Resource utilization and outcomes in emergency general surgery during the COVID19 pandemic: An observational cost analysis

Background Over the course of the COVID19 pandemic, global healthcare delivery has declined. Surgery is one of the most resource-intensive area of medicine; loss of surgical care has had untold health and economic consequences. Herein, we evaluate resource utilization, outcomes, and healthcare costs associated with unplanned surgery admissions during the height of the pandemic in 2020 versus the same period in 2019. Methods Retrospective analysis on patients ≥18 years admitted from the emergency department to General & Digestive and Gastrointestinal Surgery Services between February and May 2019 and 2020 at our center; clinical outcomes and unadjusted and adjusted per-person healthcare costs were analyzed. Results Consults and admissions to surgery declined between February and May 2020 by 37% and 19%, respectively, relative to the same period in 2019, with even greater relative decline during late March and early April. Time between onset of symptoms to diagnosis increased from 2±3 days 2019 to 5±22 days 2020 (P = 0.01). Overall hospital stay was two days less in 2020 (P = 0.19). Complications (Comprehensive Complication Index 10.3±23.7 2019 vs. 13.9±25.5 2020, P = 0.10) and mortality rates (3% vs. 4%, respectively, P = 0.58) did not vary. Mean unadjusted per-person costs for patients in the 2019 and 2020 cohorts were 5,886.72€±12,576.33€ and 5,287.62±7,220.16€, respectively (P = 0.43). Following multivariate analysis, costs remained similar (4,656.89€±390.53€ 2019 vs. 4,938.54±406.55€ 2020, P = 0.28). Conclusions Healthcare delivery and spending for unplanned general surgery admissions declined considerably due to COVID19. These results provide a small yet relevant illustration of clinical and economic ramifications of this healthcare crisis.

OR, operating room.

INTRODUCTION
Apart from its direct impact on infected patients, the SARS-CoV-2/COVID19 pandemic has had indirect collateral effects on patients with other pathologies who have accessed standard of care for their diseases 1,2 . Surgery is one of the most resource-intensive areas of clinical medicine 3 , and patients requiring surgical interventions have been particularly vulnerable in this regard.
Factors adversely affecting surgical patients have included lack of operating room personnel and space. Inability and/or fear of patients with incidental surgical pathology to present in timely fashion has resulted in a well-documented declines in surgical consults during the pandemic in many parts of the world [4][5][6][7][8][9] . Moreover, it has been suggested that these patients may have also presented later, with more advanced disease 9,10 . These issues have potential to not only impact patients' lives and quality of life but also put further strain on public money and healthcare resources at a moment when resources are severely diminished.
Cost studies in healthcare are performed with the objective of transforming the impact of a disease in economic terms. They provide empirical evidence to facilitate assessment by public decision-makers, who have limited budgets, in order to make rational decisions regarding efficient allocation of resources.
The degree to which patients with incidental surgical pathology derived from the emergency department may have been associated with a change in direct healthcare spending during the COVID19 pandemic is not well understood. We hypothesized that analysis of costs associated with unplanned general surgery admissions during 2020 would likely demonstrate differences versus those of the previous year, 2019, where no pandemic occurred. We performed an exploratory cost analysis to determine differences in resource utilization, clinical outcomes, and direct healthcare costs, aiming to identify what patterns may have existed and if they varied according to presenting diagnosis, in order to identify any potential areas for improving healthcare expenditures without compromising patient outcomes in future waves of the same disease or new pandemics of a similar nature.

Study design
This is a retrospective study on adult patients (≥18 years) admitted from the Emergency

Definition of variables and outcomes
The main objectives of the study were to assess clinical outcomes and costs associated with general surgery ED admissions during the peak of the pandemic in 2020 relative to the same period in 2019. The study was conducted from the perspective of the hospital provider; no primary care or other ambulatory healthcare costs were imputed. operating room (OR) and staff costs. Death did not result in cost-censoring and was considered to represent complete data in that no further cost was accrued; costs of patients who died before discharge were included in the analysis.

Data presentation and analysis
Unadjusted and adjusted direct healthcare costs are presented for the four-month periods in 2019 and 2020. Continuous variables are presented as arithmetic mean ± standard deviation and categorical variables as frequencies, unless otherwise specified. Univariate analysis of costs and differences in continuous and categorical variables were assessed using the parametric test of cost on untransformed (raw) scale, univariate two-sample t test, and Chi-squared test, respectively. Multivariate analysis of costs was performed to adjust for potential cost variation due to other causes (confounding variables). Generalized linear models (GLM) were used for multivariate analysis, as they allow for heteroscedasticity through a variance structure, relating variance to mean 13 . The modified Parks test was used to identify the Gamma family for the GLM 14  February 2020 (consults -48%, admissions -41%). This decline continued through April and returned to near normal by May 2020 (FIGURE 1).

Patient demographics
Baseline patient sociodemographic and clinical characteristics are listed in TABLE 1. The average age of all patients was 60±20 years, 43% were women, 78% were from Spain, and 39% were married; 46% were pensioners who had previously worked and another 37% were actively employed. Age, sex, country of birth, employment situation, profession, ASA classification, and history of prior surgery did not differ between the two cohorts; civil status was unavailable for a greater proportion of case in the 2020 cohort relative to 2019.

Clinical variables and outcomes
TABLE 2 provides data from the two cohorts related to admissions, interventions, and overall hospital stays. Among patients in the 2020 cohort, ten (3%) were concomitantly diagnosed with COVID19.
There were differences in the representation of the different diagnostic categories between the two periods, with the greatest number of admissions in 2019 related to acute appendicitis (N=106, 25%) and the greatest number in 2020 related to biliary tract pathology (N=103, 30%).
Time between onset of symptoms to diagnosis was significantly increased from an average of two days in 2019 to an average of five days in 2020 (P=0.01). The percent of cases initially managed non-operatively and rate of failure of non-operative management did not differ between the two periods. Intraoperative complications and operative times were similar. Overall hospital stay was, on average, two days less in 2020, though this difference was not statistically significant. Complications (CCI 10.3±23.7 2019 group vs. 13.9±25.5 2020 group, P=0.10) and mortality rates (3% vs. 4%, respectively, P=0.58) did not vary.

Cost outcomes
Costs for hospitalization, surgery, and readmissions were available for 409 patients from 2019 and 334 from 2020.
Median per-person costs were lower in both instances (2,532.77€ and 3,217.40€, respectively), indicating that cost distributions were skewed to the left and the majority of patients actually had costs below the arithmetic means. FIGURE 2 reflects how costs were distributed per month in each period.
Following unadjusted univariate analysis, a few significant differences between the two periods were noted, with higher costs associated with miscellaneous consumables used during hospitalization and lower costs associated with OR consumables and OR usage in 2020 relative to 2019 (TABLE 3). Following multivariate analysis and adjustment of costs to account for potential influence of confounding covariates, mean per-person healthcare costs remained similar between the two periods. Marginal but statistically significant differences were observed for costs associated with pharmacy and miscellaneous consumable materials (higher in 2020) and OR consumables (lower in 2020).

DISCUSSION
The World Health Organization declared SARS-CoV-2 a global pandemic on March 11, 2020; national lockdown in Spain was implemented three days later. At our center, ED consults and admissions for COVID19 infection rose progressively until they peaked the last week in March.
Coincidentally, consults for other urgent pathology declined considerably, and overall ED volume was reduced 5 . The peak was associated with >80% occupation of intensive care and 50% occupation of general floor beds with COVID19 patients. As of this writing, late March and early April continue to represent the maximum point of COVID19 hospitalizations in our region and center.
The objective of this study was not to evaluate costs specifically associated with COVID19 admissions but, rather, those associated with other urgent disease processes displaced or delayed for whatever reasonby COVID19. Based on this study's findings, in spite of the fact that ED consults and admissions to surgery declined by up to nearly 50% relative to the same period in 2019 and presentations were delayed by an average of three days, mean per-patient costs of hospitalization (including readmissions in addition to index admission) were no different during the peak of the pandemic relative to the same period in 2019.
A logical assumption is that patients arriving later would have arrived sicker. The concept of fewer but sicker patients has not been borne out in this study in terms of the complications, mortality, and costs we observed. Costs have been seen to be significantly associated with the number and severity of complications arising in the course of a disease process 15,16 . Fewer general surgery patients were attended and admitted from the ED between February and May 2020 relative to the same period in 2019, but those that were seen in 2020 did not present higher level of acuity, on average, than those seen in the previous period without pandemic.
We considered that while average costs might have been globally the same for the four-month periods under evaluation, a more detailed analysis comparing individual months might reveal differences. As reflected in FIGURE 2, March 2020 was the month with the lowest per-person healthcare costs. The decline in March, however, was not followed by any increase in costs in April or, for that matter, May 2020. That is to say, there was no rebound effect in the opposite direction following the initial reduction in usage of emergency general surgery services.
While total per-person costs did not vary between the two periods under evaluation, some marginal but nonetheless significant differences were observed in adjusted costs associated with consumable materials, in particular. While small, what these differences likely reflect, if anything, are changes in practice patterns in 2020 relative to 2019 (using, for example, more personal protective equipment in 2020) as opposed to higher degree of medical or surgical acuity, as clinical outcomes did not vary.
Considering the diagnostic subpopulations, admissions for biliary tract pathology were similar if not slightly higher in 2020 relative to 2019. This observation appears consistent with what is known about dietary and other lifestyle changes that occur(ed) during the pandemic that are largely known to exacerbate biliary tract disease [17][18][19] . It may be the case that there was some decline in "appendicitis" cases based on fewer presentations of right lower quadrant pain that ultimately resolved with symptomatic/medical management. Given that this study only includes costs associated with hospitalization and not outpatient costs, it is plausible that reduction in ED services for other diagnostic subsets, including obstructive and non-obstructive gastrointestinal pathology (perforations, inflammatory pathology, symptomatic tumors, etc.), was compensated to some degree by a shift toward greater utilization of outpatient resources. Outpatient care, however, appears to have experienced a similar decline as urgent/emergent care consults, and healthcare in general was lost during the height of the pandemic in various parts of the world 1,20,21 . Whether loss of healthcare for non-COVID19 patients accounts for some proportion of excess mortality observed in different countries, including Spain, where excess mortality between February and May was 38% 22 , is conceivable but difficult to assess at this point.
We attempted to evaluate cost differences in different diagnostic subpopulations. It is unclear whether cost differences observed for the appendicitis subpopulation, with higher apparent costs in 2020 relative to 2019 on univariate analysis, were truly related to period or were a consequence of confounding variables, as cost adjustments could not be performed. The fact that comprehensive complications for appendicitis admissions were no different in 2020 relative to 2019 is more indicative of the latter.
Based on the reduction in unplanned general surgery admissions and lack of simultaneous increase in per-patient costs for hospitalization, we estimate that healthcare spending at our center was reduced by nearly 650,000€ for this particular patient population between February and May 2020 relative 2019. General surgery is responsible for 12-13% of our emergency department volume, not to mention elective activity, and we can extrapolate further by considering the sum of similarly resource-intensive areas of medicine and reasonably assume that millions of Eurospotentially in excess of 5-6 millionin healthcare payments were lost at our institution during the height of the pandemic, unrelated to the actual treatment of COVID19. The ultimate repercussions of such findings depend on the healthcare structure and reimbursement system but may be considered to include everything from risk of healthcarerelated job losses and loss of income for third parties providing healthcare-related goods and services to considerable financial profits for healthcare insurance providers due to unspent premiums 23 .
This study has several limitations. Results we present are from the first wave of the pandemic in Spain and are influenced by the specific lockdown measures that were implemented at that time.
As well, it considers resource utilization and costs in emergency general surgery only and does not evaluate other areas of healthcare. A study from Asia examined orthopedic surgery charges and found similar if not reduced in-hospital costs for orthopedic procedures (both urgent and elective) during the pandemic relative to pre-pandemic 24 . Another study from China regarding acute ischemic stroke, however, observed the opposite effect, with higher hospitalization costs during the height of the pandemic there 25 . Based on these findings, further cost studies in other areas of clinical medicine, in particular those requiring direct patient intervention and other activities that cannot be completed via telemedicine, are encouraged.

CONCLUSIONS
During the height of the first wave of the COVID19 pandemic, in spite of up to an almost 50% reduction in emergency general surgery consults and admissions, no differences in outcomes or per-patient hospital-associated healthcare costs were observed at our center. It is improbable that pathology truly disappeared, and this decline in unplanned surgical activity undoubtedly represents part of the "untold burden" of patients that failed to seek care. As the pandemic progresses, it is important to continue to document not only the health-related effects but also the economic consequences of loss of healthcare, as they have important implications for patients, providers, payers, other healthcare stakeholders, and society.