Tuberculosis related disability: a systematic review and meta-analysis

We performed a systematic review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [19]. We searched PubMed, Embase, and Web of Science databases for studies that reported on permanent disability associated with TB, using pre-specified search terms. We checked the reference lists of included papers for additional relevant references and performed a backwards and forwards citation search. Our search strategy (Additional file 1) was developed by a senior research information specialist (JC) and respiratory physician (AB), both with extensive experience in conducting systematic reviews in health and medicine.

The screening of articles by title and abstract was carried out independently by three researchers (KW, KAA and SC) in Rayyan [20]. Full-text papers were then independently screened by four researchers (KW, KAA, CK, and SC) using eligibility criteria described below. Disagreement was resolved through discussion and consensus.

Data were extracted into a Microsoft Excel 2016 spreadsheet (Microsoft, Redmond, Washington, USA) by four researchers (KW, SC, CK, and KA). The data extraction spreadsheet was pilot tested and refined before extraction. The lists of variables included in the data extraction tool are available in the Additional file 3. All included articles were assessed for quality using a modified Ottawa Newcastle quality assessment scoring tool [21].

Meta-analysis was performed to estimate the pooled prevalence of each form of disability using the inverse variance heterogeneity model. Stratified analyses were conducted by country income-level and TB type, separately for each disability when two or more studies were available on the outcome of interest (see Additional file 4 for details).

This review was registered in the Prospective Register of Systematic Reviews (PROSPERO, CRD42019147488). Ethical approval was not sought for this study as it includes an analysis of secondary data.

The search strategy yielded 3485 unique publications, 619 articles remained after the title and abstract screening. After full-text review, 124 publications (comprising 164 datasets) were included in the review. A backward and forward citation search found 410 publications, of which 53 were not identified in the original search; seven of these were subsequently included in the final analysis. Some studies reported more than one type of disability, and thus a total of 175 datasets (217,475 patients) from 131 unique studies were included (Fig. 1).

The characteristics of the included studies are presented in Table 2. The included studies were conducted in 49 countries. The majority of studies were conducted in India 24.6% (n=43) followed by South Africa 9.7% (n=17) and Brazil 5.1% (n=9). The mean age of study participants was 36.7 years (±16.3) and 59.6% of cases were male. More than one-third of studies (39%, n=68) reported that their study population had DS-TB, while 29.1% (n=51) included patients with DR-TB. The site of TB disease was not reported in 41.1% of studies (n=72), while 28.6% (n=50) and 16.0% (n=28) included patients with pulmonary TB (PTB) and extra-pulmonary TB (EPTB), respectively. More than half of the studies (52.9%, n=90) reported that disabilities were diagnosed during TB treatment.

The review showed that the most common type of disabilities were mental health disorders (23.1%), respiratory impairment (20.7%), musculoskeletal impairment (17.1%), hearing impairment (14.5%), visual impairment (9.8%), renal impairment (5.7%), and neurological impairment (1.6%).

There was large heterogeneity in the prevalence of disability. Two variables, namely country income level and type of TB, were identified as the source of heterogeneity across all types of disability and therefore were used as the primary variables of stratification.

Table 3 shows the number of studies and the prevalence of disabilities associated with TB, stratified by country income level. A total of 43 studies reported respiratory impairment. Nearly two-thirds of patients in LICs (61.2%) and just over half in LMICs (56.1%) experienced some form of respiratory impairment. The prevalence was low among HICs (14.9%) and UMICs (15.3%) (Fig. 2). Similarly, the highest prevalence of patients with mental health disorders was observed among patients in LICs (42%) followed by LMICs (31.3%), UMICs (30.6%), and HICs (4.3%; Fig. 3). The prevalence of patients with neurological function impairment was highest in LMICs (25.6%) and UMICs (15.9%) and lowest in LICs (5.9%) and HICs (1.3%; Fig. 4). The highest prevalence of TB patients who experienced hearing impairment (59.1%) was reported in HICs and UMICs 27.4% and 11.0% and 5% were reported from LMICs and LICs, respectively (Fig. 5).

Among patients with DS-TB, the prevalence of respiratory impairment was 33.1% while 21.9% of patients reported mental disorders, 12.5% reported neurological impairment, 11.9% reported visual impairment, and 2.3% reported hearing impairment (Table 3). Among patients with DR-TB, the prevalence of patients reporting respiratory impairment was 58.7% while 26% reported mental disorder, 15% reported hearing impairment, 4.6% reported neurological impairment, and 2.7% reported visual impairment. Studies including patients with DS-TB and DR-TB (with the inclusion of an injectable agent) reported that 37.2% of patients had a neurological impairment, 33.1% had respiratory impairment, 25% had a hearing impairment, and 20.7% had a mental disorder, with none of the studies reporting visual impairment. Neurological impairment was also high in studies which included DS-TB and DR-TB patients who did not receive an injectable agent (28.5%). Additional information on disabilities by HIV status, timing of disability diagnosis, and study design is provided in Table 3.

The quality of included studies was low to moderate overall, with a median score of 5 points (the maximum score is 9 points). Of the included 131 studies, only 11 studies had a score of 8 or 9 points, regarded as high-quality studies. The remaining studies scored 7 points or less, with 28 of them scoring 4 points or lower, classified as a low-quality study. Additional file 5: Table S1 presents the results of the quality assessment scores and Additional file 6: Table S2 includes the quality assessment tools.

Respiratory impairment was the most common disability identified in this review. There was inconsistency in how respiratory impairment was diagnosed and reported. However, the prevalence of respiratory impairment was heterogeneous when stratified by country income level. The highest prevalence was reported in LICs (61.2%) and LMICs (56.1%). This may correlate with the high burden of TB in these countries, difficulties in accessing health care, or poverty. Poverty is widely recognized as a risk factor for TB and may also result in respiratory impairment [149‐151]. High rates of respiratory impairment in LICs and LMICs may also be partially explained by low levels of health service coverage [152, 153]. High coverage of essential health services, including early access to TB diagnosis, treatment, and care, with appropriate monitoring of patients while on treatment, may minimize the long-term sequelae and disabilities associated with TB [154]. Other causes of lung diseases such as cigarette smoking and air pollution (indoor and outdoor) may contribute to the high prevalence of respiratory disabilities observed in LICs and LMICs [155, 156]. One systematic review reported a positive association between a history of TB treatment and chronic respiratory diseases, including COPD and bronchiectasis [157]. This association was much stronger in non-smokers and in high TB incidence countries [157]. WHO has recommended an integrated strategy to manage respiratory patients in primary health care settings with a focus on priority respiratory diseases, particularly TB [158].

Respiratory impairment was also higher among those with DR-TB. We found an almost twofold increase in the prevalence of respiratory impairment among patients with DR-TB (58.7%), compared with DS-TB (33.1%). This is consistent with previous research demonstrating a greater prevalence of COPD among successfully treated MDR-TB patients compared to patients treated for DS-TB and community controls [159]. This supports the notion of integration of DR-TB programs with respiratory health care. Importantly, there are currently no international guidelines that recommend screening for respiratory impairment after TB treatment, although there is interest in this from several clinical and public health groups [160]. Post TB treatment respiratory care including outpatient pulmonary rehabilitation may be beneficial for some TB survivors, especially in countries with a high burden of DR-TB. DR-TB treatment completion may be a possible entry point into these programs, where they exist. Similarly, national TB programs may want to consider how they provide incentives and enablers to patients with DR-TB so that they can monitor patients closely, or consider how they can link patients to services such as social protection schemes and disability services. Both are key interventions included in the End TB Strategy [161] and in many national TB strategic plans. While incentives and enablers are frequently provided, linkages to social protection and disability services are less frequently implemented.

Mental health disorders have historically been neglected as a focus in TB research [162‐164]. Our review revealed a high prevalence of mental health disorders such as depression, anxiety and mood disorders, post-traumatic stress disorder (PTSD), and psychosis among TB patients, with substantial variation by country income level. The highest prevalence of mental health disorders was reported in LICs (42%) and the lowest prevalence in HICs (4.3%). Although our study did not include comparison data for the general population, based on other literature, we note that the prevalence of mental health disorders among TB patients in our review is higher than the prevalence of mental health disorders among the general population [165]. The prevalence of mental health disorders in our review was similar to the prevalence of mental health disorders among people with other chronic diseases such as HIV infection [166], diabetes mellitus [167], and cancer [168], from previous systematic reviews.

The relationship between mental health disorders and TB may be specific to the socioeconomic context or other factors such as health care affordability. Also, it has been well documented that TB patients and their families frequently face stigma and discrimination [169, 170]. Depression, anxiety, and other mental health disorders could be connected to this experience of stigma, loss of identity, ongoing symptoms, and the socioeconomic consequences of TB [171]. Mental health disorders can contribute to an inability to complete TB treatment and subsequently to disability [172]. The high burden of mental health disorders associated with TB suggests that additional efforts are required to improve TB care [173, 174].

We found hearing impairment (hearing loss) among TB survivors to be common, particularly among patients with DR-TB or after taking second-line TB medications. The prevalence of hearing impairment among patients with DR-TB was 15%, which is seven times higher than the prevalence of hearing impairment among patients with DS-TB (2.3%). The disorder of hearing in patients on second-line TB medications, such as the aminoglycosides (i.e., amikacin, kanamycin, and streptomycin), is common [36, 175]. A previously published review of aminoglycoside-induced hearing impairment among TB patients also reported a high incidence of ototoxicity (7–90%) [176]. The appropriate use of TB medications should help health care providers prevent hearing loss among patients. Therefore, WHO now recommends MDR-TB treatment without the aminoglycosides [174].

We found that the prevalence of hearing impairment was higher in HICs (59%) and UMICs (27%) compared to LMICS (11%) and LICs (5%). This could be due to differences in diagnostic methods or the availability of diagnostic (auditory) equipment in HICs and UMICs to assess hearing impairment [22]. Ascertainment and/or publication bias may also be relevant here as few studies were available from LMICs and LICs. Different audiological assessment methods were used for the diagnosis of hearing impairment in our included studies, including otoscopy, pure tone audiometry, otoacoustic product emissions, and automated auditory brainstem response testing [22]. Audiometry was not always available for all patients to assess hearing impairment at baseline, during treatment, and after completion of TB treatment, to quantify the timeline of hearing loss. As a result, it was not possible to establish the main cause of hearing loss among patients with TB in this review. However the prevalence of hearing impairment in this review (14.6%) is substantially higher than the global estimates of people with disabling hearing loss in 2018 (6.1%) [177]. It is worthwhile to highlight the definition of hearing impairment as defined in this review and the local estimates may be different [174]. The hearing impairment could have a considerable effect on the quality of life, work, and social relationships [22, 178]. Therefore, hearing assessments for TB patients receiving aminoglycosides should be included as part of the management package. In addition, rehabilitation packages for those with hearing impairment should be offered routinely [22, 179].

The patients on the second-line injectable drugs reported more than 37% neurological impairment compared to patients without an injectable TB medication (28.5%). The most common types of TB-related neurological impairments reported in our review were paraplegia, hemiplegia, cranial nerve palsies, peripheral neuropathy, hydrocephalus, and visual loss. These neurological impairments were permanent and irreversible and therefore have long-term functional, social, economic, and psychological consequences for affected patients [180, 181]. TB of the central nervous system accounts for 5–10% of all EPTB globally, with TB meningitis, intracranial TB, and spinal TB being some of the most severe forms of TB [182, 183]. TB of the spine (or Pott’s disease) affects the intervertebral discs and adjacent vertebrae, which may result in vertebral collapse, destruction, skeletal deformities, and disability [184, 185]. In addition, compression of the spinal cord and/or nerves may result in neurologic deficits [186]. To reduce the burden of neurological deficits in children from TB meningitis, improving BCG vaccination coverage in countries with low coverage of BCG is an important intervention [187].

The findings of our review suggest a pressing need to prevent or screen for TB-related disability among TB patients and survivors. Strategies to prevent or reduce TB-related disability include improving access to health care, promoting early TB diagnosis, appropriate use of TB medications, and providing training for health care workers. Adverse event monitoring, pharmacovigilance, therapeutic drug monitoring, and providing incentives and enablers for patients for treatment adherence or compliance and report adverse events should be introduced as part of the TB treatment package. After TB treatment, care including follow-up and continued monitoring for possible disability or sequelae should be initiated urgently.

This systematic review has several limitations. There was large heterogeneity in the prevalence of disabilities across studies which limited our ability to conduct a meta-analysis for all type of disabilities. There was also a large amount of missing data noted in our studies; for example, nearly 25% of studies had missing data for the type of TB, and HIV status was missing for 50% of the included studies. Therefore, we were unable to include these variables in the main analysis to explore the heterogeneity in these variables. In addition, in some studies, it was not possible to determine the temporal nature of the disability (i.e., whether disability occurred before, during, or after TB treatment) because the data were collected from studies that used cross-sectional study designs. For example, more than half of the papers (53%) included in the mental health impairment were cross-sectional studies. We may have some misclassification of disability for this reason. However, we attempted to explain this by conducting subgroup analysis. Moreover, we did not include studies published in languages other than English; 63 studies were excluded for this reason. Therefore, we may be missing important studies, from high TB and DR-TB burden countries such as China, the Russian Federation, and others. Lastly, as with all meta-analyses, the validity of the results is limited by the conduct and reporting of the studies from which the data were extracted and pooled.