Patients seeking mental health care have considerable unmet needs, and patients with mental illnesses are more likely than other patients to have general medical illnesses (
1). Studies of older patients suggest that they receive inadequate screening, diagnosis, and treatment for mental illnesses and that they prefer collaboration between their primary care provider (PCP) and mental health provider (MHP) (
2,
3). Mental health care is often provided in a fragmented system involving multiple providers who do not coordinate services (
4). Even for persons with severe mental illness, MHPs and PCPs have reported inadequate provider-to-provider communication (
5,
6).
Inadequate continuity of information between MHPs and PCPs is a particular concern for Medicare Advantage (MA) plan enrollees, for whom mental health services may be a carve-out that is separated from other forms of medical care. In this case, there is no consultant relationship between the MHP and PCP, and real and perceived barriers affect transfer of information between MHPs and PCPs. Financial barriers, differences in culture and practice between MHPs and PCPs, training issues, and barriers posed by information systems and privacy protections also contribute to discontinuity of information between MHPs and other clinicians caring for these patients (
7,
8).
The need to be aware of potential disease-drug and drug-drug interactions, as well as to monitor treatment adherence and condition severity across providers, further supports the critical importance of communication between MHPs and PCPs. For example, certain psychotropic medications, such as second-generation antipsychotics, are associated with higher risk of metabolic syndromes (
9). However, evidence suggests that persons with mental illnesses are less likely to receive appropriate testing (
10) and adequate intervention and monitoring (
11).
Interactive communication can improve the effectiveness of PCP-specialist collaboration (
12). Tested strategies between MHPs and PCPs include “as needed” communication through telephone calls and written documents (
13–
15), structured telephone and written communication (
16,
17), shared electronic records and reports (
18–
20), and more complex interventions, including joint case review, alert systems, and collaborative care (
12,
16,
21–
23). These interventions have enjoyed variable degrees of success in improving communication and clinical outcomes. Central to this work is the understanding that a shared clinical record can substantially contribute to continuity between MHPs and other continuity providers (
24,
25).
Despite the importance of communication between MHPs and PCPs, little work has explored the actual level of continuity between these clinicians. Continuity includes many domains (
26,
27), including relational continuity (ongoing relationship over time), management continuity (whether care received from various providers is connected in a coherent way), and information continuity, which is the focus of this study. As part of a larger effort to develop and test measures for the Centers for Medicare and Medicaid Services, we explored the level of communication after a new mental health consultation and information continuity concerning pharmacologic treatment of a mental health condition. We also explored the potential implications for continuity of a mutual-access electronic health record (EHR) that is shared between MHPs and PCPs.
Methods
Six geographically diverse MA plans of various sizes (two MA Part D preferred provider organizations, a dual-eligible special needs plan, and three health maintenance organizations) two in the South, one in the Midwest, two in the West, and one in multiple regions, collected medical record information about patients referred for a new mental health consultation in order to evaluate the continuity of information between the MHP and the PCP. The work carried out for this study underwent review by the RAND Human Subjects Protection Committee.
Patient Sample
Eligible patients were enrolled in the plan during 2011 and 2012 with not more than a 45-day disenrollment in either year, were not in a nursing home for more than 14 days in 2012, had no visit with any MHP in the outpatient or emergency department (ED) setting in 2011, and had a visit with their PCP between January 1, 2012, and December 31, 2012. Among this group, patients were eligible for inclusion if they had a visit with an MHP in the outpatient or ED setting between January 1, 2012, and September 30, 2012, and the visit was not for substance abuse treatment or neuropsychological testing. Each plan aimed to randomly choose 50 patients with a “new” mental health visit and a PCP visit in 2012.
Medical Record Abstraction
For each patient in the sample, participating plans requested medical records from the patient’s MHP and PCP. The PCP was the individual designated in that role by the MA plan. If a patient saw more than one MHP during the study period, then the MA plan chose the first MHP with a visit (because this was the initial mental health consultation during the eligible period). Plans requested the complete mental health record and the complete PCP record for care provided in 2012. Plans assembled abstractors with quality management and medical record abstraction experience who were trained to use an abstraction instrument developed for this study (available from the authors on request). Abstraction required that both the MHP and the PCP medical records were received.
The medical record abstractors recorded the number of mental health and primary care visits during the year-long study period and also whether the providers used a mutual-access EHR, which was defined as a single, centralized EHR that both the MHP and the PCP used to document the patient’s care. Records were abstracted to identify whether during the study period there was any indication (for example, notes and consultation reports) that the PCP was aware of a second source of care from an MHP and, if so, the first date of awareness. Abstractors also collected evidence from the PCP or MHP record of any contact between these two providers and whether this was by telephone, e-mail, or other correspondence. A plan for (or attempt at) a telephone call or evidence of a mental health consultation note or report in the PCP record was also collected. Abstractors also looked for documentation in either record that the patient refused to consent to the provision of information to the PCP about his or her mental health care or provision of information to the MHP from the PCP. The mental health record was evaluated for evidence of an ED visit or hospitalization for a mental health–related reason during 2012 and, if so, whether the primary care record indicated that the PCP was aware of the admission or visit.
Medical records were also abstracted for continuity of information about psychotropic medications. Abstractors recorded psychotropic medications (specifically, antidepressants, antipsychotic agents, antianxiety agents, sedatives, and hypnotics) listed in the mental health record at the first MHP visit. At the subsequent PCP visit, abstractors noted whether each medication was identified on the medication list or in a note. If the medication was not listed, any explanation or justification noted in the record for this discrepancy (for example, “discontinued fluoxetine due to side effects,” “was seen in ED yesterday and meds discontinued”) was recorded. If a medication was listed in the PCP record, abstractors recorded whether the dosing information was accurate (for total daily dosage). If a subsequent MHP visit in 2012 was followed by a PCP visit, information from these visits was also abstracted for psychotropic medication continuity.
A second abstractor conducted an interrater reliability assessment on a 20% random sample of completed charts. The weighted pooled kappa (
28) across reabstracted records was .86, representing very good agreement.
Analyses
We evaluated the number of patients for whom plans were able to retrieve medical records to conduct abstractions. We report the overall continuity of care between MHPs and PCPs in order to compute the percentage of patients for whom there was communication between these clinicians within three months of the first MHP visit.
For each hospital admission and ED visit for a mental health condition, we evaluated whether medical record documentation reflected evidence of PCP awareness of the admission or visit. These data were used to score the Assessing Care of Vulnerable Elders (ACOVE) continuity quality measure (
29) that states, “IF a vulnerable elder is treated at an ED or admitted to a hospital, THEN there should be documentation (during the ED visit or within the first 2 days after admission) of communication with a continuity physician, of an attempt to reach a continuity physician, or that there is no continuity physician.”
Concerning psychotropic medications present in the MHP record, we computed the medications for which there was continuity with the PCP at the following visit. We evaluated whether psychotropic medications were listed in the PCP record with the correct dosage amount or an explanation of the reason why the medication or dosage was not recorded. For providers using a mutual-access EHR, it was possible to abstract only whether the medication list including psychotropic medications had been accessed by the PCP during the period after the mental health visit; therefore, for patients with mutual-access EHRs, full credit for reviewing all psychotropic medications was conferred if the medication list was reviewed. These data were used to score the ACOVE continuity measure: “IF a vulnerable elder is under the outpatient care of ≥2 physicians, and one physician prescribed a new chronic disease medication or a change in prescribed medication, THEN the nonprescribing physician should acknowledge the medication change at the next visit.”
Continuity of care, awareness of hospitalizations and ED visits, and medication continuity were compared between patients in the one MA plan with a mutual-access EHR versus all other patients by using chi-square tests. All analyses were conducted by using Stata version 12.1.
Results
Four of the six MA plans identified 50 patients meeting study criteria. Overall, 249 patients were identified who received a new outpatient mental health consult during the first nine months of 2012. Among these patients, plans were unable to retrieve MHP records for 44 patients, plans could not obtain PCP records for 13 patients, and for 17 patients neither record was available. For one patient, an MHP record was retrieved but not from the primary MHP. For 30 patients, the original mental health consultation occurred in the inpatient setting or there was no identifiable outpatient mental health visit in 2012, one patient was found to be in long-term care, and two patients were excluded for unknown reasons. Thus 141 medical record abstractions were completed.
The 141 patients with a new outpatient mental health consultation had a mean of 4.2 mental health visits in 2012 (range 1–30) and a mean of 5.5 primary care visits in 2012 (range 1–20). For only 30 patients (21%) was there any indication on the PCP record of communication from the MHP within three months of the initial consultation. Among these 30 patients, for eight there was evidence of a telephone call, e-mail, or other correspondence between the MHP and the PCP; for four there was an indication of a plan for or attempt at a telephone call; for 11 there was a mental health consultation note or report in the PCP record; and for ten patients with a mutual-access EHR, the PCP reviewed the mental health note within the record. Multiple modes of contact between PCP and MHP were possible. For an additional five patients, there was an indication of PCP awareness of the MHP consultation, but this occurred more than three months after the new mental health consultation visit.
Seventeen of the 141 patients had a total of 24 mental health admissions and two ED visits not resulting in admission over the study period (
Table 1). Documentation of communication with the PCP was noted within the first two days after hospital admission or during the ED visit (or documentation of an attempt at contact or that there was no PCP) for 11 of these 26 encounters within one week of the encounter; for 17, such documentation was found in the PCP record by the end of the study period.
Medication Continuity Between MHP and PCP
Seventy-eight of the 141 patients were receiving 152 medications according to the MHP record during the study (
Table 1). For 103 of those medications, there was acknowledgment in the PCP record of the new medication (or for patients with a mutual-access EHR, the medication list that included psychotropic medications was accessed by the PCP at the follow-up visit).
An analysis that focused on only the 126 medications of the 54 patients whose physicians did not use a mutual-access EHR found that 50% of the psychotropic medications were antidepressants, 27% were benzodiazepines and hypnotics, and 12% were antipsychotics (
Table 2). For 40% of the medications, there was continuity concerning medication name and dosage in the PCP record; for 21%, the name of the medication but not the correct dosage was listed in the PCP record; and for 39%, there was no mention in the PCP record. Discontinuity between the MHP and PCP records was found for seven of 15 (47%) antipsychotic medications and 13 of 34 (38%) hypnotic-benzodiazepine medications.
MHP-PCP Continuity With and Without a Mutual-access EHR
One plan had a mutual-access EHR in which MHPs could see all other health information and PCPs were able to access mental health information. In addition, for one patient in another plan, the MHP and PCP used a mutual-access EHR. PCPs accessing mental health information within the mutual-access EHR needed to provide a password to see that information, and thus access specifically to mental health information could be identified during medical record abstraction. When PCPs accessed mental health records, access to medication could be evaluated only at the level of the medication list, and thus continuity of medication could be calculated only at the patient level, rather than at the medication level. Data in
Table 3 show that information continuity between MHPs and PCPs was better in the plan with the mutual-access EHR, compared with the other five plans without a mutual-access EHR. Forty-six percent of PCPs in the plan with the mutual-access EHR “received” information from the MHP (which could have been a communication from the MHP or a copied note or evidence that the PCP accessed the MHP record), compared with 11% of providers in the plans without a mutual-access EHR (p<.001). Information continuity regarding medication that was started by the MHP occurred for 100% of the patients in the plan with the mutual-access EHR compared with 57% of patients in plans without a mutual-access EHR (p<.001). No difference in communication with PCPs about mental health admissions and ED visits was found.
Discussion and Conclusions
This small but detailed study of patients in MA plans found poor information continuity between MHPs and PCPs. Disrupted continuity for patients receiving initial mental health consultations is particularly concerning because these patients may be at risk of not following up with treatments prescribed by MHPs and may be started on therapies that might interfere with medications and treatments initiated in the primary care setting. Although numerous organizations, including the Institute of Medicine (
30), the Health Resources and Services Administration (
31) and the Agency for Healthcare Research and Quality (
32), have called for greater integration of mental health care and primary care, the results of this investigation suggest that mental health and primary care are largely separate endeavors, at least in five of the six MA plans studied.
Undertaking a medical record review to explore continuity between two different providers is a complex and resource-intensive task. As seen in this implementation in six MA plans, a full set of records could be retrieved for fewer than half the projected number of patients. There are considerable legal barriers in some states to obtaining mental health records (
33), in addition to the perception of many behavioral health providers that their records may not be released without written consent even for continuity of care, although this appears to be an inaccurate perception of the legal rules in most states (
34). However, detailed review of medical records indicated that clinically relevant information, which is fundamental to information continuity, was lost between mental health and primary care settings. Information about mental health hospitalizations was transmitted (in a documented way) to PCPs in a timely fashion for fewer than half of such hospitalizations, and PCPs appeared to be unaware of the initiation of 39% of psychotropic medications, including antipsychotic and benzodiazepine medications. These findings suggest significant unmet need that would not be identified by using less intensive evaluation techniques.
It seems axiomatic that continuity would be enhanced when MHPs and PCPs have mutual-access to a combined set of records. However, under such circumstances, PCPs did not access mental health records even half the time, although information continuity was far better than among providers without a mutual-access EHR. In all cases, PCPs in the plan with the mutual-access EHR accessed the medication list that included the mental health medications, making this medical record construction a “favorable default” for ensuring information continuity. However, it bears noting that not all so-called mutual-access EHRs offer PCPs access to mental health records; these data suggest that limiting such information would impede continuity of care.
This study had several limitations, including a small sample and the small number of plans participating, potentially limiting the generalizability of findings. The records received by plans represented only half the patients identified. Furthermore, no clinical information was collected that might provide context for the study findings. PCPs may have interacted with MHPs but may not have documented this in the record. In addition, the nature of the consultation (one time or for ongoing care) was not known. In addition, all but one patient with a mutual-access EHR came from a single MA plan. As a result, the relationship of continuity to having a mutual-access EHR might have been the result of plan-level characteristics rather than the type of record.
Despite these limitations, this study of information continuity between MHPs and PCPs indicated startling unmet need that would have been identified only by such a detailed medical record review. The clinical implications of discontinuity (for example, mental health outcomes and medication adherence) require study. A mutual-access EHR served to mitigate but not eliminate the stark discontinuity between mental health and primary care.