Excess mortality among non-COVID-19 surgical patients attributable to the exposure of French intensive and intermediate care units to the pandemic

Abstract

Purpose: The mobilization of most available hospital resources to manage coronavirus disease 2019 (COVID-19) may have affected the safety of care for non-COVID-19 surgical patients due to restricted access to intensive or intermediate care units (ICU/IMCUs). We estimated excess surgical mortality potentially attributable to ICU/IMCUs overwhelmed by COVID-19, and any hospital learning effects between two successive pandemic waves.

Methods: This nationwide observational study included all patients without COVID-19 who underwent surgery in France from 01/01/2019 to 31/12/2020. We determined pandemic exposure of each operated patient based on the daily proportion of COVID-19 patients among all patients treated within the ICU/IMCU beds of the same hospital during his/her stay. Multilevel models, with an embedded triple-difference analysis, estimated standardized in-hospital mortality and compared mortality between years, pandemic exposure groups, and semesters, distinguishing deaths inside or outside the ICU/IMCUs.

Results: Of 1,870,515 non-COVID-19 patients admitted for surgery in 655 hospitals, 2% died. Compared to 2019, standardized mortality increased by 1% (95% CI 0.6-1.4%) and 0.4% (0-1%) during the first and second semesters of 2020, among patients operated in hospitals highly exposed to pandemic. Compared to the low-or-no exposure group, this corresponded to a higher risk of death during the first semester (adjusted ratio of odds-ratios 1.56, 95% CI 1.34-1.81) both inside (1.27, 1.02-1.58) and outside the ICU/IMCU (1.98, 1.57-2.5), with a significant learning effect during the second semester compared to the first (0.76, 0.58-0.99).

Conclusion: Significant excess mortality essentially occurred outside of the ICU/IMCU, suggesting that access of surgical patients to critical care was limited.

Keywords: COVID-19; Critical care; Patient safety; Surgical mortality.

Fig. 1

Crude ratios of surgical mortality rates between 2020 and 2019. a In-hospital mortality ratios over time. b In-hospital mortality ratios during national lockdowns by French region. ICU/IMCUs were defined as intensive or intermediate care units. Values in (a) are 7-day moving averages of 2020 to 2019 in-hospital mortality rate ratios from January to December. A mortality ratio of 2.0 means that the mortality rate in 2020 was twice the mortality rate in 2019. Bands mark the two periods of national lockdowns in France. The first lockdown took place from Mar 17 to May 10, 2020, and the second from Oct 30 to Dec 15, 2020. The timeframe of the maps in (b) consists of a combination of the two national lockdowns. Death rate ratios were defined as the ratio of the 2020 observed death rates divided by the 2019 observed death rates. The inside ICU/IMCU mortality ratio was greatest in the Greater Paris (1.17), followed very closely by the North–East (1.15), and the three other regions (1.14 for each). Outside ICU/IMCU, the greatest values were observed in the Greater Paris (1.80), followed by the North–East (1.51), the South–East (1.39), the North–West (1.26), and finally the South–West (1.22) of France. Circles on the map correspond to all hospitals included in this study. For each one of them, the COVID-19 exposure was calculated as the median of the daily proportion of COVID-19 patients in ICU/IMCUs during the lockdown periods

Fig. 2

Difference in standardized surgical mortality rates between years by pandemic exposure group and semester. *P < 0.05, **P < 0.001. ICU/IMCUs were defined as intensive or intermediate care units. The bar charts represent the standardized outcome rates per year, COVID-19 exposure group and semester (first Jan–June and second July–Dec). These standardized rates were calculated using estimated regression coefficients obtained from the triple-difference GEE models and a marginal standardization method to control case-mix differences between exposure groups. Differences above brackets indicate absolute differences in standardized rates between 2020 (exposure period) and 2019 (pre-exposure period) per semester in each pandemic exposure group (low-or-no [0–30%], moderate [30–60%], and high [60–100%])

Fig. 3

Comparison of surgical mortality risk between pandemic exposure groups by semester with related learning effect. ICU/IMCUs were defined as intensive or intermediate care units. Adjusted ratios of odds-ratios (RORs) and adjusted ratios of ratio of odds ratio (RRORs) were obtained from the triple-difference GEE models. The RORs captured the changes on primary and secondary outcomes from the pre-exposure period (2019) to the exposure period (2020) between high and low-or-no, high and moderate, and finally moderate and low-or-no exposure groups for each semester (first Jan–June and second July–Dec). A ROR greater than one indicated an increased in-hospital mortality risk for patients in the given exposure group in comparison to the exposure reference. The between semester RRORs captured the learning effect between the second and the first semesters for each of the previous comparisons. A RROR less than one indicated a decreased risk of mortality for the second semester of the year in comparison to the first. C-statistics obtained from models were 0.91 for mortality, 0.95 for mortality inside ICU/IMCU and 0.89 for mortality outside ICU/IMCU

Fig. 4

Deaths in excess among surgical patients in moderate to high pandemic exposure groups. ICU/IMCUs were defined as intensive or intermediate care units. Excess in-hospital deaths for moderate- and high-exposure groups were calculated in comparison to the low-or-no exposure group. They were calculated separately and subsequently summed up. Using the RORs (ratios of odds-ratios) obtained from the triple-difference GEE models, we calculated the etiologic fraction, i.e., the share of deaths that would have been avoided if the exposure was low-or-no, and multiplied this value by the observed number of deaths in 2020. The corresponding 95% CIs were computed from non-parametric bootstrap based on 1000 replications