Transitions between inpatient and outpatient health care settings are associated with elevated risks of adverse events and, therefore, are a focus of quality improvement initiatives (1). In the 30 days after discharge from a psychiatric hospitalization, approximately 10%–15% of patients are readmitted, and suicide rates are more than 100 times higher than in the general population (2,3).

Outpatient mental health care after a psychiatric hospitalization has been associated with reduced incidence of hospital readmission and could potentially reduce the risk of adverse outcomes (4). The National Committee for Quality Assurance, therefore, includes outpatient mental health follow-up within seven days of discharge from a psychiatric hospitalization as a quality measure in the Healthcare Effectiveness Data and Information Set (HEDIS) (5). To understand the clinical utility of applying health system resources toward improving this measure, it is important to assess whether timely outpatient mental health follow-up corresponds with greater receipt of evidence-based treatments or fewer adverse outcomes.

In 2008, the Veterans Health Administration (VHA) implemented a policy mirroring this HEDIS quality measure. All patients discharged from an inpatient mental health setting were required to have a follow-up outpatient contact within seven days. In 2009, VHA adopted this measure as a quality indicator to evaluate its medical centers and regional networks. These policy changes provide an opportunity to evaluate whether improved performance in providing seven-day follow-up visits is associated with improvements in other care processes and outcomes.

Prior research has demonstrated a spillover effect (also referred to as a “halo effect”) of performance monitoring, suggesting that focused improvement in one aspect of treatment may benefit other aspects of care for the same disorder (6). Therefore, we hypothesized that after implementation of the policy changes, performance would improve on the seven-day follow-up quality measure and there would be a statistically significant spillover effect on other quality measures.

To test our hypothesis, we measured performance on the seven-day follow-up measure among patients hospitalized for major depression. Patients with major depression are at particularly high risk of suicide after hospitalization, and quality indicators for pharmacologic and psychotherapeutic treatment of major depression have been established (7). Combined antidepressant medication and psychotherapy treatment for severe depression is recommended by evidence-based practice guidelines (8). We evaluated whether performance on the seven-day follow-up measure corresponded with these treatment indicators and with hospital readmissions, an indicator of recurrent severe illness.

The study cohort consisted of patients who received a clinical diagnosis of major depressive disorder (ICD-9-CM code 296.3 or 296.2) during an inpatient psychiatric hospitalization within the VHA between fiscal years 2005 and 2010. We excluded patients diagnosed as having bipolar disorder, schizoaffective disorder, or schizophrenia because measures of antidepressant coverage and psychotherapy use may not apply to them. Patient data, including demographic characteristics, diagnoses, service use, and pharmacy records, were obtained from the VHA’s National Registry for Depression, which includes all patients receiving treatment for depression within the VHA. The study was conducted with approval from the Veterans Affairs Ann Arbor Healthcare System Institutional Review Board.

For each fiscal year of the study period, we selected the patient’s first discharge from a psychiatric inpatient stay that included a diagnosis of major depressive disorder as the index discharge date for that year. Individual patients could be included in multiple years but not more than once in any year. We assessed four posthospital care measures subsequent to the patient’s index discharge date: timely outpatient follow-up, readmission, antidepressant treatment, and receipt of psychotherapy.

Timely outpatient follow-up was defined as an outpatient mental health contact within seven days of the index discharge date. Mental health contacts were identified by Decision Support System identifiers and could include any type of outpatient mental health service, such as medication management or psychotherapy, including a telephone contact. Patients who were readmitted within seven days of discharge were excluded from analyses of this measure.

Hospital readmission for a diagnosis of depression within 90 days of discharge was measured. The 90-day period was chosen because shorter periods may be more influenced by inpatient treatment decisions, such as the quality of discharge planning, than by outpatient monitoring and care.

Antidepressant treatment adequacy was defined as receiving at least 72 days’ supply of an antidepressant during the initial 90 days following the index discharge date and having at least one medication refill after the initial discharge prescription. The refill requirement prevented one large initial prescription from automatically satisfying the measure. Receiving medication for at least 72 of 90 days corresponds to a medication possession ratio of 80%, a cutoff frequently used for adequacy of medication receipt (9). A similar measure has been associated with decreased psychiatric hospitalizations among VHA outpatients initiating depression treatment (10).

We defined adequate psychotherapy as the receipt of eight individual or group psychotherapy visits (Current Procedural Terminology [CPT] codes 90804–90815, 90845, 90847, 90849, 90853, and 90857) during the 90 days following the index discharge date. These CPT codes reflect a broad spectrum of psychotherapy modalities, including behavior-modifying, insight-oriented, supportive, and family therapy. The cutoff of eight visits for adequate treatment was chosen on the basis of manualized clinical trials of psychotherapy for depression and prior retrospective studies of psychotherapy adequacy (11,12).

We included covariates shown in prior studies to be related to receipt of depression care within the VHA (12,13). These included age group (<35, 35–49, 50–64, and ≥65 years), gender, race (white, black, other, and unknown), Hispanic ethnicity, marital status (married or unmarried), and distance from nearest VHA facility (<30 or ≥30 miles). Age was categorized because prior work among veterans demonstrated nonlinear relationships between patients’ age and depression care.

General medical comorbidity was measured by using an enhanced coding algorithm for the Charlson Comorbidity Index. Scores were categorized as 0, 1 to 2, and 3 or more, with higher scores indicating a greater number or severity of comorbid medical conditions (14). Comorbid mental health conditions diagnosed during the index hospitalization were measured and included posttraumatic stress disorder (PTSD), other anxiety disorders, substance use disorders, and personality disorders. As in prior studies, length of stay of the index hospitalization was categorized as fewer than seven days, seven to 14 days, 15 to 30 days, or more than 30 days (7). We also included whether patients had a mental health visit, received antidepressant treatment, or received psychotherapy in the 90 days prior to the index hospitalization, given that these have been shown to predict posthospital care (12). We included fiscal-year quarter as a covariate to account for potential seasonal variation in treatment.

For each fiscal year, we calculated the frequency of a seven-day follow-up visit after discharge and whether patients were rehospitalized, received adequate antidepressant treatment, or received adequate psychotherapy within 90 days of discharge. Chi square tests tested for differences in outcomes across years. To adjust for potential confounding influences of changes in demographic characteristics, such as the increasing proportion of younger veterans among those receiving VHA care, or clinical characteristics of the population, the multivariable logistic regression models that were used to predict each outcome included year of discharge and patient characteristics as covariates.

We conducted two subanalyses regarding characteristics of the seven-day follow-up visit. VHA policy allows a telephone encounter to satisfy the seven-day visit requirement (although an in-person visit must occur within 14 days of discharge). To explore care received after a seven-day follow-up telephone encounter, we conducted separate logistic regression models predicting rehospitalization, antidepressant treatment, and psychotherapy among patients with a follow-up visit within seven days and included as an additional covariate whether the first encounter was by telephone. In a separate set of regression models, we explored continuity of care from before to after hospitalization by including whether a patient was seen by the same mental health provider within seven days of hospital discharge and in the 90 days prior to hospitalization. Alpha was set at .05 for all comparisons. All analyses were conducted in SAS, version 9.2.

A total of 56,785 unique patients had a psychiatric inpatient stay for a major depressive disorder between 2005 and 2010. There were 68,849 hospitalizations, including those of patients who were hospitalized on separate occasions during more than one fiscal year. The total cohort had a mean age of 51.5±13.1 years, and was 89% male, 72% white, 21% black, and 5% Hispanic. Characteristics that had a relative change of 10% or more between 2005 and 2010 included age younger than 35 years (9% versus 14%, respectively), age 35–49 years (31% versus 24%), unknown race (3% versus 7%), living more than 30 miles from a VHA clinic (10% versus 9%), PTSD (35% versus 43%), other anxiety disorders (15% versus 17%), Charlson score of 3 or more (4% versus 6%), length of hospital stay of 15 to 30 days (15% versus 13%), a mental health visit prior to hospitalization (59% versus 71%), antidepressant medication treatment prior to hospitalization (61% versus 68%), and psychotherapy prior to hospitalization (39% versus 51%).

The percentage of patients receiving a seven-day follow-up mental health visit increased from 39% in 2005 to 75% in 2010 (χ²=5,296.9, df=5, p<.001). Similar percentages of patients were rehospitalized within 90 days of discharge in 2005 (14%) and 2010 (15%) (χ²=15.3, df=5, p=.009). During the same period, the percentage of patients who received adequate psychotherapy increased from 14% to 19% (χ²=195.8, df=5, p<.001) and the percentage who received adequate antidepressant treatment increased from 59% to 62% (χ²=29.5, df=5, p<.001). After adjustment for demographic and clinical characteristics, patients had 4.86 greater odds (95% confidence interval [CI]=4.56–5.17) of receiving a seven-day follow-up mental health visit and 1.29 greater odds (CI=1.19–1.40) of receiving adequate psychotherapy in 2010 compared with 2005. However, there were no significant differences in the rates of rehospitalization or adequate antidepressant treatment over time (Table 1).

The percentage of patients who completed a seven-day follow-up encounter whose first encounter was by telephone increased from 14% in 2005 to 33% in 2010. After adjustment for other patient characteristics, patients who completed a telephone encounter were no more likely to be rehospitalized than those who did not receive their first encounter by telephone. However, patients whose first encounter was a telephone call were significantly less likely to receive adequate psychotherapy (odds ratio [OR]=.36, CI=.33–.39) and adequate antidepressant treatment (OR=.78, CI=74–.83) than patients whose first posthospital encounter was in person.

Among patients who had a mental health visit both 90 days prior to their hospitalization and within seven days of their discharge, the percentage whose first postdischarge visit was with a provider they had seen before the hospitalization ranged from 60% in 2006 to 53% in 2009. After adjustment for other patient characteristics, patients who saw the same provider before and after their hospitalization were no more likely to be rehospitalized or receive adequate antidepressant treatment than patients who did not see the same provider, yet patients who had continuity of care with the same provider before and after their hospitalization were more likely to receive adequate psychotherapy (OR=1.29, CI=1.22–1.38).

Subsequent to changes in health system policy starting in 2008, we observed a rapid change in performance on timely outpatient mental health follow-up after psychiatric hospitalization. However, there was no concurrent reduction in readmission after discharge or improvement in antidepressant treatment. Aside from a modest increase in receipt of adequate psychotherapy, which may be attributed partially to simultaneous health system efforts to improve access to evidence-based psychotherapies (15), we did not find evidence to support the hypothesized spillover effect.

Although the seven-day follow-up measure may not be intended specifically to reduce readmissions or improve engagement in evidence-based treatments for depression, the lack of a robust association between timely follow-up and the other outcomes indicates that the seven-day follow-up quality measure should not be considered a proxy for the quality of posthospital care more broadly. The lack of a spillover effect may relate to the fact that the current measure and the VHA requirements do not specify the therapeutic elements to be delivered. If they occur primarily to satisfy the measure, follow-up visits may be brief and contain minimal therapeutic content. The visits may also focus on the patient’s immediate safety or other acute concerns but may not adequately address subsequent treatment engagement. Additional research exploring the content of posthospital follow-up visits would help to clarify how this measure is implemented in practice and why it appears to have little correlation with other aspects of care.

To meet the seven-day follow-up requirement, facilities appear to have instituted more phone-based follow-up. Our finding that patients who received telephone follow-up were less likely to receive adequate psychotherapy or antidepressant medication suggests that these patients either have greater barriers to treatment or that telephone follow-up is less effective; on the basis of the latter possibility, in-person visits—when feasible—may be preferable. Continuity of care with the same provider before and after hospitalization may be advantageous primarily for delivery of adequate psychotherapy during the posthospital period.

The generalizability of our findings may be limited, given that VHA patients are demographically and clinically distinct from depressed patients in the general population and may have different preferences or barriers related to depression care. Our findings may be more generalizable to other integrated health systems that provide comprehensive mental health services and set performance measure goals for quality improvement.

These findings should not be interpreted to suggest eliminating requirements for close outpatient follow-up after a psychiatric hospitalization. We could not measure the utility of the seven-day follow-up visit in improving other important patient outcomes, such as suicide-related behaviors, depression symptoms, functioning, or quality of life. We cannot exclude the possibility that treatment and rehospitalization outcomes would have worsened over time were it not for improved seven-day follow-up, although the stable trend in these measures prior to 2008 suggests this is unlikely.

Timely outpatient mental health follow-up after hospitalization may be insufficient to reduce readmission or substantially improve longer-term depression treatment. Nonetheless, it may be possible to make more effective use of immediate posthospital follow-up visits to engage patients in care and reduce the likelihood of readmission. These findings should focus attention on how to improve the efficacy of immediate posthospitalization follow-up. In addition, other measures of evidence-based care beyond the initial follow-up visit should be considered when assessing posthospital care quality.

The authors report no competing interests.