Life Satisfaction Trend and Mortality After Traumatic Spinal Cord Injury: A Cohort Study

Abstract

Background

Individuals with traumatic spinal cord injuries (SCIs) have lower subjective well-being and diminished longevity, yet there is a lack of research on how life satisfaction and changes in life satisfaction relate to longevity.

Objectives

To identify the relationships between survival status and life satisfaction and its changing trend over a 10-year period.

Methods

Data were taken from the SCI Longitudinal Aging Study. A cohort study of 676 adult participants with SCI completed three assessments separated by 5-year intervals. We applied the survival analyses by using person-year logistic regression models.

Results

There were 135 participants (20%) who were deceased by the end of 2019. After controlling for demographic and injury factors, life satisfaction was significantly related to survival. In the final model, having an upward trend of life satisfaction was significantly associated with lower odds of mortality, whereas a downward trend in life satisfaction was not significant.

Conclusion

Our findings demonstrate the importance of current life satisfaction and having a trend toward improving satisfaction to survival. The findings indicated the importance of psychosocial adaptation to life quality and longevity after SCI.

Introduction

Spinal cord injury (SCI) typically results in significant disability, impacting nearly all aspects of an individual’s life—physical, mental, emotional, social, and work. Individuals with SCI have an increased risk of mortality compared to the general population, and studies suggest that life expectancy for those who survived the first year after injury has not improved overall over the past 40 years.^1–5 At the same time, many individuals with chronic SCI live long, full lives, and there is a growing body of literature focused on those with exceptional survival and the factors related to longevity.

The vast majority of research on mortality and SCI has focused on how health-related factors and high-risk behaviors relate to excess mortality,^6–12 including excess mortality related to demographic status (age, race-ethnicity),^1,2,13–19 severity of injury,^20–22 health behaviors and secondary health conditions,^23–30 poor socioeconomic status and unemployment,^25,27,31,32 and psychological distress.^33,34 Compared to the studies emphasizing the factors leading to excess mortality, substantially less research has focused on positive life factors that may promote survival and longevity.

One potential positive factor is life satisfaction, which refers to a cognitive judgmental process and assesses an individual’s feelings and attitudes toward life at a particular point in time^35,36 or an individual’s self-perceived richness, meaningfulness, fullness, or quality of their life. Life satisfaction is an important outcome in SCI and has been the focus of multiple studies.^37,38 Studies have found that life satisfaction of individuals with SCI is substantially lower than that of the general population, though there are large individual differences, and evidence suggests that most individuals adapt well.³⁷ Findings from the SCI Longitudinal Aging Study suggest that satisfaction increases over time.³⁹ The trajectory was associated with demographic factors including race, sex, preinjury employment status, marital status at discharge, and level of education and injury-related characteristics including days in rehabilitation, neurologic level, age at injury, and injury etiology.³⁹

There remain several unanswered questions as to the extent to which changes in life satisfaction over time are related to health and longevity after SCI. Early findings on the association between life satisfaction and mortality are mixed. In an early study, adjustment after SCI was related to survival, but none of the six individual areas of life satisfaction assessed were significant. Global rankings were based on psychological, social, and vocational adjustment.⁴⁰ In a later study, general satisfaction, but not economic satisfaction, was found to be related to mortality; those with greater satisfaction were more likely to survive.⁶ Although these studies had an important focus on potential promotion of longevity, they were highly limited, and none of the studies measured the relationship of change in satisfaction over time with survival.

Our purpose was to identify the relationship of life satisfaction and the changing pattern of life satisfaction over time with survival among individuals with SCI. We were specifically interested in identifying whether an improving (upward) or declining (downward) trend in life satisfaction over time was related to survival, controlling for most current life satisfaction. Satisfaction history may be very important to the prediction of longevity. If individuals had high satisfaction, then even a drop in current satisfaction or downward trend may be related to greater longevity because of their having “banked” good years with high satisfaction. On the other hand, individuals may experience an upward trend in satisfaction such that, even though they may be of a similar level to someone whose satisfaction has either declined or stayed the same, their history of improvement and upward trends lead to greater longevity. Or, history of satisfaction may be unrelated to longevity.

Methods

Participants

The enrollment criteria included at least 1 year after traumatic SCI, no full recovery, and at least 18 years old at assessment. All participants were included in the SCI Longitudinal Aging Study^41–43 and were enrolled from two geographic regions, the midwestern and southeastern United States. Individuals in the midwestern region cohort were identified from a rehabilitation specialty hospital and a university hospital starting in 1973, with subsequent enrollments in 1984 and 1993; those in the southeastern region cohort were selected from a rehabilitation specialty hospital starting in 1993, with additional enrollment in 2003. Further details about the study and participants can be found in earlier publications.^41–43 We utilize data beginning in 2003. There were 1539 individuals who completed a self-report assessment. Follow-up assessment work was completed in 2008 and 2013. Our final sample for this study includes 676 participants who completed all three measurements during the 10 years.

Procedures

Institutional review board approval was obtained prior to data collection. Cover letters were sent to potential participants during three follow-ups, separated by 5-year intervals (2003, 2008, 2013). The letters described the studies and let participants know that materials would be forthcoming. Self-reported assessments were sent 4 to 6 weeks later. Follow-up mailings and phone calls were used to maximize responses. Participants received between $30 and $40 to complete study materials. The National Death Index (NDI) service from the Centers for Disease Control and Prevention (CDC) was used to identify participants’ survival status and, for those who were deceased, the date of death and cause of death. The NDI is the gold standard for mortality research. Survival was determined as of December 31, 2019. There was an approximate 6-year interval between the last time of data collection and determination of mortality status.

Measures

The independent variable of interest was life satisfaction, which was measured by the Life Situation Questionnaire–Revised (LSQ-R). The LSQ-R includes 20 satisfaction items, each of which is measured on a 5-point scale, with anchor points of very dissatisfied (1), neutral (3), and very satisfied (5). The total score is calculated by adding all items. The LSQ-R has been used in the previous studies with good validity and reliability.^44–46 All study participants responded on the same instrument at 2003 (time 1), 2008 (time 2), and 2013 (time 3). Based on the three measures of LSQ-R during the 10 years, we created a life satisfaction history variable to assess change in life satisfaction, with three categories: increasing life satisfaction (LSQ-R score at time 1 < LSQ-R score at time 2 < LSQ-R score at time 3), decreasing life satisfaction (LSQ-R score at time 1 > LSQ-R score at time 2 > LSQ-R score at time 3), and other patterns where there was no consistent directional change in life satisfaction over time.

We measured self-reported injury characteristics: years post injury, injury level (cervical 1–4, cervical 5–8, all other levels), and ambulatory status (able to walk vs. not able to walk). We included four independent demographic variables: age, biological sex (male vs. female), years of education at the third measurement, and race-ethnicity (Hispanic, non-Hispanic Black, non-Hispanic White, and all other race-ethnicity). Because the other race-ethnicity group only had nine participants, we combined them with the non-Hispanic White group in the analyses.

Analyses

All data analyses were completed using SAS 9.4 statistical package (SAS Institute Inc., Cary, NC). We first compared the demographic and injury characteristics between 676 study participants and 863 participants who were lost in the follow-up. Then we compared the demographic and injury characteristics between study participants who were alive and study participants who were deceased by December 31, 2019. We used a t test for continuous variables and chi-square test for categorical variables.

We developed the survival analyses using the person-year logistic regression method, widely used in SCI research,^47,48 and classified participants’ survival status (alive = 0 vs. died = 1) into separate observations, one for each year, starting from 2013 to the year of death or to 2019 for those alive at the end of study period. We built three logistic models using the person-year observations, each considering more successive factors; conceptually more important factors were added to the latter models. The first model included demographic and injury characteristics variables, the second model added one more independent variable (LSQ-R life satisfaction score measured in 2013), and the third model added the life satisfaction history variables (i.e., upward, downward trends). In all three person-year logistic regression models, age and years post injury were time-variant variables, increasing one for each additional person-year observation.

Results

Comparing the 676 study participants and 863 participants lost in the follow-up, we found that the study participants were 1 year younger of age and 1 year more of post injury (Table 1). The study participants also had significantly lower percentage of cervical 1–4 injury level. We did not find other significant differences between the two groups.

Table 1.

Demographic and injury characteristics of participants by their follow-up status

	Participants lost to follow-up (n = 863)	Study participants (n = 676)	p
Age measured at time 1, mean ± SD	45±15	44±12	<.05
Years post injury measured at time 1, mean ± SD	15±11	16±10	<.01
Sex, %			.30
Female	53.90	46.10
Male	56.86	43.14
Race/ethnicity, %			.11
Hispanic	75.00	25.00
NH-Black	58.33	41.67
NH-White	55.18	44.82
Injury level, %			<.05
Cervical 1–4 level	62.84	37.16
Cervical 5–8 level	53.63	46.37
Thoracic, lumbar, and sacral level	55.44	44.56
Ambulatory status, %			.21
Nonambulatory	56.97	43.03
Ambulatory	53.24	46.76

Note: NH = non-Hispanic.

Among the 676 study participants, 135 (20%) were deceased by December 31, 2019. As expected, we found that the survivors were younger and were fewer years post injury compared with those who were decreased. The survivors were also more likely to be female, ambulatory, and have a neurologic injury level lower than cervical (Table 2).

Table 2.

Demographic and injury characteristics of participants by their survival status

	Alive (n = 541)	Deceased (n = 135)	p
Age measured at time 1, mean ± SD	42±11	52±11	<.01
Years post injury measured at time 1, mean ± SD	15±10	21±11	<.01
Sex, %			<.05
Female (n = 189)	85.19	14.81
Male (n = 487)	78.03	21.97
Race/ethnicity, %			.44
Hispanic (n = 6)	66.67	33.33
NH-Black (n = 120)	83.33	16.67
NH-White and others (n = 550)	79.45	20.55
Injury level, %			<.05
Cervical 1–4 level (n = 97)	79.38	20.62
Cervical 5–8 level (n = 243)	74.9	25.1
Thoracic, lumbar, and sacral level (n = 336)	83.93	16.07
Ambulatory status, %			<.01
Nonambulatory (n = 503)	77.53	22.47
Ambulatory (n = 173)	87.28	12.72

Note: NH = non-Hispanic.

Our LSQ-R instruments had consistent reliability across the three measures, with the 0.94 standardized Cronbach coefficient at time 1, 0.93 at time 2, and 0.94 at time 3. The average LSQ-R score was 70 at both time 1 and 2 and 69 at time 3. We also stratified the average LSQ-R score at three time points by three life satisfaction history groups (Table 3). There were 86 (13%) participants whose LSQ-R score increased consistently over the 10 years, 106 (16%) participants whose LSQ-R score decreased over time, and 484 (71%) participants whose LSQ-R score showed variation (other situations).

Table 3.

The average LSQ-R score of three measurements by three life satisfaction history groups

	Increasing group (n = 86)	Decreasing group (n = 106)	Other situations group (n = 484)
Time 1	62	78	69
Time 2	68	70	70
Time 3	77	65	69

Note: LSQ-R = Life Situation Questionnaire–Revised.

Our first multiple logistic model (Table 4) found older age, higher injury levels, and nonambulatory status were significantly associated with greater odds of mortality. In model 2, higher LSQ-R score at time 3 was significantly associated with lower odds of mortality, after controlling all the other injury and demographic covariates. When the life satisfaction history variable was added in the third model, we found that increasing life satisfaction history was significantly associated with lower odds of mortality. For those whose life satisfaction increased during the 10 years, odds of death were 72% lower than for others (OR = 0.28). Satisfaction score at the most recent time of measurement (2013) was marginally significant (p = .07). We calculated the Max-rescaled R-Square as the model fit for all three models in Table 3. It was 0.10 for both model 1 and 2 and 0.12 for model 3. The model fit increased 20% for model 3 compared with model 1 and 2.

Table 4.

Person-year logistic regression models for the survival status

	Model 1			Model 2			Model 3
	OR	95% CI	p	OR	95% CI	p	OR	95% CI	p
Current age	1.07	1.05–1.09	<.01	1.07	1.04–1.09	<.01	1.08	1.05–1.10	<.01
Years post injury	1.01	0.99–1.03	0.28	1.02	1.00–1.04	0.05	1.02	1.00–1.04	.07
Male (ref=female)	1.39	0.90–2.14	0.14	1.38	0.83–2.31	0.22	1.46	0.87–2.45	.15
Race/ethnicity
Hispanic (ref=NH White and others)	3.24	0.74–14.28	0.12	2.21	0.29–17.11	0.45	2.65	0.34–20.77	.35
Non-Hispanic Black (ref=NH White and others)	1.17	0.70–1.96	0.55	1.36	0.75–2.47	0.32	1.32	0.72–2.41	.37
Injury level
Cervical 1–4 injury (ref=other levels)	1.84	1.07–3.16	0.03	2.17	1.18–3.97	0.01	2.31	1.26–4.25	<.01
Cervical 5–8 injury (ref=other levels)	1.87	1.27–2.74	<.01	2.26	1.42–3.58	<.01	2.24	1.41–3.57	<.01
Ambulatory (ref=non-ambulatory)	0.40	0.24–0.66	<.01	0.37	0.20–0.68	<.01	0.33	0.18–0.62	<.01
Years of education at time 3	1.01	0.95–1.07	0.88	1.03	0.96–1.10	0.46	1.01	0.94–1.08	.90
Life satisfaction total score at time 3				0.98	0.97–1.00	<.01	0.99	0.97–1.00	.07
Life satisfaction history
Decreasing life satisfaction (ref=others)							1.02	0.56–1.87	.95
Increasing life satisfaction (ref=others)							0.28	0.13–0.64	<.01

Note: NH = non-Hispanic; OR = odds ratio.

Discussion

A well-established body of research clearly indicates that SCI leads to diminished survival, the extent of which depends on several demographic and injury characteristics, particularly the severity of SCI. Additional research has identified socioeconomic, health, and behavioral factors that favorably or unfavorably impact survival. The unique contribution of this study was the identification of the relationship of life satisfaction, which is a positive psychological characteristic, with survival. The study controlled for demographic and injury characteristics and measured life satisfaction on three occasions to identify the effects of both current satisfaction and changes in pattern of satisfaction over time. Life satisfaction is a positive indicator of subjective well-being instead of the more frequently used negative measures of psychological distress, such as depression.

The current findings indicate that life satisfaction is an important predictor of survival. This is consistent with earlier studies indicating the importance of psychosocial adaptation to survival, even when accounting for health care utilization.⁴⁰ More importantly, the pattern of changes in satisfaction was most highly related to survival, with an increasing trend toward improving satisfaction associated with a greater likelihood of survival. This suggests that as satisfaction improves over time, the probability of survival also improves. Because life satisfaction is related to multiple factors, including participation and health, there is hope that interventions that improve the circumstances for people with SCI and result in greater life satisfaction also show promise to enhance longevity. A diminishing or downward trend in life satisfaction was not significantly related to survival. This is encouraging in that, so long as people maintain a certain level of life satisfaction, the trend of downward life satisfaction itself is not a significant precursor to mortality. On the other hand, it also means that individuals do not receive added benefit for the years of successful living reflected in the higher previous satisfaction. So, the downward trend appears to balance or counteract any benefit of those quality years.

Methodologic considerations

There are several important methodologic considerations. The primary strengths include (1) the focus on survival using satisfaction, a positive indicator of subjective well-being; (2) measurement of the relationship of trends in satisfaction with survival; (3) use of a cohort with significant years post injury since inception; and (4) control for demographic and injury characteristics known related to survival. Identifying changing patterns of satisfaction in relation to longevity, rather than simply looking at satisfaction scores on multiple occasions, led to more parsimonious interpretations of the findings. There have been relatively few studies that have investigated changes in predictors over time and survival/mortality, with the other primary example investigating changes in medication use and corresponding changes in risk of mortality.⁴⁹ Because we used a cohort from the SCI Longitudinal Aging Study, participants had survived the initial years after SCI and life adjustment factors had time to positively impact their survival. The cohort is unique in terms of the number of people reaching aging milestones. The control for known demographic and injury covariates allowed us to evaluate the importance of satisfaction and changes in satisfaction above and beyond the effects of these characteristics.

There also are multiple limitations. First, the sample size was reasonably large for a longitudinal study of health outcomes, but the power was limited to identify significant predictors of survival because the overall number of cases of mortality was relatively small. A larger sample size followed over a longer period of time may have revealed yet more significant findings. Second, some cases were lost due to missing data. We considered utilizing alternative methods, such as Rasch modeling, but chose not to do this because of instability of parameters when classifying individuals based on changing trends and the potential negative effects of estimating missing data. Third, we utilized self-report, which is necessary because life satisfaction represents, by definition, subjective appraisals. However, demographic and injury characteristics were also measured with self-report, and this raises the possibility of misreporting due to errors in recall or knowledge of SCI severity. Fourth, we focused on changing trends in satisfaction, so some cases were lost when mortality occurred prior to completing three measures of life satisfaction or because of nonresponse on one of the times of measurement. This is the downside of having used an aging cohort: It does not capture deaths that occur early after SCI onset, so the findings generalize most prominently to those who have lived several years and even decades with SCI. Fifth, although the design allowed us to establish a temporal pattern between satisfaction and changes in satisfaction with longevity, the findings may not be interpreted as causal as other factors that have not been measured could have influenced both life satisfaction and survival. Lastly, the measurement of change in satisfaction was measured only by consistent change in one direction, so some of the actual changes may have been very small. A substantial portion of participants did not fall into either the upward or downward trend but had mixed patterns, so the findings apply to those who were changing consistently, upward or downward. It is possible for there to be other types of trends for participants who, for example, may have had stable life satisfaction from time 1 to 2 and then sharp changes from time 2 to 3.

Future research

Additional research is needed to better identify factors that enhance survival after SCI. Further research is needed to better understand the mechanisms by which life satisfaction is associated with enhanced survival. There is a need to further investigate positive factors in relation to survival, such as resilience and efficacy, as the predictors of long-term survival may be different than those that predict early excess mortality. Also, there is a need for interventions that address those factors that may be used to promote greater longevity. It is only through continued research that we will successfully identify factors that may promote survival and utilize this knowledge to promote greater longevity among those with SCI.

Conclusion

Our study indicates that both current life satisfaction status and the life satisfaction changing trend are associated with length of survival. The findings show the possibility to enhance longevity after SCI by improving people’s life quality.

Funding Statement

Financial Support The contents of the publication were developed under grants from the National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR) grant numbers 90IFRE0044, 90IF0015, H133G060126, and H133G020239. NIDILRR is a Center within the Administration for Community Living (ACL), Department of Health and Human Services (HHS). The contents of this publication do not necessarily represent the policy of NIDILRR, ACL, or HHS, and you should not assume endorsement by the Federal Government.