Malnutrition found in 38.8% of elderly tuberculosis patients in a meta-analysis of fifteen studies
A meta-analysis of fifteen studies with 5,346 patients aged ≥60 years found a pooled malnutrition prevalence of 38.8%, with a wide range between clinical settings.
Strictly defined malnutrition affects 38.8% (95% CI: 30.5–47.9%) of elderly patients with pulmonary tuberculosis, according to a systematic review and meta-analysis in Frontiers in Nutrition. frontiersin.org
Frontiers in Nutritionfrontiersin.orgOpen source ↗ The analysis included fifteen studies with 5,346 patients aged ≥60 years. The work was led by researchers at the Chongqing Municipal Institute of Tuberculosis and Wushan County People's Hospital of Chongqing.
The authors searched PubMed, Embase, Web of Science, the Cochrane Library, CINAHL, China National Knowledge Infrastructure and Wanfang, from inception to September 2026. They note that prevalence estimates vary widely, risk-factor evidence is fragmented, and nutritional risk is often conflated with malnutrition.
How malnutrition was defined
Eligible studies used one of three approaches: the complete GLIM framework, a GLIM-compatible low-BMI phenotype (BMI < 18.5 kg/m2 alone), or MNA-based tools (MNA < 17 or MNA-SF ≤ 7). Study quality was assessed with design-specific JBI checklists. Prevalence was pooled with a random-effects generalized linear mixed model (binomial likelihood, logit link). Sensitivity analyses used an inverse-variance model with Hartung-Knapp-Sidik-Jonkman adjustment.
Pooled estimates and heterogeneity
Heterogeneity was substantial (I2 = 96.4%), and the 95% prediction interval was 13.2–72.6%. frontiersin.org
Frontiers in Nutritionfrontiersin.orgOpen source ↗ GLIM-based definitions gave 39.2% (95% CI: 31.3–47.7%) and MNA-based tools 37.9% (95% CI: 14.7–68.5%; p = 0.936).
In the exploratory three-tier breakdown, the single full-GLIM study reported 47.6% (95% CI: 45.0–50.2%). The low-BMI studies (k = 11) gave 38.4% (95% CI: 30.0–47.6%), and the MNA-based studies (k = 3) gave 37.9% (95% CI: 14.7–68.5%).
Hospital-based studies showed higher prevalence than community-based cohorts (44.6% vs. 21.7%, p < 0.001), although there were only two community studies. Excluding four TB-diabetes studies did not materially change the estimate (39.8%, 95% CI: 28.6–52.2%).
Few studies performed multivariable adjustment. Diabetes, drug-resistant tuberculosis and frailty were identified as potential risk factors, but most factors could not be pooled because definitions were inconsistent. frontiersin.org
Frontiers in Nutritionfrontiersin.orgOpen source ↗
The authors conclude that clinical setting was the main source of heterogeneity and that the evidence was insufficient to judge whether different diagnostic definitions are comparable. They call for multicenter prospective studies using the complete GLIM workflow with standardized, age-appropriate criteria.
