Involuntary Psychiatric Admissions and Development of Psychiatric Services as an Alternative to Full-Time Hospitalization in France

Increasing the quality of care is the goal of any health system. In psychiatry, the development of alternatives to full-time hospitalization is supported by international recommendations for mental health care (1–5) as a way to increase this quality. Such alternatives encompass full-time care outside inpatient settings as well as part-time hospitalization and ambulatory care. Research has demonstrated that such alternatives are associated with a reduction in hospitalizations (6,7) and improvements in quality of life, clinical outcomes, adherence to treatment, service accessibility, continuity of care, and patients’ satisfaction (6,8–11). In France, few alternatives to full-time hospitalization have been developed (12,13), compared with other European countries (14–17). A recent assessment showed that development has been impeded by resistance from health professionals (18), despite support from policy makers (19). Resistance is attributed to a lack of consensus among various schools of thought in the mental health field regarding the benefits of alternatives to full-time hospitalization (18).

Therefore, additional evidence of the benefits of alternatives to full-time hospitalization is needed (20). In particular, little substantial research has examined the impact of deinstitutionalization and the development of such alternatives on involuntary treatment (21), which is still used in psychiatry worldwide (22,23), despite its controversial nature and the implementation of reforms to reduce its frequency (21,22). Involuntary treatment is often considered an indicator of the quality of prior care (24–27). Indeed, if patients receive adequate care, they are less likely to experience a crisis and to require compulsory admission, and patients’ consent to treatment has been promoted as a key component of the quality of the care (28,29).

Moreover, difficulties in applying the results of mental health services research that originates from other contexts have been underscored (7). The transferability of results from other countries is limited by the particularities of the French public psychiatric system, which is territorially organized into geodemographic areas in which multidisciplinary teams hosted by hospitals (sectors) coordinate the delivery of outpatient and inpatient services necessary to cover the mental health needs of their population (30,31). Between-country comparisons also require caution because of the specificity of involuntary admissions, which are regulated by different legislative frameworks worldwide (32).

It is thus necessary to assess whether the development of alternatives to full-time hospitalization is related to a decrease in involuntary admissions in French public psychiatry. However, several studies have shown that a number of variables, unrelated to the development of alternatives, could also be associated with involuntary admission. Such variables include patient, health care provider, and environmental characteristics (33–37).

In this context, the objective of our study was to assess whether the development of alternatives to full-time hospitalization provided by French psychiatric sectors was associated with a reduction in involuntary full-time admissions, taking into account other factors that may be associated with involuntary care.

Methods

Scope of the Study

French psychiatric sectors represent the cornerstone of the organization of public mental health care delivery, and they account for nearly 70% of the costs of mental health care in France (38). They are hosted by public and private nonprofit hospitals, which are either general hospitals with an activity in psychiatry or psychiatric hospitals. On average, a hospital hosts four sectors of adult psychiatry (39), which were initially created to serve a population of around 70,000 inhabitants, with some variation in population size and overall characteristics (18,31). Only sectors hosted by hospitals specifically mandated by regional health agencies can provide involuntary care. Given these factors and to ensure comparability, our study was carried out in psychiatric sectors of public and private nonprofit hospitals that provide involuntary care in mainland France and that treat adult patients.

We conducted a retrospective study for the year 2012 using administrative databases and included only psychiatric sectors hosted by hospitals for which the data were of good quality to enable analysis. Consequently, we excluded sectors hosted by hospitals that did not consistently report full-time inpatient activity or the number of psychiatric sectors in both the French national psychiatric discharge database (RIM-P) and the annual national survey on health care providers (SAE). In addition, we excluded sectors hosted by hospitals that did not report the data necessary to assess the development of alternatives to full-time hospitalization.

Population

We included patients with a diagnosis of a mental disorder from ICD-10 chapter 5 (40). We excluded patients with organic mental disorders, mental retardation, and disorders of psychological development (apart from pervasive developmental disorders) because of the specificity of the care they require (41,42).

Variables and Data Sources

Our variable of interest—whether a patient had at least one involuntary full-time admission in a psychiatric sector in 2012—was extracted from the RIM-P database. There is no direct measure of the development of alternatives to full-time hospitalization in psychiatric sectors. One way to assess this development is to determine the share of human resources allocated to such alternatives compared with the total human resources allocated to psychiatry in the hospital hosting each sector. Human resources indeed represent 70% of the hospital budget for the treatment of somatic illnesses (43,44), and this percentage is estimated to be even higher for psychiatric care (45–47). The development of alternatives was measured as the ratio of the number of full-time equivalents (FTEs) working in departments that provide alternatives to full-time hospitalizations to the total number of FTEs allocated to psychiatry in the hospital hosting each sector. These data were extracted from the SAE database. Considering the overall proportion of FTEs allocated to alternatives to full-time hospitalization allowed comparability of data across sectors by adjusting for their overall capacity.

Adjustment factors were also considered. Because medical practice is a complex decision-making process, we used a conceptual framework that grouped potential factors associated with practice into three categories: patient characteristics (the demand side), health care provider characteristics (the supply side), and characteristics of the practice context (the environment) (48–51). [A diagram of the framework is included in an online supplement to this article.]

We first considered factors that are expected to influence involuntary admission through mechanisms similar to alternatives to full-time hospitalization by directly affecting a patient’s health status. Diagnoses, which were classified into broader diagnostic groups (Table 1), were extracted from the RIM-P database. Because the case-mix characteristics of sectors can influence their practice, the repartition of diagnostic groups by sector was also considered. Variables related to the supply of additional medical and social care, which are likely to increase continuity of care in ways similar to alternatives to full-time hospitalization, included care provided by the private sector or through social care institutions. Information regarding the availability of these sources of care was extracted from administrative databases (52), and their density was calculated for the catchment area of each sector. These catchment areas, defined as the geographic zone from which patients of psychiatric sectors originate, were built for each sector by extracting patients’ zip codes from the RIM-P and by using a geographic information system (Geoconcept software).

Second, additional adjustment factors were considered in the analysis. They included demographic characteristics extracted from the RIM-P database. To overcome the lack of data on patients’ socioeconomic characteristics, we calculated a proxy index of deprivation based on patients’ residential zip codes by using a validated composite index specifically developed for the French context, the FDep (53–55). Institutional and organizational characteristics of sectors were extracted from the SAE database. Finally, we included characteristics of the overall health status of the population (56–58) as well as the level of urbanization, assessed by the density of inhabitants in the zip codes of the sectors’ catchment areas (59).

We obtained the authorization to access the databases used for this research from the French data protection authority (Decision DE-2013–077). No informed consent was required from patients because data were entirely anonymized.

Analysis

Descriptive analysis.

Characteristics of the study population were described either by the mean and standard deviation or by number and percentage. We computed the involuntary full-time admission rate per 1,000 patients by dividing the number of patients admitted involuntarily to full-time hospitalization at least once during the study period (2012) by the total number of patients in each sector. To describe variations in this rate across sectors and in the development of alternatives to full-time hospitalization across hospitals, we computed the mean, SD, median, interquartile range, range, coefficient of variation (60), and the ratio of the 90th to the 10th percentiles of the distribution (61,62).

We then assessed the association between the development of alternatives and rates of involuntary admission to full-time hospitalization by sector through the calculation of the Spearman correlation coefficient.

Multivariate analysis.

To assess the association between the development of alternatives to full-time hospitalization and involuntary admission after adjustment for other potentially associated factors, we carried out a multivariate analysis by using a logistic model in which the binary dependent variable was equal to 1 for patients admitted involuntarily at least once and equal to 0 for other patients. To account for the nested structure of the data (63), we ran a multilevel model. Because the mean number of adult psychiatric sectors per hospital was low (64–68), we considered only two levels: the patient (level 1) and the sector (level 2).

The development of alternatives to full-time hospitalization was introduced in the model as an explanatory variable, along with associated patient, sector, and environmental characteristics for which there were strong hypotheses in regard to their association with involuntary admissions. When explanatory variables were highly correlated or associated, only one of them was kept in the model. Because strong hypotheses about interactions between explanatory variables were lacking in the literature, we did not introduce interaction terms in the final model.

To confirm the existence of a random effect at the sector level, we first ran a null model without any explanatory variables (model 1). Second, we introduced patients’ characteristics in the model (model 2). Third, we added the variables calculated at the sector level (characteristics of the sectors and their environment) (model 3).

The analysis was performed with SAS software, version 9. The acceptable type I error rate was set at .05.

Results

Scope of the Study

A total of 229 hospitals meeting our inclusion criteria were reported in the RIM-P in 2012. Among those hospitals, 117 (51%) were included in the analysis on the basis of data quality. These hospitals were divided into 399 sectors of adult psychiatry with involuntary care activity [see online supplement], which accounted for 51% of all sectors of adult psychiatry in mainland France providing involuntary care. Included and excluded hospitals did not differ significantly in terms of main organizational and institutional characteristics or case mix.

Descriptive Analysis

A total of 473,587 patients matching our diagnostic criteria were treated in the selected sectors, representing 56.5% of all patients within the scope of our study seen in adult psychiatric sectors providing involuntary care. The mean age of patients was 47.4, and 53% were women. The two most common diagnoses were mood disorders, not including bipolar disorders, and anxiety disorders (Table 1).

A total of 105,059 (22.2%) had at least one full-time hospitalization, and 31,062 (6.6%) had at least one involuntary full-time hospitalization. Most patients admitted were men (N=18,595, 60%), and the most frequent diagnosis among the involuntarily admitted patients was schizophrenia (N=9,267, 30%) or other psychotic disorders (N=7,771, 25%).

Considerable variations between psychiatric sectors were observed, both for involuntary full-time admissions and the development of alternatives to full-time hospitalization. The mean involuntary admission rate was 70.1 per 1,000 patients per sector (Table 2) and varied between .4 and 366.1. The ratio between the 90th and the 10th percentiles of the distribution was close to 9. The ratio of FTEs allocated to alternatives to full-time hospitalization to the total number of FTEs by hospital had a mean value of .33, and the coefficient of variation reached 32.5%. This ratio was not significantly associated with any institutional or organizational characteristics of the hospital hosting each sector.

In the bivariate analysis, we found a decrease in the rate of involuntary full-time admissions per 1,000 patients per sector when the level of development of alternatives increased; however, it was not statistically significant.

Multivariate Analysis

In the multivariate analysis, we introduced ten individual patient characteristics at level 1. Ten characteristics of psychiatric sectors and 14 characteristics of the environment were introduced at level 2 (Table 3). Results of the null model showed statistically significant variations between sectors (p<.001), which confirmed the need to take into account the hierarchical structure of the data and to run a random-intercept model (Table 4). Twenty-eight percent of the total variation in involuntary admission rates was related to practice differences between sectors (intersector variations), and 72% resulted from differences within sectors linked to case mix (intrasector variations).

Patients’ individual characteristics explained 7% of the variations between sectors, and sector characteristics explained 23%. The level of development of alternatives to full-time hospitalization was significantly and negatively associated with involuntary admission (odds ratio=.99, p=.007) (Table 5). For each 10% increase in the level of development of alternatives, the probability of a patient’s being involuntarily admitted to full-time hospitalization decreased by 12%.

Other factors were also significantly associated with involuntary admission, in particular patients’ age, sex, and diagnostic group; the specialization in psychiatry and the number of inpatient beds of the hospital hosting each sector; and the mortality rate, the number of community-based private psychiatrists, and the capacity of housing institutions for disabled individuals per 1,000 inhabitants of a sector’s catchment area (Table 5).

Discussion

Significant variations were observed in involuntary full-time admission rates between psychiatric sectors in France, even though they are governed by the same legislation regarding this type of care. A large part of the variation was explained by characteristics of the sectors. Our results showed a significant negative association between the development of alternatives to full-time hospitalization and involuntary admissions to full-time hospitalization after adjustment for other factors associated with involuntary care. For each 10% increase in the level of development of alternatives, the probability that a patient would be involuntarily admitted decreased by 12%.

Our findings are consistent with previous work that has shown large variations between psychiatric services in involuntary admission rates (25,33,36). In particular, a study in Northern Europe showed a 13-fold difference in compulsory admission rates between psychiatric departments (36), a figure in the same order of magnitude as our results. Prior research on factors associated with such variations, albeit limited, has also shown results similar to ours; studies found that the psychiatric department where the patient was seen was one of the most important predictors of involuntary admission and that patients’ characteristics accounted for only a limited part of the variations (33,36). To our knowledge, no study has specifically focused on the association between involuntary admissions and the development of alternatives to full-time hospitalization.

Our findings suggest that the development of alternatives to full-time hospitalization provided by psychiatric sectors improved the quality of care. It was indeed negatively associated with involuntary admissions whose reductions align with international recommendations for mental health care (29). Our main hypothesis in regard to the underlying mechanisms is that alternatives to full-time hospitalization can facilitate continuity of care by providing more therapeutic options and can limit the negative consequences associated with inpatient care, such as loss of autonomy and lack of socialization. The presence of alternatives has the potential to lead to fewer crisis situations and to guarantee that patients in need of treatment can provide consent for it.

Nevertheless, some variables that can affect involuntary admissions through similar mechanisms, such as the availability of community-based psychiatrists in the catchment area, appeared to have a stronger effect than the alternatives to full-time hospitalization provided by the psychiatric sectors. Moreover, a portion of the variation remained unexplained, suggesting a greater need for implementation of evidence-based practices to avert crisis situations and for standardized legislation on involuntary admissions. The unexplained variation may also partly result from sectors hosted by hospitals that are not mandated to provide involuntary care and whose practice affects neighboring sectors; however, such effects are likely to be limited because they represent only 7.7% of hospitals in public psychiatry. Finally, since 2012, involuntary care can be provided in France by other means than full-time hospitalization; however, this change was made primarily to allow patients with long involuntary admissions to be treated outside of full-time hospitalization.

Our findings should be interpreted in light of several limitations. First, no conclusions about a causal relationship between alternatives to full-time hospitalization and involuntary full-time admissions can be made because we did not conduct a longitudinal study. Second, the study data were extracted from administrative databases and may be less precise than prospective data (69), and data on some characteristics, such as patients’ symptoms and their severity and patients’ relationship with their family, were not available. Furthermore, no information was available on the distribution of FTEs between the different types of alternatives to full-time hospitalization in the hospital hosting them. Information was also lacking on the distribution of FTEs between the different forms of care at the sector level. Differences between the sectors hosted by the same hospital could exist, even if they follow the same general policy. Moreover, even though it has been widely demonstrated that supply influences practice (70–72), use of the overall proportion of FTEs allocated to alternatives as a measure of the level of development of these alternatives did not allow us to directly determine whether practitioners referred their patients to those types of services.

Finally, some sectors were excluded from the analysis because of poor data quality. Sectors hosted by hospitals that did not report data on alternatives to full-time hospitalization may have been less likely to develop such alternatives. However, these data are not used for financial or certification purposes and excluded and included hospitals did not differ in terms of main organizational and institutional characteristics or case mix.

Conclusions

Our study is the first to focus on the association between involuntary full-time admissions and the development of alternatives to full-time hospitalization in the French context. It was carried out on a national scale, thus limiting selection bias, and results were adjusted on the basis of a wide range of characteristics of patients, health care providers, and the environment. It is our belief that our results, usefully supplemented by research carried out in other settings and on other aspects of psychiatric care, can be used by researchers, policy makers, health professionals, and patients alike to support the development of alternatives to full-time hospitalization worldwide.