Psychiatric services utilization in completed suicides of a youth centres population

The sample has been described in detail in a previous communication [2]. Subjects were 12 through 18 years of age. Of the 143 subjects who were eligible for the study, 53 had an available YCs file, other files having been cleared according to the YC archival regulations. These subjects were randomly matched to three living subjects receiving YCs services at the time of the death of the index case, for age, sex and geographic area. This strategy allowed us to identify living YCs controls with and without file references to suicidal behavior or ideation, a dichotomy permitting higher degrees of specificity in analyses. We identified 135 subjects with available YCs files (107 boys and 28 girls), while 24 files were not available. From these 135 files, 40 revealed references to a history of suicidal behavior or ideation, while 95 revealed no such mention.

For each suicide, the following information was extracted from the Quebec Coroner's Office computerized database: date of birth, sex, date of event (i.e. date of suicide), date of death, and means of suicide. For each subject (suicide and control groups), names and health insurance numbers were used for retrieving information about medical services from the physician fee-for-service database of the "Régie de l'assurance maladie du Québec" (RAMQ), a database regrouping information on medical interventions. Information pertaining to hospitalizations (reason for admission, length of stay, International Classification of Diseases Ninth Revision (ICD-9) diagnosis, and number of hospitalizations) was obtained from the Quebec Ministry of Health and Social Services (MHSS) hospitalization database system "Maintenance et Exploitation des Données pour l'Ètude de la Clientèle Hospitalière" (MED-ECHO). These data were retrieved for subjects' lifetime and for the one-year period preceding suicide or index period, using the discharge date in case of hospitalizations.

The utilization unit for outpatient medical services, "medical visit", regroups all outpatient services provided by a single physician to a single patient during one calendar day. A medical visit was classified as "psychiatric" if it included at least one mental health evaluation/intervention or otherwise, the visit was classified as "physical". The visits were further sub classified according to different time intervals: past week, past month, past 90 days, or past year. Hospitalizations from the MED-ECHO database were classified as "psychiatric" if they included at least one mental health diagnosis, or otherwise, as "physical". Analysis of the ICD-9 codes led to the exclusion of two incidences hospitalization related to fatal suicide attempts.

Data originating coming from the sources of information employed for this study have been found to be reliable [3, 4]. All data were entered into a relational database, and analyses were carried out with the SPSS statistical package [5]. Each type of visit and hospitalization, according to time intervals, have been analyzed as present or absent for each subject by group (dichotomized variables) in order to perform non parametric analyses on numbers of subjects by category. We considered results as significant when the probability of type I error was equal or lower than 5%.

This study was approved by the local Institutional Review Board and by the Commission d'accès à l'information du Québec.

We found a higher proportion of "physical" medical visits among suicides in the week before death than among the two control groups (n = 11, 20.8% vs. n = 2, 5.0% vs. n = 8, 8.4%; X² = 7.15, df = 2, p < 0.05). In the month, 90 days, and year preceding index period or suicide, there were no significant differences across the three groups for physical medical visits (Table 1).

In terms of psychiatric visits, we found a higher rate of service utilization for the suicide group in all four time intervals: last week (n = 5, 9.4% vs. n = 0, 0% vs. n = 0, 0%; X² = 13.08, df = 2, p < 0.001), last month (n = 7, 13.2% vs. n = 0, 0.0% vs. n = 1, 1.1%; X² = 14.59, df = 2, p < 0.001), last 90 days (n = 8, 15.1% vs. n = 1, 2.5% vs. n = 3, 3.2%; X² = 9.39, df = 2, p < 0.01), and last year (n = 15, 28.3% vs. n = 6, 15.0% vs. n = 6, 6.3%; X² = 13.38, df = 2, p < 0.001).

We found a higher rate of lifetime (n = 15, 28.3% vs. n = 5, 12.5% vs. n = 4, 4.2%; X² = 17.7, df = 2, p < 0.001), and last year (n = 9, 17.0% vs. n = 2, 5.0% vs. n = 1, 1.1%; X² = 14.6, df = 2, p < 0.001) hospitalizations for psychiatric reasons in the suicide group compared to the control groups. In addition, 4 suicide subjects had been hospitalized after a suicide attempt between 7 months and 3 years prior to fatality. In terms of past year and lifetime hospitalizations (pooled physical and psychiatric), we found no significant difference across the groups.

Exploratory analyses investigated gender differences in service utilization among cases and controls, respectively. In the suicide group, there were no significant differences between girls and boys for all indicators. Conversely, in the controls, there was a general trend in all indicators, girls having significantly more physical visits in the last week, last month and last 90 days (Table 2).