key: cord-0909172-a52bwuim
authors: Aranda-Michel, Edgar; Serna-Gallegos, Derek; Arnaoutakis, George; Kilic, Arman; Brown, James A.; Dai, Yancheng; Dunn-Lewis, Courtenay; Sultan, Ibrahim
title: The Effect of COVID-19 on Cardiac Surgical Volume and its Associated Costs
date: 2022-04-02
journal: Semin Thorac Cardiovasc Surg
DOI: 10.1053/j.semtcvs.2022.01.009
sha: 5f8ee4637ecdc44b45fedf86c9b6b474fea25b46
doc_id: 909172
cord_uid: a52bwuim

The COVID-19 pandemic significantly affected healthcare and in particular surgical volume. However, no data surrounding lost hospital revenue due to decreased cardiac surgical volume have been reported. The National Inpatient Sample database was used with decreases in cardiac surgery at a single center to generate a national estimate of decreased cardiac operative volume. Hospital charges and provided charge to cost ratios were used to create estimates of lost hospital revenue, adjusted for 2020 dollars. The COVID period was defined as January to May of 2020. A Gompertz function was used to model cardiac volume growth to pre-COVID levels. Single center cardiac case demographics were internally compared during January to May for 2019 and 2020 to create an estimate of volume reduction due to COVID. The maximum decrease in cardiac surgical volume was 28.3%. Cumulative case volume and hospital revenue loss during the COVID months as well as the recovery period totaled over 35 thousand cases and 2.5 billion dollars. Institutionally, patients during COVID months were younger, more frequently undergoing a CABG procedure, and had a longer length of stay. The pandemic caused a significant decrease in cardiac surgical volume and a subsequent decrease in hospital revenue. This data can be used to address the accumulated surgical backlog and programmatic changes for future occurrences.

'The COVID-19 pandemic significantly decreased cardiac operative volume across the country.

Our center underwent a fast recovery, but an nationally estimated substantial decrease in hospital revenue and case backlog occurred

The effects of COVID-19 on the healthcare system are profound. As cardiothoracic surgery is a top earner for hospital systems, decreases in operative volume likely result in significant lost revenue. The use of a national database allows for an estimate of lost revenue using institutional and published decreased cardiac surgery volume with the goal of ameliorating current and future pandemic effects. 

The COVID -19 pandemic has wrought unprecedented disturbances throughout society, not just in the United States but across the globe 1, 2 . From layoffs, city shutdowns, economic turmoil, and loss of life, almost no facet of society is untouched, and healthcare is no exception 3 .

The effects on healthcare have been variable. While the volume of some services and elective procedures has decreased 4, 5 , there has been an inundation of patients in other areas, such as the intensive care unit [6] [7] [8] . While surgery as a whole has been affected, various specialties have had different experiences with changes to case volume-based primarily on the urgency of the surgery required. This reallocation shift and eventual return to pre-pandemic protocols for patient care will continue to impact hospitals, surgeons, and patients significantly.

The American Hospital Association estimated a loss of more than 160 billion dollars from all canceled surgeries and other services between March and June of 2020 due to COVID 9 .

As cardiothoracic operations are routinely high charging procedures for hospitals 10, 11 , a decrease in case volume is likely to reduce hospital revenue significantly. In a recent survey of cardiothoracic surgery program directors and their response to COVID-19, over 95% of respondents reported canceling elective and non-urgent operations, with a further 33% restricting urgent operations during the peak of the pandemic 12 . In another survey, with 502 cardiac surgeons responding, 81% indicated that only urgent and emergent cases were scheduled, with 76% reporting a >50% decrease in surgery volume 13 . In this paper, we sought to use our institution's change in cardiac surgery volume, representative of a large volume academic center, and the National Inpatient Sample (NIS) database to estimate the nationwide effect of COVID-19 on hospital volume and the associated lost hospital revenue.

The National Inpatient Sample Database (NIS) is a prospectively maintained database created by the Healthcare Cost and Utilization Project. This database is drawn from all states and covers over 97% of the US population. The data given is a 20% stratified sample of all discharges with provided weights to estimate the national population. Furthermore, the NIS includes ICD 9 and 10 codes of the diagnosis and procedures performed for each patient and the hospital charges for each stay. The NIS was filtered for all patients with hospital stays between 2004 and 2017 who underwent cardiac surgery. The procedure-related Clinical Classifications Software (CCS) numbers used to screen for cardiac surgery cases are outlined in Table 1 . The CCS aggregates ICD 9 and 10 codes to useful groupings for broader categories of operations, facilitating a constant comparison between ICD 9 and 10 codes. These patients were stratified by month and year of their hospital stay. All hospital charges were adjusted to 2020 dollars using each year's gross domestic product 14 . The NIS provides event level weights which were applied to generate national estimates of volume and charges. Additionally, hospital level charge-to-cost ratios were provided by the database and used to convert hospital charges to hospital costs. It is important to clarify that hospital costs are a better estimate of the revenue a hospital would receive for a patient stay. As such, changes in the hospital cost due to Covid can be considered lost revenue, which will be addressed as such from here on. Additionally, our institutional Society of Thoracic Surgery Database was utilized to analyze cases between January and May of 2019 and 2020 to compare baseline features and outcomes. During COVID peak weeks in the state of Pennsylvania, all elective cases at our institutions were carefully scrutinized and only urgent and emergent cases were performed. Most patients undergoing elective surgery for coronary, valvular or aortic disease were reviewed by two surgeons to ensure that the risk of watchful waiting exceeded risk of contracting COVID-19 while an inpatient during surgery and hospitalization. Moreover, every patient was tested for covid-19 prior to elective or urgent surgeryIRB approval was waived for this project.

The NIS data was available through 2017, and thus the case volume and hospital charges were projected to 2020. Linear regression was carried on cardiac surgery volume and hospital charges using the month as the dependent variable. The resulting regression equation was used to estimate the 2020 COVID-19 period for the months January to May. Our institutional volume from January to May was compared to the previous year's (2019) to estimate volume loss. Each month of both years was compared as a percentage of our average monthly institutional volume across 2019, increasing the comparison's robustness. The subsequent change in the percentages was treated as a change in volume. This percentage loss was then applied to the monthly case and revenue projections from the linear regression and summed over this period to create cumulative loss estimates ( Figure 1 ). In short, NIS data was used to create projections for expected cardiac surgery volume and revenue in 2020. Our institutional volume experience decreased in 2020 when compared to 2019 served as a percentage change due to COVID that was then projected onto the national estimates, resulting in a national decrease due to COVID. A Gompertz function, a growth model, was fit to our institution's data starting in May and ending in June, which is when our hospital volume returned to normal.

Institution cases were compared with those in 2019 representing the control cases.

Categorial and continuous variables were analyzed with Chi Squared and Student's T Test respectively. R (4.0.0 'Arbor Day') was utilized for the analysis.

A total of 976 cases were done between January and May in 2019 and 551 cases were done during the same months in 2020 ( Table 2) . Patients were younger (67 vs 64, P<0.001) and more frequently male (66.1% vs 71.5%, P<0.001) during COVID months. Patient BMI was not statistically different and patients had similar STS predicted risk of mortality between the pre-COVID and COVID months. There was an increase in the number of CABG procedures (54.2% vs 64.8%, P<0.001) whereas the frequency of elective procedures did not differ between cohorts.

The length of stay was longer during COVID (9.5 vs 10.5 days, P=0.044) and 30-day mortality was similar (3.1% vs 4.4%, P = 0.247) between cohorts.

The linear regression for case volume and hospital revenue is presented in Figure 2 .

Projected hospital volume remained relatively unchanged over time with a regression slope of 1.68, an intercept of 41,597, and an R 2 of 0.0009. Projected hospital revenue were also relatively stable over the study period, with a slope of -0.0005, an intercept of 3, and an R 2 of 0.0002 (note units are in billions of dollars). Projected national hospital volume slightly increased from 41,922 cases in January to 41,929 cases in May (Table 3) . Projected hospital revenue marginally decreased from 2.9895 billion to 2.9892 billion. Note that there is not an R 2 for the recovery function. This is because the resolution of our institutional data is month to month and since the recovery occurred over a month, we only have the start and end of the recovery model. There are no datapoints of the dynamic between these two points, not allowing for a calculation of a R 2 .

Institutional, and thus nationwide, decreases in cases and subsequent revenue were -19.2%, -10.9%, -6.2%, -10.9%, and -28.3% for months January through May respectively ( Figure 3 ). This translated into decreased national hospital volume of 33,868.5, 37,364.9, 39,341.9, 37,367.9, and 30,074.5 respectively (Table 3) . Similarly, the reduced hospital revenue It is important to place these results in the context of other findings, such as a recent survey which places the overall decrease in cardiac surgery volume to between 45% 15 and 47% 16 of pre-COVID levels, potentially resulting in 125 thousand lost cases totaling 8.9 billion in lost hospital revenue.

In this manuscript, we used our institutional data in conjunction with the National Inpatient Sample database to estimate the effect of COVID-19 on cardiac surgery cases and hospital revenue in the United States. There are three main findings of this paper: 1) the overall NIS projections shows a steady cardiac surgery volume over time without dramatic increases or decreases in volume along with stable hospital revenue, 2) a significant estimated amount of national hospital revenue was lost due to COVID-19, and 3) recovery from the effects of the pandemic was rapid at a single institution. However, despite the rapid recovery, the case and charge loss's full effect will take time to realize. 18, 19 , their incidence has been low and without an increase for frequently performed operations in this period 17, 20 , which would result in decreased revenue. On the other end of the spectrum, cardiac surgery is intimately connected with technology, readily adopting advances in devices. These advancements, however, typically carry an increased cost 21, 22 . Yet, this increased cost may be upfront and not indicative of complications, as some studies show that transcatheter aortic valve patients have a shorter length of stay compared with surgical aortic valve patients 23 . This is of relevance given the recent surge in endovascular devices to treat aortic stenosis and aortic dissection. All these competing factors obfuscate what an expected trend of hospital revenue would look like. A final variable that significantly affects revenue prediction is hospital variability in charges and revenue 24, 25 , which can cause biases in a subsampled dataset. Nevertheless, the national estimated lost revenue is substantial, and it is promising that cardiac surgery volume was able to recover quickly.

The Gompertz function is used to model growth in numerous situations, including tumor growth, virus spreading, recovery, and can be applied to hospital volume [26] [27] [28] [29] . Recovery at our institution occurred over a month, with 99% of maximal capacity being estimated at 23 days. This is in stark contrast to a study done by Jain et al., which used Gompertz functions to estimate the recovery time in elective orthopedic surgery 28 . They did not benefit from a known endpoint for recovery and opted for Monte Carlo simulations over three model growth rates, resulting in estimated recovery times between 7 and 16 months. There are two key points to address to put this finding in the proper context. First, there is a substantial difference in estimated pre-COVID case numbers, 170,000 and 41,000 monthly cases for elective orthopedic and all cardiac surgery, respectively. Secondly, cardiac surgery patients likely have increased comorbidities than an elective orthopedic population, necessitating a more immediate intervention and differing rates of elective, urgent, and emergent procedures. This difference in the nature of the cases is easily understood when comparing the lowest hospital volumes of 3% in orthopedic surgical patients and 71.7% for the cardiac surgery cohort. The higher projected volume with a larger percentage of lost cases in orthopedic surgery leads to the longer recovery time seen in the elective orthopedic cohort. This illustrates how COVID-19 can disproportionately affect surgical specialties and is a crucial consideration for departments and hospitals developing response plans and projecting future productivity. While cardiac surgery did not have as severe a reduction in volume as others 28, 30, 31 , the lost potential revenue was substantial, amassing an estimated greater than 2 billion dollars in lost hospital revenue.

The estimates provided here can be considered conservative. While the lowest volume seen at our institution was 72.7% of pre-COVID levels, this level of decrease may not apply to all hospitals. In fact, recent national estimates have placed the overall decrease in cardiac surgery volume to between 45% 15 and 47% 16 of pre-COVID levels. The reason for this discrepancy is that COVID infection cases differed dramatically between states. As COVID and hospital policy was enacted on a state and hospital levels, differences in COVID volume (both duration and magnitude) would likely affect hospital surgical volume, leading to inter-hospital variability.

Indeed, a recent study by Ad et al. 15 with an increasing proportion of CABG procedures 15 and an increased length of stay 32 . However, it is likely that the COVID burden in our cardiac surgery population was minimal given the similar 30-day mortality compared to the previous year, and the lower 30-day mortality compared to other centers 15, 33 . Using the most significant national decrease 15 in cardiac surgery case volume (45%), the range of lost hospital cases is 37,505 -125,000, with an accompanying hospital revenue loss of 2.55 -8.9 billion.

The most significant limitation in this study is the use of a single center extrapolated to the nation. While this institution's experience is of a large volume academic center, there could be differences between such centers, their proportion of elective and urgent/emergent cases, and their changes in volumes throughout, thus biasing our results. This is of particular importance as the COVID-19 policy was enacted on a state-by-state basis. Moreover, different institutions could have different recovery dynamics that my not be governed by the Gompertz growth function used or by a generalized Gompertz growth function at all. This would result in biases for the lost cases and revenue during the recovery period. This is also likely true in that our relatively unchanged elective surgery volume was not typical of the national trend, which would make the estimates presented here conservative in nature. Additionally, we did not have access to our institutional change in costs which can bias the results. The second limitation is that the reduction in surgery was assumed to be proportional among all procedures. It is possible that some cases were affected more than others, which could influence the actual reduction in charges and revenue as different cardiac surgery operations will have differing associated economic burdens. Additionally, there are inherent limitations in using the NIS database and CCS coding, which may not capture the complete volume or heterogeneity in cardiac surgery volume as well as the fact that the data was projected to make estimates for expected 2020 volume, potentially adding bias. Another point to emphasize is the difference in volume trends from the NIS and the national STS data, with the STS data showing a larger increase. This would further make these estimates conservative. Lastly, clinical classification software may not capture the full spectrum of cardiac surgery.

Using our institutional cardiac surgery case volume in parallel with a nationwide database, we estimated the cardiac surgical case volume and hospital charge loss associated with the COVID-19 pandemic. There was a significant estimated reduction in national case volume, national loss of hospital revenue, and the cumulative backlog of patients. However, there was an expedient return to baseline case volume at our institution which could extend to other institutions. Despite the rapid recovery, the question remains as to how to address this backlog of patients. While a subset has pathology that allows them to await intervention safely, others may transition from an elective situation to an urgent one, likely carrying increased morbidity and mortality along with an increased cost to the hospital system. It will be up to hospital leadership to develop plans to develop strategies to address this backlog and react to any future loss in case volume.

To view the Webcast of this AATS 101st Annual Meeting presentation see the URL below: Webcast URL:

https://aats.blob.core.windows.net/media/21%20AM/AM21_A17/AM21_A17_01.mp4

Graphical Abstract and Figure 1 . Single center volume decrease due to covid was extrapolated to the nation using a national database. While the single center experience showed a rapid recovery, the possible national effect was a substantial amount of backlogged surgical cases with significant lost hospital revenue. 

The Global Impact of and Responses to the COVID-19 Pandemic

Impact of COVID-19 pandemic on socioeconomic and mental health aspects in Nepal

The socioeconomic implications of the coronavirus pandemic (COVID-19): A review

Elective surgery in the time of COVID-19

American College of Surgeons. COVID-19: guidance for triage of non-emergent surgical procedures

Critical Care Utilization for the COVID-19 Outbreak in Lombardy, Italy: Early Experience and Forecast During an Emergency Response

There's No Place for Them to Go': I.C.U. Beds Near Capacity Across

The role of extracorporeal life support for patients with COVID-19: Preliminary results from a statewide experience

Hospitals and Health Systems Face Unprecedented Financial Pressures Due to COVID-19

Surgeon contribution to hospital bottom line: not all are created equal

The impact of COVID-19 on thoracic surgery residency programs in the US: A program director survey

Programmatic Responses to the Coronavirus Pandemic: A Survey of 502 Cardiac Surgeons. The Annals of thoracic surgery

Adjusting Health Expenditures for Inflation: A Review of Measures for Health Services Research in the United States. Health services research

Cardiac surgery in North America and coronavirus disease 2019 (COVID-19): Regional variability in burden and impact

Trends and Outcomes in Adult Cardiac Surgery Patients During the COVID Pandemic: Analysis of the Society of Thoracic Surgeons Adult Cardiac Surgery Database

The Society of Thoracic Surgeons Adult Cardiac Surgery Database

Healthcare utilization and costs of cardiopulmonary complications following cardiac surgery in the United States

Prediction of costs and length of stay in coronary artery bypass grafting

Trends in isolated coronary artery bypass grafting: an analysis of the Society of Thoracic Surgeons adult cardiac surgery database

Cost-Effectiveness of Transcatheter Aortic Valve Replacement With a Self-Expanding Prosthesis Versus Surgical Aortic Valve Replacement

Economic Burden and Healthcare Resource Use for Thoracic Aortic Dissections and Thoracic Aortic Aneurysms&#x2014;A Population&#x2010;Based Cost&#x2010;of&#x2010;Illness Analysis

Comparison of Outcomes After Transcatheter Aortic Valve Replacement vs Surgical Aortic Valve Replacement Among Patients With Aortic Stenosis at Low Operative Risk

Understanding variability in hospital-specific costs of coronary artery bypass grafting represents an opportunity for standardizing care and improving resource use

The Variation in Hospital Charges: A Problem in Determining Cost/Benefit for Cardiac Surgery. The Annals of Thoracic Surgery

Population modeling of tumor growth curves and the reduced Gompertz model improve prediction of the age of experimental tumors

Comparison of Growth Patterns of COVID-19 Cases through the ARIMA and Gompertz Models. Case Studies: Austria, Switzerland, and Israel. Rambam Maimonides medical journal

SARS-CoV-2 Impact on Elective Orthopaedic Surgery: Implications for Post-Pandemic Recovery

The use of Gompertz models in growth analyses, and new Gompertz-model approach: An addition to the Unified-Richards family

Healthcare Professionals and the Impact of COVID-19. A Comparative Study of Revenue and Utilization: FAIR Health

Mortality and pulmonary complications in patients undergoing surgery with perioperative SARS-CoV-2 infection: an international cohort study