Increased Treatment Complexity for Major Depressive Disorder for Inpatients With Comorbid Personality Disorder

The introduction of a new remuneration system obliges all psychiatric and psychosomatic hospitals in Germany to collect standardized routine data. For research purposes, the VIPP (Versorgungsindikatoren in Psychiatrie und Psychosomatik, or care indicators in psychiatry and psychosomatics) database merges these data from a subsample of hospitals and enriches them with supplementary information (5). This study examined year 2013 VIPP data from 20 psychiatric and psychosomatic departments in general hospitals and 55 psychiatric and psychosomatic hospitals (approximately 15%−20% of all psychiatric and psychosomatic hospitals) (6) from 14 of the 16 federal states of Germany. All cases from the participating hospitals were included in the data set.

The indicators were based on diagnoses and procedure information that were initially generated for accounting purposes. All available accounting information categories that appeared helpful for this study’s aims were included as indicators. The data followed standardized definitions of ICD-10-GM (German modification) codes (7) and otherwise of the German Institute for Medical Documentation and Information’s German Procedure Classification (8, available only in German). All diagnoses were discharge diagnoses.

From 671,062 inpatient or day treatment clinic cases for patients ages 18 to 64 (all cases of the participating hospitals in 2013, with “case” meaning one admission), we selected three groups: first, the MDD group included all patients with ICD-10-GM codes F32.X or F33.X as the main diagnosis without a secondary diagnosis of F60.X or F61.X; second, the MDD+PD group included all patients with F32.X or F33.X as a main diagnosis code and F60.X or F61.X as a secondary diagnosis; and third, the personality disorder group consisted of all patients with a main diagnosis code of F60.X or F61.X.

There were several indicators. Episode severity was obtained by comparing the MDD and MDD+PD groups in percentage of cases that included diagnoses of mild, moderate, severe, or psychotic depression. Recurrent depression involved comparison between the MDD and MDD+PD groups for percentage of cases that included diagnoses of recurrent depression (ICD-10-GM code F33.X), and one depressive episode only (ICD-10-GM code F32.X). Emergency admissions indicated the admissions for acute medical reasons (suicidality, for example) without prior authorization by the patient’s health insurance fund. Discharge against medical advice was the percentage of cases discharged against medical advice. Average length of stay did not account for absences and periods of leave. Readmissions within one year indicated cases admitted a second time within 365 days to a psychiatric or psychosomatic hospital with an ICD-10-GM F-code main diagnosis (the diagnosis was not necessarily the same on the second admission).

TTUs, or Therapieeinheiten, are a standardized measure for contacts by physicians and psychologist. Both individual and group contacts are counted. For calculating the total TTUs, the weight of group contacts depends on group size. Only contacts with a duration of at least 25 minutes are taken into account, and time is counted in 25-minute steps. Thus TTUs do not give the exact amount of time spent with the patient, because short contacts are not counted. However, TTUs can serve as an estimate of how much time is provided for psychotherapeutic interventions.

Accounting codes for treatment modalities were another indicator used for the study. In the current German reimbursement system, every psychiatric and psychosomatic case gets a standardized code that indicates treatment complexity and setting for every day of treatment. The options most used are standard treatment (Regelbehandlung); intensive treatment (Intensivbehandlung), which required endangerment of self or others, special safety measures, severe increase or decrease in goal-directed activity, no autonomous liquid or food intake, acute danger by distortions of orientation and reality, or withdrawal syndrome with vital danger, or some combination of these; and day clinic treatment (Tagesklinische Behandlung). Because a single case can have different categories on different days, this indicator compares the percentage of days in the three categories.

Another indicator included the reimbursement accounting codes for special treatment efforts or complexities (“add-on codes”) that indicate an especially long diagnostic interview (complex diagnostic procedures), a crisis intervention of more than 25 minutes (crisis intervention), or the permanent presence of a staff member for surveillance (constant observation) (9). The study compared the percentage of cases that received an add-on code at least once during the stay.

Distribution of personality disorder diagnoses was compared across groups, specifically, the spectrum of personality disorder main diagnoses in the personality disorder group and secondary diagnoses in the MDD+PD group. To determine whether differences were significant, we performed chi-square or t tests, and the null hypothesis assumed no differences among the groups.

The study included 58,913 cases within 2013, with 41,556 in the MDD group, 8,730 in the MDD+PD group, and 8,627 in the personality disorder group. For the ten indicators, the results of the comparison of the three groups are shown in Table 1.

The comparison between MDD and MDD+PD groups showed no differences for emergency admissions, discharge against medical advice, and TTUs. Differences were small and without clinical significance for mild and severe depression and for accounting codes for treatment modalities. Accounting codes for special treatment efforts or complexities indicated that the MDD+PD group more frequently received complex diagnostic procedures, crisis intervention, and constant observation compared with the MDD group, although rates were on a low level for both groups. However, rates of recurrent depression were 15 percentage points higher and readmission rates within one year were 16 percentage points higher in the MDD+PD group compared with the MDD group despite a higher mean length of stay. These differences might be partially due to a more thorough diagnostic review (as indicated by the complex diagnostic procedures indicator) of more complicated cases, leading to more secondary diagnoses of personality disorder. But regardless of the underlying cause, this study thereby identified with the MDD+PD group a subgroup of MDD patients with a more complex treatment course. Indirectly, the indicators of recurrent depression and readmission within one year implied a worse treatment outcome for the MDD+PD group.

Comparison of the MDD+PD group with the personality disorder group revealed that for the MDD+PD group, many indicators, including length of stay, day clinic treatment, intensive treatment, and rates of emergency admissions, were more similar to the MDD group than to the PD group. However, the MDD+PD and personality disorder groups shared the very high rate of one-year readmissions, with the latter group faring worse. In terms of diagnoses, the MDD+PD group was much less dominated by borderline personality disorder (ICD-10-GM code F60.3); instead, mixed and other personality disorders (F61) and other specific personality disorders, including narcissistic and passive-aggressive personality disorder (F60.8), avoidant personality disorder (F60.6), and dependent personality disorder (F60.7), were more common.

This study compared indicators of inpatient routine care between three groups—all patients with MDD, patients with a main diagnosis of MDD and a secondary diagnosis of personality disorder, and patients with a main diagnosis of personality disorder—in a large sample of 58,913 cases from 75 psychiatric hospitals. The main finding was that MDD treatment was more complex, and recurrence of MDD episodes and hospital readmission occurred more often in patients with a personality disorder comorbidity.

This analysis of health care routine data was limited by several factors. It was not possible to test the validity of diagnoses (for example if personality disorder criteria in a patient with MDD were present not only during the depressive episode but prior to it as well) or exclude bias of the indicators by factoring in accounting incentives. Furthermore, the indicators provided only indirect evidence for relevant variables, such as complexity or remission. Finally, the indicators of depression recurrence and readmission are probably also strongly influenced by features of outpatient care, such as availability of integrated care concepts and outpatient psychotherapy (10).

Several questions arose in analyzing these results. First, in most evidence-based treatment strategies for personality disorder and MDD, psychotherapeutic approaches play an important role (11–13). These approaches require a certain minimum treatment time with qualified professionals (14,15). The amount of individual contacts with therapists and nursing staff measured by TTUs in this study was quite low and variable in all groups. A TTU of .25 corresponded to only 6.25 minutes of daily individual contact with doctors and a psychologist. On the one hand, a straightforward interpretation of this result is complicated, because group therapies also counted toward the TTU value, and their weight in our analysis was dependent on group size. On the other hand, the low TTU value and its large variability suggest that many different therapeutic approaches were in use and raise doubt that all were evidence based. Translating evidence-based concepts into routine practice may help to lower rates of readmission and recurrence, especially for the MDD+PD group and the personality disorder group. Therefore, more research is needed on the application of therapeutic concepts and allocation of qualified professionals such as psychotherapists, psychiatrists, and mental health nurses during psychiatric inpatient treatment and their influence on outcomes.

Second, future research should take the influence of the outpatient treatment infrastructure into account, especially for patient groups with high rates of readmission and recurrence (10). Third, the influence of other comorbidities and social determinants would be interesting to examine to complete the picture on which patient groups require improved treatment concepts in routine treatment. Finally, relating the indirect outcome parameters used in this study to direct indicators of recovery and patient-rated outcome measures would be an important contribution in future research.