Details of recruitment are presented in the figure. Eighty seven patients fulfilled entry criteria and completed assessment before treatment. Of these, 72 who completed treatment and seven who dropped out of treatment completed assessment after treatment, resulting in a total sample of 79 for whom complete data were available. In the remaining eight the last available scores were carried forward.
Baseline characteristics of the 87 patients were as follows: mean age 38.6 (SD 11.0) years; 69 men; 64 single; 24 lived alone, 17 lived with a partner, 31 lived with parents, and the remainder with others; 61 left school at 16 years; 76 were unemployed, five were in paid employment, six were in voluntary employment or similar, two had never worked; 64 were unskilled and 21 were skilled or professional; 78 had a diagnosis of schizophrenia, eight had schizoaffective psychosis, and two had delusional disorder; the median (range) duration of illness was 11 (1-42) years; median (range) number of admissions to hospital was 3 (0-20); 10 had a forensic psychiatric history; and 12 had a history of substance abuse.
Fidelity of treatment
Correct classification was made on 33 of the 34 taped sessions (97% agreement), with one early coping session being misclassified as “borderline” supportive counselling. This level of discrimination between cognitive behaviour therapy and supportive counselling is highly satisfactory and indicates with a high degree of confidence that these treatments followed protocol.
Doses of drugs over the trial were converted to mean daily equivalents of chlorpromazine and compared across groups by means of Kruskal-Wallis one way analysis of variance; this indicated no significant differences between treatment groups (medians of daily drugs in chlorpromazine equivalents: cognitive behaviour therapy 425, supportive counselling 517.75, routine care 450; χ2=0.963; P=0.62). Nine patients were taking atypical neuroleptic drugs (clozapine or risperidone) during the duration of treatment (two in the cognitive behaviour therapy group, four in the supportive counselling group, and three in the routine care group). Thus there was no evidence of systematic and significant differences between the groups in terms of medication.
Comparisons of efficacy of treatment groups
The relative efficacy of the treatment groups was analysed by means of one way analysis of variance on the changes in scores before and after treatment. Relevant changes and confidence intervals are presented in the table. A post hoc test (Tukey-HSD) indicated the location of any significant differences at the 0.05 level between pairs of comparisons. Analysis by intention to treat was completed on the 87 allocated patients, with last observations (scores before treatment) being carried forward for the eight patients for whom scores after treatment were missing.
Significant effects were found for the number of positive psychotic symptoms a patient experienced, and their severity. The cognitive behaviour therapy group showed a greater improvement than the supportive counselling group and patients in the routine care group showed a slight deterioration (table).
Clinically important improvement
Eighteen subjects achieved 50% improvement in psychotic symptoms in both severity and number of symptoms, taken as representing an important clinical improvement. Eleven (of 33) were in the cognitive behaviour therapy group, four (of 26) in the supportive counselling group, and three (of 28) in the routine care group. This difference was significant when the number of patients who showed a 50% or greater improvement was compared between those who received cognitive behaviour therapy and the other two groups combined (χ2=5.18; df=1; P=0.02). Five patients who received cognitive behaviour therapy and two who received supportive counselling were free from all positive symptoms after treatment, whereas none who received routine care alone achieved this.
Because an improvement of 50% or more in psychotic symptoms represents such an important clinical change in patients with chronic schizophrenia a logistic regression was performed to investigate which variables contributed to this improvement. Treatment group was represented as a categorical variable having three levels in the analysis, with routine care acting as the reference category. Three variables showed a significant contribution: allocation to cognitive behaviour therapy (B 2.064; SE 0.726; P=0.0045; Exp(B) 7.878); duration of illness (B −0.144; SE 0.054; P=0.0079; Exp(B) 0.866); severity of symptoms on the psychiatric assessment scale (B −1.893; SE 0.815; P=0.02; Exp(B) 0.151). Thus receipt of cognitive behaviour therapy resulted in almost eight times greater odds of showing a reduction in psychotic symptoms of 50% or more than subjects receiving routine care alone. A shorter duration of illness and less severe symptoms at allocation also significantly contributed to this outcome. The odds of subjects showing a reduction in psychotic symptoms of 50% or more decreased by a multiplication factor of 0.87 for every additional year of duration of illness, and decreased by a multiplication factor of 0.15 for every unit increase in severity of illness.
Relapse and readmission to hospital
The hypothesis that cognitive behaviour therapy would result in a reduction in subsequent relapse is more speculative as patients were recruited and treated during the residual phase of the disorder and there will be varying times since the patient’s previous acute episode and in their vulnerabilities to relapse. Furthermore, relapse in terms of exacerbation of positive symptoms is difficult to assess accurately as all patients were symptomatic. Thus readmission to hospital for clinical deterioration that resulted in functional impairment and admission for at least 5 days, although not an ideal measure, was chosen as a practical indicator of relapse.
Eighty five patients were followed up from the original 87, two having died of natural causes. From before to after treatment (3 months) there were no relapses in either cognitive behaviour therapy or supportive counselling groups and four (14%) in the routine care group. The total number of days spent in hospital for patients in the routine care group was 204, whereas one patient from both the cognitive behaviour therapy and supportive counselling groups spent 1 day in hospital.