Insurer Costs of COVID-19 by Disease Severity and Duration

ABSTRACT

Objectives: To analyze US commercial insurance payments associated with COVID-19 as a function of severity and duration of disease.

Study Design: Retrospective database analysis.

Methods: Patients with COVID-19 between April 1, 2020, and June 30, 2021, in the Merative MarketScan Commercial database were identified and stratified as having asymptomatic, mild, moderate (with and without lower respiratory disease), or severe/critical (S/C) disease based on the severity of the acute COVID-19 infection. Duration of disease (DOD) was estimated for all patients. Patients with DOD longer than 12 weeks were defined as having post–COVID-19 condition (PCC). Outcomes were all-cause payments (ACP) and disease-specific payments (DSP) for the entire DOD. Variables included demographic and comorbidities at the time of acute disease. Adjusted payments by disease severity were estimated using generalized linear models (γ distribution with log link).

Results: A total of 738,339 patients were included (374,401 asymptomatic, 156,220 mild, 180,213 moderate, and 27,505 S/C cases). DSP increased from $217 (95% CI, $214-221) for asymptomatic cases to $2744 (95% CI, $2678-$2811) for moderate cases with lower respiratory disease and $28,250 (95% CI, $26,963-$29,538) for S/C cases. ACP increased from $505 (95% CI, $497-$512) for asymptomatic cases to $46,538 (95% CI, $44,096-$48,979) for S/C cases. The DSP and ACP further increased by $50,736 (95% CI, $45,337-$56,136) and $94,839 (95% CI, $88,029-$101,649), respectively, in S/C cases with PCC vs a DOD of fewer than 4 weeks.

Conclusions: COVID-19 payments for S/C cases were more than 10-fold greater than those of moderate cases and further increased by nearly $95,000 in S/C cases with PCC vs a DOD of fewer than 4 weeks.

Am J Manag Care. 2024;30(3):124-129. https://doi.org/10.37765/ajmc.2024.89513

Takeaway Points

• Significant differences in commercial insurance payments were observed between asymptomatic, mild, moderate, and severe/critical COVID-19 cases, with asymptomatic cases incurring a mean of $505 in all-cause health care payments and severe/critical cases averaging all-cause payments of $46,538 per patient.
• Cost of care for severe/critical COVID-19 cases increased significantly with the duration of the disease.
• Severe/critical cases with post–COVID-19 condition, defined as more than 12 weeks of symptoms, incurred additional payments averaging $94,839 compared with severe/critical cases with a disease duration of fewer than 4 weeks.

In the US, the COVID-19 respiratory illness caused by SARS-CoV-2 has affected approximately 94 million patients and caused more than 1 million deaths as of September 7, 2022.¹ The actual direct health care costs of treating COVID-19 from the perspective of the American commercial payers are not well understood.

A characteristic of COVID-19 is its heterogeneous presentation: Individuals with SARS-CoV-2 infection can experience asymptomatic disease or develop symptoms ranging in severity from mild to deadly; therefore, health care costs and utilization vary greatly.² Sheinson et al published a cost-effectiveness framework for COVID-19 treatments for hospitalized patients in the US.³ This study’s findings suggested that bundled payments for patients with COVID-19 who had inpatient admissions without respiratory support or inpatient admissions with nonmechanical respiratory support averaged $8767 and $13,282, respectively, and reached $49,631 for patients requiring mechanical ventilation.³ Another study used a microcosting approach to estimate COVID-19–related costs, individually listing each payment for every visit and care.⁴ Most recently, results of a study evaluating health care costs of patients with outpatient-diagnosed COVID-19 (with and without subsequent COVID-19–related inpatient admission) suggested that patients may incur median costs of $208 for outpatient and $39,187 for inpatient care.⁵ Additional research has been published focusing on inpatient care only or on models of potential costs.^6-9 The actual paid amounts associated with the treatment of COVID-19 in both inpatient and outpatient settings from the perspective of the US private payer—including patient co-pays/coinsurance and deductibles—are not well documented.

Differentiating costs by case severity is key, as some asymptomatic patients may incur no costs at all. In addition, hospitalized patients may present with a primary diagnosis of COVID-19 or may be admitted for other causes and subsequently present with COVID-19. We did not include societal and non–health care costs in this analysis because our study focused on the perspective of the US private insurer. Because co-pay, coinsurance, and deductible represent a proportion of the negotiated rates between insurers and providers and vary significantly across patients based on their insurance plans or prior health care utilization during the same year, those payments were included in our analysis, allowing us to evaluate the total payments for care. Payments in our analysis thus include all monies paid by private insurance companies to health care providers for the treatment of COVID-19, including patient out-of-pocket costs.

The challenges associated with calculating payment for the treatment of COVID-19 are multiple—for example, case identification and cost attribution. Identification of cases was facilitated after April 2020 with the introduction of diagnosis codes specific to COVID-19. For cost attribution, we used 2 approaches: We identified disease-specific payments (DSPs), which were made for services with a diagnosis typically associated with COVID-19, and all-cause payments (ACPs), which include all services provided during the entire duration of the COVID-19 disease. The latter are important outcomes because some treatment associated with COVID-19, such as imaging or laboratory testing, may not have an associated COVID-19 diagnosis. These services would therefore be excluded from a DSP calculation, which would underestimate total cost of care. The use of both DSPs and ACPs thus provides a meaningful range of payments associated with the disease.

METHODS

Data Sources

The Merative MarketScan Commercial Claims and Encounters database was used. This database comprises enrollment information, demographics, and adjudicated inpatient medical, outpatient medical, and outpatient pharmacy claims data collected from more than 300 large self-insured US employers and more than 25 US health plans for individuals younger than 65 years and their dependents.

Study Population

Patients with a COVID-19 International Classification of Diseases, Tenth Edition, Clinical Modification diagnosis code (U071-U072, B34.2, B97.29, J12.81) from April 1, 2020, to June 30, 2021, were identified in the database. For each patient, the date of earliest diagnosis was defined as the index date. When patients had multiple diagnoses of COVID-19, the first instance was used as the index date and index site of care. All patients in the study also required at least 7 months of continuous medical enrollment preindex and 6 months of continuous enrollment post index.

Variable Creation

Patient demographics (age, sex, and geographic location within the US) at index date were captured. Thirty-one patient comorbidities as defined by Elixhauser, as well as the Elixhauser Comorbidity Index (ECI) score, were captured for all patients.^10,11 The ECI was selected instead of the Charlson Comorbidity Index because recent research suggests that the former might be more predictive of frailty and mortality than the latter.^12,13 For the ECI calculation, 7 to 12 months of data preindex were analyzed up to 1 month prior to index (to avoid including morbidities related to the COVID-19 disease).

Duration of Disease, Ongoing Symptomatic COVID-19, and Post–COVID-19 Condition

Duration of disease (DOD) was determined based on the presence of any 1 COVID-19 sign or symptom (CSS) or a COVID-19 diagnosis as per the codes listed earlier. A complete list of CSSs and corresponding diagnosis codes are shown in eAppendix Table 1 (eAppendix available at ajmc.com). For each patient with COVID-19, CSS diagnoses were identified starting 5 days before the COVID-19 code. Diagnoses listed prior to the index date were used because many patients presented with CSSs before being tested for and given a diagnosis of COVID-19. (For many patients with mild and moderate disease, most of the health care use happened prior to index.) COVID-19 was considered ongoing as long as the gap between distinct CSSs or COVID-19 diagnoses was shorter than 35 days. When a new diagnosis was identified after a 35-day gap, it was assumed that this might be a new condition. This rule was used because most CSSs are very generic in nature (eg, cough, headache), so their potential association with index COVID-19 may decline as the time gap in diagnosis dates increases. The decision to use a 35-day gap (vs a shorter or longer gap) was based on review of histograms of time gaps for CSS conditions for all patients. For patients with mild or moderate COVID-19 with no health care events in the database after their COVID-19 index code, it was assumed that they would have a DOD of 9 days following the COVID-19 diagnosis. The minimum DOD of any symptomatic patient in our study was 14 days. This was based on prior publications on disease duration.^14,15 Patients with no CSS diagnoses in the 5 days prior to and up to 35 days after index were considered asymptomatic, and for the purpose of calculating payments, their DOD was the single day when testing was performed. Ongoing symptomatic COVID-19 was defined as a DOD greater than 4 weeks (≥ 28 days) but shorter than or equal to 12 weeks (≤ 84 days). Post–COVID-19 condition (PCC) was defined as a DOD greater than 12 weeks (> 84 days). These cutoffs were based on definitions from the United Kingdom National Institute for Health and Care Excellence, which to our knowledge were the first published guidelines for characterization of potential “long COVID.”¹⁶

Disease Severity

The classification of mild, moderate, and severe/critical (S/C) disease was made based on the Janssen trial definitions and the presence of CSS diagnoses. This classification was used for Janssen’s phase 3 ENSEMBLE clinical trial (NCT04505722) COVID-19 vaccine trial and described in protocols that were thoroughly reviewed and approved by the FDA and the Operation Warp Speed Team of Vaccine Phase 3 Clinical Trial Studies—hence the decision to use these definitions here. These definitions were also described in the trial publications.^17,18 eAppendix Table 2 and eAppendix Table 3 show the criteria for each severity classification and how these criteria link to CSSs and thus code lists (as per eAppendix Table 1).^19,20 Patients with mild disease were characterized as having only 1 symptom, as listed in eAppendix Table 2. Patients with moderate disease presented with either 2 symptoms or 1 symptom indicative of pneumonia or lower respiratory disease. Further stratification of the population with moderate disease was performed to create a group of moderate cases without lower respiratory disease (typically considered “mild,” as per the National Institutes of Health [NIH] definition²⁰ and as shown in the eAppendix) and another group of moderate cases with lower respiratory disease. Patients with S/C disease experienced at minimum severe respiratory conditions, such as acute respiratory distress or failure, along with other symptoms. All hospitalized patients had either moderate or S/C disease. All patients admitted to the intensive care unit or on ventilators were categorized as having S/C disease.

Health Care Utilization and Payments

All payments for care in the inpatient or outpatient settings were analyzed for the entire DOD as defined earlier. Payments for visits or admissions with a CSS diagnosis were defined as a DSP so that 2 different total payments could be analyzed: an ACP, including all care encountered during the duration of the disease, and a DSP, a subset of ACP, including only care with CSS diagnoses.

Statistical Analyses

Descriptive statistics were reported for all study variables as means and SDs for continuous variables and as frequencies and percentages for categorical variables. ACP and DSP by severity type were analyzed using generalized linear models with γ distribution and log links. Predictive models for DSP were developed. All demographics, comorbidities, and patient CSSs were used as features in the model development. Target was set as DSP to avoid including non–COVID-19 payments in the models. Models were developed using DataRobot 6.0.0. DataRobot is a machine learning platform that allows simultaneous use of a large library of models; it also allows automated ensembling of different distinct models to maximize predictive accuracy. A detailed explanation of the DataRobot platform has previously been published.^21-23

RESULTS

A total of 738,339 patients with a COVID-19 diagnosis or a positive COVID-19 test result were analyzed; this included 374,401 asymptomatic, 156,220 mild, 180,213 moderate (85,232 without and 94,981 with lower respiratory disease), and 27,505 S/C cases. Of the 156,220 mild cases, 129 had ongoing disease on November 30, 2021, with a mean (SD) DOD of 287.9 (74.8) days. Of the 180,213 moderate and 27,505 S/C cases, 575 moderate and 827 S/C cases had ongoing disease (mean [SD] DOD, 290.4 [85.7] and 282.2 [85.9] days, respectively). Key demographic, comorbidity, and disease duration findings are shown in Table 1. As expected, age and comorbidities increased with disease severity (mean [SD] age, 38.0 [16.0] years across all severity classes and 36.2 [16.1], 36.7 [16.2], 41.1 [15.0], and 50.3 [11.0] in the asymptomatic, mild, moderate, and S/C populations, respectively; mean [SD] ECI score, 0.70 [1.19] across all severity classes and 0.54 [0.99], 0.63 [1.07], 0.94 [1.36], and 1.75 [2.04] in the asymptomatic, mild, moderate, and S/C populations, respectively). The proportion of patients with an ECI score of 5 or greater also increased with disease severity, from 0.7% of the asymptomatic population to 9.1% of the S/C population. Patients with mild disease experienced COVID-19 for a mean of 19 days vs 57 days for patients with S/C disease. Female patients represented 53%, 55%, 57%, and 44% of the asymptomatic, mild, moderate, and S/C cohorts, respectively.

Across the entire cohort, the mean (SD) unadjusted total DSP and ACP for the treatment of COVID-19 for all patients were $1713 ($17,031) and $3468 ($27,930), respectively. Table 2 shows adjusted DSP and ACP amounts for each severity. Mean DSP increased from $217 (95% CI, $214-$221) for asymptomatic cases to $28,250 (95% CI, $26,963-$29,538) for S/C cases (incremental payments from asymptomatic to mild, moderate without lower respiratory disease, moderate with lower respiratory disease, and S/C: $197 [95% CI, $189-$206], $515 [95% CI, $496-$534], $2527 [95% CI, $2461-$2594], and $28,033 [95% CI, $26,746-$29,321], respectively). ACP increased from $505 (95% CI, $497-$512) for asymptomatic cases to $46,538 (95% CI, $44,096-$48,979) for S/C cases.

Within the S/C cohort, the mean (SD) unadjusted DSP and ACP for the treatment of COVID-19, regardless of DOD, were $30,079 ($81,332) and $56,277 ($128,561), respectively. Table 3 shows the adjusted payments for patients with S/C disease as a function of DOD: fewer than 4 weeks, 4 to 12 weeks (also described as ongoing symptomatic COVID-19), and more than 12 weeks (PCC). When comparing patients with a DOD fewer than 4 weeks vs those with a DOD more than 12 weeks, DSP and ACP increased by $50,736 (95% CI, $45,337-$56,136) and $94,839 (95% CI, $88,029-$101,649), respectively.

The most accurate predictive model evaluating patient characteristics (including diagnoses and treatments) associated with increased ACP was a blended generalized linear model. Not surprisingly, the most predictive features associated with increased payments were sepsis and ventilator use, followed by repeat inpatient admission and acute respiratory failure. The US region where the patient was treated was the fifth predictor of cost. Other predictors included presence of pneumonia and acute respiratory distress syndrome.

DISCUSSION

Our study was designed to evaluate payments associated with COVID-19 from the US payer perspective based on disease severity. Asymptomatic, mild, and moderate cases of COVID-19 without lower respiratory disease required ACP of less than $2000 per person, whereas patients with moderate COVID-19 with lower respiratory disease incurred mean costs exceeding $5000. For patients with S/C disease, DSP ranged from $17,333 to $68,069 and ACP from $31,255 to $126,094 based on DOD.

Our analysis was based on disease classification (asymptomatic, mild, moderate, and S/C), with the mild, moderate, and S/C characterizations defined by the Janssen phase 3 ENSEMBLE study protocol. All CSSs, as captured in the diagnosis fields of claims databases, were analyzed, and algorithms of severity were defined based on the Janssen trial disease severity definitions. These definitions are comparable to those suggested by the NIH (see eAppendix Table 3) but provide greater detail in terms of all diagnoses that make a mild, moderate, or S/C case, so they allow for a potentially better characterization of patients. We further stratified the population with moderate disease into 2 groups: patients without and patients with lower respiratory disease. Patients in the moderate group without lower respiratory disease are classified as having mild disease by the NIH, whereas those with lower respiratory disease meet the NIH criteria for moderate disease. As expected, increasing severity was associated with increased comorbidities, male sex (vs female), and increased age, which are characteristics shown in other studies to be associated with increasing risks for severe COVID-19.^24-27 In addition, payments increased in a consistent manner with increased COVID-19 severity category, further highlighting the fact that our severity characterization was representative of increased clinical distress.

Two payment types, DSP and ACP, were evaluated in our study. DSPs were those associated with at least 1 diagnosis of a sign or symptom of COVID-19 or with COVID-19 testing, and they were potentially underestimates of total payments because they would not include related health care utilization (eg, testing, follow-up visits) that did not specifically mention those diagnoses. Because ACP did include all care received during the duration of the disease (or day of testing for asymptomatic cases), it may have included some unrelated emergency care, which was unavoidable during COVID-19; however, it is important to keep in mind that nonessential, nonurgent care during acute COVID-19 episodes was mostly postponed. It is therefore unlikely that significant unrelated, unessential care was provided during a patient’s COVID-19 disease. In most US states, elective, expensive treatments (such as hip or knee arthroplasty) were postponed during the peak of the pandemic. Afterward, many states still required patients to present with a negative COVID-19 test to undergo such procedures. Therefore, the true payments associated with COVID-19 may be closer to that of the ACP than the DSP.²⁸

Our findings are similar to those reported by Sheinson et al (who calculated $8767 for inpatient care without oxygen, $13,282 for inpatient care with oxygen, and $49,631 for inpatient care with mechanical ventilation)³ and those suggested by Scott et al, who found median payments of $39,187 for patients requiring inpatient care.⁵ Sheinson et al and Scott et al both used a bottom-up approach to evaluate cost.^3,5 In contrast, our study used all reported payments for analyses of both DSP and ACP. Importantly, at the time of our study, 129 mild, 575 moderate, and 827 S/C cases had ongoing disease, meaning that they would continue to incur payments; therefore, our analysis may show greater payments with longer follow-up time. (For those ongoing cases, the mean follow-up time was greater than 8 months, or 288, 290, and 282 days for patients with mild, moderate, and S/C disease, respectively.)

Limitations

Our study has the following limitations. Our analysis included data from late 2020 to early 2021, describing outcomes with a COVID-19 variant that may not be dominant anymore. However, our focus on disease severity and clinical presentation and our analysis of health care payments based thereof may provide a framework for understanding COVID-19 payments regardless of variant, simply based on patient disease severity. Our analysis period began in April 2020, when COVID-19 started to be identified in claims with specific codes, but it is possible that some patients with COVID-19 after April 2020 were miscoded and not represented herein. New COVID-19 codes introduced in 2021 were possibly missed, thus excluding some additional patients. An additional limitation is the lack of validation of diagnosis codes as found in claims databases. However, false diagnoses on claims constitute punishable fraud and, although possible, are unlikely on a large scale. The severity of disease (mild, moderate, and severe) was based on definitions that were thoroughly reviewed as part of a clinical trial but not validated. All signs and symptoms may not be exhaustively captured in this data set, as might be the case in a prospective clinical trial. This limitation may have affected our ability to accurately characterize some patients’ disease severity. However, severity of disease was clearly associated with comorbidities and overall payments. Finally, a key limitation of our research was the determination of disease duration. A major assumption in our analysis was the maximum 35-day gap allowed between diagnoses or care associated with CSSs. Therefore, our analysis is possibly conservative, as patients returning for care due to PCC after a break of more than 35 days would not be characterized in our research as having continued COVID-19 care. An additional limitation may be related to the generalizability of our findings because all payments were obtained from US-based commercial payers.

CONCLUSIONS

Findings from our study of nearly 740,000 patients with COVID-19 highlight the heterogeneity in payments and show that payments for COVID-19 increased substantially with increasing disease severity and duration, thus reinforcing the economic importance of reducing the severity of COVID-19 through available technologies, including vaccines.

Acknowledgments

The authors wish to acknowledge Lilit Hovhannisyan, MD, for editorial support.

Author Affiliations: Epidemiology and Real-World Data Sciences, Johnson & Johnson MedTech (CEH, JWR, PMC), New Brunswick, NJ; Janssen Global Services (BJP, FR, ACEK), Raritan, NJ; Mu Sigma (RD), Bangalore, India; Janssen Scientific Affairs (JKD, BB), Titusville, NJ.

Source of Funding: This study was funded by Johnson & Johnson.

Author Disclosures: Drs Holy, Patterson, Richards, El Khoury, DeMartino, and Coplan, Ms Ruppenkamp, and Mr Bookhart report being employees of Johnson & Johnson, which manufactures a COVID-19 vaccine. Drs Holy, Patterson, El Khoury, DeMartino, and Coplan and Mr Bookhart also report owning stock in Johnson & Johnson. Ms Debnath reports working as an analyst for Mu Sigma, paid for by Johnson & Johnson.

Authorship Information: Concept and design (CEH, BJP, JWR, FR, ACEK, JKD, BB, PMC); acquisition of data (JWR); analysis and interpretation of data (CEH, BJP, JWR, RD, JKD, PMC); drafting of the manuscript (CEH, ACEK, PMC); critical revision of the manuscript for important intellectual content (CEH, BJP, JWR, FR, RD, ACEK, JKD, BB, PMC); statistical analysis (CEH, RD, PMC); administrative, technical, or logistic support (BB); and supervision (CEH, BJP, FR, ACEK, BB, PMC).

Address Correspondence to: Chantal E. Holy, PhD, MSc, Johnson & Johnson, 410 George St, New Brunswick, NJ 08901. Email: choly1@its.jnj.com.

REFERENCES

1. Post COVID-19 condition (long COVID). News release. World Health Organization. October 16, 2021. Accessed April 1, 2022. https://www.who.int/srilanka/news/detail/16-10-2021-post-covid-19-condition

2. Taleghani N, Taghipour F. Diagnosis of COVID-19 for controlling the pandemic: a review of the state-of-the-art. Biosens Bioelectron. 2021;174:112830. doi:10.1016/j.bios.2020.112830

3. Sheinson D, Dang J, Shah A, Meng Y, Elsea D, Kowal S. A cost-effectiveness framework for COVID-19 treatments for hospitalized patients in the United States. Adv Ther. 2021;38(4):1811-1831. doi:10.1007/s12325-021-01654-5

4. Padula WV, Malaviya S, Reid NM, et al. Economic value of vaccines to address the COVID-19 pandemic: a U.S. cost-effectiveness and budget impact analysis. J Med Econ. 2021;24(1):1060-1069. doi:10.1080/13696998.2021.1965732

5. Scott A, Chambers R, Reimbaeva M, et al. Real-world retrospective analysis of patient characteristics, healthcare resource utilization, costs, and treatment patterns among unvaccinated adults with COVID-19 diagnosed in outpatient settings in the United States. J Med Econ. 2022;25(1):287-298. doi:10.1080/13696998.2022.2037917

6. Bartsch SM, Ferguson MC, McKinnell JA, et al. The potential health care costs and resource use associated with COVID-19 in the United States. Health Aff (Millwood). 2020;39(6):927-935. doi:10.1377/hlthaff.2020.00426

7. Di Fusco M, Shea KM, Lin J, et al. Health outcomes and economic burden of hospitalized COVID-19 patients in the United States. J Med Econ. 2021;24(1):308-317. doi:10.1080/13696998.2021.1886109

8. Ohsfeldt RL, Choong CK, Mc Collam PL, Abedtash H, Kelton KA, Burge R. Inpatient hospital costs for COVID-19 patients in the United States. Adv Ther. 2021;38(11):5557-5595. doi:10.1007/s12325-021-01887-4

9. Wager E, Claxton G, Amin K, Cox C. Cost of COVID-19 hospital admissions among people with private health coverage. Peterson-KFF Health System Tracker. November 14, 2022. Accessed March 1, 2024. https://bit.ly/48jJRM2

10. Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8-27. doi:10.1097/00005650-199801000-00004

11. Menendez ME, Neuhaus V, van Dijk CN, Ring D. The Elixhauser comorbidity method outperforms the Charlson index in predicting inpatient death after orthopaedic surgery. Clin Orthop Relat Res. 2014;472(9):2878-2886. doi:10.1007/s11999-014-3686-7

12. Tsai KY, Hsieh KY, Ou SY, Chou FH, Chou YM. Comparison of Elixhauser and Charlson methods for discriminative performance in mortality risk in patients with schizophrenic disorders. Int J Environ Res Public Health. 2020;17(7):2450. doi:10.3390/ijerph17072450

13. Sharma N, Schwendimann R, Endrich O, Ausserhofer D, Simon M. Comparing Charlson and Elixhauser comorbidity indices with different weightings to predict in-hospital mortality: an analysis of national inpatient data. BMC Health Serv Res. 2021;21(1):13. doi:10.1186/s12913-020-05999-5

14. Polak SB, Van Gool IC, Cohen D, von der Thüsen JH, van Paassen J. A systematic review of pathological findings in COVID-19: a pathophysiological timeline and possible mechanisms of disease progression. Mod Pathol. 2020;33(11):2128-2138. doi:10.1038/s41379-020-0603-3

15. Tenforde MW, Kim SS, Lindsell CJ, et al; IVY Network Investigators; CDC COVID-19 Response Team. Symptom duration and risk factors for delayed return to usual health among outpatients with COVID-19 in a multistate health care systems network - United States, March-June 2020. MMWR Morb Mortal Wkly Rep. 2020;69(30):993-998. doi:10.15585/mmwr.mm6930e1

16. COVID-19 Rapid Guideline: Managing the Long-Term Effects of COVID-19. National Institute for Health and Care Excellence. December 18, 2020. Updated January 25, 2024. Accessed January 23, 2023. https://www.nice.org.uk/guidance/NG188

17. Sadoff J, Gray G, Vandebosch A, et al; ENSEMBLE Study Group. Safety and efficacy of single-dose Ad26.COV2.S vaccine against Covid-19. N Engl J Med. 2021;384(23):2187-2201. doi:10.1056/NEJMoa2101544

18. Sadoff J, Gray G, Vandebosch A, et al; ENSEMBLE Study Group. Final analysis of efficacy and safety of single-dose Ad26.COV2.S. N Engl J Med. 2022;386(9):847-860. doi:10.1056/NEJMoa2117608

19. A Randomized, Double-Blind, Placebo-Controlled Phase 3 Study to Assess the Efficacy and Safety of Ad26.COV2.S for the Prevention of SARS-CoV-2-Mediated COVID-19 in Adults Aged 18 Years and Older: ENSEMBLE. Janssen Vaccines & Prevention BV. December 14, 2020. Accessed March 1, 2024. https://www.jnj.com/coronavirus/covid-19-phase-3-study-clinical-protocol

20. Clinical spectrum of SARS-CoV-2 infection. National Institutes of Health. Updated March 6, 2023. Accessed March 1, 2024. https://www.covid19treatmentguidelines.nih.gov/overview/clinical-spectrum/

21. Muhlestein WE, Akagi DS, Kallos JA, et al. Using a guided machine learning ensemble model to predict discharge disposition following meningioma resection. J Neurol Surg B Skull Base. 2018;79(2):123-130. doi:10.1055/s-0037-1604393

22. Muhlestein WE, Akagi DS, McManus AR, Chambless LB. Machine learning ensemble models predict total charges and drivers of cost for transsphenoidal surgery for pituitary tumor. J Neurosurg. 2018;131(2):507-516. doi:10.3171/2018.4.JNS18306

23. Tsuzuki S, Fujitsuka N, Horiuchi K, et al. Factors associated with sufficient knowledge of antibiotics and antimicrobial resistance in the Japanese general population. Sci Rep. 2020;10(1):3502. doi:10.1038/s41598-020-60444-1

24. Fathi M, Vakili K, Sayehmiri F, et al. The prognostic value of comorbidity for the severity of COVID-19: a systematic review and meta-analysis study. PLoS One. 2021;16(2):e0246190. doi:10.1371/journal.pone.0246190

25. Marik PE, DePerrior SE, Ahmad Q, Dodani S. Gender-based disparities in COVID-19 patient outcomes. J Investig Med. 2021;69(4):814-818. doi:10.1136/jim-2020-001641

26. Raimondi A, Fucà G, Leone AG, et al. Impact of age and gender on the efficacy and safety of upfront therapy with panitumumab plus FOLFOX followed by panitumumab-based maintenance: a pre-specified subgroup analysis of the Valentino study. ESMO Open. 2021;6(5):100246. doi:10.1016/j.esmoop.2021.100246

27. Ryan C, Minc A, Caceres J, et al. Predicting severe outcomes in Covid-19 related illness using only patient demographics, comorbidities and symptoms. Am J Emerg Med. 2021;45:378-384. doi:10.1016/j.ajem.2020.09.017

28. Brown TS, Bedard NA, Rojas EO, et al; AAHKS Research Committee. The effect of the COVID-19 pandemic on hip and knee arthroplasty patients in the United States: a multicenter update to the previous survey. Arthroplast Today. 2021;7:268-272. doi:10.1016/j.artd.2020.11.025