A Tipping Point for Measurement-Based Care

There are numerous validated, brief structured rating scales that measure the severity or frequency, or both, of psychiatric symptoms as defined in the DSM. Symptom rating scales are structured instruments that patients use to report their perceptions about psychiatric symptoms. Patient-reported symptom rating scales have been widely used in drug trials to demonstrate the clinical effectiveness of most currently used psychotropic medications approved by the Food and Drug Administration. Although some have argued that many patients with psychiatric disorders could have difficulty cooperating with the administration of symptom rating scales or lack the insight to assess their own symptom severity (17), this perspective is fundamentally incongruent with a patient-centered approach to care. Patients are in the best position to assess their own well-being. Moreover, patient-reported symptom rating scales have been shown to be equivalent to clinician-administered rating scales in their ability to identify treatment responders and remitters (18). In fact, for routine care, patient-reported symptom rating scales may be preferable to rating scales that are administered by the clinicians responsible for delivering the treatment. If clinician raters have a stake in the outcomes (for example, provider profiling) assessments may be biased (17).

Many symptom rating scales are just as practical, interpretable, reliable, and sensitive to change as commonly conducted medical tests (for example, a blood pressure cuff). For example, the Patient Health Questionnaire (PHQ-9) is a nine-item symptom rating scale that has one question for each symptom of depression (19). The overall PHQ-9 severity score is easily interpretable as minimal, mild, moderate, moderately severe, and severe. Brief symptom rating scales have been empirically validated to assess the severity and change in severity of most psychiatric disorders, including depression, bipolar disorder, anxiety disorders, posttraumatic stress disorder, schizophrenia, and substance use disorders (20). Many symptom rating scales also assess specific health-related quality-of-life domains, such as appetite, insomnia, and ability to concentrate.

Providers should be wary of developing homegrown rating scales that have not been validated psychometrically. Rating scales that are not reliable or not sensitive to change or that have poor concurrent validity could misinform clinical decision making. To be patient-centric, clinicians should choose a brief diagnosis-specific rating scale or a global-functioning rating scale that best informs clinical decision making for each patient. It is relatively straightforward to incorporate a suite of rating scales into electronic medical record systems in order to make them readily available to clinicians and patients.

Of course, people with mental illness should not be defined by their symptoms. Every patient is unique, with his or her own set of personal recovery goals. A discussion of recovery goals should be part of every clinical encounter in addition to a structured assessment of symptom severity. Both target complaints and goal attainment scaling can be used to assess individualized outcomes (21,22). Changes to the treatment plan can be made based on either a lack of recovery goal attainment or a lack of symptom improvement. However, unlike use of symptom rating scales, incorporating individualized outcome measures into an MBC program presents several barriers. First, assessing individualized outcomes (for example, target complaints and recovery goals) is somewhat more time consuming, compared with using a standardized symptom rating scale. Second, because recovery goals are highly individualized, these outcomes are more difficult to aggregate across providers and practices. Third, there is currently no evidence demonstrating that the systematic assessment of individualized outcomes alone is effective. Establishing this evidence is an important area for future research.

MBC has been defined as “enhanced precision and consistency in disease assessment, tracking, and treatment to achieve optimal outcomes” (16). MBC entails the systematic administration of symptom rating scales and uses the results to drive clinical decision making at the level of the individual patient. MBC is designed to optimize the efficiency, accuracy, and consistency of symptom assessment in order to maximize the likelihood that nonresponse to treatment is detected by the provider. MBC is not intended to be a substitute for clinical judgment (16). Symptom rating scales should be used as a starting point in the provider’s evaluation of the clinical effectiveness of the current treatment.

For the past 20 years, leaders in the field of mental health have been calling for the implementation of MBC into routine care. The Group for the Advancement of Psychiatry, founded by William Menninger, officially endorses the use of self-reported symptom rating scales to supplement clinical interviews (23). The routine administration of symptom rating scales is considered integral to most evidence-based psychotherapies. Because many psychotherapy and pharmacotherapy treatments are diagnosis-specific (for example, prolonged exposure therapy for posttraumatic stress disorder and mood stabilizers for bipolar disorder), results from rating scales that are diagnosis-specific can facilitate adjustments to the diagnosis-specific treatment plan. Psychiatric diagnoses are evolving, and research into biomarkers and precision medicine may fundamentally change diagnostic assessment and treatment in the future. Currently, however, the uncertain relationships between diagnosis, treatment, and outcomes and the limitations associated with our trial-and-error approach (for example, medication trials and stepped care) underscore the importance of MBC. In the absence of biomarkers to inform clinicians as to which treatment will work best for an individual patient, initial treatment choices are often ineffective. Thus, without clear biological causal mechanisms to guide clinical decisions and given the current limitations with trial-and-error–based clinical decision making, it is critical to closely monitor treatment response by using MBC.

For MBC programs to be clinically effective and sustainable in the long run, the symptom severity feedback must be clinically actionable (24). In other words, the symptom rating scale data must be perceived by providers to have a direct benefit to patients (25). The instruments used to measure symptom severity must be reliable (that is, consistent across repeated measurements when there is no actual change in severity) and sensitive to change (that is, able to detect clinically meaningful changes in actual severity) (26). To be able to inform clinical decision making, symptom rating scale data must also be current, interpretable, and readily available during the clinical encounter. Feeding back outdated symptom severity data to providers outside the context of the clinical encounter is not clinically actionable.

To facilitate easy interpretation, changes in symptom severity should be classified into clinically meaningful categories (for example, response, remission, nonresponse, relapse, and recurrence) to facilitate the use of treatment guidelines (27,28). Depression treatment guidelines, such as medication-prescribing algorithms, require data about symptom severity (or changes in symptom severity) at specified intervals (for example, six and 12 weeks following treatment initiation) (29,30). MBC greatly facilitates the use of algorithms, because symptom improvement can be quantified and operationalized into the decision points (23). Importantly, MBC enables the treatment-to-target philosophy of treatment guidelines by identifying which patients have achieved remission (31). Specifically, MBC facilitates the detection of residual symptoms (a known risk factor for relapse [12]) and prompts clinicians to consider intensifying the treatment plan until the patient’s symptoms have completely remitted (that is, treatment to target). MBC also focuses collaboration and coordination across providers. For example, in the team-based collaborative care model, data on the patient’s self-reported symptom severity are collected by the care manager and shared with the treating primary care provider and the consulting psychiatrist (31).

MBC also has the potential to enhance the therapeutic relationship between the patient and the provider, leading to a more informed and activated patient who can participate meaningfully in shared decision making. Patients who regularly complete self-reported rating scales are likely to become more knowledgeable about their disorders, attuned to the fluctuation of their symptoms over time, and cognizant of the warning signs of relapse or reoccurrence (23). MBC can also help patients recognize improvement early in the course of treatment that they might not notice without symptom rating scales. Patient recognition of even small decreases in symptom severity may help them feel more optimistic and hopeful and maintain better adherence to the treatment plan (12). Completing standardized symptom rating scales often validates the way patients are feeling and can mitigate the self-blame that patients sometimes experience. The use of symptom rating scales also empowers patients by helping them communicate more effectively with their providers. Specifically, the use of symptom rating scales may help address health disparities by improving the communication between providers and patients from disadvantaged groups. For MBC to enhance the therapeutic relationship for patients historically experiencing health disparities, symptom rating scales must be chosen that have been culturally validated in low-income and minority populations.

Not all approaches to structured symptom assessment and feedback improve outcomes. For example, assessing patients once by using a symptom rating scale and alerting clinicians to symptomatic patients does not improve outcomes. A Cochrane review of depression screening trials found that patients with depression who were randomly assigned to be screened did not have better outcomes than patients who were randomly assigned to no screening (26). Similarly, alerting clinicians to positive screening results and providing them with guideline-concordant treatment recommendations is no more effective than usual care (32). The suboptimal outcomes associated with this approach are likely due to the fact that initial mental health treatment choices are often ineffective. Thus screening alone is insufficient to improve outcomes without systems in place to monitor treatment response (33).

There is also evidence that symptom severity must be assessed frequently for MBC to be effective. For example, patients seeking treatment at an eating disorder clinic who were randomly assigned to an intervention that fed back self-reported symptoms to their provider midway through treatment (that is, counseling session 5 of 10) did not have better outcomes than patients randomly assigned to usual care (34). There is also evidence that symptom severity must be assessed concurrently with the clinical encounter (that is, shortly before or during the encounter). For example, specialty mental health patients randomly assigned to an intervention that fed back self-reported symptoms to their provider every three months (but not timed to coincide with a clinical encounter) had similar outcomes as those randomly assigned to usual care (35). The largest and most definitive negative trial randomly assigned 895 providers (treating 6,958 patients with depression and 5,858 patients with problem drinking) to usual care or symptom rating scale feedback at every clinical encounter. However, the symptom severity data were collected only at baseline and at three, six, and 18 months (and not timed to coincide with a clinical encounter); thus the symptom data were often not current and therefore not clinically actionable. It can be concluded from the available evidence that to be effective, MBC programs must collect symptom severity data from patients frequently and shortly before or during the clinical encounter (36).

Randomized controlled trials with frequent and timely feedback of patient-reported symptoms to the provider during the clinical encounter significantly improved outcomes (24,37–49) or showed trends toward significance (50). These findings are robust and are consistent across patient groups (for example, various disorders and ages) and provider types (for example, psychotherapists, psychiatrists, and primary care providers). Much of the evidence base is attributable to the research conducted by Lambert and colleagues, who have for many years been routinely collecting symptom severity data from patients immediately prior to their clinical encounter with psychotherapists. An early meta-analysis of six studies with nearly 300 therapists and more than 6,000 patients found that those randomly assigned to MBC had significantly and substantially better outcomes than patients randomly assigned to usual care (37,38,43,45,46,51). The weighted effect size was large (Hedges’ g=–.53) for patients adhering to treatment and medium (Hedges’ g=–.28) for all patients. Not surprisingly, the effect of MBC was found to be stronger for patients whose symptoms did not initially improve than for those who did initially improve (46). Presumably, this is because MBC facilitated changes to the treatment plan only for patients who were not responding to treatment. However, without routine administration of symptom rating scales, it is more difficult to determine which patients are improving and which are not. On the basis of this body of research, the Substance Abuse and Mental Health Services Administration added this MBC model to its registry of evidence-based practices.

Other notable studies include the work of Anker and colleagues (44), who recruited a total of 906 individuals seeking couples therapy. Couples randomly assigned to MBC had significantly better outcomes than couples randomly assigned to usual care, with a moderate effect size (Cohen’s d=.5). Bickman and colleagues (24) recruited 340 youths (ages 11–18) treated by 144 providers from 28 clinics (affiliated with a managed behavioral health organization) in ten states and randomly assigned them to MBC or usual care. Youths assigned to MBC had significantly (p<.01) greater improvements in symptoms than those assigned to usual care, with a small effect size (Cohen’s d=.18). Brodey and colleagues (47) recruited 1,374 adult outpatients with depression from a managed behavioral health organization and randomly assigned them to MBC or usual care. The MBC group had a significantly (p=.04) greater reduction in mean depression symptom severity scores than the usual care group, although the effect size was small (Cohen’s d=.09). Even in relatively small studies, the effect size of MBC can be sufficiently large to detect statistically significant differences in outcomes between groups. For example, Guo and colleagues (52) found that outpatients (N=59) with depression randomly assigned to usual care from a psychiatrist had fewer treatment adjustments and lower remission rates, compared with patients (N=61) randomly assigned to MBC (28.8% and 73.8%, respectively, p=.001).

Two meta-analyses further contribute to the growing evidence base for MBC. Knaup and colleagues (48) analyzed 12 studies and found that MBC had a small but significant effect (Hedge’s g=.10) on outcomes, compared with usual care. The small effect size observed in this meta-analysis was likely attributable to the heterogeneity of the studies with respect to the type of symptom rating scale, frequency of feedback, and to whom the feedback was directed (patients, care coordinator, or provider). A subgroup analysis revealed that effect sizes were larger for outpatient settings (versus inpatient settings), patient self-rated symptom severity (versus staff rated), feedback of changes in symptom severity over time (versus feedback of current symptom severity), and frequent monitoring and feedback (versus infrequent monitoring and feedback). Krägeloh and colleagues (49) analyzed 27 studies and categorized MBC interventions into five groups: administration of symptom severity scales with no feedback, administration with feedback to the provider, administration with feedback to the provider and the patient, administration with unstructured feedback to the provider during the encounter, and administration with structured feedback to the provider and use of treatment guidelines during the encounter. The final category of MBC interventions had the largest effect sizes, highlighting the importance of feeding back symptom severity scores to providers in a structured manner during the clinical encounter.

MBC is also feasible to implement on a large scale. Table 1 describes three-large scale examples of MBC programs in public, private, specialty, and primary care settings (53–55). MBC was also the cornerstone of the intervention tested in the largest pragmatic depression trial (Sequenced Treatment Alternatives to Relieve Depression [STAR*D]) ever conducted in routine primary care and specialty mental health settings (56). The STAR*D trial implemented MBC for 2,876 patients with depression in 23 specialty mental health and 18 primary care clinics representative of real-world settings across the United States (56). MBC was also the foundation of the intervention tested in the largest pragmatic bipolar disorder trial (Systematic Treatment Enhancement Program for Bipolar Disorder [STEP-BD]) ever conducted in routine specialty mental health care (57). The STEP-BD trial implemented MBC for 3,158 patients with bipolar disorder treated in 22 specialty mental health clinics across the United States (58). The scope, heterogeneity, and representativeness of the patients, providers, and clinics in these pragmatic trials suggest that MBC is feasible to implement on a large scale.

MBC is highly acceptable to patients. In a qualitative study conducted in 34 primary care clinics, patients with depression were very receptive to symptom rating scales and perceived them as efficient, complementary to their provider’s clinical judgment, and evidence that their provider was taking their mental health problems seriously (59). Many patients reported that the symptom rating scales helped them to better understand their illness and to express themselves to their provider (59). Symptom rating scales are feasible to administer in the waiting room, and results can be uploaded to the electronic health record for use during the encounter. A publicly funded community mental health center assessed the feasibility and acceptability of using handheld devices to collect symptom severity data. Patients (N=200) reported that the handheld devices were private and easy to use, compared with filling out paper forms (60). However, paper-and-pencil versions of scales need to be available for patients with lower computer literacy.

Although symptom rating scales are diagnosis specific, MBC itself is transdiagnostic and can be incorporated into routine care regardless of the patient population and type of treatment (20). MBC is also transtheoretical and can be incorporated into routine care regardless of the treatment philosophy and training background of providers (20). MBC is effective for both psychotherapy and pharmacotherapy. MBC can also be implemented in such a way that it does not burden providers (8). In fact, MBC can help providers streamline assessments by focusing the discussion on symptoms identified as most severe by the patient (20). In a large clinical demonstration project (1,763 patients in 17 specialty mental health clinics) that replicated the STAR*D MBC protocol, psychiatrists reported that the symptom rating scales were helpful for monitoring response to treatment (100%), assessing severity (94%), tailoring treatment (82%), monitoring suicide risk (71%), and improving the therapeutic alliance (53%) (61). In addition, the psychiatrists reported that the symptom rating scales were helpful for making treatment decisions in 93% of the 6,096 patient encounters (61). Importantly, MBC led to a treatment change in 40% of the patient encounters (61).

Practical factors are the reason providers most commonly report for not implementing MBC, including paperwork burden, amount of time required, and lack of personnel resources (62). Although it is preferable to document symptom severity in the electronic medical record, it should be relatively straightforward to pilot a paper-and-pencil version of MBC to assess clinical utility. Front-office staff can simply give patients a paper copy of the symptom rating scale in the waiting room and ask them to bring it with them to their clinical encounter. However, to assess treatment response, it is critical that the clinician be able to compare current symptom severity to past symptom severity, which is more logistically challenging with a paper-and-pencil version of MBC. Importantly, provider acceptability is notably lower when the symptom severity scores are collected and fed back by an outside organization with a notable lag time between the administration of the symptom rating scale and the clinical encounter. In a study conducted by Brodey and colleagues (47), only 47% of the providers thought that the symptom severity data collected and fed back by a managed care organization helped them monitor their patients’ response to treatment. Providers were concerned about the burden of additional paperwork and felt that the managed care organization was intruding on the patient-provider relationship.

There are also widespread expectations that MBC can be used to improve outcomes at the provider and practice level and to inform payers about the value of mental health services (63,64). Because MBC involves providers’ use of symptom severity scores to make treatment decisions, patients are incentivized to give valid responses to the rating scale questions. This ensures the accuracy of the symptom severity data when aggregated across providers, practices, and health care systems. When symptom severity data are aggregated at the provider or practice level and results compare poorly with benchmarks, it is anticipated that this poor comparison will promote the adoption of evidence-based treatments (20). Moreover, at the level of the provider, aggregated symptom severity data can be used for professional development. For example, a provider can use aggregate symptom severity data to monitor the effectiveness of specific treatments and treatment components (20). By identifying which treatments are most effective, providers can tailor their practice to their patients (20).

Furthermore, if the same rating scales are used by all clinicians in a practice, aggregate symptom severity data can be used to support quality improvement efforts. For example, by means of “practice-based evidence” methodologies (65), aggregated symptom severity data can be used to determine whether the implementation of a new clinical program improves outcomes. Similarly, when symptom severity data are aggregated at the health care system level, the data can be used to demonstrate competency to accreditation organizations and value to payers (20). Purchasers and payers can in turn use aggregated outcomes data to inform the refinement of benefit structures and reimbursement policies in order to maximize the well-being of their employees and enrollees. Thus secondary gains could be made with MBC in addition to the primary gains of improving the outcomes of individual patients.

One of the primary motivating factors for providers and practices to begin implementing MBC is that payers and accreditation organizations are demanding information about outcomes. Representatives from insurance companies and regulators participating in Kennedy Forum meetings have stated that they are looking to health care systems to design their own MBC programs, but that they are prepared to develop their own outcomes-monitoring systems if necessary to support value-based purchasing initiatives. Clearly, it is in the best interest of providers if health care systems develop their own MBC programs that are focused on informing clinical decision making rather than having an outcomes-monitoring system imposed on them by payers and regulators. When developing MBC programs, organizations should adopt psychometrically validated symptom rating scales in order to satisfy purchaser and payer requirements, as well as to maximize patient outcomes. An important reason to be an early adopter of MBC is to have sufficient opportunity to conduct professional development and quality improvement prior to being mandated by purchasers and payers to report aggregated patient-reported outcomes.

In 2015, the National Committee for Quality Assurance announced that depression symptom monitoring and depression response and remission rates will be health plan performance measures for the Healthcare Effectiveness Data and Information Set. Also in 2015, the Centers for Medicare and Medicaid Services, Anthem Blue Cross Blue Shield, and UnitedHealthcare all announced value-based payment programs that incentivize the implementation of MBC. As accreditation organizations and payers roll out these programs, it will be critical that risk adjustment methods or variable benchmarking strategies, or both, are used to correctly interpret observed differences in aggregated outcomes across providers, practices, and health care systems. This will ensure that differences in outcomes reflect differences in access and quality rather than differences in patient case mix (for example, social determinants of health) (66). Another limitation is that aggregation of outcomes data requires storage of the information in electronic health records in such a way that it is easily extractable (that is, not as text embedded in progress notes). This functionality varies across electronic health record systems, and thus it may be burdensome for some health care systems to reap the secondary benefits of MBC.

For synergistic reasons, MBC may be at a tipping point in the field of mental health. There are now numerous brief structured symptom rating scales, many in the public domain, that have strong psychometric properties and that have been validated in diverse patient populations. Technological innovations (for example, handheld devices and electronic health records) have increased the efficiency of routinely collecting symptom severity data from patients and feeding the data back to providers during the clinical encounter. There is mounting empirical evidence from trials that both pharmacotherapy and psychotherapy patients randomly assigned to MBC have better outcomes than patients randomly assigned to usual care. There is evidence from large pragmatic trials and clinical demonstration projects that MBC is acceptable to both patients and providers. There is also growing consensus from accreditation organizations, purchasers, and payers that MBC should be incorporated into performance measures and payment reforms.

Without MBC, providers will not recognize the lack of improvement of hundreds of thousands of patients nationwide, and the patients will endure ineffective treatment. This is particularly problematic for patients from low-income and minority groups who face persistent health disparities. The time is long overdue for the field of mental health to embrace MBC and live up to the medical testing and treat-to-target principles applied by other medical specialties. The cost of routinely administering symptom severity scales is minimal, yet the benefits of MBC accrue to all the stakeholders involved, including patients, providers, purchasers and payers. For patients, completing symptom rating scales and reviewing the information with providers validates the way they feel and ameliorates the self-blame that some patients experience (23). Completing symptom rating scales empowers patients by helping them more fully understand their disorder and the fluctuation in their symptom severity over time and by making them feel more involved in clinical decision making (67). Most important, the use of symptom rating scales helps patients communicate to providers when treatments are not working, thus facilitating changes to their treatment plan.

Although the primary benefit of MBC is improved outcomes for patients, a secondary benefit is the potential to use aggregated symptom rating scale data to enhance professional development, facilitate practice-level quality improvement, demonstrate the value of the mental health services to purchasers and payers, and positively influence reimbursement policies (16).

The potential exists for using aggregated outcomes data to make direct comparisons between providers, and many provides may not be comfortable reconciling their personal assessment of their effectiveness with objectively measured symptom severity data (28). Moreover, it may be challenging to adequately adjust for potential patient case-mix differences between providers, practices, and health care systems. Yet it will be critical to adjust for case-mix differences or to apply benchmarks that accurately reflect the underlying treatment resistance in the patient population. Otherwise, providers, practices, and systems serving the mostly severely ill populations will be unfairly penalized.

Unfortunately, risk adjustment is often associated with concerns about biased case-mix measurement and lack of transparency of complex statistical methodologies. Nevertheless, providers should be held accountable if their patients are not improving as expected and the provider is not revising the treatment plan, getting additional consultation, or referring the patient to a higher level of care. Ultimately, with the primary clinical benefits of MBC and the secondary gains associated with professional development and practice improvement, widespread implementation of MBC will generate evidence for purchasers and payers that mental health treatment works, and this should lead to increases in reimbursement for mental health services over time.