Effects of group musical therapy on inpatients with chronic psychoses: A controlled study
Abstract
Abstract The objective of the present study was to examine the efficacy of group musical therapy for inpatients with DSM-IV schizophrenia or schizoaffective psychosis. Thirty-four therapy group subjects in a ward for long-stay female patients received 15 group musical therapy sessions over 4 months, while 32 waiting group subjects from another ward with the same function were to wait for the sessions until the studied course was completed. The assessment included measures of psychotic symptoms, objective quality of life and subjective musical experiences. Comparison of the groups indicated that significant advantages in the therapy group subjects were detected in some measures concerning personal relations and a subjective sense of participation in a chorus activity. However, the follow-up evaluation suggested that the improvement might not be durable. These findings suggested that the musical therapy had some, but possibly only short-lived, effects on personal relations and musical experiences of chronic psychotic patients.
INTRODUCTION
The influence of music may well extend over many areas of human functions ranging from physiological ones to quality of life and psychological well-being. Deservedly, applications of music for treatment of various diseases have been recognizable.1 This generality could justify that patients with schizophrenia and other chronic psychoses would enjoy favorable effects of music. To put it more specifically, the deficits characterizing schizophrenic patients such as impoverishment of interpersonal contact and inadequately organized behavior, corresponding signs of which have been described in some areas of music performance,2,3 can be ameliorated by using the communication potentiality and harmony directedness of music.4
The beginning of a systematic undertaking of musical therapy for patients was alleged to be a consolatory activity of musicians in mental hospitals in the early years of the last century.1 Thereafter, thanks to innovative efforts to broaden its applicability, musical therapies of various kinds have been performed increasingly in clinical practice for patients with psychoses. In Japan, being influenced by accomplishments of pioneers from North American and European countries, several therapists commenced in the 1950s systematized efforts of musical therapy practice for psychiatric inpatients, most of whom were patients with schizophrenia.5 These gradually expanding activities resulted in the establishment of the Japanese Federation of Musical Therapy in 1995. The federation has unified hitherto rather separately practicing musical therapy organizations, and provided a base for the specialized training and its further dissemination over various areas of psychiatric practice.
This situation highlights the need to verify the effectiveness of musical therapy on patients with chronic psychoses. It is also indispensable for further development of the therapy to seek answers to practical questions such as what effects can be expected from the therapy, what assessment methods are of use to detect the effects, and how long the effects endure. However, the number of previous studies that addressed these issues is quite limited. For instance, we could find only three controlled studies through a literature search, which revealed rehabilitative effects of certain forms of musical therapy on the patients.6–8 To remedy this state of affairs, we carried out a controlled study to examine the effects of group musical therapy on older female inpatients with chronic psychoses. The main features of the present study were a relatively long duration of the studied therapy course and the inclusion of the evaluation of subjective experiences related to music in the assessment.
METHODS
Design
To examine the effect of group musical therapy, we used the method of comparing clinical measures of subjects in the therapy group who received a musical therapy course with those of subjects in the waiting group who were to start the course after the studied course was completed. Allocation of patients to groups was according to which of the two wards they belonged. Intention to treat analysis was applied. The assessment using clinical measures for the subjects was conducted at pre-session, post-session and 4 months after the studied course periods. During the period between the second and third assessments, the sessions were performed for the waiting group patients. For the therapy group subjects, the third assessment was considered a follow-up evaluation, and a bi-monthly assessment of ward life activities by ward nursing staff was added.
Subjects
The two wards for female long-stay patients in which this study was performed were those of Tokyo Metropolitan Matsuzawa Hospital, a large psychiatric center for regional psychiatric service in central Tokyo, which has 30 wards with various specialized functions. Both wards were composed mostly of schizophrenic patients of similar demographics who needed enduring rehabilitative services. Annually, approximately one-tenth of the patients were discharged from the wards to the community including nursing homes and group homes, and one-fifth were transferred to other wards for some reasons such as a need for more intensive rehabilitation or specialized treatment of physical illness. Routine recreational and rehabilitative activities of the same intensity and frequency (e.g. chorus or dance meetings, excursions and occupational therapies) were carried out in both wards during the study period.
Inclusion criteria of the subjects were diagnosis of schizophrenia or schizoaffective psychosis and absence of unequivocal organic cognitive impairment or mental retardation. The diagnosis of the subjects was made on the basis of examining case records according to DSM-IV criteria, and confirmed in the later assessment. Seventy-four eligible patients were approached, and 66 patients gave written informed consent to participate in the study. There was no subject who had specific musical skills.
Musical therapy
The course of the group musical therapy consisted of 15 1-h per week sessions. The objective of the therapy was facilitation of enjoyment in listening and performing music and interpersonal communications through musical activities. Two session groups from the therapy group subjects, each consisting of fewer than 20 participants, received the therapy. Seven musical therapists (YT, MN, CI, YK, MW, MO and KT) were all female, and had an average clinical experience of 8.0 years (SD 6.4; range 3–22). Two of them were in charge of each session group on a rotating basis in such a manner that at least one was in attendance successively to preserve the continuity of the course. During the sessions, participants sat in a circle, on the circumference of which a piano was placed. Occasionally, a violin, clarinet or flute was used as a leading instrument in addition to the piano. To ensure efficient therapist–patient communications, all participants wore a name badge throughout the sessions. Improvization was principally not applied.
The starting phase of the course was characterized by rather passive patient involvement and gentle and contained contact, consisting mostly of therapists performing music and patients listening. Gradually, the proportion of singing folk songs and old popular songs in chorus was increased, and then games and plays using music were introduced. In the later phase, stress was put on active patient participation and facilitating interpersonal communications by various playful games and patients joining in the percussion of bells, tambourines, triangles and castanets.
Assessment
The measures used for assessment by the psychiatrist raters (NH, SN) were Positive and Negative Syndrome Scale (PANSS)9 and the Quality of Life Scale (QLS).10
Regarding PANSS, a scale widely used for evaluating symptoms of psychotic patients, the reliability study of the Japanese version of PANSS by our research group gave favorable results.11 The raters had fulfilled Kay's criteria12 by the end of the training course applying eight video recorded interviews before this study; the concordance rate for the whole items (a deviance of 1 point was also included as concordant) must be over 80%, and scores of each subscale, within the range of 80–120% of consensus ratings. During the study period, confirmation of the skill by attaining a consensus PANSS rating between the raters was also conducted during this study.
The QLS was developed for the evaluation of the patients' objective quality of life by means of assessing the patients' deficit symptoms. It has 21 items and four subscales of interpersonal relations, instrumental role, intrapsychic foundations, and commonplace objects and activities. In addition to precedent rating training, we carried out an interrater reliability study of rating the QLS items except items of work satisfaction that were not applicable to the subjects, and commonplace objects and activities that were rated based on specified inquiry. In the reliability study, the two raters made independent ratings of the scale on the basis of a single interview for each of 20 patients during the study period, and analysis of variance (ANOVA) intraclass coefficients (ICC) were calculated to measure the level of the concordance.
To inquire about the subjects' subjective musical experience in ward life, we devised new self-report scales for assessing musical experiences: one for music listening experience and the other for a ward chorus activity which were carried out independently of the studied musical therapy. They were both composed of four 5-point Likert scale items inquiring about the degrees of motivation, concentration, aroused emotion, and emotion shared with peers. High scores of the scales meant that the subjects more actively participated in the musical experience.
The ward nursing staff also rated the therapy group patients using a 5-point, 10-item scale (developed by KS, KH, AS and IK) for assessing ward-life activity every 2 months during the study period. On the basis of a factor analytic study, this scale was divided into two subscales: ward activity and ward adjustment. Items of the ward activity subscale were less bed-liking behavior, general activity, recreational activity, interpersonal activity and vividness of facial expression, and items of ward adjustment were volition to occupational therapy, participation in ward meetings, personal hygiene, garb neatness, and tidiness of belongings. High scores in this scale indicated more pertinent ward life activity of the subjects. The nursing staff conducted the ratings of the subjects of whom they were in charge. Most of the nurses were independent of the research. The concurrent validity of this scale was sought by calculating Pearson's correlation coefficient of the score with the total general behavior score of Rehabilitation Evaluation Hall and Baker (REHAB),13 a scale for the assessment of inpatients' capability of daily life, the Japanese version of which has been confirmed to have sufficient reliability and validity.
Data analysis
Analysis of variance and χ2 test were used to detect significant differences in demographic and clinical features at baseline between the groups. For analyzing longitudinal data of the investigation, we did repeated measures ANOVA with the group as a between-subject factor and with time as a within-subject factor for comparison between clinical measures before and after the therapy, across the groups. In addition, we performed repeated measures ANOVA including polynominal contrasts for the therapy group subjects to detect significant changes in the scale scores that would give positive findings in the previous ANOVA and the ward life scale scores. In statistical analyses, we applied a significance level of 0.05. The SPSS Release 10.0.5 statistical package (SPSS Inc., Chicago, IL, USA; 1999) was used for the entire analyses.
RESULTS
The demographic and clinical characteristics of the patient groups at the pre-session assessment are shown in Table 1. Most patients were in their 60s, and had a long history of hospitalization. The diagnosis of two subjects of the therapy group and one of the waiting group was schizoaffective psychosis, and the rest were schizophrenia. The subjects were voluntarily admitted except two patients of each group who were compulsorily admitted with the consent of the family guardian. There was a significant difference between the two groups in the duration of current inpatient treatment with longer hospital stays for patients of the waiting group (F1,64 = 4.87, P < 0.05). This difference was supposed to be derived from the study design that did not include randomized patient allocation. However, as shown in Table 1, the subjects of the two wards were comparable for the most part.
| Therapy group (n = 34) Mean (SD), range | Waiting group (n = 32) Mean (SD), range | |
|---|---|---|
| Age at pre-session assessment (year old) | 66.1 (10.2), 43–84 | 69.0 (7.6), 49–83 |
| Age at onset (year old) | 32.1 (14.9), 14–70 | 29.5 (13.6), 13–65 |
| No. lifetime hospitalizations | 3.6 (2.7), 1–13 | 4.0 (3.2), 1–15 |
| Years of full-time education | 10.0 (3.0), 6–16 | 10.5 (2.3), 6–14 |
| Marital history | 12/34 (35%) | 17/32 (53%) |
| History of fulltime regular work | 18/34 (53%) | 19/32 (59%) |
| Duration of current hospitalization (day) | 4,664 (4,847)* | 7,817 (6,666) |
| Neuroleptic daily dose (mg)a | 697 (648), 12–2400 | 765 (604), 50–2100 |
- * P < 0.05, analysis of variance.
- a Chlorpromazine equivalent.
The interrater reliability study of QLS items gave satisfactory results: the average ANOVA ICC of the items was 0.76 (SD 0.12; range 0.41–0.95). The internal consistency of the two scales of music listening and a chorus experience, and the ward life scale and its two subscales of ward activity and adjustment, proved to be good to excellent. Their Cronbach's alpha coefficients at the pre-session assessment were 0.840, 0.871, 0.929, 0.839 and 0.941, respectively. Regarding the convergent validity of the ward life scale, its total score was highly correlated with the total score of the REHAB general behavior scale; the Pearson's correlation coefficient was 0.72 (P < 0.001, two-tailed).
The number of sessions the therapy group subjects participated in was between three and 15, averaging 11.8. The average chlorpromazine-equivalent neuroleptic daily doses (SD) of the therapy group and the waiting group at the post-session assessment were 729 (648) mg and 786 (604) mg, respectively. There was no significant difference in the doses between the groups during the study period. At 4 months after the therapy assessment three subjects in the therapy group could not be followed up because of discharge.
The two groups' mean scores and standard deviations on scores of PANSS, QLS and the music experience scales are shown in Table 2. There was a significant difference at the pre-session assessment in the scores of PANSS positive and general psychopathology scale scores with more severe symptoms in the waiting group subjects (F1,64 = 5.29, P < 0.05 and F1,64 = 4.15, P < 0.05, respectively). Therefore, the initial scores of these two scales were to be hereafter put into repeated measures ANOVA models as covariates if applicable.
| Pre-sessionaMean (SD) | Post-sessionaMean (SD) | 4 months after therapybMean (SD) | |
|---|---|---|---|
| PANSS positive | |||
| Therapy group | 19.6 (5.0)c | 18.6 (4.3) | 18.7 (3.7) |
| Waiting group | 22.5 (5.2) | 21.3 (4.3) | |
| PANSS negative | |||
| Therapy group | 27.6 (5.8) | 25.4 (5.6) | 25.8 (5.1) |
| Waiting group | 28.6 (5.3) | 28.2 (5.9) | |
| PANSS general | |||
| Therapy group | 46.7 (6.2)c | 43.7 (8.5) | 44.2 (7.6) |
| Waiting group | 50.0 (6.9) | 45.6 (8.4) | |
| QLS interpersonal relations | |||
| Therapy group | 10.8 (5.7) | 12.6 (6.1) | 12.1 (4.4) |
| Waiting group | 10.4 (5.9) | 10.8 (5.1) | |
| QLS instrumental role | |||
| Therapy group | 4.4 (2.5) | 4.7 (2.4) | 4.5 (2.1) |
| Waiting group | 4.4 (1.9) | 4.4 (2.1) | |
| QLS intrapsychic foundations | |||
| Therapy group | 12.0 (7.1) | 14.2 (6.0) | 11.5 (4.6) |
| Waiting group | 10.3 (5.4) | 11.5 (5.4) | |
| QLS common objects and activities | |||
| Therapy group | 4.0 (2.0) | 3.7 (2.0) | 3.7 (1.9) |
| Waiting group | 4.3 (2.2) | 3.1 (2.1) | |
| QLS total score | |||
| Therapy group | 31.1 (14.1) | 35.3 (14.0) | 31.7 (9.7) |
| Waiting group | 29.4 (13.1) | 30.7 (12.1) | |
| Music listening | |||
| Therapy group | 10.2 (3.9) | 10.9 (3.5) | 9.8 (3.4) |
| Waiting group | 10.8 (4.9) | 10.9 (5.1) | |
| Chorus activity | |||
| Therapy group | 8.5 (2.7) | 9.6 (2.5) | 8.3 (2.7) |
| Waiting group | 9.3 (3.4) | 8.9 (2.9) | |
- PANSS positive, PANSS positive subscale score; PANSS negative, PANSS negative subscale score; PANSS General, PANSS general psychopathlogy subscale score.
- a n = 34 for the therapy group, n = 32 for the waiting group,
- b n = 31,
- c Significant differences between groups (P < 0.05, analysis of variance).
Repeated measures ANOVA for the PANSS subscales revealed a significant finding of an interaction of group and time for negative symptoms (F1,62 = 10.31, P < 0.01) that demonstrated a significant advantage of the musical therapy sessions for the symptoms. Individual PANSS items that showed significant advantage of the musical sessions were emotional withdrawal (negative syndrome item 2), poor rapport (negative syndrome item 3) (P < 0.05) and passive apathetic social withdrawal (negative syndrome item 4) (P < 0.05).
As for the QLS total score and its four subscale scores, repeated measures ANOVA gave a significant finding of an interaction of group and time for the QLS total score (F1,62 = 4.11, P < 0.05). Individual items that showed a significant interaction of group and time (i.e., a significant advantage of the therapy), were less socially withdrawal (P < 0.01) and empathetic toward other people (P < 0.05). All the PANSS and QLS items that gave positive findings appeared to be related to interpersonal activities.
A significant finding of the repeated measures ANOVA for music experience scales was an interaction of group and time for the chorus activity scale (F1,62 = 4.37, P < 0.05), which also suggested an advantage of the therapy.
It is shown in Table 2 that the PANSS negative score, the QLS total score and the chorus activity scale score that gave positive findings tended to return to the baseline scores at the follow-up assessment. Confirming this finding, the repeated measures ANOVA of their three measures during the entire study period revealed significant quadratic components (F1,30 = 15.66, P < 0.01, F1,30 = 13.12, P < 0.01 and F1,30 = 18.91, P < 0.01, respectively) in addition to the significant time effects (Wilks' lambda = 0.592, F2,29 = 10.00, P < 0.01, Wilks' lambda = 0.674, F2,29 = 7.03, P < 0.01 and Wilks' lambda = 0.593, F2,29 = 9.94, P < 0.01, respectively).
Figure 1 shows the changes in the scores of the ward life subscales of the therapy group subjects. An inspection indicated that the scores went up as the course proceeded, and declined to baseline levels 4 months after the course. Consequently, the therapy was suggested to have had an influence on the subjects' ward activity and adjustment. The overall repeated measures ANOVA of the serial scores of the ward life scale indicated a significant time effect (Wilks' lambda = 0.626, F4,26 = 4.00, P < 0.05) and a significant quadratic component (F1,29 = 10.20, P < 0.01). The same analyses of the ward life subscales gave consistent significant findings. The findings of the decline in the measures after the therapy are in line with the previous trend analysis of the PANSS negative, QLS total and chorus activity scale scores, which suggested that the effects of the therapy might not be durable.
Ward life scale scores of the therapy group subjects over 8 months (n = 30).
DISCUSSION
The main finding of the present study was that the therapy group patients who had received the 15 musical therapy sessions showed a significant advantage in terms of improvement of the negative symptoms and a QOL measure: some signs of activation of personal relations, and increased subjective sense of participation in a chorus activity. The findings of the present study were almost consistent with those of previous studies conducted to verify the effectiveness of some musical therapies. The randomized controlled trial of Tang et al. demonstrated that their 19 group chorus sessions held in 1 month produced rehabilitative effects: diminishing negative symptoms and some interpersonal aspects of social disability.7 By comparing 12 Karaoke group sessions with simple singing sessions conducted for 6 weeks, Leung et al.8 found that active participation in Karaoke musical performance had effects of arousing social interactions. The study of Pfeiffer et al. did not show clear findings most likely because its sample size was small (n = 7).6 The former two studies and the present study showed that the group musical therapy had personal–relation activating effects. It has repeatedly been stressed that a potentiality of musical therapy consists of its distinctive communication establishing and preserving function.4,14,15 Especially in a group musical therapy setting, it would be no surprise that the effects of this sort were commonly seen.
The second domain where the effect was hypothesized to emerge was that of the subjective musical experience. The effects in this domain might contribute to subjective quality of life or psychological well-being. Since patients highly appraise musical therapy in general,6,16 we may well expect that the effects on subjective experiences could be seen. Therefore, it would be of some relevance that the present study identified an increased subjective sense of participation in a chorus activity as an effect of the musical therapy. However, the area covered by the assessment of the present study was only small. There must be many other areas of subjective experience where musical therapy would produce effects. In particular, esthetic or sensuous areas must take some part in its effects. Studies have shown a relaxing effect of music on schizophrenic patients17 and their tendency of perceiving music as attractive.18 The assessment methods used in those studies may be applicable for assessing the effects of musical therapies. It is necessary to seek and test new methods for assessing the effects of musical therapy.
Another question postulated prior to the present study was whether the improvement achieved by the musical therapy was durable. This study suggested that the improvement might be short-lived as shown by the decline at the follow-up assessment of the measures that had improved during the therapy. That finding suggests that the interpretation that the group musical therapy substantially reduced negative symptoms and impoverishment of social interaction might be only presumptive. An alternative explanation would be that the amelioration was only superficial, and hardly influenced the core pathology of the symptoms. For the purpose of maintaining the improvement, it is necessary to introduce a longer or more intensive musical therapy course, or to combine interventions of other modalities with the therapy.
The weaknesses of the present study need to be mentioned. Random allocation of subjects was not applied. Consequently, there was a confounding difference in some measures at baseline between the subject groups that necessitated the correction in the statistical models to detect the effect of the therapy. Second, the assessors were not blind to the treatment condition of the subjects. Third, there was no dummy treatment for the waiting group. Therefore, the effect of simply having meetings could not be distinguished from the advantage of the therapy group found in the present study. In addition, the subjects of this study were not representative of the common inpatients with chronic psychoses. Future research needs to adopt a design to overcome these drawbacks.
Despite the weaknesses of the present study, it is suggested that the results remain informative. The present findings indicated that the assessment of psychiatric raters, that of ward nursing staff, and patient subjective responses gave simultaneously positive findings. In addition, the quality of the psychiatrist ratings was favorably controlled by means of conducting an interrater reliability study or attaining consensus ratings during the study period. Moreover, even if the advantage could be attributable to simply having meetings and not specific to the therapy, it must be due to the potential of the musical therapy to realize the cohesive group experience of a high attendance rate. Further research is warranted to scrutinize characteristics of the effects that musical therapy would produce on patients with chronic psychoses.
ACKNOWLEDGEMENTS
We are grateful for the cooperation of Dr Hajime Kazamatsuri, Ms Tamiko Aoki, Ms Ichiko Shinohara, Mr Haruo Uesugi, and other treatment and administrative staff at Tokyo Metropolitan Matsuzawa Hospital.
