Data collection

Each region appointed a coordinator, who organised and supervised data collection. Information about the number and location of all public and private NHRFs was obtained from the 21 regional ministries of health and then from each of the 224 departments of mental health throughout the country. The manager of each NHRF completed a structured interview, drawn in part from the Royal College of Psychiatrists' Mental Health Residential Care Study (Reference Lelliott, Audini and KnappLelliott et al, 1996). The information was checked by the regional coordinators, who conducted further interviews when necessary.

Statistical analysis

Analysis focused on descriptive statistics. In addition, multiple logistic regression was used to identify variables associated with the probability of discharge (Reference Breslow and DayBreslow & Day, 1980). The dependent variable was the presence or absence of discharges from each NHRF during 1999; facilities opened from 1999 onwards were excluded.

Poisson regression was used to analyse the relationship between the rate of residential beds in each region (number per 10 000 inhabitants), the availability of other types of services, and two basic socio-economic indicators: number of unemployed per region, in millions of people, and overall regional income, in millions of euros (Reference Breslow and DayBreslow & Day, 1987; Reference Zheng and AgrestiZheng & Agresti, 2000). Figures used for this analysis were based on 1998 government statistics (Reference de Girolamo and Cozzade Girolamo & Cozza, 2000), while the number of beds in private in-patient facilities was obtained from the national Ministry of Health; the socio-economic data are the official data for the corresponding year. The very small Valle d'Aosta region was combined with the neighbouring Piedmont region. All analyses were performed using SPSS, version 9.0 for Windows.

Staffing

The 1370 NHRFs employed 18 666 professionals, of whom 60% (n=11 240) worked full-time (Table 1). The distribution of professional resources was variable. For example, 285 NHRFs (21%) had no nurses, and almost half had no full-time nurses. Most (57.8%) were directed by psychiatrists, while 22% had a coordinator who did not have mental health qualifications. Around 40% of staff had no specific professional qualification for working with patients with severe psychiatric conditions. The mean number of staff per NHRF was 13.7 (range 6.9-21.0); the mean number of full-time staff was 8.2 (range 0.6-15.0). The overall ratio of patients to full-time staff was 1.42:1 (range 0.82-22.3).

Patients

At the census there were 15 943 residents in the 1370 NHRFs, a mean of 11.6 per facility (Table 2). In 18% of the NHRFs there were only patients discharged from former mental hospitals, while the largest group of NHRFs (43%) housed only patients who had never been admitted to a mental hospital. The majority of NHRF residents (58.5%) had never before been admitted to a mental hospital or a forensic mental hospital, almost 40% had been admitted at least once to a mental hospital and 1.6% had been detained in a forensic mental hospital. Most residents (82.7%) had no current problems of alcohol or substance misuse. ‘Mental retardation’ was the primary problem of around 10% of residents; half of these had come from mental hospitals.

Table 2 Characteristics of residents in non-hospital residential facilities (n=15 943)

	n	%
Former places of residence
Patients never admitted to former mental hospital or forensic mental hospital	9319	58.5
Former mental hospital residents	6371	39.9
Former forensic mental hospital patients	253	1.6
Clinical categories of residents
No current problems of alcohol or substance misuse	13177	82.7
Comorbid alcohol or substance misuse problems	805	5.1
Primary alcohol or substance misuse problems	55	0.3
Primary ‘mental retardation’	1593	10.0
Other problems	313	1.9
Most numerous age group in each facility1,2
< 40 years	434	31.8
40-59 years	681	49.8
> 60 years	252	18.4
Admissions in 19911,3
None	334	24.4
1-2	360	26.3
> 2	670	49.3
Discharges in 19991,4
None	513	37.7
1-2	429	31.5
> 2	420	30.8

Turnover of residents

Turnover of residents was low. During 1999, more than a third of NHRFs (37.7%) had not discharged any patients and 31.5% had discharged only one or two (Table 2). Consequently, few new admissions were possible: 24.4% had not admitted any new residents in 1999, and 26.3% had admitted only one or two patients. The results of the multiple logistic regression analysis for 1091 NHRFs are shown in Table 3. Of the variables included in this analysis, intensity of care and type of management were not significantly associated with low turnover. The variables significantly associated with lack of discharges were the most numerous age group and the number of residents, which also showed a significant interaction. Non-discharge was inversely related to the number of residents and directly related to the mean age of residents. A first-order interaction between these variables revealed that in NHRFs in which most patients were from the oldest age group (>60 years), the probability of non-discharge was higher than in the other two age groups and was not significantly associated with the number of residents. However, in NHRFs with patients mainly from the other two age groups (<40 years and 40-59 years), non-discharge was related to having more patients.

Table 3 Logistic regression: variables associated with no turnover of residents in non-hospital residential facilities in 1999 (n=1091)

Variable		Estimate	s.e.	Z	P
Intercept		-0.674	0.255	-2.639	0.008
Former mental hospital patients	All v. not all	0.828	0.195	4.241	<0.001
Presence of full-time staff	No v. yes	0.489	0.176	2.783	0.005
Number of residents	6-10 v. 1-5 (p2)	-0.843	0.325	-2.596	0.009
	11-15 v. 1-5 (p3)	-1.915	0.529	-3.620	<0.001
	> 15 v. 1-5 (p4)	-2.873	0.737	-3.900	0.000
Most numerous age group	40-59 years v. < 40 years (e2)	0.221	0.319	0.692	0.489
	> 59 years v. < 40 years (e3)	0.287	0.413	0.695	0.487
Interaction between number of patients and most numerous age group	p2.e2	0.748	0.412	1.816	0.069
	p2.e3	1.115	0.522	2.138	0.032
	p3.e2	1.316	0.598	2.199	0.028
	p3.e3	1.903	0.719	2.648	0.008
	p4.e2	1.102	0.802	1.373	0.170
	p4.e3	2.728	0.851	3.207	<0.001

The odds for non-discharge in NHRFs hosting only former mental hospital patients was double that of the other NHRFs (95% CI 1.56-3.35) while NHRFs without full-time staff had 1.6 times the risk of non-discharge (95% CI 1.16-2.30) compared with the other NHRFs.

NHRFs and other mental health facilities

Table 4 shows the relationship between availability of residential places and provision of other mental health services in each region controlling for two basic socio-economic indicators (number of unemployed and overall regional income). A high rate of residential beds was associated with a higher rate of private in-patient beds, day hospitals and general hospital wards, and fewer community mental health centres and day centres, but the effect of former mental hospitals was not significant. More residential beds were found in regions with a lower number of unemployed people, but the effect of regional income was not significant (at the 5% level). The impact of these variables on bed numbers has been quantified using relative risk (RR). For example, with an increase of 1000 private in-patient beds, the rate of residential beds increases by a factor of 1.7 (and similar considerations apply for other variables with RR >1); with an increase of 100 in the number of community mental health centres, the rate of residential beds shows a decrease of 2.6 times (1/0.385, which is the RR for this variable; similar considerations apply for other variables with a RR <1). The regression equation predicted between-region variation in number of residential beds both in absolute and relative terms; the linear correlation coefficient between predicted and actual residential beds was 0.93, while the correlation between predicted and current occupancy rate per 10 000 inhabitants was 0.80.

Table 4 Poisson regression: predictors of the number of residential beds in non-hospital residential facilities

Variable	Estimate	s.e.	Z	P	RR	95% CI
Intercept	-7.680	0.020
Community mental health centres (× 100)	-0.955	0.148	-6.47	<0.001	0.385	0.288-0.514
Day centres (× 100)	-1.127	0.087	-12.97	<0.001	0.324	0.273-0.384
General hospital psychiatric wards (× 100)	0.735	0.194	3.79	<0.001	2.086	1.426-3.050
Beds in private in-patient facilities (× 1000)	0.539	0.025	21.18	<0.001	1.715	1.631-1.803
Day hospitals (× 100)	0.394	0.095	4.16	<0.001	1.483	1.232-1.786
Number of unemployed (× 106)	-0.303	0.031	-9.73	<0.001	0.739	0.695-0.785
Regional income ([UNK] million)	0.051	0.028	1.80	0.071	1.052	0.996-1.112

Rehabilitation centres or ‘homes for life’?

Our results suggest that many NHRFs merely provide long-term accommodation. Lack of mobility was inversely related to the exclusive presence of former mental hospital patients and to the number of residents, while it was directly related to the age of residents. Only in NHRFs providing for older patients was low turnover unrelated to the number of patients. Since the absence of full-time staff (which suggests low intensity of care) was also related to a low turnover, it seems that NHRFs hosting more patients, predominantly elderly, and having no full-time staff to provide intensive rehabilitation, have few discharges.

Similar low turnover rates have been observed in the USA (Reference Geller and FisherGeller & Fisher, 1993) and in British studies of patients discharged from mental hospitals (Reference Trieman, Smith and KendalTrieman et al, 1998). In the Team for the Assessment of Psychiatric Services (TAPS) study, 61% of the patients discharged to community placements (mostly group homes) remained in the same accommodation over the 5-year follow-up (Reference Trieman, Smith and KendalTrieman et al, 1998). Despite a tendency for long-term stay in large NHRFs, it should be stressed that several studies have shown that living in small, domestic environments is associated with better quality of life and higher satisfaction reported by patients, compared with traditional mental health wards (Reference Lehman, Possidente and HawkerLehman et al, 1986; Reference Barry and CrosbyBarry & Crosby, 1996).

Staffing: quantity and quality

Three-quarters of the NHRFs provided 24-hour care, representing a real alternative to long-stay hospital wards. The ratio of patients to full-time staff of 1.42:1 gave further support to the intensive care notion of residential services in Italy. However, about 40% of staff had no specific professional qualification for working with people with severe psychiatric illness. Similar lack of trained staff caring for patients in NHRFs has been observed in the USA and the UK (Reference Randolph, Ridgway and CarlingRandolph et al, 1991; Reference Senn, Kendal and TriemanSenn et al, 1997). Efforts should be made to train mental health workers to treat residential patients.

In-patient psychiatric care today in Italy

In May 2000 there were 27 649 psychiatric beds in Italy (including hospital and residential places). Since 1978, when Law 180 was enacted, there has been a 65% reduction in the provision of residential beds of all kinds. This change in service provision seems to have occurred without increases in suicide rates (Reference Williams, De Salvia and TansellaWilliams et al, 1986; Reference De Leo, Conforti and CarolloDe Leo et al, 1997) or in the number of patients admitted to psychiatric forensic facilities (Reference Fioritti, Melega and FerrianiFioritti et al, 2001). Although there is no detailed information, numbers of homeless people with mental illness do not appear to have increased in urban areas (Reference de Girolamo and Cozzade Girolamo & Cozza, 2000). It is also unlikely that community-based services have reduced the prevalence of serious mental disorders, as implementation of evidence-based treatment strategies has been delayed as in other countries (Falloon et al, 1999). Support for families has been provided unevenly, even though a large proportion of people with mental illness in Italy live with their families (Reference Warner, de Girolamo and BelelliWarner et al, 1998; Reference Magliano, Fadden and EconomouMagliano et al, 2000).

Regions with better provision of outpatient and day care services had lower rates of residential beds. This suggests that providing more community services might reduce the need for residential care. However, it is also possible that the availability of more residential beds decreases the need for community-oriented services, or that a third factor has an influence on overall service provision. In any case, at a regional level, service planning has focused on developing either a caring supportive approach based on long-term residential care, or a more community-based rehabilitative treatment approach.

Limitations of the study

This study has some limitations. First, the large number of researchers collecting data might have created inconsistencies. Second, the cross-sectional design of the study does not allow causal inferences, for instance regarding the relationship between provision of NHRFs and of other types of psychiatric services.

Despite such limitations, PROGRES is the first systematic attempt in Italy to fill the gap between psychiatric services planning and evaluation, by setting up a network of investigators throughout the country and evaluating an entire typology of services in a consistent fashion.

Members of the PROGRES team

National coordinators: Dr G. de Girolamo, Dr A. Picardi, Dr P. Morosini (National Mental Health Project, National Institute of Health, Rome); biostatistician: Professor R. Micciolo (University of Trento); regional coordinators: Dr P. Argentino, Professor M. Casacchia, Dr P. Ciliberti, Dr G. Civenti, Dr A. Colotto, Dr G. Dell'Acqua, Dr P. Di lorio, Dr W. Di Munzio, Dr G. Fagnano, Dr A. Fioritti, Dr N. Longhin, Dr M. Miceli, Dr M. Nicotera, Dr M. Pisetta, Dr R. Putzolu, Dr E. Rossi, Dr M. E. Rotunno, Professor G. Borsetti, Dr D. Semisa, Dr R Tomasi, Dr P. Tulli, Dr E. Zanalda. Researchers: Dr C. Barbini, Dr F. Basile, Dr G. Bazzacco, Dr R. Bracco, Dr A. Calvarese, Dr G. Canuso, Dr E. Caroppo, Dr L. Caserta, Dr M. Colangione, Dr S. Damiani, Dr T. De Donatis, Dr F. Di Donato, Dr V. Di Michele, Dr R. Esposito, Dr M. Facchini, Dr S. Ferraro, Dr P. Fracchiolla, Dr P. Gabriele, Dr D. Gallicchio, Dr G. Giardina, Dr A. Greco, Dr F. Grilletti, Dr S. Guzzo, Dr A. M. Lerario, Dr M. R. Marinelli, Dr C. Marino, Dr E. Monzani, Dr F. Picoco, Dr L. Pinciaroli, Dr C. A. Rossetti, Dr P. Rubatta, Dr G. Santone, Dr F. Scorpiniti, Dr V. Scrofani, Dr G. Soluri, Dr M. Stefani, Dr A. Svettini, Dr A. Vannini, Dr A. Zaffarano. Scientific consultants: Dr F. Amaddeo, Dr P. Bonizzato, Dr B. D'Avanzo, Dr G. Ditta, Professor I. Falloon.