Interventions to Improve Metabolic Risk Screening Among Adult Patients Taking Antipsychotic Medication: A Systematic Review

In recent years, greater attention has been given to general medical comorbidities among individuals with severe mental illness, specifically metabolic and cardiovascular diseases (1, 2). This population is at risk of developing general medical problems for a myriad of reasons, including limited access to primary care (3), illness-related difficulty in acquiring adequate medical care (4, 5), social disadvantage (6), poor health behaviors (7), and propensity for metabolic dysregulation that may be inherent to psychotic disorders (8). Antipsychotic medication use, which is the mainstay of pharmacotherapy in this group, further increases cardiometabolic risk (9, 10).

Therefore, metabolic abnormalities are prevalent among individuals with severe mental illness. One review found that one in three individuals with schizophrenia and more than half of those receiving clozapine treatment were affected by the metabolic syndrome (11), a cluster of metabolic abnormalities linked with two- to threefold risk of cardiovascular mortality (12). In response to the high prevalence of cardiometabolic risk, several health authorities have published recommendations for risk screening. U.S. providers follow the guidance of the American Diabetes Association and American Psychiatric Association consensus paper, which recommends regular monitoring of specific metabolic measures for those prescribed antipsychotics, including weight, waist circumference, blood pressure (BP), fasting blood glucose, and a lipid profile (13). In the United Kingdom, the National Institute for Health and Clinical Excellence schizophrenia guidelines call for yearly physical health assessment, which includes a similar metabolic profile screen (14).

Despite the known elevated cardiometabolic risk among individuals with severe mental illness receiving antipsychotic medication and the presence of clear recommendations for risk screening, evidence suggests that many patients remain unscreened (15, 16). Mitchell et al. (17) conducted a review and meta-analysis of the effect of guideline publication on metabolic risk screening and found partial uptake of guidelines, with only 56% of patients being screened for glucose and 29% for lipids. In response to these findings, interventions that aim to improve metabolic risk screening have been implemented into practice; however, the literature describing these interventions has been only selectively reviewed. One review was limited to a single database and did not assess the effect of the interventions on metabolic risk screening rates or the quality of evidence of the included studies (18). Another review was comprehensive but included interventions that target metabolic risk screening among only a small subgroup of young people with psychosis (19). Neither of these reviews provided a detailed characterization of the improvement strategies that were used in the interventions.

Against this background, we aimed to systematically review interventions that aim to improve metabolic risk screening of adults taking antipsychotic medication. We also aimed to characterize the improvement strategies used by the interventions, describe the intervention effect on screening rates, and comment on their methodological rigor. The goal of this review was to summarize the existing literature to help inform the development of future programs that aim to optimize antipsychotic metabolic risk screening.

Methods

Overview

We conducted a systematic review of interventions that target metabolic risk screening of adult patients taking antipsychotic medication with the objective of improving screening rates to the level recommended by clinical guidelines. We aimed to categorize and describe the various improvement strategies used by the interventions and assess intervention quality and overall efficacy in increasing metabolic screening rates. This review was written in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (20).

Search Strategy

Articles published in print or online from inception to July 11, 2018, were identified through electronic searches by using medical subject headings (MeSH) and keywords of MEDLINE (including Epub ahead of print, in-process, and other nonindexed citations), Embase, PsycINFO, CINAHL, and Cochrane Reviews. The search strategy was developed in collaboration with an information specialist (see online supplement for further details). Search terms related to antipsychotics included the MeSH terms and keywords “antipsychotic” or any known antipsychotic medication (e.g., clozapine). Metabolic monitoring search terms included “monitor,” “measure,” “track,” “metabolic,” “glucose,” “weight,” “blood pressure,” or “lipid.” Search terms for antipsychotics and search terms for metabolic monitoring were combined using the “AND” function. Bibliographies of articles and review articles were hand searched to identify primary articles that may have been missed in the initial search. Only published, peer-reviewed articles available in English were included. Two authors (O.C.M. and E.N.W.) independently completed the title-abstract search, followed by a full-text screen against a priori inclusion criteria. Any disagreements between the two were resolved by discussion and consensus.

Criteria for Study Selection

Inclusion criteria.

We included interventions that targeted metabolic risk screening among adult patients (ages ≥18) taking antipsychotic medication. Eligible interventions targeted screening practices of at least one of four metabolic measures: obesity-weight (body mass index [BMI] or waist circumference), glucose (blood glucose or HbA1C), lipids (total cholesterol, LDL cholesterol, HDL cholesterol, or triglycerides), and BP (systolic BP, diastolic BP, or both). Studies were not restricted to a minimum time duration for an intervention or to a diagnosis of a specific mental illness or type of antipsychotic medication exposure. Studies were required to have an experimental design in which the intervention was implemented into clinical practice, because studies that examined the effect of passive uptake of monitoring guidelines on metabolic screening rates have been reviewed elsewhere and were excluded from this review (17). When studies reported metabolic screening rates at multiple time points in the intervention group (e.g., 1 month postintervention or 3 months postintervention), we considered the latest time point available as the outcome for the review. In cases in which monitoring rates were reported for initiation of antipsychotic medications and ongoing use (maintenance), only the latter was considered. Studies of any design and clinical setting (inpatient or outpatient) were included.

We use the term “intervention group” for the experimental group in controlled designs and for the postintervention group in uncontrolled pre-post designs. Similarly, “comparison group” is used for the control group in controlled designs and for the preintervention group in uncontrolled pre-post designs. Electronic correspondence was sent to authors of articles that described eligible interventions but lacked reporting of either sample size or intervention or comparison group screening rates in order to make the review as exhaustive as possible following the recommendations for “best evidence synthesis” (21).

Exclusion criteria.

Studies describing interventions that do not target adult populations were excluded, as were studies examining metabolic screening of individuals taking psychotropic medications other than antipsychotics (for example, mood stabilizers and selective serotonin reuptake inhibitors), because guideline recommendations are specific for antipsychotics. Studies that described metabolic screening rates as self-reported by physicians were excluded.

Data Extraction and Classification

Data were extracted on the basis of criteria outlined in the Cochrane Handbook for Systematic Reviews of Interventions (22), and data extraction was completed by two authors (O.C.M and L.R.L). Information on the study populations, including demographic factors, psychiatric diagnosis and type of antipsychotic medication, study setting, study design, and rates of metabolic screening in the intervention and the comparison groups, were extracted from the included studies. Strategies used by the intervention to improve metabolic risk screening (i.e., improvement strategies) at the levels of the patient, provider, and system were extracted, because these were found to influence the uptake of clinical guidelines (23).

Outcome Measures

Primary outcome measures included metabolic screening rates for obesity-weight (BMI or waist circumference), glucose (blood glucose or HbA1C), lipids (total cholesterol, LDL cholesterol, HDL cholesterol, or triglycerides), and BP (systolic BP, diastolic BP, or both) and the proportion of patients screened for these four metabolic measures (obesity, glucose, lipids, and BP) in the intervention and comparison groups. Secondary outcome measures included the type and number of improvement strategies used in each intervention to influence monitoring rates at the provider, patient, and system levels.

Quality Assessment

To appraise the internal validity of the studies, each study was examined for risk of bias by using the Effective Public Health Practice Tool (EPHPP) (24). Specifically, the EPHPP assesses the following six domains: sample selection, study design, identification and treatment of confounders, blinding of outcome assessors and participants, reliability and validity of data collection methods, and withdrawals and dropouts. Each domain is rated according to a standardized guide and given a rating of 1, strong; 2, moderate; or 3, weak. Once all domains have been assessed, the study is rated according to the number of domains with a weak rating: strong (no weak ratings), moderate (one weak rating), or weak (two or more weak ratings). Quality assessment was conducted by two authors (O.C.M. and L.R.L.), with discrepancies resolved by consensus.

Data Analysis

Study characteristics and quality appraisal were summarized in tables and the online supplement, respectively. Quantitative data regarding screening rates for four metabolic indices (obesity, glucose, lipid measures, and BP measures) and the proportion of patients screened for four metabolic measures (obesity, glucose, lipids, and BP) were summarized and depicted. For each metabolic measure, the median, minimum, and maximum screening rates for the intervention and the comparison groups were calculated by using Microsoft Excel. We aimed to assess the quantitative data by using standard meta-analytic methods to produce odds ratios to estimate the effect of the intervention on metabolic screening rates when a minimum of three clinically homogeneous studies were found. Qualitative data regarding the strategies used by the intervention to improve screening rates were organized into three a priori categories according to their level of action at the provider, patient, and system levels. The lead author (O.C.M.) initially identified the improvement strategies, and in an iterative process, the research team met several times to analyze and identify these strategies in the studies by using a consensus model (25). We anchored our findings within the Theoretical Domains Framework (TDF), which is commonly used to describe theoretical constructs associated with change in health care provider behavior (26).

Results

Characteristics of Included Studies

The database search led to identification of 7,596 unique citations, and the reference lists of relevant articles and review papers yielded an additional 11 citations that had not been retrieved by the database search. (The online supplement includes a flowchart of the search process and study selection according to PRISMA guidelines (20).) Overall, 166 full-text articles were reviewed in detail, with 30 meeting inclusion criteria. However, the DelMonte et al. (27) and Lee et al. (28) publications were considered as one study because the latter is a long-term follow-up of the former.

The characteristics of the 29 included studies are summarized in Table 1 (27, 29–56). All of the included studies were conducted in mental health care systems, and most (N=23) were conducted in outpatient settings. Over two-thirds of the included studies (N=21) were conducted in the United States or the United Kingdom. A study design of clinical audit cycle with an assessment of preintervention screening rates compared with postintervention screening rates was reported in 23 studies, four reported an interrupted time series, and two were randomized controlled trials (RCTs). Five and seven studies, respectively, were reported to be quality improvement and implementation programs. All 29 included studies were published between 2007 and 2018, and more than half of all publications (N=15) were published in 2015 or later.

Quality Assessment of Included Studies

Overall, 16 studies were rated as weak (29, 31, 33, 37, 38, 40–42, 46, 48, 50–53, 55, 56), and 13 studies were rated as moderate (27, 30, 32, 34–36, 39, 43–45, 47, 49, 54) (see online supplement for a summary of the EPHPP (24) assessment). No studies were given the rating of strong. For 13 of the studies assigned a weak rating (29, 33, 37, 38, 40–42, 46, 50–53, 55), the rating stemmed from their uncontrolled audit cycle design, comparing preintervention screening rates to postintervention screening rates and the lack of adjustment for confounding factors known to influence monitoring rates, such as age and medical comorbidities (15).

Categorization and Description of Improvement Strategies

The included studies described complex interventions that used multiple strategies that aimed to improve metabolic risk screening. Providers were the target of four improvement strategies, patients were targeted by two strategies, and the health system was targeted by seven strategies. Osborn et al. (48) conducted an RCT with two active arms; therefore, our review categorized 30 interventions, which were published in 29 studies as summarized in Table 2 (27, 29–56). The number of improvement strategies utilized in the studies ranged from one to nine, with a median of five.

Improvement strategies targeting providers.

Adequate knowledge and skills regarding metabolic risk screening are key to changing providers’ practice. Concordantly, 23 interventions had an educational component that aimed to enhance providers’ knowledge regarding the importance of metabolic risk screening (27, 29–33, 36–38, 41, 42, 45, 47–56). There was large variation in the intensity and type of educational material delivered to providers, which included both in-person training sessions (ranging from a single 30-minute session [49] to an interactive series [51]) and passive dissemination of information (17, 18). Memory and attention processes of providers need to be influenced for monitoring to occur; therefore, prompts were integrated into routine clinical practice in 21 interventions and were paper based, electronic, or visual (posters) (27, 30, 31, 33–36, 38–42, 44, 45, 47, 49, 51, 52, 54–56). Most commonly, a paper sheet designated for metabolic monitoring was added into patients’ charts and served as a clinical reminder for monitoring in adherence with guidelines as well as a form for recording results (34–36, 38, 40, 41, 47, 51, 52, 54, 56). Electronic prompts for monitoring within an electronic medical record (EMR) appeared when physicians entered a prescription for antipsychotics or when monitoring was due (27, 33, 39, 42, 44, 45, 55), and large posters were displayed in clinical areas with information regarding local monitoring protocols (30, 31, 49, 51, 56). Establishing new social norms that favor metabolic risk screening was attempted by 13 interventions. In this instance, interventions presented providers with a personalized audit of their metabolic monitoring performance benchmarked with rates of monitoring by providers in similar mental health services (30, 31, 33, 36–38, 42, 45, 47, 49, 51, 54, 56).

Improvement strategies targeting patients.

Providing education to patients regarding the need and importance of metabolic risk screening was reported in 12 interventions (29–34, 37, 43, 48, 55, 56), with educational content being delivered both face to face and via paper-based materials. In four interventions, a component of a “lifestyle pack” was added to the educational materials, which included advice regarding physical activity, smoking cessation, and a healthy diet (30, 31, 34, 56). Patients were empowered to seek metabolic risk screening from their health care providers and given practical tools that could assist them in preparing for fasting blood tests (53, 55). The intervention conducted by Kreyenbuhl et al. (43) consisted solely of improvement strategies targeting patients. These strategies included patient education, a personalized feedback of one’s metabolic risk screening status, and an empowerment strategy that urged patients to ask their providers to conduct metabolic risk screening. In some studies, the intervention was codesigned by health care staff and patients to ensure that the proposed intervention was patient centered (37, 45).

Improvement strategies targeting systemic issues.

The majority of the interventions (N=23) included improvement strategies that operated at the system level. Most notably, a number of strategies aimed to create an organizational culture that promotes the practice of metabolic risk screening through social influences. A “metabolic champion,” usually a health care provider on the clinical team in charge of promoting metabolic risk screening in adherence with guidelines, was designated in 11 of the interventions (11, 15, 17, 19, 21, 23, 26–28, 30, 36), and 13 interventions included leadership support in which clinical or administrative service leaders actively supported clinicians’ adherence to monitoring guidelines (32, 33, 35, 37, 40, 42, 44, 45, 47, 50, 53, 54, 56). Leadership involvement included the endorsement of a local monitoring protocol (40), clinical site visits, and newsletters sent to clinicians to encourage metabolic risk screening (47). To further facilitate leadership support, seven studies reported that leaders had access to benchmarked performance audits for metabolic risk screening within their services (30–33, 42, 45, 47). Similarly, mental health organizations that prioritized metabolic risk screening utilized a clinical decision support system to allow for adequate metabolic risk screening upon inpatient admission (33, 46).

Some interventions targeted the environments in which providers work, and as a consequence, an assessment of local barriers to and facilitators of performing metabolic risk screening was reported in 15 interventions (29, 31, 33, 35, 37, 40, 42, 45, 48, 49, 51, 53–56). These evaluations subsequently informed actions, such as the provision of training on how to conduct anthropometric measurements (40, 49) and ensuring that clinicians’ offices contained measuring equipment (51). Promoting collaboration between mental health services and primary care services was a strategy used by four interventions (29, 34, 48, 53) and aimed to increase the supports available for providers to interpret screening results and intervene when appropriate. Finally, nine interventions employed a systematic method to identify patients who are in need of metabolic risk screening as an initial step in improving screening rates (34, 35, 38, 39, 42, 45, 47, 53, 54). These included searching paper and electronic charts to identify patients in a given setting who lack metabolic screening and flagging those charts for intervention (39, 47).

Intervention Effects on Metabolic Risk Screening

There was wide clinical heterogeneity of the included studies; the 30 identified interventions used between one and nine unique strategies to increase metabolic screening rates. Considerable variation was also noted within each strategy across studies (i.e., provider education and metabolic champion). Moreover, studies differed in baseline monitoring rates (e.g., glucose monitoring rates at baseline varied from 0% to 92%) and the target metabolic measure and the timing of postintervention screening rates (1 week to 6 years). Therefore, a meta-analysis of the results was not feasible.

The quantitative results of 28 interventions (27, 29–42, 45–56) are presented in a descriptive manner in Table 3 and depicted in Figure 1. Positive results were reported for 21 of the 30 interventions, in which significant increases to screening rates of all targeted metabolic measures had been observed (27, 29–31, 33–37, 39, 44–48, 51, 53–56). Mixed results were reported for five interventions, with at least one metabolic measure showing an increase in screening rates in the intervention group (40–42, 49, 52). A null effect on screening rates was observed for four interventions (32, 38, 43, 50).

Intervention effects on glucose and lipids screening.

Glucose was the most prevalent metabolic measure screened in the studies and was targeted by 21 interventions (27, 29–32, 34–36, 38–42, 47–50, 52, 55, 56) (Table 3). Median screening rates for glucose increased from 28% in the comparison groups to 65% in the intervention groups. Similarly, lipids were screened in 18 interventions (27, 29–32, 34–36, 38, 40–42, 48, 49, 52, 55, 56). Median screening rates for lipids increased from 22% in the comparison groups to 61% in the intervention groups.

Intervention effects on anthropometric and physical measurements.

Waist circumference screening rates were reported in six studies, with median screening rates increasing from 2% in the comparison groups to 87% in the intervention groups (32, 34, 35, 38, 40, 52). BMI/weight screening rates were reported in 16 studies, with median screening rates increasing from 19% in the comparison groups to 67% in the intervention groups (29–32, 34, 35, 37, 38, 40–42, 48, 50, 55, 56). BP screening rates were reported in 15 studies, with median screening rates increasing from 22% in the comparison groups to 80% in the intervention groups (29–32, 34, 35, 37, 38, 40, 41, 48, 52, 55, 56).

Additional summary of intervention effects.

Eight studies reported the intervention effects by presenting the change in the proportion of patients who were screened for all four metabolic measures (obesity, glucose, lipids, and BP) (30, 31, 33, 45, 46, 51, 53, 54), and the median screening rates for the four measures increased from 11% in the comparison groups to 57% in the intervention groups.

In two studies, outcomes of the intervention were not presented as screening rates. In one, Kreyenbuhl et al. (43) conducted an RCT in which the outcome was the proportion of days in which patients were adherent with monitoring guidelines. All metabolic parameters were appropriately monitored for over 75% of the days in both intervention and comparison groups, and no beneficial effect was observed in the intervention group. Similarly, Lai et al. (44) described the effect of the intervention as a proportion of “qualified patient visits” (QPVs), in which patients were in concordance with monitoring guidelines within 6 months of the visit date. A sevenfold increase in the percentage of QPVs was found among the patients taking clozapine or olanzapine, compared with those taking first-generation antipsychotics. The types of measures used in these two studies precluded the inclusion of the studies in the intervention summaries provided in Table 3 and Figure 1.

Discussion

This review conducted a comprehensive knowledge synthesis of interventions that target metabolic risk screening of adults taking antipsychotic medication. Using a theoretical framework (26), the review identified improvement strategies used by the interventions at the levels of the provider, patient, and system, highlighting that organizational commitment to monitoring is a key issue. Common methodological limitations were observed in most interventions, including uncontrolled design and lack of adjustment for confounders, which reduced our ability to draw firm conclusions regarding the true effect of the interventions. Findings suggest that interventions were overall successful in increasing screening rates; however, up to a third of patients remained unscreened.

This review identified 30 unique interventions that were carried out in mental health services and targeted at improving metabolic risk screening of adults taking antipsychotics. Given that antipsychotic metabolic screening falls short of recommended guidelines (15, 17), we reported on programs that addressed this guideline-to-practice gap. Interventions varied widely on the “continuum of quality improvement” (57). That is, some interventions were local staff initiatives involving a small unit (34, 50) while others were resource-intensive organizationwide programs (32, 56). Over time, it was evident that some programs advanced from local practice improvement initiatives to large-scale theory-driven implementation science programs (29, 33, 37, 42, 45, 47, 54). Some included external experts and advisors who worked longitudinally in collaboration with clinicians, managers, and patients to achieve predefined measurable improvements in metabolic risk screening. Inherently, implementation programs are contextualized to existing resources, processes, and structure, which limits their transferability between organizations (26). Nevertheless, interventions reviewed here shared certain similarities in objectives, tools, and improvement strategies that have been used to narrow guideline-to-practice gaps in other domains of clinical practice (58). This review suggests that use of improvement strategies is beneficial in reducing the guideline-to-practice gap for metabolic risk screening of patients who take antipsychotics; however, additional reports are needed to support the knowledge base.

The quality of the studies included in this review was found to be weak (N=16) or moderate (N=13), and no studies were rated as strong. Quality improvement programs are gaining momentum in health care and approximately 1,000 quality improvement studies are published each year (59). Nevertheless, scholars are divided about whether findings from quality improvement programs hold similar scientific rigor as other methods of inquiry, such as RCTs (60). Given that quality improvement programs are often multifaceted and implemented into intricate systems, it is challenging to ascertain the effect of an intervention with traditional highly controlled designs. This is evident in the studies reviewed here; over three-quarters of them chose a clinical audit cycle to identify the effect of the intervention, which contributed to their lower quality ratings. Recently, burgeoning methodological advancements are likely to increase the scientific rigor of quality improvement programs to a degree that allows us to infer causality (61)—for example, use of interrupted time-series analyses, cluster RCTs, and theory-driven improvement strategies (60, 61). These advances are timely and should be incorporated into future interventions that aim to improve antipsychotic risk screening in mental health services.

Using a theoretical framework (26), this review identified the improvement strategies used by the interventions to influence metabolic risk screening at the provider, patient, and system levels. For providers, considerable efforts were made to incorporate screening into practice by influencing memory and attention processes with prompts and monitoring tools. Physician reminders, both paper and electronic, have been known to influence physician behavior and thus are frequently used in medical practice (62). Similarly, nearly all interventions included provider education regarding metabolic risk and recommended antipsychotic monitoring; however, most would not be considered as continuing medical education (CME) activities. A recent review found that CME had a profound effect on physician performance (63); therefore, CME activities have the potential to further improve antipsychotic monitoring (64).

Patients were empowered to seek metabolic risk screening from their health care providers (43, 53) by provision of both written materials and verbal encouragement. These approaches are in line with contemporary trends positing that individuals with severe mental illness benefit from acquiring self-management skills to better manage their health and health care, as reported in the Health and Recovery Peer program (65). Future research is needed to examine the effect of peer-support self-management programs on reducing cardiometabolic risk and improving screening among individuals with severe mental illness.

Interventions attempted to change metabolic risk screening by using social influences to alter providers’ behavior. Social influence is a construct targeting interpersonal processes that are known to change providers’ thoughts, feelings, and behaviors (26). Interventions reported commitment at the mental health system level, where leaders and managers took an active part in promoting physical health (32, 42, 54), thus creating an organizational culture that encouraged metabolic risk screening. For example, managers and leaders who had access to data comparing the metabolic risk screening rates in their organization or unit with those in similar services could discuss this with their clinical teams. Similarly, the appointment of a clinician-leader for metabolic monitoring (i.e., metabolic champion) helped facilitate change to clinicians’ professional role to one that includes general medical issues (27, 37). The environmental context, in terms of resources and capability to conduct metabolic risk screening, was targeted by some interventions. Some conducted a barrier assessment and used the results to inform the intervention, because practical issues, such as lack of measurement equipment in clinic rooms, often prevent monitoring (40, 42). Others used available external resources, such as primary care providers in a shared care model, to promote metabolic risk screening (29, 48). This was seen in studies originating from the United Kingdom, where primary and mental health care services are geographically aligned. Such an approach may be difficult to replicate in the United States because of fragmentation of the two systems. Future research is needed to test whether integration of primary care activities into mental health care services would improve metabolic risk screening and management in that setting.

Most interventions were successful, as evidenced by the increase in median metabolic screening rates for glucose (28% to 65%), lipids (22% to 61%), weight (19% to 67%), and BP (22% to 80%) (Table 3 and Figure 1). However, after implementation of the intervention, up to one-third of patients still remained unscreened. This is consistent with other reports that describe suboptimal medical care for diabetes, hypertension, and ischemic heart disease and suboptimal preventive medicine among people with severe mental illness (66, 67). Anthropometric measurements were easier to obtain, compared with blood tests, especially when this task was delegated to nonmedical providers, such as administrative staff and home care workers (29, 40). Blood tests for glucose and lipids pose a burden for individuals, mostly because they require fasting samples but also because of access issues in services that do not offer on-site testing. Overcoming these barriers can be achieved by the use of point-of-care (POC) testing. Conducting a POC random glucose measurement, along with weight and BP assessments, during a routine mental health outpatient visit has been used successfully to screen for metabolic risk (68).

Studies differed substantially in the metabolic measure targeted for improvement—while some targeted a single measure (e.g., glucose only), others targeted four (glucose, lipids, weight, and BP). Individuals with severe mental illness are at risk of abnormalities in all metabolic measures; therefore, future antipsychotic surveillance programs should aim to improve monitoring that is concordant with published guidelines.

Most of the interventions focused on improving metabolic risk screening, and only a few reported linking screening programs with management of identified risk (38, 45, 46, 51, 56, 68). This is of concern because far too often metabolic risk factors are left untreated. For example, at the time of patient enrollment into the Clinical Antipsychotic Trials of Intervention Effectiveness schizophrenia trial, no-treatment rates were 30% for diabetes, 62% for hypertension, and 88% for dyslipidemia (69). Psychiatrists acknowledge the importance of conducting metabolic risk screening; however, most reported difficulty in arranging medical follow-up for identified risk, and only about a third were willing to prescribe medications such as statins or metformin (70). Thus a clear rationale exists for linking metabolic risk screening with management of identified risk by local resources that are available to mental health services. Close collaboration with primary care services is recommended in the guidelines (71). When primary care services are not available, delegation of responsibility for metabolic risk screening to clinical pharmacists, who frequently work alongside psychiatrists, has been used successfully (27, 39, 68). The role of clinical pharmacists offers advantages beyond screening. This includes pharmacotherapy recommendations for treatment of identified risk and patient education.

Similarly, identification and management of tobacco use, although not directly recommended in antipsychotic monitoring guidelines, should be routinely conducted alongside metabolic risk screening programs, given its strong link with cardiovascular risk and metabolic dysregulation (72). Indeed, nearly a third of all interventions screened for tobacco use, which is a prerequisite for linking patients with smoking cessation programs, preferably those adapted to meet the needs of individuals with severe mental illness (73, 74).

This review, which offered a knowledge synthesis of intervention studies that aimed to improve metabolic risk screening of patients taking antipsychotic medication, had some limitations. Because of the heterogeneity of the included interventions in terms of design, duration, outcomes, and improvement strategies, we were not able to combine the effects of the interventions in meta-analyses or assess for publication bias. This is in line with Cochrane guidance, which discourages use of meta-analyses in heterogeneous studies (22). The fact that many interventions used multiple improvement strategies in an uncontrolled design puts the study results at high risk of bias. This limited our ability to draw firm conclusions regarding the true effect or the superiority of one type of intervention over another. A strength of this review was its attempt to describe the improvement strategies of the intervention at the levels of the provider, patient, and system. To establish the credibility of this process we followed expert guidance (75), which included anchoring our findings within a theoretical framework (26), meeting several times to discuss the results, and working iteratively to analyze the findings.

Conclusions

Interventions to increase metabolic risk screening among individuals taking antipsychotic medications include multiple improvement strategies that are shared by other programs in clinical areas that address an evidence-to-practice gap (57, 58). Although interventions were successful in improving median screening rates, up to one-third of patients remained unscreened. Improvement strategies that are key to the success of metabolic risk screening programs create a culture within the organization that prioritizes general medical care by means of education, leadership support, clinician leadership, and metabolic risk screening audit and feedback. Current knowledge is based on studies that lack scientific rigor, which plagues many quality improvement programs. However, timely advances in theoretical understanding of implementation problems (26) and methodological issues (60) offer guidance for the future design of scientifically sound programs. These advances include the use of interrupted time-series analyses and cluster RCTs.

Treatment of identified metabolic risk is of utmost importance and should be addressed by all risk screening programs in consideration of the availability of local resources, including but not limited to primary care providers (29, 45), allied health clinicians in mental health services (27, 48), and psychiatrists and aided by a robust clinical decision support system (76). Mental health organizations are urged to provide adequate monitoring and management of metabolic risk of patients taking antipsychotic medication, and the knowledge synthesis presented here can inform the design and evaluation of future programs in this area.