Timing of neurorehabilitation and subsequent Alzheimer’s disease risk in patients with moderate to severe traumatic brain injury: A nationwide retrospective cohort study in the United States

Abstract

Background:

Traumatic brain injury (TBI) is associated with an increased risk of Alzheimer’s disease (AD). It is unknown if prompt neuro-rehabilitative treatment following moderate or severe TBI mitigates this risk compared with delayed treatment.

Objective:

To determine whether immediate neuro-rehabilitative treatment following moderate or severe TBI reduces the risk of AD and related cognitive outcomes compared with delayed treatment.

Methods:

We conducted a retrospective cohort using the TriNetX Analytics Platform, which includes health records from over 100 million US patients. Adults aged 50–90 years with moderate or severe TBI were included if they received immediate treatment (within 1 week) or delayed treatment (>1 week). Outcomes were AD risk at 3- and 5-year follow-up, with additional outcomes of mild cognitive impairment (MCI), dementia, and AD-related medication prescriptions. Cox proportional hazards models were applied to propensity score–matched cohorts, and hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated.

Results:

Of 37,081 eligible patients, 17,636 remained after propensity score matching. Immediate treatment was associated with lower AD risk compared with delayed treatment (HR, 0.59; 95% CI, 0.41–0.86 at 3 years; HR, 0.70; 95% CI, 0.52–0.94 at 5 years). Similar risk reductions were observed for MCI, dementia, and AD-related medication use.

Conclusions:

Immediate treatment following moderate or severe TBI was associated with significantly reduced risk of AD and related cognitive decline. These findings suggest that prompt intervention may mitigate long-term neurodegenerative consequences of TBI.

Introduction

Traumatic brain injury (TBI) is defined as a form of acquired brain injury that occurs when a mechanical force causes damage to the brain including direct impact, rapid acceleration-deceleration, and blast-induced injuries often seen in military servicemembers.^1,2 TBI is a significant public health concern worldwide, estimated to affect 69 million individuals each year.³ Furthermore, TBI has been implicated in severe socioeconomic consequences including increased risk of job loss, healthcare utilization, and divorce in the years following TBI.⁴ Typical mechanisms of TBI include falls, motor vehicle accidents, sports injuries, and assaults.⁵ Younger adults are the most common demographic to receive a TBI through interpersonal violence or motor vehicle accident, and the elderly are most likely to receive a TBI from falling. Across all age groups and injury mechanisms males have been found to receive TBIs at nearly twice the incidence as females.⁶ Diagnosis of TBI typically involves a clinician assessing the severity of the TBI by using the Glasgow Coma Scale (GCS), which is a tool used to assess coma and impaired consciousness.⁷ This scale is often combined with neuroimaging, loss of consciousness at the time of injury, and the presence of posttraumatic amnesia to determine the overall severity of TBI.⁸ Recovery after a TBI is complex due to the heterogeneous nature of TBI injuries and the varied presentation of symptoms. In general, patients receive targeted treatment based on their deficits post-injury that may include cognitive rehabilitation, speech-language pathology services, physical or occupational therapy, and psychotherapy for mental health symptoms.⁹ Elderly patients receive worse outcomes post-TBI than younger patients, however recent studies are encouraging that significant gains can be made with rehabilitation.^10,11

Dementia is a general term used to describe a decline in one’s cognitive abilities severe enough to affect daily living, with Alzheimer disease (AD) being the most prevalent form of dementia.¹² AD is characterized by progressive memory loss, disorientation, impaired executive functioning, and behavioral disturbances. Pathologically, AD is defined by the accumulation of extracellular amyloid-beta plaques and intracellular neurofibrillary tangles composed of hyperphosphorylated tau protein.¹³ Diagnosis is typically based on clinical evaluation, neuropsychological testing, and supportive imaging findings such as cortical atrophy or amyloid PET scans.¹⁴ Established risk factors for AD include advancing age, the presence of the APOE ϵ4 allele, cardiovascular disease, and low educational attainment.¹⁵ Protective factors include physical activity, cognitive stimulation, and adherence to a Mediterranean-style diet.¹⁶ An estimated 6.7 million Americans are living with AD today, and it is estimated this number could grow to 13.8 million by 2060.¹⁷ Recently, epidemiological studies established a link between TBI and an increased risk for AD and dementia later in life.¹⁸ Several studies have found that the risk of AD varies with the type and severity of TBI, and that moderate and severe TBI have a higher risk than mild TBI.^19–21 Furthermore, even remote TBI may lead to an earlier age of onset of AD.^22,23

Neurorehabilitation, encompassing physical therapy, occupational therapy, cognitive rehabilitation, and speech-language therapy, is a mainstay of TBI treatment that is designed to leverage the neuroplasticity of the brain to improve motor and cognitive functions that may have been damaged from the injury.^24,25 Duration typically ranges from several weeks to months depending on injury severity and patient progress.²⁶ Patients who receive neurorehabilitation during acute hospitalization from a head injury have been found to have significantly higher cognitive levels at discharge compared to those who did not receive neurorehabilitation treatment.²⁷ Furthermore, several studies have illustrated that timely treatment of a TBI leads to lower scores on Post-Concussion Symptom Scales, which assess cognitive and emotional symptoms like memory difficulties, concentration deficits, and emotional lability, compared to a delay between injury and neuro-rehabilitative treatment.^28–31 However, not all patients receive immediate treatment following TBI. One study in Colorado found that 29% of patients did not receive neurorehabilitation services following TBI while another study found that 40% of patients had persistent cognitive impairments in the year following TBI.^32,33

Despite the established link between TBI and an increased risk of AD, it remains unknown how the timing of TBI treatment is associated with AD risk. We hypothesized that those who received immediate neuro-rehabilitative treatment following TBI had a reduced risk of AD and other dementias compared to those who received delayed treatment. In this study, we performed a retrospective cohort study using a nation-wide database of patient electronic health records (EHRs) to investigate the long-term risk of AD, mild cognitive impairment (MCI), all-cause dementia, and AD-related medication prescriptions in patients who received treatment within one week following TBI compared with those who received treatments between one-week and 6 months after TBI.

Methods

Database description

We used TriNetX, a global federated health research network providing access to statistics on electronic health records (diagnoses, procedures, medications, laboratory values, genomic information) from approximately 118 million patients in 69 large Healthcare Organizations in the US covering diverse geographic regions, age, race and ethnicity, income and insurance groups, and clinical settings.³⁴ Given potential differences across countries in cultural and socioeconomic factors, diagnostic criteria for AD and TBI, treatment guidelines, healthcare utilization, and clinical practice patterns, which we could not explicitly control for, this study focused exclusively on U.S. patients. As a federated network, TriNetX received a waiver from Western IRB since only aggregated counts and statistical summaries of de-identified information, but no protected health information was received, and no study-specific activities were performed in retrospective analyses.

Study population and cohort definitions

The study population comprised 37,081 patients aged 50–90 who had medical encounters for a moderate or severe TBI diagnosis with healthcare organizations from January 1, 2000 until December 31, 2019. Patients who have had a diagnosis of MCI, AD, dementia, or AD-related medicine prescriptions prior to TBI were excluded. The population was divided into two cohorts: 1) immediate treatment group: 28,157 patients who received neurorehabilitation within one week of TBI. 2) delayed treatment group: 8924 patients received no neurorehabilitation within in the first week but received treatment between 1 week and 6 months of TBI (Figure 1). Diagnoses of TBI, MCI, AD, and Dementia were determined by the International Classification of Diseases, 10^th Revision (ICD-10) codes. AD-related drug data were recorded using the National Drug Code (NDC), the FDA’s drug identification system. Neurorehabilitation treatment was determined by Current Procedural Terminology (CPT) codes (Supplemental Table 1).

Figure 1.

Flowchart of study cohort selection.

The time periods used to define immediate or delayed treatment were chosen to reflect the critical window for addressing acute brain injuries and initiating neurorehabilitation and were also influenced by the available sample size in each group. While alternate timeframes such as treatment within the first day or within two weeks could be considered, these may be impractical in clinical workflows where logistical delays are common. The delayed treatment window of one week to six months was selected to represent a transitional period in which subacute interventions may still occur, capturing patients who do not receive timely treatment but still receive some rehabilitative care before prolonged gaps. In the delayed treatment group, the median time for receiving treatment after TBI diagnosis is approximately 1 month. Future work should explore how alternative definitions of treatment timing may impact outcomes.

Covariates and outcomes

Cohorts were propensity-score matched for demographics (age, race, sex), socioeconomic factors (education level, employment status, social and psychosocial environment, and housing conditions), and pre-existing conditions that are risk factors of TBI and Dementia, including diabetes mellitus, obesity, cardiovascular disease, chronic obstructive pulmonary disease, and substance abuse treatment.^35–39

Statistical analysis

The main outcome was AD (ICD-10 code G30). Additional outcomes of MCI (ICD-10 code G31.84), all-cause dementia (ICD-10 codes F01-F03), and AD-related medication prescriptions (NDC code N06D) were also examined. The immediate and delayed treatment groups were propensity-score matched (1:1 using nearest neighbor greedy matching with a caliper of 0.25 times the Standard Deviation) for baseline covariates. Cox proportional hazard analyses were used to compare rates of time-to-events daily during three- and five-year follow-up time after the TBI diagnosis (index event). Patients in matched groups were followed starting after the index event (TBI) until the occurrence of the outcome, death, loss to follow-up, or five years after the index event, whichever occurred first. Each outcome was analyzed separately (no multiple comparisons or competing outcomes). Hazard ratios (HRs) and 95% CIs were calculated. All the analyses were done using built-in functions within the TriNetX Analytics platform that are implemented using Survival 3.2–3 in R 4.0.2 and libraries and utilities for data science and statistics in Python 3.7 and Java 11.0.16.

A sensitivity analysis was done to examine the possibility that those with delayed treatment are having more serious procedures done, such as craniotomy or craniectomy. A separate cohort was created based on the immediate and delayed treatment groups to exclude any patients that received a craniotomy or craniectomy, based on CPT code. The full list of codes used can be found in Supplemental Table 1.

Results

Comparison of AD risk between patients who received immediate versus delayed neurorehabilitation following a moderate to severe TBI

Among the 28,157 patients who received immediate treatment (mean [SD] age, 65.8 [9.31]), 39.3% were female, 5.8% were Asian individuals, 10.7% were Black individuals, 5.3% were Hispanic individuals, and 75.0% were White individuals. Among the 8924 patients with delayed treatment (mean [SD] age, 63.9 [9.04]), 40.9% were female, 3.2% were Asian individuals, 11.7% were Black individuals, 5.7% were Hispanic individuals, and 73.6% were White individuals. Compared with patients in the immediate treatment, patients with delayed treatment faced more socioeconomic and environmental challenges. Additionally, patients in the delayed treatment group had a higher prevalence of chronic illness such as diabetes, cerebrovascular disease, and chronic respiratory disease. After propensity-score matching, the immediate and delayed treatment groups (8818 patients in each group) were balanced (Table 1).

Table 1.

Baseline characteristics of immediate vs. delayed treatment cohorts before and after propensity-score matching. SMD: standardized mean difference; SD: standard deviation.

	Before matching			After matching
	Immediate	1 week – 6 months	SMD	Immediate	1 week – 6 months	SMD
Total number of patients	28,157	8924		8818	8818
Age at Index (years, mean ± SD)	65.8 ± 9.31	63.9 ± 9.04	0.21*	64 ± 9.09	63.9 ± 9.05	0.008
Sex (%)
Female	39.3	40.9	0.03	41.1	40.9	0.004
Male	60.1	56.5	0.07	57.0	56.9	0.001
Unknown	0.64	2.6	0.15*	1.9	2.2	0.02
Ethnicity (%)
Hispanic/Latinx	5.3	5.7	0.02	6.4	5.7	0.02
Not Hispanic/Latinx	77.9	76.4	0.03	75.5	76.7	0.02
Unknown	16.8	17.9	0.03	18.1	17.6	0.01
Race (%)
African American/Black	10.7	11.7	0.03	11.7	11.7	0.0003
American Indian/Alaska Native	0.39	0.62	0.03	0.54	0.59	0.006
Asian	5.8	3.2	0.12*	3.4	3.2	0.007
Native Hawaiian/Pacific Islander	1.09	0.76	0.03	0.78	0.77	0.001
White	75.0	73.6	0.03	73.4	74.00	0.01
Other	3.0	2.7	0.01	3.1	2.8	0.01
Unknown	4.1	7.4	0.14*	7.2	7.0	0.009
Adverse social determinants of health (%)	2.4	4.4	0.11*	4.3	4.2	0.002
Problems related to lifestyle (%)	2.9	5.0	0.11*	4.9	4.8	0.005
Comorbidities (%)
Hypertensive diseases	33.9	45.7	0.24*	46.9	45.3	0.03
Diabetes mellitus	15.6	19.5	0.10*	20.3	19.4	0.02
Ischemic heart diseases	15.0	19.6	0.12*	20.3	19.5	0.02
Other forms of heart disease	22.0	30.4	0.19*	31.4	30.0	0.03
Chronic rheumatic heart diseases	3.4	4.6	0.06	5.0	4.6	0.01
Cerebrovascular diseases	14.2	18.0	0.10*	18.5	17.7	0.01
Chronic lower respiratory diseases	13.6	21.6	0.21*	21.5	21.2	0.008
Diseases of arteries, arterioles and capillaries	10.2	14.5	0.13*	15.0	14.3	0.01
Transient cerebral ischemic attacks and related syndromes	2.4	3.5	0.06	3.4	3.5	0.004
Disorders of lipoprotein metabolism and other lipidemias	25.0	37.2	0.26*	37.4	36.7	0.01
Cerebral infarction	5.6	7.0	0.05	7.0	6.9	0.0004
Down syndrome	0.04	0.11	0.02	0.11	0.11	0
Sleep disorders	11.7	20.6	0.24*	20.3	20.0	0.007
Occlusion and stenosis of precerebral arteries	3.5	4.8	0.06	4.9	4.7	0.006
Occlusion and stenosis of cerebral arteries	0.56	0.71	0.01	0.71	0.70	0.001
Other and unspecified hearing loss	3.2	5.9	0.12*	5.8	5.6	0.008
Conductive and sensorineural hearing loss	2.8	5.8	0.15*	5.6	5.5	0.004
Depressive episode	12.5	20.7	0.22*	20.9	20.1	0.018
Major depressive disorder, recurrent	2.5	5.2	0.13*	5.1	4.9	0.006
Anxiety	12.6	22.8	0.26*	22.3	22.2	0.001
Injuries to the neck	7.3	12.7	0.18*	12.1	12.2	0.003
Crushing injury of head	0.04	0.11	0.02	0.11	0.11	0
Avulsion and traumatic amputation of part of head	0.04	0.11	0.02	0.11	0.11	0
Traumatic cerebral edema	0.04	0	0.02	0.11	0	0.04
Diffuse traumatic brain injury	0.10	0.16	0.01	0.19	0.15	0.01
Focal traumatic brain injury	0.27	0.33	0.01	0.31	0.32	0.002
Epidural hemorrhage	0.22	0.11	0.02	0.16	0.11	0.01
Traumatic subdural hemorrhage	2.3	2.4	0.01	2.6	2.4	0.01
Traumatic subarachnoid hemorrhage	0.52	0.48	0.004	0.50	0.48	0.003
Other specified intracranial injuries	0.04	0.11	0.02	0.11	0.11	0
Unspecified intracranial injury	1.0	1.85	0.06	1.8	1.8	0.001
Traumatic brain compression and herniation	0.04	0.11	0.02	0.11	0.11	0
Other specified injuries of head	2.5	4.7	0.11*	4.6	4.5	0.005
Overweight and obesity	7.6	14.4	0.22*	14.3	13.9	0.01
Alcohol related disorders	7.6	9.6	0.07	10.09	9.6	0.01
Nicotine dependence	11.3	15.5	0.12*	16.0	15.4	0.01
Procedure (%)
Substance Abuse Treatment	0.54	1.0	0.05	l.l	1.0	0.007
Occupational therapy evaluation	3.0	3.6	0.03	4.0	3.6	0.02
Treatment of speech, language, voice, communication	2.2	2.4	0.01	2.5	2.4	0.006
Neuropsychological testing evaluation services	0.04	0.11	0.02	0.11	0.11	0

Compared with patients who received delayed treatment following TBI, those who received immediate treatment had a significantly lower risk of subsequent AD, with hazard ratios of 0.59 (95% CI, 0.41–0.86) at 3-year and 0.7 (95% CI, 0.52–0.94) at 5-year of follow-up. Similar results were observed for all-cause dementia, MCI and AD-related medicine prescription (Figure 2). Additionally, we conducted a stratified analysis by gender, which found that males in the immediate treatment group had significantly lower risk of dementia, MCI, and AD-related prescriptions while females had significantly lower risk of MCI and AD-related prescriptions. These results are presented in Supplemental Figure 1.

Figure 2.

Comparison of the risk of AD, dementia, MCI and receiving AD-related medicine treatment between propensity-score-matched patients who received immediate vs delayed treatments following moderate or severe TBI (index event). Individuals in the matched groups were followed following the index event until the occurrence of the outcome, death, loss to follow-up, or 3 or 5 years after the index event, whichever occurred first. Hazard rates were calculated using a Cox proportional hazards model with censoring applied.

Sensitivity analysis

Since intensive surgical treatment such as craniotomy or craniectomy might disrupt neurorehabilitation patterns, we performed a sensitivity analysis by excluding patients who received either a craniotomy or craniectomy after receiving a TBI. A total of 2056 (7.3%) patients in the immediate treatment group, and 878 (9.8%) patients in the delayed treatment group were excluded from the sensitivity analysis. The characteristics of patients in the immediate treatment and delay treatment cohorts are shown in Supplemental Table 2. In Figure 3, patients who received immediate treatment had a reduced risk of AD at both 3 years (HR: 0.66, 95% CI: 0.46–0.95) and 5 years (HR: 0.70, 95% CI: 0.52–0.95). Significant risk reductions were also observed for other outcomes, with hazard ratios ranging from 0.75 to 0.88 for all-cause dementia, 0.76 to 0.78 for MCI, and 0.70 to 0.76 for AD-related medication use. These results suggest that surgical procedures following TBI likely do not significantly impact AD risk after neurorehabilitation.

Figure 3.

Comparison of the risk of AD, dementia, MCI, and receiving AD-related medication treatment between propensity-score-matched patients who received immediate versus delayed treatments without undergoing craniotomy or craniectomy following a moderate or severe traumatic brain injury (TBI) (index event).

Discussion

This study suggests that patients who sustained a moderate or severe TBI and received immediate treatment within one week had a lower risk of developing AD, all-cause dementia, MCI, and receiving AD-related medication treatment over the 3- and 5-year periods following injury, compared to those who received delayed treatment.

There have been several studies linking TBI that occurs in middle age to dementia later in life, however there are no other large epidemiological studies that have examined how prompt neurorehabilitation treatment of TBI may reduce the risk of dementia after a TBI.^40–48 Findings from this study suggest an association between earlier treatment following TBI and a lower risk of developing AD compared to delayed treatment.

There have been several proposed mechanisms for the development of AD and dementia after TBI. These include the accumulation of amyloid-β (Aβ) and the development of tau pathology, both of which are hallmark features of AD.^49,50 In addition, TBI may lead to a chronic inflammatory state and a weakened blood-brain barrier, increasing susceptibility to neurodegenerative diseases.^51–53 Another possibility is that individuals may experience functional impairments, anxiety, depression, and other long-term cognitive deficits after TBI, which may further elevate the risk of developing dementia.^54–58

While the primary analyses were conducted after propensity-score matching, the delayed treatment cohort exhibited higher prevalence of adverse social determinants of health and comorbid medical conditions such as hypertension, diabetes, cardiovascular disease, cerebrovascular disease, and chronic lower respiratory disease (Table 1). Each of these factors is independently associated with increased risk of dementia and AD, potentially compounding vulnerability after TBI.^59–64 Although matching substantially attenuates these differences, residual confounding cannot be excluded. The persistence of a significant association between delayed treatment and higher AD risk after matching suggest treatment timing may independently contribute to risk reduction, but this should be interpreted alongside the broader health context of the cohorts.

Neurorehabilitation plays a central role in the recovery process following TBI and typically includes a combination of physical therapy, occupational therapy, speech-language therapy, and cognitive rehabilitation.⁹ Neurorehabilitation leverages the brain’s plasticity to make functional recovery after TBI and other brain injuries.⁶⁵ Neuroplasticity describes the brain’s ability to make adaptive changes, particularly during growth and development but also after an injury.⁶⁶ This process involves changes in the existing connections between neurons and the development of new connections and synapses between others allowing for the capacity to adapt structurally and functionally in response to injury.⁶⁷ Recent studies have shown that early, intensive neurorehabilitation leads to better functional outcomes in patients with TBI.^68,69 Findings from our study further show that immediate treatment could have long-term benefits in mitigating the risk of developing AD, MCI, all-cause dementias and receiving AD-related medicine treatment.

An additional mechanism linking early neurorehabilitation to lower AD risk may involve the concept of cognitive reserve. Cognitive reserve reflects the brain’s resilience to neuropathological damage, allowing individuals to maintain function despite accumulating pathology.⁷⁰ Neuroplasticity, enhanced though neurorehabilitation, may contribute to building or maintaining cognitive reserve by strengthening neural networks, promoting synaptogenesis, and supporting compensatory reorganization.^71,72

This study focuses on patients aged 50–90. This age range was selected because middle-aged and older adults have a higher baseline risk of dementia, which allowed for diagnoses to be measured within the relatively short follow-up period of 5 years. Although TBI can occur at any age, younger patients have lower dementia risk and would require decades of observation to detect. Additionally, older individuals may be more susceptible to cognitive decline or already have age-related changes that amplify the effects of a TBI, increasing the likelihood of a diagnosis of AD or dementia. Future research incorporating broader age ranges and longer follow-up times could clarify if there are long-term neuroprotective benefits from early treatment in younger cohorts.

This study compared the long-term risk of AD, all-cause dementia, and MCI in patients who receive immediate treatment, or delayed treatment following MCI. These results have important implications for the management and treatment of TBI in adults of middle age, as the risk of dementia in later life may be mitigated in the treatment of such injuries.

Limitations

This study has several limitations. First, the nature and intensity of treatments received in the “immediate” and “delayed” groups were not fully standardized. Variability in the quality and type of rehabilitation services, medical management, and follow-up care could have contributed to differences in outcomes. While many patients likely received neuro-rehabilitative care in the hospital, it is possible they were referred for outpatient and did not attend. Neurorehabilitation treatment was based on procedural codes, which does not provide information about the nature or intensity of treatment. Future studies should attempt to control for treatment variability to better isolate the impact of treatment timing. In addition, due to limited sample size and the focus of our study on timing, we did not further compare single vs multiple injuries as well as how treatment timing was associated with AD outcomes in single vs multiple injuries. Future studies leveraging larger-scale EHR data are warranted to explore these associations. Second, TBI severity was determined using ICD-10 codes, which may not be as accurate as other diagnostic measures like the Glasgow Coma Scale. The use of administrative codes might lead to misclassification or insufficient granularity in capturing injury severity, potentially introducing bias. Third, we used ICD-10 diagnosis codes for AD as the primary outcome and corroborated findings using AD-related medication prescriptions. However, AD diagnosis in real-world clinical settings is complex and often lacks standardized documentation in EHRs. Diagnostic accuracy may be influenced by various factors, including socioeconomic status, healthcare access, and the use of biomarkers, all of which could introduce misclassification or bias. Fourth, this is a retrospective observational study which has limitations due to residual confounding and biases; therefore, causality cannot be established. While our propensity-score matching approach aimed to balance observed characteristics, unmeasured or inadequately captured socioeconomic, cultural, and environmental factors may still confound the observed association. In addition, there are inherent limitations in patient EHR-based studies that include over, under, and misdiagnosis. Fifth, due to sample limitations, we were unable to examine how immediate treatment following TBI had an impact on frontotemporal dementia, Lewy body dementia, for both of which TBI is a modifiable risk factor.^73,74 We were also unable to evaluate the impact on Chronic Traumatic Encephalopathy, because there is currently no formal ICD-10 code. Finally, the patient data from the TriNetX database is limited to those who have had medical interactions within healthcare systems that participate in the TriNetX network. This dataset may not accurately reflect the United States population. Consequently, it is important to corroborate these findings with studies of other population groups.

Conclusion

Results from this cohort study suggest that among patients aged 50 to 90 years diagnosed with moderate or severe TBI, those who received treatment within the first week had a significantly lower risk of developing dementia compared to those who received delayed treatment. These results highlight the need for immediate treatment of TBI and for long-term monitoring for the development of AD and other dementias. Future studies are needed to continue to explore the underlying biological mechanisms present in both TBI and dementia to understand how to further mitigate risk.

Supplementary Material

Supplemental material for this article is available online.

Data availability statement

This study used population-level aggregate and de-identified data collected by the TriNetX Platform and are available from TriNetX, LLC (https://trinetx.com/) but third-party restrictions apply to the availability of these data. The data were used under license for this study with restrictions that do not allow for the data to be redistributed or made publicly available. To gain access to the data, a request can be made to TriNetX (join@trinetx.com), but costs might be incurred, and a data-sharing agreement would be necessary. Data specific to this study including diagnosis codes and group characteristics in aggregated format are included in the manuscript as tables, figures, and supplemental files.