Impact of Symptom Remission on Outpatient Visits in Depressed Primary Care Patients Treated with Collaborative Care Management and Usual Care

Introduction

Depression contributes to significant morbidity in the US population, with 8% of persons aged 12 years or older reporting depressive symptoms within the last 2 weeks. ¹ Depression has a significant economic impact and is associated with increased absenteeism and decreased work productivity. ² Likewise, depression contributes to significant health care utilization, serving as the primary diagnosis of 7.9 million ambulatory medical visits annually. ³ The Institute for Healthcare Improvement (IHI) triple aim states that health care reform should focus on improving patient outcomes, improving the experience of health care, and decreasing the total cost of care. ⁴ Thus, changes in practice should consider improvements in clinical outcomes as well as decreased overall health care utilization as mechanisms to control health care costs. Studies have shown that collaborative care management (CCM) has been effective in the treatment of depression by decreasing depression symptomatology. ^{5 –8} CCM also has been demonstrated to improve work-related outcomes (absenteeism and presenteeism) in primary care patients with depression. ⁹

Multiple studies have demonstrated increased health care utilization for individuals with mental health disorders. In a 1996 study of health maintenance organization (HMO) patients who were high utilizers (based on medical costs), those diagnosed with depression were found to accrue greater costs than those without a depression diagnosis when controlling for medical comorbidities. ¹⁰ In a US study of 3 HMO sites that identified high utilizers of outpatient health care (defined by greater than 6 ambulatory visits in each of the previous 2 years), 20% of the high utilizers were found to have current major depression or major depression in partial remission. ¹¹ After controlling for comorbidities, the study found significantly more ambulatory office visits, hospitalized days, and percentage of individuals hospitalized for participants with current depression compared to those with past or no history of depression. In a 2009 study of depressed veterans, those with post-traumatic stress disorder (PTSD) and depression demonstrated significantly more outpatient utilization than those who only had depression. ¹² Although outcomes were not delineated, the group with PTSD had more severe depression symptoms and other comorbidities. In addition, a self-reported patient survey study of urban family medicine practices found a positive association between mental health conditions and increased utilization of health care services. ¹³ The authors' prior studies of primary care patients in CCM also have shown that those patients who have increased depression severity and other comorbidities tend to have worse clinical outcomes. ^{14 –16}

Previous studies have demonstrated that depression treatment makes a positive impact on health care costs and outpatient utilization. A review of 9 studies in Washington State, including 1814 patients meeting criteria for depression, found mean health service costs 6 months after treatment to be significantly less for those achieving remission ($2012) compared to persistent depression ($3094, P<0.001) or partial remission ($2571, P=0.007). ¹⁷ Of note, these studies utilized classic depression treatment strategies with no CCM. As the authors have noted in a prior study, those patients whose depression improved to remission while in CCM had decreased odds of outlier status for outpatient utilization at 6 months (defined as >8 visits) and 12 months (defined as >12 visits) compared to those who were not in remission. This was demonstrated while controlling for baseline severity of depression, clinical diagnosis, anxiety symptoms, and prior utilization patterns. ¹⁸

Although CCM has been shown to be effective in the treatment of depression and that achieving depression remission is correlated with decreased outpatient utilization, some studies have demonstrated increased short-term costs after implementation of CCM. A randomized controlled study of 407 patients with depression demonstrated an increase in total medical costs 12 months after enrollment in a CCM program compared to usual care (UC). ¹⁹ However, a more recent randomized controlled study of 551 patients diagnosed with major depressive disorder demonstrated a long-term cost reduction with CCM compared to UC over 4 years. ²⁰ The increase in short-term cost is thought to be related to patient “activation,” or increased patient engagement with treatment. ^20,21

In the present study, the authors compared outpatient visit counts of patients with depression in CCM to patients with depression receiving UC while controlling for comorbidities and prior outpatient visit patterns. Based on the authors' prior results that clinical remission was associated with fewer outpatient visits at 6 and 12 months in patients enrolled in CCM, ¹⁸ their hypothesis was that patients with poor depression outcomes at 6 months would have increased outpatient visit counts, regardless of treatment type.

Methods

Primary care patients from 4 clinics in a large integrated practice (∼100,000 adult patients) in Rochester, Minnesota comprised the study group. Those who were diagnosed with major depressive disorder or dysthymia from March 1, 2008 through December 31, 2011, had a Patient Health Questionnaire (PHQ-9) ²² with a score of ≥10, and had a complete set of study data were included. The PHQ-9 is a 9-item patient assessment tool that scores each item from 0 to 3. A PHQ-9 score of <5 is classified as asymptomatic; 5–9 as mild depressive symptoms; 10–14 as moderate depressive symptoms; 15–19 as moderately severe; and ≥20 as severe symptoms of depression. The study cohort included 1733 patients. Treatment was by UC with the patient's primary care provider (PCP) or enrollment in CCM. The only exclusionary criterion was a clinical diagnosis of bipolar disorder.

Patients in the UC arm were treated (medical and/or therapeutic) and had follow-up as directed by their PCP. For those enrolled in CCM, a registered nurse care manager reviewed the patient with the supervising psychiatrist, who met with the care managers on a weekly basis. Recommendations regarding therapeutic changes were communicated to the patient's PCP. All treatment decisions were the responsibility of the PCP. Care manager to patient contact varied based on clinical need. Full details of the development and implementation of this study's CCM have been described previously. ^21,23,24

The dependent variable for this study was outlier status (>80th percentile) for the number of outpatient visits at 6 months. Outpatient visits included those for all concerns, not just behavioral health visits. The independent demographic variables were age, sex, race, and marital status. The clinical predictor variables were baseline PHQ-9, clinical diagnosis, and remission status at 6 months (PHQ-9<5). Prior studies in the authors' institution have demonstrated that outpatient visits were impacted by the individual patient's prior visit counts. ^{25 –29} In order to control for this, the number of outpatient visits (primary, secondary, or tertiary care) for the 6 months prior to enrollment also was measured. The presence of high-risk clinical comorbidities also was recorded, as defined by Colla et al in 2012. ³⁰ A diagnosis of diabetes, cancer, chronic obstructive lung disease, dementia, or coronary artery disease was determined by screening of the electronic medical record. Following Rohrer et al, the number of comorbidities was categorized as none, 1, or ≥2. ³¹

Categorical data were analyzed with chi-square testing while Mann-Whitney testing was utilized for statistical analysis of the continuous variables (because of non-normal distribution). Multiple logistic regression modeling for outpatient visit outlier status was performed while retaining all independent variables studied. Calculations were performed on MedCalc software (<www.medcalc.org>, version 12.3.0.0). This study was reviewed and approved by the authors' institutional review board.

Results

Of the 1733 patients included in the study, 819 (47.3%) were in remission at 6 months and 914 (52.7%) were not (Table 1). At baseline, the remission group was slightly older, more likely married, with a diagnosis of first episode of major depression of moderate severity. The remission group had fewer outpatient visits compared to the group that was not in remission, and were less likely to be a prior outpatient visit outlier(>6 visits) for the 6 months prior to diagnosis.

For the 6 months after diagnosis, the remission group demonstrated a significantly lower number of outpatient visits compared to those patients not in remission. Outpatient visit outlier status was defined as >80th percentile (>8 visits) for the 6 months following diagnosis. Patients in remission were approximately 50% as likely to be an outlier as those who were not in remission (Table 1).

Figure 1 demonstrates the multiple logistic regression analysis for outpatient visit outlier status (>8 visits) at 6 months after diagnosis. Severe depression (odds ratio [OR] 1.775, confidence interval (CI) 1.259–2.503, P<0.01) and increased pre-diagnosis outpatient utilization outlier status (OR 4.892, CI 3.655–6.548, P<0.01) were independent predictors for outpatient visit outlier status. In this model, remission at 6 months demonstrated an OR for outpatient visit outlier status of 0.609 (CI 0.461–0.805, P<0.01), while controlling for age, sex, marital status, clinical diagnosis, enrollment into CCM versus UC, and clinical location (for site-to-site variability in outcomes). The OR of those patients not in remission at 6 months having outpatient visit outlier status was the inverse of this at 1.643 (CI 1.243–2.173).

OPEN IN VIEWER

FIG. 1.

Multiple regression model for odds ratio of outpatient visit outlier status (>8 visits) 6 months after diagnosis of depression, by variable. Multiple regression modeling with all other variables, controlling for age, sex, race, marital status, clinical diagnosis, medical comorbidity, enrollment into collaborative care management versus usual care, and clinical location (to control for site-to-site variability in outcomes). Model overall fit with null model −2 log likelihood of 1701.742 (P<0.0001). PHQ-9, Patient Health Questionnaire.

Discussion

In patients diagnosed with depression, the number of all-cause outpatient visits after the diagnosis of depression was significantly impacted by the severity of depression as well as the number of outpatient visits prior to diagnosis. Clinical remission at 6 months, as suggested by the original hypothesis, was associated with lower odds of having more than 8 visits during the 6 months after diagnosis. This finding was demonstrated while controlling for patient comorbidities, depression severity, and clinical diagnosis. It was noted that a simple variable with 3 categories proved to be useful for comorbidity adjustment, as was demonstrated previously in a study of polypharmacy. ³¹

In addition, the authors adjusted for the likelihood to use medical care by including the number of visits in the previous year as a covariate, as was done in their previous work. ^{25 –29} This variable reflects both the biomedical need for services and any psychological inclination to seek care. By adjusting for prior use of service, the authors were able to separate out how much depression remission changes use of service above and beyond the normal pattern.

This study is validated by its consistency with other studies that demonstrated similar findings. For instance, it was shown in a prior study that achieving remission status in CCM was inversely related to high outpatient visit counts 6 and 12 months after enrollment. ¹⁸ In the present study, the authors find that this continues to be the case for all patients (CCM or UC) who had a 6-month follow-up PHQ-9 score for their depression. Further integration of behavioral health services continues to be strongly emphasized in Patient-Centered Medical Home standards ³² and the results of the present study support early intervention and effective treatment of depression in primary care patients as a mechanism to improve clinical outcomes and decrease all-cause outpatient visit utilization.

Future studies could examine long-term effects of CCM for depression on outpatient visits, as previous studies have suggested economic benefits beyond this study's follow-up time period. Future studies of the total health care costs by per member per month (PMPM) in individuals with depression within the CCM program also would be interesting. A 2011 study of PMPM costs for patients with bipolar disorder compared to other chronic conditions found that only those with both diabetes and coronary disease had higher total PMPM costs. Also, patients with depression had higher adjusted PMPM costs for primary care and nonpsychiatric inpatient services. ³³ In addition, further investigation into the effect of patient complexity on outpatient utilization may be made by examining alternative measures or study design. For example, the case-control design can be an efficient approach to studying the outcomes of practice innovations. ³⁴

Limitations of this study include the lack of racial and economic (such as health insurance coverage) diversity in the population. Because this study encompassed 1 large, midwestern US medical practice and the implementation of CCM may be different in other organizations, the results may not be generalizable to other clinical environments. However, patients in the present study were treated in several clinical locations, assuring some internal variation in practice styles, attitudes, and other unmeasured variables. An additional limitation could include potential referral bias because this cohort study did not randomize patients into CCM or UC. Also, CCM patients had more consistent depression severity follow-up than UC, which may have produced a bias in patient selection.

Conclusions

In primary care patients treated for depression, successful treatment to remission (PHQ-9 score <5) at 6 months decreased the likelihood of the patient having more than 8 visits during the 6 months after diagnosis. This holds true even when controlling for the individual patient's prior outpatient visit counts, health care comorbidities, and enrollment in CCM versus UC. Although this is encouraging for this developing model of care, future studies are encouraged.