Volume 90, Issue 8 , Pages 1371-1378, August 2009
Influence of Race/Ethnicity on Divorce/Separation 1, 2, and 5 Years Post Spinal Cord Injury
Article Outline
- Abstract
- Methods
- Results
- Description of the Sample
- Included Versus Excluded
- Differences in Covariates Among the Races/Ethnicities
- Effects of Demographic and Injury Characteristics on Marital Status and Changes in Marital Status Over Time
- Adjusted Model
- Effects of Demographic and Injury Characteristics on Marital Status Postinjury and Changes in Marital Status Over Time
- Discussion
- Conclusions
- References
- Copyright
Abstract
Arango-Lasprilla JC, Ketchum JM, Francis K, Premuda P, Stejskal T, Kreutzer J. Influence of race/ethnicity on divorce/separation 1, 2, and 5 years post spinal cord injury.
Objectives
(1) To compare the proportions of divorce/separation between races/ethnicities at 1, 2, and 5 years post spinal cord injury (SCI); (2) to examine changes in proportions of divorce/separation over time within each race/ethnicity group; and (3) to compare the changes in proportions of divorce/separation over time between races/ethnicities.
Design
Retrospective study.
Setting
Model Spinal Cord Injury Systems.
Participants
A sample of participants married preinjury (N=1528; 1108 whites, 258 blacks, 162 Hispanics) was selected from the National Spinal Cord Injury Statistical Center database from 1988 to 1998.
Interventions
Not applicable.
Main Outcome Measures
Proportion of separation/divorce at 1, 2, and 5 years postinjury for each race/ethnic group.
Results
At all postinjury years (1, 2, 5y), blacks had significantly greater odds of divorce/separation versus staying married than Hispanics. In addition, whites had significantly greater odds of divorce/separation versus staying married compared with Hispanics at 1 and 2 years postinjury. People with SCI of all races/ethnicities showed significantly greater increases in the odds of divorce/separation versus staying married over time (1–2, 2–5, 1–5y postinjury). Although there was evidence that the races/ethnicities were significantly different at each postinjury year, and that each race/ethnicity showed significant increases in the proportion of divorce/separation over time, there was no indication that the increases in the divorce/separation over time were significantly different among the race/ethnic groups.
Conclusions
Family therapists and rehabilitation professionals should work together to reduce the separation and divorce rates in all subjects with SCI, with special attention paid to meeting the specific needs of those with minority backgrounds.
Key Words: Divorce, Rehabilitation, Spinal cord injuries
List of Abbreviations: AIS, American Spinal Injury Association Impairment Scale, ANOVA, analysis of variance, CI, confidence interval, NSCISC, National Spinal Cord Injury Statistical Center, OR, odds ratio, QOL, quality of life, SCI, spinal cord injury
SPINAL CORD INJURY is usually the result of an acute traumatic event, such as a motor vehicle collision, fall, or violent incident, which may result in partial or complete loss of mobility and sensation, depending on the level of injury and the completeness of the spinal cord lesion.1, 2 In the United States, the incidence of SCI is estimated at nearly 12,000 new cases per year, and it currently affects 227,080 to 300,938 people.3
An SCI results in physical, emotional, psychosocial, and economic changes that affect subjects with SCI and their families.4, 5, 6, 7 Examples of physical changes can include bladder and bowel dysfunction, problems with sexual function, respiratory impairment, spasticity, pain, and pressure ulcers.8, 9 The emotional impact of SCI can include depression,10, 11, 12 anxiety,8, 13 and substance abuse with possible dependency.14, 15, 16 The psychosocial consequences often include increased social isolation17 and lower QOL.18 Finally, economic changes might include medical and home modification costs,5, 19 as well as decreased employment and associated income.20
Experiencing an SCI often results in a lack of autonomy that causes the person with SCI to require some degree of caregiving. Family members are likely to assume the role of caregiver, which involves assisting the person with SCI in a variety of activities of daily living.21, 22 The physical, emotional, psychosocial, and economic changes experienced after SCI can be especially difficult for these caregivers.5, 23 Furthermore, family-member caregivers are often informal caregivers who are untrained and largely unprepared to assume the caregiving role. Not surprisingly, taking on these new responsibilities with inadequate resources can provoke stress in them.5, 23
Being a caregiver for a person with SCI has been associated with a wide range of negative effects, including the following: decreased QOL,24 less choice and control over life,22 increased burden,22 increased financial instability,22 anxiety/depression,25 and psychologic morbidity.26 In most cases, the spouse assumes the role of primary caregiver.27 Research on spousal caregivers of subjects with SCI shows increased emotional impact and more health problems, particularly for wives.28 Caregiver spouses of people with SCI reported more symptoms of burnout, fatigue, anger, and resentment than spouses who were not caregivers.29 Schulz et al30 reported that the daily amount of time spouses (who were also primary caregivers) spent aiding the person with SCI was strongly correlated with depression in the spouse, even after controlling for health and income measures. In addition, this study found that the degree of perceived burden was inversely related to life satisfaction in the caregiver.
The negative consequences that SCI causes on the person and his/her spousal caregiver may lead to relationship problems and possibly separation/divorce. Thus, it is not surprising that various studies have found that subjects with SCI have higher divorce rates compared with the general population.31, 32 For example, DeVivo and Fine31 studied 143 single and 133 married subjects with SCI over 3 years postinjury to determine the short-term impact of SCI on marital status. Approximately 90% of those single subjects with SCI remained single, compared with an expected 80% in the general population. Also, approximately 19% of married subjects with SCI divorced/separated, compared with an expected 6% according to US population divorce/separation rates. A decade later, divorce was still more common in subjects with SCI. People with SCI were 1.7 times more likely to divorce than the general population.32
Although there is a prevalent belief that remaining single and/or being separated/divorced is more common after SCI than remaining married, the actual rates reported in the literature are quite variable.23 Divorce rates for subjects with SCI range from 17% to 48%.31, 32, 33, 34, 35 The reported marriage rate for single SCI patients is nearly 3 times smaller than that of their able-bodied peers.15
Several factors have been identified that influence marital status after SCI, including age, sex, education, discharge disposition, employment status, having children (or not), level of functional ability, previous divorce,31, 32, 33 and cultural traditions.36 Because of the relationship between race/ethnicity and cultural traditions, this variable may be an important moderator of marital status after SCI. Unfortunately, only 2 SCI research studies examining marital status have included race/ethnicity in their analyses. Among subjects with SCI who were married at the time of injury (n=133), discriminant analysis of postinjury divorce and separation rates found that being black was significantly related to obtaining a divorce within the first 3 years postinjury compared with whites.31 DeVivo et al32 studied 622 persons from the model Spinal Cord Injury Systems of Care (502 whites, 80 blacks, and 40 subjects of Native American, Asian, or Hispanic descent) who were single before injury and got married after SCI and were followed 1 and 15 years after their marriage. Although race was not a significant predictor of divorce in the multivariate regression model, univariate analysis showed that 23 out of the 80 blacks in their sample divorced. This rate was 1.6 times higher than the divorce rate of whites and the highest of the 3 race groups. The divorce rate of the other racial/ethnic group was not significantly different relative to whites.
There are various limitations of the previous studies examining race/ethnicity and marital status after SCI. For instance, the minority groups were small in number, consisting mainly of blacks.31, 32 Consequently, the analyses may not have had enough power to detect true differences in marital status after SCI, and/or the effects may have been confounded by combining subjects from various ethnicities into a single racial/ethnic category (ie, other). One study did not control for potential confounders and used a cross-sectional analysis of marital status in couples married after the injury for varying amounts of time.32 This latter study did not take into consideration fluctuations in marital status that may occur over time uniquely for each race/ethnicity.
To our knowledge, this is the first study to examine the relationship between race and divorce/separation 1 to 5 years post-SCI in a large longitudinal sample of diverse subjects (whites, blacks, and Hispanics) who were married before injury. The purpose of the present study was to examine the relationship between race/ethnicity and subsequent divorce/separation at 1, 2, and 5 years post-SCI. Specifically, the study had 3 aims: (1) to compare the odds of divorce/separation versus staying married between races/ethnicities at each follow-up year; (2) to examine changes in odds of divorce/separation versus staying married over time within each race/ethnicity group; and (3) to compare the changes over time in odds of divorce/separation versus staying married between races/ethnicities. Demographic and injury characteristics that differ significantly among race/ethnicities, significantly affect divorce/separation, and/or significantly affect divorce/separation over time were controlled for the final analysis.
Methods
Participants
All participants were Model Spinal Cord Injury System patients enrolled in the NSCISC database. The database stores the records of patient rehabilitation and follow-up information from 25 designated model systems nationwide. Although eligibility criteria have been revised several times since the database's creation in 1975,37 as of January 2005, all participants must (1) receive system inpatient care, (2) be treated at a model system within 1 year of injury, (3) have a clinically discernible degree of neurologic impairment after a traumatic event, (4) have provided informed consent, (5) reside in the geographic catchment area of the model system at the time of injury, and (6) be a citizen of the United States.
A sample of 1664 participants was initially selected from the NSCISC database on the basis of the following criteria: (1) 18 years or older at injury, (2) subjects classified as either Hispanic, non-Hispanic white, or non-Hispanic black, (3) married at injury, (4) injury occurred between September 30, 1988, and September 30, 1998, and (5) subjects not classified as widowed or other at any of the follow-up visits. In addition, 1 subject was excluded from the analysis data set who was classified at discharge as normal on the AIS and had a normal category of neuro impairment. The initial date of September 30, 1988, was chosen because this was the date the measure FIM was added to the database. The end date of September 30, 1998, was chosen because participants' second-year follow-up was no longer required 2 years after this (October 1, 2000). For the sake of brevity, we will use the terms “black” and “white” to refer to non-Hispanic blacks and non-Hispanic whites, respectively.
To assess marital status 1 to 5 years postinjury, the subjects used in this analysis must have at least 1 year of follow-up information. A total of 136 (8.17%) subjects could not be included in this analysis because marital status at all 3 follow-up visits (1, 2, 5y) was unknown. Thus, 1528 subjects were left for analysis.
Measures
The measures and modeling types for the independent and dependent variables used in this study are summarized below.
Dependent variableThe dependent variable used for this analysis was a categorical variable indicating whether the subject was married or divorced/separated at the specified postinjury year. This measure was collected from participants at 1, 2, and 5 years postinjury.
Independent variablesThe primary independent variable was a categorical variable indicating whether the subject was classified as Hispanic, white, or black. Subjects who were classified as non-Hispanic, not white or non-Hispanic, and not black were not considered in this study.
Other demographic variables available for analyses included age at injury, sex, level of education at injury, and employment status at injury. Highest formal level of education completed was categorized into less than high school (grades 1–11), high school/general equivalency diploma, or more than high school (associate's degree, bachelor's degree, master's degree, doctoral degree). Primary occupational, educational, or training status was categorized as employed, unemployed, or other (homemaker, on-the-job training, sheltered workshop, retired, student, other/unclassified).
Measures of injury characteristics available for analysis included cause of injury, FIM total motor score at discharge from inpatient rehabilitation, category of neuroimpairment at discharge, and AIS at discharge. Cause of injury was dichotomized into not violent (vehicular, sports related, falls, flying/falling objects, medical/surgical complications, pedestrian accidents, other) and violent (ie, gunshot wounds, penetrating wounds, person-to-person contact, explosions). The FIM is the most widely accepted measure of functional assessment in the rehabilitation community. The FIM total motor score at discharge is a 13-item measure collected by raters at discharge with scores for each item ranging from 1 (total assist) to 7 (complete independence) with the exceptions of the transfer items (tub and shower), which have minimum values of 0. Category of neuroimpairment at discharge was dichotomized as paraplegia or tetraplegia, and AIS at discharge was dichotomized as incomplete or complete.
Statistical Analyses
All statistical analyses were conducted by SAS (version 9.1),a and graphics were produced by SigmaPlot (version 11.0).a There were a total of 136 subjects who could have been included in the analysis if their marital status had not been missing at all 3 follow-up years. These subjects were compared with the 1528 included subjects with respect to demographic and injury characteristics by chi-square analyses for categorical variables and t tests for continuous variables.
Preliminary analyses that used ANOVA models for continuous variables and chi-square tests for categorical variables were conducted to identify differences between the race/ethnicity groups with respect to age at injury, sex, level of education at injury, employment status at injury, cause of injury, FIM at discharge, neuroimpairment, and AIS impairment.
To test the hypotheses of interest, a repeated-measures logistic regression model was fit using the framework of generalized linear mixed-effects models.38 The model included effects for follow-up year, race/ethnicity, and the 2-way interaction between race/ethnicity and year. To correctly understand the effects of ethnicity on marital status and changes in marital status over time, the model should adjust for the demographic and injury characteristics (covariates) that are significantly different among the races/ethnicities as well as those that may affect marital status and changes in marital status over time.
Because there were a variety of potential covariates and interaction effects that could be added to this model, model-building strategies similar to those used for logistic regression as outlined by Hosmer and Lemeshow39 were used. Briefly, the steps are as follows: (1) Models for each covariate were fit, including main effects for follow-up year and the covariate of interest. Any covariate with P less than .25 was considered for the adjusted model. (2) Models for each covariate identified in step 1 were fit and included main effects for follow-up year and the covariate of interest, as well as the follow-up year by covariate 2-way interaction effect. Any interaction effect with P less than .10 was considered for the adjusted model. (3) The adjusted model was initially fit with effects for race/ethnicity, follow-up year, the race/ethnicity by follow-up year interaction effect, and all potential covariates and interaction effects obtained from steps 1 and 2 as well any additional covariates found to differ significantly among the races/ethnicities. (4) Any covariate (if it was not included in a significant interaction) or covariate by follow-up year interaction effect that no longer contributed to the fit of the model (P<.05) was removed in a backward selection manner.
Results
Description of the Sample
The demographic and injury characteristics of the sample of 1528 are summarized irrespective of race/ethnicity in table 1. The mean age ± SD at injury of subjects was 43.9±14.5 years, 72.5% were white, 16.9% were black, 10.6% were Hispanic, 79.8% were men, and 74.0% had at least a high school education. There was some missing data for level of education (n=86), employment at admission (n=5), cause of injury (n=1), FIM at discharge (n=80), neuroimpairment (n=15), and AIS impairment (n=22).
Table 1. Demographic and Injury Characteristics of Sample
| Characteristics | White (n=1108) | Black (n=258) | Hispanic (n=162) | Overall (N=1528) |
|---|---|---|---|---|
| Age at injury (N=1528) | 44.6±14.2 | 45.6±15.4 | 36.5±12.2 | 43.9±14.5 |
| Discharge FIM (n=1448) | 57.0±22.8 | 51.7±23.7 | 61.6±22.3 | 56.6±23.0 |
| Sex (N=1528) | ||||
 Male | 859(77.5) | 218(84.5) | 142(87.7) | 1219(79.8) |
| Female | 249(22.5) | 40(15.5) | 20(12.3) | 309(20.2) |
| Employment (n=1523) | ||||
| Employed | 856(77.5) | 165(64.2) | 129(79.6) | 1150(75.5) |
| Unemployed | 40(3.6) | 35(13.6) | 20(12.4) | 95(6.2) |
| Other | 208(18.8) | 57(22.2) | 13(8.0) | 278(18.3) |
| Education (n=1442) | ||||
| Less than HS | 186(17.8) | 92(37.4) | 97(63.8) | 375(26.0) |
| HS/GED | 612(58.6) | 134(54.5) | 50(32.9) | 796(55.2) |
| More than HS | 246(23.6) | 20(8.1) | 5(3.3) | 271(18.8) |
| AIS (n=1506) | ||||
| Incomplete | 622(57.0) | 152(59.8) | 74(46.0) | 848(56.3) |
| Complete | 469(43.0) | 102(40.2) | 87(54.0) | 658(43.7) |
| Neurologic impairment (n=1513) | ||||
| Paraplegia | 554(50.5) | 85(33.2) | 91(56.5) | 730(48.2) |
| Tetraplegia | 542(49.5) | 171(66.8) | 70(43.5) | 783(51.8) |
| Cause of injury (n=1527) | ||||
| Not violent | 1050(94.9) | 200(77.5) | 115(71.0) | 1365(89.4) |
| Violent | 57(5.1) | 58(22.5) | 47(29.0) | 162(10.6) |
Included Versus Excluded
The 136 subjects who were excluded because of missing marital status information at all 3 follow-up visits were compared with the 1528 subjects used for analyses. There was no evidence that the included and excluded subjects differed significantly with respect to age at injury (t=.57, df=1662, P=.571), FIM total at discharge (t=.49, df=1569, P=.625), sex (χ2=2.40, df=1, P=.122), education at admission (χ2=.40, df=2, P=.820), AIS (χ2=.55, df=1, P=.457), and neurologic impairment (χ2=2.93, df=1, P=.087). Included and excluded subjects were significantly different with respect to employment at admission (χ2=12.18, df=2, P=.002) and cause of injury (χ2=6.23, df=1, P=.013). Specifically, the excluded subjects had lower proportions of employment (63.2% vs 75.5%) and greater proportions of violent injuries (17.65% vs 10.61%) than the included subjects.
Differences in Covariates Among the Races/Ethnicities
The demographic and injury characteristics of the sample of 1528 are summarized by race/ethnicity in table 1. The 3 races/ethnicities were different with respect to sex (χ2=13.27, df=2, P=.001), employment at admission (χ2=61.21, df=4, P<.001), education at admission (χ2=185.66, df=4, P<.001), AIS (χ2=8.51, df=2, P=.014), neurologic impairment (χ2=29.94, df=2, P<.001), cause of injury (χ2=130.99, df=2, P<.001), age at injury (F2,1525=25.40, P<.001), and FIM total at discharge (F2,1445=9.49, P<.001).
Effects of Demographic and Injury Characteristics on Marital Status and Changes in Marital Status Over Time
Neurologic impairment and FIM total at discharge were not significant univariate predictors of marital status postinjury (P≥.25), whereas sex (P=.158), education at admission (P=.052), employment at admission (P<.001), AIS (P<.001), cause of injury (P<.001), and age at injury (P<.001) were at least marginal predictors (ie, P<.25). Sex, neurologic impairment, education at admission, employment at admission, cause of injury, and age at injury were not significant univariate predictors of changes in marital status over time (P≥.10), whereas AIS (P=.001) and FIM total at discharge (P=.018) were at least marginal predictors (ie, P<.10).
Adjusted Model
The distribution of marital status for each follow-up year is summarized in table 2. These unadjusted data do not reflect adjustment for relevant demographic and injury characteristics.
Table 2. Observed Marital Status 1, 2, and 5 Years Postinjury by Race/Ethnicity
| Characteristics | 1-Year Follow-up | 2-Year Follow-up | 5-Year Follow-up | |||
|---|---|---|---|---|---|---|
| n | % | n | % | n | % | |
| White | ||||||
| Married | 995 | 93.8 | 790 | 90.7 | 528 | 81.7 |
| Not married | 66 | 6.2 | 81 | 9.3 | 118 | 18.3 |
| Divorced | 37 | 3.5 | 58 | 6.7 | 101 | 15.6 |
| Separated | 29 | 2.7 | 23 | 2.6 | 17 | 2.6 |
| Black | ||||||
| Married | 222 | 89.2 | 156 | 81.2 | 90 | 71.4 |
| Not married | 27 | 10.8 | 36 | 18.8 | 36 | 28.6 |
| Divorced | 11 | 4.4 | 14 | 7.3 | 24 | 19.1 |
| Separated | 16 | 6.4 | 22 | 11.5 | 12 | 9.5 |
| Hispanic | ||||||
| Married | 149 | 93.7 | 117 | 91.4 | 61 | 75.3 |
| Not married | 10 | 6.3 | 11 | 8.6 | 20 | 24.7 |
| Divorced | 4 | 2.5 | 8 | 6.3 | 16 | 19.8 |
| Separated | 6 | 3.8 | 3 | 2.3 | 4 | 4.9 |
| Overall | ||||||
| Married | 1366 | 93.0 | 1063 | 89.3 | 679 | 79.6 |
| Not married | 103 | 7.0 | 128 | 10.7 | 174 | 20.4 |
| Divorced | 52 | 3.5 | 80 | 6.7 | 141 | 16.5 |
| Separated | 51 | 3.5 | 48 | 4.0 | 33 | 3.9 |
A repeated-measures logistic regression model was fit with effects for follow-up year, race/ethnicity, and 2-way interaction between race/ethnicity and year. To correctly understand the effects of race/ethnicity on marital status and changes in marital status over time, the model should also adjust for the demographic and injury characteristics that are significantly different among the races as well as those that may affect marital status and changes in marital status over time. As previously reported, ANOVA and chi-square tests indicate that all of the demographic and injury characteristics are significantly different among the races/ethnicities (P≤.014). Univariate repeated-measures logistic regression models indicated that sex, education at admission, employment at admission, AIS, cause of injury, and age at injury significantly affect marital status postinjury (P≤.052). The univariate repeated-measures models further showed that AIS and FIM total at discharge significantly affect changes in marital status over time (P≤.018). Therefore, additional effects for each of the demographic and injury characteristics, as well as interactions between year and AIS and between year and FIM total at discharge, were initially included in the model.
The main effects for sex (P=.124), education at admission (P=.120), neurologic impairment (P=.113), FIM total at discharge (P=.999), and the interaction effect for year by FIM total at discharge (P=.052) did not significantly contribute to the fit of the adjusted model and were removed. The final model then included main effects for year, race/ethnicity, employment at admission, AIS, cause of injury, and age at injury, as well as interaction effects for year by race/ethnicity and year by AIS. The effect tests for the final adjusted model are summarized in table 3.
Table 3. Effect Tests for Adjusted Model
| Effects | F | NDF, DDF | P |
|---|---|---|---|
| Year | 35.93 | 2, 1953 | <.001 |
| Race/ethnicity | 6.41 | 2, 1493 | .002 |
| Year×race/ethnicity | 0.88 | 4, 1953 | .474 |
| Employment at admission | 5.57 | 2, 1493 | .004 |
| AIS | 2.02 | 1, 1493 | .155 |
| Year×AIS | 7.27 | 2, 1953 | .001 |
| Cause of injury | 6.38 | 1, 1493 | .012 |
| Age at injury | 68.04 | 1, 1493 | <.001 |
At 1 year postinjury, blacks (OR=3.11; 95% CI, 1.36–7.11) and whites (OR=2.17; 95% CI, 1.01–4.69) had significantly greater odds of divorce/separation compared with Hispanics. Blacks and whites did not have significantly different odds of divorce/separation at 1 year postinjury.
At 2 years postinjury, blacks (OR=4.60; 95% CI, 2.06–10.27) and whites (OR=2.96; 95% CI, 1.40–6.27) had significantly greater odds of divorce/separation compared with Hispanics. Blacks and whites did not have significantly different odds of divorce/separation at 2 years postinjury.
At 5 years postinjury, blacks had significantly greater odds of divorce/separation compared with Hispanics (OR=2.21; 95% CI, 1.11–4.38). Whites did not have significantly different odds of divorce/separation compared with blacks or Hispanics at 5 years postinjury.
Changes in marital status over time for each race/ethnicity (aim 2)For whites, the odds of divorce/separation versus staying married were 1.64 times greater at 2 years postinjury than at 1 year postinjury (95% CI, 1.29–2.09), 1.94 times greater at 5 years postinjury than at 2 years postinjury (95% CI, 1.55–2.42), and 3.18 times greater at 5 years postinjury than at 1 year postinjury (95% CI, 2.49–4.07).
For blacks, the odds of divorce/separation vs staying married were 1.78 times greater at 2 years postinjury than at 1 year postinjury (95% CI, 1.17–2.72), 1.57 times greater at 5 years postinjury than at 2 years postinjury (95% CI, 1.03–2.40), and 2.80 times greater at 5 years postinjury than at 1 year postinjury 1 (95% CI, 1.79–4.40).
For Hispanics, the odds of divorce/separation versus staying married were 3.27 times greater at 5 years postinjury than at 2 years postinjury (95% CI, 1.72–6.23) and 3.95 times greater at 5 years postinjury than at 1 year postinjury (95% CI, 2.06–7.57); however, the odds of divorce/separation were not significantly different for 2 years postinjury compared with 1 year postinjury.
Comparison of changes in marital status between races/ethnicities (aim 3)There was no evidence of a significant interaction effect between race/ethnicity and year (F4,1979=1.14, P=.336). Thus, the differences in the proportion of divorce/separation between the race/ethnicity groups at each year postinjury do not change significantly over time, and the changes over time in the proportion of divorce/separation are not significantly different among the race/ethnicity groups.
Effects of Demographic and Injury Characteristics on Marital Status Postinjury and Changes in Marital Status Over Time
The final adjusted model further indicated that employment status at admission, cause of injury, and age at injury also had significant effects on marital status postinjury, and furthermore, that AIS has an effect on changes in marital status over time.
For employment status at admission, the odds of divorce/separation vs staying married were significantly greater for those who were unemployed versus employed (OR=2.47; 95% CI, 1.44–4.25) and for those who were unemployed versus other (OR=2.68; 95% CI, 1.28–5.61). The odds of divorce/separation versus staying married were not significantly different for those classified as other compared with those who were employed.
With respect to cause of injury, the odds of divorce/separation versus staying married were 1.77 times greater for participants with violent injuries compared with those with nonviolent injuries (95% CI, 1.14–2.77).
For age at injury, comparisons were made between the 25th (32y) and 75th (55y) percentiles of the distribution of age. The odds of divorce/separation versus staying married were 4.57 times greater (95% CI, 3.19–6.56) for younger participants (32y) compared with older participants (55y).
Finally, with respect to AIS, there was a significant interaction between year and AIS. Thus, the differences in the proportions of divorce/separation between the incomplete and complete groups change over time, and the changes over time are different for the complete and incomplete groups. The probability of divorce/separation increase over time at a greater rate for those with complete AIS compared with those with incomplete AIS. More specifically, for those with complete AIS, the odds of divorce/separation versus staying married were 1.80 times greater at 2 years postinjury than at 1 year postinjury (95% CI, 1.29–2.50), 2.70 times greater at 5 years postinjury than at 2 years postinjury (95% CI, 1.99–3.67), and 4.87 times greater at 5 years postinjury than at 1 year postinjury (95% CI, 3.49–6.79). For those with incomplete AIS, the odds of divorce/separation versus staying married were 1.71 times greater at 5 years postinjury than at 2 years postinjury (95% CI, 1.20–2.44), and 2.21 times greater at 5 years postinjury than at 1 year postinjury (95% CI, 1.54–3.17). However, the odds of divorce/separation versus staying married were not significantly different for 2 years postinjury versus 1 year postinjury.
Discussion
The first aim of the present study was to assess the differences in the proportion of divorce/separation between races/ethnicities at 1, 2, and 5 years postinjury, after adjusting for relevant demographic and injury characteristics. The present study found that at all postinjury years (1, 2, and 5y), blacks had significantly greater odds of divorce/separation vs staying married than Hispanics. In addition, whites had significantly greater odds of divorce/separation versus staying married compared with Hispanics at 1 and 2 years postinjury. The odds of divorce/separation versus staying married were not significantly different for blacks versus whites at 1, 2, and 5 years postinjury or for whites versus Hispanics at 5 years postinjury.
One explanation why Hispanics had lower odds of divorce/separation than blacks or whites postinjury could be the similar traditional values and customs that unite them, including familism, a sense of responsibility for the welfare of other family members, and religious beliefs.40 In general, Hispanics have a common respect for elders and people with disabilities as well as a sense of obligation for their care.41 Familism, or valuing family over individual and/or community needs, is a strong, almost universal value in Hispanic subjects. Additionally, for Hispanic spouses, SCI and other disabilities may be viewed as God's will, and coping with difficult life circumstances may be a way of demonstrating worthiness.42
The results of the present study differ from those reported by DeVivo et al,32 who found that there were no significant differences in the divorce rates between different race/ethnic groups. However, 3 important methodologic differences between the 2 studies may account for this discrepancy. First, different definitions are used for marital status. DeVivo et al32 defines termination of marriage as any change from married to divorced/widowed on successive annual evaluations, but this study defines it as subjects divorced/separated at 1, 2, and 5 years postinjury. Second, the time when participants were married and follow-up evaluations differ. Although the participants in the present study were married pre-SCI with follow-up 1 to 5 years post-SCI, participants in the DeVivo32 study got married post-SCI and were followed 1 to 15 years after marriage. Finally, different sample sizes and minority group representations were used. DeVivo32 studied 622 persons (502 whites, 80 blacks, 12 Native Americans, 5 Asians, 23 Hispanics) and clumped the latter 3 races/ethnicities into one heterogeneous minority group. Conversely, the present study sampled 1528 participants (1108 whites, 258 blacks, 162 Hispanics) and compared each racial/ethnicity group separately.
The present study's findings also contrast with DeVivo and Fine's conclusion31 that blacks are more likely than whites to divorce within 3 years postinjury. Again, several methodologic differences may explain this divergence. The first important distinction involves inclusion criteria, with DeVivo31 only considering divorces and the present study considering both divorces and separations. Second, sample sizes and data collection methods differ greatly. DeVivo's31 small sample used 130 subjects from an Alabama center, whereas this study's larger sample uses 1528 subjects from a national, multicenter database. Third, minority representation and reporting also differ. Namely, DeVivo31 included only blacks and whites in his study without reporting the number of minority subjects, whereas the present study includes blacks, whites, and Hispanics, with a clear description of racial/ethnic group sizes. Finally, distinctions in time of data collection may contribute to different results. The data reported in this study were collected from patients whose SCI occurred between 1988 and 1998, whereas in DeVivo's31 study, the injuries had occurred between 1973 and 1980.
The second aim of this study was to examine changes in proportions of divorce/separation over time within each race/ethnicity group. The results of the present study found that all races/ethnicities showed significantly greater increases in the proportion of divorce/separation over time (1–2, 2–5, 1–5y postinjury). One possible explanation for these results may stem from the challenges that occur over time postinjury. Chronic stress and caregiving may accrue considerably over time such that challenges get worse instead of better, leading to increased proportions of separation/divorce rates at 5 years or longer. Because partners of someone with a SCI begin to realize that the physical, behavioral/emotional, psychosocial, and economic problems are going to be long-term consequences, the probability of separation or divorce often increases.
The third aim was to assess the differences in the changes over time (1–5y postinjury) between races/ethnicities in the proportion of divorce/separation, after adjusting for relevant demographic and injury characteristics. The results of the present study found no evidence that the increases in the divorce/separation rates over time were significantly different among the race/ethnicity groups. In other words, the likelihood of divorces and separations increased over time regardless of race/ethnicity. This coincides with previous findings that the percentage of divorces among people with SCI who were married at the time of injury increased in the first 3 years after injury.31, 43 However, unlike their findings, in this sample, we found that divorces and separations continued to increase up to 5 years postinjury for all of 3 racial/ethnic groups. Thirty years' worth of follow-up data from the Model Spinal Cord Injury Care Systems also confirm a rise in divorces and separation over time.44
Other interesting findings of the present study are that being unemployed, younger, and having SCI as a result of a violent cause increases the odds of divorce/separation. In addition, for people with complete AIS (more severe injury), the proportion of divorce/separation across the 5 follow-up years increased at a significantly greater degree than for those with incomplete AIS. In the general population, being married has been associated with being healthier45 and having lower levels of depression.46 For people with SCI, being married has shown to be related to less depression, higher self-reported QOL, and lower self-reported handicap (especially in social integration and economic self-sufficiency areas).27, 47 It seems that for people with SCI, being married is related to positive psychosocial outcomes, such as life satisfaction, return to productive activities, and independent living.23, 47, 48, 49
The findings of the present study have clinical relevance because they indicate a potential area of intervention for rehabilitation professionals in order to avoid separation/divorce in marriages of people with SCI. Rehabilitation professionals can easily screen for signs of stress in the marriage by including questions on the present functioning of the marriage in their periodic follow-up visits, and can help the family problem-solve and implement more effective coping strategies if needed. Counseling, as well as referral to community resources (eg, personal care attendants, home care services, vocational rehabilitation agencies), should be made available to persons with SCI and their families by rehabilitation professionals.
Study Limitations
The results of this study must be interpreted with caution because of the following limitations: (1) Confounding factors, including social support, acculturation level, income, length of marriage, length of preinjury relationship, religious preference and spiritual beliefs held by the couple, and neurobehavioral sequela exhibited by the person with SCI were not extensively measured and therefore could not be controlled. (2) All SCI survivors in the present study were recruited from the NSCISC database regardless of ethnicity and received inpatient rehabilitation/treatment from these standardized centers from 1988 to 1998; thus, the results may not generalize to the at-large U.S. population of SCI survivors. It is possible that the average black and Hispanic SCI survivor does not receive such state-of-the-art rehabilitation or top-quality care; therefore, one might expect additional differences in outcomes for these groups compared with whites when examining marital status in the population of SCI as a whole. (3) Hispanics and blacks with SCI were the only minority groups considered in this study; as such, these findings cannot be extended to other minority subgroups, such as Native Americans and Asians. (4) The present study includes Hispanics with SCI in the United States that are registered as aliens or citizens, but may not represent all of the Hispanics with SCI in the United States, given the large undocumented population of this group living in the United States.
The current study provides insight into how divorce/separation is changing over time within white, Hispanic, and black subjects with SCI. However, this study does not lend itself to comparisons to the general population. To assess how the SCI population differs with respect to non-SCI populations, a future longitudinal study involving both groups would be necessary. Today, the U.S. general population divorce/separation rate is approximately 11% to 13%.50, 51 Although divorce/separation rates for subjects with SCI were lower for follow-up years 1 (7.01%) and 2 (10.75%) in our study, divorce/separation rates increased at follow-up year 5 (20.40%).
Conclusions
The present study found that at all postinjury years (1, 2, 5y), blacks had significantly greater odds of divorce/separation versus staying married than Hispanics, after adjusting for relevant demographic and injury characteristics. In addition, after adjustment, whites had significantly greater odds of divorce/separation versus staying married compared with Hispanics at 1 and 2 years postinjury. With the exception of Hispanics from 1 to 2 years postinjury, people with SCI of all races/ethnicities showed significantly greater increases in the proportion of divorce/separation over time (1–2, 2–5, 1–5y postinjury). Although there was evidence that the races/ethnicities were significantly different at each postinjury year and that each race/ethnicity showed significant increases in the proportion of divorce/separation over time, there was no indication that the increases in divorce/separation over time were significantly different among the race/ethnicity groups. Family therapists and rehabilitation professionals should work together to reduce the separation and divorce rates in all subjects with SCI, with special attention given to meeting the specific needs of those with minority backgrounds.
Supplier
References
- . Medical and psychosocial aspects of chronic illness and disability. 3rd ed.. Salisbury: Jones & Bartlett; 2005;
- . The spinal cord injury handbook: for patients and their families. 2nd ed.. Birmingham: HealthSouth Pr; 1998;
- . Spinal cord injury: facts and figures at a glance. Birgmingham: National Spinal Cord Injury Statistical Center; 2008;
- . Social needs of the patient and his family. BMJ. 1986;292:546–548
- . Coping with spinal cord injury: personal and marital adjustment in the Hong Kong Chinese setting. Spinal Cord. 2000;38:687–696
- . Family adaptation to traumatic spinal cord injury: response to crisis. Fam Relat. 1980;29:558–565
- . Family adjustment to spinal cord injury. http://images.main.uab.edu/spinalcord/pdffiles/FamilyAdjustment.pdfAccessed September 1, 2008
- . Spinal cord injury. In: Frank RG, Elliot TR editor. Handbook of rehabilitation psychology. Washington (DC): American Psychological Association; 2000;p. 11–28
- . Spinal cord injury rehabilitation: state of the science. Am J Phys Med Rehabil. 2006;85:310–342
- Utilization of health services following spinal cord injury: a 6-year follow-up study. Spinal Cord. 2004;42:513–525
- . Treatment of depression following spinal cord injury: an evidence-based review. Rehabil Psychol. 2004;49:134–139
- . Anxiety and depression after spinal cord injury: a longitudinal analysis. Arch Phys Med Rehabil. 2000;81:932–937
- . Post traumatic stress disorder and spinal cord injuries. Spinal Cord. 2001;39:1–10
- . Do preinjury alcohol problems predict poorer rehabilitation progress in persons with spinal cord injury?. Arch Phys Med Rehabil. 2004;85:1488–1492
- . Spinal cord injury. In: Zaretsky H, Richter E, Eisenberg M editor. Medical aspects of disability: a handbook for the rehabilitation professional. 3rd ed.. New York: Springer; 2005;p. 611–647
- . Postinjury substance abuse among persons with brain injury and persons with spinal cord injury. Brain Inj. 2002;16:583–592
- . Psychosocial outcome following spinal cord injury. Paraplegia. 1994;32:771–779
- . Quality of life of spinal cord injured patients in Taiwan: a subgroup study. Spinal Cord. 1997;35:841–849
- . Spinal cord injury: an analysis of medical and social costs. New York: Demos Medical Publishing; 1998;
- . Return to work after spinal cord injury: a review of recent research. Neurorehabilitation. 2002;17:177–186
- . Spinal cord injury. In: Nezu A, Nezu C, Geller P editor. Handbook of psychology. Vol 9:Hoboken (NJ): Wiley; 2003;Health psychology
- . The impact of being a support provider to a person living in the community with a spinal cord injury. Rehabil Psychol. 2005;50:397–407
- . Spinal cord injury and partner relationships. Spinal Cord. 2000;38:2–6
- . Health-related quality of life of primary caregivers of persons with paraplegia. Spinal Cord. 2007;45:399–403
- . Depression among African-Americans, Latinos and whites with spinal cord injury: an exploratory study. Rehabil Psychol. 1999;44:235–247
- . Leisure satisfaction and psychiatric morbidity among informal carers of people with spinal cord injury. Spinal Cord. 2006;44:676–679
- . Ageing with spinal cord injury: the impact of spousal support. Disabil Rehabil. 1999;21:250–257
- . Spouse adjustment to spinal cord injury: long-term medical and psychosocial factors. Rehabil Couns Bull. 1988;32:4–21
- . Spouses of spinal cord injury survivors: the added impact of caregiving. Arch Phys Med Rehabil. 1997;78:822–827
- . The social psychology of caregiving: physical and psychological costs of providing support to the disabled. J Appl Soc Psychol. 1987;17:401–428
- . Spinal cord injury: its short-term impact on marital status. Arch Phys Med Rehabil. 1985;66:501–504
- . Outcomes of post spinal cord injury marriages. Arch Phys Med Rehabil. 1995;76:130–138
- . Marriage rates among persons with spinal cord injury. Rehabil Psychol. 1996;41:321–339
- . Improving the long-term adjustment of spinal cord injured persons. Spinal Cord. 1999;37:345–350
- . Partner relationships, functioning, mood and global quality of life in persons with spinal cord injury and traumatic brain injury. Spinal Cord. 1998;36:252–261
- . How does spinal cord injury affect marital relationship? (A story from both sides of the couple). Disabil Rehabil. 2000;22:764–775
- . Race, employment, and spinal cord injury. Arch Phys Med Rehabil. 2004;85:1782–1792
- . Models for discrete longitudinal data. New York: Springer; 2006;
- . Applied logistic regression. 2nd ed.. New York: Wiley; 2000;
- . Normative beliefs about marriage and cohabitation: a comparison of non-Latino whites, Mexican-Americans, and Puerto Ricans. J Marriage Fam. 1996;58:49–62
- . Discharge to home among Hispanic and non-Hispanic stroke survivors: does family make a difference?. Int J Rehabil Res. 1999;22:317–320
- . Socioeconomic and cultural factors in rehabilitation of Latinos with disabilities. In: Garcia JG, Zea MC editor. Psychological interventions and research with Latino populations. Boston: Allyn & Bacon; 1997;p. 217–234
- . Spinal cord injury (Rehabilitation adds life to years). Western J Med. 1991;154:602–606
- . Annual report for the Model Spinal Cord Injury Care Systems. Birmingham: National Spinal Cord Injury Statistical Center; 2006;
- . Marital status and health: United States, 1999-2002. Adv Data. 2004;351:1–32
- . Marital status and depression: the importance of coping resources. Soc Forces. 1982;61:484–507
- . Marital status and adjustment 1 year post-spinal-cord-injury. J Clin Psychol Med Settings. 2001;8:101–107
- . The relationship between productivity and adjustment following spinal cord injury. Rehabil Couns Bull. 1990;33:188–199
- . Independent living outcomes in spinal cord injury: multivariate analyses. Arch Phys Med Rehabil. 1984;65:66–73
- . Marital status: 2000 (Census 2000 brief). US Census Bureau; 2003;
- . Household relationships and size: marital status. http://factfinder.census.gov/servlet/STTable?_bm=y&-geo_id=01000US&-qr_name=ACS_2007_3YR_G00_S1201&-ds_name=ACS_2007_3YR_G00_&-redoLog=falseAccessed September 2, 2008
- a SAS Institute, 100 SAS Campus Dr, Cary, NC 27513.
Supported by the National Institute on Disability and Rehabilitation Research, United States Department of Education (grant no. H133A060039).
No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit on the authors or on any organization with which the authors are associated.
PII: S0003-9993(09)00284-6
doi:10.1016/j.apmr.2009.02.006
© 2009 American Congress of Rehabilitation Medicine. Published by Elsevier Inc. All rights reserved.
Volume 90, Issue 8 , Pages 1371-1378, August 2009
