Marital Stability Over 10 Years Following Traumatic Brain Injury

Abstract

Objective: 

To examine the stability of marriage from the time of traumatic brain injury (TBI) to 10 years postinjury.

Design: 

Retrospective cohort.

Setting: 

TBI Model Systems centers.

Participants: 

In total, 1423 participants in the TBI Model Systems National Database who experienced TBI 10 years prior and were married at the time of injury.

Interventions: 

Not applicable.

Main Outcome Measure: 

Remaining married to the same partner from the time of injury to 10-year follow-up.

Results: 

At 10 years, 66% (938) remained married to the same person. Significant associations were found with age at injury (P < .0001), sex (P = .0028), and preinjury problematic substance use (P = .0092). Marital stability over the 10-year period was higher for those who were older, were female, and had no problematic substance use history. Marital instability was greatest in the first year postinjury.

Conclusions: 

Most married adults who received inpatient rehabilitation for TBI remained married to the same individual 10 years later. Those who were younger, were male, and had a history of problematic substance use were at a highest risk for relationship dissolution. Findings have implications for content, timing, and delivery of marital interventions. Substance use education and prevention appear to be important aspects of marital support.

MARITAL STABILITY has been shown to play a vital role in community reentry, return to work, adaptive emotional functioning, and rehabilitation success in persons with traumatic brain injury (TBI).¹ TBI often leads to changes (eg, in cognition, social-emotional functioning, personality, financial status, role changes) that may strain relationships.^2,3 However, research characterizing marital stability after TBI is limited. We found 10 studies^4–13 that examined marital stability after TBI. Marital stability has been operationally defined as staying married since the time of injury and not separating or divorcing. Marital stability varied widely from 22% to 85% as a function of study factors. Differences in samples may partially account for the inconsistencies across studies. One critical variation is sample size, which ranged from 9 to 977, with 50% of sample sizes of fewer than 50 (n = 9, 10, 23, 31, and 48),^4–7 restricting the validity and generalizability of findings. When only studies with more than 100 participants are taken into consideration (n = 5),^5,8–12 the marital stability findings become less disparate, spanning 51% to 85% remaining married. Another difference across studies is time postinjury. Of the 5 studies with sample sizes of more than 100, 2 studies examined marital stability at 2 years post-TBI,^8,9 1 study examined participants on average 4 years postinjury,¹⁰ and 2 studies included participants who were an average of 8 years postinjury.^11,12 Confounding differences in the follow-up period is the historical epoch during which the studies took place—varying by more than 25 years.

The 2 studies that reported on marital stability at 2 years post-TBI^8,9 were samples from Traumatic Brain Injury Model Systems (TBIMS) databases; one study was from the civilian TBIMS database (n = 977) (which overlaps with some participants in the present study), and the other from the VA TBIMS database (n = 357). The civilian TBIMS study found that 85% were still married 2 years postinjury, while the VA TBIMS study reported a slightly lower but similar finding of 78% marital stability. Kreutzer and colleagues¹⁰ examined marital status in 120 couples with mild to severe TBI and found that 75% were still married 2.5 to 8 years postinjury (mean = 4.1 years). Of the 2 studies that involved participants who were an average of 8 years post-TBI, one was of a civilian sample of 131 participants recruited in the United Kingdom¹² and the other involved 626 veterans with mild TBI.¹¹ Interestingly, both studies reported a 51% stability rate. It appears that the longer the study period, lower the marital stability. However, these findings allow limited conclusions due to a variety of confounding factors, and further investigation is warranted of time postinjury as a factor affecting marital stability. In previous studies, when the sample was studied less than (or, on average, less than) 5 years after the injury, stability rates ranged between 75% and 85%. However, in studies that were conducted in a sample when time postinjury was greater (average of 8 years), stability percentages were lower, indicating approximately only half of couples stayed together. These findings raise the question whether marital stability declines with time post-TBI.

Beyond the likelihood of marital stability after TBI, we also sought to understand predictors of marital stability. Two studies^8,9 found younger age, male gender, those belonging to racial or ethnic minority group, moderate injury severity, violent injury, lower education levels, and higher mental health service use prior to injury to be associated with marital instability. Kreutzer and colleagues¹⁰ reported similar predictors 4 years postinjury and also found that the longer the marriage preinjury, the more likely marital stability postinjury. Similarly, Wood and Yurdakul¹² found that the longer the relationship prior to injury, the more improved the couple's chances of staying together postinjury. In veterans¹¹ 8 years post–mild TBI, working full-time and not abusing substances were associated with greater marital stability.

No published data were found relative to rate or timing of marital dissolution following TBI. Although timing of dissolution was not examined, Wood and Yurdakul¹² found in a small convenience sample that average time post-TBI was significantly greater for participants who were separated or divorced (6.16 years) than those who stayed together (4.81 years).¹²

Given the potential for greater instability with increased time postinjury, there is need to examine stability in later years postinjury. To date, marital stability data 10 years post-TBI have not been published. While 2 studies have reported on samples that were on average 8 years postinjury, there are limitations to generalization. For instance, the large veteran study of 626 participants with mild TBI may not generalize to civilians. Veterans often have unique, polytraumatic injuries and life stressors that can influence marital stability (e.g., posttraumatic stress disorder, physical injuries, past deployments).⁹ Cultural differences may play a role in marital outcomes, including the country in which the study takes place.^5,6,12

Given the relevance of marital stability to living well after brain injury, combined with the variability and short postinjury duration of prior studies, it is important to assess marital stability over a longer time span. The objective of the current study was to extend findings from the civilian TBIMS reported at 2 years post-TBI to examine marital stability, rate of change, and predictors of marital stability over 10 years postinjury. Taking into consideration findings of prior marital studies in the TBI population, as well as US general population marriage data (with the exception of race/ethnicity), we hypothesized that marital dissolution would be associated with younger age, male gender, belonging to racial or ethnic minority group, lower education level, and substance abuse.^8,9,14–16 Also supported by the TBI literature, we further hypothesized associations of marital dissolution with violent cause of injury and moderate injury severity.^8,9 Finally, we hypothesized relations with greater motor, cognitive, and overall functional dependence with the presumption that these would impose greater care burden.

METHODS

Participants

Participants were individuals with TBI consecutively enrolled into the TBIMS National Database based on the following enrollment criteria: age at least 16 years at the time of injury; moderate-severe TBI (defined as posttraumatic amnesia [PTA] >24 hours, trauma-related intracranial neuroimaging abnormalities, loss of consciousness exceeding 30 minutes, or Glasgow Coma Scale [GCS] score in the emergency department of <13); and received acute care hospitalization within 72 hours, followed by inpatient rehabilitation in designated TBIMS facilities. For the current retrospective study, participants needed to be married at the time of injury, eligible for their 10-year follow-up window between June 25, 1998, and June 19, 2019, and had marital status data collected at baseline and 10-year follow-up. Participants who were widowed at any follow-up were excluded. Each center received approval for human subjects research from their respective institutional review boards.

Measures

The primary outcome was the participant's marital stability determined by marital status across each assessment epoch (at injury; years 1, 2, 5, and 10). If the participant's marital status remained “married” to the same partner at each time point, the participant was considered married throughout. If the participant reported having a divorce or being separated at any time point, then that participant was classified as not married throughout.

Several demographic and injury-related characteristics were considered as predictive variables for marital stability including the participant's age at injury, sex, race/ethnicity, education level, employment status, problematic substance use, and cause of injury. Injury severity was measured using the GCS, number of days in PTA, number of days not following commands (time to follow command [TFC]), length of stay in acute care (LOS acute), length of rehabilitation stay without interruptions (LOS rehab), and total length of stay (LOS total). Functional independence was measured with the Disability Rating Scale and the Motor, Cognitive, and Total FIM (Functional Independence Measure) scores (with possible scores ranging for Motor [13-91], Cognitive [5-35], and Total [18-126]). Sex was used as a binary variable. Race/ethnicity was categorized as White, Black, Hispanic, and other. The participant's education level was classified as less than a high school diploma; high school diploma or General Education Development (GED); associate or bachelor's degree; and graduate degree. Employment status was dichotomized as employed and not employed. The participant's cause of injury was grouped as vehicular, violence, falls, and other. GCS score¹⁷ was categorized as mild (13-15), moderate (9-12), and severe (3-8). Problematic substance use was dichotomized as “yes” or “no,” defined as heavy alcohol consumption (ie, >14 drinks per month for males and >7 drinks per month for females), or use of illicit drugs, or binge drinking in the past month.^18,19

Data analysis

All statistical analyses were conducted using SAS v.9.4 (SAS Institute Inc, Cary, North Carolina),²⁰ assuming a significance level of α = .05, unless otherwise specified. The demographic and injury characteristics for the sample were summarized using means and standard deviations for the continuous variables and frequency counts and percentages for the categorical variables.

Bivariate relations were examined between each of the demographic and injury characteristics with marital stability using logistic regression models. With marital stability as the outcome, the results were summarized using odds ratios (OR), 95% CIs, concordance index (C-Index), and significance level (P value). For the categorical variables, the relations with marital stability were displayed between each of the subgroups, and pairwise comparisons were tested using a Bonferroni correction.

After the bivariate analysis, a multivariate logistic regression model was created for marital stability. This model controlled for the participant's age at injury, sex, race/ethnicity, education level, employment status, problematic substance use, cause of injury, TFC days, FIM Motor score, and FIM Cognitive score. The model results were also summarized using OR, 95% CI, C-Index, and P value. Pairwise comparisons were again tested using a Bonferroni correction.

RESULTS

Sample

In total, 1923 TBIMS participants were married at the time of injury and eligible for their 10-year follow-up window between June 25, 1998, and June 19, 2019. Of these, 16 were not eligible as they reported being single (never married) at any follow-up and 109 were not eligible having been widowed at any follow-up, leaving 1798 eligible participants. Three hundred seventy-five participants were excluded because of missing marital status data at 10-year follow-up for a final sample of 1423. Of 1798 eligible participants, 1423 (79.1%) had marital status data. Reasons for being excluded were being lost to follow-up (15.8%), withdrew from the study (2.6%), incarcerated (1.1%), refused the interview (0.8%), and followed but marital status data were missing (0.1%). The demographic characteristics of those eligible but not included resembled those of the analytic sample except that they were on average 2.2 years younger, more likely male (78.1% vs 75.3%) and non-White (42.5% vs 22.3%), less educated, and more likely unemployed at injury (26.6% vs 18.7%). Injury severity characteristics did not differ, nor did acute and rehabilitation length of stay and discharge functional status. Those not included were slightly less likely to be injured in a vehicular crash (51.8% vs 58.4%) and more likely to be injured in a violent act (11.9% vs 7.1%). The demographic differences between those included and those not included were consistent with previous reports of persons lost to follow-up in longitudinal studies of moderate and severe TBI.^21–23

Characteristics of the total sample before excluding the widows are available in Supplemental Digital Content Table 1 (available at: https://links.lww.com/JHTR/A432), and characteristics of the study sample excluding widows are displayed in Table 1. Participants had an average age at injury of 44 years, and the majority were male, White in race/ethnicity, employed, did not have a substance abuse problem, and had a GCS score for mild TBI. The average number of days in PTA was 32 and TFC days was 8. The average FIM scores were 70, 25, and 94 for Motor, Cognitive, and Total, respectively. These scores correspond to a level between supervision and modified independent for Motor and Total mean scores and supervision level for Cognitive mean score. At 10 years postinjury, 66% remained married to the same partner while 34% were not married to the same partner throughout the 10-year period.

TABLE 1 - Summary of the demographic and injury-related variables for the sample (N = 1423)

	n	Mean	SD
Age at injury	1423	44.43	12.86
Posttraumatic amnesiaduration, d	1347	31.73	30.49
Time to follow commands, d	1391	8.14	14.11
Disability Rating Scale	1399	5.80	3.28
Length of stay acute, d	1423	20.45	16.10
Length of stay rehab, d	1391	26.95	24.77
Length of stay total, d	1390	47.34	34.02
FIM Motor	1396	69.88	16.61
FIM Cognitive	1407	24.59	6.30
FIM Total	1393	94.41	20.42
		n	%
Marital stability
Stayed married throughout		938	65.92
Did not stay married throughout		485	34.08
Sex
Male		1071	75.26
Female		352	24.74
Race/ethnicity
White		1096	77.68
Black		168	11.91
Hispanic		98	6.95
Other		49	3.47
(Missing)		12
Education level
Less than high school		231	16.65
High school diploma or General Education Development		751	54.15
Associate or bachelor's degree		285	20.55
Graduate degree		120	8.65
(Missing)		36
Employment status
Employed		1150	81.27
Not employed		265	18.73
(Missing)		8
Problematic substance use
Yes		407	28.60
No		811	56.99
Missing		205	14.41
Cause of injury
Vehicular		822	58.38
Violence		100	7.10
Falls		383	27.20
Other		103	7.32
(Missing)		15
Glasgow Coma Scale score
Mild		431	30.63
Moderate		148	10.52
Severe		203	14.43
Intubated		279	19.83
Sedated		346	24.59
(Missing)		16

Abbreviation: FIM, Functional Independence Measure.

Timing of change in marital status

Of those whose marital status changed to not married during the 10 years following injury, the greatest change occurred from injury to year 1, with substantially reduced rate thereafter and continual decline in rate over time. Among those who experienced marital status change, 68% experienced this during the first 5 years compared with 32% occurring between years 5 and 10 (see Table 2). Figure 1 depicts the average percentage per year of those who change to not married, showing decline in rate over time postinjury.

Figure 1.: Average annual rates of marital loss among those with change in marital status.

TABLE 2 - Proportion of change from married to not married by time period

Timing post-TBI	No longer married
	n	%
Injury to year 5	332	68.45
Injury to year 1	131	39.46
Year 1 to year 2	80	24.10
Year 2 to year 5	121	36.45
Year 5 to year 10	153	31.55
Total	485	100

Abbreviation: TBI, traumatic brain injury.

Bivariate relations

For the bivariate relations with marital stability, statistically significant associations were found with age at injury (P < .0001), sex (P = .0030), race/ethnicity (P = .0177), education level (P = .0007), employment status (P = .0305), problematic substance use (P < .0001), cause of injury (P < .0001), GCS score (P < .0001), PTA days (P = .0349), TFC days (P = .0159), LOS acute (P = .0193), and FIM Total score (P = .0440). The odds of staying married were shown to be higher for older participants and females. After controlling for the pairwise comparisons, Black participants were less likely to stay married throughout than White participants. Participants with higher levels of education were more likely to stay married, while employed participants were less likely to remain married. Participants who did not have a substance abuse problem were more likely to remain married than those who reported a problem using substances. With respect to cause of injury, participants who sustained a fall had higher odds of staying married. The odds of staying married were lower for participants who had a GCS score for severe TBI than those with a GCS score for mild TBI. A higher number of PTA days, TFC days, and LOS acute resulted in lower odds of staying married, while participants with higher FIM Total scores were more likely to remain married. Table 3 illustrates all bivariate results.

TABLE 3 - Logistic regression model results for the bivariate relationships between marital stability (stayed married throughout vs did not stay married throughout) and the demographic and injury-related variables

	n	OR	95% CI	C-Index	P
Age at Injury	1423	1.070	1.059-1.081	0.717	<.0001a
Sex	1423	...	...	0.536	.0030a
Male vs female	...	0.670	0.514-0.873	...	.0030a
Race/ethnicity	1411	...	...	0.535	.0177a
White vs Black	...	1.577	1.134-2.193	...	.0068b
White vs Hispanic	...	0.884	0.565-1.382	...	.5878
White vs other	...	0.650	0.335-1.261	...	.2025
Black vs Hispanic	...	0.560	0.331-0.949	...	.0311
Black vs other	...	0.412	0.201-0.846	...	.0156
Hispanic vs other	...	0.735	0.337-1.605	...	.4400
Education level	1387	...	...	0.555	.0007a
High school diploma or GED vs less than high school	...	1.073	0.792-1.453	...	.6514
High school diploma or GED vs associate or bachelorʼs degree	...	0.727	0.542-0.976	...	.0337
High school diploma or GED vs graduate degree	...	0.428	0.267-0.685	...	.0004b
Associate or bachelor's degree vs less than high school	...	1.475	1.022-2.128	...	.0380
Associate or bachelorʼs degree vs graduate degree	...	0.588	0.352-0.984	...	.0431
Graduate degree vs less than high school	...	2.507	1.491-4.216	...	.0005b
Employment status	1415	...	...	0.524	.0305a
Employed vs not employed	...	0.723	0.539-0.970	...	.0305a
Problematic substance use	1423	...	...	0.548	<.0001a
No vs Yes	...	2.147	1.676-2.750	...	<.0001c
No vs missing	...	1.254	0.903-1.741	...	.1771
Yes vs missing	...	0.584	0.411-0.829	...	.0026c
Cause of injury	1408	...	...	0.575	<.0001a
Vehicular vs falls	...	0.560	0.428-0.734	...	<.0001b
Vehicular vs violence	...	1.464	0.965-2.222	...	.0730
Vehicular vs other	...	0.617	0.391-0.974	...	.0379
Falls vs violence	...	2.614	1.658-4.121	...	.2323
Falls vs other	...	1.101	0.673-1.799	...	.7023
Violence vs other	...	0.421	0.234-0.756	...	.0038b
Glasgow Coma Scale score	1407	...	...	0.591	<.0001a
Mild vs moderate	...	1.382	0.911-2.098	...	.1281
Mild vs severe	...	2.125	1.484-3.043	...	<.0001c
Moderate vs severe	...	1.538	0.979-2.415	...	.0617
Posttraumatic amnesia, d	1347	0.996	0.993-1.000	0.560	.0349a
Time to follow commands, d	1391	0.990	0.982-0.998	0.581	.0159a
DRS score at discharge	1399	0.984	0.952-1.018	0.523	.3519
Length of stay acute, d	1423	0.992	0.985-0.999	0.557	.0193a
Length of stay rehab, d	1391	0.999	0.994-1.003	0.522	.5556
Length of stay total, d	1390	0.997	0.994-1.001	0.545	.1170
FIM Motor at discharge	1396	1.006	1.000-1.013	0.528	.0699
FIM Cognitive at discharge	1407	1.012	0.995-1.030	0.529	.1625
FIM Total at discharge	1393	1.006	1.000-1.011	0.537	.0440a

Abbreviations: C-Index, concordance index; DRS, Disability Rating Scale; FIM, Functional Independence Measure; GED, General Education Development; OR, odds ratio.

^aSignificant level (α = .05).

^bBonferroni correction significance level (α = .0125).

^cBonferroni correction significance level (α = .0167).

Multivariate model

The multivariate logistic regression model revealed similar relations between the model covariates and marital stability as did the previous bivariate analyses. The significance levels did not change for age at injury, sex, problematic substance use, FIM Motor, and FIM Cognitive scores after controlling for the other covariates. However, no longer significant in this multivariable model were race/ethnicity, education level, employment status, cause of injury, and TFC days. Significant relations were found with age at injury (P < .0001), sex (P = .0028), and problematic substance use (P = .0092). Again, older participants were shown to have higher odds of staying married. Females were more likely to stay married throughout than males. The odds of staying married throughout were higher for those participants who do not have a problem using substances than those who do. Full model results are shown in Table 4. There was no evidence of collinearity among the predictor variables within the model.

TABLE 4 - Multivariate logistic regression model results for marital stability (stayed married throughout vs did not stay married throughout)

	(N = 1298, C-Index = 0.740)
	OR	95% CI	P
Age at injury	1.070	1.058-1.083	<.0001a
Sex	...	...	.0028a
Male vs female	0.621	0.455-0.849	.0028a
Race/ethnicity	...	...	.0775
White vs Black	1.292	0.876-1.905	.1974
White vs Hispanic	0.623	0.371-1.047	.0739
White vs other	0.665	0.316-1.401	.2826
Black vs Hispanic	0.483	0.265-0.880	.0174
Black vs other	0.515	0.229-1.157	.1079
Hispanic vs other	1.066	0.443-2.564	.8860
Education level	...	...	.8326
High school diploma or GED vs less than high school	0.977	0.679-1.405	.8995
High school diploma or GED vs associate or bachelor's degree	0.935	0.671-1.303	.6902
High school diploma or GED vs graduate degree	0.787	0.469-1.323	.3665
Associate or bachelor's degree vs less than high school	1.045	0.675-1.618	.8433
Associate or bachelor's degree vs graduate degree	0.842	0.481-1.475	.5490
Graduate degree vs less than high school	1.241	0.683-2.252	.4787
Employment status	...	...	.0864
Employed vs not employed	1.373	0.956-1.971	.0864
Problematic substance use	...	...	.0092a
No vs Yes	1.507	1.133-2.004	.0048b
No vs missing	0.939	0.637-1.383	.7488
Yes vs missing	0.623	0.414. 0.937	.0231
Cause of injury	...	...	.4527
Vehicular vs falls	0.899	0.657-1.232	.5085
Vehicular vs violence	1.166	0.713-1.908	.5410
Vehicular vs other	0.707	0.420-1.190	.1920
Falls vs violence	1.296	0.759-2.214	.3423
Falls vs other	0.786	0.444-1.389	.4069
Violence vs other	0.606	0.306-1.199	.1503
Time to follow commands, d	1.001	0.991-1.010	.8662
FIM Motor at discharge	1.005	0.996-1.014	.3230
FIM Cognitive at discharge	1.010	0.986-1.034	.4316

Abbreviations: FIM, Functional Independence Measure; GED, General Education Development; OR, odds ratio.

^aSignificant level (α = .05).

^bBonferroni correction significance level (α = .0167).

Bivariate and multivariate analyses were also conducted including the 109 widowed subjects. These results are available in Supplemental Digital Content Tables 2 and 3 (available at: https://links.lww.com/JHTR/A433 and https://links.lww.com/JHTR/A434, respectively). Among the 1532 participants that included widows, 39% (594) were not married to the same partner throughout the 10-year period, with 109 (18%) of the 594 widowed. For the bivariate relations with marital stability, higher odds of staying married were found with older age at injury, non-Black race, higher education level, fall as the cause of injury, GCS score for mild or moderate TBI, no problematic substance use, and higher FIM Motor and FIM Total scores. Table 3 illustrates all bivariate results. The multivariate logistic regression model revealed similar relations between the model covariates except that education level and cause of injury were found to have a nonsignificant relation with marital stability in the multivariable model.

DISCUSSION

The current study is the first to examine rate of change and predictors of marital stability over 10 years post-TBI in a large sample. We found a high rate of marital stability, with 66% married to the same partner during the 10-year period. Marital stability was higher for those who were older, female, and without problematic substance use. The greatest rate of marital status change occurred from injury to year 1 (39%), with substantially reduced rate thereafter and continual decline in rate of marital status dissolution over time.

Currently, it is unclear how these marital outcomes directly compare with the general US population. What is known is that nearly half of marriages in the United States end in divorce, with an estimated lifetime disruption probability between 40% and 50%.²⁴ Since 1960, Americans have delayed marriage, with a steady rise of the median age at first marriage²⁵ and increase in earlier-life cohabitation.²⁵ While marital stability has increased over the years since the 1970s,²⁶ US Census Bureau American Community Survey estimates do not reveal substantial change in marital dissolution rates from 2010 to 2018. In 2010, divorce/separation was 11.1% among males and 14.2% among females who were 15 years and older compared with 11.2% of males and 14.4% of females in 2018. Comparison with the prior decade is not possible as marital status was not asked during the US Census Bureau 2000 survey.^27,28

Consistent with prior studies examining stability at 2 years (85%⁸ [shared participants with the current study] and 78%⁹) and 4 years average (75%¹⁰) postinjury, our study found similarly high rates of cumulative marital stability at 2 years (83%) and 5 years (74%). The level of stability 10 years post-TBI (66%) is higher than other studies (51% at 8 years). Supported by the large sample size and length of follow-up, our data dispel myths about risk of divorce after TBI and suggest a message of hope.

This is the first study to assess rate of change of marital stability after TBI. When marital loss was broken down by year, the greatest rate of marital instability occurred by year 1 (9.21% of the total sample not married). The average rate of marital loss each year was much smaller for future years. Annual breakdown of marriage was 2.8% between years 5 and 10. Specifically examining the timing of marital status change among the “not married” sample revealed the largest percentage of relationship dissolution occurred in the first year (39%). When comparing the first 5 years postinjury to the latter 5 years postinjury, two-thirds of breakdowns occurred in the first 5 years. In contrast, Wood and Yurdakul¹² examined marital stability in a convenience sample of participants who were 5 to 8 years post-TBI and found those who were separated/divorced were 6.16 years postinjury compared with those still married being 4.81 years postinjury. Differences in the findings of these 2 studies may be attributed to sample size (n = 1423 current study vs 131) or recruitment mechanism (referral for neuropsychological and/or rehabilitation assessment vs consecutive admissions for inpatient rehabilitation).

Our findings provide insight into who may be at a greatest risk of marital instability, supporting some previously known risk factors (younger age, male, and substance use), and not others (race/ethnicity, lower education level, employment status, cause of injury, injury severity).^8–11 Substance use and TBI are known to have a bidirectional link,^29,30 with an increased risk for TBI due to substance abuse and an increased risk for substance abuse after TBI. In one study,³¹ abstinence improved relationship stability in couples with one substance-abusing partner, whereas relationship stability declined because of abstinence in relationships where both partners used substances. While substance use itself may not cause marital instability, a spouse's perception that substance use is problematic may contribute to marital instability. Finally, consistent with other studies,^8–11,13 older age at the time of injury was associated with marital stability. It is possible that those older at the time of injury were married for longer duration prior to the injury, or that with age come the qualities it takes to commit to marriage even during adversity.

It is important to note that there are likely other factors beyond what we were able to investigate with this study that contribute to marital stability. For instance, a systematic literature review on marriage in the general population conducted by Karimi et al³² identified factors such as spirituality and religion, commitment, sexual relationship, communication, children, love, attachment, intimacy, and conflict resolution approach as protective factors associated with marital stability. Alternatively, Gottman³³ found negative behaviors such as criticism, contempt, defensiveness, and stonewalling to be highly predictive of marital dissolution. Data on such potentially protective and negative factors are not available in the current study but should be taken into consideration in future studies, including those investigating marital interventions. In addition, long-lasting, permanent, and even temporary life circumstances such as family member death and car accidents may be pivotal events that lead people to divorce or hasten divorce, and this should also be examined in future studies.^34–37

Implications

A better understanding of timing and predictors of marital stability informs delivery of relationship interventions aimed at preventing avoidable divorces within a larger framework of TBI as a “chronic disease condition.”³⁸ Marital loss was greatest in the first 2 years, suggesting that psychoeducation and support may be beneficial early after the injury and potentially prior to the onset or elevation of marital discord. It is not known whether aggressive treatment of neuropsychiatric TBI impairments and symptoms during the first 2 years would affect marital stability, especially if the marriage was already strained pre-TBI. It seems important to openly discuss the possible and likely brain injury–related long-term outcomes and treatments to help couples adjust.

Immediately postinjury, couples may be more focused on “surviving” the TBI than on how the injury is affecting their relationship. Accordingly, the nature of support early on may involve check-ins on relationship status, education about natural changes in relationships, and counseling on managing relationship stress as needed. As time progresses, if the couple discloses deeper relationship and intimacy issues, interventions would need to morph to fit those needs. Clinical discretion should be used regarding readiness to start these discussions (based on the survivor's functioning). Alternatively, with marital loss being greatest in the first 2 years, it is possible the marriage was dysfunctional prior to the TBI but was previously tolerated by one or both parties. Furthermore, quality of the marriage, which was not studied here, would be important clinically. While marital stability is often viewed as the goal, sometimes the most appropriate outcome may be dissolution.

Clinicians should be cognizant of who might be at a greater risk of marital instability and therefore require further assistance, especially those younger in age at the time of injury, male, and with a history of substance abuse. Because persons with alcohol or illicit drug abuse history are at an increased risk of marital instability, a first step to enhancing marital stability is identifying those with problematic use. We recommend substance abuse screening with referrals and resources available for those at risk.

Limitations

Results from this study and their implications should be interpreted with a few limitations in mind. The findings may not generalize to individuals with TBI who do not receive inpatient rehabilitation or are treated at less specialized rehabilitation centers. This study only included individuals requiring inpatient rehabilitation who survived at least 10 years post-TBI, excluding those who did not survive to follow-up. Causality between TBI and marital dissolution was not studied. There is no comparison noninjured control group. Data were collected via self-/proxy report, which carries the risk of inaccuracies. In particular, self-report of preinjury substance use is prone to inaccuracy. It is unknown who in the sample may have received marital counseling/intervention at any point during the 10 years. Analyses were limited to those variables collected for the TBIMS data set, and as such some variables that may contribute to marital stability were unavailable to be studied (eg, age at marriage, marital history, longevity of marriage prior to injury, pre- and postinjury relationship quality/satisfaction, children, religion, supports, finances, extramarital relationships, partner-specific demographics, well-being, and interpersonal behaviors such as criticism, contempt, defensiveness, and stonewalling). Of note, in our study, there was missing information regarding problematic substance use involving 172 participants. As eliminating these subjects would have introduced bias with regard to race, these subjects were maintained. Missing data were analyzed for significance with regard to marital stability, and none was found. Finally, for those “still married” 10 years postinjury, it was known that the participant was married at the time of injury, at 10 years, and at years 2 and 5 follow-ups (if successfully contacted). It is possible that a few participants were married but to a different partner, having divorced, or their spouse had died and they had remarried during the 10-year span without detecting this. This chance of someone marrying a different partner between follow-up epochs without us detecting this is minimized as, since 2010, the TBIMS has specifically asked the participants whether their marital status has changed since last follow-up.

Directions for future research

It would be useful to study marital adjustment and satisfaction in the individuals who remain together in contrast to those who divorce. It will be important to gain a clearer picture of the risk factors³³ and interpersonal and intrapersonal factors³² that are likely protective or promote commitment to each other (eg, perceived blame regarding the injury, presence of children, faith values, support, history of infidelity). Also of importance are the couple's self-perceptions of factors related to remaining committed, things they have done to sustain a relatively well-adjusted marriage, and resources they wish they had. Together, these factors could inform a more thorough theoretical model that could then be tested and, in turn, would offer a more comprehensive and detailed understanding of their significance to marital stability and quality.³⁹ While the relations of marital loss with global physical and cognitive burden of care were studied, future studies examining objective functional impairments (such as communication, awareness, initiation, behavior, and emotion) may be informative. Given the trend that more couples live together without being married,³² in the future, it will also be important to apply a broader perspective to studying the experience of post-TBI relationships that include couples in committed, intimate relationships beyond that of marriage.

CONCLUSION

Among adults who receive inpatient rehabilitation for TBI and who are married at the time of injury, most remain married 10 years later. Those of younger age, who are male, and who have a history of problematic substance use were found to be at a highest risk for relationship dissolution. The findings inform content, timing, and delivery of marital interventions with potential implications for patient and family education, relationship counseling, support, and treatments. Interventions aimed at substance use prevention and functional improvement may also have relevance to facilitating marital stability.