Possible confounding factors
Mental state may bias the subjective perceptions of patients while mental state and global function may also influence the views of clinicians. At baseline (T1, six months before the next hearing) GAF was lower (worse) for those detained under the Mental Health Act 2001 when compared to those detained under the Criminal Law (Insanity) Act (mean 48.3(SD 19.3), n = 17 vs 63.3(18.0), n = 58, ANOVA F = 8.8, df = 73, p = 0.004), the PANSS total score was higher (worse) for those detained under the Act of 2001 (62.5(20.9) vs 50.2(16.6), F = 6.4, p = 0.014) and the PANSS positive symptom score was significantly higher (worse) in those detained under the Act of 2001 (15.5(6.9) v 11.5(5.0) F = 6.9, p = 0.010). PANSS negative and PANSS general symptom scores tended to be higher also in those detained under the Act of 2001 and who would go on to appear before the MHT.
Remission status is defined as scoring below threshold level on eight key symptoms of mental illness on the PANSS and sustaining this status for at least six months . Remission status did not differ significantly at baseline between those detained under either Act. The patients' rating of WAI concerning their consultant psychiatrist was higher (more positive) at baseline (T1, six months before the hearing) in those who were then in remission (n = 20) compared to those who were not in remission (n = 55) (not in remission 55.6(SD 15.6), in remission 66.6(13.9) ANOVA F = 4.9, p = 0.03). This was true also for the patients' rating of ITP concerning their consultant psychiatrist (not in remission 37.4(7.3), in remission 41.7(5.5), ANOVA F = 5.7, p = 0.019). Primary nurses also tended to be rated higher by those in remission though this did not reach statistical significance, and the patients' rating of ITP for doctors in general did not differ between those in remission and those not in remission.
Perceived coercion, perceived procedural justice and impact of hearing
The MacArthur scales [2, 3] for perceived coercion, perceived procedural justice and impact of hearings were minimally modified for use in this context (Additional file 2).
The MacArthur Perceived Coercion Scale elicits positive or negative appraisals of five items concerning subjective control or subjective coercion over the outcome of the hearing-freedom, choice, initiative, control and influence. Each item is rated from 1 (no person control) to 7 (personal control), so 4 can be taken as neutral. These items were later recoded as dichotomous (0/1), where a rating of four or more counted as positive. Cronbach's alpha was 0.714 for patients and 0.623 for the consultant psychiatrist (0.655 for the dichotomised scale).
The MacArthur Perceived Procedural Justice Scale elicits six items concerning the role of an actor in the hearing. These are voice (whether the subject was able to express themselves to the person in question), interest (whether the person in question was interested in the subject), respect, fairness, satisfaction with the person in question and satisfaction with the procedure overall. These items are rated on a Likert self report scale from 1 ('not at all') to 7 ('definitely'). Patients were asked to rate these items first regarding the legal chair of the hearing and their legal representative, then separately regarding the role of their treating consultant psychiatrist in the hearing. A further scale rated in the same way elicits the impact of the hearing on the patient.
The treating consultant psychiatrist was asked to complete the same scales, including the Perceived Procedural Justice Scale regarding the legal chair of the hearing and patient's legal representative, and separately regarding the patient's role in the hearing. In order to analyse perceived coercion, perceived procedural justice and the impact of the hearing as outcomes, these items were later recoded as dichotmous (0/1), where a rating of '4' or more counted as positive. Cronbach's alpha was 0.904 for the patients rating the legal actors in the hearing (0.820 for the dichotomised scale) and 0.858 for the consultant psychiatrist (0.523 for the dichotomised scale). When the patient rated their treating consultant psychiatrist in the hearing, Cronbach's alpha was 0.925 (dichotomised alpha = 0.871) and when the consultant psychiatrist rated their patient in the hearing Cronbach's alpha was 0.864 (dichotomised alpha = 0.817).
The MacArthur Impact of Hearing Scale elicits ratings for six items regarding the person's subjective feelings after the hearing on self report Likert scales from 1 to 7, worse/better, upset/calm, less respected/more respected, confused/informed, less hopeful/more hopeful, and 'globally overall' good/bad. These items were also later recoded as dichotomous (0/1), where a rating of '4' or more counted as positive. When patients rated the impact of the hearing, Cronbach's alpha was 0.888 (dichotomised alpha = 0.847) and when consultant psychiatrists rated this, alpha = 0.906 (dichotomised alpha = 0.825).
The scales composed of summated dichotomised items generally correlated better than the scales of 'raw' data. The patients' and psychiatrists' ratings of perceived coercion correlated Spearman r = -0.322, p = 0.005. Patients and psychiatrists' ratings of perceived procedural justice regarding the legal actors in the hearing Spearman r = -0.11 (NS). The patients' rating of perceived procedural justice in the hearing concerning their treating psychiatrist's role correlated with the psychiatrists' rating of the patients' role in the same hearing r = 0.695 p < 0.001 (r = 0.304, p = 0.008 for the full scale items summated) and comparing the patients and psychiatrists' rating of the impact of the hearing, r = 0.189 (NS). When the patient and psychiatrist ratings for each dichotomised item were compared, there was a high degree of agreement. For perceived coercion, patient and psychiatrist agreed for three or more items in only 13 out of 75 cases. However, for perceived procedural justice regarding the role of the legal actors in the hearings, patient and psychiatrist agreed for 4 or more of the 6 items in 47 (63%) of 75 cases (binomial exact probability compared to expected random 50% p = 0.037). For perceived procedural justice concerning the roles of patient and psychiatrist, the patient and consultant ratings agreed in 4 or more items in 56 (75%) of 75 cases, (exact binomial probability compared to random 50% p < 0.001). For the impact of the hearing, patient and psychiatrist agreed on four or more items in 55 (73%) of 75 cases (exact binomial probability compared to random 50% p < 0.001). This suggests that the mutual perceptions of patient and psychiatrist regarding the hearing agreed very well concerning the nature of the hearings in specific domains such as perceived procedural justice and impact, but not concerning perceived coercion.
The dichotomised items were next used to divide the overall scale scores for subjective ratings of aspects of the hearings into 'positive' and negative'. For the five item perceived coercion scale (dichotomised) a score of 3 or more was taken as 'positive,' while for the remaining six item scales, a score of 4 or more was taken as 'positive'. Once again, the patient and psychiatrist did not agree regarding perceived coercion, with 12 (16%) of the pairs both rating 'positive' and 17 (23%) of the pairs both rating negative (X2 = 2.6, df = 1, NS). For perceived procedural justice regarding the legal actors in the hearings, the psychiatrist and patient both rated this positive in 59/75 (79%), both negative in none (X2 = 0.3, NS). For perceived procedural justice regarding the role of patient and psychiatrist, the mutual ratings agreed that this was positive in 63 (84%) and negative in 6 (8%), an agreement of 69/75 (92%) overall (X2 = 31.1, p < 0.001). For the impact of the hearing, the patient and psychiatrist both rated this positive in 62 (83%), both rated it negative in 3 (4%), an agreement overall in 65/75 (87%) (X2 = 10.9, p < 0.001, all df = 1, all n = 75 pairs)