Predictors for quality of life improvement in schizophrenia: a prospective study of patients in hospital-based rehabilitation programs

Abstract

Background

Enhancing quality of life (QoL) is a crucial goal in the rehabilitation of patients with schizophrenia (SCZ). Identifying factors associated with the outcomes of rehabilitation is of clinical importance. This study examined predictors of subjective QoL (SQoL) outcomes in SCZ patients undergoing a structured 24-week rehabilitation program.

Methods

A total of 235 patients with SCZ completed the program. SQoL was assessed using the Schizophrenia Quality of Life Scale (SQLS), comprising three subdomains: psychosocial (PS), motivation/ energy (ME), and symptoms/side effects (SS). Clinical symptoms were evaluated using the Brief Psychiatric Rating Scale (BPRS), Patient Health Questionnaire-9 (PHQ-9), and Generalized Anxiety Disorder Scale (GAD-7). Cognitive impairment was assessed using the Ascertain Dementia 8 (AD8). Medication side effects were assessed using the Treatment Emergent Symptom Scale and Rating Scale for Extrapyramidal Side Effects.

Results

After 24 weeks, the response rate (≥ 20% reduction of SQLS total score post-training from baseline was 72.8% (171/235). Multiple regression analysis revealed that SQoL improvement was negatively associated with baseline scores of BPRS negative symptoms (P = 0.017) and PHQ-9 (P = 0.034), while positively associated with changes in BPRS negative symptoms (P = 0.031) and PHQ-9 (P = 0.034), and baseline PS subdomain (P = 0.021). The predictive model yielded an area under the curve of 0.798, which increased to 0.926 when changes in clinical symptoms were included.

Conclusion

Structured rehabilitation training may significantly improve SQoL in patients with SCZ. Both baseline levels and changes in psychiatric symptoms, cognition, anxiety, and baseline PS subdomain levels significantly predict SQLS improvement, suggesting they could be targets for future comprehensive and individualized interventions.

Trial registration

This protocol was registered on February 21, 2021, at chictr.org.cn (Identifier: ChiCTR2100043537).

Introduction

Schizophrenia (SCZ) is a chronic and highly disabling mental disorder that significantly impairs cognitive, emotional, and social functioning, leading to a substantial decline in quality of life (QoL) for affected individuals and imposing a considerable societal burden [1]. As QoL is a critical outcome measure in SCZ and often predictive of relapse [2], enhancing QoL in these patients is a crucial goal. Similar to symptoms, the QoL of SCZ patients often fluctuates. For example, a report from the Netherlands showed that more than half (56%) of older Dutch individuals with SCZ reported clinically relevant changes in subjective QoL within 5 years of follow-up, with some experiencing improvement and others deterioration [3].

Rehabilitation training is essential and widely recommended for improving functional outcomes and promoting recovery in SCZ after acute treatment, which often involves different medications [4–8]. Rehabilitation training has shown promise in improving QoL by targeting various domains, including self-monitoring of symptoms, drug self-management, social communication skills, and independent living [9, 10]. Prior studies have identified various predictive factors associated with QoL in SCZ patients undergoing rehabilitation training. These factors include cognitive function, satisfaction with psychiatric treatment, social and emotion-related symptoms [11], along with physical functioning [12]. Additionally, demographic variables (such as age, gender, education, marital status, and employment status), as well as treatment-related factors, such as medication types and dosages, and rehabilitative training, have also been linked to QoL outcomes [13, 14]. Beyond treatment, strong social support, particularly family support [15], is also associated with improved QoL.

However, most of the studies mentioned above used cross-sectional designs [11, 12], or observation studies [13, 14], limiting in identification of predictive factors [16]. Only a few longitudinal studies have explored the predictive factors affecting the efficacy of rehabilitation training in SCZ patients. For instance, Ehrminger et al. showed that baseline insight predicted changes in depression and also independently predicted changes in QoL [17]. Similarly, Prouteau et al. reported that baseline cognitive abilities, such as visual memory and attention, predict different aspects of QoL during rehabilitation in 55 SCZ patients who participated in an integrated rehabilitation program. Specifically, visual memory predicted better community function, while poorer attention correlated with higher subjective QoL, verbal memory and facial affect recognition were linked to objective QoL improvements [18].

QoL includes both subjective and objective aspects. The objective QoL is usually measured with externally observable indicators such as employment status, living conditions, and social functioning, which are typically assessed through standardized measures or reports from reliable informants [19]. Subjective QoL (SQoL) is a self-reported personal reflection, combining cognition and affect [20]. This personal appraisal of life, also referred to as subjective well-being or life satisfaction, can be as important as a healthcare professional’s assessment of a health condition’s impact. Emphasizing SQoL encourages a more accurate evaluation of an individual’s experiences, prioritizing the person over their health condition. Despite its importance, the holistic predictive factors influencing SQoL during rehabilitation training remain unclear, particularly regarding the longitudinal changes involving clinical symptoms and cognitive function [18, 21]. This highlights a significant research gap, as there is a lack of comprehensive studies that investigate the interplay between sociodemographic factors, disease characteristics, clinical symptoms, and cognitive function in predicting SQoL improvements for SCZ patients undergoing rehabilitation training [8].

Given these gaps in the literature, our study aimed to investigate the predictors of SQoL in SCZ patients undergoing rehabilitation training, focusing on longitudinal changes in clinical symptoms. Specifically, our goals were to explore how demographic factors, clinical symptoms, cognition, and side effects at baseline predict changes in SQoL after training. We hypothesized that baseline clinical symptoms, cognition, and QoL would influence SQoL improvements in these patients. To test the hypothesis, we first analyzed the impact of baseline clinical symptoms, side effects, and cognition on SQoL outcomes following rehabilitation. We then examined the association between changes in these factors and changes in SQoL. Finally, we developed predictive models to identify the key determinants of SQoL improvement in SCZ patients.

Methods

Sample and training

The study sample consisted of patients diagnosed with SCZ from 18 psychiatric hospitals across 13 provinces and regions of China. One of our previous studies reported factors related to QoL of SCZ patients using cross-sectional baseline information [22]. Inclusion criteria were as follows: patients aged 18–65 years, diagnosed with schizophrenia according to DSM-IV diagnostic criteria. Exclusion criteria included unstable medical conditions, blindness or deaf-muteness, other severe physical disabilities, comorbidities with other mental disorders, including mental retardation, dementia, or other severe cognitive impairment, and active alcohol/substance use disorders; unable to complete relevant tests.

The study was registered (Chictr.org.cn: ChiCTR2100043537) and approved by the Ethics Committee of Beijing Anding Hospital, Capital Medical University (2020-research-45), and in line with the principles of the Declaration of Helsinki. All participants or their family members provided written informed consent.

Rehabilitation training

The rehabilitation training modules were based on psychiatric rehabilitation by Liberman [9] which included social skills training, skills for self-management of illness involving symptoms of their illness, together with recommendations for evidence-based treatment with medications. Patients underwent a 24-week rehabilitation training. The program was divided into three phases: the first 8 weeks focused on medication management skills, the next 8 weeks focused on symptom management skill training, and the final 8 weeks focused on social skills training. The training consisted of 3 sessions per week, each lasting 45 min.

Measures

Data collection

Data were collected through face-to-face interviews using a locally developed questionnaire, which included socio-demographic characteristics (e.g., gender, age, education, marital status, employment, social support, etc.), disease characteristics (e.g., age of onset, duration of illness, and antipsychotic medications.) and clinical symptoms (e.g., psychiatric symptoms, anxiety, depression, and cognition). Data collection of clinical symptoms was recorded at baseline and after 24 weeks of rehabilitation training.

Dependent variable

SQoL assessment

The SQoL was assessed using the Schizophrenia Quality of Life Scale (SQLS) [17], which consists of 30 self-assessment items and provides scores in three domains: psychosocial (PS, 15 items), motivation/ energy (ME, 7 items), and symptoms/ side effects (SS, 8 items). Each item is scored on a scale of 0 to 4 (0: never, 1: rarely, 2: sometimes, 3: often, 4: always). The total score ranges from 0 (best status) to 100 (worst status). Clinically significant SQoL change is marked by a half Standard Deviation shift in SQLS post-rehabilitation [23]. Participants are divided into responders (half SD or greater improvement) and non-responders (less than half SD improvement) based on their SQLS score changes post-rehabilitation.

Covariates measures

Clinical assessment

Psychiatric symptoms were assessed using the Brief Psychiatric Rating Scale (BPRS) [24]. The BPRS is a clinician-rated score of schizophrenia symptoms, comprising 18 items across 5 domains. Each item is rated between 1 (symptom absent) and 7 (symptom extremely severe), yielding a total score between 18 and 126. The five domains are Affect (anxiety, guilt, depression, somatic); Positive Symptoms (thought content, conceptual disorganization, hallucinatory behavior, grandiosity); Negative Symptoms (blunted affect, emotional withdrawal, motor retardation); Resistance (hostility, uncooperativeness, suspiciousness); and Activation (excitement, tension, mannerisms–posturing) [25].

Depression and anxiety symptoms were evaluated using the Patient Health Questionnaire-9 (PHQ-9) and Generalized Anxiety Disorder Scale (GAD-7) self-report scales. The PHQ-9 has nine items with a total score from 0 to 36 and the GAD-7 has seven items with a total score from 0 to 21 [26, 27].

Cognitive function assessment

The Ascertain Dementia 8 (AD8) was used to assess patients’ cognitive and functional abilities [28]. It included eight items to assess memory, orientation, and judgment [29]. The total score ranges from 0 to 8, and the lower score means the better cognition function.

Side effects assessment

To assess the side effects of medication therapy, we used the Treatment Emergent Symptom Scale (TESS) (National Institute of Mental Health. TESS [30] and Rating Scale for Extrapyramidal Side Effects (RSESE). TESS is a 35-item scale that measures the severity and management of side effects symptoms on a 0–4 and 0–6 scale, respectively, with a total score of 0-350. RSESE is a 10-item that evaluates the extrapyramidal reaction on a 0–4 scale, with a total score of 0 to 40 [31]. RSESE was included to account for the complex medication regimens used in this study.

Patients were assessed at baseline and after rehabilitation training using the above-mentioned scales.

Before the study began, all data collectors and supervisors received training on data collection tools. The supervisors and the principal investigator reviewed the data for completeness and consistency regularly during data collection.

Statistical analysis

All study data were analyzed using SPSS software (version 25.0, SPSS Inc., Armonk, NY, USA). Descriptive data were used to summarize each factor across different levels of efficacy in QoL, as indicated by changes in the SQLS score. Logistic regression was performed to identify potential predictors of SQoL improvement after rehabilitation training, distinguishing between responders and non-responders. Standardized Beta (β) coefficients, and odds ratios (ORs) with a 95% confidence interval (CI) were computed to assess the strength of association and statistical significance, with P < 0.05 indicating statistical significance. A stepwise selection method was used to assess the baseline clinical symptoms, their changes, and baseline SQLS total and subdomain scores as predictors. This analysis aimed to determine whether sociodemographic information, clinical features, symptoms, and baseline SQoL could predict SQLS improvement post-rehabilitation Training. The Operating Characteristic (ROC) curve was used to assess the model’s accuracy. Bonferroni correction was used to adjust the significance of multiple comparisons, such as the comparisons in SQLS, BPRS and their subscales, and the significance level was set at 0.05/ n, (n=number of comparisions).

Results

A total of 400 SCZ patients were initially enrolled in the rehabilitation training. Of these, 105 were excluded for not meeting the inclusion criteria, leaving 295 patients who began the rehabilitation training. Among the 295 patients, 60 patients discontinued: 48 due to incomplete medication treatment or side effect data, and 12 due to missing cognitive or clinical symptom assessments. The final analysis comprised 235 patients who completed full rehabilitation training and assessments.

Of the final sample, 128 (54.5%) were male, with a mean age of 44.2 ± 9.1 years. A majority (N = 125, 53.2%) had an education level above high school, and 180 patients (76.2%) had an onse to age older than 18 years. Their clinical characteristic features are presented in Table 1.

Table 1.

Demographic and clinical characteristics of the participants

Variable		N	%
Sex	Male	128	54.5
	Female	107	45.5
Marital status	Married	108	46.0
	Unmarried	127	54.0
Education	Below high school	110	46.8
	Above high school	125	53.2
Duration of illness	< 10 years	65	27.7
	≥ 10 years	170	72.3
Age of onset	< 18 years	56	23.8
	≥ 18 years	180	76.2
Family history of mental illness	Yes	37	15.7
	No	198	84.3
Chronic physical disease	Yes	33	14.0
	No	202	86.0
Employ	Yes	15	6.4
	no	220	93.6
Live alone	Yes	25	10.6
	No	215	89.4
Activating antipsychotics	Yes	47	20.0
	No	188	80.0
Economic support	Yes	87	37.0
	No	148	63.0
		Mean ± SD	Min	Max
Age, years		44.2 ± 9.1	23	65
Baseline BPRS	Total	31.7 ± 11.2	18	92
	Anxiety/Depression	8.4 ± 3.7	4	20
	Negative Symptoms	7.7 ± 3.4	4	22
	Positive Symptoms	7.0 ± 3.3	4	22
	Activation	4.9 ± 2.4	3	16
	Resistance	5.3 ± 2.8	3	16
Baseline PHQ-9		4.9 ± 4.2	0	27
Baseline GAD-7		3.8 ± 3.3	0	21
Baseline AD		2.4 ± 2.1	0	8
Baseline TESS		11.4 ± 13.8	0	84
Baseline RSESE		1.6 ± 3.02	0	21
Baseline DDD		13.6 ± 6.1	2	36
Baseline SQLS	Total	31.7 ± 16.0	6	95
	PS	29.4 ± 16.2	2	88
	ME	46.5 ± 15.4	7	86
	SS	20.8 ± 14.8	3	72
Monthly household income		2875 ± 1685	0	8000

Economic support: Financial aids from the government or the disabled persons’ federation in China for mentally disabled patients; BPRS, Brief Psychiatric Rating Scale; PHQ-7, Patient Health Questionnaire-9; GAD-7, Generalized Anxiety Disorder Scale; AD8, The Ascertain Dementia; TESS, Treatment Emergent Symptom Scale; RSESE, Rating Scale for Extrapyramidal Side Effects; DDD: defined daily doses; SQLS: Quality of Life Scale, PS: Psychosocial, ME: Motivation/ Energy, SS: Symptoms/ Side effects

Comparison of demographic and clinical characteristics between responders and non-responders

After rehabilitation training, the mean change in SQLS score was 14.9 ± 15.7 (the reduction rate was 0.36 ± 0.39) compared to the score of baseline, indicating a modest, but significant improvement from baseline scores (t = 9.52, P = 0.002). Responders were determined as individuals exhibiting 20% of greater improvement (equivalent to half an SD) in their total SQLS score post-rehabilitation. Conversely, non-responders were defined by less than 20% improvement in SQLS total score after rehabilitation. The response rate for the total SQLS score was 72.77% (171/235), and it was 74.47% (175/235) for the PS subdomain, 61.7%(145/235) for the ME subdomain, and 82.98% (195/235) for the SS subdomain.

There were no significant differences in age, age of onset, duration of illness, education level, family history of mental illness, chronic physical disease, employment, and activating antipsychotics between responders and non-responders. However, there was a significant difference in marital status between the two groups, with more responders being married. There were also significant differences in the total SQLS scores (P = 0.013), and ME subdomain scores (P = 0.047), but not SS subdomain scores (p > 0.05) between the two groups.

There were significant differences in baseline PHQ-9, GAD-7, AD-8, and TESS scores (all P < 0.05) between the two groups. Significant differences after Bonferroni correction were also observed in baseline BPRS factors scores of anxiety/depression for the total SQLS score (P = 0.003) between the two groups, and negative symptoms (P = 0.005) for the ME subdomain. There were no significant differences in baseline RSESE scores, antipsychotic daily doses, and rate of activating antipsychotics between the two groups. See Table 2.

Table 2.

Comparison of demographic and clinical characteristics between responder and non-responder patients

Variable		SQLS(Mean ± SD)						PS(Mean ± SD)							ME(Mean ± SD)							SS(Mean ± SD)
		Responders	Non-responders		t	P	Responders		Non-responders			t	P	Responders		Non- responders			t	P	Responders		Non- responders			t	P
individuals, n		171	64					175		60					145		90					195		40
Age, year		42.9 ± 7.7	45.7 ± 10.9	0.82		0.976		44.7 ± 8.9		44.1 ± 10.4	0.30		0.453		44.4 ± 9.6		43.8 ± 7.8	-0.43		0.287		43.9 ± 9.0		45.4 ± 9.4	0.98		0.297
Age of one set, year		25.1 ± 7.8	23.7 ± 8.9	-0.82		0.976		25.2 ± 8.1		26.9 ± 13.9	-1.23		0.393		24.8 ± 7.8		25.2 ± 8.3	-0.57		0.428		24.8 ± 7.9		24.9 ± 8.0	1.04		0.168
Duration of illness, year		16.8 ± 8.7	20.1 ± 14.4	1.31		0.105		16.8 ± 8.8		20.1 ± 14.3	-1.79		0.145		17.5 ± 10.3		16.7 ± 8.0	0.33		0.726		16.9 ± 8.9		18.9 ± 12.8	0.09		0.683
Baseline SQLS
Total SQLS		37.4 ± 16.1	23.7 ± 2.5	-4.76		0.013		36.1 ± 16.8		29.9 ± 14.7	-1.94		0.123		36.2 ± 16.4		32.7 ± 14.2	1.41		0.195		36.9 ± 16.8		28.8 ± 14.2	-3.01		0.033
PS		20.2 ± 13.1	30.7 ± 16.2	-3.28		0.041		30.6 ± 16.2		20.9 ± 13.9	-3.43		0.014		29.4 ± 16.3		29.1 ± 16.2	-0.13		0.925		31.1 ± 16.3		21.8 ± 13.4	-3.83		0.023
ME		47.9 ± 14.9	38.5 ± 16.0	-3.41		0.047		46.7 ± 15.7		45.7 ± 15.7	-0.31		0.199		48.6 ± 15.7		40.5 ± 12.8	- 4.97		0.004		46.2 ± 14.9		48.1 ± 17.1	0.70		0.375
SS		21.5 ± 14.7	15.6 ± 14.0	-1.791		0.409		21.5 ± 14.9		16.9 ± 13.4	-1.56		0.330		20.1 ± 14.7		23.1 ± 14.9	1.16		0.214		21.9 ± 14.6		11.5 ± 12.6	-3.33		0.027
Baseline PHQ-9		5.9 ± 5.0	3.3 ± 4.7	-3.36		0.039		5.7 ± 4.9		5.0 ± 6.6	-0.63		0.476		5.9 ± 5.2		4.6 ± 4.2	-1.99		0.024		5.9 ± 5.0		4.5 ± 5.0	-1.66		0.425
Baseline GAD-7		4.8 ± 4.2	2.2 ± 3.0	-4.63		0.005		4.5 ± 4.1		4.1 ± 4.6	-0.56		0.482		4.8 ± 4.4		3.4 ± 3.0	-2.69		0.012		4.6 ± 4.0		4.1 ± 4.9	-0.57		0.953
Baseline AD8		1.26 ± 2.0	2.7 ± 2.2	-3.85		0.022		1.9 ± 2.3		2.5 ± 2.2	-1.42		0.134		2.7 ± 2.2		2.0 ± 3.0	-2.07		0.015		2.6 ± 2.2		2.2 ± 2.3	-0.86		0.256
Baseline TESS		6.4 ± 8.3	4.3 ± 8.2	-4.16		0.001		11.9 ± 14.3		6.9 ± 9.2	-2.94		0.012		14.0 ± 1.1		13.2 ± 1.7	-1.65		0.106		12.2 ± 14.5		6.8 ± 9.8	-2.99		0.036
Baseline RSESE		1.6 ± 2.9	1.6 ± 3.5	0.08		0.434		1.6 ± 2.9		1.4 ± 3.3	-0.32		0.561		1.6 ± 2.9		1.6 ± 3.3	0.13		0.972		1.7 ± 2.8		1.1 ± 3.6	-0.92		0.695
Baseline DDD		13.0 ± 6.5	12.6 ± 6.6	0.05		0.423		12.6 ± 6.6		13.0 ± 6.6	0.34		0.290		12.3 ± 6.8		13.6 ± 6.1	1.26		0.243		12.1 ± 6.5		14.3 ± 7.1	1.75		0.087
Baseline BPRS
Total BPRS		34.0 ± 12.7	29.9 ± 10.8	-2.01		0.046		33.8 ± 12.3		29.7 ± 13.2	-1.80		0.744		33.0 ± 12.1		34.5 ± 13.8	0.59		0.195		33.0 ± 11.6		27.8 ± 14.8	-3.43		0.009
Anxiety/Depression		8.8 ± 3.7	6.7 ± 3.3	-3.28		0.003		8.6 ± 3.8		7.7 ± 3.6	-1.27		0.446		8.6 ± 3.6		8.0 ± 4.1	-1.01		0.676		8.7 ± 3.7		7.2 ± 3.6	-2.46		0.026
Negative Symptoms		7.7 ± 3.2	7.3 ± 4.1	-0.63		0.408		7.6 ± 3.3		8.3 ± 3.9	1.03		0.617		7.5 ± 3.3		9.1 ± 3.6	4.07		0.005		7.7 ± 3.4		7.6 ± 3.4	-0.18		0.857
Positive Symptoms		7.0 ± 3.3	7.0 ± 3.5	-0.51		0.552		6.8 ± 3.2		7.8 ± 3.9	1.37		0.410		6.9 ± 3.2		7.1 ± 3.6	0.57		0.214		7.2 ± 3.3		6.6 ± 3.2	-1.04		0.173
Activation		4.9 ± 2.4	4.4 ± 2.4	-1.15		0.233		4.8 ± 2.4		5.2 ± 2.5	0.74		0.723		4.8 ± 2.2		5.1 ± 2.8	0.65		0.423		4.9 ± 2.5		4.4 ± 2.1	-1.62		0.144
Resistance		5.4 ± 2.7	4.8 ± 2.7	-1.20		0.287		5.3 ± 2.7		5.8 ± 3.2	0.88		0.471		5.3 ± 2.7		5.5 ± 3.0	0.50		0.769		5.5 ± 2.9		4.8 ± 2.2	-1.80		0.123
		n	n	χ2		P		n		n	χ2		P		n		n	χ2		P		n		n	χ2		P
Sex
Male		89	39	1.48		0.242		90		38	2.55		0.133		80		48	0.08		0.789		106		22	0.01		0.999
Female		82	25					85		22					65		42					89		18
Education
High school or below		86	24	3.06		0.106		83		27	0.11		0.765		72		38	1.23		0.284		96		14	2.70		0.121
Above high school		85	40					92		33					73		52					99		26
Marital status
Unmarried		84	43	6.12		0.018		90		37	1.89		0.873		71		56	3.93		0.059		103		24	0.69		0.487
Married		87	21					85		23					74		34					92		16
Family history of mental illness
Yes	30		7	1.53		0.314		32		5	3.34		0.099		20		17	1.09		0.357		35		2	4.19		0.054
No	141		57					143		55					125		73					160		38
Chronic physical disease
Yes	22		11	0.72		0.404		23		10	0.46		0.521		16		17	2.84		0.122		27		5	0.06		0.999
No	149		53					152		50					129		73					167		35
Employment
Yes	11		4	0.03		0.999		11		4	0.01		0.099		12		3	2.27		0.173		11		4	1.06		0.294
No	160		60					164		56					133		87					184		36
Live alone
Yes	12		8	1.80		0.194		14		6	0.23		0.601		13		7	0.10		0.814		17		3	0.06		0.999
No	159		56					161		54					132		83					178		37
On Activating antipsychotics
Yes	38		9	1.94		0.201		40		7	3.49		0.064		34		13	2.81		0.130		37		10	0.75		0.390
No	133		55					135		53					111		77					158		30
Economic support
Yes	71		16	4.45		0.023		73		14	6.47		0.013		52		35	0.22		0.678		71		16	0.18		0.720
No	100		48					102		46					93		55					124		24

The significance level after Holm-Bonferroni correction was set at 0.05/4 for SQLS, and 0.05/ 6 for BPRS

Abbreviation: SQLS: Quality of Life Scale; PS: Psychosocial; ME: Motivation/ Energy; SS: Symptoms/ Side effects; PHQ-7, Patient Health Questionnaire-9; GAD-7, Generalized Anxiety Disorder Scale; AD8, The Ascertain Dementia; TESS, Treatment Emergent Symptom Scale; RSESE, Rating Scale for Extrapyramidal Side Effects; DDD: defined daily doses; BPRS, Brief Psychiatric Rating Scale. Economic support: Financial aids from the government or the disabled persons’ federation in China for mentally disabled patients

Data are reported as Mean ± SD, unless indicated otherwise

Correlation Between Changes in Clinical Symptoms and SQLS Scores After Rehabilitation Training

Compared to baseline scores, there were significant improvement in clinical symptoms, cognition, and side effects at the end of training: PHQ-9 (5.58 ± 5.06 vs. 2.32 ± 3.43), GAD-7 (4.49 ± 5.06 vs.2.32 ± 3.43), BPRS (33.22 ± 12.46 vs.24.65 ± 9.32), Anxiety/Depression(8.44 ± 3.75 vs.6.02 ± 2.56), Negative Symptoms(7.68 ± 3.38 vs.5.72 ± 2.22), Positive Symptoms(6.95 ± 3.32 vs.5.31 ± 2.39), Activation(4.85 ± 2.39 vs.3.75 ± 1.65), Resistance (5.34 ± 2.76 vs.3.97 ± 1.93). AD8 (2.50 ± 2.27 vs.1.07 ± 1.69), TESS(11.14 ± 13.87 vs. 5.16 ± 7.75). All P < 0.001.

Significant correlations were found between changes in PHQ-9, GAD-7, TESS, and BPRS (total BPRS, anxiety, and depression factor, negative symptom factor) with changes in the SQLS total scores and all subdomain scores (all P < 0.05). Significant correlations were also found between changes in BPRS activation factor scores and changes in SQLS total scores and PS and SS (but not ME) subdomains (all P < 0.01). There was a significant correlation between the reduction in AD8 scores and the reduction of SQLS total, as well as the subdomains (PS, ME, and SS (All P < 0.05). See Fig. 1.

Fig. 1.

Correlation between the change of clinical symptoms(PHQ-9, GAD-7, AD8, TESS, RSESE, BPRS) and the change of SQLS after rehabilitation training compared with baseline. Abbreviations: PHQ-9, Patient Health Questionnaire-9; GAD-7, Generalized Anxiety Disorder Scale; AD8, The Ascertain Dementia; TESS, Treatment Emergent Symptom Scale; RSESE, Rating Scale for Extrapyramidal Side Effects. Results are Pearson’s correlation coefficient; BPRS, and Brief Psychiatric Rating Scale. *P < 0.05, **P < 0.01

Bivariate and multivariate regression analysis of factors and SQLS improvement

Bivariate regression showed significant associations between baseline scores of PHQ-9, GAD-7, AD8, TESS, and BPRS with changes in SQLS total and subdomain scores (P < 0.05). There was no significant association between the baseline score of medication doses and changes in SQLS total or subdomain scores. The baseline RSESE score was significantly associated with changes in the SS subdomain score (P < 0.001). Significant correlations were found between changes in SQLS total scores and changes of the scores of PHQ-9, AD8, and BPRS positive symptoms factor scores (all P < 0.05). Correlations were found between changes in PS subdomains scores and GAD-7 scores and negative symptoms scores and positive symptoms scores in BPRS (all P < 0.05). There was a correlation between the reduction in scores of PHQ-9, AD8, and BPRS anxiety/depression symptoms factor scores with the reduction of ME subdomains scores (All P < 0.05). These results are summarized in Table 3.

Table 3.

Bivariate regressions analysis of SQLS-response to the symptom score change after 24-week rehabilitation training

Variable	SQLS			PS			ME			SS
	B(95%CI)	P		B(95%CI)	P		B(95%CI)	P		B(95%CI)	P
Baseline PHQ-9	0.07(-0.65-0.79)	0.848		0.96(-0.01-2.0)	0.074		-0.60(-1.80-0.60)	0.327		0.78(0.04–1.51)	0.040
Baseline GAD-7	-0.12(-1.08-0.83)	0.804		0.36(-0.46-1.18)	0.391		0.88(-0.73-2.50)	0.282		-0.15(-1.15-0.84)	0.762
Baseline AD8	-0.25(-1.27-0.76)	0.628		1.4(1.34–1.71)	< 0.001		-0.16(-1.85-1.54)	0.857		0.50(-0.55-1.54)	0.352
Baseline TESS	-0.08(-1.06-0.16)	0.229		-0.09(-0.24-0.05)	0.188		-0.09(-0.30-1.20)	0.403		-0.13(-0.34-0.09)	0.237
Baseline BPRS
Anxiety/Depression	0.43(-0.88-1.84)	0.365		0.96(-0.09-1.18)	0.139		-1.08(-2.62-0.47)	0.172		0.60(-0.35-1.55)	0.214
Negative Symptoms	0.38(-0.61-1.37)	0.447		0.24(-0.93-1.33)	0.772		1.37(-0.61-3.34)	0.174		0.72(-0.30-1.74)	0.166
Positive Symptoms	-1.19(-2.88-0.28)	0.112		-0.91(-2.59-0.76)	0.284		-1.25(-3.70-1.20)	0.315		-1.40(-2.90-0.18)	0.069
Activation	-1.63(-3.27-0.11)	0.052		-0.99(-1.13-1.28)	0.246		-0.01(-2.75-2.73)	0.994		2.02(0.34–3.70)	0.019
Resistance	0.96(-0.33-2.23)	0.141		0.94(-0.56-2.44)	0.220		0.86(-1.40-3.05)	0.443		1.05(-0.29-2.39)	0.126
Improvement PHQ-9	1.07(0.33–1.81)	0.005		0.36(-0.56-2.44)	0.391		1.19(0.03–2.04)	0.016		0.52(-0.27-1.29)	0.182
Improvement GAD-7	0.72(-0.22-1.65)	0.131		1.23(0.159–2.31)	0.025		-0.25(-1.82-1.32)	0.757		-0.04(-1.01-0.95)	0.989
Improvement AD8	1.82(0.69–2.95)	0.002		1.05(-0.24-2.34)	0.110		2.71(0.82–4.60)	0.005		0.07(-1.09-1.23)	0.907
Improvement TESS	-0.78(-0.20-0.05)	0.121		-0.10(-0.24-0.05)	0.188		-0.09(-0.29-0.12)	0.403		1.19(-0.02-0.394)	0.078
Improvement BPRS
Anxiety/Depression	0.60(-0.38-1.58)	0.228		0.28(-0.84-1.39)	0.627		1.67(0.03–3.29)	0.046		-0.35(-1.35-0.66)	0.498
Negative Symptoms	0.30(-0.95-1.40)	0.702		2.21(0.01–2.44)	0.049		1.37(-0.61-3.34)	0.174		-0.60(-1.82-0.62)	0.334
Positive Symptoms	1.84(0.35–3.34)	0.018		1.72(0.35–3.09)	0.014		1.20(-1.31-3.70)	0.347		2.507(0.97–4.04)	0.001
Activation	1.48(-0.44-3.41)	0.129		0.18(-1.17-1.54)	0.799		-0.62(-3.83-2.59)	0.703		1.82(-0.16-3.80)	0.071
Resistance	-1.68(-2.88-0.47)	0.067		-0.91(-2.59-0.76)	0.284		-1.52(-3.52-0.49)	0.137		-1.68(-2.92-0.448)	0.094

PHQ-9, Patient Health Questionnaire-9; GAD-7, Generalized Anxiety Disorder Scale; AD8, The Ascertain Dementia; TESS, Treatment Emergent Symptom Scale; RSESE, Rating Scale for Extrapyramidal Side Effects. DDD: defined daily doses; BPRS, Brief Psychiatric Rating Scale; SQLS: Quality of Life Scale, PS: Psychosocial, ME: Motivation/ Energy, SS: Symptoms/ Side effects

Multivariate linear regression analysis further examined the effects of multiple factors on SQLS and its subdomains, including socio-demographic variables, baseline clinical symptoms, cognition, and side effects. The overall model fit for the total SQLS score was 157.1(R² =0.23, p < 0.01). The individual model fit was 27.5 for PS (R² =0.17, p < 0.01), 25.1 for ME (R² =0.16, p = 0.001), and 31.4 for SS (R² =0.22, p < 0.01). The total SQLS score change was negatively associated with baseline BPRS negative symptom factor score (P = 0.017), baseline score of PHQ-9 (P = 0.034), while positively associated with changes in BPRS negative symptom factor score (P = 0.031), PHQ-9 scores (P = 0.034), as well as baseline PS subdomain score (P = 0.021) (See Fig. 2a). The change in PS subdomain score was positively associated with baseline PS subdomain score (P = 0.002) and changes in GAD-7 score (P = 0.002) (See Fig. 2b). Changes in the ME subdomain were negatively associated with baseline BPRS negative symptom factor score (P = 0.008), while positively associated with changes in BPRS negative symptom factor score (P = 0.010), as well as baseline ME subdomain score (P = 0.013) (See Fig. 2c). The change in SS subdomain score was negatively associated with baseline BPRS positive symptom factor score (P = 0.038) and activation factor score (P = 0.002), and GAD-7 score (P = 0.012), while positively associated with changes in PHQ-9 score (P = 0.023), GAD-7 score (P = 0.028), AD-8 score (P = 0.048), baseline SS subdomain score (P = 0.004) and economic support(P = 0.034) (See Fig. 2d).

Fig. 2.

Multivariate regression analysis of factors(baseline symptom and symptom change after 24-week rehabilitation training)associated with SQLS change and its domains in patients with schizophrenia.(a)Effect of factors on the total SQLS change. (b) Effect of factors on the domain of PS.(c) Effect of factors on the domain of ME. (d) Effect of factors on the domain of SS. Abbreviations: PS: Psychosocial, ME: Motivation/ Energy, SS: Symptoms/ Side effects. Results are Unstandardized Beta (β) coefficients with a 95% confidence interval (CI)

Receiver operating characteristic curve (ROC) analysis was used to test the model’s accuracy. The improvement in SQLS total and its subdomains after rehabilitation training was considered the positive status variable. Baseline levels of total SQLS and its subdomains, education level, baseline BPRS negative symptom factor score, baseline PHQ-9, GAD-7, AD8, and TESS scores were used as test variables. The results showed that the values of area under the curve (AUC) for SQLS total score, PS, ME, and SS subdomain scores were 0.798, 0.802, 0.729, and 0.689, respectively. See Fig. 3a.

Fig. 3.

a The improvement of SQLS and its domains after rehabilitation training was taken as the positive status variable, baseline SQLS and its domains (PS, ME, and SS), baseline BPRS(Anxiety/Depression, Negative Symptoms and Positive Symptoms), baseline PHQ-9, baseline GAD-7, baseline AD8, and baseline TESS were used as test variables for receiver operating characteristic curve (ROC). b The improvement of SQLS and its domains after rehabilitation training was taken as the positive status variable, baseline SQLS and its domains(PS, ME, and SS), baseline BPRS (Anxiety/Depression, Negative Symptoms and Positive Symptoms), baseline PHQ-9, baseline GAD-7, baseline AD8, baseline TESS, and change of symptoms after 24-week rehabilitation training (BPRS: Anxiety/Depression, Negative Symptoms and Activation, PHQ-9, GAD-7, AD8 and TESS) were used as test variables for receiver operating characteristic curve (ROC). Abbreviations: SQLS: Quality of Life Scale; PS: Psychosocial, ME: Motivation/ Energy, SS: Symptoms/ Side effects; PHQ-9, Patient Health Questionnaire-9; GAD-7, Generalized Anxiety Disorder Scale; AD8, The Ascertain Dementia; TESS, Treatment Emergent Symptom Scale; RSESE, Rating Scale for Extrapyramidal Side Effects

ROC was also performed to explore the changes in clinical symptoms (change in symptoms after 24-week rehabilitation training in BPRS factor subscore of negative symptoms, positive symptoms, and activation, PHQ-9, GAD-7, AD8, and TESS), along with aforementioned variables, as predictors of improvement in SQLS after rehabilitation training. The result showed that the AUC values of SQLS total score, PS, ME, and SS subdomain scores were improved to 0.926, 0.932, 0.812, and 0.885, respectively. See Fig. 3b.

Discussion

The study presents a novel prospective interventional approach to understanding the interrelationships between demographical factors, clinical symptoms, and cognitive function (considering both baseline levels and changes) in predicting improvement of SQoL following rehabilitation training in hospital-based SCZ patients. Our findings highlight several key points: (1) Overall high response rate: The response rate to rehabilitation training, in terms of improved SQoL was 72.77% for SQLS total score. The subdomain response rates were 74.47% for PS, 61.7% for ME, and 82.98% for SS (82.98%). (2) Predictive effects of baseline SQLS: Baseline SQLS total and subdomain scores significantly predicted improvement in SQoL total score and each specific subdomain. (3) Impact of clinical symptoms and cognition: clinical symptoms and cognition, both baseline levels and changes, were significantly associated with SQoL improvement. (4) Predictive models: The models for the SQLS total score and its subdomain demonstrated clinical validity for early prediction of outcomes, highlighting the potential for targeted interventions to improve SQoL effectively.

Impact of Rehabilitation Training on SQoL and Effects of Baseline SQLS on SQoL Improvement

Our study supports the effectiveness of rehabilitation training, consistent with previous studies [4–6]. Our rehabilitation involved management skills training, symptom management skills training, and social skills training, which proved to be beneficial for psychosocial functioning and managing symptom/side effects, while less effective for motivation and energy levels. The training, based on social learning theory and cognitive science, empowered patients to better manage their illness and make informed treatment decisions. However, the different response rates across SQoL subdomains, particularly the modest improvement for ME, suggest that rehabilitation may need to be tailored to address specific QoL dimensions. This is in line with previous findings [32] that indicated that combining medication with recreational therapy may yield better outcomes for ME and SS, while PS showed notable effectiveness when the rehabilitation training was only used. Therefore, PS and SS may be particularly beneficial following rehabilitation, whereas ME’s effectiveness was relatively lower. Additionally, we also found a positive correlation between baseline QoL and the post-training improvement of SQoL, which is consistent with the findings reported by Nevarez-Flores et al. [33], who found that global functioning was associated with QoL in people with psychotic disorders. Therefore, it is crucial to assess overall QoL and individual subdomains before rehabilitation training to implement targeted training for the most impacted areas in quality of life.

Importance of Clinical Symptoms’ Impact on SQoL

The improvement in SQoL observed in our study was associated with baseline clinical symptoms and changes after rehabilitation training. For example, lower baseline negative symptom scores and improvement in depression (PHQ-9) and negative symptoms were associated with improvement in SQoL, consistent with previous reports [22, 33]. Additionally, lower baseline scores for negative symptom scores and anxiety (GAD-7) were associated with improvement in the PS subdomain. Similarly, lower baseline anxiety and depression scores, as well as greater improvement in anxiety and depression, were associated with improvement in ME subdomains. Higher baseline depression scores (PHQ-9) were associated with improvement in the SS subdomain. These findings emphasize that clinical symptoms, particularly anxiety, depression, negative symptoms, and activation, play a crucial role in determining the efficacy of rehabilitation. This aligns with previous findings [32, 35–37], suggesting that depressive and positive symptoms may be important factors influencing SQoL in patients with SCZ. Notably, changes in BPRS negative symptom scores were negatively associated with the improvement of total SQLS and the PS subdomain, changes in depression and anxiety scores, and anxiety and depression symptom scores were negatively associated with improvement in ME subdomains. This suggests that the persistence or worsening of negative symptoms (e.g. diminished expression, amotivation) may hinder the benefits of rehabilitation, aligning with previous studies that linked severe negative symptoms to poor subjective well-being [15, 38, 39]. It is noteworthy that some other studies showed a negative correlation between the duration of untreated psychosis and the severity of symptoms with QoL in people experiencing first-episode psychosis [40]. While psychiatric or psychotic symptoms are common predictors of QoL, the specific symptoms that best predict QoL remain inconclusive [25, 33, 41, 42, 38]. Overall, it is challenging for most people to maintain high SQoL when concurrently experiencing significant symptoms, underscoring the need for effective treatment of these clinical symptoms to improve the SQoL of patients with schizophrenia.

Our study found that there was a significant association between baseline TESS score with the improvement of SQLS PS subdomain. Previous studies also suggested that some medication side effects, such as weight gain and sexual dysfunction, have been associated with a worse outcome [43]. Medication side effects may be one of the targets in future rehabilitation training, especially in patients with lower PS subdomain levels.

Importance of cognitive function on SQoL and Its improvement

Our study found that better cognitive function (as measured by AD8) at baseline and improvements after rehabilitation training were negatively associated with the improvement of SQLS. This may seem counter-intuitive, but it reflects the complexity of the relationship between cognitive function and QoL. This is reflected in previous reports. For example, Prouteau et al. [18] found that a better baseline associative visual memory predicted better community functioning over the rehabilitation follow-up period. Several other previous studies have shown that cognition may contribute to QoL [13, 44, 45], especially information processing speed, memory, and executive function [17, 46, 47]. However, other studies reported negative associations [41, 42, 38] and some reported no associations [41, 39]. This discrepancy could be partly explained by the fact that most of the studies were cross-sectional the associations were with QoL, not QoL improvement. Further analysis of our data revealed positive correlations between the reduction in AD8 scores and the decrease in SQLS total scores and all subdomain scores, indicating that the improvement in cognitive function contributes to better subjective QoL. Additionally, our study used AD8 as a self-report cognitive assessment tool, which reflects memory, computation, and social cognitive functions. Compared with neurocognition, social cognition has a greater impact on patients [48], which further impacts the quality of life [46]. In conclusion, cognitive functions played important roles in psychosocial functioning improvement and should be noted during a rehabilitation program.

Negative findings

A few negative findings in our study are also worth noting. We did not find significant associations between SQoL improvement and demographic factors or medication side effects. One possible explanation is due to the sample and training. Patients in this study were relatively stable with strong support, which ensured their participation in the 24-week training with few dropouts. Moreover, we did not observe significant associations between the type and dosages of medications and the improvement of SQoL, contrary to the evidence in many previous studies and common expectations. One possible explanation is that we examined the effects of different factors on the improvement in QoL, not QoL per se. Another possible explanation is that the relationship between medications and QoL is complex, while antipsychotics can improve symptoms (especially positive symptoms), leading to improvement in QoL, the side effects may also be associated with poor QoL [49]. Our study suggests that the relationships between the clinical variables and improvement of SQoL are quite different from cross-sectional relationships [50], highlighting the need for further longitudinal studies to clarify these associations.

Strengths and Limitations

Several strengths of our study need to be em[phasized. First, our study is a prospective interventional study with a large sample size (> 200), which allowed us to identify predictive factors for the improvement in QoL. Second, patients in our study underwent weekly rehabilitation training for 24 weeks, and the dropout rate (20.33%) was fairly low. Third, we assessed symptoms and cognition at baseline and after rehabilitation training (week 24), allowing us to examine the relationship between changes in clinical symptoms and cognition with the improvement of SQoL.

However, several limitations should also be mentioned in this study. First, to minimize the number of statistical comparisons, we did not evaluate some clinical factors, such as the number of hospitalizations [51] and insight [52], which are associated with QoL in previous studies. It should be noted that although the dropout rate (20.33%) was fairly low, those who dropped out might have lower QoL at baseline or showed no improvement with the program, a large sample is be needed in future studies. Second, the assessment of cognition is less comprehensive. A previous study [18] suggested that the specific domains of cognitive deficits in schizophrenia reflect a pattern of correlative relationships and that in the regression analysis, verbal memory remained a predictor of improvements in subjective life satisfaction. Third, the samples mainly come from psychiatric hospitals in specific regions, which may affect the generalizability of the results to a wider population or different healthcare settings. Fourth, we did not include a control group that did not receive rehabilitation training, and thus it remains unclear to what degree these predictor variables are specific to this integrated behavioral intervention. Fifth, incorporating additional assessments of subjective and objective functioning would provide a more comprehensive evaluation. In a previous study [19], we employed the Personal and Social Performance Scale (PSP) to assess objective social functioning and found that cognitive training with interventions and pharmacological treatments may offer synergistic benefits for long-term objective functional improvements in schizophrenia. It is valuable to integrate clinician-rated instruments alongside self-reported QoL measures in order to capture a broader spectrum of patient outcomes in future study. Additionally, it is conceivable that a portion of the observed improvement may be ascribed to the effects of medications, considering all participants were consistently maintained on antipsychotic medication regimens. Regarding the abundant clinical resources in China, a larger sample, homogenous participants such as the first episode psychosis will be enrolled in future studies, and the results would be more valuable.

Conclusion

In summary, the findings indicate structured rehabilitation training substantially enhances SQoL, and the improvements in SQoL are associated with clinical symptoms, cognition, and baseline SQoL. Specifically, both baseline and changes of specific psychiatric symptoms, anxiety/depression, negative symptoms, positive symptoms, along with cognitionstatus and psychosocial subdomain scores, emerged as significant predictors of SQoL outcomes.These results underscore the importance of individualized treatment by targeting clinical symptoms and cognitive function in rehabilitation programs to enhance SQoL in SCZ patients, suggesting “One size does not at all” approach is inadequate. Comprehensive assessment of specific attributes of each individual, including their clinical symptoms and cognitive function, are essential for optimizing rehabilitation outcomes. By targeting these areas, psychosocial rehabilitation interventions can be become evidence-based practices that form a major part of the standard treatment of schizophrenia, ultimately improving the patients’ quality of life [53].

Abbreviations

SCZ

Schizophrenia

QoL

Quality of life

SQLS

Quality of Life Scale

PS

Psychosocial

ME

Motivation/ Energy

SS

Symptoms/ Side effects

FGAs

First-generation antipsychotics

SGAs

Second-generation antipsychotics

DDDs

Defined daily doses

BPRS

Brief Psychiatric Rating Scale

DSM-IV

Diagnostic and statistical manual for mental disorders. Fourth Edition

PHQ-9

Patient Health Questionnaire-9

GAD-7

Generalized anxiety disorder scale

AD8

Ascertain dementia 8

TESS

Treatment emergent symptom scale

RSESE

Rating Scale for Extrapyramidal Side Effects

SD

Standard deviation

CI

Confidence interval

n (%)

Composition ratio

OR

odds ratios

ROC

The Operating Characteristic curve

Author contributions

XW and XY designed the study and wrote the first draft of the manuscript. WJ and D-nZ administered neuropsychological and clinical measures and assisted in writing. W-qJ and WW helped design the study and recruited participants. Y-pR and C-lY designed the study and assisted in writing and editing the manuscript. Y-lT assisted in data interpretation and critical editing of the manuscript. All authors contributed to the article and approved the submitted version.

Funding

The research was supported by Beijing Municipal Science & Technology Commission (No.Z191100008319009)to Y-pR).

Data availability

Data are available from the corresponding authors.

Declarations

Ethics statement

The study protocol was reviewed and approved by the Ethics Committee of Beijing Anding Hospital, Capital Medical University. Written informed consent to participate in this study was provided by participants or their legal guardians. The study was registered (Chictr.org.cn: ChiCTR2100043537).Dr. Yi-lang Tang received research funding from the U.S. Department of Veterans Administration for a medication clinical trial for alcohol use disorder.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.