Impact of smoking status on disease severity and mortality of hospitalized patients with COVID-19 infection: a systematic review and meta-analysis

Abstract

The recent COVID-19 pandemic has raised concern regarding risk factors associated with disease severity and case fatality. In this context, the association of smoking with severity and mortality of COVID-19 infection remains controversial.1 We conducted this systematic review and meta-analysis to summarize relevant studies and quantify the impact of smoking on disease severity and mortality of hospitalized COVID-19 patients. The primary outcome was a composite for disease severity including severe disease (including critical cases), ICU admission or invasive ventilation, and adverse disease progression or refractory disease. The secondary outcome was death. Inclusion criteria were as follows: (1) full-length original research publications in peer-reviewed journals in English; (2) inclusion of hospitalized adult patients with COVID-19 infection; (3) reported incidence of smoking stratified by a component of the primary outcome or mortality status; and (4) sample size > 20 patients. Details of literature search, data extraction, and statistical analysis are available in online Supplementary Material. Briefly, a systematic review of literature from September 1, 2019 to May 4, 2020 was performed, and additional data sources were identified through manually searching references. Two independent reviewers (A.K. and K.A.) recorded the incidence of smoking per group and performed an assessment of study quality using seven indicators (see online Supplementary Material for details). The initial search identified 652 results. After screening titles and/or abstracts, 166 articles remained for full-text assessment from which 146 were subsequently excluded. Two additional studies were identified by searching references. Overall, 22 studies met our inclusion criteria, 17 reporting on severity,2–18 4 on mortality,19–22 and 1 on both23 (Supplementary Table). Study quality was generally low (Figure 1; online Supplementary Material). For the primary outcome of disease severity, 18 studies were included with 6310 patients.2,4–19,21 Smoking modestly increased the risk for the combined end point of disease severity (odds ratio [OR] = 1.34, 95% confidence intervals [CI] = 1.07–1.67, I2 = 45%) (Figure 1A). Sensitivity analyses were consistent (online Supplementary Material). After restricting the analysis in studies explicitly reporting current smoking, the association of smoking with disease severity was not statistically significant (10 studies with 4152 patients; OR = 1.12, 95% CI = 0.84–1.50, I2 = 38%). Forest plot and risk of bias tables examining in hospitalized COVID-19 patients the association of smoking with (A) the composite end point of disease severity in China and in US regions and (B) mortality. Boxes represent odds ratio (OR) and lines represent the 95% confidence interval [CI] for individual studies. Diamonds and their width represent pooled ORs and the 95% CI, respectively. There was a difference (p = .03) between Chinese studies (16 studies—4423 patients; OR = 1.48, 95% CI = 1.17–1.87; I2 = 40%)2,4,6–12,14–19,21 and US studies (2 studies—1887 patients; OR = 0.65, 95% CI = 0.33–1.29; I2 = 0%).5,13 Visual inspection of the funnel plot did not reveal publication bias. Meta-regression analyses are reported in online Supplementary Material. Age (Z = −0.91, p = .05) and diabetes (Z = −2.81, p = .005) had significant negative associations with log-OR of smoking for disease severity (Supplementary Figure). In studies with low (<15%) prevalence of diabetes, smoking increased the risk for severe disease (OR = 1.66, 95% CI = 1.26–2.18, I2 = 34%), whereas in studies with high (≥15%) prevalence of diabetes, there was a trend for a negative association (OR = 0.70, 95% CI = 0.46–1.08, I2 = 0%) (χ2 for subgroup differences = 10.82, p = 0.001) (Supplementary Figure). For mortality, five studies with 838 patients were included.3,9,20,22,23 Smoking was not significantly associated with increased mortality (OR = 1.45, 95% CI = 0.78–2.72, I2 = 18%) (Figure 1B). Results were similar for studies explicitly reporting current smoking (2 studies—465 patients; OR = 1.57, 95% CI = 0.75–2.31, I2 = 0%). Our main findings were (1) that smoking modestly increases the risk of severe disease in hospitalized COVID-19 patients, whereas mortality data suggest a similar effect size but are currently inconclusive due to a low sample size and (2) that this increased risk for disease severity is more prominent in younger patients without diabetes. Although smoking may enhance lung inflammation, increase epithelial cell permeability, and cause mucus overproduction and impaired mucociliary clearance,24 its specific impact on COVID-19 disease is controversial. Hypotheses support both a potentially hazardous impact via overexpression of the ACE2 receptor gene24 and a potentially protective effect via attenuation of the commonly observed in critically ill patients with COVID-19 infection excessive immune response.1 Our findings demonstrate an adverse impact