Inpatient Volume and Quality of Mental Health Care Among Patients With Unipolar Depression

There is a growing interest in the association between inpatient volume and the quality of care, and the organizational structure in the health care sector, such as inpatient volume in hospital units, is a topic of discussion worldwide. The objective of the discussion is to ensure better treatment quality, higher patient satisfaction, and more efficiency (1). Several observational studies have examined the association between inpatient volume and clinical outcomes for surgical procedures and various medical conditions. The evidence from these studies indicates that higher inpatient volume is associated with better clinical outcomes, including reduced complications and lower mortality (2–5). Despite the increasing interest, an analogous association between inpatient volume and the quality of mental health care has not been examined properly.

To our knowledge, only six published studies have examined this association (1,6–10). However, the results of these studies are inconsistent, making it difficult to draw conclusions. Furthermore, only one of the existing studies has examined the association specifically for patients with depression, and there is a general lack of studies with detailed data regarding the quality of the provided mental health care (1,6–10). To further assess the role of inpatient volume in mental health care, we conducted a nationwide population-based cohort study to examine the association between inpatient volume per psychiatric hospital ward and the quality of mental health care, as reflected by the receipt of specific guideline-based processes of care among Danish patients admitted with depression.

Methods

Danish Health Care System

The Danish health care system is mainly tax-funded and provides health coverage that is free of charge (11). In particular, health equity is a stated priority, with equal reimbursement across all institutional levels. If psychiatric treatment is required, patients with depression can be admitted to public psychiatric hospitals, and their use of inpatient services is recorded in national registers with the patient’s unique, 10-digit civil registration number (12,13).

Danish Depression Database

The Danish Depression Database was established in 2011 with the objective to monitor, document, and improve the quality of treatment and care among patients with depression. It is mandatory by law for all Danish psychiatric hospital wards to report data to the registry on all patients with depression who received treatment. The database holds information on process performance measures reflecting recommendations from national clinical guidelines (14–17). The following process performance measures were obtained for inpatients (Table 1): examination by a psychiatrist, somatic examination, assessment by a social worker, assessment of depression (soon after admission and upon discharge), assessment of suicide risk (upon admission and before discharge), contact with relatives, and plans for psychiatric aftercare. The registry also contains information on gender and age. The process performance measures are not necessarily causally linked with improved clinical outcomes, but they reflect key recommendations from the national clinical guidelines. The measures were selected from the national clinical guidelines by an expert panel consisting of psychiatrists, psychologist, nurses, occupational therapists, and social workers (15,17). The data are collected prospectively from documentation contained in the medical records by using a registration form with detailed instructions from the Danish Depression Database. The process performance measures for inpatients with depression are continuously collected during the hospitalization by the health care professionals responsible for the care of the individual patient (16).

Of the 21 processes of care monitored by the Danish Depression Database, nine are relevant to inpatients (15,17).

Study Population

The study population included all patients ages 18 and older who were admitted with unipolar depression as a primary diagnosis and registered in the Danish Depression Database between January 1, 2011, and December 31, 2016. Depression was defined according to the ICD-10 (codes F32.0– F32.99, F33.0– F33.99, F34.1, and F06.32) (18). A total of 18,389 patients were identified. The patients were admitted to 94 different psychiatric hospital wards with 24,395 admissions during the study period. However, we included only the first recorded admission in each calendar year per patient at each psychiatric hospital ward because readmissions are not comparable with a similar number of one-time admissions of different patients. We excluded hospitals wards with fewer than 20 recorded admissions during the entire study period, corresponding to 418 patients. This exclusion was made because hospital wards with only sporadic admissions were potentially less likely to report data to the Danish Depression Database because of possible inadequate routines.

Inpatient Volume

The inpatient volume was defined as the average number of admissions to each psychiatric hospital ward per year from 2011 to 2016, on the basis of the entire study population, including 17,971 patients with 21,120 admissions. The inpatient volume was divided into four quartiles and referred to in this study as low volume (quartile 1, <102 inpatients per year), medium volume (quartile 2, 102–172 inpatients per year), high volume (quartile 3, 173–227 inpatients per year), and very high volume (quartile 4, >227 inpatients per year).

Statistical Analyses

The quality of mental health care was measured by the receipt of process performance measures of care for each admission, and it was assessed both overall and for the individual processes of care. The overall quality of care was calculated by dividing the number of process performance measures that a patient received during each recorded admission by the nine process performance measures that were relevant to inpatients with depression. The association between inpatient volume and the quality of mental health care—the overall quality of care as well as the individual process performance measures —was examined by using binomial regression while adjusting for gender and age. Information on other patient- or hospital-related covariates was not available. High overall quality of care was defined as a patient’s receipt of ≥80% of all relevant recommended process performance measures. The Wald test was used to test for trends across the quartiles of inpatient volume. The analysis was also repeated with alternative cut off points of quality of care, varying from 60% to 90% of process performance measures received. The association was likewise examined in a sensitivity analysis for the inpatients at the excluded hospital wards.

A number of additional analyses were performed to examine the robustness of the primary analysis. First, inpatient volume was examined as a continuous variable. Second, a multinomial logistic regression was used to examine the association between inpatient volume and the quality of mental health care, with the quality of mental health care defined as 0%−50%, 51%−70%, 71%−90%, and >90%. Third, an analysis was performed based on the number of unique patients rather than the number of admissions. In this analysis, we excluded patients from 2011 to ensure a minimum wash-out period of previous admissions of 12 months.

All 95% confidence intervals [CIs] were corrected for the clustering of patients within psychiatric hospital wards by using robust estimates of the variance. The analyses were adjusted and stratified according to gender and age. A two-sided p value of ≤.05 was considered significant.

Results

Characteristics of patients in the four quartiles of inpatient volume did not vary substantially. Most patients were older than 60 years, and the majority were women. On the overall quality of care by inpatient quartiles, shown as the proportion of relevant process performance measures received by the patients, the largest proportion of patients was in the group receiving 0% to 20% of the relevant recommended process performance measures. The proportion of patients receiving ≥80% of the recommended process performance measures varied between 11.8% and 21.0%.

The association between volume of inpatient admissions and the overall quality of care are presented in Table 2. Patients admitted to very-high-volume psychiatric hospital wards had a higher likelihood of receiving high overall quality of care (≥80% of the process performance measures), compared with patients admitted to low-volume wards (gender- and age-adjusted relative risk (ARR)=1.78, CI=1.02–3.09). The test for trends across the inpatient volume categories did not reach statistical significance (Table 2). When alternative cut off points for quality of care were used, the association between inpatient volume and quality of care was likewise confirmed. That is, patients in very-high-volume wards were more likely than those in low-volume wards to receive >90% of the process performance measures (ARR=2.02, CI=1.03–3.97). No statistical significant association was found when treating inpatient volume as a continuous variable.

Table 3 presents data on the receipt of the nine individual process performance measures by inpatient volume quartiles. The proportion of admissions in which patients received the individual process performance measures was low, ranging from 18% to 66%. Only 35% to 44% of admitted patients are seen by a psychiatrist within seven days, and only about one-half of the patients received a suicide risk assessment at admission and discharge. Likewise only about 40% had planned psychiatric aftercare.

The association between inpatient volume and the individual process performance measures are presented in Table 4. Patients admitted to low-volume psychiatric hospital wards were less likely than the other three inpatient quartiles to have received the individual processes of care; however, not all associations reached statistical significance. Patients admitted to very-high-volume wards had a higher likelihood of being somatically examined, compared with those admitted to low-volume wards (ARR=1.35, CI=1.03–1.78). In addition, a multinomial logistic regression demonstrated that admission to very-high-volume wards was associated most strongly with the highest level of quality of care (>90% of the process performance measures). Hence, admission to a very-high-volume ward was associated with an adjusted RR of 1.11 (CI=.68–1.83) for receiving >50%−70% of the process performance measures, whereas the adjusted RR for receiving >90% of the process performance measures was 2.39 (CI=1.00–5.69). Thus, the difference between high- and low-volume wards with regards to quality of care was most pronounced with regards to the chance of receiving optimal or near-optimal quality of care.

We found no evidence of systematic interaction when the analyses stratified patients according to gender and age. In addition, the findings from the primary analysis were confirmed when performing an analysis based on the volume of unique patients at the hospitals, rather than the admissions, during the study period (data not shown). The sensitivity analysis showed that the characteristics of the 418 excluded inpatients were similar to those of the patients in our study. For the overall quality of care, only 5.5% of the patients who were excluded received ≥80% of the recommended process performance measures, whereas 72% received 0%−20%.

Discussion

Overview

Our results showed a small but statistical significant association between inpatient volume and quality of mental health care among patients with depression. Patients with depression who were admitted to very-high-volume wards were more likely than those admitted to low-volume wards to receive 80% or more of process performance measures, including a somatic examination. However, no significant association with quality of mental health care was found with a continuous increase in inpatient volume. The differences in care between hospitals above the lowest patient volume category were small or nonexisting, and the findings may therefore imply that low-volume psychiatric hospital wards are challenged with delivering optimal care for inpatients with depression—at least in some areas of care.

Strengths and Limitations

The study strengths include prospectively collected data, a relatively large study population, and a nationwide population-based design. The Danish Depression Database has high coverage—i.e., in 2015, it was estimated to include records for 100% of inpatients with depression in the Danish psychiatric health care system (19). The risk of confounding is a concern in our study, as it is in any observational study. The only available covariates in the database were gender and age, and residual and unaccounted confounding from other patient- or hospital-related variables can therefore not be excluded. However, it should be noted that the included process of care performance measures are relevant in principle for all patients with depression, regardless of their characteristics and independent of inpatient volume.

Data validity is always a relevant concern in registry-based studies. The data in the Danish Depression Database are collected by a large number of clinicians during routine clinical work, and registration errors and variation in registration practice can occur. Extensive efforts, however, are made to ensure data validity and uniformity through detailed instructions, with explicit data definitions, standardized registration forms, and systematic structured audit processes conducted on a local, regional, and national basis. The audit processes evaluate the quality of the data and provide feedback to the psychiatric hospital wards (15,19). Moreover, the quality of care was simplified because of the dichotomous data. In a clinical setting, variations may occur in delivering processes of care to patients, but detailed instructions included in the registration forms were intended to reduce such differences. In this study, hospital wards with fewer than 20 recorded admissions during the study period were excluded. This may be a study limitation because the objective was to examine the association between inpatient volume per psychiatric hospital ward and the quality of mental health care. We did, however, conduct a sensitivity analysis, which confirmed the association found in the main analyses. Furthermore, the process of care performance measures in this study have not been shown to be directly associated with clinical outcome and may not necessarily apply to other countries or patient populations. However, the measures do reflect recommendations from a national expert panel, and they are in line with similar standards of care used in other comparable health care systems (20).

Comparison With Other Studies

Among the few existing studies on the topic is a Danish study, which examined the relationship between admission volume per ward per year and quality of mental health care among Danish patients recently diagnosed as having schizophrenia (1). The quality of care in that study was also defined as fulfillment of processes of care that reflected the national guidelines. The study found that patients admitted to very-high-volume psychiatric hospital wards were 1.40 times more likely than patients admitted to low-volume wards to receive a high overall quality of care (≥80% of the relevant recommended processes). Furthermore, patients admitted to very-high-volume psychiatric hospital wards were more likely to receive several of the individual processes of care.

A U.S. study has examined the relationship between inpatient volume per ward and five measures of mental health care quality (8). The study measured the quality of care by seven-day and 30-day follow-up after hospitalization for mental illness, the management of antidepressant medication by prescriptions filled during a 12-week period and for at least six months, and at least three follow-up visits in the 12 weeks after diagnosis of a new episode of depression. For all five measures, patients admitted to wards with a low inpatient volume received poorer quality of care than did those admitted to wards with high inpatient volume.

Other studies have examined the relationship between inpatient volume per psychiatrist and the quality of mental health care by length of stay and readmission (6,7,9,10). The studies found that a high inpatient volume per psychiatrist was associated with both a shorter stay and a higher readmission rate. However, increased length of stay and readmissions do not necessarily indicate poor quality of mental health care. If a psychiatric patient is severely ill or psychotic, an extended stay may be required and readmissions may be essential for stabilization of the patient.

In the study presented here, we examined inpatient volume, defined as the average number of inpatients in each psychiatric hospital ward per year from 2011 to 2016. This measure reflects the experience and capacity of an organization and not of an individual psychiatrist. Other studies have examined the provider-level volume, defined as the total number of mental health inpatients treated by a given psychiatrist (6,7,9,10). The inpatient volume per psychiatrist used in these studies reflects the experience of the individual psychiatrist rather than the organization within which the psychiatrist works, as was used in our study.

Underlying mechanisms in high-volume psychiatric hospital wards, such as specialization, greater clinical experience, and better resources, might explain the observed association between psychiatric hospital wards with very high volume and the highest overall quality of care for patients with depression. Furthermore, a greater number of beds and shorter stays may characterize high-volume psychiatric hospital wards.

On the basis of the results of this study, it is not possible to estimate whether there is an upper limit for how much inpatient volume a psychiatric hospital ward should have. We cannot automatically assume that bigger is always better, because hospitals with very high patient volume may also encounter other challenges (such as insufficient resources and communication difficulties within a very large staff group). This scenario is also illustrated by another Danish study on ward volume of patients admitted for hip fractures. That study found that patients admitted with hip fractures to high-volume wards had higher mortality rates, received a lower quality of in-hospital care, and had longer length of hospital stay, compared with those admitted to low-volume wards (21).

We encourage further studies of the association between inpatient volume and quality of care to confirm the generalizability of our findings for specific mental disorders, including depression. A final question remains about the cost-effectiveness of qualified diagnosis, treatment, and care in very-high-volume psychiatric hospital wards. The cost of providing higher-quality care, health consequences, and specific short- and long-term costs need to be clarified.

Conclusions

This nationwide, population-based cohort study demonstrated that patients with depression who were admitted to psychiatric hospitals with very-high-volume wards were more likely to receive care in accordance with clinical guidelines, compared with those admitted to low-volume wards. Still, for most of the examined process performance measures, the absolute differences were modest, and further studies are necessary to determine the clinical implications of the possible differences in care.