Change in Life Satisfaction of Adults With Pediatric-Onset Spinal Cord Injury

Participants 

Study participants were recruited from patients who had received care at the SCI Programs of the Shriners Hospital for Children in Chicago, IL, Sacramento, CA, or Philadelphia, PA. Care is provided free of charge by this hospital system, and there are no financial or insurance restrictions. Patients are permanently discharged from the system when they reach the age of 21 years. To be eligible for the present study, patients had to be injured at age 18 years or younger, be 24 years or older at the time of enrollment to this study, be capable of understanding and responding to the questions, and be living in the United States or Canada. The restriction of age 24 years or older was an arbitrary decision because we felt the early 20s were still a transition period from adolescence and that by age 24 years, we were looking at adulthood. In addition, individuals are usually finished with college and have become employed by 24 years of age.

Of 605 patients who were potentially eligible, 246 (41%) were not enrolled for various reasons (deceased, declined, unable to locate). After further excluding 81 subjects who had baseline interview data but had not completed any follow-up interviews during the study period (1996–2006), a total of 278 (46%) patients were included in the present analyses. There were no significant differences between study participants and nonparticipants in age, sex, and completeness of injury. Those who were included, however, were more likely to be white (89% vs 75%, P<.0001) and have tetraplegia (58% vs 46%, P=.003) than nonparticipants. Details of the study design, subject recruitment and enrollment procedures, and participation versus nonparticipation have been reported previously.17

Data Collection Procedures 

This study was performed with the approval of the local institutional review board. Informed consent was obtained from all participants prior to enrollment. We conducted a structured telephone interview on study participants on an annual basis to obtain information on sociodemographics and physical and psychosocial conditions. The injury-related information was retrieved by medical chart reviews and from the Shriners SCI database.

A total of 1171 interviews from 278 patients were included in the analysis, with a median of 4 interviews a subject (range, 2–8). Participants were followed for an average of 4.1 years (range, 1–9y). The age at the initial interview averaged 27.1 years (range, 24–37y). The years since injury averaged 12.8 years (range, 5–30y) at baseline. Selective characteristics of study participants are shown in table 1.

Table 1. Subject Characteristics at Baseline (N = 278)

	Characteristics	n (%)
	Age (y)
	24–30	227(81.7)
	31–37	51(18.3)
	Sex
	Men	184(66.2)
	Women	94(33.8)
	Race
	White	243(89.3)
	Black	15(5.5)
	Other	14(5.2)
	Age at injury (y)
	0–13	72(25.9)
	14–18	206(74.1)
	Years since injury
	5–15	202(72.7)
	16–30	76(27.3)
	Level of injury
	C1–T6	210(75.5)
	T7–S5	68(24.5)
	Completeness of injury
	Complete	189(68.2)
	Incomplete	88(31.8)
	Presence of medical complications⁎ in the past year
	Yes	144(58.8)
	No	101(41.2)
	Marital status
	Married/living with a partner	56(20.1)
	Single	207(74.5)
	Divorced	12(4.3)
	Separated	3(1.1)
	Education level
	Less than college	187(67.3)
	College or higher	91(32.7)
	Employment status
	Employed	127(45.7)
	Student	48(17.3)
	Unemployed	103(37.0)
	Independent living
	Yes	167(60.1)
	No	111(39.9)
	Illicit drug use
	Yes	37(13.4)
	No	239(86.6)

		Mean ± SD
	CHART
	Physical independence	90.6±14.7
	Mobility	89.0±17.8
	Occupation	80.6±28.3
	Social integration	91.4±14.8
	FIM
	Self-care	31.6±11.8
	Sphincter control	8.5±3.8
	Mobility	12.6±7.1
	Locomotion	7.4±1.6
	Total motor	60.2±21.9
	SF-12
	Physical health	45.6±9.1
	Mental health	52.1±7.9
	SWLS	24.2±6.9

NOTE. Unknown/missing data: 6 persons for race; 2 for illicit drug use; 1 for completeness of injury; 33 for medical complications; 42 for FIM self-care, sphincter control, locomotion, and SF-12; 43 for FIM mobility and total motor; 14 for CHART physical independence; 3 for CHART mobility; 9 for CHART occupation; 5 for social integration.

⁎ One or more of the following complications: (1) urinary tract infections requiring intravenous antibiotics or hospitalization, (2) pressure ulcers requiring hospitalization or surgery, (3) pain that limited the activities, and (4) any other medical conditions that required surgery or hospitalization.

Measures 

All information was self-reported and updated on an annual basis except the injury variables and neurologic status. Demographic information included age, sex, and race. Social factors included marital status, education level, employment status, and living situation (living with or away from parents). Study participants were also asked to identify current use of illegal drugs.

Impairment variables, including level and completeness of injury, were assessed with the International Standards for Neurological Classification of Spinal Cord Injury,30 as documented in the Shriners SCI database. Medical complications were measured during the past year. This report was limited to relatively severe complications, including (1) urinary tract infections requiring intravenous antibiotics or hospitalization, (2) pressure ulcers requiring hospitalization or surgery, (3) pain that limited the activities, and (4) any other medical conditions that required surgery or hospitalization.

Instruments used to assess the physical and psychosocial functioning are well validated with acceptable reliability.31, 32, 33, 34 The motor component of the FIM instrument was used to quantify activity limitation by assessing performance in 4 areas: self-care, sphincter control, mobility, and locomotion. Total FIM motor scores may range from 13 (complete dependence) to 91 (complete independence). Self-perceived health status was assessed with the SF-12, from which mental and physical component scores are generated. Each of the component scores has a range of 10 to 70, with higher scores indicating greater perceived health.35

Social participation was measured by 4 subscales of the CHART: physical independence, mobility, occupation, and social integration.36 Economic self-sufficiency was not used in this analysis because of a large number of missing records. Each of the subscale scores ranges from 0 to 100, with higher scores reflecting greater participation and less handicap. Because of a skewed distribution, each CHART subscale was dichotomized into handicap (<100) versus no handicap (=100) for statistical analysis. It also allowed us to assess the advantage of no handicap, compared with any level of handicap, in life satisfaction.

Subjective overall life satisfaction was assessed with the Diener SWLS, which consists of 5 statements that are each rated on a 7-point Likert-type scale with responses ranging from “strongly disagree” to “strongly agree.”9 Total scores may range from 5 to 35, with higher scores reflecting greater life satisfaction. Psychometric qualities have been found to be satisfactory with internal reliability (Cronbach α) between 0.80 to 0.89 and test-retest reliability from 0.54 (4-y interval) to 0.83 (2-wk interval). The SWLS has also been shown to be sensitive to changes in life circumstances.34, 37

Statistical Procedures 

The outcome of this study was SWLS, which was assessed on an annual basis for up to 9 years. The repeated measures of SWLS on individuals enable a direct assessment of changes over time within individuals. The longitudinal data, nevertheless, require special statistical methods because the set of measurements on 1 subject tends to be correlated. This correlation must be taken into account to draw valid scientific inferences.38 The HLM was therefore chosen to adjust for the potential dependence within the same participants across measurements. The analysis was conducted using the Proc MIXED procedure of SAS 9.1.a

Because of different times of enrollment, loss to follow-up, or other reasons, the sample size for each follow-up year varied. It was 278 patients at baseline, 206 for year 1, 158 for year 2, 170 for year 3, 129 for year 4, 88 for year 5, 71 for year 6, 35 for year 7, 25 for year 8, and 11 for year 9. The HLM is robust to these unequal numbers of interviews and unequally spaced interviews per individual.

In the analyses, each person can be thought of as serving as his or her own control. The HLM model characterized individual person-specific time paths and estimated the average rate of change in SWLS over time across the 278 study participants. The potential contributing factors to the individual time paths were also examined, including time-invariant variables (sex, race, neurologic status, injury-related factors) and time-varying variables (age at interview, years since injury, social factors, medical complications, FIM, SF-12, CHART). Because of unknown/missing records, the sample size for the analysis of each factor was slightly different. As an example of how this analysis was done, consider the case of a person whose status changed from employed at baseline and year 1 follow-up to unemployed at years 2, 3, and 5. This individual contributed observations to the calculation of the regression line for the employed during year 0 through year 1 follow-up, but during years 2 through 5, this person contributed data to the estimation of the regression line for the unemployed.

The HLM estimated the following effects in each model: main effect of the variables, time factor, and interaction between the variables and time. More specifically, the HLM examined the statistical significance of the following:

Satisfaction With Life Scale at Baseline 

The mean SWLS score among 278 patients was 24.2 at baseline, which is somewhat higher than the mean of 19.4 to 21.6 reported in a normative study of persons with adult-onset SCI whose demographic characteristics were similar to the present sample.23

The initial SWLS was significantly higher for women and those who were married or living with a partner; had college or higher education level; were employed or students; did not use illicit drugs; were free of medical complications in the past year; had higher FIM scores; had higher self-perceived mental health; and had a perfect score in CHART mobility, occupation, and social integration, compared with their counterparts (Table 2, Table 3, Table 4). The initial status of life satisfaction did not differ by age, race, living situation, age at injury, duration of injury, level and completeness of injury, FIM sphincter control, FIM locomotion, SF-12 physical health, or CHART physical independence (P>.05).

Table 2. Change in Satisfaction With Life Scale in Relation to Demographic and Medical Characteristics Considered Individually: Results of Hierarchical Linear Model

	Factor	Initial Status		Rate of Change
		Estimated Mean SWLS Score	P Value⁎	Estimated Mean Change per Year	P Value⁎
	Age at interview (y)
	24–30	24.1		0.05
	31–42	24.2	.91	0.15	.53
	Sex
	Men	23.3		0.08
	Women	25.5	.01	0.26†	.30
	Race
	White	24.3		0.11
	Nonwhite	22.8	.27	0.39	.38
	Age at injury (y)
	0–13	25.0		0.09
	14–18	23.7	.16	0.15	.74
	Years since injury
	5–15	24.0		0.14
	16–35	24.1	.89	0.13	.96
	Level of injury
	C1–T6	23.8		0.04
	T7–S5	24.8	.27	0.49‡	.02
	Completeness of injury
	Complete	24.1		0.17
	Incomplete	24.0	.89	0.09	.65
	Presence of medical complications§ in the past year
	Yes	23.4		0.22‡
	No	24.5	.02	0.09	.32

NOTE. Unknown/missing data: 23 person-interviews for race, 4 for completeness of injury, 36 for medical complications.

⁎ P value indicates the significance level of the difference in SWLS score at baseline or change per year between the 2 groups under comparison. † .05<P<.10 for the rate of change for the group under investigation. ‡ P<.05 for the rate of change for the group under investigation. § One or more of the following complications: (1) urinary tract infections requiring intravenous antibiotics or hospitalization, (2) pressure ulcers requiring hospitalization or surgery, (3) pain that limited the activities, and (4) any other medical conditions that required surgery or hospitalization.

Table 3. Change in Satisfaction With Life Scale in Relation to Social Factors and Participation Considered Individually: Results of Hierarchical Linear Model

	Factor	Initial Status		Rate of Change
		Estimated Mean SWLS Score	P Value⁎	Estimated Mean Change per Year	P Value⁎
	Marital status
	Married/living with a partner	26.3		0.14
	Otherwise	23.5	<.01	0.07	.71
	Education level
	Less than college	23.6		0.20‡
	College or higher	24.9	.04	−0.01	.20
	Employment status
	Employed/student	24.7		0.25†
	Otherwise	22.8	<.01	0.04	.13
	Independent living
	Yes	24.4		0.21†
	No	23.4	.07	−0.001	.19
	Illicit drug use
	Yes	22.0		0.11
	No	24.3	<.01	0.13	.94
	CHART: physical independence
	No handicap (=100)	24.3		0.19
	Otherwise (<100)	23.9	.36	0.11	.60
	CHART: mobility
	No handicap (=100)	24.8		0.10
	Otherwise (<100)	23.0	<.01	0.20‡	.46
	CHART: occupation
	No handicap (=100)	24.8		0.14
	Otherwise (<100)	23.1	<.01	0.15	.96
	CHART: social integration
	No handicap (=100)	24.7		0.13
	Otherwise (<100)	23.2	<.01	0.14	.94

Note. Unknown/missing data: 4 person-interviews for illicit drug use, 24 for CHART physical independence, 3 for mobility, 13 for occupation, 5 for social integration.

⁎ P value indicates the significance level of the difference in SWLS score at baseline or change per year between the 2 groups under comparison. † P<.05 for the rate of change for the group under investigation. ‡ .05<P<.10 for the rate of change for the group under investigation.

Table 4. Change in Satisfaction With Life Scale in Relation to Physical Independence and Self-Perceived Health Considered Individually: Results of Hierarchical Linear Model

	Factor	Initial Status		Rate of Change
		Regression Coefficient⁎	P Value	Regression Coefficient†	P Value
	FIM
	Self-care	0.11	<.01	0.003	.68
	Sphincter control	0.13	.07	0.018	.34
	Mobility	0.14	<.01	0.008	.50
	Locomotion	−0.22	.24	−0.010	.84
	Total motor	0.06	<.01	0.003	.45
	SF-12
	Physical health	0.002	.58	−0.001	.36
	Mental health	0.16	<.01	0.012	.12

NOTE. Unknown/missing data: 42 person-interviews for FIM self-care, sphincter control, locomotion, and SF-12; 43 for FIM mobility and total motor.

⁎ Indicates the magnitude of increase or decrease in SWLS score with 1 unit increase in the variable under investigation. † Indicates the magnitude of increase or decrease in the rate of change with 1 unit increase in the variable under investigation.

The correlation between predictor variables was also assessed using Spearman correlation analysis. As anticipated, there was moderate to strong correlation (r≥0.40) between marital status and independent living, between level of injury and FIM total motor, between employment status and CHART occupation subscale, and between FIM total motor and CHART physical independence subscale.

Change in Satisfaction With Life Scale Over Time 

The SWLS score was estimated to increase by 0.14 a year over the course of follow-up (P=.10). The trend toward increasing life satisfaction over time was statistically significant for those who were employed or students (0.25 a year, P=.02), lived independently (0.21 a year, P=.05), had injury level at T7-S5 (0.49 a year, P=.006), and had medical problems in the past year (0.22 a year, P=.02; see Table 2, Table 3).

The estimated rate of change in SWLS over time was significantly higher for those with T7-S5 injury than those with C1-T6 injury (0.49 vs 0.04 a year, P=.02; see table 2). The rate of change in SWLS was not significantly different by sociodemographic variables, presence of any medical complications, or physical and psychosocial functioning. It suggests that people of various characteristics (except level of injury) seem to have a parallel life satisfaction trend over time (fig 1).

View Large Image Download to PowerPoint

Fig 1. Projected trends in the SWLS over the course of follow-up: impact of selective factors. NOTE. The reference group: those who were not married, were unemployed, were free of medical complications in the past year, or had injury at the C1-T6 level.

Multivariable Analysis 

As depicted in table 5, after accounting for all potential contributing factors (P≤.15 while being considered individually in the model), the SWLS at baseline was significantly associated with sex, marital status, employment status, illicit drug use, FIM motor score, SF-12 mental health, and CHART social integration subscale. The rate of change in SWLS was not significantly associated with any factors under investigation (P>.05).

Table 5. Change in Satisfaction With Life Scale: Results of Multivariate Hierarchical Linear Model

	Factor	Initial Status		Rate of Change
		Estimated Difference	P Value	Estimated Difference	P Value
	Sex
	Women versus men	1.86	.01
	Age at injury (y)
	0–13 versus 14–18	0.63	.42
	Level of injury
	T7–S5 versus C1–T6	−0.04	.96	0.103	.59
	Medical complications⁎
	Presence versus otherwise	−0.47	.12
	Marital status
	Married/living with a partner versus otherwise	2.49	<.01
	Education level
	College or higher versus otherwise	0.38	.49
	Employment status
	Employed/student versus otherwise	1.23	.02	0.130	.35
	Living status
	Independent versus otherwise	0.33	.50
	Illicit drug use
	Users versus nonusers	−2.08	<.01
	CHART: mobility
	No handicap versus otherwise	0.46	.28
	CHART: occupation
	No handicap versus otherwise	0.44	.25
	CHART: social integration
	No handicap versus otherwise	1.12	<.01
	FIM
	Total motor score	0.034†	.03
	SF-12
	Mental health	0.171†	<.01	0.007	.36

NOTE. All variables with a significance level ≤0.15 while being considered individually were included in the multivariable model: intercept = 9.89, slope = −0.34 (P=.38).

⁎ One or more of the following complications: (1) urinary tract infections requiring intravenous antibiotics or hospitalization, (2) pressure ulcers requiring hospitalization or surgery, (3) pain that limited the activities, and (4) any other medical conditions that required surgery or hospitalization. † Indicates the magnitude of increase or decrease in SWLS score with 1 unit increase in the variable under investigation.

Longitudinal Trend in Life Satisfaction 

Using 9 years of observations and HLM modeling, the present study examined individual trajectories of life satisfaction and found a small, marginally significant increase in SWLS over the course of follow-up (0.14 a year, P=.10). Among several groups, life satisfaction seemed to increase in meaningful magnitude with statistical significance (P<.05), including individuals who were employed or in school (0.25 a year), lived independently (0.21 a year), and had injury at the T7-S5 level (0.49 a year). It is possible that these individuals have a relatively productive lifestyle, available financial resources, good social support, and/or access to various services, and consequently their life satisfaction tends to increase gradually over time.

Interestingly, the life satisfaction of individuals who experienced medical complications during the past year (including urinary tract infection, pressure ulcer, pain, surgery, or hospitalization) tended to significantly increase by 0.22 a year, although their SWLS at baseline was lower than that of persons who were free of medical complications (see fig 1). This increasing trend needs further confirmation, but seems to suggest that a recent history of medical illnesses would not necessarily have a negative impact on subject well being over time. Also, there might be illness-associated factors that play a more critical role in determining life satisfaction.

Compared with persons with C1-T6 injury, those with T7-S5 injury seem to have a greater increase in SWLS over the course of follow-up, which could be explained by the possibility that several important determinants of greater life satisfaction, such as employment and social participation, are easier to achieve over time for persons with a lower level of injury. This hypothesis also seems to be supported by the finding of a nonsignificant difference in the rate of change between T7-S5 and C1-T6 injury after adjusting for potential confounding factors in the multivariable model.

The rate of change in SWLS was not significantly different among people of various sociodemographic, disability, and participation characteristics. Although the nonsignificance could be a result of an insufficient study power and limited sample size, it suggests a relatively parallel trend in life satisfaction over time—for instance, between the employed/students and the unemployed and between those married/living with a partner and otherwise (see fig 1). Without a change in the status of critical predictors, life satisfaction tends to remain consistently lower for the disadvantaged group.

Based on our knowledge, there was only 1 study previously conducted to determine the individual time paths in life satisfaction in an SCI cohort. Putzke et al28 used the Life Satisfaction Index-A39 to measure life satisfaction among 67 patients with adult-onset SCI, who completed follow-up evaluation at 1, 2, 4, and 5 years postinjury. The result of the analysis of variance with repeated-measure analysis showed a nonsignificant change in life satisfaction across the assessment time points, which is similar to the present finding. There were, nevertheless, no data reported on predictors of life satisfaction in a cross-sectional or longitudinal manner. An indirect evidence of a stable trend in life satisfaction is also provided by several studies based on the SCI Model System Database that indicated a positive and significant association between the 2 consecutive measures of life satisfaction with 1 or more years apart, with those who initially had a high initial SWLS score reporting a high satisfaction level at the follow-up.10, 14, 27

Overall Life Satisfaction 

Consistent with most previous studies of persons with pediatric-onset or adult-onset SCI, we observed that multiple demographic and social factors were significantly associated with overall life satisfaction, even after accounting for potential confounding factors. Those who were married or living with a partner tended to have better life satisfaction than unmarried subjects, as supported by prior studies.17, 27, 40 This finding highlights the importance of the spouse support in successful rehabilitation, perhaps through a number of roles and functions that spouses can contribute, including the facilitation of other social contacts.27

Similar to other studies that used the SCI Model System Database,10, 14 we found that women tended to score higher in SWLS by 1.86 points (SD, 0.72) than men, even after adjusting for various social factors and impairment, disability, and handicap variables. Multiple mechanisms have been proposed to explain sex difference in life satisfaction. For example, women may have a lower standard used to judge for satisfactory accomplishments and to value how satisfactory their life is. The relatively sedentary and homebound lifestyle after SCI may fit better into the traditional female role than the male role.10, 14 Moreover, our previous study suggests that being a woman is predictive of satisfaction with income and sexual experience, which may contribute to greater overall life satisfaction for women than men.41

Although there have been few studies that assessed the impact of the illicit drug use on life satisfaction in the SCI population, the existing research indicated a significantly negative association,16, 17 which is supported by the present study that found a lower SWLS among the illicit drug users than nonusers. Because the information was obtained by telephone interview, it is possible that we underestimated the actual usage of illicit drugs even with confidentiality assured. The direction and magnitude of the bias introduced by the potential underreporting is not clear in the present analysis.

The present finding of a nonsignificant relationship between injury severity and life satisfaction, after accounting for other important factors, is not surprising and requires little discussion because it has been thoroughly covered in the literature.10, 11, 14, 16, 17, 18, 19, 20, 21, 22 The nonsignificant effect of medical complications on life satisfaction was consistent with some studies13, 14, 16 but not others.10, 17, 18, 21 The lack of consistence in the SCI literature could be explained, to some degree, by the differences in types of medical complications under investigation, reporting period (during the past year, month, and so forth), and measures (any complication, total number of complications, and so forth).

Likewise, physical health measured by SF-12 did not play a significant role in life satisfaction in the present analysis. The mental health component of the SF-12 nevertheless impacted life satisfaction significantly while considered individually and also jointly with other factors. This finding indicates the advantage of better psychologic functioning, as suggested by previous research.10, 22

The various dimensions of social participation have received much attention as an important predictor of life satisfaction in recent years, and the evidence is consistent in the SCI literature.10, 11, 12, 14, 15, 16, 17, 18, 19, 20, 22 In the present study, among the 3 dimensions of CHART, social integration seemed to play the most meaningful role in life satisfaction, even considered jointly with marital status and other important variables. The advantage of community mobility and engagement in meaningful activities seemed to decrease after accounting for the effect of employment status, functional independence, and other variables. Because employment status is not synonymous with CHART occupation subscale, the present finding of a greater life satisfaction in individuals who were employed or in school suggests that employment or study may be the key to a satisfied life among those activities included in the CHART occupation measure.

Based on a relatively small number of studies, the impact of physical disability on life satisfaction was inconclusive.10, 11, 13, 14, 17, 18, 19, 20 In the present study, we observed a minimal but significant impact of FIM motor score on life satisfaction, even after adjusting for CHART variables. The mechanism is not clear and needs further confirmation.

The nonsignificant effects of education, independent living, and age at injury on life satisfaction observed in the present study also parallel, to some degree, previous reports.10, 11, 14, 16 The present finding of a nonsignificant relationship of life satisfaction with current age and duration of injury was consistent with some studies,10, 11, 16 but not others.10, 11, 13, 16, 21 The relatively homogenous distribution of age (range, 24–37y at baseline) may limit the present study to examine appropriately the age effect on life satisfaction.

This is the first study conducted to examine a longitudinal change in life satisfaction among adults with pediatric-onset SCI. To address appropriately the sensitive-to-change nature of life satisfaction, the present study updated the change in personal and clinical characteristics on a yearly basis and included them as time-varying predictor variables in the longitudinal analyses. Moreover, this study population is unique in the young age of injury (≤18y) and young adults (27y on average at baseline) but long duration of injury (13y on average at baseline).

Study Limitations 

The study findings however need to be interpreted with caution because of the following limitations. The present study somewhat overrepresents people with non-Hispanic white race/ethnicity and a relatively high level of education. The projected change in life satisfaction may not apply to a long-term trend (10 or more years) because the maximal follow-up was 9 years in the present study and because chronologic aging may influence the trend when people grow much older. The skew distribution and potential ceiling effect limit the present analysis to examine appropriately the linear relationship between the CHART subscales and life satisfaction. As a result, the use of the dichotomous CHART variables (100 vs <100) may in part explain the inconsistence of the present study findings with previous research in which continuous CHART variables were primarily used in their analyses.

Because of constraints with limited data obtained in the present study, we were not able to address the impact of other important variables on life satisfaction, including depression and access to environment. Moreover, because of the lack of a control group, the observed small increase in SWLS over time may be an artifact of repeated testing. Finally, the present findings do not necessarily suggest a causal relationship. In contrast, several characteristics, including social participation, marital status, and employment, could be the consequences of life satisfaction, but not causes. For example, individuals who feel good about life may be more likely to engage in meaningful activities.