Abstract
The psychiatric reform implemented in Italy since the late 1970s has led to a community-based care system characterized by closure of psychiatric hospitals, establishment of general hospital acute psychiatric units with limited numbers of beds, community mental health centres and clinics, day-hospitals and day-centres, and community residential facilities, all supposed to work according to principles of integration and continuity of care, and organized within the Department of Mental Health in each Health District [1], [2].
National standards were defined for the various components of community mental health services [2]. Whereas for some types of services the actual rates were lower than the standards (the actual number of general hospital acute in-patient beds was 0.7/100 000 versus the standard of 1/100 000, and day-hospital beds 0.1/10 000 versus 1/10 000), the opposite was true for community mental health centres and clinics (1.8/150 000 versus 1/150 000), day-centres (1.3/150 000 versus 1/150 000), and community residential beds (3/10 000 versus 1–2/ 10 000), which were implemented at rates higher than the standard ones. Community residential facilities doubled between 1997 and 2000 [3], either to host patients discharged from psychiatric hospitals (finally closed in 1999), or to respond to the therapeutic and rehabilitative needs of patients treated in the community.
Community residential facilities for psychiatric patients are intended for short, medium and long-term stays of non-acute patients, according to a continuity of care model in which all community mental health services and agencies contribute to the community care of the mentally ill [2]. In them, patients are meant to stay only a limited time, so as to avoid an ‘institutionalization’ effect, and to have beds available for other patients needing care. According to the National Mental Health Plan [4], housing needs should not be a reason for admission to community residential facilities. Notwithstanding these indications, the function of community residential facilities is not clear, and the issue is a matter of debate among the professionals of the mental health services.
Despite its importance, there are few studies on the use of community residential facilities, the patients they host, the interventions they deliver and the rate at which they discharge patients [3],[5–8]. These studies were mainly concerned with the description of the existing provision of non-hospital residential settings in the mental health community services of various health systems, with regard to types of facilities, level of staffing, staff qualifications, services and interventions offered and patients’ characteristics in the US [5], England [6–8] and Italy [3].
The overall picture of the residential facilities appeared very heterogeneous and definitions were not consistent across studies and health systems. Some studies raised the problem of scant available provisions to the most severely disable patients [6–9]; all observed long durations of stay, and some a limited turn-over [3], [8], [9]. In the study by Trieman and Leff [7], [8] it was found that, in 5 years, 39% of patients had left the specialized facilities for sheltered accommodations in the community and that these patients were significantly more skilled than those who were not discharged, although they showed similar measures of mental state. A nationwide Italian survey of psychiatric community residential facilities gave the numbers of such facilities in all Italian regions and their main characteristics; 38% of the facilities in Italy had discharged no patient in the past year, and 32% no more than two patients [3].
To further investigate this issue, we analyzed data from a longitudinal study conducted in 2000–2001 on all the community residential facilities in Lombardy, which is the most populated region in Italy, with nine million inhabitants. This study served to describe the population in 179 community residential facilities at the end of 2000 and addressed questions on the use and function of facilities with different intensity of care. We took discharge from any of the community residential facilities surveyed as the outcome variable. This is of particular interest for various related reasons: (i) assuming that treatment in those facilities improves patients’ clinical conditions, discharge to lower-protection settings should be the expected outcome; (ii) discharge always represents a change in the patient or in the treatment delivered; and (iii) it is necessary to know which factors facilitate turnover of patients and to make a more extensive use of this resource.
Method
The study population was drawn from all the patients hosted in the community residential facilities identified in Lombardy in the framework of a national survey, the PROGRES Project [3], which included the great majority of Italian regions. Between February and April 2000, 196 facilities (for a total of 2117 beds) satisfying the accreditation criteria defined by the Regional Health Authority of Lombardy were identified and described according to various indicators. In October 2000, the professionals working in all the facilities identified were requested to participate in a prospective study aiming at describing the population hosted and register patients discharged or who had died in the period between recruitment and the end of 2001. Of the 196 community residential facilities identified in Lombardy, 91% agreed to participate. Among the 17 facilities which did not participate, four facilities did not host patients at the index day and the others refused to participate because of practical difficulties due to shortage of staff. The characteristics of these facilities (public or private management, intensity of care, number of patients) did not differ from those of the whole group.
The community residential facilities were grouped into four categories (Table 1), according to accreditation criteria defined by the Regional Health Authority, and were classified as: residential care centres (24-hour staffed facilities, often for post-acute, relatively short admissions); high-staffed facilities for longer stays and for severe long-term conditions (24-hour); mid-staffed (12-hour) facilities for relatively independent patients who can stay alone overnight; and low-staffed facilities (≤ 8-hour) for more independent patients (the so-called group homes were included in mid- or low-staffed facilities, and supported flats in the low-staffed ones). In low-staffed facilities, care could be delivered as needed and sometimes quite infrequently. Residential care centres and high-staffed facilities had to have not more than 20 beds, whereas mid- and low-staffed facilities not more than eight. In the facilities where the study was conducted, the mean number of patients was 11. The high-staffed facilities and the residential care centres, although both 24-hour staffed, are intended to have quite different functions. Whereas residential care centres are used for admissions of post-acute patients, high-staffed facilities are intended for longer admissions of chronic patients and working style is supposed to be different.
Type of community residential facilities in Lombardy
Professionals of the participating facilities completed a structured questionnaire, developed specifically for this study, describing all the patients hosted in the facilities on 15 November 2000. This made up the study sample. If a patient was not physically present on that date, because s/he had been admitted to a general hospital for psychiatric or medical/surgical treatment, or was on holiday, but the bed was still kept, that person was included in the study sample. Patients who used the community residential facilities only for day-care were not included. A total of 1792 patients were recruited and described.
The questionnaire collected information about: socio-demographic characteristics; primary and secondary diagnosis according to ICDX [9] and illness severity according to the Health of the Nation Outcome Scale (HoNOS) [10–12]; duration of the admission until 15 November 2000; housing and family conditions; employment; lifetime suicide attempts; admissions to psychiatric acute units and other psychiatric institutions or prison; adequacy of the current accommodation in the staff's opinion; physical health; psycho-pharmacological therapy prescribed on the index day; autonomy according to several indicators; and psychosocial interventions delivered to the patients during the current admission. Among these, daily life activity interventions included measures aimed at improving skills such as washing and dressing themselves, cleaning, cooking, shopping, communication and other social skills. Social and recreational activities included going out together, practising sports, going for dinner, organizing and going to parties and more structured activities, such as gardening or other semivocational activities. Psychological interventions included individual therapy, group therapy, self-help groups, family therapy, psycho-education sessions, social skills training groups and meetings between relatives and professionals. All these activities were investigated in the month before recruitment and were considered in the analyses only if offered at least once a week.
The follow-up finished on the date of death or discharge or on 31 December, 2001.
If a patient died, the date and cause of death were recorded; and if discharged, the date, HoNOS score at discharge and accommodation at discharge were entered. Checks were made regularly in all the participating community residential facilities, in order to ascertain completeness of recording of the events of interest.
Accommodations at discharge were: home (either with the family or alone); other community residential facility; nursing home; facility for patients with learning disabilities or mental retardation; forensic mental hospital; prison; and other and unknown accommodation. In this analysis, we focused on two types of discharges: (i) those that represented an improvement in the patient's condition, that is discharges to their own home or to a lower intensity of care facility; and (ii) discharges to higher intensity of care facilities, that is from mid- or low-staffed facilities to high-staffed facilities, residential care centres, nursing homes and facilities for patients with mental retardation, and from low- to mid-staffed facilities. Both the groups of discharge were contrasted with the group of undischarged patients, which included only patients who on 31 December 2001 were still present in the same facility as on the index day (1459 patients).
Among the psychiatric residential facilities where patients were discharged, there were also facilities not included in the study: for instance, facilities open after the beginning of the study, those out of the region or those which did not agree to participate.
Statistical analysis was done in order to assess which factors were associated with discharge of patients. Odds ratios (OR) and corresponding 95% confidence intervals (CI) were computed by unconditional logistic regression models, in which the dependent variable was discharge or not. The models included terms for sex, age, duration of the current admission, placement before current admission and intensity of care of the community residential facilities.
The significance of the OR estimate was represented by the corresponding confidence interval: if it did not include unity, the OR estimate was statistically significant. In order to evaluate a linear trend in the association of risk with the factors considered, a χ2 for trend test (χ2 trend) was performed, and the corresponding p-value was also shown.
Results
There were 1061 men and 731 women, and mean and median age was 47 years, standard deviation (SD) = 14. Diagnosis was schizophrenia or other psychosis for 67% of the patients, mental retardation for 10%, personality disorders for 10%, affective disorders for 7% and other disorders for 6%. A total of 393 people had also a secondary diagnosis: 90, personality disorder; 76, mental retardation; 48, anxiety; 40, some organic mental disorder or dementia; 35, bipolar disorder; 32, psychosis; 30, depression; 23, substance abuse; and 19, eating disorder. Thirty-seven percent of the patients had never worked in their life, the majority were not working before the current admission and 3% had a regular job on the index day. Violent behaviour in the last two weeks was reported as a major problem (scores 3 and 4 of item 1 of the HoNOS) for 6% of the patients; 7% had been in a forensic psychiatric institution or in jail at some time during their life. Risk of suicide according to the HoNOS (in the last 15 days) was reported for only two patients and suicide attempts lifelong were reported for 30%. Forty-five percent of patients reported some health problem; all but 61 patients were on medication on the index day: of these, 1561 took at least one antipsychotic, 325 antidepressants and 1181 benzodiazepines. The proportion of patients who came directly from a psychiatric hospital was 37%. Mean length of the current admission was 730 days (SD = 1108), the median was 444 days and the range was 1–9943 days (excluding a patient who had 12 193 days of admission, as he was there since 1967). For 23% of the patients, staff stated that the current accommodation was inadequate.
Table 2 shows the distribution of the above variables according to different types of residential facilities. Differences were apparent for age, with older people in low- and mid-staffed facilities and younger in high-staffed facilities and residential care centres (p < 0.001); placement before the current admission, with 63% of patients coming from a psychiatric hospital in mid-staffed facilities, and 4% in the residential care centres (p < 0.001); duration of admission, which was shorter in low-staffed facilities and residential care centres (p < 0.001); HoNOS score, higher in high-staffed facilities and residential care centres (p < 0.001); number of types of interventions in daily life activities, more frequent in low and mid-staffed facilities (p < 0.001), and social and recreational activities and psychological interventions, which were more frequent in high-staffed facilities and in particular in residential care centres (p < 0.001).
Characteristics of the four types of community residential facilities and their patients in Lombardy in 2000–2001
The variability in length of stay in the four types of facilities was also expressed as mean, SD and median values in days: in the residential care centres the mean was 631 days, the SD = 773, and the median, 329; in the high-staffed facilities (excluding the outlier value of 12 193 days) the mean was 730, SD = 1103 and median, 529; in the mid-staffed facilities the values were 970, 1529 and 684 days, respectively; and in the low-staffed facilities 771, 1209, and 350 days. In the four types of facilities, the ranges were, respectively: 1–4594 days, 1–9917; 1–7793; and 2–7280 days.
In the period from 15 November 2000 − 31 December 2001, a total of 17 patients died, and 316 patients (18%) left one of the residential care facilities for various different destinations: 172 went to their own home (alone or with relatives); 19 went to a lower intensity of care facility; 49 to a higher intensity of care facility (9 patients) or to a nursing home (36 patients) or facility for people with mental retardation (4 patients); 56 to an equally staffed community residential facility; and 20 to other or unspecified places (among whom only one patient was homeless).
Table 3 shows the distributions of patients discharged to lower and higher intensity of care facilities and undischarged patients, and the corresponding OR for discharge. For discharge to lower intensity of care facilities significant OR estimates emerged for employment at admission, with those not working having half the probability of discharge of those who were working; placement before entering the community residential facility, with 17 times the probability of discharge for those who came from home, and eight times for those coming from other institutions compared to those who came from a psychiatric hospital; duration of current admission, with a strong inverse association; HoNOS total score in the most severe group; type of facility, with patients in low-staffed facilities having a 70% decreased probability to be discharged compared to those in residential care centres. Associations of discharge and psychosocial and psychological interventions were not significant.
Odds ratios (and 95% confidence intervals) of discharge from community residential facilities to lower and higher intensity of care settings in Lombardy in 2000–2001
The probability of discharge to higher intensity of care facilities was significantly associated with: older age, with a nearly fourfold increase for people aged 60 or more compared to those under 40, with a significant trend in risk; opinion of the staff that the current accommodation was not suitable for the patient; being hosted in a private facility, with an 80% decrease of probability to be discharged. No significant association emerged with the interventions delivered during the admission, HoNOS score, violent behaviour and lifelong suicide attempts.
Discussion
In the period considered, 11% of the patients hosted in 179 community residential facilities in Lombardy were discharged to lower-protection settings and 3% to higher-protection ones. Level of intensity of care of the community residential facility of the current admission, placement before the current admission, duration of current admission on the index day, HoNOS score and employment at the time of admission were significantly associated with the probability of discharge from community residential facilities to a lower intensity of care setting (home or lower intensity of care psychiatric residential facilities) in this population.
Old age, inadequate accommodation in staff's opinion, and the public sector managing the facility were associated with probability of discharge to higher intensity of care settings.
Some estimates of risk were nonetheless not significant, possibly because of the small numbers, especially for discharges to higher intensity of care facilities.
Duration of the current admission was a strong determinant of discharge to lower intensity of care and its effect was not explained by other potential risk factors, as suggested by the absence of differences between the crude (not shown) and adjusted estimates. The effect of duration of stay was also true across different types of community residential facilities, and in all types of facility the OR estimates of discharge were lower for patients with longer stays.
The type of facility where the patient was hosted strongly influenced the chance of discharge to a lower intensity of care setting. There were 114 discharges (23%) from the residential care centres, 67 (7%) from high-staffed facilities and five (4%) either from midstaffed and low-staffed facilities.
Being hosted in a residential care centre was associated with higher probability of discharge, but no significant associations were observed with factors that markedly characterized the population of the residential care centres. In fact, age was not a determinant of discharge to lower-protection facilities, but patients were younger, and in a statistically significant way, in residential care centres; patients in residential care centres also had more people closely in touch with them, but this was not a determinant of discharge; patients in residential care centres had higher HoNOS scores, and high HoNOS score was inversely associated with discharge. This suggests that the effect of the type of facility on the probability of discharge was not mediated by the effect of patients’ characteristics, and it represented a determinant of discharge in itself.
Psychiatric hospitals in Italy were all closed in 1999 and in Lombardy about 28% of their patients were discharged to community residential facilities [13]. This population tended to have rather severe clinical features, a mean age of 60 years, very long hospital stays (30 years on average) and a weak social network. It is not surprising therefore, that it was much more difficult to discharge these patients than patients coming from their own home.
The great majority of discharges to higher intensity of care facilities was to nursing homes and the strongest determinant of these discharges was old age: this finding is reassuring, since it suggests more an increased need for geriatric care, than the failure of projects in mid- or low-staffed facilities. These discharges were less frequent from private facilities, and the effect was not explained by the presence of only high-staffed ones in the private sector. It is likely that private facilities had fewer connections with other public services, such as nursing homes and that some of them were more available to provide non-specifically psychiatric care.
The absence of association with other factors should be commented upon. It is remarkable that the frequency of psychosocial and rehabilitative interventions had no material effect on the probability of discharge. However, this finding does not mean that interventions were not useful in improving patients’ performance and abilities, since we did not measure them. It does, however, mean that the probability of discharge was not affected by the number and type of interventions offered. This is consistent with the finding that the type of facility in itself has an effect on probability of discharge and that rather than the interventions delivered in the facilities, it was the a priori attitude toward turnover and discharge of patients that facilitated discharges. This effect also raises the question about adequate training for delivering rehabilitative treatment [6], but perhaps more important is the question how single interventions can usefully be put in the framework of truly individual projects [14].
Probability of discharge to lower-protection settings was only weakly related to public or private sector management of the community residential facilities. It should be recalled that private facilities considered were owned and managed by a private, often non-profit enterprise, but still paid for by the national health system according to agreed rates which are the same for public and private facilities. Our sample did not include therefore strictly private facilities where the patients or their families paid out-of-pocket. For chronic severe patients like those in this study, private residential facilities not financed by the national health system are not widely used in Italy.
Mean and median values of duration of stay in the four types of facilities indicate a large variability of lengths of stay. This may in turn be related to great variations in patients’ characteristics and clinical features in each type of facility. In fact, in all facilities time-limited treatments were implemented for some patients, whereas admission was intended as a long-lasting accommodation for others.
Comparisons with the findings of another Italian investigation [3] are problematic because of some methodological differences: the different population base of that study (whole of Italy); the period when the national survey was conducted (many community residential facilities had just started operating, and this may have limited discharges); and the retrospective nature of the question about discharge. The period of our study may have influenced our findings too, since it is possible that community residential facilities had a higher turnover in the most recent years.
Some weaknesses in this study must be noted. The longitudinal design had an artificial baseline for each patient: this was 15 November 2000 for all patients and not the day of admission of each patient. Information on severity of illness at follow-up was collected only for discharged patients, so no comparison could be made between discharged and undischarged patients. There was no true assessment of those characteristics which could predict improvement through rehabilitation interventions. Nonetheless, it is reasonable to assume that such interventions were delivered in the presence of some scope for improvement and therefore to expect some meaningful changes.
Also, the follow-up did not include information on patients subsequently readmitted to the same or another facility. This was reported by several professionals, but we could not collect detailed information.
Some problems may be related to the definitions of the types of facilities and we realize that it may be hard to have a clear idea of the facilities studied. Nonetheless, the definitions we used have two advantages: they correspond to the shared official definition of the health authority of the Lombardy region; and they are based only on time, staff and number of patients criteria, thus acknowledging that different uses for different patients can be made; definitions given by other authors also included the specific functions, which can be inconsistent and different in different health systems and countries [5–7],[15].
Despite these shortcomings, the study can be useful for the turnover pattern and activity of these facilities, which are among the most meaningful features of the community mental health services in Italy. This was strengthened by almost complete participation. The paucity of studies conducted on this issue adds interest to our work, although it limits comparison with data from other populations.
These findings confirm the great variations of the characteristics of psychiatric residential facilities and the patients they hosted and a limited turnover and suggest some directions for future activity and research. The psychiatric residential facilities represented a continuum of several functions: they were a permanent accommodation for those persons too difficult to be cared for outside a residential setting; they also offered a protected setting for a prolonged period of time with intensive rehabilitation and social integration activities; and, at the other extreme of the continuum, were a short-period rehabilitative and therapeutic setting with specific shortterm aims. Our findings show that long-term care was prevalent. Some reasons for the limited turnover of patients in the psychiatric residential facilities of the community services in Lombardy can be argued. Part of the difficulty may – at least in the areas where there was a psychiatric hospital – be explained by the presence of patients who were in psychiatric hospitals before the current admission and who had lost any tie with the community; part may be explained by the lack of alternative affordable accommodation for the more independent patients. The small number of patients transferred from high-staffed facilities to mid- or lowstaffed ones may be related to the small availability of such lower staffed facilities. Some role may also be played by the culture and working style of the services, which may be more concerned with the outcomes shown by patients within the facility and the good management of the facility itself than with outcomes of real autonomy as represented by discharge. It is also possible that community non-residential services were perceived by professionals as unable to assure the care necessary for patients who might live at home, but needed intense aftercare and rehabilitation.
All these points need to be addressed while discussing the role played by other psychiatric and social services in the community care of their severely mentally ill citizens.
Footnotes
Acknowledgements
This work was supported by the Regione Lombardia.