Hazard function curves of active tuberculosis for different body mass index categories in the overall Cox model.
Leung CC, Lam TH, Chan WM, Yew WW, Ho KS, Leung G, Law WS, Tam CM, Chan CK, Chang KC. Lower Risk of Tuberculosis in Obesity. Arch Intern Med. 2007;167(12):1297-1304. doi:10.1001/archinte.167.12.1297
Copyright 2007 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2007
Obesity is increasingly prevalent in both developed and developing areas. Although undernutrition is well associated with tuberculosis, few studies have systematically examined the association with obesity.
A cohort of 42 116 individuals 65 years or older enrolled at 18 health centers for elderly patients in Hong Kong, China (which has a tuberculosis incidence of approximately 90 per 100 000 population), in 2000 were followed up prospectively through the territory-wide tuberculosis registry for the development of active tuberculosis from 3 months after enrollment until December 31, 2005, using the identity card number as the unique identifier. The association with body mass index (BMI; calculated as weight in kilograms divided by the square of height in meters), as categorized by the Asian standards, was assessed with the control of other baseline characteristics.
Obese (BMI ≥30) and overweight (BMI, 25 to <30) individuals were at significantly lower risks of developing active tuberculosis than normal-weight individuals (BMI, 18.5 to <25), with hazard ratios (95% confidence intervals) of 0.36 (0.20-0.66) and 0.55 (0.44-0.70), respectively, after adjustment for baseline demographic, social, and clinical variables. An inverse linear association was observed predominantly for pulmonary but not extrapulmonary tuberculosis. This association persisted after controlling for potential confounders or excluding individuals with known tuberculosis risk factors.
Obesity is associated with a lower risk of active pulmonary tuberculosis in the older population of Hong Kong. The presence of such a strong but selective association across the whole spectrum of BMI could have major biological, clinical, and/or epidemiological implications. Further studies are indicated to explore the underlying mechanisms, potential clinical utilities, and possible epidemiological consequences.
Tuberculosis is commonly associated with poverty and undernutrition in both developed and developing countries.1,2 In addition, obesity is an increasing problem that is associated with a wide range of chronic degenerative conditions,3- 5 notably, diabetes mellitus, a well-reported predisposing factor for active tuberculosis.6,7 Few studies have systematically examined the effect of obesity and overweight on tuberculosis, especially in Asian populations.
In Hong Kong, active tuberculosis is statutorily notifiable to the Department of Health, and the notification database has been computerized, using the identity card number as the unique identifier. The annual tuberculosis notification rate is approximately 90 per 100 000 population and is substantially higher in male individuals and those older than 65 years.8 The Elderly Health Service provides a community-based health maintenance program to elderly patients through 18 centers.9 For each enrolled client, a trained nurse administers a standardized questionnaire, collecting information on the Hong Kong identity card number, sociodemographic variables, and general health status. Body weight and height are measured. The questionnaire is followed by medical examination, chest x-ray examination, and regular screening for hypertension, blood glucose level, and serum total cholesterol level. These key variables are entered into a structured database and regularly checked for completeness. After enrollment, clients may attend the health centers for health education and curative care at any time. Patients with symptoms suggestive of active tuberculosis or radiological abnormalities are referred to 1 of the 18 chest clinics under a centralized Tuberculosis and Chest Service. The availability of appropriate service infrastructure provides a unique opportunity to prospectively examine the relationship between body mass index (BMI; calculated as weight in kilograms divided by the square of height in meters) and the development of active tuberculosis.
A cohort of clients who enrolled at the 18 health centers for elderly patients from January 1, 2000, to December 31, 2000, was retrospectively assembled as previously reported.9 The date of enrollment, name, sex, age, identity card number, smoking status, alcohol use, language spoken, educational level, marital status, housing situation, work status, public means–tested financial assistance status, coexisting medical conditions, recent weight loss, hospital admissions during the past 12 months, activities of daily living score, cholesterol level, and hemoglobin A1c level were retrieved from the baseline health assessment database of the Elderly Health Service. The baseline database was cross-matched prospectively with the death registry and the tuberculosis notification registry using the identity card number as the identifier, supplemented by name and age, from 3 months (arbitrarily considered 91 days) after enrollment to December 31, 2005. The BMI was classified into the following categories, which are in line with the World Health Organization expert consultation on the appropriate BMI for Asian populations: less than 18.5 (underweight), 18.5 to less than 23 (normal), 23 to less than 25 (at risk), 25 to less than 30 (overweight), and 30 or higher (obese).10 Hypertension and diabetes mellitus were defined as the corresponding physician-diagnosed conditions with or without treatment, and the updated diagnoses after the screening at enrollment were used in the analysis. The hemoglobin A1c level and cholesterol level at enrollment were also included. The duration of follow-up in days was defined as the period from the start of matching (91 days after enrollment) to the date of notification of tuberculosis, the date of death, or December 31, 2005, whichever was the earliest. Information on date of tuberculosis notification, bacteriological status, form of tuberculosis, and previous tuberculosis history was retrieved from the notification registry. The diagnosis and clinical information for all identified tuberculosis cases were verified by reviewing medical records retrieved from chest clinics and other relevant sources as well as the public health records of the Tuberculosis and Chest Service. An active case of tuberculosis was defined as disease proved by isolation of Mycobacterium tuberculosis or, in the absence of bacteriological confirmation, disease diagnosed on clinical, radiological, and/or histological grounds with an appropriate response to antituberculosis treatment.
Univariate analysis was first performed to analyze the relationship between BMI and other baseline variables. The χ2and Fisher exact tests were used as appropriate for categorical variables, and analysis of variance was used for numerical variables. The incidence of active and culture-confirmed tuberculosis was calculated by assuming a Poisson distribution in the rate of occurrence of the events. This was followed by Cox proportional hazards analysis of the effects of BMI on the development of active tuberculosis, culture-confirmed tuberculosis, pulmonary tuberculosis, and extrapulmonary tuberculosis (alone) after adjustment by sex and age. Adjustment was also made for the effects of potential confounders by entering them together with BMI in the overall Cox proportional hazards model. Multicolinearity was considered, with analysis repeated after exclusion of individuals with risk factors at baseline. A 2-tailed P<.05 was considered statistically significant. Data are presented as mean ± SD.
The attributable risk was derived by applying the adjusted hazard ratios (HRs) to a modified version of the Levin formula11:
Population-attributable risk of a factor = 1 − (rate in absence of factor/observed rate) = 1 − [1/Σ(HRi × Pi)],
where HRi and Pi are the adjusted HR and prevalence, respectively, of level i of the factor, and HR0 (adjusted HR of reference level) is 1.
The study was approved by the Ethics Committee of the Department of Health of Hong Kong.
A total of 42 659 clients 65 years or older were recruited into the health maintenance program of the Elderly Health Service in 2000. Twenty-three duplicate entries; 304 clients with missing or invalid identification numbers; 127 with missing or incomplete information on sex, age, weight, and/or height; and 89 with known active tuberculosis on presentation or active tuberculosis notified within 3 months of enrollment were excluded, leaving 42 116 individuals for analysis.
The background characteristics of the cohort are given in Table 1, with stratification by the 5 BMI categories commonly used for Asian populations.12 The data were more than 99.9% complete for the variables listed. Almost all (>99.8%) were ethnically Chinese, speaking Cantonese, Mandarin, or other Chinese dialects.
A total of 3893 deaths (9.2%) were identified from the death registries. The mean ± SD duration of follow-up from the day of enrollment to death, development of active tuberculosis, or December 31, 2005 (whichever was earlier), was 1827 ± 325 days. A total of 510 new tuberculosis notifications were identified from the tuberculosis notification registry during the period of follow-up. Eight cases were duplicate entries because of early relapses. Twenty-five cases (6 cases of carcinoma of the lung, 13 cases of nontuberculous mycobacterial infection, and 6 cases of old lung scars) were found to have an incorrect diagnosis after reviewing the records, leaving 477 cases of active tuberculosis for analysis, 326 (68.3%) of which were culture confirmed. The mean ± SD interval from enrollment to notification of active tuberculosis was 881 ± 583 days. The incidence of active and culture-confirmed tuberculosis by various BMI categories is summarized in Table 2. There were 395 new cases (82.8%) and 82 retreatment cases (17.2%). Pulmonary involvement was found in 426 cases (89.3%) and extrapulmonary involvement in 87 (18.2%), including 36 cases (7.5%) with both. Of 394 patients who underwent voluntary human immunodeficiency virus (HIV) testing, only 1 (0.25%) was HIV infected.
Those individuals who developed active tuberculosis had a significantly greater mean ± SD baseline body height (1.56 ± 0.08 vs 1.54 ± 0.08; P<.001) but lower body weight (54.9 ± 10.3 vs 57.6 ± 10.2; P<.001) and BMI (22.5 ± 3.7 vs 24.3 ± 3.6; P<.001) than those who did not. Similar results were obtained for culture-confirmed tuberculosis. However, when the analysis was confined to either sex, only the baseline body weight and BMI (P<.001) but not body height (P = .11 for men; P = .40 for women) remained statistically significantly different between the 2 groups. No significant differences were found in the mean ± SD baseline BMI of new and retreatment cases (22.5 ± 3.71 vs 22.0 ± 3.5; P = .25). A higher mean baseline BMI was found in pulmonary- than in extrapulmonary-only cases (22.3 ± 3.6 vs 24.1 ± 3.9; P = .001), but no statistically significant differences were found in their mean age, duration of follow-up, weight, and height (all P>.05).
Table 3 gives the incidence of active tuberculosis and culture-confirmed tuberculosis by sex and age. Table 4 gives the sex- and age-adjusted HRs of different types of tuberculosis for different categories of BMI before and after exclusion of individuals with diabetes mellitus and other known risk factors at baseline. No significant interaction was found between BMI and sex or age groups for active, culture-confirmed, pulmonary, and extrapulmonary tuberculosis (all P>.05). The interaction terms of sex × BMI, age group × BMI, and sex × age group × BMI were therefore excluded from the Cox proportional hazards models. Table 5 gives the adjusted HRs for different BMI categories after adjustment for all sociodemographic and clinical variables in a combined Cox proportional hazards analysis. The Figure depicts the hazard function curves of active tuberculosis for different BMI categories in the overall Cox model. A strong linear dose-response effect was observed per unit change of BMI (as a continuous variable) for active tuberculosis (adjusted HR, 0.90; 95% confidence interval [CI], 0.87-0.93; P<.001), culture-confirmed tuberculosis (adjusted HR, 0.89; 95% CI, 0.85-0.93; P<.001), and pulmonary tuberculosis (adjusted HR, 0.88; 95% CI, 0.85-0.92; P<.001) but not for extrapulmonary tuberculosis alone (adjusted HR, 1.01; 95% CI, 0.92-1.10; P = .91) after exclusion of individuals with BMIs less than 18.5 and controlling for the same confounders. The adjusted HRs (95% CIs) per unit change of BMI became 0.87 (0.82-0.93), 0.87 (0.81-0.94), 0.84 (0.78-0.90), and 1.01 (0.90-1.14), respectively, after exclusion of all individuals with diabetes mellitus and the risk factors listed in Table 4.
On applying the adjusted HRs to a modified version of the Levin formula, BMI outside the range of 18.5 to 23 decreased the active tuberculosis risk by 23.5% (95% CI, 3.7%-43.6%) of the observed level. Baseline BMI obesity at 25 or above was associated with a 30.1% (95% CI, 12.7%-47.7%) decrease in risk, whereas BMI lower than 18.5 increased the risk by 6.6% (95% CI, 4.1%-9.0%).
In this study, obesity and overweight were associated with a significantly lower risk of both clinically active and culture-confirmed tuberculosis, even after exclusion and adjustment for a series of potential confounding variables at baseline (Table 4 and Table 5). A strong linear dose-response relationship was also observed with a 10% reduction in risk of active tuberculosis for each unit change of BMI above 18.5.
Being underweight has been associated with a higher risk of tuberculosis in several studies,12- 14 but few of them were able to control for the myriad possible confounders as in the current study. Malnutrition has traditionally been regarded as a causal link in the association between body composition and tuberculosis, but it does not readily explain why obesity is associated with a lower risk, despite all the associated comorbidities (Table 1).
The risk-lowering effect of high BMI was relatively specific for pulmonary tuberculosis but not for extrapulmonary tuberculosis (Tables 4 and 5). Although a type II error was possible with the limited number of extrapulmonary tuberculosis cases, the magnitude of the HRs and the absence of a consistent trend across different BMI categories did not suggest any major effect. Such findings also concurred with a previous report on chest x-ray screening in Switzerland.14 Extrapulmonary involvement is a hallmark of most immunocompromising conditions. The use of the tumor necrosis factor inhibitors infliximab and etanercept has been associated with a high tuberculosis risk,15,16 but a high proportion of extrapulmonary involvement was observed.15- 17 Among HIV-infected individuals, extrapulmonary tuberculosis involvement is a feature of advanced immunodeficiency.18 The differential effect of BMI on pulmonary tuberculosis is therefore unique and could point to a novel mechanism of interaction between the pathogen and host defense.
This is a study on a cohort of elderly individuals in community settings. Although our cohort has a lower percentage of male subjects compared with the 2001 census (34.8% vs 46.2%)19 and does not include institutionalized clients, the sex- and age-stratified tuberculosis incidence (Table 3) was comparable to the reported population age-specific notification rates in 2003.8 Although the observation in this study does not necessarily apply to younger adults, a previous study12 of young military recruits reported a 140% increase in risk among those underweight by 15% or more and a 36% reduction in risk among those overweight by 5% or more. The effect of body build therefore appears to be consistent among both young and old adults. Although only modest numbers of individuals had baseline BMI at the uppermost and lowest BMI categories, a consistent risk reduction with increasing BMI was demonstrated across all BMI categories. The difference in baseline BMI (22.5 vs 24.3) between those who did and did not develop active tuberculosis might have been only modest, but this would be expected from the bell-shaped distribution of BMI in the parent cohort.
Undernotification is always a concern for notifiable diseases.20 The cohort was already under the care of a service in the Department of Health and had ready access to the chest clinics in the same department for free tuberculosis treatment. In a local audit of tuberculosis notifications, the undernotification rate was only 3% in the chest clinics even before the introduction of specific improvement measures.21 Because we examined the differences between subgroups in this cohort, this study should have good internal validity.
A more fundamental question is whether the observed association is the result of tuberculosis on body build rather than the reverse. With the prospective design, systematic clinical and x-ray examination at enrollment, exclusion of all prevalent cases up to 3 months after recruitment, and 5 years of follow-up, such direction of causation is unlikely. Subclinical disease could also be the culprit, but the relationship persisted even after exclusion of all individuals with significant weight loss at baseline (Table 4), and the risk differentials among different BMI categories persisted throughout the entire period of follow-up (Figure).
Systematic screening for diabetes mellitus, hypertension, and hypercholesterolemia was also performed at enrollment. Despite the absence of formal cancer-screening programs, patients with unexplained baseline abnormalities on clinical examination or chest x-ray examination were referred for further workup. Only 1 (0.25%) of the 394 tuberculosis patients who underwent voluntary HIV testing had a positive HIV test result, and such a result concurred with the low HIV prevalence (well below 1%) found by unlinked anonymous HIV assays.8 As contrasted with communities with a high HIV prevalence, HIV is unlikely to confound the relationship between BMI and tuberculosis in Hong Kong. No information was available about the tuberculin status or the newer interferon assays because such screening was not regularly performed. However, with the significant past burden of tuberculosis,8 a high proportion of latent tuberculosis infection is expected among this elderly cohort. A previous study22 has reported a tuberculin reactivity rate (≥10 mm) of 68.6%, in the absence of recent contact history, among inmates of old-age homes after 2-stage tuberculin skin testing. A history of tuberculosis was not specifically included for all enrolled individuals, but only 17.2% of observed patients with tuberculosis had a history of the disease, and no significant difference in the mean baseline BMI was found between new and retreatment cases.
Only a small percentage of infected individuals develop active tuberculosis after infection.23,24 A delicate balance presumably exists between the pathogen and the human host. Adipose tissue is now recognized to have major endocrine functions.25 Increasing evidence suggests that leptin differentially regulates metabolic, neuroendocrine, and immune functions in humans.26,27 The administration of leptin in patients with human congenital leptin deficiency has demonstrated beneficial effects on obesity, T-cell hyporesponsiveness, and neuroendocrine and metabolic dysfunction.28 Leptin-deficient ob/ob mice were shown to be highly susceptible to pulmonary infection by Mycobacterium tuberculosis.29 Besides the possible immune-modulating effects of obesity, a family of fatty acid transporters are highly conserved from mycobacterium to humans,30 and a cholesterol-rich diet has also been shown to accelerate bacteriologic sterilization in pulmonary tuberculosis.31 A better understanding of the complex interplay between the tubercle bacillus and our bodies is likely an important step for the discovery of new intervention targets in our battle against this long-standing enemy of humanity.
Low body weight has been associated with risk of tuberculosis disease,12- 14 severity of disease,32,33 unfavorable response to treatment,33,34 and relapse.35 Body weight has been mentioned briefly as a modifier of disease risk in the recent Centers for Disease Control and Prevention guidelines,36 but little attention has been paid to this readily measurable prognostic index in clinical practice.37 Endogenous reactivation is becoming a major source of tuberculosis disease among the older generations. A previous molecular epidemiological study in Hong Kong attributed less than 24% of tuberculosis cases to recent transmission.38 Although treatment of latent infection is a possible strategy against endogenous reactivation, the decision to screen is not always easy to make in a high-prevalence setting, especially in view of the limitations in the currently available diagnostic and treatment tools. The cost-effectiveness of such screening is likely to be improved if BMI can be taken into consideration. Aggressive measures may also be indicated against elderly malnutrition to reduce the risk of reactivation of tuberculosis, and nutritional supplementation might be an important adjunct to chemotherapy in this regard. Although the gain in tuberculosis prevention with weight gain must be balanced carefully against the increased risk of diabetes mellitus and cardiovascular disease, it might still be appropriate to guard against an increased risk of tuberculosis with aggressive weight reduction. Further studies are warranted to inform clinical decision making.
Although improving socioeconomic conditions have been associated with the decline of tuberculosis well before the advent of effective treatment in the last century, little consensus exists on the underlying mechanism.39 Increased availability of cheaper food has been reported with improving socioeconomic conditions.39 The observed effect of BMI in this study may therefore provide a possible link that could help to address the ongoing controversy. Although malnutrition is a critical problem in many resource-limited countries, obesity is becoming a major epidemic in both developed and developing areas.3,40- 42 Even in this cohort in which neither of the extremes is prevalent, BMI contributes a net 23% risk reduction, predominantly to the more infectious pulmonary form. It may therefore be worthwhile to explore the potential impact of the ongoing changes in our body build on the future tuberculosis trend.
With the potential biological, clinical, and epidemiological implications of our observations, further cohort studies are warranted to confirm these findings in other population groups and communities. More basic research and animal studies are also indicated to explore the underlying mechanisms.
Correspondence: Chi C. Leung, MBBS, Wanchai Chest Clinic, 99 Kennedy Rd, Wanchai, Hong Kong, China (email@example.com).
Accepted for Publication: February 22, 2007.
Author Contributions: Dr C. C. Leung had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: C. C. Leung, Lam, W. M. Chan, Ho, Tam, and C. K. Chan. Acquisition of data: C. C. Leung, W. M. Chan, Ho, Tam, and C. K. Chan. Analysis and interpretation of data: C. C. Leung, Lam, Yew, G. Leung, Law, C. K. Chan, and Chang. Drafting of the manuscript: C. C. Leung. Critical revision of the manuscript for important intellectual content: C. C. Leung, Lam, W. M. Chan, Yew, Ho, G. Leung, Law, Tam, C. K. Chan, and Chang. Statistical analysis: C. C. Leung, Lam, G. Leung, Law, and Chang. Administrative, technical, and material support: W. M. Chan, Ho, Tam, and C. K. Chan. Study supervision: C. C. Leung, Yew, and Tam.
Financial Disclosure: None reported.