Incidence and Predictors of Tuberculosis among Adult Diabetic Patients, Debre Markos Referral Hospital, Northwest Ethiopia, 2018: A Retrospective Cohort Study

Diabetes as a risk factor for tuberculosis disease

BACKGROUND

Tuberculosis (TB) is amongst the leading causes of death from an infectious disease, with an estimated 1.3 million deaths from TB in 2022. Approximately 25% of the global population is estimated to be infected with the TB bacterium, giving rise to 10.6 million episodes of TB disease in 2022. The prevalence of diabetes influences TB incidence and TB mortality. It is associated not only with an increased risk of TB disease but also death during TB treatment, TB relapse after treatment completion and multidrug-resistant TB. Since 2011, the World Health Organization (WHO) has recommended collaborative TB and diabetes activities as outlined in the Collaborative Framework for Care and Control of TB and Diabetes.

OBJECTIVES

To determine the prognostic value of diabetes mellitus (DM) in the general population of adults, adolescents and children for predicting tuberculosis disease.

SEARCH METHODS

We searched the literature databases MEDLINE (via PubMed) and WHO Global Index Medicus, and the WHO International Clinical Trials Registry Platform (ICTRP) on 3 May 2023 (date of last search for all databases); we placed no restrictions on the language of publication.

SELECTION CRITERIA

We included retrospective and prospective cohort studies, irrespective of publication status or language. The target population comprised adults, adolescents and children from diverse settings, encompassing outpatient and inpatient cohorts, with varying comorbidities and risk of exposure to tuberculosis.

DATA COLLECTION AND ANALYSIS

We used standard Cochrane methodology and the Quality In Prognosis Studies (QUIPS) tool. Prognostic factors assessed at enrolment/baseline included diabetes, as defined by the individual studies, encompassing patient-reported status, abstracted from medical records or claims data, or diagnosed by plasma glucose/glycosylated haemoglobin. The primary outcome was the incidence of tuberculosis disease. The secondary outcome was recurrent TB disease. We performed a random-effects meta-analysis for the adjusted hazard ratios, risk ratios, or odds ratios, employing the restricted maximum likelihood estimation. We rated the certainty of the evidence using the GRADE approach.

MAIN RESULTS

We included 48 cohort studies with over 61 million participants from the six WHO regions. However, the representation was variable as eight population-based studies were from South Korea and 19 from China, with overlapping study periods, and only one from the African region (Ethiopia). All studies included adults, and nine studies also included children and adolescents. Most studies diagnosed DM based on clinical records, including fasting blood glucose levels or glucose-lowering treatments. The studies did not distinguish between type 1 and type 2 DM; only one study focused on type 1 DM. Diagnosis and exclusion of TB were performed using culture or molecular WHO-recommended rapid diagnostic tests (mWRD) in only 12 studies, which could have biassed the effect estimate. The median follow-up time was five years (interquartile range 1.5 to 10, range 1 to 16.9), and the studies primarily reported an adjusted hazard ratio from a multivariable Cox-proportional hazard model. Hazard Ratios (HR) The HR estimates represent the highest certainty of the evidence, explored through sensitivity analyses and excluding studies at high risk of bias. We present 95% confidence intervals (CI) and prediction intervals, which show between-study heterogeneity represented in measuring the variability of effect sizes (i.e. the interval within which the effect size of a new study would fall considering the same population of studies included in the meta-analysis). DM may increase the risk of tuberculosis disease (HR 1.90, 95% CI 1.51 to 2.40; prediction interval 0.83 to 4.39; 10 studies; 11,713,023 participants). The certainty of the evidence is low, due to a moderate risk of bias across studies and inconsistency. Considering a risk without diabetes of 129 cases per 100,000 population, this represents 102 more (59 to 153 more) cases per 100,000. When stratified by follow-up time, the results are more consistent across < 10 years follow-up (HR 1.52, 95% CI 1.47 to 1.57; prediction interval 1.45 to 1.59; 7 studies; 10,380,872 participants). This results in a moderate certainty of the evidence due to a moderate risk of bias across studies. However, at 10 or more years of follow-up, the estimates yield a wider CI and a higher HR (HR 2.44, 95% CI 1.22 to 4.88; prediction interval 0.09 to 69.12; 3 studies; 1,332,151 participants). The certainty of the evidence is low due to the moderate risk of bias and inconsistency. Odds Ratio (OR) DM may increase the odds of tuberculosis disease (OR 1.61, 95% CI 1.27 to 2.04; prediction interval 0.96 to 2.70; 4 studies; 167,564 participants). Stratification by follow-up time was not possible as all studies had a follow-up < 10 years. The certainty of the evidence is low due to a moderate risk of bias and inconsistency. Risk Ratio (RR) The RR estimates represent the highest certainty of the evidence, explored through sensitivity analyses and excluding studies at high risk of bias. DM probably increases the risk of tuberculosis disease (RR 1.60, 95% CI 1.42 to 1.80; prediction interval 1.38 to 1.85; 6 studies; 44,058,675 participants). Stratification by follow-up time was not possible as all studies had a follow-up < 10 years. The certainty of the evidence is moderate due to a moderate risk of bias.

AUTHORS' CONCLUSIONS

Diabetes probably increases the risk of developing TB disease in the short term (< 10 years) and may also increase the risk in the long term (≥ 10 years). As glycaemic control and access to care may be potential effect modifiers of the association between diabetes and the risk of TB disease, the overall estimates should be interpreted with caution when applied locally. Policies targeted at reducing the burden of diabetes are needed to contribute to the aims of ending TB. Large population-based cohorts, including those derived from high-quality national registries of exposures (diabetes) and outcomes (TB disease), are needed to provide estimates with a high certainty of evidence of this risk across different settings and populations, including low- and middle-income countries from different WHO regions. Moreover, studies including children and adolescents and currently recommended methods for diagnosing TB would provide more up-to-date information relevant to practice and policy.

FUNDING

World Health Organization (203256442) REGISTRATION: PROSPERO registration: CRD42023408807.

Undernutrition as a risk factor for tuberculosis disease

BACKGROUND

Tuberculosis (TB) is a leading cause of mortality due to an infectious disease, with an estimated 1.6 million deaths due to TB in 2022. Approximately 25% of the global population has TB infection, giving rise to 10.6 million episodes of TB disease in 2022. Undernutrition is a key risk factor for TB and was linked to an estimated 2.2 million TB episodes in 2022, as outlined in the World Health Organization (WHO) Global Tuberculosis Report.

OBJECTIVES

To determine the prognostic value of undernutrition in the general population of adults, adolescents, and children for predicting tuberculosis disease over any time period.

SEARCH METHODS

We searched the literature databases MEDLINE (via PubMed) and WHO Global Index Medicus, as well as the WHO International Clinical Trials Registry Platform (ICTRP) on 3 May 2023 (date of last search for all databases). We placed no restrictions on the language of publication.

SELECTION CRITERIA

We included retrospective and prospective cohort studies, irrespective of publication status or language. The target population comprised adults, adolescents, and children from diverse settings, encompassing outpatient and inpatient cohorts, with varying comorbidities and risk of exposure to tuberculosis.

DATA COLLECTION AND ANALYSIS

We used standard Cochrane methodology and the Quality In Prognosis Studies (QUIPS) tool to assess the risk of bias of the studies. Prognostic factors included undernutrition, defined as wasting, stunting, and underweight, with specific measures such as body mass index (BMI) less than two standard deviations below the median for children and adolescents and low BMI scores (< 18.5) for adults and adolescents. Prognostication occurred at enrolment/baseline. The primary outcome was the incidence of TB disease. The secondary outcome was recurrent TB disease. We performed a random-effects meta-analysis for the adjusted hazard ratios (HR), risk ratios (RR), or odds ratios (OR), employing the restricted maximum likelihood estimation. We rated the certainty of the evidence using the GRADE approach.

MAIN RESULTS

We included 51 cohort studies with over 27 million participants from the six WHO regions. Sixteen large population-based studies were conducted in China, Singapore, South Korea, and the USA, and 25 studies focused on people living with HIV, which were mainly conducted in the African region. Most studies were in adults, four in children, and three in children and adults. Undernutrition as an exposure was usually defined according to standard criteria; however, the diagnosis of TB did not include a confirmatory culture or molecular diagnosis using a WHO-approved rapid diagnostic test in eight studies. The median follow-up time was 3.5 years, and the studies primarily reported an adjusted hazard ratio from a multivariable Cox-proportional hazard model. Hazard ratios (HR) The HR estimates represent the highest certainty of the evidence, explored through sensitivity analyses and excluding studies at high risk of bias. We present 95% confidence intervals (CI) and prediction intervals, which present between-study heterogeneity represented in a measurement of the variability of effect sizes (i.e. the interval within which the effect size of a new study would fall considering the same population of studies included in the meta-analysis). Undernutrition may increase the risk of TB disease (HR 2.23, 95% CI 1.83 to 2.72; prediction interval 0.98 to 5.05; 23 studies; 2,883,266 participants). The certainty of the evidence is low due to a moderate risk of bias across studies and inconsistency. When stratified by follow-up time, the results are more consistent across < 10 years follow-up (HR 2.02, 95% CI 1.74 to 2.34; prediction interval 1.20 to 3.39; 22 studies; 2,869,077 participants). This results in a moderate certainty of evidence due to a moderate risk of bias across studies. However, at 10 or more years of follow-up, we found only one study with a wider CI and higher HR (HR 12.43, 95% CI 5.74 to 26.91; 14,189 participants). The certainty of the evidence is low due to the moderate risk of bias and indirectness. Odds ratio (OR) Undernutrition may increase the odds of TB disease, but the results are uncertain (OR 1.56, 95% CI 1.13 to 2.17; prediction interval 0.61 to 3.99; 8 studies; 173,497 participants). Stratification by follow-up was not possible as all studies had a follow-up of < 10 years. The certainty of the evidence is very low due to the high risk of bias and inconsistency. Contour-enhanced funnel plots were not reported due to the few studies included. Risk ratio (RR) Undernutrition may increase the risk of TB disease (RR 1.95, 95% CI 1.72 to 2.20; prediction interval 1.49 to 2.55; 4 studies; 1,475,867 participants). Stratification by follow-up was not possible as all studies had a follow-up of < 10 years. The certainty of the evidence is low due to the high risk of bias. Contour-enhanced funnel plots were not reported due to the few studies included.

AUTHORS' CONCLUSIONS

Undernutrition probably increases the risk of TB two-fold in the short term (< 10 years) and may also increase the risk in the long term (> 10 years). Policies targeted towards the reduction of the burden of undernutrition are not only needed to alleviate human suffering due to undernutrition and its many adverse consequences, but are also an important part of the critical measures for ending the TB epidemic by 2030. Large population-based cohorts, including those derived from high-quality national registries of exposures (undernutrition) and outcomes (TB disease), are needed to provide high-certainty estimates of this risk across different settings and populations, including low and middle-income countries from different WHO regions. Moreover, studies including children and adolescents and state-of-the-art methods for diagnosing TB would provide more up-to-date information relevant to practice and policy.

FUNDING

World Health Organization (203256442).

REGISTRATION

PROSPERO registration: CRD42023408807 Protocol: https://doi.org/10.1002/14651858.CD015890.