Table 3.
Results of the multivariable analysis (negative binomial regression)
| Variable | Estimated value of the coefficient | Standard error | 95% Confidence interval of the coefficient | Exponentiated coefficient | P-value | |
|---|---|---|---|---|---|---|
| Lower bound | Upper bound | |||||
| Intercept | −9.6329 | 0.9758 | −11.5455 | −7.7203 | < 0.0001 | |
| Epidemiological characteristics | ||||||
| Psychiatric health status of the population | ||||||
| Number of individuals suffering from chronic mental disorders (per 100,000 inhabitants)a | 0.0000 | 0.0001 | −0.0002 | 0.0003 | 1.0000 | 0.8069 |
| Percentage of deaths by suicide among total deaths | 0.0837 | 0.0283 | 0.0282 | 0.1391 | 1.0873 | 0.0031 |
| Overall health status of the population (per 100,000 inhabitants) | ||||||
| Acute admission rate for somatic disorders | 0.0000 | 0.0000 | −0.0000 | 0.0000 | 1.0000 | 0.2079 |
| Mortality rate | 0.0014 | 0.0009 | −0.0004 | 0.0032 | 1.0014 | 0.1223 |
| Number of individuals suffering from chronic somaticdisorders | 0.0000 | 0.0000 | −0.0001 | 0.0000 | 1.0000 | 0.1868 |
| Demographics of the population | ||||||
| Number of women (per 100,000 inhabitants) | 0.0000 | 0.0000 | −0.0000 | 0.0000 | 1.0000 | 0.9899 |
| Mean age of individuals aged over 16 | 0.0166 | 0.0163 | −0.0155 | 0.0486 | 1.0167 | 0.3109 |
| Socio-economic characteristics of the population | ||||||
| Quintile of the mean deprivation index (FDep) (from lower to higher deprivation), reference: 5th quintile | ||||||
| 1 | 0.2679 | 0.1399 | −0.0064 | 0.5422 | 1.3072 | 0.0556 |
| 2 | 0.2941 | 0.1117 | 0.0751 | 0.5130 | 1.3419 | 0.0085 |
| 3 | 0.2939 | 0.1102 | 0.0779 | 0.5098 | 1.3416 | 0.0076 |
| 4 | 0.2944 | 0.1050 | 0.0886 | 0.5002 | 1.3423 | 0.0051 |
| Characteristics of the supply of health and social care | ||||||
| Supply of public mental health care | ||||||
| Characteristics of the hospital to which each sector was linked | ||||||
| Private non-profit (vs. public) | 0.4939 | 0.1990 | 0.1039 | 0.8840 | 1.6387 | 0.0131 |
| Participation to teaching activities (vs. no participation) | −0.3929 | 0.1001 | −0.5891 | −0.1968 | 0.6751 | < 0.0001 |
| Specialization in psychiatry (vs. general hospital) | −0.3476 | 0.0721 | −0.4890 | −0.2062 | 0.7064 | < 0.0001 |
| Participation to emergency care (vs. no participation) | 0.2215 | 0.1068 | 0.0122 | 0.4307 | 1.2479 | 0.0380 |
| Number of inpatient beds in the hospital (per 100,000 inhabitants)c | 0.0046 | 0.0025 | −0.0002 | 0.0095 | 1.0046 | 0.0609 |
| Supply of private mental health care (per 100,000 inhabitants) | ||||||
| Number of community-based private psychiatristsb | −0.0202 | 0.0055 | −0.0310 | −0.0094 | 0.9800 | 0.0003 |
| Number of psychologists | 0.0011 | 0.0008 | −0.0005 | 0.0026 | 1.0011 | 0.1678 |
| Number of psychiatric inpatient beds in private for-profit hospitalsa | −0.0013 | 0.0020 | −0.0052 | 0.0026 | 0.9987 | 0.5059 |
| Supply of non-specialized care (per 100,000 inhabitants) | ||||||
| Number of general practitionersb | −0.0009 | 0.0023 | −0.0054 | 0.0035 | 0.9991 | 0.6777 |
| Number of non-psychiatric inpatient beds | 0.0001 | 0.0001 | −0.0001 | 0.0002 | 1.0001 | 0.5699 |
| Supply of social care (per 100,000 inhabitants) | ||||||
| Number of beds in housing institutions for disabled individuals | 0.0012 | 0.0006 | 0.0001 | 0.0023 | 1.0012 | 0.0335 |
| Capacity of centres providing care through employment | 0.0003 | 0.0007 | −0.0010 | 0.0017 | 1.0003 | 0.6101 |
| Capacity of housing and social rehabilitation centres | −0.0001 | 0.0008 | −0.0017 | 0.0015 | 0.9999 | 0.8884 |
| Level of urbanization | ||||||
| Level of urbanization (from lower to higher urbanization), reference: 6th quantile | ||||||
| 1 | −0.0818 | 0.1147 | −0.3066 | 0.1429 | 0.9215 | 0.4754 |
| 2 | −0.0664 | 0.1448 | −0.3503 | 0.2174 | 0.9358 | 0.6463 |
| 3 | −0.0456 | 0.3700 | −0.7709 | 0.6796 | 0.9554 | 0.9018 |
| 4 | −0.1111 | 0.5310 | −1.1518 | 0.9296 | 0.8948 | 0.8343 |
| 5 | −0.1557 | 0.1118 | −0.3749 | 0.0634 | 0.8558 | 0.1637 |
a and b Significant correlations were observed between these variables. However, corresponding correlation coefficients were weak and there were strong hypotheses on associations of these variables with psychiatric inpatient admission rates so they were all introduced in the model. cThe number of inpatient beds per 100,000 inhabitants of the catchment area was highly correlated with the total number of full-time equivalents allocated to psychiatric care by the hospital to which each sector was linked per 100,000 inhabitants of the catchment area. We therefore only introduced the number of beds in the model