key: cord-0323313-ce0w9m6l authors: Chalmers, K.; Brownlee, S.; Gopinath, V.; SAINI, V. title: Overuse in US Medicare during the COVID-19 pandemic date: 2022-05-26 journal: nan DOI: 10.1101/2022.05.25.22275006 sha: 936c0b673891d80821d6fe9e8c325d49133bf4ad doc_id: 323313 cord_uid: ce0w9m6l Importance: While the COVID-19 pandemic disrupted regular hospital care and decreased overall volume of hospitalized patients, its impact on low-value care has not been investigated. Objective: To examine the impact of the COVID-19 pandemic on overuse rates. Design: A retrospective cohort study using Medicare fee-for-service claims. Setting: All outpatient and inpatient claims. Participants: Medicare beneficiaries aged 65 and over who met the criteria for one of 10 overuse measures were included, if they had a claim between January 1, 2019 to December 31, 2020. Exposure: Claims meeting overuse measure criteria over various time periods in 2020 were compared to the same period in the previous year: during the initial COVID-19 surge and shutdown (March 15 to May 2, 2020) and periods where states had high (if the mean 7-day case incidence was greater than 175 cases per 100,000 persons) or low COVID-19 incidence. Main Outcomes and Measures: The overuse measures in this study had two components: a denominator encompassing a particular patient cohort and a numerator to capture overuse of each service among patients within this cohort. We report claim volumes for both the denominator cohorts and overuse numerators as well as their rate differences (2020 versus 2019) using incidence rate ratios (IRRs) estimated using Poisson regressions. Results: There were 2,053,792 patients in 2019 and 1,699,807 in 2020 included across all 10 measure cohorts, with 2,112,904 (61.0%) female patients and a mean (SD) age of 76.5 (8.1) years. Across the 10 measures, 302,379 (14.7%) patients had a claim meeting the overuse criteria in 2019 and 234,481 (13.8%) in 2020. The COVID-19 shutdown had a large impact on overall cohort volume; there were 2,341,017 patients during this period, a decrease of 52.3% from 4,912,453 patients in the corresponding 2019 period. Both overuse volume and rates declined in 2020 compared to 2019. In April 2020, there were 3,955 overuse procedure claims (including spinal fusion/laminectomy, carotid endarterectomy, knee arthroscopy, hysterectomy and vertebroplasty) compared to 14,663 in 2019; an IRR of 0.64 (95% CI 0.62 to 0.67; p <0.001). After the COVID-19 shutdown period, 2020 overuse rates were mostly similar to rates in the corresponding 2019 period. Conclusions and Relevance: The COVID-19 shutdown period during March through May in 2020 had a drastic impact on both the overall volume of patients meeting one of 10 measure criteria and the rate of overuse for these patients. Overuse rates, however, returned to 2019 levels shortly after this shutdown period even as COVID incidence rose. The COVID-19 pandemic has caused substantial and drastic disruption to usual hospital care around the world. During 2020 in the United States, hospitals were inundated with COVID-19 patients and from April to June and this triggered a national stay at home order. Infection among hospital personnel meant staff shortages and many hospitals had temporary bed shortages. 1 While devastating on many levels, the resulting scarcity of hospital resources during this time may have reduced the use of overused or lowvalue services. Overuse involves the use of tests or procedures that risk patient harm beyond the potential patient benefits and increase health care spending without improving health outcomes. 2 Estimates of health care overuse have been published on data prior to the COVID-19 pandemic and have found substantial variation in its use across US regions. 3, 4 In this article, we investigated geographic and time variation in overuse in the US traditional Medicare population during 2020. Previous research has shown that there was a substantial decrease in surgical procedures and hospital admissions in the US during 2020. 5, 6 This decrease occurred even in counties where there were low levels of COVID-19 cases relative to other parts of the country. 6 Surgery volumes vastly decreased compared to 2019 levels during the initial shutdown period from March 2020. 5 We therefore expected to find a decrease in overall patient and service volume. We investigated whether health care providers and patients avoided overuse more than or the same amount as other care during the COVID-19 pandemic. We used a set of 10 overuse measures 7 and examined changes in the rates of overuse and in the volumes of underlying care delivered. 3 WCG Institutional Review Board approved this study. This study followed Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies. We used a 100% sample of Medicare fee for service inpatient and outpatient claims from January 1, 2019 to December 31, 2020. We excluded claims if the beneficiary was younger than 65 at the date of service or did not live in the US, or if claims were from hospitals not in the 50 US states or D.C. We used Centers for Disease Control and Prevention (CDC) weekly COVID-19 case data 8 to distinguish COVID-19 surge periods in states. We defined the initial COVID-19 shutdown period as March 15 to May 2, 2020. This was the same period selected by, 5 which encompassed the period where most states had directives to postpone elective surgical procedures. After this period, states were defined as high COVID-19 incidence if the seven day COVID-19 incidence rate was greater than 175 cases per 100,000 people in the state, and a low COVID-19 incidence otherwise. Like, 6 we selected the threshold of 175 cases per 100,000 population based on the CDC's hot spot standard of seven-day incidence. We included 10 overuse measures, which we provide in supplementary table 1. The overuse measures we used have two components: a denominator (defined by procedure or diagnosis codes) and a numerator (the subset of the denominator cohort meeting the overuse criteria). For example, the denominator cohort for low-value carotid imaging for syncope included all patients with claims where syncope was recorded as a primary diagnosis on the claim and no exclusion diagnosis codes were recorded (supplementary table 1 ). The overuse numerator for this measure was any patient within this cohort with carotid imaging. We included all claims meeting the criteria for at least one of the denominator cohorts. This included claims with the following services: hysterectomy, knee arthroscopy, carotid endarterectomy, coronary stenting, spinal fusion/laminectomy, vertebroplasty; or a claim for a person presenting with syncope or headache (with some exclusions, the full criteria list is in supplementary table 1). We defined two rates for each overuse measure. The overuse rate was defined as above: numerator of the number of inappropriate services divided by its definition denominator. To adjust for changes in the yearover-year volume of the denominator, we calculated a second rate, the size of the denominator cohort itself relative to the 65 and over Medicare beneficiary population. In order to understand the COVID-19 impact on overuse, we compared these two rates in 2020 to their 2019 levels using incidence rate ratios. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) We calculated incidence rate ratios (IRRs) with 95% confidence intervals using Poisson regressions. For each measure, we found the total state beneficiary counts who met the measure cohort definition during each period (month or state COVID-19 period) in 2020 as well as the corresponding period in 2019. In the regressions, we included an offset of the log of the state counts of the number of enrolled fee-forservice Medicare beneficiaries who were 65 or older. We then found the state beneficiary counts during the same periods who met the overuse criteria, and included in these regressions an offset of the definition denominator. The regression standard errors were clustered at the week and state level. The study cohort included 3,464,994 beneficiaries in one of the 10 denominator cohorts; 2,053,792 in 2019 and 1,699,807 in 2020. There were 2,112,904 (61.0%) women and the mean (SD) age was 76.5 (8.1) years. Table 1 is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275006 doi: medRxiv preprint relative to Medicare beneficiary counts (and absolute volumes) returning to 2019 levels after May 2020. For procedure measures, the overuse volume declined in April 2020 compared to April 2019 (14,663 claims in 2019 versus 3,955 in 2020). The overuse rate also declined in April (IRR 0.64 (95% CI 0.62 to 0.67; p <0.001)). After April, the 2020 procedure overuse rates started to return to 2019; in July 2019 there were 13,584 procedure-related claims meeting overuse criteria versus 12,318 in 2020 (IRR 1.00 (95% CI 0.98 to 1.03; p = 0.736)). The overuse rate declined during the shutdown period across most procedure-related measures ( Figure 1 The lack of decline in knee arthroscopy overuse rate was due to the substantial decline in the overall volume of knee arthroscopies (IRR 0.11 (95% CI 0.086 to 0.14; p <0.001)), with only 42 claims in April 2020 meeting the overuse criteria. The number of patients who were seen for syncope declined during the 2020 COVID-19 shutdown period . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) Figure 3 shows the 2020 versus 2019 IRRs for each state and the mean 7-day COVID-19 incidence in that state for each incidence period. We omit head imaging for syncope as the Figure 1 results showed that this overuse measure rate did not change in 2020. As expected based on our previous results, the decline in rates per beneficiary in the procedure denominator cohorts were greatest in states during the 2020 COVID-19 shutdown period compared to other periods in 2020. New York, New Jersey, Massachusetts and Connecticut had both the highest mean 7-day cumulative COVID-19 incidence during this period and the largest decrease in the denominator . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275006 doi: medRxiv preprint cohort rate. The correlation between COVID-19 incidence and procedure denominator cohorts IRRs was -0.66 (95% CI -0.79 to -0.47; p <0.001). During the shutdown period, there was also a decline across all states' procedure overuse rates. There was a negative linear correlation between the COVID-19 incidence rate and procedure overuse rate IRRs; -0.36 (95% CI -0.58 to -0.098; p = 0.009). New Jersey, New York, Connecticut and Wyoming had the largest decrease in procedure overuse rates during the shutdown period. Unlike the others, Wyoming had a low mean 7-day COVID-19 incidence during the shutdown period (13.44 cases per 100,000 population, opposed to New York's 217.36 cases per 100,000 population) After the shutdown period, state procedure overuse IRRs were not significantly correlated with COVID-19 incidence rates: correlation coefficients were -0.022 (95% CI -0.30 to 0.26; p = 0.880) for the high COVID-periods and 0.091 (95% CI -0.19 to 0.36; p = 0.526) for the low COVID periods. Maine and Oregon had the lowest mean 7-day COVID-19 incidence across all states during their high COVID-19 period, yet had some of the largest declines in procedure overuse rates compared to other states. The state imaging denominator cohorts all declined during both the shutdown periods and low COVID-19 periods. These declines were fairly consistent across states, and the state IRRs were negatively correlated with the mean COVID-19 incidence during the shutdown period (-0.30 (95% CI -0.53 to -0.026; p = 0.033)). During high COVID-19 periods, however, there was actually a positive correlation between state denominator IRRs and COVID-19 incidence (0.54 (95% CI 0.31 to 0.71; p <0.001)). Like procedure overuse during the shutdown, there was a negative linear correlation between the imaging overuse IRRs and the state mean COVID-19 incidence (-0.23 (95% CI -0.48 to 0.049; p = 0.104)). This means that the greater the COVID-19 incidence in the state, the greater the decline in imaging overuse rates relative to 2019 during the shutdown period. On the other hand, there was no significant correlation between state imaging overuse IRRs and mean COVID-19 incidence in either low or high COVID-19 periods. The disruption of business as usual at hospitals during the COVID-19 pandemic has served as a natural experiment providing a window on overuse practices. 9 As expected based on previous research, our results showed the claim volume of procedures and certain conditions decreased during the initial COVID-19 pandemic shutdown. There was also a substantial decrease in overuse rates of most services . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275006 doi: medRxiv preprint during the shutdown period. Later in 2020, however, services overuse rates were mostly similar to previous rates prior to the pandemic. Overuse is driven by multiple factors, some of which may be more discretionary than others. 10 The changes in overuse rates observed during the shutdown period may be explained by the government directive to delay non-urgent care. When the federal government declared a national emergency and hospitals were to provide urgent and necessary care only, fewer inappropriate procedures and imaging/diagnostic services were provided. For example, we observed the proportion of coronary stents that were performed on patients with stable coronary disease declined during the shutdown period. This is presumably because these procedures were more likely to be delayed or canceled due to the shutdown compared to stenting for patients with myocardial infarction. The more urgent and evidence-based care may have taken priority. Some services had a smaller or no decline in overuse rates during the shutdown period, including vertebroplasty for osteoporotic fractures, EEGs for headaches and head imaging for syncope. Although there was a decrease in the volume of beneficiaries with claims for these conditions, the proportion of patients receiving an overuse procedure or imaging service remained similar to 2019 levels. There was no evidence that clinical-decision making changed for these services. There was evidence, however, that clinical-decisions for syncope patients did change in the shutdown period. While these patients were just as likely to receive low-value head imaging as they were prior to the pandemic, they were less likely to receive a low value EEG or carotid imaging study. This might reflect the imaging resources available through the shutdown. Clinicians may have delayed or decided against the use of EEG and carotid imaging studies, while head imaging (CT or MRIs) were more accessible through this time. The change in overuse rates of some services were associated with high or low periods of COVID-19 incidence. Hysterectomy, knee arthroscopy, and the four imaging services all had greater declines in overuse rates during periods of high COVID-19 incidence compared to low COVID-19 incidence. Spinal fusion/laminectomy, carotid endarterectomy, coronary stenting and vertebroplasty had no changes in overuse rates in high versus low COVID-19 incidence periods. Again, this points to the discretionary overuse of some services in certain conditions, but not other services. Northeastern states had the most substantial state-specific decline in denominator cohorts and overuse rates during the shutdown period. These states all had the highest COVID-19 case incidence rates during this time, and the highest hospitalization and death rates. 11 COVID-19 hospitalizations and/or death rates may predict impact on hospital capacity (and therefore the denominator and overuse changes) more than . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275006 doi: medRxiv preprint case rates, particularly since in the early stages of the pandemic testing was less available than later in 2020 and cases may have been higher than reported. 12 These overuse measures are indicators of low-value care, and are limited by the diagnostic information available in claims data. We used measures from the Lown Institute overuse metric that had a specific denominator cohort. 7 We did not include the two measures from this metric that have a denominator of all patients at the hospital over the selected time period (inferior vena caval filters and renal stenting). We adjusted for and investigated state-level beneficiary counts and COVID-19 incidence rates. A more granular approach could have used counts at the hospital referral region or county level and explored within and across region differences. COVID-19 incidence at more local regional levels may be more predictive of overuse or patient volumes than at the state level. This could be future research built from this current study, which presents a more high-level overview of specific denominator and numerator overuse measure rates throughout 2020. This study only investigated the impact on services through 2020. By the end of 2020, the US was entering another COVID-19 surge that lasted most of the winter. Later in 2021, the COVID-19 Omicron variant wave caused another surge in cases and hospitalizations. The changes to denominator and overuse rates of these measures may be different during each pandemic stage. For most investigated overuse measures, we observed the largest decrease in overuse rates during the COVID-19 shutdown period. These were larger declines than later in 2020 when COVID-19 case incidence was higher in the US. This suggests that the shutdown, an administrative intervention external to health care delivery, perhaps coupled with the initial uncertainty of the pandemic, had a larger impact on overuse than actual COVID-19 case levels. Notably, the overuse rates had all increased to 2019 levels shortly after the end of the shutdown period despite the ongoing pandemic. Seven out of the ten investigated measures had a significant decline in overuse rates as well as volume during the shutdown period, meaning that some selection or distinction of low-value care was apparent in clinical decision making for these measures compared to the others. The varying results, however, demonstrate that efforts to reduce low value care may require evaluations and interventions that are tailored to each service. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275006 doi: medRxiv preprint Left panel shows the incidence rate ratios (IRRs) for the rate of beneficiaries in each cohort in2020 compared to 2019, while the right panel shows the IRRs the overuse rates for this cohort. *Measures where the cohort is defined as claims for patients with a certain condition are marked with an asterisk (vertebroplasty, EEG for syncope, carotid imaging for syncope, head imaging for syncope and EEG for headache patients). Other measures define the cohort as all patients with the service. IRRs and 95% CIs (error bars) were estimated from Poisson regressions by comparing total beneficiary counts during epidemiological weeks in 2020 with corresponding weeks in 2019. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 26, 2022. ; https://doi.org/10.1101/2022.05.25.22275006 doi: medRxiv preprint Hospital capacities and shortages of healthcare resources among US hospitals during the coronavirus disease 2019 (COVID-19) pandemic, National Healthcare Safety Network (NHSN) Highlighting a Common Quality of Care Delivery Problem: Overuse of Low-value Healthcare Services Measuring low-value care in Medicare Low-Value Care at the Actionable Level of Individual Health Systems Trends in US Surgical Procedures and Health Care System Response to Policies Curtailing Elective Surgical Operations During the COVID-19 Pandemic National Trends In ED Visits, Hospital Admissions, And Mortality For Medicare Patients During The COVID-19 Lown Institute. 2021 Winning Hospitals: Avoiding Overuse. Lown Institute Hospital Index United States COVID-19 Cases and Deaths by State over Time | Data | Centers for Disease Control and Prevention The COVID-19 Pandemic Can Help Us Understand Low-Value Health Care | Health Affairs -6736the author/funder, who has granted medRxiv a license to display the preprint in perpetuity