Hamann 2006.
| Methods | Cluster randomised controlled trial (unit of randomisation = ward). | |
| Participants | Inpatients with ICD‐10 diagnosis of schizophrenia or schizophreniform disorder (F20/F23). Mean age intervention group: 35.5 yrs (SD 11.9); mean age control 39.6 years (SD 10.8). Intervention group 41% female, control group 53% female. | |
| Interventions | Patients ‐ Decision aid ‐ 16 page booklet. Patients were assisted in working through this by nurses. Duration 30 to 60 minutes. Patients met with their physicians within 24 hours afterwards for a planning talk. Nurses ‐ instructed on use of decision aids. Physicians ‐ two information sessions on SDM and the required communication skills. |
|
| Outcomes |
Physician rated: Psychopathology: Positive and Negative Syndrome Scale for Schizophrenia (PANSS ‐ validated, Kay 1987) (Baseline and at discharge). Global functioning (GAF, validated APA 2000) and severity of illness (Clinical Global Impressions Scale, validated Guy 1976) (6 and 18 months after discharge Rating of time spent per week with patient (at discharge). Rehospitalization (6 and 18 months after discharge ‐ dichotomous outcome) Provider satisfaction (unvalidated 5 point rating scale at point of discharge) Patient rated: Patient satisfaction (ZUF‐8, German version of the CSQ ‐ validated, Schmidt 1989) (at discharge). Risk communication and confidence in decision (COMRADE ‐ validated, Edwards 1999) (immediately after the intervention and at discharge). Patient knowledge (unvalidated questionnaire, at discharge). Composite measure: Patient concordance with treatment plan ‐dichotomous outcome (based on patient completion of MARS questionnaire, patient compliance as rated by the physician on a 4‐point scale, and plasma levels of antipsychotics) rated at 6 and 18 months post discharge. |
|
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Adequate sequence generation? | High risk | Sequence generation took place after wards had been paired based on their characteristics so this is not truly random. |
| Allocation concealment? | High risk | Randomisation was at the ward level. Adequate allocation concealment at the level of the patient would not be possible. |
| Blinding? All outcomes | Unclear risk | Participants, providers, outcome assessors were not blinded, unclear whether data analysts were blinded. |
| Incomplete outcome data addressed? All outcomes | High risk | Significant loss to follow up |