Bi-directional screening for tuberculosis and diabetes: a systematic review

Abstract

Objective To assess the yield of finding additional TB or diabetes mellitus (DM) cases through systematic screening and to determine the effectiveness of preventive TB therapy in people with DM. Methods We systematically reviewed studies that had screened for active TB or implemented preventive therapy for TB among people with DM, and those that screened for DM among patients with TB. We searched published literature through PubMed and EMBASE and included studies that reported the number of TB cases identified among people with DM; the number of DM cases identified among patients with TB, or the relative incidence of TB between people with DM who received a TB prophylaxis and those who did not. We assessed the yield of screening by estimating the prevalence of TB or DM in each study, the prevalence ratio and difference where comparison populations were available, and the number of persons to screen to detect an additional case of TB or DM. Results Twelve studies on screening for TB in people with DM and 18 studies on screening for DM in patients with TB met our inclusion criteria. Screening for TB in persons with DM demonstrated that TB prevalence in this population is high, ranging from 1.7% to 36%, and increasing with rising TB prevalence in the underlying population as well as with DM severity. Screening patients with TB for DM also yielded high prevalences of DM ranging from 1.9% to 35%. Two studies examining the role of TB preventive therapy in people with DM did not provide sufficient details for clear evidence of the effectiveness. Conclusion Active screening leads to the detection of more TB and DM with varying yield. This review highlights the need for further research in screening and preventive therapy. Dépistage bidirectionnel de la tuberculose et du diabète: une revue systématique Objectif: Evaluer le rendement du dépistage systématique pour la détection de cas supplémentaires de TB ou de diabète sucré (DS) et déterminer l’efficacité du traitement préventif de la TB chez les personnes atteintes de DS. Méthodes: Nous avons passé en revue systématiquement les études portant sur le dépistage actif de la TB ou l’implémentation d’un traitement préventif de la TB chez les personnes atteintes de DS et celles portant sur un dépistage du DS chez les patients TB. Nous avons recherché la documentation publiée sur PubMed et EMBASE et comprenant des études portant sur le nombre de cas de TB identifiés chez les personnes atteintes de DS, le nombre de cas de DS identifiés chez les patients TB, ou l’incidence relative de la TB chez les personnes atteintes de DS qui ont reçu une prophylaxie contre la TB et ceux qui n’en ont pas reçu. Nous avons évalué le rendement du dépistage en estimant la prévalence de la TB ou du DS dans chaque étude, le rapport et la différence de prévalence là où des comparaisons de populations étaient disponibles et le nombre de personnes à tester afin de détecter un cas supplémentaire de TB ou de DM. Résultats: 12 études sur le dépistage de TB chez les personnes atteintes de DS et 18 études sur le dépistage du DS chez les patients TB ont rempli les critères d’inclusion. Le dépistage de la TB chez les personnes atteintes de diabète a démontré que la prévalence de TB dans cette population est élevée, variant de 1,7%à 36% et augmentant avec la prévalence croissante de TB dans la population générale ainsi qu’avec la sévérité du DS. Le dépistage du DS chez les patients TB a également révélé des prévalences élevées de DS variant de 1,9%à 35%. Deux études portant sur le rôle du traitement préventif de la TB chez les personnes atteintes de DS n’ont pas fourni suffisamment de détails pour une preuve concrète de son efficacité. Conclusion: Le dépistage actif conduit à la détection de plus de TB et de DS avec un rendement variable. Cette revue met en évidence la nécessité de recherches supplémentaires dans le dépistage et le traitement préventif. Tamizaje Bidireccional para tuberculosis y diabetes: una revisión sistemática Objetivo: Evaluar el rendimiento de encontrar casos adicionales de TB o diabetes mellitus (DM) mediante el tamizaje sistemático, y determinar la efectividad de la terapia preventiva de la TB en personas con DM. Métodos: Hemos realizado una revisión sistemática de estudios en los que se tamizó para TB activa o se implementó una terapia preventiva para TB entre personas con DM, y aquellos que tamizaron para DM entre pacientes con TB. Se realizó una búsqueda de literatura a través de PubMed y EMBASE y se incluyeron los estudios que reportaban el número de casos de TB identificados en personas con DM; el número de casos de DM identificados entre pacientes con TB, o la incidencia relativa de TB entre personas con DM que recibieron una profilaxis para TB y aquellos que no la recibieron. Evaluamos el rendimiento del tamizaje estimando la prevalencia de TB o DM en cada estudio, la tasa de prevalencia y la diferencia donde había poblaciones disponibles para hacer la comparación, y el número de personas que han de ser tamizadas para detectar un caso adicional de TB o DM. Resultados: 12 estudios con tamizaje para TB en personas con DM y 18 estudios con tamizaje para DM en pacientes con TB cumplían con los criterios de inclusión. El tamizaje para TB en personas con diabetes demostró que la prevalencia de TB en esta población es alta, con rangos de 1.7% a 36%, y que aumenta con el aumento de la prevalencia de TB en la población subyacente así como con la severidad de la DM. El tamizaje de pacientes con TB para DM también arrojó una alta prevalencia de DM con un rango de 1.9% a 35%. Dos estudios que examinaban el papel de la terapia preventiva de TB en personas con DM no proveyó suficientes detalles que pudiese aportar una evidencia clara de la efectividad. Conclusión: El tamizaje activo lleva a la detección de más casos de TB y DM con un rendimiento variable. Esta revisión subraya la necesidad de más investigaciones sobre el tamizaje y la terapia preventiva. Currently, an estimated 285 million people live with diabetes mellitus (DM), a number which is expected to grow to at least 439 million by the year 2030 (IDF, 2010). At the same time, 9.6–13.6 million people live with tuberculosis [TB] disease and 1.1–1.7 million people die from the disease every year (WHO, 2009). Previous studies have demonstrated that DM not only increases the risk of active tuberculosis but also puts co-affected patients at increased risk for poor outcomes (Alisjahbana et al. 2007; Stevenson et al. 2007; Wu et al. 2007; Jeon & Murray 2008; Leung et al. 2008; Dooley et al. 2009). The potential for ‘syndemics’ of DM and TB in countries with high burdens of both diseases raises the question of how best to integrate management of these diseases (Restrepo 2007; Dooley & Chaisson 2009; Harries et al. 2009). Over the past two decades, TB control efforts have focused on the strategy of Directly Observed Therapy Short-Course (DOTS), a package of interventions, including diagnosis by sputum smear microscopy and supervised chemotherapy. While this strategy has improved both the detection rate of smear-positive cases and the outcomes of treatment, its impact on TB prevalence is less clear (Dye et al. 2005). Experts suggest that TB control would be further improved by intervening patients with known determinants of TB, including those with diabetes (Lonnroth et al. 2009, 2010). This entails both TB prevention through actions to diminish the prevalence of risk factors and targeted diagnostic and treatment interventions in risk groups, such as people with DM. Screening for active TB in people with DM could hasten case detection, which could lead to earlier therapy and prevention of transmission; the administration of preventive TB therapy in TB-infected people with DM could avert progression to TB. Conversely, screening for DM in patients with TB could improve case detection, early treatment and tertiary prevention of DM, and indirectly lead to better TB-specific treatment outcomes. Recognizing the opportunities offered by screening and preventive therapy, we systematically reviewed studies that had implemented screening or preventive therapy for TB among people with DM and those that screened for DM among patients with TB to assess the yield of finding additional TB or DM cases through active screening and to determine the effectiveness of preventive TB therapy in patients with diabetes. We conducted this systematic review to address three distinct aims: (1) to assess the yield of screening people with DM for the detection of TB in various settings, (2) to assess the yield of screening the patients with TB for the detection of DM in various settings and (3) to measure the effectiveness of TB preventive therapy in people with DM. For these aims, we conducted a literature search in PubMed from 1965 to May 2009 and in EMBASE from 1974 to May 2009 using the search strategy outlined below with no language restriction. For the PubMed search, we used the MeSH Terms: 1. ‘diabetes mellitus’ and 2. ‘tuberculosis’ [majr] and the Text terms 3. ‘prevent*’ OR ‘isoniazid*’ OR ‘chemoproph*’; 4. ‘detect*’ OR ‘screen*’ OR ‘diagnos*’ with the search string: 1 AND 2 AND (3 OR 4 in abstract). For the EMBASE search, we used the subject terms 1. ‘diabetes mellitus’ and 2. ‘tuberculosis’ [majr] and the text terms 3. ‘prevent*’ OR ‘isoniazid*’ OR ‘chemoproph*’; 4. ‘detect*’ OR ‘screen*’ OR ‘diagnos*’, with the search string: 1 AND 2 AND (3 OR 4 in abstract). We chose the root terms ‘detect*’, ‘screen*’ and ‘diagnos*’ to find articles that described a screening programme for people with DM. We also employed the root terms ‘prevent*’, ‘isoniazid*’ and ‘chemoproph*’ to find articles that described preventive therapy against TB among people with DM. We specifically included ‘isoniazid*’ as one of the search terms, given that isoniazid is the most commonly used preventive therapeutic drug against TB. We also searched the bibliographies of relevant literature and the abstracts of World Lung Conferences held in 2007 and 2008, limiting the search to these years because we expected high-quality studies reported in prior years to be published by May 2009. Among the retrieved citations, we examined the full texts of articles with abstracts that described or mentioned (1) the screening of people with DM for TB, (2) the screening of patients with TB for DM or glucose intolerance, (3) the effectiveness of preventive therapy in people with DM or (4) ‘diabetes’ and ‘tuberculosis’ but did not provide enough detail on the methods to determine if screening or preventive therapy were implemented. Our initial aim was to identify studies that assessed the effectiveness of TB screening in people with DM for preventing TB-related morbidity or mortality, but because of the lack of such studies, we focused on assessing the yield of screening for TB disease among people with DM. For this aim, we included studies that screened people with DM for TB using any of the following methods of identification: X-ray consistent with TB, positive sputum smear microscopy, positive mycobacterial culture and clinical evaluation. We excluded studies that did not allow computation of TB prevalence or TB incidence in the screened population and those that did not describe the age distribution of the screened population. Because our focus was on summarizing the yield of screening for TB in people with DM, we included studies even if they did not provide a non-diabetic control group; in the latter case, we searched for published estimates of TB prevalence in a comparable population based on surveys conducted within ±5 years in the same population that gave rise to the screened DM patients and conducted with the same screening methods as the screening study. To assess the yield of screening for DM among patients diagnosed with active TB, we included studies that screened for DM using the following types of blood glucose tests: random blood glucose, fasting blood glucose and oral glucose tolerance test. We excluded studies that did not describe the age distribution of the screened population and studies that only diagnosed impaired glucose tolerance. Here again, we did not require that studies include a control group but used an external estimation of the population prevalence of DM as a comparator if the prevalence was estimated from a survey conducted within ±5 years in the same population that gave rise to the screened TB cases and employed the same definition of DM as the study. To assess the efficacy of TB preventive therapy in people with DM, we included studies that had compared the incidence of TB in a diabetic population receiving any form of preventive therapy to a control population with DM. We included studies whether or not TB infection was confirmed to consider evidence on the potential effect of preventive therapy in people with DM even in settings where TB infection is not a prerequisite for preventive therapy. Two investigators (CYJ, SG) independently extracted data from the studies using standardized extraction forms. Evaluation and data extraction of non-English papers was carried out in conjunction with translators fluent in the language of the paper. For all studies, we extracted information on the study population, location, study period and method of recruitment of the screened population. For the studies on screening for TB among people with DM, we extracted information on the proportion of people with DM that were insulin dependent, the method of TB diagnosis, the sample size, the number of patients with TB identified in the screened populations (and control population, if available), as well as the age and sex distribution of the screened DM patients and identified TB patients. We noted whether studies included prevalent or incident TB cases and if incident TB, we extracted information on the period during which patients were followed for TB. For the studies on screening for DM among patients with TB, we extracted information on the method of DM diagnosis, the timing of DM diagnosis relative to the onset of TB treatment, the sample size, the number of people with DM identified in the studied TB population (and control population, if available), as well as the age and sex distribution of the screened TB patients and DM patients identified through screening. For studies on preventive therapy, we extracted information on the duration and type of regimen, the follow-up period, the incidence of TB in the intervention group and control groups, and the criteria by which intervention was assigned. Differences in the extracted information were by between the two data for non-English any information on extracted data was to be reviewed by the same We studies on screening for TB among people with DM those that assessed TB prevalence and those that assessed TB incidence through We the prevalence or the incidence of TB in people with DM and where data were For follow-up studies, we the TB by the incidence by the number of years of For studies that the number of TB cases diagnosed by X-ray from those diagnosed by we used the latter definition to TB We prevalence or incidence and prevalence or incidence to assess the relative and in yield of finding TB cases between the screened diabetic populations and the comparison populations where we the number of people to screen to detect one additional case of TB in a diabetic population by the of the prevalence or incidence difference for each study. To the potential yield in screening for TB in people with DM in various settings, we how patients with DM would need to be screened to detect one additional case of TB, TB prevalence ranging from 1 to for prevalence estimated from the screened populations and Because DM be in patients with TB a at the of treatment, we studies on screening for DM among people with TB (1) those that screening TB treatment (2) those that had screened the population prior to TB treatment and (3) those in which the timing of screening relative to the of TB treatment was not We the DM prevalence for the screened diabetic populations and We the prevalence and to assess the relative and between the screened TB populations and the comparison populations where We also the number of people to screen to detect one additional case of DM in a TB population by the of the prevalence or incidence difference for each study. To the potential yield in screening for DM in patients with TB in various settings, we how patients with TB would need to be screened to detect one additional case of DM, DM prevalence ranging from to for prevalence estimated from the screened populations and The PubMed and EMBASE search yielded we examined the full text of articles and excluded studies as they did not describe a screening study. of the bibliographies yielded additional studies for full text review of of we included 12 studies that reported the of TB among people with of these reported TB prevalence et al. et al. et al. et al. et al. et al. et al. & et al. et al. and 2 TB incidence et al. and the methods of screening in 1. The prevalence of active TB among people with DM from 1.7% in in to 36% in in the incidence from people with DM in to people with DM in of the 12 studies screened a non-diabetic control group or an of the TB prevalence in the population that gave rise to the study group within years of the study. prevalence from in and to in Among the studies that were by the of TB was more in those with insulin compared to those with diabetes with prevalence from to 2 the number of people with DM that would need to be screened to detect one additional case of TB based on prevalence from 2 to for varying of TB settings in which TB prevalence is at least people with DM would need to be screened to find a additional case of TB. in with TB such as where TB prevalence is estimated at screening people with DM would yield one or more cases of TB. of people with diabetes to screen to detect one additional case of tuberculosis by varying tuberculosis given prevalence in screening We identified 18 studies that met our inclusion criteria on screening for DM among patients with studies conducted screening TB treatment et al. et al. & et al. et al. et al. with of these screening for DM at during TB therapy et al. & et al. screened for DM only treatment & et al. et al. et al. and did not screening had & & et al. et al. et al. et al. 2009). and methods of screening in 4. of the 18 studies also reported the of screening for DM in a population TB. we were to find DM prevalence estimates for a comparison population based on prevalence surveys using the same method of screening for three of the studies that did not a control group et al. et al. et al. DM prevalence reported in studies that screened TB treatment from 1.9% in in to in in The prevalence of DM was in patients with TB in with prevalence of and from studies with a control population that had a for people with DM Among studies that conducted screening prior to TB treatment, the prevalence of DM from in to in Among those that did not the timing of DM screening relative to TB treatment, estimates from in to in these latter two of studies, diabetes prevalence was in patients with TB in the where available, with prevalence ranging from in to in from the studies that screened for DM at in the of TB treatment in et al. & et al. the prevalence of the of TB in in patients with tuberculosis tuberculosis 4 the number of TB patients that would need to be screened to detect one additional case of DM, varying DM prevalence and prevalence settings with a DM prevalence of (IDF, screening as as patients with TB would lead to the detection of at least one additional case of DM. of patients with tuberculosis to screen to detect one additional case of diabetes by varying diabetes given prevalence in screening We identified two studies in which people with DM had for prevention of TB. a study conducted in & in the patients with diabetes who had a of treatment for active TB were with isoniazid for a comparison group of people with DM treatment for active TB was not While the comparison group 18 cases of TB a of years of follow-up time, the intervention group no TB years of follow-up a study conducted in in the & investigators a of to patients with diabetes and compared to who were not patients were reported to incidence of TB compared to who did not any during a follow-up period of The study did not of TB whether the follow-up period during intervention or the for not in the control this systematic we the published of screening for TB in people with DM and screening for DM in patients with TB a and We that the prevalence of TB in screened people with DM was high, with estimates for populations in which active case finding is such as & et al. and in countries et al. 2009). The yield of screening for TB in people with DM increases with the prevalence of TB in the as well as with the of as by insulin that screened for DM among patients with TB also reported a of DM prevalence ranging from 1.9% to as high as with the reported for in which DM prevalence is For in where the DM prevalence is high at of the screened TB patients were to have DM et al. The studies reviewed on how to screening efforts in people with DM and patients with they were in of the studies followed an population to measure the effectiveness of screening in preventing TB or in disease outcomes. studies did not a control group and an This the computation of prevalence to a the studies did not provide details on the of the screened population, which the of finding TB and DM. would be to factors populations to screening. DM have in studies that for glucose prior to the of TB treatment, as TB disease an that be as DM. DM prevalence estimates were in studies that had screened prior to TB treatment or did not screening had that glucose screening for DM diagnosis be more TB treatment has the methods of screening for TB and diabetes to the TB or DM prevalence the studies used methods to TB, which have lead to to the of TB prevalence estimates was by more such as culture or smear microscopy those studies that reported from both X-ray and prevalence as assessed by was by culture et al. et al. TB screening studies in people with DM focus on TB TB preventive therapy the incidence of TB in two the lack of details of intervention method evidence for TB preventive therapy in people with DM. This systematic review a that research screening be a high on the research for both DM and TB et al. 2010). studies need to be conducted in diabetes with a focus on by of glucose patients with TB, research is to determine the and best methods for DM, on and by type of TB The question of TB preventive therapy only be through a which be and to Screening and to better DM control be a more of preventing TB and DM This review was by the and Lung the World and the World and by the World