness analyses of the various healthcare-and community-level service delivery models are important to maximize value for money in a time of scarce resources [56].

Reducing mortality among TB patients living with HIV
In addition to ART, cotrimoxazole preventive therapy (CPT) has been found to reduce mortality in people living with HIV and treated for TB [57]. However, the provision of both ART and CPT has not been fully scaled up, and operational research can assist in determining the best ways to do so. Operational research can also help determine if there are other context-dependant and modifi able risk factors, such as undiagnosed multi-drugresistant TB, malnutrition and poor treatment adherence, that are associated with death among patients living with HIV on TB treatment despite receiving ART and CPT.  1 Seriously ill refers to the presence of danger signs, including: respiratory rate >30/min, temperature >39°C, heart rate >120/min and unable to walk unaided. 2 Among adults and adolescents living with HIV, a TB suspect is defi ned as a person who reports any one of current cough, fever, weight loss or night sweats. Among children living with HIV, a TB suspect is defi ned as a person who reports one of poor weight gain, fever, current cough, or history of contact with a TB case. 3 In all persons with unknown HIV status, HIV testing should be performed according to national guidelines. In high HIV prevalent settings, seriously ill patients should be tested using Xpert MTB/RIF as the primary diagnostic test regardless of HIV status. 4 The highest priority should be to provide the patient with life-sustaining supportive therapy, such as oxygen and parenteral antibiotics. If life-sustaining therapy is not available at the initial point of care, the patient should be transferred immediately to a higher level facility before further diagnostic testing. 5 Antibiotics (except fl uoroquinolones) to cover both typical and atypical bacteria should be considered. 6 PCP= Pneumocystis jirovecii pneumonia. 7 CPT = cotrimoxazole preventive therapy. 8 ART = antiretroviral therapy. All TB patients living with HIV are eligible for ART irrespective of CD4 count. Start TB treatment fi rst, followed by ART as soon as possible within the fi rst 8 weeks of TB treatment. See ART guidelines. 9 In low MDR-TB prevalence setting, a confi rmatory test for Rifampicin resistance should be performed. See MDR-TB algorithm. 10 An HIV treatment assessment includes WHO clinical staging and/or CD4 count to assess eligibility for antiretroviral therapy. See ART guidelines. 11 Additional investigations for TB may include chest x-ray, liquid culture of sputum, lymph node aspiration for acid-fast bacilli microscopy and culture, abdominal ultrasound. Non-tuberculosis mycobacterial infection should be considered in the diff erential diagnosis of patients who have a negative Xpert but a sputum or extra-pulmonary specimen with acid-fast bacilli. Prospective assessment of clinical data could help identify risk factors associated with mortality rates at fi xed time points.

Childhood TB and HIV
Most infants born to HIV-infected mothers still receive the Bacille Calmette-Guérin (BCG) vaccine, as HIV infection is usually not ruled out in the fi rst weeks of life. BCG can cause disseminated mycobacterial disease in HIV-infected children, including those on ART [58], and its effi cacy has been questioned in HIV-infected infants up to one year of age [59]. While waiting for a better and safer vaccine to prevent TB, the role of the BCG vaccine in infants living with HIV should be further investigated. Similarly, while isoniazid preventive therapy has been found to reduce the risk of TB disease by 72% and mortality by 56% in children living with HIV [60], little is known about the benefi t of IPT in the context of ART, the optimal duration and regimen for preventive TB therapy, the duration of protective eff ect, and the longterm adverse events among children living with HIV.
In the absence of mycobacterial confi rmation, the diagnosis of childhood TB relies on clinical features, exposure history, tuberculin skin test, relevant investigations for suspected pulmonary or extrapulmonary TB, and HIV testing in areas of high HIV prevalence [61]. Th e use of Xpert MTB/RIF on sputum samples and others specimens to diagnose TB among children living with HIV should be urgently evaluated. Preliminary results show that Xpert MTB/RIF performs better in those children living with HIV than in HIV-negative children (Catharina Boehme, personal communication).
Issues related to combined TB and ART treatment in children with HIV, such as optimal time to initiate ART, drug interactions, overlapping side eff ects, and diagnosis and management of immune reconstitution infl amma tory syndrome, have been poorly investigated. Current recommen dations are often drawn from data from HIV-uninfected children and adults [62] or rely on expert opinion. Th ere is an urgent need to conduct operational research to assess the feasibility, eff ectiveness and impact of these recom mendations and further inform national and international guidance on TB treatment, but also prevention and diagnosis in infants and children living with HIV.
Lastly, control of HIV-associated TB in children will not succeed without tackling their mothers. In the prevention of mother to child transmission of HIV, prevention of TB is crucial since maternal TB has been associated with increased vertical transmission of HIV [63].

Monitoring and evaluation
Functional TB/HIV monitoring and evaluation system with standardized indicators is crucial to inform policy and programming. Th e capacity to cross check and reconcile accurate data between TB and AIDS programmes will depend on the existence of standardized and integrated recording and reporting formats (patient cards, clinical fi les and registers), and will assist in identifying recording and reporting gaps so as to improve the monitoring and evaluation of collaborative TB/HIV activities. Eff ective and regular joint supervision by TB and AIDS programmes may greatly improve the process.

Access to care
Th ere are several structural and social barriers of access to TB and HIV services for people living with HIV [64]. Qualitative research to better understand the barriers to TB and HIV care for people living with HIV, their children and their families, both from a patient and health care worker's perspective, is important in improving access to and use of diagnostic services.

Civil society and public-private sector engagement
Enhancing the role of civil society organizations in implementation and research [65] and the promotion of sustainable collaboration between all public and private practitioners for TB care and control are essential for scaling up collaborative TB/HIV activities [66]. Th e best way to achieve these goals needs to be worked out.

Capacity building for operational research
Finally, enhanced capacity building for operational research within programmes in countries with high burdens of TB and HIV diseases and a transfer of northto-south expertise is urgently needed [67]. Trainings and courses on operational research for collaborative TB/HIV activities need to take this into consideration, and ensure that there is leadership by researchers, programme managers and other stakeholders based in the countries affl icted by the dual epidemic. Th is will result in ownership of the research that will facilitate its translation into policy and programming.

Conclusions
Promoting research is a key component of the Stop TB Strategy, which includes conducting "programme-based operational research" and "research on introducing new tools into practice" [68]. Th e importance of programmebased operational research is increasingly recognized, as exemplifi ed through the insertion of a concise Operational Research section in the new Global Plan to Stop TB 2011-2015 [4], and was recently identifi ed as a major area in which global action is urgently needed [69].
In this paper, we have reviewed the main operational research priorities that need to be addressed to improve the implementation of collaborative TB/HIV programme activities in areas aff ected by both epidemics [16,17,19].
We have seen how operational research is crucial in answering the translation gaps in implementing combina tions of existing and new tools and technologies to improve the prevention, early diagnosis and prompt treatment of TB among people living with HIV. Optimal control of TB in high HIV burden areas requires strong collaborative TB/HIV interventions through sound policy and programme environments that give due consideration to the local context and the respective epidemiology of TB and HIV. Moreover, since both TB and HIV control activities are placed within the larger frame of health services, due consideration should also be given to the health system infrastructure, as it has implications in terms of general health system functions (e.g., fi nances, regulations, policy development, management of human resources, procurement of drugs and supplies, and maintenance of health infrastructure) [70]. It is therefore important to address these questions within the larger context of health services research to identify measures that would facilitate wider uptake and scaling up of collaborative TB/HIV interventions for prevention, diagnosis and treatment, through eff ective service delivery models, including community-based interventions [71] In this respect, national TB and AIDS pro grammes should develop their operational research agendas and conduct the research they consider crucial for improving TB and HIV control in their settings in collaboration with research stakeholders. Support from academic institutions and international organizations is essential for implementing research questions that aim to assess new tools and strategies and to inform global policy, as well as to build local capacity in conducting operational research. Financial resources to implement these operational research questions should be mobilized and drawn from existing funding mechanisms (among others, the Global Fund to fi ght AIDS, Tuberculosis and Malaria, the President's Emergency Plan for AIDS Relief, and the Bill & Melinda Gates Foundation), and new funding mechanisms need to be sought. Enhancing fund allocation for operational research by national governments of countries highly aff ected by both diseases and ensuring the expedited translation of results of the research into policy and programming are points of equally crucial importance.