Health Care Utilization and Costs Associated With Blue-Light vs White-Light Cystoscopy: A Real-World Bladder Cancer Population

Abstract

Purpose:

Compared with white-light cystoscopy (WLC), blue-light cystoscopy (BLC) enhances detection of non–muscle-invasive bladder cancer (NMIBC), especially carcinoma in situ (CIS). However, its effects on health care resource utilization and cost burden have not been fully elucidated. This study assessed the economic implications of BLC vs WLC.

Materials and Methods:

We conducted a retrospective cohort analysis with claims data from the Optum Research Database between June 2011 and May 2023. Patients who underwent BLC or WLC were identified, and white-light patients were matched 6:1 to blue-light patients by index year and time between bladder cancer diagnosis and cystoscopy. Inverse probability of treatment weighting was used to adjust for baseline differences. Health care resource utilization and total and bladder cancer–specific costs were calculated on a per-patient-per-month basis.

Results:

The final cohort included 794 blue-light and 4764 white-light patients. Before weighting, claims coded for CIS were more frequent in blue-light patients (19.6% vs 8.8%; P < .001). Blue-light patients underwent more upper-tract imaging (71.8% vs 57.8%; P < .001), BCG (24.2% vs 15.8%; P < .001), and biomarker testing (65.0% vs 34.9%; P < .001). After weighting, blue-light patients had a higher mean number of bladder cancer–related ambulatory visits per month (1.3 vs 1.0; P = .004). However, neither all-cause health care resource utilization (3.47 vs 3.21; P = .11) nor total cost ($2987.93 vs $2886.16, per patient per month, P = .65) differed significantly between cohorts.

Conclusions:

BLC is associated with greater health care resource utilization for NMIBC without significantly elevating health care costs, suggesting that enhanced diagnostic strategies can be implemented without added financial burden in real-world practice.

INTRODUCTION

Bladder cancer is one of the most commonly diagnosed cancers worldwide, ranking among the top 10 cancers in terms of incidence and contributing substantially to morbidity and mortality.¹ Most cases are urothelial carcinoma, and approximately 75% of these are classified as non–muscle-invasive bladder cancer (NMIBC).² NMIBC has a high recurrence rate (estimated at 50%-70%) and is a heterogeneous disease with varying risk of progression to muscle-invasive bladder cancer.³ Although 5-year survival rates for NMIBC exceed 90%, the need for long-term surveillance and repeated intervention imposes a substantial burden on patients and the health care system. Indeed, bladder cancer is the most expensive cancer to manage, largely because of the need for intensive monitoring and treatment.⁴

Cystoscopic surveillance is the cornerstone of NMIBC management, and white-light cystoscopy (WLC) is the diagnostic standard. WLC has high sensitivity for detecting papillary tumors but is less effective at identifying carcinoma in situ (CIS). Therefore, it may yield false-negative results. To address this limitation, blue-light cystoscopy (BLC) with hexaminolevulinate hydrochloride has emerged as a promising alternative, with extensive evidence showing its superiority for detecting tumors that WLC may miss, including CIS and small papillary tumors,^5-9 and for this reason, it is recommended on various guidelines.^10,11 Heightened detection may lead to enhanced tumor staging and allow for appropriate guideline-endorsed therapy. However, evidence regarding the impact of BLC on oncologic outcomes such as recurrence and progression has been variable across studies and remains an area of active investigation.^12-14 Accordingly, understanding the economic implications of BLC in routine practice is essential, independent of its effects on disease control. In contrast to the extensive body of research evaluating the clinical efficacy of BLC, its economic and health care resource implications have not been broadly evaluated in real-world settings.

This study aimed to examine the health care resource utilization (HCRU) and costs associated with BLC vs WLC comparatively. Using a retrospective cohort design, we analyzed claims data to evaluate whether enhanced diagnostic accuracy is associated with differences in health care resource utilization and costs in routine clinical practice. This study provides a more comprehensive understanding of the economic effect of BLC and its value in optimizing NMIBC management.

MATERIALS AND METHODS

Study Population and Design

This retrospective cohort analysis used claims data from the Optum Research Database to compare HCRU and costs for patients undergoing BLC vs WLC. This study was reviewed by the Mayo Clinic Institutional Review Board and was deemed exempt from full board review because of its use of deidentified administrative claims data. The requirement for written informed consent was waived because the study involved secondary analysis of deidentified data. For this study (Figure 1), we included patients who were at least 18 years old; had evidence of at least 1 BLC (with hexaminolevulinate hydrochloride) or WLC procedure from June 1, 2011, through May 31, 2023 (termed the identification period); had at least 2 nondiagnostic claims for bladder cancer, separated by at least 30 days, during the full study period (January 1, 2011, through November 30, 2023); and had continuous enrollment in any health plans within the Optum Research Database for at least 6 months before (baseline period) and after (follow-up period) their bladder cancer diagnosis as well as 6 months of enrollment before and after their index cystoscopy (for BLC patients only). Patients were excluded if they had evidence of pregnancy, a diagnosis of another primary cancer, missing demographic information, or if they had participated in a clinical trial.

Figure 1.

Flowchart of patient selection.

The index date for the BLC cohort was the earliest medical claim for BLC during the identification period. BLC procedures were identifiable beginning in 2018, corresponding to the introduction of a specific CPT code. WLC patients were then matched to BLC patients on the basis of the year of the cystoscopy and the time between index cystoscopy and bladder cancer diagnosis (categorized into 30-day intervals). This match was performed for the purpose of comparing outcomes between BLC and WLC patients; cystoscopy year and time between index cystoscopy and bladder cancer diagnosis were used in lieu of cancer stage and other clinical disease information, which was not available in the claims data. If the date of bladder cancer diagnosis was the same as the index date, then the index date was used. A pairing of the 30-day intervals and same year of cystoscopy was used to match 1 BLC patient to 10 WLC patients. A subset of WLC patients was then randomly selected to be used in the final cohort at a 1:6 match ratio. Patients in the WLC cohort who had undergone BLC at any point during the study period were excluded to maintain mutually exclusive groups for analysis.

Weighting

Inverse probability of treatment weighting (IPTW) was used to address imbalances of baseline characteristics that remained between the cohorts after matching. Model covariates included demographic characteristics such as age, sex, race, geographic region, and insurance type; baseline clinical characteristics, including nondiagnostic medical claims for CIS of bladder or malignant neoplasm of bladder (Supplemental Table, http://links.lww.com/JU9/A192), BCG vaccine use, upper-tract imaging (including intravenous pyelography, retrograde pyelography, kidney ultrasonography, abdominal and pelvic computed tomography, and magnetic resonance imaging), urinary biomarkers (including urine cytologic testing or other urine biomarker assessment), transurethral resection of bladder tumor, cystectomy, and intravesical therapy; baseline comorbid conditions, including the 10 most frequent conditions within the study population as listed by the Agency for Healthcare Research and Quality and the Charlson comorbidity index¹⁵ score; and baseline HCRU and costs for ambulatory visits (office visits and outpatient hospital visits), emergency department (ED) visits, inpatient hospitalizations, bladder cancer–specific utilization, and total health care expenditures. After IPTW adjustment, most baseline characteristics were balanced between the BLC and WLC cohorts, although small imbalances remained for the proportion of patients with CIS claims, geographic region (Northeast and Midwest), and ambulatory health care utilization.

Outcomes and Statistical Analysis

The primary outcomes of interest were HCRU and total costs associated with BLC vs WLC during the follow-up period, which began 90 days after the index cystoscopy. Utilization was measured with per-patient-per-month (PPPM) metrics for ambulatory visits, ED visits, inpatient hospitalizations, and pharmacy fills. Health care costs were calculated as inflation-adjusted 2023 US dollars by using the Consumer Price Index,¹⁶ and reported PPPM, for all medical care costs. Costs reported represented combined payments made by the health care plan and the patient to providers. Cost categories included total medical costs (which included physician office visits, hospital outpatient visits, inpatient care, and ED visits), pharmacy costs, and total health care expenditures. Ambulatory utilization and costs include the combined sum of both office visits and outpatient visits. Bladder cancer–related HCRU and costs were defined as any nonhospital claim with a primary or secondary diagnosis of bladder cancer, inpatient hospital stay with a primary claim for bladder cancer or claim for a bladder cancer–specific treatment such as chemotherapy.

All statistical analyses were performed by using SAS version 9.4 (SAS Institute, Inc). Ordinary least squares regression with robust standard errors was used to analyze continuous outcomes; the Wald test was used to evaluate the significance of regression coefficients. The Rao-Scott test was used to analyze binary measures. A 2-tailed P value < .05 was considered statistically significant.

RESULTS

The final analytic cohort comprised 794 BLC patients and 4764 matched WLC patients (Figure 1). Among BLC patients, 785 (98.9%) also underwent WLC on the index date. The mean age was 73 years, and the cohort was predominantly male (75%) (Table 1). Before IPTW, significantly more BLC than WLC patients had CIS (ICD9 dx code 233.7, ICD10 dx code D090) claims during the baseline (4.7% vs 1.5%) and study (19.6% vs 8.8%) periods (P < .001 for both) (Table 2). In addition, during the baseline period, a higher proportion of BLC patients than WLC patients underwent upper-tract imaging (71.8% vs 57.8%; P < .001) and cytologic biomarker testing (65.0% vs 34.9%; P < .001). After IPTW, differences in baseline proportions between BLC and WLC were no longer significant for CIS (3.4% vs 2.0%), upper-tract imaging (61.8% vs 59.9%), and cytologic biomarker testing (42.4% vs 39.4%). However, the proportion of BLC vs WLC patients with CIS claims during the study period remained significantly different after IPTW (18.2% vs 9.4%; P < .001).

Table 1.

Demographic Characteristics^a

Characteristic	Unweighted			Weightedb
	BLC (n = 794)	WLC (n = 4764)	P value	BLC (n = 794)	WLC (n = 4764)	P value
Age, y	72.6 (10.1)	73.9 (9.1)	< .001	73.7 (9.4)	73.7 (9.2)	.90
Sex			.89			.50
Female	194 (24.4)	1173 (24.6)		210 (26.4)	1170 (24.6)
Male	600 (75.6)	3591 (75.4)		584 (73.6)	3594 (75.4)
Race/ethnicity
White	605 (76.2)	3750 (78.7)	.07	614 (77.3)	3737 (78.4)	.68
Black	65 (8.2)	340 (7.1)	.27	69 (8.7)	347 (7.3)	.51
Hispanic	66 (8.3)	348 (7.3)	.33	62 (7.8)	352 (7.4)	.84
Asian	22 (2.8)	82 (1.7)	.03	13 (1.6)	89 (1.9)	.56
Unknown	22 (2.8)	119 (2.5)	.64	24 (3.0)	118 (2.5)	.50
Missing	14 (1.8)	125 (2.6)	.18	12 (1.5)	121 (2.5)	.12
Region
Northeast	200 (25.2)	882 (18.5)	< .001	194 (24.4)	933 (19.6)	.06
Midwest	186 (23.4)	1372 (28.8)	.006	185 (23.3)	1332 (28.0)	.06
South	283 (35.6)	1947 (40.9)	.004	323 (40.7)	1908 (40.1)	.82
West	125 (15.7)	560 (11.8)	.002	93 (11.7)	588 (12.3)	.61
Otherc	0	< 5	NA	0	< 5	NA
Insurance type			.006			.89
Commercial	160 (20.2)	779 (16.4)		137 (17.3)	807 (16.9)
Medicare	634 (79.8)	3985 (83.6)		657 (82.7)	3957 (83.1)
Index year			NA			NA
2018	23 (2.9)	138 (2.9)		20 (2.5)	139 (2.9)
2019	143 (18.0)	858 (18.0)		121 (15.2)	865 (18.2)
2020	173 (21.8)	1038 (21.8)		185 (23.3)	1041 (21.9)
2021	175 (22.0)	1050 (22.0)		167 (21.0)	1042 (21.9)
2022	206 (25.9)	1236 (25.9)		237 (29.8)	1230 (25.8)
2023	74 (9.3)	444 (9.3)		64 (8.1)	447 (9.4)

Abbreviations: BLC, blue-light cystoscopy; CIS, carcinoma in situ; NA, not applicable; WLC, white-light cystoscopy.

^aCategorical data are expressed as No. of patients (%), and continuous data are expressed as mean (SD).

^bBecause weighted values included decimal places, rounded values may not add up to the total number of patients.

^cIncludes Armed Forces, American Samoa, Federated State of Micronesia, Guam, Marshall Islands, Commonwealth of the Northern Mariana Islands, Puerto Rico, Palau, and the Virgin Islands.

Table 2.

Clinical Characteristics^a

Characteristic	Unweighted			Weightedb
	BLC (n = 794)	WLC (n = 4764)	P value	BLC (n = 794)	WLC (n = 4764)	P value
Time with bladder cancer diagnosis before index cystoscopy, d	374.4 (569.9)	374.6 (570.1)	.42	399.9 (596.5)	365.8 (557.6)	.39
Baseline CIS claims	37 (4.7)	72 (1.5)	< .001	27 (3.4)	96 (2.0)	.22
Study period CIS claims	156 (19.6)	417 (8.8)	< .001	144 (18.1)	449 (9.4)	< .001
Baseline Charlson Comorbidity Index score	2.7 (1.6)	2.6 (1.8)	.25	2.8 (1.8)	2.6 (1.8)	.29
Patients undergoing WLC during baseline	722 (90.9)	3234 (67.9)	< .001	578 (72.8)	3393 (71.2)	.66
Unique WLC procedures performed during baseline	1.6 (0.8)	1.6 (0.8)	.02	1.6 (0.8)	1.6 (0.8)	.13
Baseline cystectomy	16 (2.0)	161 (3.4)	.05	23 (2.9)	152 (3.2)	.70
Patients undergoing BCG vaccine use during baseline	192 (24.2)	753 (15.8)	< .001	144 (18.1)	810 (17.0)	.53
Baseline BCG vaccine treatments	1.1 (2.2)	0.7 (1.9)	< .001	0.8 (2.0)	0.8 (1.9)	.50
Baseline upper-tract imaging	570 (71.8)	2753 (57.8)	< .001	491 (61.8)	2854 (59.9)	.54
Baseline cytologic tests or urinary biomarker assessments	516 (65.0)	1665 (34.9)	< .001	336 (42.3)	1876 (39.4)	.36

Abbreviations: BLC, blue-light cystoscopy; CIS, carcinoma in situ; NA, not applicable; WLC, white-light cystoscopy.

^aCategorical data are expressed as No. of patients (%), and continuous data are expressed as mean (SD).

^bBecause weighted values included decimal places, rounded values may not add up to the total number of patients.

After IPTW, BLC patients had higher utilization than WLC patients of all-cause visits during follow-up; however, these differences were not statistically significant (Table 3). The mean number of bladder cancer–related ambulatory visits was significantly higher for the BLC group than the WLC group (1.30 vs 1.00 visits PPPM; P = .004). However, these differences were offset by similar inpatient, ED, and pharmacy utilization between groups, resulting in no significant difference in total costs. Despite higher utilization rates for the BLC group, the mean number of all-cause ambulatory health care visits was similar between BLC and WLC groups (3.47 vs 3.21 visits PPPM; P = .11). Mean bladder cancer–related outpatient costs were higher for BLC than WLC patients ($621.86 vs $564.70 PPPM), but the difference was not significant (Table 4). Mean bladder cancer–related total medical and pharmacy costs did not differ significantly between BLC and WLC groups ($1301.10 vs $1246.60 PPPM; P = .75), nor did all-cause costs ($2987.93 vs $2886.16 PPPM; P = .65).

Table 3.

All-Cause and Bladder Cancer–Related Health Care Resource Utilization^a

Follow-up visitb	All-cause			Bladder cancer–related
	BLC (n = 794)	WLC (n = 4,763c)	P value	BLC (n = 794)	WLC (n = 4,763c)	P value
Ambulatory	3.47 (2.80)	3.21 (2.69)	.11	1.30 (1.72)	1.00 (1.55)	.004
Office	1.69 (1.31)	1.61 (1.28)	.32	0.50 (0.71)	0.45 (0.71)	.26
Outpatient	1.79 (2.33)	1.61 (2.20)	.16	0.81 (1.47)	0.55 (1.23)	.003
Emergency department	0.15 (0.30)	0.12 (0.26)	.17	0.02 (0.08)	0.02 (0.08)	.27
Inpatient stays	0.05 (0.10)	0.04 (0.10)	.08	0.02 (0.05)	0.01 (0.05)	.07
Inpatient days	0.50 (1.12)	0.45 (1.45)	.45	0.19 (0.73)	0.17 (1.02)	.66
Pharmacy fills	2.83 (2.16)	2.80 (2.23)	.82	0.01 (0.06)	0.01 (0.09)	.04

Abbreviations: BLC, blue-light cystoscopy; WLC, white-light cystoscopy.

^aMean (SD) visits per patient per month, weighted.

^bAt least 90 d after index cystoscopy.

^cOne WLC patient was excluded from analysis for having fewer than 90 d of follow-up after the index cystoscopy.

Table 4.

All-Cause and Bladder Cancer–Related Health Care Costs^a

Costb	All-cause			Bladder cancer–related
	BLC (n = 794)	WLC (n = 4,763c)	P value	BLC (n = 794)	WLC (n = 4,763c)	P value
Total medical and pharmacy	2987.93 (4177.72)	2886.16 (5316.09)	.65	1301.10 (3169.18)	1246.60 (4108.14)	.75
Medical	2594.12 (3886.82)	2531.34 (5175.33)	.76	1274.24 (3129.73)	1217.21 (4062.73)	.74
Ambulatory visits	1449.37 (2383.10)	1417.73 (3415.35)	.81	859.07 (2179.33)	832.82 (3160.19)	.83
Office visits	457.61 (1215.28)	470.31 (1359.35)	.86	237.21 (1154.72)	268.12 (1283.88)	.64
Outpatient visits	991.76 (1983.00)	947.41 (3054.28)	.68	621.86 (1820.24)	564.70 (2821.19)	.56
Emergency department visits	100.40 (229.52)	97.18 (247.25)	.79	18.04 (98.74)	14.94 (127.56)	.55
Inpatient stays	912.41 (2020.28)	880.72 (2875.89)	.75	348.90 (1381.46)	330.11 (1911.45)	.77
Other medical	131.95 (587.94)	135.71 (763.46)	.89	48.24 (479.87)	39.34 (396.84)	.68
Pharmacy	393.81 (1345.28)	354.83 (850.26)	.41	26.86 (551.62)	29.39 (465.68)	.89

Abbreviations: BLC, blue-light cystoscopy; WLC, white-light cystoscopy.

^aAt least 90 d after index cystoscopy.

^bMean (SD) 2023 inflation-adjusted US dollars, per patient per month, weighted.

^cOne WLC patient was excluded from analysis for having fewer than 90 days of follow-up after the index cystoscopy.

DISCUSSION

This claims-based study provides a real-world evaluation of health care resource utilization and costs associated with blue-light vs WLC in patients with NMIBC. We observed higher rates of claims coded for CIS and greater bladder cancer–related ambulatory utilization among patients undergoing BLC. However, despite increased utilization, total health care costs did not differ significantly between groups.

A notable finding was the significantly different rate of coding for CIS in the BLC group both before and after weighting. This pattern is consistent with previous clinical studies demonstrating improved visualization of CIS with BLC.^5-9 Identification of CIS can have an impact on patient management. First, CIS is associated with higher-risk disease categories and more intensive management pathways.^17,18 Second, its detection would require additional monitoring. Although treatment patterns must be interpreted cautiously given the limitations of claims data, the observed elevation in CIS coding compared with WLC suggests a more advanced patient cohort. Because CIS coding remained substantially more common among BLC patients even after weighting, residual differences in underlying disease severity or clinician selection of higher-risk patients for BLC cannot be excluded. Therefore, observed differences in utilization may reflect both enhanced detection and differences in patient case-mix.

Consistent with this, patients undergoing BLC were also more likely to have received intravesical BCG therapy. The observed claims indicate that patients receiving BLC are likely higher risk than those receiving WLC. This scenario would be consistent with BLC identifying more patients who should receive BCG and as a result driving additional HCRU. In exploratory analyses (data on file), we further observed that BCG during follow-up was at a higher rate in the BLC cohort than the WLC cohort and median time to BCG was shorter (7.67 months vs 13.19 months, P < .001).

Despite increased HCRU, our cost analysis showed no significant difference in overall health care expenditures between the BLC and WLC groups. This finding challenges the assumption that increased diagnostic intensity necessarily translates into higher cost. These findings also suggest that greater procedural intensity in NMIBC surveillance can occur within existing cost structures. Mean bladder cancer–related outpatient visit costs were slightly higher for BLC patients ($621.86 vs $564.70), but differences in total medical and pharmacy costs were not statistically significant. This suggests that although BLC may lead to more interventions in the short term, these do not translate into excessive financial burden. Despite higher bladder cancer–related utilization, differences in inpatient care, emergency department use, and pharmacy costs were small and not statistically significant.

There are existing published health economics and outcomes models on BLC. For example, Garfield et al¹⁹ created a probabilistic decision tree model for BLC vs WLC estimating costs and utility based on detection and recurrence rates. That analysis found that the use of BLC led to net positive savings as a result of overall improved downstream patient management and fewer recurrences. The Garfield model was based on Blue Light pivotal data from Grossman et al,⁸ not claims data or real utility measures. Creswell et al,²⁰ using a 5-year, Medicare-based model, calculated that BLC usage can create saving over a 5-year timeline vs WLC. The study concluded that recurrence rates were the strongest driver of cost. Further studies factor in the impact of BLC cost based on directly observed reductions in recurrence.^21-24

BRAVO, a recent propensity-score matched cohort from the Veterans Affairs Healthcare System, the largest real-world equal-access setting in the United States, compared the costs of BLC exposure vs WLC in NMIBC management in 622 patients.²⁵ On one hand, this study showed BLC exposure was associated with higher 5-year costs, driven in particular by additional outpatient care. However, offsetting these costs are lower recurrence rates, resulting in similar overall expenditures because of reduction in expensive clinical interventions. Thus, incorporating BLC into NMIBC management could provide an improved path to care while remaining affordable.

Our study demonstrates similar cost neutrality in a more generalizable real-world practice. It has several strengths, including the use of a large real-world data set, rigorous matching and IPTW to balance baseline characteristics, and an economically focused evaluation of cystoscopy approaches. However, this study does have limitations inherent to administrative claims. Primarily, claims data lack clinical information (eg, tumor stage, grade, and risk stratification), limiting the ability to fully characterize baseline disease severity. The study mitigated these differences by using weighting to balance patient characteristics across the treatment cohorts. The study lacks the long-term follow-up of other retrospective and prospective studies, which provide financial data over a more extended period.^14,25 Finally, other limitations include that the costs reported reflect paid amounts and do not capture institutional cost accounting or profitability, and although IPTW adjustment accounted for many baseline imbalances, some residual confounding remained, particularly for variables such as baseline CIS, geographic distribution, and prior bladder cancer treatments.

Despite these limitations, this study's findings were able to describe real-world utilization patterns and paid costs associated with BLC vs WLC, controlling for those clinical factors that were available within claims using robust matching and weighting procedures.

Several areas warrant further investigation. Future studies should (1) prospectively assess whether BLC reduces rates of long-term recurrence and progression in high-risk patients and (2) incorporate pathology-confirmed outcomes to better quantify the true clinical effect of BLC. In addition, a cost-effectiveness analysis could provide greater clarity about whether increased use of BLC ultimately translates into lower downstream costs by preventing high-cost interventions such as radical cystectomy or systemic therapy. Finally, further exploration of demographic and geographic variations in BLC utilization could help identify disparities in access and guide efforts to optimize its adoption across different health care settings.

CONCLUSION

This study demonstrates that although BLC is associated with higher bladder cancer–related health care resource utilization, it is not associated with higher total health care costs compared with WLC in real-world claims data. Higher rates of claims coded for CIS were observed among patients undergoing BLC; however, the clinical implications of this finding cannot be determined from claims data. Future studies incorporating clinical outcomes and pathology-confirmed end points are needed to further define the value of BLC in NMIBC management.

FUNDING

This study was supported by grant P30 CA015083 from the National Cancer Institute and by the Mayo Clinic Robert D. and Patricia E. Kern Center for the Science of Health Care Delivery (to M.D.T.), the Christian Haub Family Career Development Award in Cancer Research Honoring Dr. Richard Emslander (to M.D.T.), Eric and Gail Blodgett (to M.D.T.), the Endras family (to M.D.T.), and Photocure.

CONFLICT OF INTEREST DISCLOSURES

None. Photocure provided funding to Optum for this study. Barnes, Engel-Nitz, Johnson are employees of Optum and shareholders in UnitedHealth Group.

ETHICS STATEMENT

This study was reviewed by the Mayo Clinic Institutional Review Board and was deemed exempt from full board review because of its use of deidentified administrative claims data. The requirement for written informed consent was waived because the study involved secondary analysis of deidentified data.

DATA AVAILABILITY STATEMENT

All relevant data supporting the findings of this study are reported within the article or are available from the corresponding author upon reasonable request.

AUTHOR CONTRIBUTIONS

Conceptualization: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Data curation: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Formal analysis: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Funding acquisition: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Investigation: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Methodology: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Project administration: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Resources: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Software: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Supervision: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Validation: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Visualization: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Writing—original draft: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes. Writing—review & editing: Tyson, Choudry, McKee, Kalaria, James, Engel-Nitz, Johnson, Barnes.