[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure.
Weighted Kaplan-Meier Curves of Change in Coverage Comparing Enrollees With Traumatic Brain Injury (TBI) With Enrollees Without TBI
Weighted Kaplan-Meier Curves of Change in Coverage Comparing Enrollees With Traumatic Brain Injury (TBI) With Enrollees Without TBI

All P values are log-rank P values. A, Survival functions are significantly different (P < .001). B, Each stratum of Abbreviated Injury Scale (AIS) score among enrollees with TBI is significantly different from enrollees without TBI (P < .001). Enrollees with TBI who had an AIS score of 3 and those who had an AIS score of 4 are not significantly different (P = .70). Enrollees with TBI who had an AIS score of 5 or 6 are significantly different from enrollees with TBI who had an AIS score of 3 or 4 (P = .01).

Table.  
Accelerated Failure Time Model Comparing Coverage Time Among 13 558 Enrollees With Traumatic Brain Injury
Accelerated Failure Time Model Comparing Coverage Time Among 13 558 Enrollees With Traumatic Brain Injury
1.
Faul  M, Xu  L, Wald  MM, Coronado  VG.  Traumatic Brain Injury in the United States: Emergency Department Visits, Hospitalizations and Deaths 2002–2006. Atlanta, GA: US Dept of Health and Human Services/Centers for Disease Control and Prevention/National Center for Injury Prevention and Control; 2010.
2.
Selassie  AW, Zaloshnja  E, Langlois  JA, Miller  T, Jones  P, Steiner  C.  Incidence of long-term disability following traumatic brain injury hospitalization, United States, 2003.  J Head Trauma Rehabil. 2008;23(2):123-131.PubMedGoogle ScholarCrossref
3.
Reynolds  WE, Page  SJ, Johnston  MV.  Coordinated and adequately funded state streams for rehabilitation of newly injured persons with TBI.  J Head Trauma Rehabil. 2001;16(1):34-46.PubMedGoogle ScholarCrossref
4.
Masel  BE, DeWitt  DS.  Traumatic brain injury: a disease process, not an event.  J Neurotrauma. 2010;27(8):1529-1540.PubMedGoogle ScholarCrossref
5.
Adamson  DM, Chang  S, Hansen  LG.  Health Research Data for the Real World: The MarketScan Databases. New York, NY: Thompson Healthcare; 2008.
6.
Iacus  SM, King  G, Porro  G.  Causal inference without balance checking: coarsened exact matching.  Polit Anal. 2011;20(1):1-24. doi:10.1093/pan/mpr013.Google ScholarCrossref
7.
Cox  C, Chu  H, Schneider  MF, Muñoz  A.  Parametric survival analysis and taxonomy of hazard functions for the generalized gamma distribution.  Stat Med. 2007;26(23):4352-4374.PubMedGoogle ScholarCrossref
Research Letter
July 2016

Continuity of Private Health Insurance Coverage After Traumatic Brain Injury

Author Affiliations
  • 1Johns Hopkins Surgery Center for Outcomes Research, Department of Surgery, Johns Hopkins School of Medicine, Baltimore, Maryland
  • 2Center for Surgery and Public Health, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts
 

Copyright 2016 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.

JAMA Surg. 2016;151(7):678-680. doi:10.1001/jamasurg.2016.0040

Traumatic brain injury (TBI) accounts for 2.5 million emergency department visits, 280 000 hospitalizations, and 52 000 deaths in the United States each year, with approximately 40% of survivors developing some level of disability.1,2 Post-TBI disability in adults can lead to loss of employment and disruption of employment-based health insurance, which may affect ongoing care and can be devastating to policyholders and dependents. Even individuals sufficiently disabled by TBI to qualify for Medicare may face substantial delays in eligibility and access to care after loss of private coverage.3 The chronic nature of many post-TBI health issues underscores the importance of coverage continuity.4 We examined TBI-related factors associated with accelerated coverage change and differences in time to coverage change among individuals with employer-provided private insurance.

Methods

Using MarketScan,5 a claims database of privately insured Americans 0 to 65 years of age, we identified patients treated for TBI from January 1, 2010, to December 31, 2012. We excluded individuals with no Abbreviated Injury Scale (AIS) score for the head region (with an AIS score of ≥1 for nonhead regions) whose initial enrollment in MarketScan began after January 1, 2010, or whose death had been recorded during the study period. This study was approved by the institutional review board of the Johns Hopkins Medical Institutions and was conducted under the requirements of the data use agreement associated with the MarketScan data set. The primary outcome was coverage change, defined as an individual dropping out of MarketScan before turning 65 years of age or before the end of the study period. To examine the effect of TBI on time to coverage change, we created comparable times to coverage change for patients without TBI. To do so, we used the distribution of time from January 1, 2010, to the date of TBI among enrollees with TBI in order to randomly assign offsets to non-TBI enrollees. For each individual without TBI, the assigned offset was added to January 1, 2010, creating a “non-TBI” time origin. This enabled comparison of time to coverage change between enrollees with TBI and enrollees without TBI. Individuals without TBI were matched to those with TBI on age, sex, policyholder status, and time origin (TBI/assigned) using coarsened exact matching.6 We examined time to coverage change using both the Kaplan-Meier method with log-rank testing and parametric accelerated failure time modeling. We felt that accelerated failure time analysis yielded more meaningful estimates than Cox regression7; furthermore, the proportional hazard assumption of the latter was violated between AIS strata. We used R version 3.0.2 (The R Foundation for Statistical Computing) for data management, and Stata version 14 (StataCorp) for statistical analysis. Statistical significance was set at P < .05.

Results

Enrollees with TBI were more likely to change coverage than those without (30.7% vs 27.6%, respectively; P < .001) and to demonstrate accelerated coverage change (P < .001). Those with severe TBI (ie, AIS score of 5-6) demonstrated the shortest median time to coverage change compared with enrollees without TBI (145 vs 258 days) (Figure). The severity of the TBI was associated with accelerated coverage change in a dose-response manner; compared with patients without TBI, patients with TBI who had an AIS score of 2 demonstrated 8% shorter coverage (95% CI, 1%-14%; P = .03), patients with TBI who had an AIS score of 3 demonstrated 19% shorter coverage (95% CI, 14%-25%; P < .001), patients with TBI who had an AIS score of 4 demonstrated 23% shorter coverage (95% CI, 18%-27%; P < .001), and patients with TBI who had an AIS score of 5 to 6 demonstrated 44% shorter coverage (95% CI, 24%-59%; P < .001). Among patients with TBI (n = 13 558), time to coverage change was 17% shorter for policyholders than dependents and 32% shorter for those with an AIS score of 5 to 6 than those with an AIS score of 3 (P = .05). Compared with adult patients 40 to 49 years of age, those patients who were 0 to 9 years of age, 20 to 39 years of age, or 60 to 65 years of age experienced accelerated coverage change. The burden of comorbid disease was also associated with accelerated coverage change (Table).

Discussion

Traumatic brain injury accelerated coverage change, especially among the most severely injured. MarketScan does not report the nature (or lack) of postdropout coverage, so postchange insurance coverage could not be examined. Also, patients’ race and clinical measures of TBI-related physiologic dysfunction (eg, Glasgow Coma Scale) are not reported. Despite these limitations, increasing severity of TBI was associated with accelerated coverage change among individuals with employer-provided health insurance. Future studies should examine variability in access to care and subsequent coverage among patients with TBI-related change in insurance coverage.

Back to top
Article Information

Corresponding Author: Eric B. Schneider, PhD, Center for Surgery and Public Health, Brigham and Women’s Hospital, Harvard Medical School, 1620 Tremont St, Ste 4-020, Boston, MA 02120 (eschneider3@partners.org).

Published Online: March 2, 2016. doi:10.1001/jamasurg.2016.0040.

Author Contributions: Dr Schneider had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Lin, Schneider.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Lin, Schneider.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: All authors.

Administrative, technical, or material support: Canner.

Study supervision: Schneider.

Conflict of Interest Disclosures: None reported.

References
1.
Faul  M, Xu  L, Wald  MM, Coronado  VG.  Traumatic Brain Injury in the United States: Emergency Department Visits, Hospitalizations and Deaths 2002–2006. Atlanta, GA: US Dept of Health and Human Services/Centers for Disease Control and Prevention/National Center for Injury Prevention and Control; 2010.
2.
Selassie  AW, Zaloshnja  E, Langlois  JA, Miller  T, Jones  P, Steiner  C.  Incidence of long-term disability following traumatic brain injury hospitalization, United States, 2003.  J Head Trauma Rehabil. 2008;23(2):123-131.PubMedGoogle ScholarCrossref
3.
Reynolds  WE, Page  SJ, Johnston  MV.  Coordinated and adequately funded state streams for rehabilitation of newly injured persons with TBI.  J Head Trauma Rehabil. 2001;16(1):34-46.PubMedGoogle ScholarCrossref
4.
Masel  BE, DeWitt  DS.  Traumatic brain injury: a disease process, not an event.  J Neurotrauma. 2010;27(8):1529-1540.PubMedGoogle ScholarCrossref
5.
Adamson  DM, Chang  S, Hansen  LG.  Health Research Data for the Real World: The MarketScan Databases. New York, NY: Thompson Healthcare; 2008.
6.
Iacus  SM, King  G, Porro  G.  Causal inference without balance checking: coarsened exact matching.  Polit Anal. 2011;20(1):1-24. doi:10.1093/pan/mpr013.Google ScholarCrossref
7.
Cox  C, Chu  H, Schneider  MF, Muñoz  A.  Parametric survival analysis and taxonomy of hazard functions for the generalized gamma distribution.  Stat Med. 2007;26(23):4352-4374.PubMedGoogle ScholarCrossref
×