Predictive typology of subjective quality of life among participants with severe mental disorders after a five-year follow-up: a longitudinal two-step cluster analysis

Overall, 437 individuals were recruited at baseline (T0), with 352 (80.5 %) consenting to participate. Of the 352 participants enrolled at baseline (T0), 297 (84.4 %) were interviewed at T1 [45]. The 85 individuals who refused to participate were compared with the study participants with respect to age and type of housing. No statistically significant differences were found with respect to the following variables: (1) intermediary resource participants (Chi-square: 5.999 (P = 0.199)); (2) foster home participants (Chi-square: 4.482 (P = 0.482)); or (3) other types of housing (Chi-square: 3.229 (P = 0.665)). Participants were also compared with respect to gender distribution (total sample). No statistically significant difference was found (Chi-square: 1.210; P = 0.271). Of the 297 participants at T1, 207 (69.6 %) were interviewed at T2; 32 (10.8 %) refused to participate; 12 (4.0 %) were excluded due to incapacity; 6 (2.0 %) were deceased, and 40 (13.5 %) had moved away from the study area or were lost to follow-up. Of the 207 participants at T2, 3 were excluded because of the large amount of missing data. The retention rate after five years was thus 59 %. Comparison analyses using cross-tabulation on categorical variables showed no differences in terms of gender and education, either between T0 and T2 samples (gender: X ² = 0.001, p = 0.982; education: X ² = 0.606, p = 0.436) or between T1 and T2 samples (gender: X ² = 0.099, p = 0.754; education: X ² = 0.942, p = 0.344). Moreover, comparison analyses using one-way ANOVA on MCAS scores (community ability) at T0, T1 and T2 yielded no significant differences between the three measurement times (p = 0.406). Comparison analyses were also carried out between participants lost-to-follow-up and those remaining in the sample, with baseline characteristics for age (p = 0.174), and MCAS score (p = 0.533). No statistically significant differences were found.

Participant socio-demographic, clinical and health service use characteristics at baseline are displayed in Table 1. Males represented 52 % of the sample. The mean age was 51 years. Only 40 % had some post-secondary education. More than 50 % of the sample lived in independent apartments and 85 % were single. The main diagnoses were psychotic disorders (schizophrenia: 36 % of the participants, schizophrenia spectrum disorders: 14 %, delusion: 7 %; Total: 49 % of the participants) and mood disorders (42 %). Several participants had more than one mental disorder, the mean being 1.6 (s.d. = 0.9). Table 1 also shows participant characteristics according to deterioration, stability or improvement in relation to the SQOL score. Only 18.6 % (n = 38) of participants saw an improvement in their SQOL over time, while 41.7 % (n = 85) showed stability and 39.7 % (n = 81) experienced a deterioration. Females were more likely to experience improvements in their SLDS scores over time, compared with males, while older participants (50 yrs. and over) were more likely to report a deterioration in their SQOL, compared with younger participants (under 50 yr.). Deterioration in the SLDS score was found in participants living in supervised apartments or other types of supervised housing (group homes, foster homes), whereas stability was found in participants living in independent apartments. Participants with mood disorders and addiction to drugs and alcohol also experienced improvements in their SLDS scores, whereas those with psychotic disorders seemed to deteriorate, and those with personality disorders showed stability.

Table 2 displays the four classes of participants generated from the two-step cluster analysis. The cluster model retained twenty-three variables based on their importance for the characterization of the participants and their discriminative results between clusters. This process led to the elimination of the functional status in the community (MCAS) score, the social support (SPS) score, and help received from relatives (MANQ) score, all of which had yielded very similar mean scores among clusters.

Class 1 contained a higher proportion of participants who were older, female, who had a higher level of education, and who had sources of income other than welfare (i.e., employment income). This class had the highest proportion of individuals with mood and anxiety disorders, and no individuals with psychotic disorders. People in this class had the lowest ACSS and addiction scores (AUDIT and DAST), and the least amount of help from services. They ranked second in terms of SQOL score at T2. This class was labelled as “highly functional older females having mood and anxiety disorders receiving little help from services and with little continuity of access to services but with a mean SQOL.”

Class 2 contained a higher proportion of participants who were younger, female, less educated, mostly affected by personality disorders, mood disorders and addiction, and all living in independent apartments. People in this class had the highest severity of needs scores and the lowest SQOL scores at T2. They ranked second in terms of mood disorders. This class was labelled as “young females with serious needs and co-occurring mental and addiction disorders living in independent apartments and with a very low SQOL.”

Participants in Class 3 were mostly middle-aged (40 to 49 yr.), predominantly male, and well educated, but living on welfare. The majority lived alone. Most of them had psychotic disorders. This class had the third highest SQOL scores at T2. This class was labelled as “middle-aged males, well-educated, but poor and single with psychotic disorders and a mean SQOL.”

Class 4 was composed mainly of older participants (50 yr. and over). They were all single and poorly educated, and the majority were males receiving welfare and living mostly in supervised apartments or in other types of supervised housing. The highest proportion of people with psychotic disorders was in this class. They had the highest ACSS-MH scores, the highest amount of help from services, and the lowest severity of needs scores. They ranked first in terms of SQOL at T2. This class was labelled as “older, poorly educated single males living in supervised apartments or in other types of supervised housing, with psychotic disorders but few serious needs, receiving a large amount of help from services and with a very high SQOL.”

The proportion of participants who saw improvements (18.6 %) in their SQOL after five years was lower in the present study than the 26.4 % reported in the study by Ruggeri et al. [6]. Moreover, the proportion of individuals who saw deterioration in their SQOL was higher than that found in Ruggeri et al. [6] (39.7 % vs 19.8 %). Differing study timeframes (five-year vs two-year follow-up) and choice of instruments (the SLDS vs the Lancashire Subjective Quality of Life Profile [58]) may explain these differences. Moreover, our study was conducted during a major reform of the Quebec mental healthcare system [74]. Under Quebec’s 2005–2010 Mental Health Action Plan, patients living in specialized psychiatric services had to be assessed, and those deemed sufficiently stable were identified and transferred to the primary care sector in order to facilitate their integration into the community. This reform may very well have affected some participants, and may explain deterioration in their SQOL.

Four profiles emerged from the cluster analysis. In one of them (Class 4), the SQOL was higher than the mean at T2; in another one (Class 2), SQOL at T2 was lower; and in the last two (Class 1 and Class 3), SQOL scores at T2 were at the mean. Classes 2 and 4 did not share any of the variables included in the cluster analysis, except for education level, civil status and continuity of care. Interestingly, a large majority of the participants included in the cluster with the highest SQOL (Class 4) lived in supervised apartments or other types of supervised housing, were over the age of 50 years, were single and poorly educated, and had psychotic disorders. A possible explanation for these high scores may be that these individuals were satisfied with living in supervised apartments or other types of supervised housing and had consequently lowered their life expectations [1, 75]. Moreover, given that they all lived in supervised apartments or other types of supervised housing, Class 4 participants received more help from services and benefitted from greater continuity of care. Access to a regular source of healthcare is strongly associated with service use [76] and helps build a good therapeutic alliance between individuals with severe mental disorders and health professionals [77]. This situation often leads to a reduction in unmet needs and an improvement in health outcomes. Not surprisingly, the severity of needs among Class 4 participants was much lower than that in the other classes. According to the literature, the absence of serious needs is the most important predictor of a high SQOL in individuals with severe mental disorders [26, 39]. Furthermore, individuals with psychotic disorders, more particularly schizophrenia, often have a higher SQOL than those with mood disorders or other mental disorders [6, 75], even though their living conditions and life circumstances may be more adverse [75, 78]. This perception may, however, also be the result of poor insight and affective blunting among this clientele [75], or an overestimate of their level of functioning [79].

Conversely, Class 2 contained a higher proportion of personality disorders and addiction, and the second highest proportion of mood and anxiety disorders. It is recognized that individuals with personality disorders are more likely to have a lower SQOL than those affected by other mental disorders [80]. They are also more likely to have a greater number of unmet needs, mainly in the areas of self-care, psychotic symptoms, psychological distress, safety to self, safety to others, alcohol use, sexual expression, and money [81]. Individuals with co-occurring addiction and mental disorders are also very likely to experience unmet needs and are less likely to view their treatment as effective in comparison with those who have mental disorders only [82, 83]. Unlike the Class 4 participants, who lived mainly in supervised apartments or other types of supervised housing, all the Class 2 participants lived in independent apartments. Given that this type of housing offers less access to services, individuals living in independent apartments are required to interact more with their social environments in order to reduce their isolation and improve their SQOL [17].

Like the Class 2 participants, those in Class 1 were mainly females receiving little help from services. Moreover, no participant in these two classes was affected by psychotic disorders. The main differences between Class 1 and Class 2 emerged in the areas of diagnosis, severity of needs, age, education level, sources of income and civil status. Class 1 participants were more affected by mood disorders and anxiety disorders. However, they were less affected by addiction and personality disorders, which may explain their higher SQOL. Moreover, the Class 1 participants were older, better educated, non-single and likely to have better incomes, all variables associated with a higher SQOL [30‐32]. They were also less likely than the Class 2 participants to be suffering from loneliness and isolation.

Like Class 4 participants, those in Class 3 were mainly male, single, receiving welfare as a source of income, and affected by psychotic disorders. The main differences between Class 3 and 4 were in the areas of education level, age, type of housing and severity of needs. Unlike the Class 4 participants, a large majority of the Class 3 participants had pursued a post-secondary education. The fact that most of the Class 3 participants received welfare benefits as their source of income may thus have been a major source of frustration, which could account for both their higher severity of needs and their lower SQOL. Individuals with severe mental disorders living in the community are often victims of stigma in the job market [84]. It is also possible that the individuals with higher education, mainly those who were middle-aged, were more aware of their rights and had difficulty accepting jobs and social conditions below their educational status. Finally, it is possible that, like the Class 2 participants, those in Class 3 living in independent apartments lacked the ability to interact with their social environments.

To the best of our knowledge, this study is the first to establish a predictive typology of SQOL in individuals with severe mental disorders. Using cluster analysis was also innovative in that it included socio-demographic, clinical and health service use variables at T0, along with SQOL measured at a five-year interval (T2). Four different profiles were identified, showing considerable heterogeneity in SQOL after a five-year follow-up, mainly in terms of age, gender, diagnoses, type of housing and severity of needs. Given that predictive SQOL varies in relation to different combinations of socio-demographic, clinical and health service use variables, it would be useful for programs to target predictors specific to the various profiles. The results also show that SQOL at five-year follow-up was particularly low in the class that included more cases of co-occurring mental disorders and addiction and greater severity of needs (Class 2).

Our study highlights the need for mental health service planning and delivery to be better adapted to client profiles in order to improve SQOL. According to the severity of substance abuse problems, priority should be given to the provision of a variety of different services or interventions, including integrated dual disorder treatment, harm reduction or self-help groups such as Alcoholic Anonymous. Concerning individuals in Class 1, who were affected mainly by mood and anxiety disorders and exhibited low severity of needs, access to a regular source of healthcare, such as a family physician, could be sufficient to improve or maintain their SQOL. For individuals in Class 3, who have a high level of education but live on welfare, access to a job integration program seems to be a priority in order to improve their SQOL. Finally, our results show that supervised apartments or other types of supervised housing are adequate for older individuals with psychotic disorders, singles and those with low education, such as the participants in Class 4. Residential services matching the level of functioning of individuals with severe mental disorders should be maintained or developed.