Five-Year Quality of Life Trajectories for Persons With Disabilities

Abstract

The current study identified the longitudinal quality of life (QOL) trajectories of individuals with disabilities as well as the predictors of those trajectories. It examined a secondary data set, the Panel Survey of Employment for the Disabled (PSED), conducted by the Korea Employment Agency for the Disabled (KEAD). Data were gathered over 5 years from individuals with disabilities and analyzed using growth mixture modeling to identify the best QOL trajectory model. Covariates, physical dependence, experience of discrimination, emotional stability, self-esteem, religion, and degenerative type of disability were explored as trajectory predictors. Analysis revealed four latent classes: a high and stable QOL class, a high and varied QOL class, a low and stable QOL class, and a low and varied QOL class. Analysis of predictors indicated degenerative type of disability, physical dependence, discrimination, emotional stability, and self-esteem differentiated the high and stable QOL group from other groups. Significance, limitations, and implications for future practice and research are also discussed.

Keywords

people with disabilities quality of life trajectories growth mixture model

Rehabilitation professionals recently proposed that quality of life (QOL) plays a crucial role in understanding the process of psychosocial adaptation to disability (Bishop, 2005). It is defined as “individuals’ perceptions of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards, and concerns” (World Health Organization, 1998, p. 1), and the term QOL is often used interchangeably with subjective well-being or life satisfaction (Roessler, 1990). Individual QOL perception has frequently been used to measure response to disability and serves as an important indicator of the impact of disability on daily life (Nemec & Gagne, 2005). As a result, QOL is recognized as one of the most important rehabilitation outcomes (G. K. Lee, Chronister, & Bishop, 2008), an underlying goal of all interventions (Crewe, 1980), and a useful way to measure the effectiveness of rehabilitation services (Chan, Rubin, Kubota, Chronister, & Lee, 2003; Degeneffe & Lee, 2010; Frisch, 2004; Lustig & Crowder, 2000). Individuals with disabilities with successful rehabilitation outcomes report higher QOL than their counterparts (Chapin & Holbert, 2010). Researchers have also noted the connection between QOL and a degenerative type of disability (Kemp, 1999). Closer examination of the relationship between this type of disability and QOL may be useful to rehabilitation practitioners to promote successful rehabilitation outcomes.

Despite the obvious importance of QOL as an indicator of response to disability and rehabilitation outcomes, it is still unclear if QOL can change over an individual’s lifetime. Some research suggests QOL is a changeable and subjective perception (Boswell, Dawson, & Heininger, 1998; Degeneffe & Lee, 2010; Jones, Dagnan, Trower, & Ruddick, 1996), which is influenced by daily life events (Boswell et al., 1998; Jones et al., 1996). For example, Niemi, Laaksonen, Kotila, and Waltimo (1988) indicated QOL significantly decreased after the onset of a disability and failed to fully rebound after 4 years. Conversely, other research indicates that although QOL can be influenced by life events such as the occurrence of disabilities, it is a stable concept (Ehrhardt, Saris, & Veenhoven, 2000). Ehrhardt et al. (2000) reported the QOL level indicated by participants was maintained over 10 years. Research also concludes that QOL is influenced by innate traits (Lykken, 1999) or acquired dispositions (Lieberman, 1970), which are not easily changed. In other words, QOL is a fixed concept influenced by innate or acquired features. The inconsistent results about QOL trajectory suggest the possibility of heterogeneous patterns among people with disabilities.

Predictors of QOL

Current evidence suggests QOL trajectories may differ among individuals with disabilities, and certain variables may serve as predictors of these trajectories. For example, research indicates a relationship between QOL and physical dependence (Lez-Salvador et al., 2000), experience of discrimination (Jeong, 2012; J. Lee, 2011; J. S. Lee, 2009), emotional stability (Gorzkowski, Kelly, Klaas, & Vogel, 2010; Kortte, Gilbert, Gorman, & Wegener, 2010; Krause & Edles, 2014), self-esteem (Coyle, Lesnik-Emas, & Kinney, 1994; Dlugonski & Molt, 2012), degenerative type of disability (Germano, Misajon, & Cummins, 2001), and religion (Holt-Ashley, 2000).

Physical Dependence

Previous research provides evidence regarding the relationships reviewed above. For example, Tariah, Hersch, and Ostwald (2006) found when people with disabilities perceive themselves as a burden to others and feel negatively about physical dependence, QOL can decrease. In addition, Lez-Salvador et al. (2000) tested the association between physical dependence and QOL and reported greater physical dependence led to a lower QOL.

Discrimination

Evidence also highlights the relationship between QOL and experience with discrimination. It is clearly evident that discrimination and stigma serve as a barrier to rehabilitation interventions and services, especially for individuals with severe mental illness (Shrivastava, Bureau, Rewari, & Johnston, 2013). Staring, Van der Gaag, Van den Berge, Duivenvoorden, and Mulder (2009) also found the experience of discrimination for individuals with mental illness can be a risk factor for decreased QOL. As experience with discrimination increases, QOL may continue to decrease for individuals with mental illness (Muñoz, Sanz, Pérez-Santos, & Quiroga, 2010).

Emotional Stability

Psychological variables, such as emotional stability, are identified as predictors for QOL. Gorzkowski et al. (2010) showed the role of emotional stability as a mediator between social and job-related participation and QOL. Social and job-related participation influences QOL by reducing depression level. In addition, more positive feelings such as joy or pleasure are associated with higher QOL (Kortte et al., 2010).

Self-Esteem

Research also points to a relationship between self-esteem and QOL among people with disabilities. For example, Nicolson and Anderson (2003) found that people with multiple sclerosis often feel a sense of blame, which results in decreased self-esteem. People with low self-esteem are less receptive to positive stimuli and more receptive to negative stimuli (Ferris et al., 2011), and according to Dlugonski and Molt (2012), QOL can be a consequence of self-esteem. Significant associations between self-esteem and QOL were also revealed among individuals with spinal cord injuries (van Leeuwen, Kraaijeveld, Lindeman, & Post, 2012).

Degenerative Disability

The type of disability, especially if it is an unstable or degenerative condition, can be associated with QOL. For the people with arthritis, the most common chronic disease among the elderly, the pain caused by the degenerative physical state was associated with lower QOL (Germano et al., 2001). More specifically, degenerative physical disabilities, such as arthritis, are associated with decreased functional status (Downe-Wamboldt & Melanson, 1995) and a decreased ability to participate in physical activities (Liang et al., 1984), which are critical for adjustment to disability and can result in decreased QOL (Hewlett, Young, & Kirwan, 1995).

Religion

Furthermore, religious factors are related with QOL. In rehabilitation literature, the emerged disability adjustment models reflected the role of religious beliefs and practices. The Disability Centrality Model (Bishop, 2005) also considered religious expression as one of the important life domains that may influence and even predict QOL. Recent studies provided empirical evidence to support the significant role in QOL for people with disabilities played by religious domains. According to Marini, Glover-Graf, and Millington (2010), most people with spinal cord injury considered religion an important part of their lives. These results were consistent with those reported by Rodriguez, Glover-Graf, and Blanco (2013), which revealed that a number of survey respondents reported themselves as religious persons.

Purpose of the Study

Certain research indicates these variables can not only influence QOL at specific points but also QOL trajectories across time. Resch et al. (2009) showed that the QOL of individuals with lower levels of functional independence was more likely to decrease over time than individuals with higher levels of functional independence. As a result of the relationships between the predictive variables and QOL, the potential influence of these variables on QOL trajectory, and the potential for variable patterns of the QOL trajectory, the current study sought to further investigate these patterns. More specifically, the following research questions were addressed:

Research Question 1: Are there heterogeneous patterns of QOL for individuals with disabilities over time?

Research Question 2: What is the influence of physical dependence, experience of discrimination, emotional stability, self-esteem, degenerative type of disability, and religion on QOL trajectory?

Method

Sample

The current study used a secondary data set, the Panel Survey of Employment for the Disabled (PSED), collected by the Korea Employment Agency for the Disabled (KEAD). The PSED is used to collect data annually, and the data collection process is regulated by the standards of the Korean National Statistical Office. This study analyzed 5 years of the data set: 2008, 2009, 2010, 2011, and 2012, which included 5,092 participants. The data consisted of information on economic activities and employment-related variables for people with disabilities. The ages for participants ranged from 15 to 75 years of age. Table 1 provides information on the type of disability for all participants in the current study.

Table 1.

Disability Type.

Disability type	PD	VI	A	MH	HI	NI	SI	CP	HD	Total
n	2,582	590	30	157	582	221	44	511	375	5,092
%	50.7	11.6	0.6	3.1	11.4	4.3	0.9	10.0	7.4	100.0

Note. PD = physical disability; VI = visual impairment; A = autism spectrum disorder; MH = mental health disorder; HI = hearing impairment; NI = neurological impairment (i.e., cognitive disorder); SI = speech impairment; CP = cerebral palsy; HD = health-related disabilities.

Instrumentation

QOL

Nine items provided measurements of QOL by asking to what extent are you satisfied with family, friends, community, health, income, leisure activities, job, marriage, and overall life satisfaction. Each item was measured using a Likert scale ranging from 1 (very dissatisfied) to 5 (very satisfied). Responses for these items were collected from the 2008, 2009, 2010, 2011, and 2012 data sets. After averaging scores for the nine items in each year, five QOL scores were obtained. Internal consistency estimates ranged from .72 to .83 for each of the 5 years.

Physical dependence

The physical dependency level was measured by asking participants whether they needed outside support each year. Support was defined by the individual’s perception of both formal and informal supports. A binary scoring system was used, and individuals requiring no support received scored 0, whereas individuals requiring support scored 1. Responses were collected from the 2008, 2009, 2010, 2011, and 2012 data sets. After compiling the data, the scores were averaged across the years.

Experience of discrimination

To determine participants’ experience of discrimination, a questionnaire consisting of three items was completed. The first question asked participants about the experience of discrimination in daily activities; the second question asked about experience of discrimination when seeking employment; and the last question asked about the experience of discrimination in employment settings. Each item was measured using an ordinal scale ranging from 1 to 4 (1 = not at all, 2 = rarely, 3 = often, 4 = very often). Responses were collected from the 2010, 2011, and 2012 data sets, rather than over all 5 years due to an increased interest in this variable in 2010 by the KEAD. Scores were averaged for the three items in each year. The internal consistency estimate using Cronbach’s alpha was .68.

Emotional stability

The participants’ emotional stability was only measured in the 2011 data set due to limited interest in this variable by the KEAD. It was measured using a survey consisting of five items inquiring about neurosis, helplessness, stability, depression, and happiness. For example, one item asked, “Have you experienced depression in the last month?” Each item was scored using a 5-point Likert scale (1 = always, 2 = almost always, 3 = often, 4 = rarely, 5 = never). Negative emotions including neurosis, helplessness, and depression were reverse-scored, and all items were averaged to obtain a single score for emotional stability. The internal consistency estimate using Cronbach’s alpha was .85.

Self-esteem

Self-esteem was only measured in the 2012 data set due to limited interest in this variable by the KEAD. The Rosenberg Self-Esteem Scale, which was translated to Korean, was used to measure this variable (Jon, 1974; Rosenberg, 1965). The scale consists of 10 items measuring self-esteem, and each item was scored using a 4-point ordinal scale (1 = never, 2 = somewhat, 3 = almost, 4 = always). The items were then averaged to obtain a single score for self-esteem, and higher scores indicated higher levels of self-esteem. The internal consistency estimate using Cronbach’s alpha was .92.

Degenerative type of disability

Degenerative type of disability was measured by asking respondents whether they felt they had a stable or degenerating condition. We used dummy codes to score the responses with 1 referring to a degenerating condition and 0 referring to a stable condition. Responses were collected from the 2008 data set.

Religion

Participants were asked about religious beliefs in 2008, 2009, 2010, 2011, and 2012. A score of 1 meant that an individual had religious beliefs, and a score of 0 indicated an absence of religious beliefs. After compiling the data, the scores were averaged across the years. Table 2 provides descriptive information for each variable.

Table 2.

Descriptive Information.

Variables	Years	n		M	SD	Minimum	Maximum
QOL	2008	5,092		1.98	.59	0	4
	2009	4,676		2.02	.58	0	4
	2010	4,566		2.00	.56	0	4
	2011	4,395		1.99	.54	0	4
	2012	4,297		1.99	.55	0	4
Physical dependence	2008–2012	5,092		1.39	.91	0	3
Experience of discrimination	2010–2012	4,657		0.80	.73	0	3
Emotional stability	2011	4,396		2.33	.76	0	4
Self-esteem	2012	4,273		1.73	.53	0	3
Religion	2008–2012	4,657		0.45	.43	0	1
Disability type (degenerative condition)	2008	Yes	1,178 (23.1%)
Disability type (degenerative condition)	2008	No	3,119 (61.3%)

Note. QOL = quality of life.

Data Analysis: Growth Mixture Modeling (GMM)

To investigate the potential heterogeneous trajectories of QOL over time, GMM was used in the current study. GMM enables identification of homogeneous subsamples within a heterogeneous sample to distinguish meaningful groups of individual variation over time (Jung & Wickrama, 2008). The current study applied GMM to identify heterogeneous subtrajectories of QOL (outcome variable) at five time points (i.e., 2008, 2009, 2010, 2011, and 2012). GMM approaches do not predetermine that all samples may belong to a single homogeneous population. GMM can identify heterogeneous patterns of QOL that represent, in effect, disabilities (Muthen, 2004). The mean growth curves of these distinct QOL patterns for people with disabilities can then be modeled separately, allowing for unusual flexibility and precision in identifying trajectories of QOL for individuals with disabilities across 5 years. GMM can also be used to test predictors of membership in these trajectories (Muthen, 2004; B. Muthen & Muthen, 2000; Wang & Wang, 2012).

The GMM analyses for QOL consisted of three steps. First, to facilitate model specification, simple growth models with both intercept and linear slopes were used, and quadratic slope was included to help determine the growth parameters for the GMMs. Chi-square, root mean square error of approximation (RMSEA), comparative fit index (CFI), Tucker–Lewis Index (TLI), and standardized root-mean-square residual (SRMR) were reported (L. K. Muthen & Muthen, 2007–2012). RMSEA and SRMR values less than 0.06 were considered ideal, and values less than 0.08 were considered acceptable (Hu & Bentler, 1998, 1999). CFI and TLI values greater than 0.95 were also used, but values greater than 0.90 were considered acceptable (Hu & Bentler, 1998, 1999). If the simple latent curve model (both linear and quadratic) demonstrated a poor trend across years, a rationale for GMM analysis can be provided. Specifically, GMM analysis can be conducted if the result from the general growth curve model shows no overall trend across the sample but remains diverse in QOL.

Second, one-to-five-class unconditional GMM models were compared (with no covariates) and fit indices were assessed. Nylund, Asparouhov, and Muthén (2007) suggested the Bootstrap Likelihood Ratio Test (BLRT; McLachlan, 1987; McLachlan & Peel, 2000) be performed to determine the most appropriate number of classes. It is also useful for several model fit indices to assist in identifying the best number of classes in the initial exploratory step (Tofighi & Enders, 2008). As a result, the combination of model fit indices such as Akaike information criterion (AIC; Akaike, 1973, 1983), Bayesian information criterion (BIC; Schwarz, 1978), sample-size-adjusted BIC (saBIC), entropy value, and the Lo–Mendell–Rubin Likelihood Ratio Test (LRT; Lo, Mendell, & Rubin, 2001) were compared, rather than including the BLRT in the initial exploration steps. Following Nylund et al.’s (2007) recommendation, the combination of indices proved the relatively best fitting model solution. In addition, 1,000 random starting values were specified for initial stage optimizations, 250 random starting values sets for final stage optimizations, and the maximum number of optimization iterations was set at 20.

In the third and final step, the GMM was extended to add covariates as predictors of class membership (Muthen, 2004) by using multinomial logistic regression. Several covariates can prevent model convergence, and as a result, only a subset of the available predictors were included. Figure 1 depicts the structural model used in the current study.

Figure 1.

Structural model.

Results

Initially, a model solution was built by estimating a simple growth model. During the first step, factor loadings for the time points were specified. Using both the likelihood ratio chi-square test as the absolute fit and CFI, TLI, and RMSEA as the relative fit to determine a better model, the models were examined with intercept and slope parameters (linear growth), and intercept, slope, and quadratic parameters (nonlinear growth). Table 3 displays the simple latent linear and quadratic growth models. The quadratic model solution provided a significant improvement in fit over the linear model for life satisfaction (Δχ²/df = 50.084/4 df, p < .05).

Table 3.

Simple Latent Linear and Quadratic Growth Models.

Fit index	Linear	Quadratic
χ²/df	77.384/10 (p < .000)	27.300/6 (p < .000)
CFI	0.995	0.998
TLI	0.995	0.997
RMSEA	0.036	0.026
SRMR	0.017	0.007

Note. CFI = comparative fit index; TLI = Tucker–Lewis Index; RMSEA = root mean square error of approximation; SRMR = standardized root-mean-square residual.

Classes Solution Selection

Based on the recommendations of the model testing, one-to-five-class unconditional models without covariates were compared for both linear and quadratic forms of QOL. To determine the appropriate class solution, the log-likelihood value, the saBIC indices, entropy values, the LRT (Lo et al., 2001), and the BLRT were examined. A model with lower values for the log-likelihood and saBIC criterion indices, higher entropy values, and significant p values for both the LRT and the BLRT was preferred.

Nylund et al. (2007) recommended one could use fit indices such as log likelihood, saBIC, and entropy during the beginning phase because of the increased amount of computing time for the LRT and BLRT. It is equally possible to use fit indices as guides to approximate possible solutions, which could then be reanalyzed by requesting the LRT and BLRT.

As illustrated in Table 4, the information criterion indices showed lower values for each additional class going from two to five classes for both the linear and quadratic forms of QOL. This suggested that the four-class linear (saBIC = 26,213.653) and the three-class quadratic (saBIC = 26,154.195) solutions are optimal.

Table 4.

Fit Indices for One-to-Five-Class Growth Mixture Models for QOL (Unconditional).

Fit index	1 Class	2 Classes	3 Classes	4 Classes	5 Classes
Linear
Log likelihood	−13,114.338	−13,083.134	−13,066.857	−13,055.928	−13,052.596
saBIC	26,282.254	26,235.919	26,219.439	26,213.653	26,223.063
Entropy	1	0.957	0.864	0.712	0.719
LRT p value	NA	.0036	.0011	.0065	.0300
BLRT p value	NA	.0000	.0000	.0000	.3333
# people/class (% of total sample)
1	5,092 (100)	5,032 (98.83)	4,810 (94.46)	4,131 (81.13)
2		60 (01.17)	215 (04.22)	57 (01.12)
3			67 (01.32)	693 (13.60)
4				211 (04.15)
5
Quadratic
Log likelihood		−13,034.986	−13,018.162	−13,019.699
saBIC		26,166.412	26,154.195	26,178.699
Entropy		0.956	0.920	0.757
LRT p value		.0160	.8118	.4968
BLRT p value		.0000	.0000	1.0000
# people/class (% of total sample)
1		5,026 (98.71)	4,921 (96.65)	4,258 (83.62)
2		66 (01.29)	101 (01.99)	58 (01.14)
3			70 (01.37)	776 (15.24)
4				0 (0.00)

Note. QOL = quality of life; saBIC = sample-size-adjusted Bayesian information criterion; LRT = Lo–Mendell–Rubin Likelihood Ratio Test; BLRT = Bootstrap Likelihood Ration Test.

The model with a low saBIC (four linear and three quadratic classes) value and a significant BLRT p value comparing the k and the k-1 class models indicates the statistically best solution. More specifically, comparing the current model against the model with one less class yields a BLRT p value less than .05. For the linear solution, the BLRT showed no statistically significant difference (p = .3333) between the four-class and five-class models. This suggests that the five-class model demonstrated no significant improvement in model fit over the four-class model. For the quadratic solution, the BLRT showed no statistically significant difference (p = 1.000) between the three-class versus four-class models. Ultimately, this indicates the four-class model does not significantly improve the model fit over the three-class version.

In determining the number of classes, it is crucial that one not merely rely on statistical factors such as fit indices, but that the research question, parsimony, theoretical justification, and interpretability must also be evaluated. The totality of these indices, combined with the interpretability and theoretical coherence of a given class solution, guides the final model selection (Muthen, 2004). The quadratic three-class model solution had the lowest log-likelihood and saBIC values, an appropriate value of entropy level, and a BLRT p value. Although the quadratic three-class solution functioned in categorizing changing classes, it only did so with a small proportion of the total data sample (n = 171, 3.36%). Additionally, as Table 5 displays, few estimations in the quadratic model were significant including intercepts or quadratic and linear slopes compared with the linear solution. Thus, the results in the quadratic model were less empirically interpretable.

Table 5.

Comparison of Linear and Quadratic Growth Mixture Solution.

Estimate	Class 1	Class 2	Class 3	Class 4
Linear M/SE
I	2.114 (0.039)**	2.372 (0.139)**	1.531 (0.065)**	1.277 (0.229)**
S	−0.008 (0.006)	−0.343 (0.050)**	−0.016 (0.024)	0.187 (0.056)**
Quadratic M/SE
I	1.994 (0.028)**	1.966 (0.320)**	1.769 (1.530)
S	0.016 (0.007)*	0.308 (0.188)	−0.409 (1.534)
Q	−0.004 (0.002)	−0.135 (0.048)**	0.143 (0.337)

p < .05. **p < .01.

The linear four-class solution offered more detail in classifying each group. The stable QOL class was divided into two groups. The first represented the high intercept of stable QOL, whereas the second represented the low intercept of stable QOL. The stable QOL class was successfully interpreted using the linear four-class solution. The functional form of each class solution was inspected, and the four-class solution revealed high and low intercepts of QOL classes, along with two stable or changing slope classes: (a) a high and stable QOL class, (b) a high and varied QOL class, (c) a low and stable QOL class, and (d) a low and varied QOL class. For the sake of parsimony and interpretability, we selected the optimal four-class solution.

As depicted in Figure 2 and Table 6, the high and stable QOL class, capturing 81.13% (n = 4,131) of the sample, had a trajectory of high intercept and stable slope class with high initial scores and a flat and nonsignificant slope across time. The high and varied QOL class (1.12%, n = 57) had a trajectory of high intercept and changing slope class with high QOL level and a pronounced drop (from 2.37 to 1.00) in QOL for 5 years. The low and stable QOL class (13.60%, n = 693) had a trajectory of low intercept and stable slope class with initially low scores and a flat and nonsignificant slope across time. Finally, the low and varied QOL class (4.15%, n = 211) had a trajectory of low intercept and changing slope class with initially low scores and a gradual increase (from 1.28 to 2.00) in QOL for 5 years.

Figure 2.

Linear class solution.

Table 6.

Parameter Estimates for Each Class in Linear Classes Solution.

Class	n (%)	Intercept		Slope
Class	n (%)	Est. (SE)	p≤	Est. (SE)	p≤
High and stable QOL	4,131 (81.13)	2.114 (0.039)	.01	−0.008 (0.006)
High and varied QOL	57 (1.12)	2.372 (0.139)	.01	−0.343 (0.050)	.01
Low and stable QOL	693 (13.60)	1.531 (0.065)	.01	−0.016 (0.024)
Low and varied QOL	211 (4.15)	1.277 (0.229)	.01	0.187 (0.056)	.01

Note. QOL = quality of life.

The global maximum of model estimation for the four-class linear solution was also tested. The global maximum of model estimation was established by specifying 1,000 initial and 250 final random starting values. The 250 sets of random starting values final stage optimizations found –13,055.928 (i.e., the best log-likelihood value) multiple times. Thus, the four-class linear solution provided evidence for a successful model convergence (Wang & Wang, 2012). To ensure the global maximum of model estimation, two different random seed models were tested. The test indicated the two different random seed model parameters were identical, implying the model estimation had established the global maximum of likelihood (Wang & Wang, 2012).

Prediction of QOL Trajectories

In addition to selecting the best class-number solution, the relationships of predictive variables with QOL class trajectories was also analyzed. The high and stable QOL class was established as the reference class due to its large sample size and tested through multinomial logistic regressions to assess the degree to which the probability of being in the high and stable QOL class was associated with each of the covariates.

As seen in Table 7, the experience of discrimination (odds ratio [OR] = 2.681, p < .01) was strongly associated with increased probability of membership in the high and varied QOL group compared with the high and stable QOL group. The emotional stability (OR = 0.507, p < .01) and self-esteem (OR = 0.270, p < .01) of persons with disabilities both strongly associated with a membership in the low and stable QOL class. People with higher levels of perceived discrimination were more likely to be in the varied QOL class, rather than the stable QOL group. Additionally, individuals with disabilities who had higher levels of emotional stability and self-esteem were more likely to be in the stable QOL class rather than varied QOL group.

Table 7.

Prediction of Life Satisfaction Trajectories.

Predictor	Class #2 (high and varied QOL)		Class #3 (low and stable QOL)		Class #4 (low and varied QOL)
Predictor	Est/SE	OR (95% CI)	Est/SE	OR (95% CI)	Est/SE	OR (95% CI)
Degenerative disability	0.140 (0.406)	1.151 [0.590, 2.245]	0.362 (0.164)*	1.437 [1.098, 1.880]	0.101 (0.263)	1.107 [0.718, 1.706]
Physical dependence	0.122 (0.227)	1.130 [0.778, 1.641]	0.304 (0.104)**	1.356 [1.142, 1.609]	1.016 (0.179)**	2.763 [2.059, 3.708]
Discrimination	0.986 (0.228)**	2.681 [1.841, 3.903]	0.959 (0.098)**	2.609 [2.220, 3.067]	0.214 (0.165)	1.239 [0.945, 1.625]
Emotional stability	−1.310 (0.421)**	0.507 [0.304, 0.845]	−0.377 (0.118)**	0.686 [0.565, 0.833]	−0.488 (0.185)**	0.614 [0.453, 0.833]
Self-esteem	−4.147 (0.576)**	0.270 [0.135, 0.539]	−1.782 (0.197)**	0.168 [0.122, 0.233]	0.190 (0.323)	1.209 [0.711, 2.057]
Religion	−0.040 (0.451)	0.961 [0.457, 2.019]	−0.323 (0.198)	0.724 [0.523, 1.003]	−0.189 (0.299)	0.828 [0.506, 1.354]

Note. QOL = quality of life; OR = odds ratio; CI = confidence interval.

p < .05. **p < .01.

When compared with the high and stable QOL trajectory, degenerative type of disability (OR = 1.437, p < .05), physical dependence (OR = 1.356, p < .01), and experience of discrimination (OR = 2.609, p < .01) were significantly associated with increased probability of membership in the low and stable QOL class. People with disabilities with higher levels of physical dependence, degenerative disabilities, and experience of discrimination were more likely to be in the low and stable QOL class than the high and stable QOL class. In addition, the emotional stability (OR = 0.686, p < .01) and self-esteem (OR = 0.168, p < .01) of people with disabilities were significantly associated with decreased probability of membership in the low and stable QOL class compared with the higher and stable QOL class. Individuals with disabilities with higher levels of emotional stability and self-esteem were more likely to be in the high and stable QOL class than the low and stable class.

Finally, compared with the high and stable QOL class, physical dependence (OR = 2.763, p < .01) was associated with significantly increased probability membership in the low and varied QOL class. Level of emotional stability (OR = 0.614, p < .01) was also significantly associated with decreased probability of membership in the low and varied QOL class compared with the higher intercept and stable trajectory. People with disabilities with higher levels of physical dependence were more likely to experience lower QOL. People with disabilities with higher levels of emotional stability were more likely to have higher QOL.

Discussion

Results indicate four linear trajectories of QOL patterns among individuals with disabilities over a 5-year time period. The first distinct domain in these patterns was trend of slope (stable or varied), and the second domain was related to intercept (high intercept or low intercept). Two classes showed stable trends, and two indicated varied trends. Overall, results yielded four types of QOL trajectories: high and stable QOL, high and varied QOL, low and stable QOL, and low and varied QOL. The results indicate QOL cannot be simply understood as stable or varied, rather the results point to heterogeneous QOL patterns, which aligns closely with previous research in this area (Brown & Vandergoot, 1998; Degeneffe & Lee, 2010; Niemi et al., 1988).

The current study further tested the conditional model with several predictive variables to provide a rationale for the distinct patterns in QOL for individuals with disabilities. Results showed experience of discrimination, degenerative type of disability, physical dependence, emotional stability, and self-esteem can differentiate between high and stable QOL and other classes. Religion did not influence classification of QOL trajectories for individuals with disabilities. Despite the significant role of religious activities for individuals with disabilities (Breslin & Lewis, 2008; Glover-Graf, Marini, Baker, & Buck, 2007; Johnstone, Glass, & Oliver, 2007; Reyes-Ortiz, 2006; Rodriguez et al., 2013), the current study did not find statistically significant associations with each QOL trajectory.

Among predictor variables in this study, experience of discrimination and self-esteem played an important role in the distinction between the high and stable QOL class and high and varied QOL class. When comparing low and varied QOL with high and stable QOL, physical dependence was significant, as high physical dependency was associated with low QOL, while experience of discrimination and self-esteem were not significant. A degenerative type of disability influences the distinction between low and stable QOL and high and stable QOL. These results imply experience of discrimination, self-esteem, and physical dependency influence change in QOL over time, as well as the level of QOL. Whereas a degenerative type of disability does not contribute to change in QOL, but it does influence level of QOL. Therefore, interventions focusing on strategies that decrease experience of discrimination and physical dependency and/or increase self-esteem may be more effective than strategies focused on improving a degenerative disability.

Limitations and Implications for Future Research

There are several limitations in the current study. First, the findings in this study can differ by disability types (e.g., mental health disability, hearing impairment, or physical disability). For example, physical dependency may not be a concern for individuals with mental health disabilities, but emotional stability and self-esteem may have more influence on QOL for this population. In addition, over 50% of the sample for the current study was made up of individuals with physical disabilities. As a result, the results may not accurately reflect the influence of predictor variables on other disability types. Future research must examine the differential effect of predictors on QOL for persons with various types of disability.

Although, results identified QOL trajectory and indicated this trend can be stable and varied, the level of QOL for participants in this study did not extend beyond an average level of QOL. For example, the QOL level for individuals with disabilities who belong to the high and stable QOL class typically scored in the midrange (the score of intercept is 2.114). It is possible that subclasses exist within the high and stable QOL class. Additionally, the possibility of improving QOL must not be ignored as the slope was positive and statistically significant among classes. Previous research highlights the ability to improve QOL (Ansbacher & Ansbacher, 1956; Maslow, 1964; Rogers, 1963), and as a result, future research must continue to investigate strategies to improve and maintain QOL, perhaps by categorizing latent classes into subclasses.

QOL was measured by only nine items in the current study, including areas such as family, friends, health, and income. These items may not be sufficient to fully detect overall level of QOL among individuals with disabilities. The WHOQOL-BREF (The World Health Organization Quality of Life–Brief; WHOQOL Group, 1995, 1998a, 1998b) is used to measure QOL in several countries (Hao, Fang, & Power, 2006) and is composed of 26 items with four subscales; Physical, Psychological, Social, and Environment. The WHOQOL-BREF was also validated in Korea among individuals with disabilities as well people without disabilities (Min et al., 2002). Future research may use the WHOQOL-BREF or other QOL measures to provide a more accurate measure of QOL with stronger psychometric properties.

Implications for Practice

Results indicate individuals with disabilities often experience low to midrange levels of QOL, and trajectories can be stable and varied. Rehabilitation practitioners must begin to assess QOL to fully understand the experience of individuals with disabilities, and interventions can be designed accordingly (Bishop, 2005). For example, if assessment results indicate concerns related to social support from family and friends, rehabilitation professionals must prioritize strategies to increase social support among individuals with disabilities.

Attention to the predictor variables examined in this study may also aid in increasing QOL for individuals with disabilities. Results point to the experience of discrimination as an influential factor in the level and variability of QOL. Unfortunately, the resolution of discrimination and attitudinal barriers is a challenging task, which cannot be “fixed” through any single action. Deliberate forward progress in education and advocacy efforts will reduce the impact of discrimination on individuals with disabilities. Chan, Livneh, Pruett, Wang, and Xi Zheng (2009) highlight the importance of using multidimensional strategies around (a) increasing contact, (b) education, (c) social influence, (d) disability simulations, (e) protests, (f) political efforts, and impression management to decrease the impact of negative attitudes toward individuals with disabilities. Rehabilitation professionals may also find benefit to involving individuals with disabilities in these efforts.

Physical dependency was also found to influence level and variability of QOL in the present study. In addition, degenerative disabilities impact the level of QOL according to the results. Interventions designed to increase personal control may address physical dependency and degenerative type of disability and the influence of these variables on QOL. For example, the provision of information related to the degenerative nature of the disability, assistance in the development of self-management skills, and developing accommodations that increase the experience of control for individuals with disabilities may serve as useful strategies to address the influence of physical dependency and degenerative type of disability on QOL (Bishop, 2005).

Results of the current study also point to the importance of self-esteem on the level and variability of QOL experience by individuals with disabilities. Interventions that increase meaningful activities (e.g., employment) may lead to improved self-esteem and further impact QOL (Viemero & Krause, 1998). It may be necessary to assist individuals with disabilities in exploring new interests, new social outlets, and new ways of engaging in life to increase perceptions of self-esteem and QOL (Bishop, 2005). Additional efforts to improve coping strategies, goal setting, and reframing expectations may also lead to improved self-esteem, along with improvement in the other variables discussed above (Schwartz, Andresen, Nosek, & Krahn, 2007). These strategies will directly or indirectly influence the individual’s perception of QOL, and as a result, it is essential for practitioners to design plans with attention to interventions such as those discussed above.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by research fund from Honam University, 2015.

References

Akaike

(1973). Information theory and an extension of the maximum likelihood principle. In Petrov

B. N.

Csaki

(Eds.), Second International Symposium on Information Theory (pp. 267–281). Budapest, Hungary: Akademiai Kiado.

Akaike

(1983). Information measures and model selection. International Statistical Institute, 44, 277–291.

Ansbacher

H. L.

Ansbacher

R. R.

(1956). The individual psychology of Alfred Adler. New York, NY: Basic Books.

Bishop

(2005). Quality of life and psychosocial adaptation to chronic illness and disability: Preliminary analysis of a conceptual and theoretical synthesis. Rehabilitation Counseling Bulletin, 48, 219–231.

Boswell

B. B.

Dawson

Heininger

(1998). Quality of life as defined by adults with spinal cord injuries. Journal of Rehabilitation, 64, 27–32.

Breslin

M. J.

Lewis

C. A.

(2008). Theoretical models of the nature of prayer and health: A review. Mental Health, Religion & Culture, 11, 9–21. doi:10.1080/13674670701491449

Brown

Vandergoot

(1998). Quality of life for individuals with traumatic brain injury: Comparison with others living in the community. Journal of Head Trauma Rehabilitation, 13, 1–23.

Chan

Livneh

Pruett

Wang

Xi Zheng

(2009). Societal attitudes toward disability: Concepts, measurements, and interventions. In Cardoso

E. D.

Chronister

J. A.

(Eds.), Understanding psychosocial adjustment to chronic illness and disability: A handbook for evidence-based practitioners in rehabilitation (pp. 333–367). New York, NY: Springer.

Chan

Rubin

S. E.

Kubota

Chronister

Lee

(2003). Counselors’ and consumers’ derived criteria for assessing the effectiveness of rehabilitation services. International Journal of Rehabilitation Research, 26, 335–337. doi:10.1097/00004356-200312000-00014

10.

Chapin

M. H.

Holbert

(2010). Employment at closure is associated with enhanced quality of life and subjective well-being for persons with spinal cord injuries. Rehabilitation Counseling Bulletin, 54, 6–14.

11.

Coyle

C. P.

Lesnik-Emas

Kinney

(1994). Prediction life satisfaction among adults with spinal cord injuries. Rehabilitation Counseling Bulletin, 39, 95–112.

12.

Crewe

N. M.

(1980). Quality of life: The ultimate goal in rehabilitation. Minnesota Medicine, 63, 586–589.

13.

Degeneffe

C. E.

Lee

G. K.

(2010). Quality of life after traumatic brain injury: Perspectives of adult siblings. Journal of Rehabilitation, 76(4), 27–36.

14.

Dlugonski

Molt

R. W.

(2012). Possible antecedents and consequences of self-esteem in persons with multiple sclerosis: Preliminary evidence from a cross-sectional analysis. Rehabilitation Psychology, 57, 35–42.

15.

Downe-Wamboldt

B. L.

Melanson

P. M.

(1995). Emotions, coping and psychological well-being in elderly people with arthritis. Western Journal of Nursing Research, 17, 250–265.

16.

Ehrhardt

J. J.

Saris

W. E.

Veenhoven

(2000). Stability of life-satisfaction over time. Journal of Happiness Studies, 1, 177–205.

17.

Ferris

D. L.

Rosen

C. C.

Johnson

R. E.

Brown

D. J.

Risavy

Heller

(2011). Approach or avoidance (or both?): Integrating core self-evaluations within an approach/avoidance framework. Personnel Psychology, 64, 137–161.

18.

Frisch

M. B.

(2004). Quality of life assessments and interventions: The Quality of Life Inventory. In Maruish

M. E.

(Ed.), The use of psychological testing for treatment and outcome assessment (3rd ed., pp. 1277–1332). Mahwah, NJ: Lawrence Erlbaum.

19.

Germano

Misajon

Cummins

R. A.

(2001). Quality of life and sense of coherence in people with arthritis. Journal of Clinical Psychology in Medical Settings, 8, 253–261.

20.

Glover-Graf

N. M.

Marini

Baker

Buck

(2007). Religious and spiritual beliefs and practices of people with chronic pain. Rehabilitation Counseling Bulletin, 51, 21–33.

21.

Gorzkowski

J. A.

Kelly

E. H.

Klaas

S. J.

Vogel

L. C.

(2010). Girls with spinal cord injury: Social and job-related participation and psychosocial outcomes. Rehabilitation Psychology, 35, 58–67.

22.

Hao

Y. T.

Fang

J. Q.

Power

M. J.

(2006). The equivalence of WHOQOL BREF among 13 culture versions. Chinese Mental Health Journal, 20, 71–75.

23.

Hewlett

Young

Kirwan

(1995). Dissatisfaction, disability, and rheumatoid arthritis. Arthritis Care and Research, 8(1), 4–9.

24.

Holt-Ashley

(2000). Nurses pray: Use of prayer and spirituality as a complementary therapy in the intensive care setting. AACN Clinical Issues, 11, 60–67.

25.

L.-T.

Bentler

P. M.

(1998). Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification. Psychological Methods, 3, 424–453.

26.

L.-T.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55.

27.

Jeong

(2012). The effect of environmental factors regarding employment on acceptance of disabilities. Disability & Employment, 22, 323–350.

28.

Johnstone

Glass

B. A.

Oliver

R. E.

(2007). Religion and disability: Clinical, research and training considerations for rehabilitation professionals. Disability & Rehabilitation, 29, 1153–1163.

29.

Jon

B. J.

(1974). Self-esteem: A test of its measurability. Journal of Yonsei Review, 11, 107–130.

30.

Jones

Dagnan

Trower

Ruddick

(1996). People with learning disabilities living in community-based homes: The relationship of quality of life with age and disability. International Journal of Rehabilitation Research, 19, 219–227.

31.

Jung

Wickrama

K. A. S.

(2008). An introduction to latent class growth analysis and growth mixture modeling. Social and Personality Psychology Compass, 2, 302–317.

32.

Kemp

B. J.

(1999). Quality of life while aging with a disability. Assistive Technology, 11, 158–163.

33.

Kortte

K. B.

Gilbert

Gorman

Wegener

S. T.

(2010). Positive psychological variables in the prediction of life satisfaction after spinal cord injury. Rehabilitation Psychology, 55, 40–47.

34.

Krause

J. S.

Edles

P. A.

(2014). Injury perceptions, hope for recovery, and psychological status after spinal cord injury. Rehabilitation Psychology, 59, 176–182.

35.

Lee

G. K.

Chronister

Bishop

(2008). The effects of psychosocial factors on quality of life among individuals with chronic pain. Rehabilitation Counseling Bulletin, 51, 177–189.

36.

Lee

(2011). The effect of experienced discrimination and self-stigma on life satisfaction of people with disabilities. Social Science Research Review, 27, 277–299.

37.

Lee

J. S.

(2009). The analysis of factors influencing on the life satisfaction of poor disabled persons—Focus on registration disabled persons of Gwang-ju city in south Korea. Journal of the Korean Association for Governance, 16, 79–105.

38.

Lez-Salvador

T. G.

Lyketsos

C. G.

Baker

Hovanec

Roques

Brandt

Steele

(2000). Quality of life in dementia patients in long-term care. International Journal of Geriatric Psychiatry, 15, 181–189.

39.

Liang

Larso

Thompson

Eaton

McNamara

Katz

Taylor

(1984). Costs and outcomes in rheumatoid arthritis and osteoarthritis. Arthritis and Rheumatism, 27, 522–529.

40.

Lieberman

L. R.

(1970). Life satisfaction in the young and the old. Psychological Reports, 27, 75–79.

41.

Mendell

Rubin

(2001). Testing the number of components in a normal mixture. Biometrika, 88, 767–778.

42.

Lustig

D. C.

Crowder

(2000). The quality of life of persons with severe and persistent mental illness: A typology based on cluster analysis. Journal of Applied Rehabilitation Counseling, 31(3), 22–29.

43.

Lykken

D. I.

(1999). Happiness: What studies on twins shows us about nature, nurture and the happiness setpoint. New York, NY: Golden Books.

44.

Marini

Glover-Graf

N. M.

Millington

(2010). Psychosocial aspects of disability: Insider perspectives and strategies for counselors. New York, NY: Springer

45.

Maslow

A. H.

(1964). The superior person. Transaction, 1, 17–26.

46.

McLachlan

G. J.

(1987). On bootstrapping the likelihood ratio test statistic for the number of components in a normal mixture. Applied Statistics, 36, 318–324.

47.

McLachlan

Peel

(2000). Finite mixture models, New York, NY: John Wiley.

48.

Min

S. K.

Kim

K. I.

Lee

C. I.

Jung

Y. G.

Suh

S. Y.

Kim

D. K.

(2002). Development of the Korean versions of WHO Quality of Life Scale and WHOQOL BREF. Quality of Life Research, 11, 593–600.

49.

Muñoz

Sanz

Pérez-Santos

Quiroga

M. A.

(2010). Proposal of a socio-cognitive-behavioral structural equation model of internalized stigma in people with severe and persistent mental illness. Psychiatry Research, 186, 402–408.

50.

Muthen

(2004). Latent variable analysis: Growth mixture modeling and related techniques for longitudinal data. In Kaplan

(Ed.), Handbook of Quantitative Methodology for the Social Sciences (pp. 345–368). Newbury Park, CA: SAGE.

51.

Muthen

L. K.

(2000). Integrating person-centered and variable-centered analyses: Mixture modeling with latent trajectory classes. Alcoholism: Clinical and Experimental Research, 24, 882–891.

52.

Muthen

L. K.

Muthen

B. O.

(2007–2012). Mplus user’s guide (6th ed.). Los Angeles, CA: Author.

53.

Nemec

P. B.

Gagne

(2005). Recovery from psychiatric disabilities. Journal of Applied Rehabilitation Counseling, 36(4), 4–10.

54.

Nicolson

Anderson

(2003). Quality of life, distress and self-esteem: A focus group study of people with chronic bronchitis. British Journal of Health Psychology, 8, 251–270.

55.

Niemi

M. L.

Laaksonen

Kotila

Waltimo

(1988). Quality of life 4 years after stroke. Stroke, 19, 1101–1107.

56.

Nylund

K. L.

Asparouhov

Muthén

B. O.

(2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14, 535–569.

57.

Resch

J. A.

Villarreal

Johnson

C. L.

Elliott

T. R.

Kwok

Berry

J. W.

Underhill

A. T.

(2009). Trajectories of life satisfaction in the first 5 years following traumatic brain injury. Rehabilitation Psychology, 54, 51–59.

58.

Reyes-Ortiz

C. A.

(2006). Spirituality, disability, and chronic illness. Southern Medical Journal, 99, 1172–1173.

59.

Rodriguez

V. J.

Glover-Graf

N. M.

Blanco

M. S.

(2013). Conversations with God: Prayer and bargaining in adjustment to disability. Rehabilitation Counseling Bulletin, 56, 215–228. doi:10.1177/0034355213477477

60.

Roessler

R. T.

(1990). A quality of life perspective on rehabilitation counseling. Rehabilitation Counseling Bulletin, 34, 82–91.

61.

Rogers

C. R.

(1963). The concept of the fully functional person. Psychotherapy: Theory, Research, and Practice, 1, 17–26.

62.

Rosenberg

(1965). Society and adolescent child. Princeton, NJ: Princeton University Press.

63.

Schwartz

C. E.

Andresen

E. M.

Nosek

N. A.

Krahn

G. L.

(2007). Response shift theory: Important implications for measuring quality of life in people with disability. Archives of Physical Medicine and Rehabilitation, 88, 529–536.

64.

Schwarz

(1978). Estimating the dimension of a model. Annals of Statistics, 6, 461–464.

65.

Shrivastava

Bureau

Rewari

Johnston

(2013). Clinical risk of stigma and discrimination of mental illnesses: Need for objective assessment and quantification. Indian Journal of Psychiatry, 55, 178–182.

66.

Staring

A. B.

Van der Gaag

Van den Berge

Duivenvoorden

H. J.

Mulder

C. L.

(2009). Stigma moderates the associations of insight with depressed mood, low self-esteem, and low quality of life in patients with schizophrenia spectrum disorders. Schizophrenia Research, 115, 363–369.

67.

Tariah

H. A.

Hersch

Ostwald

S. K.

(2006). Factors associated with quality of life: Perspectives of stroke survivors. Physical and Occupational Therapy in Geriatrics, 25(2), 33–50.

68.

Tofighi

Enders

C. K.

(2008). Identifying the correct number of classes in growth mixture models. In Hancock

G. R.

Samuelsen

K. M.

(Eds.), Advances in latent variable mixture models (pp. 317–341). Charlotte, NC: Information Age.

69.

van Leeuwen

C. M. C.

Kraaijeveld

Lindeman

Post

M. W. M.

(2012). Associations between psychological factors and quality of life ratings in persons with spinal cord injury: A systematic review. Spinal Cord, 50, 174–187.

70.

Viemero

Krause

(1998). Quality of life in individuals with physical disabilities. Psychotherapy and Psychosomatics, 67, 317–322.

71.

Wang

(2012). Structural equation modeling: Applications using Mplus. New York, NY: John Wiley.

72.

World Health Organization. (1998). WHOQOL User manual. Geneva, Switzerland: Author.

73.

The World Health Organization Quality of Life Group. (1995). The World Health Organization Quality of Life assessment (WHOQOL): Position paper from the World Health Organization. Social Science Medicine, 41, 1403–1409.

74.

The World Health Organization Quality of Life Group. (1998a). Development of the World Health Organization WHOQOL-BREF quality of life assessment. Psychological Medicine, 28, 551–558.

75.

The World Health Organization Quality of Life Group. (1998b). The World Health Organization Quality of Life assessment (WHOQOL): Development and general psychometric properties. Social Science Medicine, 46, 1569–1585.