The research team followed the PRISMA 2020 reporting guidelines and searched six major databases—PubMed, Embase, PsycINFO, Web of Science, Scopus, and the WHO Global Index Medicus—up to August 2025, with no language restrictions. To be eligible, studies had to be observational in design, assess depression using either a validated rating scale or a clinical diagnosis, evaluate patients undergoing anti-tuberculosis therapy, and report treatment outcomes. Thirteen studies ultimately met these criteria. Risk of bias in the included studies was appraised with the Newcastle-Ottawa Scale, a standard instrument for non-randomized research, and the pooled estimates were generated using random-effects modeling with the DerSimonian-Laird method, which assumes that the true effect may vary from study to study rather than being a single fixed quantity.

The headline result is striking. Across the pooled data, depression was associated with substantially higher odds of an unfavorable end-of-treatment outcome, with an odds ratio of 3.66 and a 95 percent confidence interval spanning 1.51 to 8.90. In practical terms, depressed tuberculosis patients were roughly three and a half times more likely to experience treatment failure, abandonment of therapy, or death than their non-depressed counterparts. Heterogeneity between studies was high, with an I-squared statistic of 81.2 percent, indicating that the underlying effect varied considerably across populations and settings. The authors also calculated a 95 percent prediction interval ranging from 0.20 to 65.99, a sobering reminder that in a future study the effect could, in principle, be very small or extraordinarily large depending on context.

Because heterogeneity and small-study effects can distort meta-analytic conclusions, the team subjected their findings to a battery of sensitivity analyses. Leave-one-out analyses, in which each study is removed in turn and the pooled estimate recalculated, produced odds ratios ranging from 1.31 to 4.79 for unfavorable outcomes. While this range confirms that the direction of the effect remained consistent, it also reveals how sensitive the magnitude is to individual studies. The Doi plot, a graphical alternative to the traditional funnel plot for detecting publication bias, showed major asymmetry, quantified by an LFK index of 6.96—a value well above the thresholds conventionally interpreted as indicating minor or no asymmetry. This suggests that smaller studies reporting stronger associations may be overrepresented in the literature, a common problem in observational meta-analyses that readers should keep in mind when weighing the headline figure.

The analysis of treatment non-compliance, based on seven of the included studies, told a similar but somewhat more stable story. Depression increased the odds of non-compliance with anti-tuberculosis therapy, with a pooled odds ratio of 2.85 and a confidence interval of 1.48 to 5.46. Heterogeneity was again high at 80.6 percent, and the prediction interval stretched from 0.34 to 23.79. The Doi plot showed only minor asymmetry here, with an LFK index of 1.85, and leave-one-out estimates clustered more tightly, ranging from 1.99 to 2.67. Taken together, these checks suggest that the link between depression and failure to adhere to the demanding months-long drug regimen is robust, even if its precise size varies across settings.

Perhaps the most technically interesting result emerged when the authors restricted the non-compliance analysis to the four studies that reported adjusted odds ratios—estimates that statistically account for confounding variables such as age, sex, HIV status, socioeconomic position, and disease severity. When these adjusted estimates were pooled, the association not only persisted but strengthened, yielding an adjusted odds ratio of 3.48 with a confidence interval of 2.51 to 4.45. Crucially, heterogeneity collapsed to zero percent. The disappearance of heterogeneity in the adjusted analysis is notable because it implies that much of the variability seen in the crude estimates may stem from differences in the confounder profiles of the study populations rather than from genuine differences in the underlying biological or behavioral relationship. In other words, once confounders are properly controlled, depression appears to act as a strong and remarkably consistent independent predictor of non-compliance.

The mechanistic plausibility of this link is not difficult to construct. Tuberculosis treatment typically requires six months or more of daily multidrug therapy, and in drug-resistant cases the regimen can stretch to two years with injectable agents and burdensome side effects. Depression is known to sap motivation, impair executive function, and promote social withdrawal, all of which can erode the daily discipline that successful therapy demands. There are also biological threads worth exploring: inflammatory cytokines such as interleukin-6 and tumor necrosis factor-alpha, which are elevated in tuberculosis, have been implicated in depressive symptomatology, raising the possibility of a bidirectional relationship in which infection worsens mood and poor mood worsens treatment engagement. The included studies used a variety of validated instruments, including the Patient Health Questionnaire-9, the Beck Depression Inventory, the Hospital Anxiety and Depression Scale, the Center for Epidemiologic Studies Depression Scale, the Hamilton Depression Rating Scale, and the Aga Khan University Anxiety and Depression Scale, as well as clinical diagnoses under DSM and ICD frameworks, lending credibility to the exposure assessment.

The public health implications are considerable. The World Health Organization’s End TB Strategy and the classic DOTS framework of directly observed treatment were built on the assumption that patients, once diagnosed, will complete therapy if drugs and supervision are available. This meta-analysis suggests that assumption fails a substantial subset of patients whose untreated depression quietly sabotages their recovery. The authors argue that routine depression screening at the start of tuberculosis treatment, followed by integrated, stepped mental health care within tuberculosis services, is likely to improve both adherence and treatment success. Stepped care models, which match the intensity of psychological and pharmacological intervention to symptom severity, have proven feasible in other chronic disease settings and could be adapted for tuberculosis programs in high-burden countries where mental health resources are scarce.

Several caveats temper the enthusiasm. All thirteen included studies were observational, so causality cannot be definitively established; reverse causation and residual confounding remain possible, and the major asymmetry detected in the outcome analysis hints at publication bias. The high heterogeneity in the crude analyses and the wide prediction intervals mean that the pooled odds ratios should be read as central tendencies within a broad distribution of context-dependent effects rather than as universal constants. The authors themselves call for prospective trials of integrated depression screening and treatment models within tuberculosis care, which would provide the interventional evidence needed to move from association to actionable policy.

Even with those limitations, the study lands at a consequential moment. Drug-resistant tuberculosis is rising in many regions, and every patient who abandons therapy risks not only personal relapse but the selection of resistant strains that are far costlier and harder to cure. If depression triples the odds of that abandonment, then mental health care stops being a luxury adjunct and becomes core tuberculosis infrastructure. The message for clinicians, program managers, and funders is blunt: a curable disease cannot be cured by drugs alone when the mind carrying the regimen is left untreated. Screening a tuberculosis patient for depression takes minutes; the analysis suggests it could be one of the highest-yield interventions in the entire treatment cascade.

Subject of Research: The association between depression and treatment compliance and outcomes in tuberculosis patients

Article Title: Association between depression and compliance to treatment of tuberculosis and its outcomes: a systematic review and meta-analysis

Article References: Han, J., Yang, F., Wang, J., & Chen, Y. (2026). Association between depression and compliance to treatment of tuberculosis and its outcomes: a systematic review and meta-analysis. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08627-4

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08627-4

Keywords: tuberculosis, depression, meta-analysis, treatment compliance, mental health, systematic review, odds ratio, public health, adherence, BMC Psychiatry, drug-resistant TB, screening