Introduction

Reliable and timely health information is the foundation for effective health services management and public health action [1]. In the area of mental health strengthening information systems is one of the ten recommendations issued by the World Health Organization (WHO) to make a difference in mental healthcare [2]. When their potential is fully harnessed, health information systems can generate data to assess and monitor the structure, processes, and outcomes of care [3].

In mental healthcare, one area of major concern identified by health information systems is the appropriate use of antipsychotic (AP) medications [4]. High variability in AP use in community mental healthcare has often been documented, with concerns regarding the appropriateness of therapy, including misuse, underuse and overuse [5]. Interestingly, while most studies have investigated patient- and prescriber-related factors as possible reasons for variability in AP use, almost no studies have investigated facility-level characteristics.

One facility characteristic that may be associated with AP use is staffing levels, as already documented in nursing homes [6]. In these hospital-based settings studies have demonstrated an inverse relationship between staffing levels and AP use, that is, as staffing levels decrease AP drug use increases [6,7]. However, in community mental healthcare there are no data investigating this association. Against this background, given the availability of a recently implemented Italian national mental health information system, this study investigated whether staffing level is associated with AP prescribing in community mental healthcare.

Methods

Results

In Italy in 2015 there were on average nearly 60 mental health professionals per 100,000 inhabitants, with wide regional variations (range 21 to 100) (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Distribution of mental health staff, psychiatric beds, treated incidence and treated prevalence of mental disorders, poverty index, employment rate and rate of individuals treated with antipsychotics by Italian regions.

https://doi.org/10.1371/journal.pone.0193216.t001

On average, there were 10 psychiatric beds per 100,000 inhabitants (range 5 to 21). The treated prevalence of schizophrenia was 0.31% (range 0.13% to 0.44%) and the treated prevalence of bipolar disorder was 0.12% (range 0.05% to 0.25%), while the treated prevalence of any psychiatric disorder was 1.59% (1.07 to 2.05) The average rate of individuals prescribed AP drugs was 2.33%, with wide regional variations (1.04% to 4.01%) (Table 1).

Univariate analysis showed that the rate of individuals prescribed AP drugs was inversely associated with the rate of mental health professionals available in Italian regions (Kendall's tau -0.438, p = 0.006), with lower rates of AP users in regions with higher rates of mental health professionals (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Relationship between staff availability per 100,000 inhabitants and rate of individuals prescribed antipsychotic drugs per 1,000 inhabitants in Italian regions.

https://doi.org/10.1371/journal.pone.0193216.g001

A similar relationship was observed for mental health nurses (Kendall's tau -0.485, p = 0.002) but not for psychiatrists (Kendall's tau -0.181, p = 0.263), psychologists (Kendall's tau 0.228, p = 0.155) and educators (Kendall's tau -0.040, p = 0.833). After adjustment for possible confounders (Table 2), the total availability of mental health professionals, and the availability of nurses, but not of psychiatrists, psychologists and educators, significantly predicted the rate of individuals exposed to AP drugs (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Multivariate linear regression models investigating the association between mental health staff availability (total, mental health nurses, psychiatrists) and the rate of individuals prescribed antipsychotic drugs in Italy.

https://doi.org/10.1371/journal.pone.0193216.t002

Discussion

Analysis of Italian national statistics found an inverse relationship between mental health staffing levels and AP prescribing. Intriguingly, when the analysis was carried out by type of staff, a statistical association was found for nurses but not for other professionals. We note, however, that the association coefficients for psychiatrists and psychologists were similar in magnitude to that for nurses, but with wider confidence intervals, therefore suggesting a similar general trend.

To our knowledge, this is the first study highlighting an association between staffing levels and AP prescribing in community mental healthcare. Interestingly, however, very similar results were found by a number of epidemiological studies carried out in nursing home settings, where staffing level was found to be inversely associated with AP use [6,7]. This association was particularly evident for nurses, who play a critical role in the caring of nursing home residents [6,7].

In community mental healthcare, however, the interpretation of this association is not straightforward, and the Italian data would suggest the following considerations. First, the rate of individuals exposed to AP medications is five times higher than the treated prevalence of schizophrenia and bipolar disorder considered together, and 1.5 higher than the treated prevalence of any mental disorder. Second, the rate of individuals prescribed AP medications is extremely variable in Italian regions, ranging from 1 to 4%. Third, one of the main challenges to studying AP utilization as a variable is determining appropriate versus inappropriate use. It would be incorrect, based on these aggregate statistics, to a priori imply that higher AP prescribing is negative. The existence of an inverse relationship does not guarantee inappropriate use as other factors may impact prescriber decisions to use APs. It remains therefore unclear what these data imply in terms of AP prescribing quality. For all these reasons, the possibility that increasing mental health staff may lead to better community mental health care and, consequently, less psychopathology and less AP use, remains a challenging hypothesis generated, but not addressed, by the present analysis. In any case, we demonstrated that a facility-level factor was associated with AP prescribing, which is a new key finding from a health system perspective.

This study has limitations. A first issue is that only data aggregated at regional level were accessed and analysed. This inevitably decreased statistical power, especially when multiple variables were included in the regression models. Additionally, information aggregated at regional level implies that within each Italian region service and AP prescribing data were homogeneous. We acknowledge that this assumption could not be verified and that the same analysis should be replicated at a local rather than regional level in order to increase statistical power, and to be based on fewer assumptions. A second limitation is that no outcome data were analysed to investigate the implications of different rates of AP prescribing at a population level. A third limitation is that although we adjusted for a number of potential confounders, we cannot exclude that residual confounding might have acted in an unpredictable way. For example, we could not adjust the analysis for disease severity. Finally, we cannot exclude that the rate of AP prescribing has been slightly overestimated as Italian regions collect prescribing data from two different sources. Additionally, as with all analyses of prescription databases, the lack of data on whether patients eventually took the prescribed agents should be highlighted, since a relevant proportion of the medicines prescribed for people with chronic conditions are not taken.

This study has practical implications. The finding that staffing level, a variable related to the organization of mental health systems, is associated with the observed variability in the rate of AP prescribing, suggests that any actions aimed at decreasing such variability need to take into account, in addition to patient- and prescriber-related factors, aspects related to the organization of the mental health system, in particular its staffing level. It is perhaps not surprising that mental health care exhibited a high degree of geographic variability. It has been shown that dissemination and adoption of treatments are influenced by many factors other than patient needs and scientific evidence, including marketing, profit motives, health care cultures, professional and guild issues, spending levels, organizational structures, financing mechanisms, and implementation difficulties [11,12]. In our analysis it is possible that spending levels influenced staffing level and organizational structures which, in turn, may have affected the provision of psychosocial interventions. Similarly, mental health care culture and professional issues may have affected AP prescribing as well as the provision of psychosocial interventions.

Unfortunately, the present study was not able to identify any potential mechanisms that may underlie the association between staffing levels and AP prescribing. It might be speculated that in regions with high staffing levels psychological, psychosocial and rehabilitation interventions were provided on a more regular basis, or more intensively or more appropriately, than in regions with low staffing level. As a consequence, individuals with mental disorders might have required less AP prescribing given the provision of non-pharmacological interventions. By contrast, in regions with low staffing levels, paucity of human resources might have implied limited provision of psychosocial and rehabilitation interventions, leaving AP prescribing as one of the few available therapeutic options.

In terms of implications for research, we argue that current evidence of correlation between staffing and AP prescribing warrants ad hoc studies to demonstrate potential causality.

Acknowledgments

We wish to acknowledge the contribution of Flavia Baccari (Department of Mental Health and Drug Abuse, AUSL Modena, Italy) who provided expert input on the analysis.