Aim: To investigate the factor structure underlying the Camberwell Assessment of Need–Patient Version (CANSAS-P) items in schizophrenia and schizoaffective disorder.
Method: Factor, correlation and regression analyses were performed for dimensions of CANSAS-P, illness, personality and quality of life (QOL) related variables in 95 stabilized patients with chronic schizophrenia and schizoaffective disorder.
Results: Exploratory factor analysis revealed a four-factor model that explains 50.4% of the total variance of the 20 CANSAS-P items. The factors ‘Social disability’, ‘Information processing disability’, ‘Emotional processing disability’, and ‘Coping disability’ showed acceptable internal consistency (Cronbach's α coefficient 0.67–0.77). The CANSAS-P subscale scores positively correlated with severity of symptoms, distress (r ranged from 0.34 to 0.45), while negatively associated with general functioning (r = −0.34), friend (r = −0.46) and family support (r = −0.41), satisfaction with medicine (r = −0.35), general activities (r = −0.40), and general QOL (r = −0.35) (all P < 0.001). Severity of illness, symptoms, emotional distress and emotion-oriented coping were positive predictors; friend support, QOL general activities, life satisfaction and satisfaction with medicine were negative predictors of the CANSAS-P subscale scores. The effect size (f2) for these predictors ranged from medium to quite large (f2 = 0.28–1.13), and they explain from 23% to 46% of the variability in CANSAS-P subscales.
Conclusions: A four-factor structure mode, including social and cognitive functioning, emotion responsivity and coping with daily challenges, appears to fit CANSAS-P items. These subscales may contribute to research and improve treatment of psychiatric patients.
THE CAMBERWELL ASSESSMENT OF NEED (CAN) and the Camberwell Assessment of Need Short Appraisal Schedule, Patient Version (CANSAS-P), a patient-rated scale, are the most widely used tools for the assessment of needs for people with severe mental health problems.1–5
The most frequently detected needs involve psychotic symptoms, psychological distress, house upkeep, food, information on condition and treatment, needs involving company and daytime activities.6–11 Individual unmet needs appear to be quite sensitive to change over time.12 Two previous studies found different factor structures for the CANSAS. For instance, a principle component analysis by Slade et al.13 yielded seven rather vague components with several cross-loadings, of which three were difficult to interpret. Wennström et al.14 investigated the factor structure on a sample of 741 outpatients with severe mental illness (only 68% schizophrenia or other psychotic disorders) and found three main factors, comprising only 13 of the 22 items in the CAN. Thus, previous studies of the CANSAS produced inconsistent results.
The present study aimed to detect the association of illness, personality, and quality of life variables with any significant latent factors underlying the CANSAS-P items in schizophrenia and schizoaffective disorder.
The study sample was drawn from a database of patients participating in the 10-year follow-up stage of an ongoing naturalistic prospective investigation of quality of life of patients with major psychiatric disorders (schizophrenia, and schizoaffective disorder) that was initiated in 1998. A detailed description of the design, data collection, measures and findings was reported elsewhere.15–18 Briefly, the initial sample was systematically selected from the hospital case register according to the following inclusion criteria: (i) fulfillment of DSM-IV criteria19 for schizophrenia and schizoaffective disorder; (ii) age 18–65 years; and (iii) inpatient status in closed, open or rehabilitation hospital departments of a university hospital. Patients with mental retardation, organic brain disease, severe physical disorders, drug/alcohol abuse, and those with low comprehension skills were not enrolled. Patients that met the inclusion criteria were assessed three times: prior to discharge from hospital (initial assessment), and about 2 and 10 years later. The Sha'ar Menashe Internal Review Board and the Israel Ministry of Health approved the study. All participants provided written informed consent for participation in the study, after receiving a comprehensive explanation of study procedures.
Of 108 outpatients who were assessed at the 10-year follow-up examination, 95 (88%) of the patients completed all 22 items of the CANSAS-P, while 13 patients (12%) left one or more items of the CANSAS-P unrated. Thus, the present sample included 95 subjects, 73 (76.8%) men, mean age 48.2 ± 9.1 years (range: 30–69); 56 individuals (58.9%) were single, 20 (21.1%) were married, and the remaining 19 (20.0%) were divorced, separated or widowed. Mean extent of education was 10.4 ± 2.3 years. Mean age of application for psychiatric care was 28.8 ± 7.7 years, and mean duration of disorder was 25.1 ± 9.2 years (range: 11–49). Among 95 patients in the sample, 55 (57.9%) presented with DSM-IV paranoid type schizophrenia, 18 (18.9%) with residual type, one (1.1%) with disorganized type, two (2.1%) with undifferentiated type schizophrenia, and 19 (20.0%) with schizoaffective disorder. Patients were treated with first-generation antipsychotic agents (FGA, 51 patients), second-generation antipsychotics (SGA, 21 patients) and with a combination of FGA and SGA (23 patients).
Needs were assessed using the CANSAS-P.4,5,13,20 The CANSAS-P assesses needs over the past month in 22 health and social items. The need rating for each item is 0, ‘no need’; 1, ‘met need’; or 2, ‘unmet need’.
Severity of illness and psychopathology were assessed using the Clinical Global Impression Scale (CGI-S),21 and the Positive and Negative Syndromes Scale (PANSS).22 The presence and severity of adverse effects of medication as well as psychological responses to them were measured with the Distress Scale for Adverse Symptoms (DSAS).15,23,24 The overall level of functioning was evaluated with the Global Assessment of Functioning Scale (GAF).19
Assessment of emotional distress and somatization was done using the Talbieh Brief Distress Inventory (TBDI).25,26 Responses are 0 to 4 with higher scores indicating greater intensity of six distress symptoms: obsessiveness, hostility, sensitivity, depression, anxiety, and paranoid ideation. The Somatization Scale is derived from the Brief Symptom Inventory–Somatization Scale (BSI-S).27 The Coping Inventory for Stressful Situations (CISS)28 assesses ways in which people react to various difficulties, or stressful situations. Responses are scored on a five-point scale ranging from ‘not at all’ to ‘very much’. Three basic coping styles are evaluated: task-oriented, emotion-oriented and avoidance-oriented coping. The Multidimensional Scale of Perceived Social Support (MSPSS)29 was used as a measure of social support (support from family, friends, and others). Quality of Life Enjoyment and Life Satisfaction Questionnaire (Q-LES-Q)30 was used to assess subjective quality of life, that is, general (Q-LES-Qindex) and the following domains: physical health, subjective feelings, leisure time activities, social relationships, general activities, medication satisfaction, and life satisfaction. Responses are scored on a 1–5-point scale (1 = ‘not at all or never’ to 5 = ‘frequently or all the time’), with higher scores indicating better enjoyment and satisfaction with specific life domains.
The principle axis method of factor analysis was applied to identify the factors structure underlying the CANSAS-P items. Variables with an absolute loading greater than the amount set in the minimum loading option (≥0.4) were selected.
Pearson correlation coefficients were evaluated between the mean score of items involved in each CANSAS-P main factor (subscale) and 25 dimensions of illness, personality and quality of life: CGI-S, PANSS, DSAS, GAF, TBDI, BSI-S, CISS, MSPSS, and Q-LES-Q (correlations with Q-LES-Q domains were adjusted for CGI-S, PANSS, and DSAS total scores). Pearson correlation coefficients are presented after the Bonferroni correction for the 25 dimensions (P = 0.05/25 = 0.002).
Finally, a multiple regression analysis with step-wise backward selection was performed with a set of illness, personality, and QOL independent variables for prediction of the CANSAS-P subscale scores (the dependent variable). The effect size for a multiple regression study (Cohen's f2) was calculated using a value of R2.31 Cohen32 suggests at three different levels of the effect size for the multiple regression model: small, medium and large effects (0.1, 0.25 and 0.4, respectively). Mean values with standard deviation (SD) are presented. For all analyses, the level of statistical significance was defined as P < 0.05 except correlation coefficients (P < 0.002). Statistical analysis was performed using the Number Cruncher Statistical Systems.33
Responses on the CANSAS-P items
The most frequently detected unmet needs involved psychological distress (33.7%), sexual expression (33.7%), intimate relationships (31.9%), company (31.6%), physical health (26.3%), psychotic symptoms (25.3%), daytime activities (24.2%), information on condition and treatment (21.1%), accommodations (21.1%), and money (20.0%). The mean number of needs identified per patient was 7.29 ± 4.8 (range 0–20).
An exploratory factor analysis yielded a four-factor solution with virtual overlaps of the factor loading scores for three items (self-care, daytime activities, and psychotic symptoms), whereas two items (‘alcohol’ and ‘money’) did not reach the minimum loading option (≥0.4) (Table 1). The four main factors were identified on the highest Eigenvalues. The first factor included negative loadings of accommodation, food, looking after the home, physical health, information on condition and treatment scores. The second factor was constructed using basic education, telephone, transport, and welfare benefits scores with negative loadings. The third factor included daytime activities, psychotic symptoms, psychological distress, company, intimate relationships, and sexual expression factor scores (all positive loadings). The fourth factor was constructed using self-care, safety to self, safety to others, drugs, and child care factor scores (all negative loadings). These four main factors were labeled ‘Social disability’, ‘Information processing disability’, ‘Emotional processing disability’, and ‘Coping disability’. Correspondingly, they accounted for 14.9%, 11.4%, 12.7%, and 11.4% of the total variance of the 20 items. The sum of the scale scores of the four CANSAS-P factors called subscales showed high internal consistency: Cronbach's α coefficient ranged from 0.67 to 0.77. A small correlation was found between these subscales (r range from 0.25, P = 0.014 to 0.38, P < 0.001).
Table 1. Factor loadings and communalities after varimax rotation of variable values CANSAS items among 95 patients with schizophrenia and schizoaffective disorder
| 1. Accommodation|| −0.78354 ||0.6139||−0.2318||0.0537||−0.1139||0.0129||0.0450||0.0020|
| 2. Food|| −0.76246 ||0.5812||−0.0224||0.0005||0.1940||0.0376||−0.0569||0.0032|
| 3. Looking after the home|| −0.57697 ||0.3328||0.0684||0.0046||0.3186||0.1015||−0.2351||0.0552|
| 4. Self-care||−0.44096||0.1943||−0.1272||0.0161||0.2021||0.0408|| −0.4796 ||0.2300|
| 5. Daytime activities||−0.45261||0.2048||0.1892||0.0358|| 0.5610 ||0.3146||−0.2396||0.0574|
| 6. Physical health|| −0.5333 ||0.2844||−0.1268||0.0160||0.1660||0.0275||−0.0352||0.0012|
| 7. Psychotic symptoms||−0.3330||0.1108||−0.0868||0.0075|| 0.4326 ||0.1871||−0.4005||0.1604|
| 8. Information on condition and treatment|| −0.6196 ||0.3839||−0.0876||0.0076||0.0190||0.0003||−0.0870||0.0075|
| 9. Psychological distress||−0.1766||0.0312||−0.0239||0.0005|| 0.7436 ||0.5530||−0.1442||0.0208|
| 10. Safety to self ||−0.0777||0.0060||−0.1735||0.0301||0.0871||0.0075|| −0.7127 ||0.5080|
| 11. Safety to others ||−0.2031||0.0412||−0.0913||0.0083||0.0492||0.0024|| −0.5802 ||0.3366|
| 12. Alcohol ||−0.0960||0.0092||0.1720||0.0295||0.2758||0.0760||−0.3461||0.1198|
| 13. Drugs ||0.0210||0.0004||−0.0053||0.0001||0.0525||0.0027|| −0.5034 ||0.2533|
| 14. Company ||−0.3327||0.1107||−0.0422||0.0017|| 0.6954 ||0.4836||−0.0093||0.0001|
| 15. Intimate relationships ||0.1894||0.0359||−0.4337||0.1881|| 0.6412 ||0.4112||0.0496||0.0024|
| 16. Sexual expression ||0.1585||0.0251||−0.5204||0.2708|| 0.6108 ||0.3730||0.07238||0.0052|
| 17. Child care ||0.0844||0.0071||−0.3347||0.1120||−0.0908||0.0082|| −0.7086 ||0.5022|
| 18. Basic education ||−0.0523||0.0027|| −0.6458 ||0.4171||0.0829||0.0068||0.0506||0.0025|
| 19. Telephone ||−0.1869||0.0349|| −0.7669 ||0.5882||−0.1066||0.0113||−0.2584||0.0668|
| 20. Transport ||−0.2381||0.0567|| −0.6818 ||0.4649||−0.0204||0.0004||−0.3654||0.1335|
| 21. Money ||−0.1626||0.0264||−0.2978||0.0886||0.3306||0.1093||−0.1523||0.0231|
| 22. Welfare benefits ||−0.0652||0.0042|| −0.4036 ||0.1629||0.1668||0.0278||−0.0962||0.0092|
| Factors' contribution (%) ||14.9||–||11.4||–||12.7||–||11.4||–|
Correlation with illness, personality and QOL related variables
Mean scores with SD, minimum and maximum values of illness, personality and QOL related variables are presented in Table 2. All of the following correlations were significant with P-value < 0.001. ‘Social disability’ sub-scale scores positively correlated with PANSS total score (r = 0.37), and with three distress symptoms: sensitivity (r = 0.34), depression (r = 0.38), and somatization (r = 0.36). ‘Coping disability’ sub-scale also shows only positive association with TBDI hostility (r = 0.54), sensitivity, depression, and anxiety (all r = 0.40). ‘Information processing disability’ sub-scale positively correlated with four distress symptoms, such as sensitivity (r = 0.43), depression (r = 0.44), anxiety (r = 0.43), and somatization (r = 0.45), while negatively associated with family support (r = −0.41), Q-LES-Q general activities (r = −0.43) and satisfaction with medicine (r = −0.35). ‘Emotional processing disability’ scores positively correlated with anxiety (r = 0.39), and somatization (r = 0.38), but negatively associated with general functioning (r = −0.34), friend support (r = −0.46), general quality of life (Q-LES-Q index; r = −0.35), satisfaction with general activities (r = −0.40) and with medicine (r = −0.38). Other illness, personality, and QOL dimensions, education, age and age at onset, and illness duration were not associated with these sub-scales (all P-values > 0.002).
Table 2. Mean scores of illness, personality, and quality of life related variables of 95 patients with schizophrenia and schizoaffective disorder
|Illness severity (CGI-S)||4.1||1.1||2||7|
|Emotional distress (TBDI):|| || || || |
| Paranoid ideation||1.4||1.0||0||4|
|Coping styles: Task-oriented coping||55.1||17.1||4||80|
| (CISS) Emotion-oriented coping||42.0||13.7||11||72|
| Avoidance oriented coping||48.2||14.3||6||79|
|Social support (MSPSS):|| || || || |
| Family support||20.1||6.7||4||28|
| Friend support||15.2||8.2||3||28|
| Other significant support||20.3||7.3||4||28|
|GENERAL QUALITY OF LIFE (Q-LES-Q INDEX)||3.5||0.8||1.5||4.9|
| PHYSICAL HEALTH||3.5||0.8||1.5||4.9|
| Subjective feelings||3.6||0.9||1||5|
| Leisure time activities||3.4||1.1||1||5|
| Social relationships||3.4||0.9||1.4||5|
| General activities||3.4||0.8||1.3||5|
| Life satisfaction||3.7||1.1||1||5|
| Satisfaction with medicine||3.6||1.2||1||5|
Predicting CANSAS-P subscale scores
Table 3 presents a summary of regression models. In particular, symptom severity (β = 0.28, P < 0.01), and TBDI depression (β = 0.27, P < 0.015) were positively associated with the ‘Social disability’ scale. Illness severity (β = 0.28, P = 0.048), emotional distress (β = 0.53, P < 0.001), and emotion-oriented coping (β = 0.41, P < 0.01) were positively associated, while friend support (β = −0.28, P < 0.01), general activities (β = −0.34, P = 0.012), and satisfaction with medicine (β = −0.44, P < 0.001) were negatively associated with ‘Information processing disability’. Two positive (illness severity, β = 0.22, P = 0.015; and emotion-oriented coping, β = 0.26, P = 0.013) and two negative predictors (friend support, β = −0.53, P < 0.001; and life satisfaction, β = −0.22, P < 0.001) were significantly associated with the ‘Emotional processing disability’ subscale, whereas TBDI anxiety (β = 0.23, P = 0.049), and friend support (β = −0.33, P < 0.001) were associated with the ‘Coping disability’ subscale. The effect size for the multiple regression models is between medium (f2 = 0.28; illness severity) and quite large (f2 = 1.13; emotional distress total score, and friend support). The obtained model explains 23% of the variability in the ‘Social disability’, 45% of the variability in the ‘Information processing disability’, 46% of the variability in the ‘Emotional processing disability’ and 36% of the variability in the ‘Coping disability’ subscale scores.
Table 3. Summary of multiple regressions to predict CANSAS-P sub-scale scores among 95 patients with schizophrenia and schizoaffective disorder from illness and personality related variables
| Social disability ||Symptom severity (PANSS total score)||0.28||2.8||0.006||0.39|
|R2 = 0.23, F = 6.2, P < 0.001||Depression (TBDI)||0.27||2.5||0.015||0.37|
| Information processing disability ||Illness severity (CGI-S)||0.28||2.0||0.048||0.39|
|R2 = 0.45, F = 8.1, P < 0.001||Emotional distress total score (TBDI)||0.53||4.4||0.001||1.13|
|Emotion-oriented coping (CISS)||0.41||3.4||0.001||0.69|
|Friend support (MSPSS)||−0.29||2.7||0.008||0.41|
|General activities (Q-LES-Q)||−0.34||2.6||0.012||0.52|
|Satisfaction with medicine (Q-LES-Q)||−0.44||3.5||0.001||0.78|
| Emotional processing disability ||Illness severity (CGI-S)||0.22||2.5||0.015||0.28|
|R2 = 0.46, F = 9.2, P < 0.001||Emotion-oriented coping (CISS)||0.26||2.5||0.013||0.35|
|Friend support (MSPSS)||−0.53||4.8||0.001||1.13|
|Life satisfaction (Q-LES-Q)||−0.22||2.0||0.048||0.28|
| Coping disability ||Anxiety (TBDI)||0.23||1.9||0.049||0.30|
|R2 = 0.36, F = 6.3, P < 0.001||Friend support (MSPSS)||−0.33||3.2||0.001||0.49|
In the present study unmet needs were found among 20–34% of the subjects that seem in line with previous studies of patients with schizophrenia and related disorders.2,6 The mean number of needs in the present study was 7.29 ± 4.8 (range 0–20) that also replicated data from previous publications.1,11,34
An exploratory factor analysis yielded a four-factor solution comprising 20 of 22 items that explain 50.4% of the total variance. The interpretation of these factors is complex.
The first factor or subscale, labeled ‘Social disability’, generally captures difficulties in basic social needs, and covers many everyday social and practical skills in patients' independent living. Poor social functioning showed at admission for a first psychosis.35–37
The second factor is generally defined as any difficulty linked to ‘basic education’, ‘telephone’, ‘transportation’, and ‘welfare benefits’. It is reasonable to assume that unmet needs in these areas might relate to cognitive or information processing impairment; therefore, it is called ‘Information processing disability’ that contributes significantly to functional impairments.38–40
The third factor, ‘Emotional processing disability’, consisted of lack of assistance with ‘daytime activities’, ‘psychotic symptoms’, ‘psychological distress’, ‘company’, ‘intimate relationships’, and ‘sexual expression’ item scores, that reflect deficits in the treatment processing of negative emotional information.41–47 Elevated emotional distress in schizophrenia has been found to be significantly associated with symptom expression,48–50 side-effects of antipsychotic agents,51 quality of life, coping abilities, and social support.15,18
‘Coping disability’ or fourth factor was constructed using ‘self-care’, ‘safety to self’, ‘safety to others’, ‘drugs’ and ‘child care’ scores. This factor related to lacking protective behavior and skills, and to decreased self-esteem, self-efficacy, and coping abilities in the context of feeling safe.52 Research has indicated that schizophrenia patients are inflexible in their use of coping strategies or styles, and they tend to use maladaptive or emotion-oriented coping styles.53–60
The present factor analysis only partly supports findings from two previous studies that did find some different factor structures within the CANSAS13 and CAN items14 that may be explained by using different versions of this scale, heterogeneity of samples, and factor loadings. The present sample, which included 30–69-year-old individuals with schizophrenia or schizoaffective disorder of 11–49 years' duration, looks quite different from those in previous studies.
Consequently, four main factors were scored and analyzed as CANSAS-P's subscales. The sum of the scale scores of the four CANSAS-P factors called subscales showed high internal consistency: Cronbach's α coefficient ranged from 0.67 to 0.77. A small correlation was found between these subscales (r range from 0.25, P = 0.014 to 0.38, P < 0.001).
The CANSAS-P subscale scores positively correlated with severity of symptoms, sensitivity, depression, anxiety, and somatization (r ranged from 0.34 to 0.45), but were negatively associated with general functioning (r = −0.34), friend (r = −0.46) and family support (r = −0.41), general QOL (r = −0.35) and two domains: satisfaction with medicine (r = −0.35), and general activities (r = −0.40) (all P < 0.001).
Lastly, severity of illness, symptoms, emotional distress, depression, anxiety and emotion-oriented coping were positive predictors, whereas friend support, general activities, life satisfaction and satisfaction with medicine were negative predictors of the CANSAS-P subscale scores. The effect size for the multiple regression models is between medium (f2 = 0.28; illness severity) and quite large (f2 = 1.13; emotional distress total score, and friend support). The obtained regression models explain 23–46% of the variability in these subscales.
Overall, findings from the present study are consistent with other studies of the patient's needs, and extend previous research regarding the correlation between CANSAS measures and clinical symptoms, psychological distress, coping styles, general functioning, social support, quality of life and disability scores.4,8–10,61–64
The present study has several limitations. First, acute psychotic patients were unable to participate in the study. Second, the results of the present study might apply only to adult (30–69 years old) individuals with chronic schizophrenia and schizoaffective disorder (illness duration: 11–49 years) who are more treatment compliant and more cooperative patients. The third limitation is common for most studies using self-reporting methodology for investigating needs in severely ill psychiatric patients. Finally, the cross-sectional design of this study cannot establish the direction of causality among the variables assessed.
In conclusion, this study suggests that a factor structure model appeared to fit the CANSAS-P items; it includes four dimensions that may represent disabilities in social and cognitive functioning, emotional responsivity and coping with everyday life challenges. These plausible subscales may be useful for the research and care needed to improve the treatment of psychiatric patients.