key: cord-0821100-3hkmakee
authors: Watanabe, S.; Shin, J.-h.; Okuno, T.; Morishita, T.; Takada, D.; Kunisawa, S.; Imanaka, Y.
title: Medium-term impacts of the waves of the COVID-19 epidemic on treatments for non-COVID-19 patients in intensive care units: a retrospective cohort study in Japan
date: 2022-02-28
journal: nan
DOI: 10.1101/2022.02.28.22271604
sha: 6de9a1f7cee4ccebffd551081553f526e1481b59
doc_id: 821100
cord_uid: 3hkmakee

Background: Maintaining critical care for non-Coronavirus-disease-2019 (non-COVID-19) patients is a key pillar of tackling the impact of the COVID-19 pandemic. This study aimed to reveal the medium-term impacts of the COVID-19 epidemic on case volumes and quality of intensive care for critically ill non-COVID-19 patients. Methods: Administrative data were used to investigate the trends in case volumes of admissions to intensive care units (ICUs) compared with the previous years. Standardized mortality ratios (SMRs) of non-COVID-19 ICU patients were calculated in each wave of the COVID-19 epidemic in Japan. Results: The ratios of new ICU admissions of non-COVID-19 patients to those in the corresponding months before the epidemic: 21% in May 2020, 8% in August 2020, 9% in February 2021, and 14% in May 2021, approximately concurrent with the peaks in COVID-19 infections. The decrease was greatest for new ICU admissions of non-COVID patients receiving mechanical ventilation (MV) on the first day of ICU admission: 26%, 15%, 19%, and 19% in the first, second, third, and fourth waves, respectively. No statistically significant change in SMR was observed in any wave of the epidemic; SMRs were 0.990 (95% confidence interval (CI), 0.962-1.019), 0.979 (95% CI, 0.953-1.006), 0.996 (95% CI, 0.980-1.013), and 0.989 (95% CI, 0.964-1.014), in the first, second, third, and fourth waves of the epidemic, respectively. Conclusions: Compared to the previous years, the number of non-COVID-19 ICU patients continuously decreased over the medium term during the COVID-19 epidemic. The decrease in case volumes was larger in non-COVID-19 ICU patients initially receiving MV than those undergoing other initial treatments. The standardized in-hospital mortality of non-COVID-19 ICU patients did not change in any waves of the epidemic.

data from April 2018 to September 2021 were included in our study. Among the patients 1 admitted to these hospitals, ICU admissions from April 2018 to July 2021, aged 18 2 years or older, were included. 3

In this study, an ICU was defined as per international standards as the wards 4 which can at least provide oxygen, noninvasive monitoring, and more intensive nursing 5 care than usual beds [18] . Within the general bed reimbursement categories, which are 6 not beds for specified diagnoses (such as stroke) in Japan, three categories meet the ICU 7 criteria as below (the summary description in each category refers to the minimum 8 requirements for reimbursement). 9 -Specialized-care ICU (sICU): require the most intensive resourcing, including a 1:2 10 patient-nurse ratio and constant placement of a doctor. 11 -Emergency-care ICU (eICU): require the facility to deal with emergency patients, a 12 1:4 patient-nurse ratio, and constant placement of a doctor. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. admitted to ICUs. The case volume in each wave of COVID-19 infections was analyzed. 1

To take seasonality into account, a trend of the ratio of the case volume in each month to 2 the case volume in the corresponding month before the pandemic was analyzed. More 3 specifically, monthly case volumes until March 2021 were compared with those in 4 corresponding months one year before, and monthly case volumes from April 2021 to 5

July 2021 were compared with those in corresponding months two years before. 6

Additionally, the ratios of case volumes of non-COVID-19 patients initially admitted to 7 an sICU (admitted to an sICU on the first day of ICU admissions), of non-COVID-19 8 ICU patients initially undergoing mechanical ventilation (MV), of non-COVID-19 ICU 9 patients initially administered vasopressors, and of all ICU patients, the ratios of the 10 total case numbers (patients times days) of all ICU patients, and the absolute numbers of 11 COVID-19 ICU patients were described. The classification of COVID-19 patients was 12 based on the diagnoses recorded in the DPC/PDPS data. For diagnoses, the International 13 Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-14 10) was applied, and patients with diagnoses of B34.2 and U07.1, except for suspected 15 diagnoses, were classified as In addition, the association between hospitals' acceptance of patients and impacts on non-COVID-19 ICU patient volumes was investigated. 18 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 Specifically, the included hospitals were classified into three categories described below, 1 which were used to stratify the ratios of case volumes of non-COVID-19 patients to 2 before the epidemic. 3 -Hospitals that were continuously accepting COVID-19 ICU patients: more than ten 4 patient days of COVID-19 patients in every wave of COVID-19. 5 -Hospitals that were accepting few COVID-19 ICU patients: less than ten patient 6 days of COVID-19 patients in total in the study period. 7 -Hospitals that were intermediately accepting COVID-19 ICU patients: hospitals 8 meeting neither of the categories above. 9

Moreover, another type of hospital classification was employed as a sensitivity 10 analysis as below (hereinafter, referred as the month criteria). 11 -Hospitals that were continuously accepting COVID-19 ICU patients: at least one 12 COVID-19 ICU patient in every month in the study period. 13 -Hospitals that were accepting non COVID-19 ICU patients: non COVID-19 ICU 14 patient in the study period. 15 -Hospitals that were intermediately accepting COVID-19 ICU patients: hospitals 16 meeting neither of the categories above. 17

To account for regional variation in the COVID-19 epidemic, the restricted 18 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 analysis above was performed in limited areas where the impact of COVID-19 was 1 considered to be the largest. These areas included Hokkaido, Tochigi, Saitama, Chiba, 2 Tokyo, Kanagawa, Gifu, Aichi, Osaka, Kyoto, Hyogo, Okayama, Hiroshima, Fukuoka, 3 and Okinawa, where the duration of the first or second states of emergency was longer 4 than other areas of Japan [17] . 5 6 Changes in initial treatments for non-COVID-19 ICU patients at the start of the 7 epidemic 8

We investigated changes in the initial treatments for non-COVID-19 ICU 9 patients at the start of the epidemic. Specifically, the proportion of initial treatments that 10 were distinctive for ICU patients, such as extracorporeal membrane oxygenation 11 (ECMO), MV, noninvasive positive pressure ventilation or nasal high-flow therapy 12 (NIPPV/NHF), renal replacement therapy (RRT), and vasopressors [19, 20] , were 13 examined. Initial treatments were defined as treatments received on the day of 14 admission to the ICU. The proportion of initial treatments in each wave of the COVID-15 19 epidemic in Japan were compared with those in the corresponding months in the 16 years before the epidemic -from April 2018 to March 2020. 17 18 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 Standardized mortality ratios of non-COVID-19 ICU patients during the COVID-1 19 epidemic 2

We investigated the change in the standardized mortality ratio of non-COVID-3 19 ICU patients from the beginning of the epidemic, based on indirect standardization. 4

First, a prediction model for the in-hospital mortality of non-COVID-19 patients was 5 developed based on observations before the epidemic. Second, based on observations 6 after the COVID-19 epidemic began, the ratio of observed in-hospital deaths to 7 expected in-hospital deaths from the prediction model was calculated as a standardized 8 mortality ratio (SMR). The SMR was calculated in each wave of COVID-19 infections, 9 stratified into the categories of hospitals based on acceptance of COVID-19 patients in 10 the ICU. 11

Although the DPC/PDPS data from the target population of this study did not 12 include risk scores for ICU patients (such as a sequential organ failure assessment 13 (SOFA) score and acute physiology and chronic health evaluation (APACHE) Ⅱ 14 scores), the available data had an acceptable performance with regards to risk 15 adjustment for ICU patients [19, 20] . In this study, sex, age, smoking history, body mass 16 index (BMI), major diagnosis category (MDC), ICU category of initial admission (sICU, 17 eICU, or HCU), initial ICU treatment (ECMO, MV, NIPPV/NHF, RRT, and 18 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101/2022.02.28.22271604 doi: medRxiv preprint vasopressors), admission process (post-emergency operation, post elective operation, or 1 medical indication), months of admissions to ICUs, e.g., January or February, were used 2 for predictive analysis. 3 4

A chi-square test was performed to compare the proportion of initial treatments 6 of non-COVID-19 ICU patients before and after the epidemic. A statistical significance 7 level of 5% was set. 8 A logistic regression model was employed to develop the prediction model of 9

in-hospital mortality among non-COVID-19 ICU patients. To account for potential 10 clustering by hospitals, a multilevel model with random intercepts for each hospital was 11 applied [21] . For point estimates and confidence intervals, bootstrap methods were 12 employed [21] . Resampling with replacement from the observed data was repeated 1,000 13 times, and percentiles (2.5% and 97.5%) from the distribution of SMRs were calculated 14 as the lower and upper limits of the confidence intervals. The fiftieth percentile of the 15 distribution was calculated as the point estimate of the SMRs. 16 SAS software version 9.4 (SAS Institute Inc., Cary, NC) was used for all 17 statistical analyses; PROC GLIMMIX was used for the multilevel logistic regressions. 18 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint was in April 2020, the second peak was in August 2020, the third peak was in January 17 2021, and the fourth peak was in May 2021. The third and fourth peaks were larger in 18 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101/2022.02.28.22271604 doi: medRxiv preprint size. 1

The ratios of new ICU admissions of non-COVID-19 patients declined around 2 the same time as the four peaks in the number of COVID-19 patient admissions to 3

ICUs: a 21% decrease in May 2020, an 8% decrease in August 2020, an 9% decrease in 4

February 2021, and a 14% decrease in May 2021. Similarly, the ratios of new ICU 5 admissions of non-COVID-19 patients initially receiving MV decreased in the same 6 months, but to a greater degree: a 26% decrease in May 2020, a 15% decrease in August 7 2020, an 19% decrease in February 2021, and 19% decrease in May 2021. 8 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ;  Changes in initial treatments for non-COVID-19 ICU patients at the start of the 1 COVID-19 epidemic 2 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101/2022.02.28.22271604 doi: medRxiv preprint Table 2 shows the standardized mortality ratios of ICU patients admitted in 1 each wave of the epidemic. The confidence intervals of SMRs included one in all waves 2 of the epidemic. Supplementary Table 6 shows the SMRs stratified into the hospital 3 categories. There was no statistically significant increase in SMRs, in any of the 4 categories of hospitals, and in any of the waves. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101/2022.02.28.22271604 doi: medRxiv preprint DISCUSSION 1

Our study investigated a large claims database in Japan and revealed the 2 medium-term impact of waves of the COVID-19 epidemic on the case volumes, initial 3 treatments, and in-hospital mortality ratios of non-COVID-19 patients admitted to ICUs. 4 Descriptive analysis revealed that the waves of the COVID-19 epidemic 5 negatively impacted the volume of non-COVID-19 ICU admissions. A previous study 6

[6] suggested that the reduced volume of ICU patients was due to patients' hesitancy to 7 visit hospitals. In fact, a reduction in case volumes has been observed in many areas, 8

including acute coronary syndrome [22] , pneumonia[23], and surgeries [24] . However, 9

the reasons behind the reduction in admissions have not been clarified. In addition to 10 patient hesitancy, another potential mechanism of reduced case volumes is the 11 postponement of non-emergency treatments or tests to reserve capacity for critically ill 12 COVID-19 patients. Postponing non-essential procedures was implemented in various 13 countries around the world [3, 4] , and the Japanese government also requested health 14 care providers to postpone non-emergency medical procedures during waves of 15 COVID-19 infections [25] . We observed that the decrease in non-COVID-19 ICU 16 patient admissions was greater for hospitals continuously accepting COVID-19 ICU 17 patients, results that are consistent with the latter mechanism of postponement of non-18 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; emergency procedures. 1

This study observed that the impact on case volumes was largest in the first 2 wave compared to subsequent waves. In other countries, decreased patient volumes 3 during subsequent waves of COVID-19 have been reported, but comparisons with the 4 impact of the first wave are not in agreement with the present study [26, 27] . One 5 potential explanation for the smaller impact in subsequent waves is the more organized 6 management of ICU beds. For instance, during subsequent waves, health professionals 7 might take advantage of lessons learned during the first wave. Another possible 8 explanation is less hesitancy to visit hospitals during subsequent waves. Residents may 9 have gradually become accustomed to the COVID-19 epidemic. In fact, reduced 10 mobility in public spaces in Japan was reported to be smaller in subsequent waves than 11 in the first wave [28] . is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint Our finding that the confidence intervals of SMRs included one in all waves of 5 the epidemic implies that treatment quality for non-COVID-19 ICU patients was 6 maintained in Japan. Global evidence has been inconsistent about the quality of 7 intensive care during the COVID-19 pandemic [7, 8, [12] [13] [14] . Many factors influence the 8 epidemic's impact on the quality of intensive care. Confirmed COVID-19 patient 9 numbers were smaller in Japan than in other high-income countries [30] , which might 10 explain our findings. 11

This study has several limitations. First, although the characteristics of the 12 data-providing hospitals were varied, the data collection relied on the voluntary 13 participation of the hospitals. This may introduce selection bias and limit the 14 generalizability of our findings. Second, the DPC/PDPS data of the study population did 15 not include risk scores of ICU patient severity, such as SOFA or APACHE Ⅱ scores. 16

Although the prediction performance was good in our study, the risk was adjusted in 17 different ways compared to other studies [7, 8] . Third, data about the demand for 18 . CC-BY-NC-ND 4.0 International license It is made available under a perpetuity.

is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101/2022.02.28.22271604 doi: medRxiv preprint treatments is not available. As mentioned in the previous section, it is difficult to 1 distinguish between the suppression of required treatments and a decline in the demand 2 for treatments. Further research is warranted, including an investigation of the trend in 3 disease volumes in the general population. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint We revealed that the number of non-COVID-19 ICU patients continuously 2 decreased over the medium term during the COVID-19 epidemic, compared to the 3 previous year. The decrease in case volumes was larger among non-COVID-19 ICU 4 patients initially receiving MV than those undergoing other initial treatments. The 5 standardized in-hospital mortality of non-COVID-19 ICU patients did not change in any 6 waves of the epidemic. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101/2022.02.28.22271604 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 ratio to before the epidemic, all patients* ratio to before the epidemic, non-COVID-19 patients* ratio to before the epidemic, non-COVID-19 patients initially admitted to sICU* ratio to before the epidemic, non-COVID-19 patients initially undergoing MV* ratio to before the epidemic, non-COVID-19 patients initially administering vasopressor* Ratio to before the epidemic ratio to before the epidemic, non-COVID-19 patients (COVID-19 acceptance, non)* ratio to before the epidemic, non-COVID-19 patients (COVID-19 acceptance, intermediate)* ratio to before the epidemic, non-COVID-19 patients (COVID-19 acceptance, continuous)* is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted February 28, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 

Coronavirus disease 2019; ICU, intensive care unit; sICU, specialized-care ICU; MV, mechanical ventilation * From

the ratios of case volumes of non-COVID-19 patient admissions to ICUs in each month to the same month in the previous year, stratified by hospitals Case numbers

Non-COVID-19 patients (COVID-19 acceptance, few)** Non-COVID-19 patients

acceptance, intermediate)** Non-COVID-19 patients

Trends in the ratios of case volumes of non-COVID-19 patient admissions to ICUs in each month to the same month in the previous year, stratified by hospitals, in the prefectures with proactive COVID-19 policies Case numbers

Non-COVID-19 patients (COVID-19 acceptance, few)** Non-COVID-19 patients

acceptance, intermediate)** Non-COVID-19 patients

the ratios of case volumes of non-COVID-19 patient admissions to ICUs in each month to the same month in the previous year, stratified by hospitals (classified by the month criteria) Case numbers

Non-COVID-19 patients (COVID-19 acceptance, non)** Non-COVID-19 patients

acceptance, intermediate)** Non-COVID-19 patients