Advertisement

Rehospitalization After Traumatic Brain Injury: A Population-Based Study

      Abstract

      Objective

      To examine, from a Canadian population-based perspective, the incidence and etiology of long-term hospital utilization among persons living with traumatic brain injury (TBI) by age and sex.

      Design

      Retrospective cohort study.

      Setting

      Acute care hospitals.

      Participants

      Index cases of TBI (N=29,269) were identified from the Discharge Abstract Database for fiscal years 2002/2003 through 2009/2010 and were followed-up until 36 months after injury.

      Interventions

      Not applicable.

      Main Outcome Measures

      Rehospitalization was defined as admission to an acute care facility that occurred up to 36 months after index injury. Diagnoses associated with subsequent rehospitalization were examined by age and sex.

      Results

      Of the patients with TBI, 35.5% (n=10,390) were subsequently hospitalized during the 3-year follow-up period. Multivariable logistic regression (controlling for index admission hospital) identified men, older age, mechanism of injury being a fall, greater injury severity, rural residence, greater comorbidity, and psychiatric comorbidity to be significant predictors of rehospitalization in a 3-year period postinjury. The most common causes for rehospitalization differed by age and sex.

      Conclusions

      Rehospitalization after TBI is common. Factors associated with rehospitalization can inform long-term postdischarge planning. Findings also support examining causes for rehospitalization by age and sex.

      Keywords

      List of abbreviations:

      ADG (Aggregated Diagnosis Group), DAD (Discharge Abstract Database), ICD-10-CA (International Classification of Diseases–10th Revision–Canadian Enhancement), MVC (motor vehicle collision), TBI (traumatic brain injury)
      Traumatic brain injury (TBI) is a major cause of death and disability in the United States. Approximately 1.7 million people will sustain a TBI each year in the United States, resulting in 1.4 million emergency department visits, 275,000 hospitalizations, and 52,000 deaths.
      • Faul M.
      • Xu L.
      • Wald M.M.
      • Coronado V.
      Traumatic brain injury in the United States: emergency department visits, hospitalizations, and deaths, 2002–2006.
      The economic costs associated with TBI, including direct medical and rehabilitation costs and indirect societal costs, are estimated to be $60 billion in the United States.
      • Finkelstein E.
      • Corso P.
      • Miller T.
      • et al.
      The incidence and economic burden of injuries in the United States.
      Recurrence of hospitalization (ie, rehospitalization) is a common problem among individuals with a chronic condition or injury and is associated with increased health care costs and reduced health care quality.
      • Jencks S.F.
      • Williams M.V.
      • Coleman E.A.
      Rehospitalizations among patients in the Medicare fee-for-service program.
      • Weissman J.S.
      • Ayanian J.Z.
      • Chasan-Taber S.
      • Sherwood M.J.
      • Roth C.
      • Epstein A.M.
      Hospital readmissions and quality of care.
      Canadian Institute for Health Information
      All cause readmission to acute care and return to the emergency department.
      A reported 15% of individuals with chronic conditions (eg, stroke, TBI, hip fracture) were readmitted to hospital within 2 months after discharge.
      • Dai Y.T.
      • Wu S.C.
      • Weng R.
      Unplanned hospital readmission and its predictors in patients with chronic conditions.
      Those who suffered from an injury had an increased risk of reinjury.
      • Worrell S.S.
      • Koepsell T.D.
      • Sabath D.R.
      • Gentilello L.M.
      • Mock C.N.
      • Nathens A.B.
      The risk of reinjury in relation to time since first injury: a retrospective population-based study.
      Older age, being a man, greater injury severity,
      • Worrell S.S.
      • Koepsell T.D.
      • Sabath D.R.
      • Gentilello L.M.
      • Mock C.N.
      • Nathens A.B.
      The risk of reinjury in relation to time since first injury: a retrospective population-based study.
      • Moore L.
      • Stelfox H.T.
      • Turgeon A.F.
      • et al.
      Derivation and validation of a quality indicator for 30-day unplanned hospital readmission to evaluate trauma care.
      and longer initial length of stay
      • Westert G.P.
      • Lagoe R.J.
      • Keskimäki I.
      • Leyland A.
      • Murphy M.
      An international study of hospital readmissions and related utilization in Europe and the USA.
      • Cakir B.
      • Gammon G.
      Evaluating readmission rates: how can we improve?.
      were associated with higher rates of readmission, with most readmissions being the result of congestive heart failure and chronic obstructive pulmonary disease.
      • Westert G.P.
      • Lagoe R.J.
      • Keskimäki I.
      • Leyland A.
      • Murphy M.
      An international study of hospital readmissions and related utilization in Europe and the USA.
      • Shipton S.
      Risk factors associated with multiple hospital readmissions.
      These trends, however, apply to a diverse population with various conditions and may not necessarily generalize to patients with a TBI.
      Persons with TBI have higher rates of readmission and are more likely to incur subsequent injuries compared with nonbrain injured controls, with numbers ranging from 16% to 18% at 6 months,
      • Boutin A.
      • Francisque K.
      • Moore L.
      • Lauzier F.
      • Neveu X.
      • Turgeon A.
      Hospital readmissions following traumatic brain injury.
      • Moore L.
      • Stelfox H.T.
      • Turgeon A.F.
      • et al.
      Rates, patterns, and determinants of unplanned readmission after traumatic injury: a multicenter cohort study.
      17% to 20% at 1 year,
      • Moore L.
      • Stelfox H.T.
      • Turgeon A.F.
      • et al.
      Rates, patterns, and determinants of unplanned readmission after traumatic injury: a multicenter cohort study.
      • Nakase-Richardson R.
      • Tran J.
      • Cifu D.
      • et al.
      Do rehospitalization rates differ among injury severity levels in the NIDRR Traumatic Brain Injury Model Systems program?.
      and 20% to 23% over a 3-year period.
      • Cifu D.X.
      • Kreutzer J.S.
      • Marwitz J.H.
      • et al.
      Etiology and incidence of rehospitalization after traumatic brain injury: a multicenter analysis.
      Similar rates of rehospitalization have been established at 5 years postinjury, with a trend toward elective surgery early on and general health maintenance after the first year.
      • Marwitz J.H.
      • Cifu D.X.
      • Englander J.
      • High W.M.
      A multi-center analysis of rehospitalizations five years after brain injury.
      Children and adults who sustained a TBI also appear more likely to sustain another injury at 1- and 10-year follow-up, respectively, compared with nonbrain injured controls.
      • Cameron C.M.
      • Purdie D.M.
      • Kliewer E.V.
      • McClure R.J.
      Ten-year outcomes following traumatic brain injury: a population-based cohort.
      • Swaine B.R.
      • Tremblay C.
      • Platt R.W.
      • Grimard G.
      • Zhang X.
      • Pless I.B.
      Previous head injury is a risk factor for subsequent head injury in children: a longitudinal cohort study.
      Previous studies, however, have examined patients with TBI who were discharged from selected trauma
      • Boutin A.
      • Francisque K.
      • Moore L.
      • Lauzier F.
      • Neveu X.
      • Turgeon A.
      Hospital readmissions following traumatic brain injury.
      • Moore L.
      • Stelfox H.T.
      • Turgeon A.F.
      • et al.
      Rates, patterns, and determinants of unplanned readmission after traumatic injury: a multicenter cohort study.
      or rehabilitation centers,
      • Nakase-Richardson R.
      • Tran J.
      • Cifu D.
      • et al.
      Do rehospitalization rates differ among injury severity levels in the NIDRR Traumatic Brain Injury Model Systems program?.
      • Cifu D.X.
      • Kreutzer J.S.
      • Marwitz J.H.
      • et al.
      Etiology and incidence of rehospitalization after traumatic brain injury: a multicenter analysis.
      • Marwitz J.H.
      • Cifu D.X.
      • Englander J.
      • High W.M.
      A multi-center analysis of rehospitalizations five years after brain injury.
      which predominately treat a small segment of moderate to severe cases and are likely affected by attrition.
      • Boutin A.
      • Francisque K.
      • Moore L.
      • Lauzier F.
      • Neveu X.
      • Turgeon A.
      Hospital readmissions following traumatic brain injury.
      • Moore L.
      • Stelfox H.T.
      • Turgeon A.F.
      • et al.
      Rates, patterns, and determinants of unplanned readmission after traumatic injury: a multicenter cohort study.
      • Nakase-Richardson R.
      • Tran J.
      • Cifu D.
      • et al.
      Do rehospitalization rates differ among injury severity levels in the NIDRR Traumatic Brain Injury Model Systems program?.
      • Cifu D.X.
      • Kreutzer J.S.
      • Marwitz J.H.
      • et al.
      Etiology and incidence of rehospitalization after traumatic brain injury: a multicenter analysis.
      • Marwitz J.H.
      • Cifu D.X.
      • Englander J.
      • High W.M.
      A multi-center analysis of rehospitalizations five years after brain injury.
      • Colantonio A.
      • Croxford R.
      • Farooq S.
      • Laporte A.
      • Coyte P.C.
      Trends in hospitalization associated with traumatic brain injury in a publicly insured population, 1992-2002.
      A subset of these studies was based on U.S. data,
      • Cifu D.X.
      • Kreutzer J.S.
      • Marwitz J.H.
      • et al.
      Etiology and incidence of rehospitalization after traumatic brain injury: a multicenter analysis.
      • Marwitz J.H.
      • Cifu D.X.
      • Englander J.
      • High W.M.
      A multi-center analysis of rehospitalizations five years after brain injury.
      and those conducted in Canada are limited to a specific age group
      • Cameron C.M.
      • Purdie D.M.
      • Kliewer E.V.
      • McClure R.J.
      Ten-year outcomes following traumatic brain injury: a population-based cohort.
      • Swaine B.R.
      • Tremblay C.
      • Platt R.W.
      • Grimard G.
      • Zhang X.
      • Pless I.B.
      Previous head injury is a risk factor for subsequent head injury in children: a longitudinal cohort study.
      or include only unplanned readmissions.
      • Boutin A.
      • Francisque K.
      • Moore L.
      • Lauzier F.
      • Neveu X.
      • Turgeon A.
      Hospital readmissions following traumatic brain injury.
      • Moore L.
      • Stelfox H.T.
      • Turgeon A.F.
      • et al.
      Rates, patterns, and determinants of unplanned readmission after traumatic injury: a multicenter cohort study.
      Consequently, the aim of the current study was to investigate, from a Canadian population-based perspective, the incidence and etiology of all rehospitalizations among those who sustained a TBI at 1- and 3-year follow-up times. Given that the incidence rates of TBI vary by age and sex,
      • Colantonio A.
      • Croxford R.
      • Farooq S.
      • Laporte A.
      • Coyte P.C.
      Trends in hospitalization associated with traumatic brain injury in a publicly insured population, 1992-2002.
      • Colantonio A.
      • LaPorte A.
      • Croxford R.
      • Coyte P.
      Who receives in-patient rehabilitation after traumatic brain injury? A population-based study.
      rehospitalization rates were stratified by these factors.

      Methods

      Participants

      Participants included in this retrospective cohort study were patients discharged alive from Ontario hospitals between April 1, 2003, and March 31, 2010 (fiscal years 2002/2003–2009/2010) for a TBI. Patients with a TBI were identified from acute care hospitalization records using the following International Classification of Diseases–10th Revision–Canadian Enhancement (ICD-10-CA) injury codes: a fracture of the skull and facial bones (S02.0, S02.1, S02.3, S02.7, S02.8, S02.9), injury to the optic nerve and pathways (S04.0), intracranial injury (S06.0–S06.9), or crushing injury of the head (S07.0–S07.9). To ensure integrity of the incident cases, patients were excluded from analyses if they sustained a TBI the year prior to their index injury. A 1-year clearance is consistent with prior studies that investigate readmissions after injury.
      • Moore L.
      • Stelfox H.T.
      • Turgeon A.F.
      • et al.
      Derivation and validation of a quality indicator for 30-day unplanned hospital readmission to evaluate trauma care.
      • Cameron C.M.
      • Purdie D.M.
      • Kliewer E.V.
      • McClure R.J.
      Ten-year outcomes following traumatic brain injury: a population-based cohort.
      Further exclusion criteria included patients who died in hospital from the index TBI, patients who resided outside of Ontario, and patients whose records contained unspecified sex information or erroneously recorded date of death. Ethics approval was received from the Toronto Rehabilitation Institute and Institute for Clinical Evaluative Sciences.

      Data sources

      Hospital admission data were accessed at the Institute for Clinical Evaluative Sciences using the Canadian Institute of Health Information Discharge Abstract Database (DAD),
      Canadian Institute for Health Information
      Data quality documentation: Discharge Abstract Database 2002-2003.
      which contains information on the following variables: age, sex, residential postal code, date of admission, date of discharge, primary ICD diagnostic codes, and secondary and tertiary ICD diagnostic codes. Index cases of TBI were identified from the DAD using data from 141 hospitals (ranging in size from 1 to 2743 patients being treated per institution). Excellent agreement for primary diagnoses has been found between administrative hospitalization data and chart audit.
      • Williams J.I.
      • Young W.
      A summary of studies on the quality of health care administrative databases in Canada.
      The Registered Persons Database was used to identify mortality information of participants. It contains the date of birth and cause and date of death of all persons with a valid Ontario Health Card. Patients were linked via unique encoded identifiers and analyzed at the Institute for Clinical Evaluative Sciences (100% match).

      Variables

      Information was extracted from the DAD and Registered Persons Database and categorized based on preinjury, injury, and postinjury study variables.

      Preinjury variables

      Demographic information (eg, age at the time of initial injury, sex, rurality) was collected. Age was divided into 5 groups: children (<15y), adolescents/young adults (15–24y), adults (25–49y, 50–64y), and seniors (>65y). Rural versus nonrural residence was determined using postal codes from administrative databases.
      • Kralj B.
      Measuring ‘rurality’ for the purposes of health care planning: an empirical measure for Ontario.

      Injury variables

      Variables collected at the time of injury included mechanism of injury, Abbreviated Injury Score,
      initial length of acute care stay based on the duration of the index hospitalization, comorbidity, and psychiatric comorbidity. The mechanisms of injury were identified using injury codes and grouped into 4 major categories: falls, motor vehicle collision (MVC), struck by/against, and other mechanism of injury. Discharge destination was also examined to describe the cohort. In Ontario, patients may be discharged to another facility, transferred to long-term care or ambulatory/palliative care, or discharged home with or without services. Rehabilitation can occur in acute care hospitals and/or at freestanding inpatient rehabilitation hospitals. Year of injury was added to multivariable models to control for practice changes over time.
      The Abbreviated Injury Score was used as a measure of TBI severity. Severity was measured on a 6-point scale and categorized as mild (1–2), moderate (3), or severe (≥4) injury. If a patient had >1 diagnosis of TBI on their hospital record, the injury with the highest severity was used to classify the severity. The Johns Hopkins Adjusted Clinical Group Case-Mix System was used to account for patient comorbidity.
      Johns Hopkins University
      The Johns Hopkins ACG system.
      • Weiner J.P.
      • Starfield B.H.
      • Steinwachs D.M.
      • Mumford L.M.
      Development and application of a population-oriented measure of ambulatory care case-mix.
      • Starfield B.
      • Mumford L.
      Ambulatory care groups: a categorization of diagnoses for research and management.
      This method of adjusting for case mix has previously been used in Canada
      • Reid R.J.
      • MacWilliam L.
      • Verhulst L.
      • Roos N.
      • Atkinson M.
      Performance of the ACG case-mix system in two Canadian provinces.
      and has been validated in the United States.
      • McCrone P.
      • Hallam A.
      • Knapp M.
      • et al.
      Service use and costs of supporting the most socially disabled patients in a hospital reprovision programme: a two-hospital comparison.
      The system uses individual-level data to assign measures of resource use and comorbidity from diagnoses during a specified time period, obtained from patient records. In the present study, the Adjusted Clinical Group algorithm used physician billing claims from the Ontario Health Insurance Plan administrative database and hospital admissions from 2 years prior to index to determine a comorbidity score for patients equal to the sum of the Aggregated Diagnosis Group (ADG) indicators. Psychiatric comorbidities were additionally examined as a separate variable.

      Outcome variables

      Transfers between facilities for the incident TBI were censored to minimize the potential for misclassifying readmissions. Therefore, for this analysis, patients with incident TBI between fiscal years 2002/2003 and 2009/2010 were followed-up from 4 days after the initial acute care discharge until March 31, 2013, to identify readmission to an acute care facility for a hospital stay. Emergency department visits were not included as a hospital admission; however, a 1-night admission was included as an outcome variable. Participants were followed from index injury up to a maximum of 36 months. ICD-10-CA codes at admission were examined to identify causes of hospitalization and categorized according to ICD-10-CA chapter headings.

      Data analyses

      Descriptive statistics of patient characteristics at time of index injury were stratified by age and sex. Rehospitalization rates were also stratified based on duration since index injury (1 vs 3y post-TBI). A multivariable logistic regression model was run using a backward entry method to determine which variables contributed to the prediction of rehospitalization rates at 1 year and 3 years. Rehospitalization because of pregnancy/childbirth was not included in the model because this was not considered a negative health event. The models were fit using a generalized estimating equation (autoregressive correlation structure) to account for the correlation between rehospitalization and index admission hospital (n=138; 3 hospitals had patients with only maternity outcomes). Adjusting by index hospital allowed for any undue influence related to the admission hospital (eg, common patient characteristics attending certain hospitals, differences in hospitals practices) to be covaried out of the models. Initial length of acute care stay was excluded from the model based on poor fit as determined by Hosmer-Lemeshow goodness of fit test. The Abbreviated Injury Score was the only measure of severity entered into the analyses. ADG was entered as a continuous and quadratic variable. Although the interaction term between age and sex was entered for both the 1-year and 3-year models, it was not significant and therefore was excluded from both analyses. A P value ≤.05 was considered statistically significant. Analyses were performed using SAS version 9.3.a

      Results

      Descriptive statistics

      There were 29,269 patients with TBI identified from the DAD records between fiscal years 2002/2003 through 2009/2010. Most patients with TBI were men (table 1). Falls were the most common mechanism of injury for patients <15 and >49 years of age, whereas patients between 15 and 49 years were, on average, more likely to sustain a TBI because of a MVC. Patterns were similar across men and women, with the exception that among the 25- to 49-year-old age group, falls and MVC were the most common mechanisms of injury for men and women, respectively. The percentage of severely injured patients increased by age, as did the number of comorbid conditions. Most patients had no psychiatric comorbidity (96.5%) and lived in a nonrural setting (83.9%).
      Table 1Patient characteristics at index time of injury stratified by age and sex (%)
      Characteristic<15y15–24y25–49y50–64y≥65y
      M (n=2637)F (n=1501)M (n=3453)F (n=935)M (n=5039)F (n=1616)M (n=3064)F (n=1273)M (n=5062)F (n=4689)
      Rurality
       Rural16.715.319.521.616.618.316.618.014.211.4
       Nonrural83.284.580.578.483.281.783.381.985.688.5
       Unknown0.10.10.00.00.20.10.10.20.10.1
      ADG comorbidity
       0–342.340.445.225.933.819.222.213.76.75.5
       4–533.532.631.729.630.425.524.821.815.816.7
       6–715.217.815.124.718.926.422.723.021.623.4
       8–96.16.15.811.710.015.715.620.021.821.3
       ≥102.83.02.28.16.813.214.821.534.233.0
      Psychiatric comorbidity
       Yes0.30.11.02.13.44.74.84.65.05.7
      Mechanism of injury
       MVC12.913.334.956.725.440.118.326.79.48.5
       Fall47.456.915.315.126.730.351.556.777.684.6
       Struck by/against19.214.129.812.125.411.812.65.33.61.9
       Other19.815.119.015.521.417.016.310.08.03.2
       Missing0.80.50.90.61.10.91.31.31.41.7
      Abbreviated Injury Scale
       Mild (1–2)50.153.037.438.131.132.121.727.413.217.8
       Moderate (3)5.35.610.69.214.114.513.013.410.513.1
       Severe (≥4)37.934.047.747.349.847.760.153.171.963.5
       Unknown6.77.34.35.54.95.75.26.14.35.6
      Length of stay (d)
       1–271.372.549.339.138.938.231.333.121.619.3
       3–515.815.320.720.021.920.022.622.422.120.7
       6–116.26.312.317.416.617.118.320.622.623.7
       ≥126.85.917.823.422.624.627.824.033.736.3
      Discharge destination
       Transferred to another facility3.72.98.512.512.512.614.514.018.314.4
       Transferred to long-term care1.21.15.46.37.59.51110.721.228.0
       Transferred to ambulatory care/palliative1.61.41.92.83.02.52.92.12.52.2
       Discharged home with services5.54.15.67.75.96.67.39.013.819.7
       Discharged home without services87.990.377.370.367.767.161.963.143.835.3
      Abbreviations: F, female; M, male.
      Of the 29,269 patients with TBI, 22.9% (n=6703) were rehospitalized within 1 year, and 35.5% (n=10,390) were rehospitalized within the 3 years after index injury. As seen in table 2, percentages of rehospitalization increased steadily by age group. Although there were no sex differences in the number of rehospitalizations at 1 year, women between the ages of 15 and 49 years had a near 15% increase in the number of rehospitalizations relative to men at 3-years follow-up. Overall, the diagnoses most often reported at the time of rehospitalization were in the injury/poisoning ICD category. Poisoning is defined as a drug overdose (eg, poisoning by systemic antibodies, poisoning by other systemic anti-infectives and antiparasitics, poisoning by anesthetics and therapeutic gas). These were not intentional self-poisonings that would have resulted from an attempted suicide.
      Table 2Rehospitalization within 1 and 3 years from index TBI, stratified by age and sex (%)
      Follow-up Period<15y15–24y25–49y50–64y≥65y
      M (n=2637)F (n=1501)M (n=3453)F (n=935)M (n=5039)F (n=1616)M (n=3064)F (n=1273)M (n=5062)F (n=4689)
      Hospitalization within 1y (%)7.86.311.814.017.921.624.624.337.335.1
      Hospitalization within 3y (%)12.911.118.428.627.135.938.038.356.154.3
      When stratified by age and sex, differences were observed in the causes of rehospitalization (table 3). Under the age of 15 years, boys were most likely to be readmitted because of injury/poisoning (χ21,909=3.13, P=.077), and girls were most likely to be readmitted for health status related to follow-up care (χ21,909=20.26, P<.001). Cause of rehospitalization for patients aged 15 to 24 years also varied by sex: for women, the main cause was pregnancy, childbirth, and puerperium, followed by injuries, poisoning, and other external causes; for men, the main cause was injuries, poisoning, and other external causes (χ21,1475=38.15, P<.001), followed by disorders of the musculoskeletal system or health status (P=.555). The main causes of rehospitalization for patients aged 25 to 49 years were injury/poisoning, followed by mental and behavioral disorders for men (χ21,4121=4.66, P=.031) and digestive disorders for women (χ21,4121=3.92, P=.048). For those 50 to 64 years of age, the most common cause of rehospitalization for both sexes was injury/poisoning, followed by circulatory complaints for men (χ21,3755=7.6, P=.006) and digestive complaints for women (χ21,3755=2.91, P=.088). Over the age of 64 years, circulatory problems were the most common cause for both men and women.
      Table 3Causes of rehospitalization based on ICD-10 codes, stratified by age and sex (%)
      Reason for Rehospitalization<15y15–24y25–49y50–64y≥65y
      M (n=610)F (n=299)M (n=1010)F (n=465)M (n=2838)F (n=1283)M (n=2619)F (n=1136)M (n=6183)F (n=5173)
      Injury/poisoning19.514.731.716.32517.416.817.312.815.5
      Digestive system97.77.398.510.410.312.288.8
      Abnormal findings10.711.45.95.87.299.58.59.510.8
      Health status
      Related to follow-up care/convalescence.
      916.111.611.87.57.16.85.46.86.7
      Respiratory16.215.44.4<24.457.1811.910.8
      Circulatory2<21.81.54.23.711.88.721.619.2
      Musculoskeletal5.9511.610.510.97.16.87.73.53.6
      Mental/behavioral3.34.38.14.111.99.68.17.33.33
      Nervous system7.53.77.22.27.55.55.743.82.7
      Pregnancy/childbirth0<2026.7090000
      Genitourinary system2<22.52.225.53.94.566.3
      Neoplasms150.9<21.52.84.55.24.93.3
      Infectious/parasitic4.661.5<22.21.82.32.733.2
      Endocrine/nutritional/metabolic1.8<21.63.42.62.33.74.82.93.8
      Other
      Causes that represented <4% of hospitalizations per age/sex group were collapsed into an other category.
      7.76.1444.73.82.83.71.92.4
      NOTE. Data were missing or invalid for <6 rehospitalizations. All other causes of rehospitalizations have been classified.
      Abbreviation: ICD-10-CA, International Classification of Diseases–10th Revision–Canadian Enhancement.
      Related to follow-up care/convalescence.
      Causes that represented <4% of hospitalizations per age/sex group were collapsed into an other category.

      Predictors of rehospitalization

      Multivariable logistic regression analysis using the number of rehospitalizations at 1 year, while controlling for index admission hospital, demonstrated that being a man, older age, greater injury severity, suffering from an unknown mechanism of injury, and ADG and psychiatric comorbidity were significant predictors of rehospitalization, whereas persons who were struck by or against or had a later year of index injury (2006–2008) were less likely to be hospitalized (table 4). An odds ratio for ADG was not reported given the quadratic nature of the variable. There was a trend for those with a fall to have a greater risk of rehospitalization. Similar risk factors were found at 3 years: age, sex, mechanism of injury, comorbidity, and later year of index injury. Although most predictors were similar to those that were found at 1 year, being stuck by or against and unknown mechanism of injury were no longer significant predictors. Additional variables that contributed to the 3-year model were falls and rural dwelling (table 5).
      Table 4Multivariable logistic regression analysis of predictors of rehospitalization within 1-year postindex injury (N=27,696)
      Autoregressive generalized estimating equation was used to control for index admission hospital.
      VariableOR95% CIP
      Female (vs male)0.890.83–0.95<.001
      Age 15–24y (vs <15y)1.751.48–2.06<.001
      Age 25–49y (vs <15y)2.492.16–2.87<.001
      Age 50–64y (vs <15y)2.912.56–3.30<.001
      Age ≥65y (vs <15y)3.903.43–4.44<.001
      Rural (yes vs no)1.070.94–1.22.300
      No. of ADGNDND<.001
      No. of ADG squaredNDND<.001
      Psychiatric comorbidity (vs none)1.701.48–1.96<.001
      Fall (vs MVC)1.121.00–1.25.056
      Struck by/against (vs MVC)0.840.73–0.97.014
      Other (vs MVC)0.990.92–1.07.865
      Unknown mechanism (vs MVC)1.451.14–1.86.003
      Injury severity moderate (vs mild)1.050.94–1.17.374
      Injury severity severe (vs mild)1.191.10–1.29<.001
      Year 2004 (vs 2003)0.920.85–1.01.071
      Year 2005 (vs 2003)0.890.79–1.01.072
      Year 2006 (vs 2003)0.870.78–0.98.023
      Year 2007 (vs 2003)0.870.78–0.97.013
      Year 2008 (vs 2003)0.930.86–1.00.045
      Year 2009 (vs 2003)0.930.82–1.06.263
      Abbreviations: CI, confidence interval; ND, not done; OR, odds ratio.
      Autoregressive generalized estimating equation was used to control for index admission hospital.
      Table 5Multivariable logistic regression analysis of predictors of rehospitalization within 3 years postindex injury (N=27,696)
      Autoregressive generalized estimating equation was used to control for index admission hospital.
      VariableOR95% CIP
      Female (vs male)0.910.85–0.96.002
      Age 15–24y (vs <15y)1.661.50–1.84<.001
      Age 25–49y (vs <15y)2.452.20–2.73<.001
      Age 50–64y (vs <15y)3.192.87–3.54<.001
      Age ≥65 (vs <15y)4.994.58–5.45<.001
      Rural (yes vs no)1.111.03–1.20.006
      No. of ADGNDND<.001
      No. of ADG squaredNDND<.001
      Psychiatric comorbidity (vs none)1.701.44–2.01<.001
      Fall (vs MVC)1.121.01–1.24.033
      Struck by/against (vs MVC)0.910.81–1.04.160
      Other (vs MVC)0.990.93–1.06.846
      Unknown mechanism (vs MVC)1.200.93–1.56.158
      Injury severity moderate (vs mild)1.080.98–1.18.131
      Injury severity severe (vs mild)1.091.02–1.17.012
      Year 2004 (vs 2003)0.920.84–1.01.074
      Year 2005 (vs 2003)0.860.77–0.95.005
      Year 2006 (vs 2003)0.850.78–0.93.001
      Year 2007 (vs 2003)0.870.80–0.95.001
      Year 2008 (vs 2003)0.930.84–1.02.123
      Year 2009 (vs 2003)0.940.85–1.05.258
      Abbreviations: CI, confidence interval; ND, not done; OR, odds ratio.
      Autoregressive generalized estimating equation was used to control for index admission hospital.

      Discussion

      Our results demonstrate that rehospitalizations are common among men and women after a TBI. Approximately 35.5% of patients were rehospitalized within the first 3 years after a TBI, resulting in 10,390 patients being rehospitalized. Common causes of rehospitalization varied by age group and sex, which highlights the importance of examining the causes of rehospitalization based on these factors. Risk factors for rehospitalization at both 1- and 3-year post-TBI were sex, age, mechanism of injury, injury severity, and general and psychiatric comorbidity. These findings provide a valuable framework for the development of targeted intervention programs for those at greatest risk for recurrent hospitalization after TBI.
      The rate of rehospitalization for patients with TBI in the present study was higher than previously reported rates of 16% to 23%.
      • Boutin A.
      • Francisque K.
      • Moore L.
      • Lauzier F.
      • Neveu X.
      • Turgeon A.
      Hospital readmissions following traumatic brain injury.
      • Moore L.
      • Stelfox H.T.
      • Turgeon A.F.
      • et al.
      Rates, patterns, and determinants of unplanned readmission after traumatic injury: a multicenter cohort study.
      • Nakase-Richardson R.
      • Tran J.
      • Cifu D.
      • et al.
      Do rehospitalization rates differ among injury severity levels in the NIDRR Traumatic Brain Injury Model Systems program?.
      • Cifu D.X.
      • Kreutzer J.S.
      • Marwitz J.H.
      • et al.
      Etiology and incidence of rehospitalization after traumatic brain injury: a multicenter analysis.
      • Marwitz J.H.
      • Cifu D.X.
      • Englander J.
      • High W.M.
      A multi-center analysis of rehospitalizations five years after brain injury.
      This might be because all hospitalizations could be effectively tracked within a publicly insured population. These rates are also consistent with rehospitalization rates for patients with spinal cord injury at 1-year postinjury based on data from the same database.
      • Jaglal S.B.
      • Munce S.E.
      • Guilcher S.J.
      • et al.
      Health system factors associated with rehospitalizations after traumatic spinal cord injury: a population-based study.
      Contrary to other trauma patients readmitted to hospital as a result of infection,
      • Dai Y.T.
      • Wu S.C.
      • Weng R.
      Unplanned hospital readmission and its predictors in patients with chronic conditions.
      • Morris D.S.
      • Rohrbach J.
      • Sundaram L.M.
      • et al.
      Early hospital readmission in the trauma population: are the risk factors different?.
      congestive heart failure, or chronic obstructive pulmonary disease,
      • Westert G.P.
      • Lagoe R.J.
      • Keskimäki I.
      • Leyland A.
      • Murphy M.
      An international study of hospital readmissions and related utilization in Europe and the USA.
      • Shipton S.
      Risk factors associated with multiple hospital readmissions.
      the sample in this study was more likely to be rehospitalized because of injury/poisoning. This is consistent with other studies that found that both adults
      • Cameron C.M.
      • Purdie D.M.
      • Kliewer E.V.
      • McClure R.J.
      Ten-year outcomes following traumatic brain injury: a population-based cohort.
      and children
      • Swaine B.R.
      • Tremblay C.
      • Platt R.W.
      • Grimard G.
      • Zhang X.
      • Pless I.B.
      Previous head injury is a risk factor for subsequent head injury in children: a longitudinal cohort study.
      with a TBI were more likely to suffer from a subsequent injury. Persons who sustain a TBI often report fatigue and dizziness,
      • Selassie A.W.
      • McCarthy M.L.
      • Ferguson P.L.
      • Tian J.
      • Langlois J.A.
      Risk of posthospitalization mortality among persons with traumatic brain injury, South Carolina 1999-2001.
      and these conditions could make them more prone to injury than other patient populations. The likelihood of reinjury accentuates the importance of targeting injury prevention programs for patients with TBI and additionally suggests that return to work or play could require greater long-term monitoring.
      Stratification of data by age and sex revealed important differences in the causes of rehospitalization. For instance, those <15 years old were more likely to be rehospitalized for diseases of the respiratory system, whereas those >64 years were more likely to be rehospitalized for diseases of the circulatory system. This is consistent with findings in the general population that circulatory problems are associated with older age.
      • Van Rensbergen G.
      • Nawrot T.
      Medical conditions of nursing home admissions.
      Women of childbearing years presented with a high number of hospitalizations related to pregnancy and childbirth. This indicates the extent to which women have children after TBI and further demonstrates the importance of clarifying the cause of rehospitalization. Between the ages of 15 and 49 years, men were often rehospitalized for musculoskeletal problems, and women were often rehospitalized for digestive problems. Adults <65 years also had a notable number of rehospitalizations for mental health and behavioral disorders. This information is important for prevention efforts and demonstrates the need for intervention programs tailored to specific age groups and sexes.
      Overall, the results indicate that the risk of rehospitalization increased with being a man, older age, greater injury severity, sustaining a TBI because of a fall, and having additional physical or mental health conditions. The interaction between age and sex was not significant at 1-year or 3-year follow-up. Age
      • Dai Y.T.
      • Wu S.C.
      • Weng R.
      Unplanned hospital readmission and its predictors in patients with chronic conditions.
      • Corrigan J.M.
      • Martin J.B.
      Identification of factors associated with hospital readmission and development of a predictive model.
      and longer initial length of acute care stay
      • Westert G.P.
      • Lagoe R.J.
      • Keskimäki I.
      • Leyland A.
      • Murphy M.
      An international study of hospital readmissions and related utilization in Europe and the USA.
      • Cakir B.
      • Gammon G.
      Evaluating readmission rates: how can we improve?.
      are well-known determinants of rehospitalization. Consistent with previous studies, being a woman is protective against rehospitalization.
      • Moore L.
      • Stelfox H.T.
      • Turgeon A.F.
      • et al.
      Derivation and validation of a quality indicator for 30-day unplanned hospital readmission to evaluate trauma care.
      • Moore L.
      • Stelfox H.T.
      • Turgeon A.F.
      • et al.
      Rates, patterns, and determinants of unplanned readmission after traumatic injury: a multicenter cohort study.
      Lack of adequate services/supports in rural areas may contribute to the increased number of readmissions among these residents. Nonetheless, these results must be interpreted with caution given that the odds ratio for rural versus nonrural dwelling was close to 1 at 3-year follow-up and was nonsignificant at 1-year follow-up. Previous research has also indicated that mental health conditions are among the most prevalent comorbidities found with TBI.
      • Kim H.
      • Colantonio A.
      • Chipman M.
      Traumatic brain injury occurring at work.
      Patients discharged from acute care hospitals with mental health conditions were found to have a higher rate of readmission than patients discharged without these conditions
      • Shipton S.
      Risk factors associated with multiple hospital readmissions.
      • Madi N.
      • Zhao H.
      • Li J.F.
      CIHI survey: hospital readmissions for patients with mental illness in Canada.
      ; as such, comorbidity, and more specifically psychiatric comorbidity, seen in the present study may be an important risk factor for rehospitalization.
      Relative to falls and other mechanisms of injury, the risk of rehospitalization was reduced if a person was involved in a MVC, controlling for age and injury severity. In Ontario, persons injured in a MVC might be eligible for supplemental medical benefits through mandatory automobile insurance; such benefits include case management and community-based services, which are not publicly funded to the same degree. An Ontario study by Kim et al
      • Kim H.
      • Colantonio A.
      • Deber R.
      • Vernich L.
      Discharge destination from acute care after traumatic brain injury.
      showed that persons with a TBI injured by a MVC were 56% more likely to be discharged home with support services than patients who sustained at TBI from other causes, after controlling for age, severity of injury, and comorbidities. The additional support may have potentially aided recovery and therefore prevented subsequent injuries for MVC patients, limiting the number of rehospitalizations.

      Study limitations

      A number of study limitations exist. First, the DAD does not include records of Ontario residents who sustained a TBI outside of the province, who were hospitalized outside of Ontario, who died prior to being taken to an emergency department or admitted to hospital, or who were treated for a TBI in physician offices or prisons. Another limitation of using administrative data is the possibility that a TBI diagnosis may be missed and that certain factors that capture severity (eg, Glasgow Coma Score), which are not mandatory to report, may be underreported. This could result in underestimates of rehospitalization after TBI. Patients with unknown injury severity were also excluded from the multivariable analysis; although this was a limitation in our study, the number missed represented only 5% of the cohort. Another limitation was the difficulty determining whether the rehospitalization had to do with a consequence of the index injury, whereby a TBI diagnostic code would be included in the discharge abstract, or because it was a new TBI. Nevertheless, the primary purpose of our investigation was to determine the long-term hospital utilization of patients with a TBI diagnosis and not specifically to examine reinjury.

      Conclusions

      Our findings demonstrate that the rates of rehospitalization after a TBI are higher than previously reported and that a number of factors can increase the risk of being rehospitalized. Prevention efforts for reinjury and mental health issues are important for the TBI population. The study also suggests that persons with access to more services are less likely to be hospitalized, supporting the need for better postacute care across geographic areas. In addition, greater support is needed for more vulnerable persons, particularly those living with mental health issues, which should be tailored by age and sex. Future work should seek to identify rehospitalization risk using other health care services after TBI (eg, emergency departments, outpatient facilities) and address the causes of reinjury using a longitudinal approach.

      Supplier

      • a.
        SAS version 9.3; SAS Institute.

      References

        • Faul M.
        • Xu L.
        • Wald M.M.
        • Coronado V.
        Traumatic brain injury in the United States: emergency department visits, hospitalizations, and deaths, 2002–2006.
        CDC, National Center for Injury Prevention and Control, Atlanta2010
        • Finkelstein E.
        • Corso P.
        • Miller T.
        • et al.
        The incidence and economic burden of injuries in the United States.
        Oxford Univ Pr, New York2006
        • Jencks S.F.
        • Williams M.V.
        • Coleman E.A.
        Rehospitalizations among patients in the Medicare fee-for-service program.
        N Engl J Med. 2009; 360: 1418-1428
        • Weissman J.S.
        • Ayanian J.Z.
        • Chasan-Taber S.
        • Sherwood M.J.
        • Roth C.
        • Epstein A.M.
        Hospital readmissions and quality of care.
        Med Care. 1999; 37: 490-501
        • Canadian Institute for Health Information
        All cause readmission to acute care and return to the emergency department.
        Canadian Institute for Health Information, Ottawa2012
        • Dai Y.T.
        • Wu S.C.
        • Weng R.
        Unplanned hospital readmission and its predictors in patients with chronic conditions.
        J Formos Med Assoc. 2002; 101: 779-785
        • Worrell S.S.
        • Koepsell T.D.
        • Sabath D.R.
        • Gentilello L.M.
        • Mock C.N.
        • Nathens A.B.
        The risk of reinjury in relation to time since first injury: a retrospective population-based study.
        J Trauma. 2006; 60: 379-384
        • Moore L.
        • Stelfox H.T.
        • Turgeon A.F.
        • et al.
        Derivation and validation of a quality indicator for 30-day unplanned hospital readmission to evaluate trauma care.
        J Trauma Acute Care Surg. 2014; 76: 1310-1316
        • Westert G.P.
        • Lagoe R.J.
        • Keskimäki I.
        • Leyland A.
        • Murphy M.
        An international study of hospital readmissions and related utilization in Europe and the USA.
        Health Policy. 2002; 61: 269-278
        • Cakir B.
        • Gammon G.
        Evaluating readmission rates: how can we improve?.
        South Med J. 2010; 103: 1079-1083
        • Shipton S.
        Risk factors associated with multiple hospital readmissions.
        Home Care Provid. 1996; 1: 83-85
        • Boutin A.
        • Francisque K.
        • Moore L.
        • Lauzier F.
        • Neveu X.
        • Turgeon A.
        Hospital readmissions following traumatic brain injury.
        Crit Care. 2013; 17: P331
        • Moore L.
        • Stelfox H.T.
        • Turgeon A.F.
        • et al.
        Rates, patterns, and determinants of unplanned readmission after traumatic injury: a multicenter cohort study.
        Ann Surg. 2014; 259: 374-380
        • Nakase-Richardson R.
        • Tran J.
        • Cifu D.
        • et al.
        Do rehospitalization rates differ among injury severity levels in the NIDRR Traumatic Brain Injury Model Systems program?.
        Arch Phys Med Rehabil. 2013; 94: 1884-1890
        • Cifu D.X.
        • Kreutzer J.S.
        • Marwitz J.H.
        • et al.
        Etiology and incidence of rehospitalization after traumatic brain injury: a multicenter analysis.
        Arch Phys Med Rehabil. 1999; 80: 85-90
        • Marwitz J.H.
        • Cifu D.X.
        • Englander J.
        • High W.M.
        A multi-center analysis of rehospitalizations five years after brain injury.
        J Head Trauma Rehabil. 2001; 16: 307-317
        • Cameron C.M.
        • Purdie D.M.
        • Kliewer E.V.
        • McClure R.J.
        Ten-year outcomes following traumatic brain injury: a population-based cohort.
        Brain Inj. 2008; 22: 437-449
        • Swaine B.R.
        • Tremblay C.
        • Platt R.W.
        • Grimard G.
        • Zhang X.
        • Pless I.B.
        Previous head injury is a risk factor for subsequent head injury in children: a longitudinal cohort study.
        Pediatrics. 2007; 119: 749-758
        • Colantonio A.
        • Croxford R.
        • Farooq S.
        • Laporte A.
        • Coyte P.C.
        Trends in hospitalization associated with traumatic brain injury in a publicly insured population, 1992-2002.
        J Trauma. 2009; 66: 179-183
        • Colantonio A.
        • LaPorte A.
        • Croxford R.
        • Coyte P.
        Who receives in-patient rehabilitation after traumatic brain injury? A population-based study.
        Arch Phys Med Rehabil. 2005; 86: E21-E22
        • Canadian Institute for Health Information
        Data quality documentation: Discharge Abstract Database 2002-2003.
        Canadian Institute for Health Information, Ottawa2004
        • Williams J.I.
        • Young W.
        A summary of studies on the quality of health care administrative databases in Canada.
        in: Goel V. Williams J.I. Anderson G.M. Patterns of health care in Ontario: the ICES practice atlas. 2nd ed. Canadian Medical Association, Ottawa1996: 339-346
        • Kralj B.
        Measuring ‘rurality’ for the purposes of health care planning: an empirical measure for Ontario.
        Ont Med Rev. 2000; 67: 37-40
      1. Association for the Advancement of Automotive Medicine. Abbreviated Injury Scale 2005, update 2008. Association for Advancement of Automatic Medicine, Barrington2008
        • Johns Hopkins University
        The Johns Hopkins ACG system.
        John Hopkins Univ, Baltimore2013
        • Weiner J.P.
        • Starfield B.H.
        • Steinwachs D.M.
        • Mumford L.M.
        Development and application of a population-oriented measure of ambulatory care case-mix.
        Med Care. 1991; 29: 452-472
        • Starfield B.
        • Mumford L.
        Ambulatory care groups: a categorization of diagnoses for research and management.
        Health Serv Res. 1991; 26: 53-74
        • Reid R.J.
        • MacWilliam L.
        • Verhulst L.
        • Roos N.
        • Atkinson M.
        Performance of the ACG case-mix system in two Canadian provinces.
        Med Care. 2001; 39: 86-99
        • McCrone P.
        • Hallam A.
        • Knapp M.
        • et al.
        Service use and costs of supporting the most socially disabled patients in a hospital reprovision programme: a two-hospital comparison.
        Soc Psychiatry Psychiatr Epidemiol. 2006; 41: 656-661
        • Jaglal S.B.
        • Munce S.E.
        • Guilcher S.J.
        • et al.
        Health system factors associated with rehospitalizations after traumatic spinal cord injury: a population-based study.
        Spinal Cord. 2009; 47: 604-609
        • Morris D.S.
        • Rohrbach J.
        • Sundaram L.M.
        • et al.
        Early hospital readmission in the trauma population: are the risk factors different?.
        Injury. 2014; 45: 56-60
        • Selassie A.W.
        • McCarthy M.L.
        • Ferguson P.L.
        • Tian J.
        • Langlois J.A.
        Risk of posthospitalization mortality among persons with traumatic brain injury, South Carolina 1999-2001.
        J Head Trauma Rehabil. 2005; 20: 257-269
        • Van Rensbergen G.
        • Nawrot T.
        Medical conditions of nursing home admissions.
        BMC Geriatr. 2010; 10: 46
        • Corrigan J.M.
        • Martin J.B.
        Identification of factors associated with hospital readmission and development of a predictive model.
        Health Serv Res. 1992; 27: 81-101
        • Kim H.
        • Colantonio A.
        • Chipman M.
        Traumatic brain injury occurring at work.
        NeuroRehabilitation. 2006; 21: 269-278
        • Madi N.
        • Zhao H.
        • Li J.F.
        CIHI survey: hospital readmissions for patients with mental illness in Canada.
        Healthc Q. 2007; 10: 30-32
        • Kim H.
        • Colantonio A.
        • Deber R.
        • Vernich L.
        Discharge destination from acute care after traumatic brain injury.
        Can J Neurol Sci. 2006; 33: 48-52