1
Mortality and Life Expectancy after Traumatic Brain Injury: The Influence of Demographic, Etiology, Discharge Disability, and Socio- environmental Factors James S. Krause, PhD 1 ; Yue Cao, PhD 1 ; Cindy Harrison-Felix, PhD 2 ; Lee L. Saunders, PhD 1 ; Gale Whiteneck, PhD 2 1 Medical University of South Carolina, Charleston, SC; 2 Craig Hospital, Englewood, CO Objective Our purpose was to identify factors associated with differential risk of mortality and life expectancy after traumatic brain injury (TBI). Our objectives were to: 1)Identify the effects of demographic and injury characteristics with mortality. 2)Identify the effects of the socio-environmental factors with mortality, after controlling for demographic and injury characteristics. Methods Participants were adults (18 years or older) who sustained a TBI July, 2001 to December, 2009 and were alive 1-year post-injury. Identification was through the TBI Model Systems (TBIMS) National Database, a network of institutions conducting research and providing specialty care in the United States. Mortality was identified using the Social Security Death Index. All data on risk and protective factors were taken from the Form I, collected during inpatient hospitalization. Person year logistic regression analysis was used to identify the odds of mortality. The predictors in the logistic regression analysis included three demographic variables (gender, race, age at injury), three disability variables (independence in feeding, walking, and ability to follow motor commands), two types of injury etiology (fall, violence), and three socio-environmental variables (marital status, education, and pre-injury income). Mortality status was the outcome variable in the logistic regression analysis. Results There were 5,806 1-year survivors with 19,683 person-years and 362 deaths. After excluding those with missing data, there were 17,522 person-years and 333 deaths. The rescaled-R 2 was 0.15 and the C-statistic (concordance) was 0.81. Risk factors for mortality included being male, being white, and older ages (table 2). Being injured as the result of a fall or an act of violence were also risk factors, as was non- independence in feeding. There was a non- significant trend for those married to have lower odds of mortality. Walking ability, ability to follow motor commands, and education (bachelor’s degree) were all non-significant. Compared to those unemployed, only the lowest income of less than $10,000/yr. was not significant. The odds of mortality, compared with those who were not working, decreased with each successive level of income starting with $10,000- $29,999 where the odds of mortality was only 0.58 Table 2. Odds Ratio Estimates Effect Point Estimate 95% Wald Confidence L imits Male (vs. Female) 1.83 1.40 2.38 White (vs. Non-white) 1.32 1.00 1.73 Age (vs. <35 years) 35-44 years 2.81 1.74 4.54 45-54 years 5.30 3.44 8.16 55-64 years 5.00 3.08 8.13 65-74 years 9.06 5.57 14.74 75+ years 15.19 9.47 24.38 Non-independent feeding (vs. independent ) 1.51 1.18 1.94 Walking ability (vs. independent) No walking ability 1.27 0.87 1.85 Non-independent walking 1.22 0.91 1.63 Unable to follow motor commands (vs. able) 1.57 0.71 3.48 Etiology (vs. other) Violence 1.78 1.20 2.62 Fall 1.61 1.22 2.12 Married (vs. non-married) 0.80 0.63 1.02 Annual earnings (vs. not competitively employed) $9,999 or less 0.78 0.48 1.28 $10,000-$29,999 0.58 0.40 0.85 $30,000-$49,999 0.45 0.29 0.71 ≥ $50,000 0.30 0.19 0.50 Bachelor’s degree (vs. other) 0.87 0.63 1.18 Discussion Among one year TBI survivors, a number of factors related to differential mortality. Age at injury onset was strongly related to mortality. Gender (female) was particularly protective of mortality. Although there was a greater odds of mortality among whites, these differences may relate to the differential validity of the mortality search (i.e., more information on whites). The strength of the etiologic factors with mortality and significance of one of the disability indicators (independence in feeding) indicate the importance of the nature and severity of the disability resulting from the TBI itself. Independence in feeding could be a focal point for intervention. We cannot determine the extent to which the relationship of etiology with mortality relates to differences in complications, as opposed to behavioral and personality factors leading to those types of injuries. A particularly important finding was that pre- injury income was a powerful predictor of future mortality. The availability of income has been both conceptually and empirically linked with longevity after neurologic injury. 1,2,3 Access to income may come from multiple sources including earnings and settlements, and these resources may be used as a protective factor for mortality. Methodological Considerations First, all data were collected during inpatient rehabilitation, so the status on some variables could have changed since first measured. Similarly, we used person year logistic regression which assumes no change in status of the variables. Further, whereas the majority of studies of income after neurologic injury use post-injury income, we used pre-injury earnings. In terms of mortality status, the TBIMS used the Social Security Death Index which has an approximate two-month lag in information on mortality. The National Death Index may be more reliable, but also is more expensive and has a greater lag in terms of mortality updates. Lastly, we chose to include only participants who were 18 and older, as the socio- environmental variables often develop throughout adulthood. For instance, many of those who are unmarried will become married. Participants who were younger are likely to increase both their education and income over time. Future Research Additional research is needed that updates the status of predictors based on changes after TBI. This is particularly relevant for socio- environmental factors that may change over time. References The TBI Model Systems National Database is supported by the U.S. Department of Education, National Institute on Disability and Rehabilitation Research (NIDRR) in collaboration with the TBI Model Systems Centers. The contents of this presentation were also developed under a grant from the Department of Education, NIDRR grant number H133A080064. However, these contents do not necessarily reflect the opinions or views of the TBI Model Systems Centers, NIDRR, or the U.S. Department of Education. For more information, please visit our website: www.longevityafterinjury.com or contact Dr. James Krause ([email protected] ). Table 1. Characteristics of Dependent and Independent Variables Variables Frequency Percentage Deceased 362 6.2 Female 1534 26.4 White 4066 70.0 Age <35 years 2548 43.9 35-44 years 945 16.3 45-54 years 935 16.1 55-64 years 598 10.3 65-74 years 371 6.4 75+ years 409 7.0 Non-independent feeding 2045 35.4 Walking ability Independent walking 2148 37.4 No walking ability 2524 43.9 Non-independent walking 1075 18.7 Unable to follow motor commands 86 1.5 Etiology Violence 657 11.3 Fall 1531 26.4 Other 3618 62.3 Married 2017 34.7 Annual earnings Not competitively employed 1890 35.0 $9,999 or less 584 10.8 $10,000-$29,999 1340 24.8 $30,000-$49,999 821 15.2 ≥ $50,000 763 14.1 Bachelor’s degree 862 15.0

Mortality and Life Expectancy after Traumatic Brain Injury: The Influence of Demographic, Etiology, Discharge Disability, and Socio-environmental Factors

Embed Size (px)

Citation preview

Page 1: Mortality and Life Expectancy after Traumatic Brain Injury: The Influence of Demographic, Etiology, Discharge Disability, and Socio-environmental Factors

Mortality and Life Expectancy after Traumatic Brain Injury:The Influence of Demographic, Etiology, Discharge Disability, and Socio-

environmental Factors

James S. Krause, PhD1; Yue Cao, PhD1; Cindy Harrison-Felix, PhD2; Lee L. Saunders, PhD1; Gale Whiteneck, PhD2

1Medical University of South Carolina, Charleston, SC; 2Craig Hospital, Englewood, CO

ObjectiveOur purpose was to identify factors associated with differential risk of mortality and life expectancy after traumatic brain injury (TBI). Our objectives were to:1)Identify the effects of demographic and injury characteristics with mortality.2)Identify the effects of the socio-environmental factors with mortality, after controlling for demographic and injury characteristics.

MethodsParticipants were adults (18 years or older) who sustained a TBI July, 2001 to December, 2009 and were alive 1-year post-injury. Identification was through the TBI Model Systems (TBIMS) National Database, a network of institutions conducting research and providing specialty care in the United States. Mortality was identified using the Social Security Death Index. All data on risk and protective factors were taken from the Form I, collected during inpatient hospitalization. Person year logistic regression analysis was used to identify the odds of mortality. The predictors in the logistic regression analysis included three demographic variables (gender, race, age at injury), three disability variables (independence in feeding, walking, and ability to follow motor commands), two types of injury etiology (fall, violence), and three socio-environmental variables (marital status, education, and pre-injury income). Mortality status was the outcome variable in the logistic regression analysis.

ResultsThere were 5,806 1-year survivors with 19,683 person-years and 362 deaths. After excluding those with missing data, there were 17,522 person-years and 333 deaths. The rescaled-R2 was 0.15 and the C-statistic (concordance) was 0.81.

Risk factors for mortality included being male, being white, and older ages (table 2). Being injured as the result of a fall or an act of violence were also risk factors, as was non-independence in feeding. There was a non-significant trend for those married to have lower odds of mortality. Walking ability, ability to follow motor commands, and education (bachelor’s degree) were all non-significant.

Compared to those unemployed, only the lowest income of less than $10,000/yr. was not significant. The odds of mortality, compared with those who were not working, decreased with each successive level of income starting with $10,000-$29,999 where the odds of mortality was only 0.58 compared to those who were unemployed at injury. The odds of mortality for those with incomes of $30,000-$49,999 and ≥$50,000 were less than half of that of those who were unemployed.

 Table 2. Odds Ratio Estimates

EffectPoint

Estimate

95% WaldConfidence Li

mitsMale (vs. Female) 1.83 1.40 2.38White (vs. Non-white) 1.32 1.00 1.73Age (vs. <35 years)

35-44 years 2.81 1.74 4.5445-54 years 5.30 3.44 8.1655-64 years 5.00 3.08 8.1365-74 years 9.06 5.57 14.7475+ years 15.19 9.47 24.38

Non-independent feeding (vs. independent ) 1.51 1.18 1.94Walking ability (vs. independent)

No walking ability 1.27 0.87 1.85Non-independent walking 1.22 0.91 1.63

Unable to follow motor commands (vs. able) 1.57 0.71 3.48Etiology (vs. other)

Violence 1.78 1.20 2.62Fall 1.61 1.22 2.12

Married (vs. non-married) 0.80 0.63 1.02Annual earnings (vs. not competitively employed)

$9,999 or less 0.78 0.48 1.28$10,000-$29,999 0.58 0.40 0.85$30,000-$49,999 0.45 0.29 0.71≥ $50,000 0.30 0.19 0.50

Bachelor’s degree (vs. other) 0.87 0.63 1.18

DiscussionAmong one year TBI survivors, a number of factors related to differential mortality. Age at injury onset was strongly related to mortality. Gender (female) was particularly protective of mortality. Although there was a greater odds of mortality among whites, these differences may relate to the differential validity of the mortality search (i.e., more information on whites). The strength of the etiologic factors with mortality and significance of one of the disability indicators (independence in feeding) indicate the importance of the nature and severity of the disability resulting from the TBI itself. Independence in feeding could be a focal point for intervention. We cannot determine the extent to which the relationship of etiology with mortality relates to differences in complications, as opposed to behavioral and personality factors leading to those types of injuries.

A particularly important finding was that pre-injury income was a powerful predictor of future mortality. The availability of income has been both conceptually and empirically linked with longevity after neurologic injury.1,2,3 Access to income may come from multiple sources including earnings and settlements, and these resources may be used as a protective factor for mortality.

Methodological ConsiderationsFirst, all data were collected during inpatient rehabilitation, so the status on some variables could have changed since first measured. Similarly, we used person year logistic regression which assumes no change in status of the variables. Further, whereas the majority of studies of income after neurologic injury use post-injury income, we used pre-injury earnings. In terms of mortality status, the TBIMS used the Social Security Death Index which has an approximate two-month lag in information on mortality. The National Death Index may be more reliable, but also is more expensive and has a greater lag in terms of mortality updates. Lastly, we chose to include only participants who were 18 and older, as the socio-environmental variables often develop throughout adulthood. For instance, many of those who are unmarried will become married. Participants who were younger are likely to increase both their education and income over time.

Future ResearchAdditional research is needed that updates the status of predictors based on changes after TBI. This is particularly relevant for socio-environmental factors that may change over time.

References1.Krause JS, Saunders LL. Life expectancy estimates in the life care plan: Accounting for economic factors. J Life Care Plan. 2010;9(2):15-28.2.Krause JS, Saunders LL, DeVivo MJ. Income and risk of mortality after spinal cord injury. Arch Phys Med Rehabil. 2011;92(3):339-345.3.Harrison-Felix C, Whiteneck G, DeVivo M, Hammond FM, Jha A. Mortality following rehabilitation in the Traumatic Brain Injury Model Systems of Care. NeuroRehabil. 2004;19(1):45-54.

The TBI Model Systems National Database is supported by the U.S. Department of Education, National Institute on Disability and Rehabilitation Research (NIDRR) in collaboration with the TBI Model Systems Centers. The contents of this presentation were also developed under a grant from the Department of Education, NIDRR grant number H133A080064. However, these contents do not necessarily reflect the

opinions or views of the TBI Model Systems Centers, NIDRR, or the U.S. Department of Education. For more information, please visit our website: www.longevityafterinjury.com or contact Dr. James Krause ([email protected]).

Table 1. Characteristics of Dependent and Independent Variables

Variables Frequency PercentageDeceased 362 6.2Female 1534 26.4White 4066 70.0Age    

<35 years 2548 43.935-44 years 945 16.345-54 years 935 16.155-64 years 598 10.365-74 years 371 6.475+ years 409 7.0

Non-independent feeding 2045 35.4Walking ability    

Independent walking 2148 37.4No walking ability 2524 43.9Non-independent walking 1075 18.7

Unable to follow motor commands

86 1.5

Etiology    Violence 657 11.3Fall 1531 26.4Other 3618 62.3

Married 2017 34.7Annual earnings    

Not competitively employed 1890 35.0$9,999 or less 584 10.8$10,000-$29,999 1340 24.8$30,000-$49,999 821 15.2≥ $50,000 763 14.1

Bachelor’s degree 862 15.0