Research Letter
May 2, 2023
Body Mass Index and Risk of Hospitalization or Death Due to Lower or Upper Respiratory Tract Infection
Bastian Bohrmann, M. Sofia Massa, Stephanie Ross, et al
JAMA. 2023;329(17):1512-1514. doi:10.1001/jama.2023.2619
High body mass index (BMI) is a risk factor for several noncommunicable diseases.1 In addition, high BMI has been associated with severe COVID-19; however, associations with other severe infectious respiratory diseases remain unclear.2,3 We assessed the associations between BMI and risk of hospitalization for or death due to COVID-19, lower respiratory tract infections (LRTIs), and upper respiratory tract infections (URTIs).
Methods
The UK Biobank is a prospective cohort of more than 500 000 individuals recruited from the general population with baseline assessment performed in 2006-2010.4,5 Follow-up from baseline was completed on February 2, 2021, via linkage with national electronic health records. The UK Biobank received ethical approval from the multicenter North West Research Ethics Committee; participants provided written informed consent.
The analyses excluded participants with prevalent chronic respiratory diseases, previous hospitalization for infectious respiratory diseases, and missing or extreme values for key variables. Cox proportional hazards models yielded hazard ratios (HRs) for the association between BMI (calculated as weight in kilograms divided by height in meters squared) categories (14-24.9, 25-29.9 [reference], 30-34.9, and 35-60; there were too few events at <25 to subdivide further) and risk of hospitalization for primary diagnosis of or death due to COVID-19, LRTIs (eg, influenza, pneumonia), or URTIs (eg, laryngitis, tonsillitis); the diagnostic codes appear in the eTable in Supplement 1. In the continuous analyses, the HRs are per 10-unit higher BMI.
The HRs were adjusted for age at risk, sex, region, race, Townsend deprivation score, education, smoking, and alcohol intake; the 95% CIs were calculated using the variance of the log risk. The Cochran-Armitage test for trend across BMI categories was used to calculate the Pvalues. The sensitivity analyses assessed for residual confounding and reverse causality by excluding people who ever smoked, those with major chronic disease at baseline, and the first 5 years of follow-up.
Analyses were performed using Stata/SE version 17 (StataCorp), and a 2-sided P < .05 was considered statistically significant.
Results
After exclusions, 476 176 participants remained (median age, 58 years [IQR, 50-63 years]; 54% were female; 94% were of White race). The mean BMI was 27.4 (SD, 4.7). At baseline, BMI was significantly associated with potential confounders, adiposity-related biomarkers, and chronic disease (Table 1). The Pearson correlation of BMI at baseline and at resurvey (approximately 4 years later) was r = 0.92.
Over a mean follow-up of 11.8 years, 20 302 participants were hospitalized for or died of severe infectious respiratory disease: 1296 (144 died without prior hospitalization) for COVID-19, 17 085 (273 died without prior hospitalization) for LRTIs, and 1504 (0 died without prior hospitalization) for URTIs. For COVID-19, the crude incidence rates per 1000 person-years ranged from 1.2 for BMI less than 25 to 5.8 for BMI in the range of 35 to 60; for LRTIs, from 25.6 to 55.5; and for URTIs, from 2.2 to 4.4. For COVID-19, the adjusted HRs ranged from 0.62 (95% CI, 0.54-0.71) for BMI less than 25 to 2.38 (95% CI, 2.07-2.74) for BMI in the range of 35 to 60; for LRTIs, from 1.02 (95% CI, 0.99-1.05) to 1.91 (95% CI, 1.83-2.00); and for URTIs, from 0.79 (95% CI, 0.72-0.88) to 1.40 (95% CI, 1.20-1.64).
In the continuous analyses, there were positive log-linear associations of BMI with risk of severe COVID-19 and URTIs throughout the BMI categories examined, and with BMI greater than 25 for LRTIs; the HR per 10-unit higher BMI was 2.26 (95% CI, 2.05-2.49) for COVID-19, 1.74 (95% CI, 1.68-1.80) for LRTIs, and 1.37 (95% CI, 1.24-1.51) for URTIs (P < .001 for trend for each; Table 2).
The sensitivity analyses did not materially alter the associations in the continuous analyses but slightly lowered the HR in the lowest BMI group for LRTIs in the categorical analyses (Table 2).
Discussion
This study found log-linear associations between higher BMI and increased risk of hospitalizations for or death due to severe COVID-19 and URTIs throughout the BMI categories examined, and with LRTIs for BMI greater than 25.
This study supports previous findings that 10-unit higher BMI is associated with about double the risk of severe COVID-19.6 In contrast, the limited previous evidence on the association of BMI with risk of severe LRTIs or URTIs shows substantial variation in the strength of these associations.3 Several studies have described lower risk for LRTIs with higher BMI, but the current study found no such evidence in the BMI categories examined. The findings suggest that the proposed mechanisms (including obesity-related chronic disease or compromised respiratory function) between BMI and COVID-19 might be shared by other LRTIs.
Study limitations include that associations could not be explored by subtypes of LRTIs. The UK Biobank is not representative of the general population. Underweight and healthy-weight groups could not be separated. The findings support individual and population-level approaches to reduce obesity and targeted vaccination programs for respiratory infections for individuals with a high BMI.
