Of the 616 patients meeting inclusion criteria, 94 patients (15.3%) either suffered from pre-existing psychiatric illness (Stable), needed psychiatric consultation during hospitalization (Acute), or suffered both pre-existent from a psychiatric illness and needed psychiatric consultation during hospitalization (Chronic).

The greatest expenses during hospitalization were Clinical costs, which mainly included nursing costs on the inpatient ward and intensive care unit (ICU) (Table 1). The median Clinical costs for the psychiatric cohort were significantly (50%) higher than those for the control cohort (€12000 (IQR: €21900) versus €7400 (IQR: €12200), p < 0.01). The psychiatric cohort also generated significantly greater Laboratory costs than the control cohort (€1100 (IQR: €2100) versus €500 (IQR: €1400), p < 0.01). Operating room expenses were also significantly higher for psychiatric patients than for the control cohort (€1500 (IQR: €5800) versus €1100 (IQR: €3600), p = 0.04).

All costs of treatment given in the patients Emergency Department (ED) were listed under ED costs. The psychiatric patients had higher overall ED costs than the control cohort (€500 (IQR: €700) versus €400 (IQR: €500), p = 0.01).

Psychiatricv patients had higher Other costs than the control group (€3100 (IQR: €3900) versus €1900 (IQR: €2300), p < 0.01). Finally, average total in-hospital costs for psychiatric patients were significantly higher than the control group (€22000 (IQR: €35000) versus €15200 (IQR: €20200) p < 0.01) (Fig. 1).

(a): Total in-hospital costs and Clinical costs by psychiatric comorbidity status. (b): Other, Image, Surgical, Laboratory and Emergency Department costs by psychiatric comorbidity status. *Denotes statistical significance between groups (p < 0.05)

Within the three different psychiatric sub-cohorts, there were significant differences in median Clinical costs: Acute €29,700 (IQR: €33400), Stable €7000 (IQR: €8200), Chronic €13,100 (IQR: €18300) versus Control €13,000 (IQR: €16200), p = 0.04.

Acute and Chronic psychiatric patients encountered significantly higher Clinical costs in comparison with the Stable and Control psychiatric sub-cohort, but there were no differences in costs between the Acute and Chronic groups (p = 0.08) (Fig. 2).

Comparison of total in-hospital costs and Clinical costs between sub-cohorts

There were also significant differences in total in-hospital costs between the psychiatric sub-cohorts (Fig. 2). The active psychiatric sub-cohorts (Acute and Chronic) generated higher total in-hospital costs in comparison with the inactive psychiatric sub-cohorts (Stable and Control) (Acute €47000 (IQR: €62300), Stable €11700 (IQR: €16000), Chronic €25900 (IQR: €33800) versus Control of €15200 (IQR: €20200), p < 0.01). There were no significant differences in in-hospital costs between the Stable psychiatric cohort and Control cohort.

Finally, the independent effect of various clinically relevant as well as significantly different factors (p < 0.05) were assessed (Table 4). Because the R-squared of the multivariable linear regression model was 0.86, the independent variables used explained a great percentage of the higher total in-hospital costs. The model showed that despite the marked differences in in-hospital costs, psychiatric comorbidity was not an independent predictor for total in-hospital costs (p = 0.88). However, the number of days in hospital (p < 0.01), as well as the ISS score (p < 0.01) and the number of surgical interventions during hospitalization (p < 0.01) independently predicted the total in-hospital expenses of our psychiatric cohort (Table 4). Since there was a distinct difference between the Stable and Control cohorts compared to the Acute and Chronic cohorts, a multivariate model was created that compared psychiatric sub-cohorts. This model did not reveal active psychiatric diseases (i.e. Acute and Chronic sub-cohorts) as an independent predictor for in-hospital costs (Supplement 1). Also in this model the R-squared was 0.86.