Abstract
The objective was to examine the impact of a multipayer patient-centered medical home (PCMH) on health care utilization for behavioral health patients seen at a tertiary care emergency department (ED). A retrospective health records review was performed for PCMH and non-PCMH patients who presented and received a psychiatric consultation during a 2-year period in the ED of the Mayo Clinic Hospital in Rochester, Minnesota. Univariable and multivariable associations with the outcomes of admission and return visits within 72 hours were evaluated using logistic regression models and summarized with odds ratios (ORs) and 95% confidence intervals (CIs). There were 5398 visits among 3815 patients during the study period. Among these, there were 2440 (45%) PCMH patient visits. There were 2983 (55%) total patient visits resulting in an admission. In a univariable model, PCMH patients (53%) were less likely to be admitted from the ED compared with non-PCMH patients (57%) (OR 0.84; 95% CI 0.76–0.94; P = 0.002) and this remained statistically significant (OR 0.83; 95% CI 0.74–0.93; P = 0.001) after multivariable adjustment. Among the 2415 non-admitted patients, there was no significant difference in returns within 72 hours between PCMH patients (13%) and non-PCMH patients (12%) (OR 1.12; 95% CI 0.83–1.43; P = 0.36). PCMH membership was associated with a lower probability of inpatient hospitalization from the ED. PCMH interventions may be associated with a reduction in health care utilization.
Introduction
B
The 24-hour accessibility of the ED makes it attractive for those who want and need care, including psychiatric patients. According to Hazlett et al, the rate of psychiatric-related ED visits increased by 15% between 1992 and 2000, with the majority of those presenting being on Medicaid. 3 From a health care utilization perspective, costs appear to be accrued from recurrent ED visits and inpatient hospitalizations. Strong primary care has been postulated as central to reducing multiple ED visits and hospitalization for patients with chronic illness.
The patient-centered medical home (PCMH) model has been touted as the new standard of care that enhances linkage between the ED and primary care services. 4 It is defined by the provision of patient-centered, team-oriented, coordinated, and comprehensive care that focuses on the whole patient, including addressing behavioral health needs and conditions. 5 –7 PCMH interventions have been suggested to lead to a reduction in multiple ED visits for uninsured low-income residents in Orange County, 8 and to an increase in virtual visits (eg, telephone, electronic communication) for patients in Group Health 9 in Seattle. Evidence from a systematic review by Jackson et al, though equivocal, suggests that PCMH implementation may be associated with the potential to reduce cost by decreasing unnecessary ED visits and preventing inpatient admissions by employing improvements in patient and staff experiences and preventive care services. 10
On the other hand, research by Werner et al showed that PCMH intervention was not associated with significant changes in health care utilization and cost. 11 Friedberg et al measured associations between participation in a multipayer PCMH pilot in which participating practices adopted new structural capabilities and received National Committee for Quality Assurance (NCQA) certification and found no reductions in utilization of hospital, ED, or ambulatory care services. 12
The PCMH model may provide the appropriate environment necessary to support integrated behavioral health (IBH) services and potentially reduce visits and subsequent boarding in the ED. 13 –15 However, to the best of the authors' knowledge, little is known about the impact of the PCMH on the behavioral health population seen in the ED.
The aim of this study is to examine the impact of a multipayer PCMH on health care utilization for behavioral health patients seen in the ED of Mayo Clinic, Rochester, Minnesota. Analyzing clinical and financial claims data from electronic health records to study the association between PCMH and inpatient admissions and return ED visits for behavioral health patients seen in the ED will add to the evidence on current health care reform initiatives designed to deliver the Triple Aim of lowering cost and improving population health and the patient experience. 16
The research question addressed in this article is: Does PCMH membership impact the probability of inpatient admissions or return ED visits within 72 hours for behavioral health patients seen in the ED?
Methods
Study setting
The Mayo Clinic, one of the most integrated care systems in the world, has been a forerunner in adopting the PCMH model. Employee and Community Health is the primary care service of the Mayo Clinic. It is a recognized PCMH that has been endorsed by NCQA. The Mayo Clinic PCMH is served by a dedicated collocated IBH team comprising psychiatrists, psychologists, social workers, nurse care coordinators, and other allied health staff who are accessible via various mediums including e-consults.
The ED of the Mayo Clinic Hospital, St. Mary's Campus, in Rochester, Minnesota, is a tertiary care academic ED that provides emergency services for PCMH members who require acute treatment for physical and behavioral health conditions. It is the only ED in the area with the capability to do comprehensive psychiatric evaluation; hence, it also serves non-PCMH members who belong to a heterogeneous group consisting of patients without a designated primary care provider and those who are members of practices serving the local neighborhood and regional communities. This diverse group can serve as a comparison group to the Mayo clinic's PCMH for the cohort of patients who present to the Mayo Clinic ED.
Context
As this is an observational study, patients within the community were not assigned by the investigators; rather, this was a natural selection in a real-world setting wherein patients presented to their local ED. Their PCMH status was determined primarily by whether or not they had a primary care provider in the PCMH. The main criteria for having a primary care provider in the Mayo Clinic PCMH was patient choice and whether they live in Rochester or surrounding towns and had no other primary care. Patients with PCMH and non-PCMH membership are treated in a similar manner when they arrive in the ED. A triage process is undertaken by nursing staff and patients who present with behavioral health problems are roomed in the behavioral health wing (or other areas, depending on availability and severity) of the ED. ED providers are then alerted to review the patient, rule out any primary medical etiology, and subsequently determine whether or not a psychiatric consultation is required. If the case is straightforward (eg, anxiety presentations, no safety concerns, obvious intoxication, further psychiatric evaluation unnecessary), the patient may be discharged without psychiatric involvement as determined by the ED consultant. If the ED consultant assesses that a psychiatric consultation is required, the patient is seen by the psychiatry resident on call, who then staffs the case with a psychiatric consultant. The determination of hospitalization–which applies to all patients–is based primarily on severity, risk of presentation, collateral information, and available resources. That said, one central feature of PCMH is care coordination across different specialties within the health care setting.
When a patient from the PCMH is seen in the ED, an automated message is sent to his/her primary care provider. Additionally, during the process of medical record review, the psychiatry resident seeing the patient may send inbox messages to IBH psychiatrists who are collocated with primary care providers. If the PCMH patient is discharged home, primary care providers can discuss these patients with collocated IBH psychiatrists or ask for them to be seen. Other critical IBH team members include care coordinators who can follow up with patients with phone calls and IBH social workers and psychologists who can deliver short-term therapy for PCMH patients.
This care coordination with IBH team members who are collocated with primary care providers is not available to non-PCMH patients. PCMH and non-PCMH members do have access to usual care from providers such as psychiatrists, social workers, and psychologists outside of the IBH team described.
Patient selection
This was a retrospective health records review of consecutive patients who presented to the ED of the Mayo Clinic Hospital, St. Mary's Campus, in Rochester, Minnesota, which is a tertiary care academic ED with 73,000 annual patient visits. Behavioral health patients who presented to the ED between January 1, 2012, and December 31, 2013, and who provided research authorization were included. A review of the electronic medical record and claims data of PCMH and non-PCMH patients who presented and received a psychiatric consultation over a 2-year period was performed. After identifying those who met the inclusion criteria, electronic data were abstracted from the electronic medical record. The use of an electronic tool reduced input error and allowed for easier centralization and access to data.
A coding manual was developed that provided an explicit protocol enabling data abstractors to define, locate, and capture the appropriate variables of interest, including dealing with missing data. These variables included age group, date of presentation, chief complaint, diagnosis, insurance types, disposition, and Mayo Clinic PCMH and non-PCMH membership. Outcome measures for health care utilization in this study were defined as inpatient admission and 72-hour return visits (a frequently cited quality measure) following ED presentation for behavioral health issues. 17 The study was approved by the Mayo Clinic Institutional Review Board. Patients who had declined research consent per Minnesota Statute 144.335 were excluded, irrespective of psychiatric consultation or intervention.
Statistical methods
Continuous features were summarized with medians, interquartile ranges (IQRs), and ranges; categorical features were summarized with frequency counts and percentages. Comparisons of baseline features between patients classified as PCMH and those who were not were evaluated using chi-square tests. Univariable and multivariable associations with the outcomes of admission and return within 72 hours were evaluated using logistic regression models and summarized with odds ratios (ORs) and 95% confidence intervals (CIs). Because some patients had multiple visits during the study, the multivariable models also were evaluated using generalized estimating equations to verify that the associations observed were similar after accounting for any correlation that might occur among visits from the same patient. Statistical analyses were performed using SAS software (SAS Institute, Inc., Cary, NC). All tests were 2-sided and P values <0.05 were considered statistically significant.
Results
There were 5398 ED behavioral health visits among 3815 patients during the study period. Baseline features collected for the 5398 patient visits and comparison by PCMH status is shown in Table 1. The median number of visits per patient during the time frame under study was 1 (IQR 1–1; range 1–21). Median age was 31 years (IQR 20–47; range 4–93). Among the 4381 visits by adults the median age was 36 years (IQR 26–50; range 18–93); among the 1017 visits by children the median age was 15 years (IQR 13–16; range 4–17). There were 2440 (45%) patients classified as PCMH members.
PCMH, patient-centered medical home.
Of the 5398 visits included in this study, 2983 (55%) resulted in a hospital admission. In the univariable analysis (Table 2), PCMH patients (53%) were less likely to be admitted compared with non-PCMH patients (57%) (OR 0.84; 95% CI 0.76–0.94; P = 0.002). A multivariable model accounting for all baseline features studied is shown in Table 3. After adjusting for other features of interest, PCMH patients remained statistically significantly less likely to be admitted compared with non-PCMH patients (OR 0.83; 95% CI 0.74–0.93; P = 0.001). The results were similar after incorporating generalized estimating equations to account for any correlation among visits from the same patient.
CI, confidence interval.
CI, confidence interval.
Overall, 457 (8%) patients returned to the ED within 72 hours. As expected, patients who were admitted to the hospital were less likely to return within 72 hours; 155 (5%) of the 2983 patients who were admitted returned within 72 hours compared with 302 (13%) of the 2415 patients who were discharged (P < 0.001). Therefore, the study team evaluated associations with return visits within 72 hours among the subset of 2415 visits that resulted in discharge (Tables 4 and 5). There was no significant difference between rates of return visits within 72 hours among PCMH patients (13%) compared to non-PCMH patients (12%) (OR 1.12; 95% CI 0.83–1.43; P = 0.36) in a univariable setting; the difference remained non-statistically significant (OR 1.19; 95% CI 0.93–1.53; P = 0.17) after multivariable adjustment. Similar results were observed after incorporating generalized estimating equations to account for any correlation among visits from the same patient.
CI, confidence interval.
CI, confidence interval.
Discussion
Statement of principal findings
The Mayo Clinic's PCMH has been endorsed by NCQA. The major findings from this research indicate that behavioral health patients seen in the ED who were members of the PCMH were less likely to be admitted to the hospital compared to those who did not have PCMH membership. This finding supports the hypothesis that PCMH models may be associated with a reduction in inpatient hospitalization for behavioral health patients seen in the ED. Although the present study involved a different clinical population, this finding is similar to those in the Rosenthal et al study, 18 which showed downward trends in inpatient admissions among patients with diabetes, though trends did not reach statistical significance.
No significant difference was found between PCMH members and non-PCMH members with regard to returns to the ED within 72 hours. However, a reduction in return ED visits also was not found, as was suggested by Roby et al, 8 although that study lacked a comparison group. The findings from this research are similar to those of the Friedberg et al study, 12 which suggest that achieving NCQA certification was not associated with reductions in ED utilization. The research setting is similar to other studies conducted in integrated health systems, such as Geisinger Health 19 and Group Health 9 in Seattle. The generalizability of these findings to less integrated health systems is limited as independent primary care practices do not have the same degree of relationship and shared electronic health records with the ED in their medical neighborhood.
Strengths and limitations
This study is different from other studies examining the PCMH concept in a number of ways. Firstly, it focuses on behavioral health patients, a high-risk group frequently known to have comorbid medical conditions, who typically are high utilizers of health care, particularly recurrent ED users. Advocates of PCMH have stressed that the deployment of PCMH should focus on this particular group of patients. 20 Secondly, this study examines the PCMH concept at the level of the ED, which is different from other research that typically has focused on studying pilot primary care practices undergoing the PCMH transformation to determine its effects on health care utilization compared to practices that have not volunteered to participate in the intervention. 9,11,18 The ED is an essential component of the medical neighborhood that is central to the success of the PCMH.
This study's emphasis at the ED enables examination of the secondary downstream effect of the PCMH model on health care utilization for the cohort of patients who do present to the ED. Particularly questions such as, “If PCMH members do present to the ED, are they more likely to be admitted to the hospital or are they more likely to revisit the ED?” are central to understanding the benefits of the model. Thirdly, unlike other studies that were conducted during the implementation process of the PCMH, the present data extraction and analysis was for a 2-year data period (2012–2013) after the Mayo Clinic PCMH achieved NCQA recognition in 2011 but before implementation of the Affordable Care Act.
The assumption is that the PCMH intervention had already been fully implemented and integrated as “practice as usual” in the primary care practices. This study is unable to comment on whether the PCMH practices are being followed “to the letter” as it focused on ED data only. Fourthly, though other studies identify whole-person orientation as central to PCMH there has been no specific mention of whether behavioral health services were integrated into their interventions. Fortunately, Mayo clinic's PCMH has a strong IBH focus in its practice; hence, the study team is able to examine the downstream effects of the whole-person orientation advocated by the Joint Principles on PCMH and on ED utilization for this group of patients.
A number of factors place limitations on this study. The lack of randomization and a true control group in this study limits the validity of the findings, although PCMH interventions are complex social phenomena requiring a fine balance between achieving control of factors versus allowing the influence of contextual factors in the evaluation process. The use of electronic and medical claims data not originally designed for research had its challenges, including reliance on the accuracy of written records, problems with coding, and the inability to deal adequately with missing data.
This research was not designed to determine whether there was an association between PCMH and inpatient hospitalization for all behavioral health patients who were members of the PCMH. Instead, it focused only on the cohort of behavioral health patients seen in the ED. There is the possibility that PCMH patients were more likely to be admitted directly to the hospital and so avoided the ED. Future research can investigate these gaps further as well as examine the impact of the PCMH on quality and cost outcomes for behavioral health patients who are PCMH members.
The core strength of the PCMH model is integration and coordination. This inherently means that organizations such as Mayo Clinic, which already have an integrated care focus and design, are far more easily able to attain PCMH status. Although this is a strength, it also is a limitation as the findings cannot be generalized to other settings that are not as integrated as the Mayo Clinic. Further randomized prospective studies, preferably in other settings, are needed to determine the true impact of the PCMH. That said, the addition of the collaborative care model to PCMH is rapidly becoming more common, especially because the Centers for Medicare & Medicaid Services has started to reimburse teams for providing IBH services. 21
Implication for clinicians and policy makers
Overall, PCMH membership was associated with a reduction in hospital admissions for behavioral health patients seen in the ED, indicating a possible reduction in health care utilization. Primary care providers should advocate for integration of behavioral health services within PCMH, as this may reduce the need for their patients with behavioral health disorders to go to the ED and provide alternative options of management.
Regardless of whether the health care law is changed, policy makers should support the provisions of the Affordable Care Act, which allowed parity between behavioral health and physical health services, as this would enable behavioral health integration within PCMH to flourish.
Policy makers should support efforts designed to enroll behavioral health patients into PCMH as well as increase community or step-down behavioral health transition programs for patients who are discharged from the ED. Transitions programs have been identified as a key component of the medical neighborhood that is required for the success of PCMH interventions. 22 Policy makers should note that without the inclusion of behavioral health in the current health care reforms, it will be more difficult to deliver whole-person care, which is a central tenet of the PCMH concept 23 and delivering the Institute of Healthcare Improvement Triple Aim of lowering cost, and improving population health and the patient experience. 16
Conclusion
A solid primary care system based on the PCMH model has been publicized as central to reducing costly and unnecessary ED visits and reducing hospital readmissions. Mayo Clinic's Employee and Community Health is an NCQA-certified PCMH. This research examined the association between PCMH membership and inpatient admission and 72-hour return visits for behavioral health patients seen in the ED. The findings indicate that PCMH membership was associated with fewer hospitalizations among discharged patients. This suggests that overall health care utilization may be reduced by PCMH interventions, although transitional care programs may be required to support behavioral health patients who are not admitted to the hospital to prevent repeated ED visits.
Footnotes
Author Disclosure Statement
Drs. Adaji, Melin, Campbell, and Katzelnick, and Ms. Lohse, and Ms. Westphal declared no conflicts of interest with respect to the research, authorship, and/or publication of this article. The authors received no financial support for this article.