Skip to main content
. 2020 Aug 28;159(1):112–115. doi: 10.1016/j.chest.2020.08.2065

Table 2.

Predictors of Mortality, Low O2 Saturation, and Sepsis Among MSHS COVID-19 Patients

Relative Odds of Deatha
Relative Odds of Hypoxemia at Admissionb
Relative Odds of Sepsis at Admissionc
Variable ORadj (95% CI) Variable ORadj (95% CI) Variable ORadj (95% CI)
Sex: Male vs female 1.37 (1.15-1.64) Sex: Male vs female 1.37 (1.13-1.66) Sex∗ Agec Females Males
Age, y Age, y Age, y
 <40 1 (ref)  <40 1 (ref)  <40 1 (ref) 3.50 (1.79-6.82)
 40-69 8.45 (3.43-20.81)  40-69 2.14 (1.52-3.01)  40-69 1.26 (0.78-2.05) 1.84 (1.15-2.80)
 ≥70 24.21 (9.83-59.63)  ≥70 2.90 (2.03-4.15)  ≥70 0.90 (0.56-1.46) 1.05 (0.65-1.70)
History of CKD, y/n 1.38 (1.10-1.73) Race Race
BMI ≥ 35 kg/m2, y/n 1.53 (1.21-1.94)  White 1 (ref)  White 1 (ref)
 Black 0.59 (0.45-0.77)  Black 1.05 (0.85-1.30)
 Other/unknown 0.86 (0.67-1.10)  Other/unknown 1.41 (1.17-1.70)
History of COPD, y/n 1.62 (0.98-2.70) History of cancer, y/n 1.53 (1.17-2.00)
History of cancer, y/n 1.44 (1.00-2.06) History of hypertension, y/n 0.88 (0.75-1.03)
BMI ≥ 35 1.72 (1.31-2.26)

CKD = chronic kidney disease.

a

Stepwise logistic regression model performed. Race, smoking status, hypertension, history of asthma, COPD, diabetes, HIV, cancer were originally included in the model and were removed; the final model was adjusted for sex, age, history of CKD, and BMI.

b

Stepwise logistic regression model performed. Smoking status, hypertension, asthma, CKD, and HIV were originally included in the model and were removed. The final model was sex, age, race, history of COPD, history of cancer, and BMI. Hypoxemia was defined as patients with ≤92% peripheral O2 saturation.

c

Stepwise logistic regression model performed. Smoking status, asthma, COPD, CKD, HIV, and obesity were originally included in the model and were removed. The final model was adjusted for sex∗age, age, race, history of cancer, and history of hypertension. The age∗sex interaction term (P < .05) and the sex and age group interaction terms (40-69 vs <40 years; P = .0182; ≥70 vs <40 years, P = .0018) were statistically significant.