DPP4 inhibitors and COVID19 outcomes in patients with type II diabetes: a multicenter retrospective cohort study in Saudi Arabia

Table 3 Clinical outcomes results

Outcome	Unadjusted analysis			Applying overlap weighting*
Outcome	DPP-4i (n = 166)	Non- DPP-4i (n = 351)	Unadjusted HR †(95% CI)	DPP-4i (ESS = 148)	Non- DPP-4i (ESS = 259)	Adjusted HR† (95% CI)	Benjamini-Hochberg (BH) procedure ‡
Mortality during hospitalization, %	9.6	18.8	0.55 (0.31 to 0.97, p = 0.040)	11.4	15.6	0.73 (0.40 to 1.34, p = 0.319)	0.563
ICU admission, %	26.5	37.0	0.66 (0.47 to 0.93, p = 0.021	30.2	33.5	0.83 (0.57 to 1.21, p = 0.338)	0.563
Progression to mechanical ventilation, %	7.2	20.8	0.30 (0.16 to 0.56, p < 0.001)	7.2	17.8	0.40 (0.21 to 0.77, p = 0.006)	0.030
Length of stay in days, median (IQR)	9.0 (5–18)	8.0 (4–19)	0.99 (0.72 to 1.37, p = 0.973	5 (10–21)	5 (9 − 11)	1.14 (0.80 to 1.62, p = 0.461)	0.576
ICU length of stay in days, median (IQR)	12.5 (7.5–22.5)	12.0 (5–22)	1.18 (0.66 to 2.11, p = 0.574)	7 (12–21)	5 (12–22)	1.19 (0.64 to 2.20, p = 0.576)	0.576

*Covariate balance was achieved via overlap weights methodology according to Li (2017) by weighting each unit proportional to its probability of assignment to the opposite group. The estimand is referred to the average treatment effect in the overlap (ATO). After overlap weighting, cox proportional hazard model was fitted and robust estimation for the standard errors was used to account for the survey design. For the ICU admission endpoint, the weights were created without the outcome, use of IV tocilizumab, and resulted in adequate balance effective sample size was 259 for non-dPP4i and 149 DDP4i
† Cox proportional hazard models were fitted for the outcomes of interest (mortality, ICU admission and progression to mechanical ventilation). The length of stay and ICU length of stay was analyzed using proportional odds model and the results are displayed as OR for these two outcomes. The weighted analysis was the same types of models but with supplying weights
‡ Benjamini-Hochberg (BH) procedure to adjust p-values for multiple hypothesis testing. The BH method ranks the observed p-values in ascending order and adjusts them using an increasing threshold to control the expected proportion of false positives
Abbreviations: DPP-4i: Dipeptidyl peptidase-4 inhibitors; ESS: Effective sample size; HR: Hazard ratio; CI: Confidence interval; IQR = interquartile range
Note: in all analyses, the non-DPP4-I was the reference level

ISSN: 3059-4057