Randomised trial of personalised computer based information for patients with schizophrenia

Study population and sample

All 420 patients with a diagnosis of schizophrenia (F2 on ICD-10, international classification of diseases, 10th revision) who were living in the community in south Glasgow15 were considered for inclusion in the study. Patients were excluded if they were aged over 65 years, had an uncertain diagnosis, were judged by community psychiatric nurses at the time of contact to be acutely ill, had chronic symptoms or physical problems restricting participation, were persistent defaulters, or had recently been involved in an education programme. This left a population of 301 (72%). Greater Glasgow Community and Mental Health Trust gave ethical approval for the study.

Recruitment and randomisation

Patients were sent a letter inviting them to join the study and could opt out without further contact. Researchers attempted to arrange meetings either at home or at an NHS facility to explain the study and seek consent. Some patients were excluded at this stage as new information became available. Of the 230 who seemed eligible and were contacted, 118 (51%) refused to participate. The remaining 112 patients were randomly allocated to one of the three interventions: 56 (50%) to computer education, 28 (25%) to sessions with a community psychiatric nurse, and 28 (25%) to the combination intervention. We had aimed to recruit 140 patients, giving 84% power at the 5% significance level to find differences in drop out rate of 10% v 30% for the group given nurse based education and those given computer based education. (See appendix B on BMJ's website for further details.)

Interventions

All interventions involved five sessions and were intended to increase patients' knowledge about schizophrenia. Sessions on the computer were held in a dedicated room at a resource centre through which all community services are organised. Sessions with the community psychiatric nurse could be at home or in the resource centre.

Patient flow through study

• Download figure
• Open in new tab
• Download powerpoint

Computer only—Patients were shown how to use the computer by a researcher (LP). Patients could only go forwards to the next screen—until the end of the session, when they could recap. There were three types of screen display: (a) general information, (b) personal information from the viewing patient's medical record “embedded” in more general information, and (c) questionnaires (including medical record audit), plus feedback displays (see appendix A on BMJ's website for examples). At the end of the session, LP printed out the information displayed if the patient requested it.

Community psychiatric nurse only—The hour long sessions with the community psychiatric nurse (KMcK) covered the same content as the computer system. Personal issues could be introduced by the patient. Patients could also be given a printed summary, but this did not include any personal information.

Combination of community psychiatric nurse and computer—The first session was with the community psychiatric nurse, sessions 2–4 were on the computer, and the last session was again with the nurse. Patients were given relevant printed summaries from sessions.

Data collection

We used four psychological measures. We assessed patients' knowledge with a structured interview, the knowledge and information about schizophrenia schedule (KISS), which was developed for the pilot study partly on the basis of the KASI used for carers (see appendix C on BMJ's website).16 We carried out this interview three times: before the first session, at the end of the fifth session or at a separate contact shortly after (patient choice), and three months later. A consultant psychiatrist (DAC) assessed the patients' mental state (with the brief psychiatric rating scale (BPRS)17), insight (insight and treatment attitudes questionnaire (ITAQ)18), and functioning (global assessment of functioning (GAF)19^–21) in a separate interview before the intervention and three months after the last session.

We assessed patients' opinions about the interventions using questionnaires, specifically about the computer for the computer group at the end of the first session and, for all groups, at the end of the fifth session and three months later.

The computer automatically monitored the time taken by patients on each screen display, and the community psychiatric nurse recorded time spent on sessions. Times spent in contacting patients, transporting them to their intervention if necessary, and preparing for the intervention were also recorded. Patients' contacts with community health services were routinely recorded (by Joan Jamieson, information nurse in primary care trust, extracting information from Peak system (Peak Systems, Irvine, Scotland)).

Data analysis

In our intention to treat analysis (112 patients), we assumed that the values for psychological variables for patients who did not complete follow up had not changed. We compared changes in scores between groups using analysis of variance F tests and Kruskal-Wallis tests. We used the χ² test to compare patients classified as improved or not improved on the psychological and patient satisfaction measures.

Economic analysis

We estimated the “long run opportunity marginal cost” of interventions.22 Patients not taking part in the intervention had zero intervention costs. Community health service contacts were costed for three months before consent until three months after the intervention for each patient. We modelled NHS costs over four years assuming that computer hardware was “written off” over this period and that future costs were discounted at 6%. Eight options were modelled: the three interventions in this study, the same three interventions assuming the existence of an electronic patient record (with no need to extract data from a manual case record), and two options using a computer based approach as an addition to other routine contacts. (See appendix D on BMJ's website for more details.)

Patients completing the intervention and non-attendance

The 112 participants were aged from 18 to 65 years, and 67% were men. The figure shows their passage through the trial. Sixty seven (60%) of the patients completed all five sessions, and completion rates in the three interventions were not significantly different (71% for combination education, 61% for computer only, 46% for community psychiatric nurse only). Thirty six patients (32%) were lost from the study before the first session. These were more likely to be women (49% of women v 24% of men; χ²=6.9, df=1, P=0.009). Nine (8%) of the patients, all men, attended the first session but dropped out subsequently.

Use of, and reaction to, computer sessions

The overall time spent on the five computer sessions was quite short (median 69 minutes (range 34-143)). Of the 39 patients who completed the first session, 27 found the touch screen easy to use, nine found it “OK,” and three found it moderately difficult. All knew where to touch the screen, and all felt “OK” or keen to use the computer again. Twenty two thought that the computer was easy to use, and the other 17 thought it “OK.” Nearly a third (12/39) encountered a display or part of a display in which they were uncertain about what to do next. (See appendix E on BMJ's website for more details.)

Patients' opinions about interventions

Among the patients who completed all five sessions, all of those given nurse based education only thought that the information was definitely relevant, compared with three quarters of those given computer based education only and half of those given the combination (table 1). For nearly all other items, the patients given nurse based education only were more satisfied, although no measure reached significance. Of the 20 patients given the combination, 13 preferred the nurse based sessions and seven preferred the computer for most items asked (see appendix F on BMJ's website). One patient preferred the computer sessions because the sessions with the nurse were too long, another because it helped to see information “in black and white.” Patients who preferred the nurse sessions mentioned the personal touch, empathy, helped to have a chat, and preferring to talk than read.

Psychological outcomes

The patient groups showed no differences in baseline psychological measures, but the range of scores was high (table 2). Mean scores for all four measures improved for the 67 patients who completed the five sessions, although the level of improvement was fairly small (table 3). Using an intention to treat analysis, we found that fewer of the patients allocated to nurse based education only had improved knowledge at three months, but those who completed the sessions showed greater improvement in insight and mental state.

Intervention costs

The main costs were staff time in contacting patients and carrying out the intervention—about £200 for each patient who completed treatment (table 4). There were no differences between groups for NHS or patient costs. A major contributor to the cost of the computer sessions was travel time: for patients who required transport, staff had to make four journeys per session. Of the 76 patients who attended one or more sessions, 19 required no transport, eight required transport for some sessions, and 49 required it for all the sessions they attended. The capital costs of the computer and its maintenance over four years was £3300.

Effect of intervention on other community health service costs

We compared other community health service costs with intervention costs. One patient, living in the community when identified, was admitted to our study with an approximate three month inpatient cost of £9720. After excluding this patient, we found that the patients' mean cost in the three months before the intervention was £55 (median £30 (range 0-616)), compared with £65 (median £37 (0–825)) in the three months after. The mean number of attendances at community health services before the intervention was 7.6 (median 6 (0–66)) and was 7.7 (5 (0–84)) afterwards. For the patients given nurse based education only, community health service costs declined in the three months after the intervention whereas the costs rose for the other two groups. However, costs generally displayed a wide range, and some of the difference seen might have been due to “regression to the mean.”

Modelling of alternative interventions

We modelled the cost of the interventions over four years for eight alternative methods of delivery: the three interventions in this study, the same three interventions but with an electronic patient record system, and the computer intervention used as an additional activity at an already planned health service contact (with or without an electronic patient record).

The three interventions used in this study would cost between £360 000 and £380 000 over four years. With an electronic patient record, these costs would be reduced to between £210 000 and £290 000. However, patients tended to have at least seven contacts with health services in three months, and if computer based educational sessions were combined with these other visits staff workloads and costs would be substantially reduced and patient costs would also be less. The option assuming an electronic patient record and computers based at each of the seven clinics in south Glasgow that the patients routinely attended would cost only £40 000 over four years (see appendix D on BMJ's website).

Psychological measures and patients' opinion

Nearly two thirds of the patients who completed the intervention and 38% of those randomised to an intervention showed improved knowledge at six months after randomisation, but there was little difference in this or other psychological outcomes between the three intervention groups. Patients given nurse based education showed more improvement in mental state at the end of the sessions, even though the intervention was not designed to make any difference. On the whole, patients were more satisfied with the nurse based education, but the difference was not great. Although most patients in the combination group preferred the nurse's sessions, a third preferred the computer. Most patients given computer based education only found the computer reasonably easy to use, and drop out from this group was similar to that in the group given nurse based education only. There were no major problems with including information from the patients' medical records in the computerised material.

Cost

From our pilot study, we predicted that the computer based intervention would be less costly, but, because of the need to transport patients to the computer, this was not the case. In general, the cost of the educational interventions organised as separate sessions was high compared with other community care contacts (£200 for intervention v £120 annual contact costs), and this did not include recruitment costs. However, our study was short term, examining costs over only six months, and improvements in schizophrenic patients' functioning may be associated initially with greater use of psychiatric services.23 Schizophrenia costs the NHS nearly £500m a year to treat.24 The patient in our study who had an inpatient stay costing over £9000 shows that the potential savings from improving schizophrenic patients' functioning are substantial, but evaluation of educational interventions, whether computer based or not, needs to be carried out for longer than six months and in a larger population to measure any impact on cost.

Recruitment and follow up

The effectiveness of educational interventions in routine practice depends on their being able to recruit patients and maintain their participation. After exclusions and refusals, only a quarter of the original population took part in our study, and this is not unusual for studies of patients with schizophrenia. Furthermore, of the 112 patients who agreed to participate, most required transport by the researchers if they were to attend. Even then, only 67 (60%) completed their intervention.

Conclusions

Computer based educational sessions for patients with schizophrenia were no less effective than sessions with a community psychiatric nurse and were acceptable to most, and preferred by some. However, running special educational sessions, whether by computer or by community psychiatric nurse, is costly and does not seem to be the way forward. It would be worth investigating the feasibility and effectiveness of giving computer based education as an addition to other health service contacts. Although this has higher capital costs, the overall cost would be less and it might lead to greater patient participation.