|Home | About | Journals | Submit | Contact Us | Français|
Concurrent peer review visits are structured office visits conducted by clinician peers of the primary care clinician that are specifically designed to reduce competing demands, clinical inertia, and bias. We assessed whether a single concurrent peer review visit reduced clinical inertia and improved control of hypertension, hyperlipidemia, and diabetes control among underserved patients.
We conducted a randomized encouragement trial to evaluate concurrent peer review visits with a community health center. Seven hundred twenty-seven patients with hypertension, hyperlipidemia, and/or diabetes who were not at goal for systolic blood pressure (SBP), low-density lipoprotein cholesterol (LDL-C), and/or glycated hemoglobin (A1c) were randomly assigned to an invitation to participate in a concurrent peer review visit or to usual care. We compared change in these measures using mixed models and rates of therapeutic intensification during concurrent peer review visits with control visits.
One hundred seventy-one patients completed a concurrent peer review visit. SBP improved significantly (p < .01) more among those completing concurrent peer review visits than among those who failed to respond to a concurrent peer review invitation or those randomized to usual care. There were no differences seen for changes in LDL-C or A1c. Concurrent peer review visits were associated with statistically significant greater clinician intensification of blood pressure (p < .001), lipid (p < .001), and diabetes (p < .005) treatment than either for control visits for patients in either the nonresponse group or usual care group.
Concurrent peer review visits represent a promising strategy for improving blood pressure control and improving therapeutic intensification in community health centers.
Control of blood pressure—particularly systolic blood pressure (SBP), low-density lipoprotein cholesterol (LDL-C), and glycated hemoglobin (A1c)—for patients with hypertension, hyperlipidemia, and diabetes has been shown to reduce morbidity and/or mortality.1–3 Achieving target goals for these conditions has proved challenging,4–7 particularly for poor and minority patients.8–10 Competing visit demands and clinician inertia (ie, failure to initiate change in treatment when clinically indicated) during 15-minute visits make it difficult for clinicians to focus on achieving these targets.11,12 Cognitive overload may also contribute to implicit clinician bias (ie, activation of unconscious stereotypes that affect clinical behavior),13 particularly in underresourced practices where working conditions may be especially stressful.14
To address these challenges, we devised a quality-improvement initiative using concurrent peer review visits. Concurrent peer review visits are semistructured, billable patient care visits conducted by a clinician peer of the primary provider explicitly designed to improve disease control by minimizing competing patient demands, clinical inertia, and bias.
These visits minimize competing demands by nearly exclusive visit focus on chronic disease management of the condition not at goal (barring an urgent problem). Other problems are deferred to the primary care clinician. Clinical inertia is minimized through a live, independent review of management by a clinician peer and by reducing competing demands.15,16 Clinical inertia is further reduced by providing generic decision support at the point of care in the form of guidelines for intensification of treatment17 and by requesting the patient to bring in medications they are actually taking. Bias is addressed through an independent review by a clinician peer who typically has fewer preconceived beliefs about the patient (including their past adherence),16 reduction in competing demands, and decision support.
To assess the impact of concurrent peer review visits on cardiovascular care among underserved patients, we conducted a pragmatic, randomized controlled encouragement trial. A pragmatic trial is designed to demonstrate effectiveness (success in the real world) rather than simply efficacy (success under optimal, if not idealized conditions).18 We assessed the impact of concurrent peer review visits on disease control and clinician treatment intensification.
The study was approved by the University of Rochester institutional review board. The study was registered at www.trials.gov (NCT00508014).
The study was conducted within 2 sites at a federally qualified health center (FQHC). Patients at the 2 sites are seen by clinicians (12 family physicians, 3 nurse practitioners, and 4 physician assistants). The FQHC has been participating in the Health Disparities Collaborative (HDC), designed to improve quality of care for cardiovascular-related conditions, including diabetes.19,20 At the time of the study, relevant clinical data for patients with a diagnosis of hypertension, hyperlipidemia, and diabetes were entered into a stand-alone electronic patient registry (Patient Electronic Care System, Aristos Group, Inc, Austin, Texas). All charting at the time of the study was done using paper charts. Patient visits were conducted using a preprinted template that was generated from the HDC registry. It included a summary of patient medication classifications, laboratory values, and key care processes. Values that were out of range or recommended preventive services were highlighted, thus providing a prompt for provider action. Thus, concurrent peer review visits were introduced into a setting where providers received modest decision support at the point of care and overall performance was being tracked.
Patients were identified using the HDC patient registry. Patients were eligible if they were aged 18 years or older; were presumed an active patient in the practice (had 1 visit at the FQHC in the past 12 months); and had SBP of at least 140 mm Hg (≥130 for diabetics), LDL-C of at least 130 mg/dL (≥100 for diabetic or those with coronary heart disease), or A1c of at least 7.0% (diabetics).
Using computer-generated random numbers weighted in favor of the intervention letter 1.4:1, the project statistician randomly assigned patients, stratified by site, to receive an invitation to participate in a concurrent peer review visit or usual care (control). Patients randomized to the concurrent peer review visit received a letter signed by the primary care provider inviting them to see a peer clinician for a concurrent peer review visit. Two weeks following this mailing, a research assistant called the patient and offered to schedule them an appointment. Appointments were randomly assigned to clinicians weighted based on the number of sessions per week the clinician worked.
Visits were scheduled in 30-minute blocks (though the actual visit was often shorter due to double booking). All visits used the same preprinted encounter templates used for all patients in the HDC with diabetes, hyperlipidemia, or hypertension (not just those assigned to concurrent peer review visits). Three training sessions for concurrent peer review visits for clinicians were held during regularly scheduled provider meetings. Providers conducting concurrent peer review visits were instructed to confine the content of the visit to chronic disease management for these 3 conditions. They were instructed to begin the visit with a medication review, ideally based on a review of medication bottles provided by the patient. Concurrent peer review providers were further instructed to intensify treatment when the patient was not at goal for any of the 3 conditions using generic guidelines that were appended to the paper chart. The concurrent peer review provider then summarized medication and other changes made using a form available in triplicate. One copy was filed in the chart after being used for program evaluation and analysis. A second copy was given to the primary care clinician, and the third copy was given to the patient.
A research assistant extracted data from charts using an abstraction tool and used data entered into the PECS system. Data included SBP, LDL-C, and A1c values, visit dates for up to 12 months following randomization. To assess whether concurrent peer review visits were associated with improved control, changes in SBP, LDL-C, and A1c were compared among patients assigned to usual care, those who completed the concurrent peer review visit, and those invited but who did not complete a concurrent peer review visit. This latter group was mostly patients who failed to respond to the invitation letter (we often could not verify an address listed in the medical record) and who could not be reached by phone but also included a few patients who agreed to a concurrent peer review visit but repeatedly failed to keep their appointment.
To assess whether concurrent peer review visits reduced clinical inertia, we compared rates of intensification of treatment where SBP, LDL-C, and A1c values were not at goal at the time of the visit. Data were abstracted from medical records from 519 (70% of sample) randomly selected patients. Funding constraints precluded examination of the entire sample. For patients completing concurrent peer review visits, data were abstracted from that visit. For patients assigned to usual care and patients who failed to respond to an invitation for a concurrent peer review visit, we selected a visit that addressed 1 of chronic conditions that was nearest 40 days of randomization (based on the mean time to complete a concurrent peer review visit following randomization) and addressed patient chronic conditions. Intensification was defined as an increase in medication dose or addition of a new medication for the condition not at goal.
Baseline measurements were taken within 90 days prior to the date of randomization. One-way analysis of variance (ANOVA) was used to assess baseline differences in SBP, LDL-C, and A1c among the control, concurrent peer review completed, and nonresponder groups. Subsequent measures were taken up to 365 days post randomization at different and unequally spaced time points with a varying number of measurements for each participant.
A longitudinal repeated measures mixed-model ANOVA with a compound symmetry covariance structure was used to assess the change in SBP, LDL-C, and A1c over time among the control, concurrent peer review, and concurrent peer review nonresponse groups. This design accounts for the correlation within individual participants and measurements taken at closer-spaced time points as well as the unbalanced design resulting from the varying number of measurements.
Our sample size was constrained by the number of eligible patients at both sites. A power calculation indicated that an anticipated sample of 750 with a 30% non-participation rate would provide 80% power to detect differences in changes of 4.9 mm Hg in SBP, 6.8 mg/dL in LDL-C, and 0.8% in A1c.
All analyses were done using SAS version 9.2 on the Windows XP Professional platform (Cary, North Carolina). A 95% level of confidence was used to determine significance in all statistical tests.
Participant characteristics, including screening, enrollment, and dropouts, are shown in the Figure. From 914 patients with hypertension, diabetes, or hyperlipidemia in the registry, 727 met eligibility criteria and were randomized to an invitation to a concurrent peer review visit or usual care (control). Four hundred thirty-four patients had blood pressure not at goal, 400 patients had LDL-C values not at goal, and 286 diabetic patients had A1c values that were equal or greater than 7.0%. These numbers total more than 727 due to some patients having several of these conditions.
Of the 423 patients invited to a concurrent peer review visit, 171 actually completed the visit. Among patients failing to complete a concurrent peer review visit, the most common reason was inability to reach the patient through follow-up phone calls after mailing of the C concurrent peer review PR invitation letter. All 19 clinicians participated. Twelve were family physicians and 7 were either nurse practitioners or physician assistants.
The characteristics of the groups are shown in Table 1. The sample is older, largely female, and minority. Most of the patients were insured through public insurance (Medicaid and/or Medicare). The 1-way ANOVA resulted in no statistical difference in baseline SBP, LDL-C, or A1c values among the 3 groups (ie, patients completing concurrent peer review visit, patients assigned but not completing a concurrent peer review, and patients not invited to a concurrent peer review visit). Patients completing a concurrent peer review visit were slightly older and more likely to receive care at 1 of the 2 FQHC sites.
Crude outcomes assessed at 1 year following initial randomization are shown in Table 2. There were 2174 SBP measurements taken on 434 unique participants during follow-up. The longitudinal analysis showed a statically significant difference in SBP (p = .01) over time among the 3 groups. Notably, patients randomized to concurrent peer review who had a concurrent peer review visit had lower SBP than either of the other 2 groups.
The longitudinal analysis showed no significant difference in LDL-C over time among any of the groups. The 1-way ANOVA resulted in no statistical difference in A1c (p = .4) among the 3 randomization groups at baseline. The longitudinal analysis showed no significant difference in A1c over time among the groups.
Last, we compared rates of intensification of treatment for hypertension, hyperlipidemia, and diabetes for patients not at goal. Results are shown in the lower portion of Table 2. Rates of clinician intensification of treatment were statistically higher during concurrent peer review visits than during control visits for patients in the other 2 groups. Intensification rates during concurrent peer review visits were roughly double those in control visits among patients in usual care. These findings were not appreciably altered by adjustment for patient characteristics (results not shown).
Findings from this randomized encouragement trial suggest that concurrent peer review visits offer promise for improving blood pressure control among underserved patients. Findings showed no significant improvement during postrandomization period between groups for control of hyperlipidemia or diabetes, but comparison of data from concurrent peer review visits with control visits suggested that clinicians intensified treatment for each of these 3 conditions more often than during control visits. This suggests the possibility that changes made during a single visit conducted by a clinician peer may not have been sufficient to improve control for these conditions despite improvements in intensification of therapy. Lower than expected rates of participation by patients in concurrent peer review visits also contributed to reduced power to detect differences in primary outcomes.
To our knowledge, use of concurrent peer review visits to improve care is a novel concept. Previous studies have shown that trained nurses using protocols combined with decision support improved blood pressure and cholesterol, but not A1c, among diabetics in primary care practices.21 Physician collaboration with pharmacists may also promote blood pressure control,22 and feedback to medical residents has been associated with improvement in diabetic control.23 However, the literature is also replete with failed efforts to improve control of these conditions, and there is a paucity of interventions that have been shown to decrease clinical inertia among community clinicians. We are not aware of previous studies that have aimed to improve care and reduce clinician inertia using clinician peers.
Concurrent peer review visits offer several major advantages. They provide a means to build quality improvement into billable office visits. Rather than requiring clinicians to review each others’ charts retrospectively—ie, after a visit where there is less opportunity to affect patient care—these visits provide time during regularly scheduled visits with patients for quality improvement. In addition, they allow quality improvement to be fully integrated into busy practices with minimal extra time or loss of clinical revenue. This is especially important in safety-net practices given pressing time demands and resource constraints.14,24 Lastly, concurrent peer review visits offer a means for promoting genuine collaboration between clinician colleagues in the same practice.
Findings of higher rates of treatment intensification during concurrent peer review visits suggest that these visits had the intended effect of reducing clinician inertia, although we do not know which aspect of this multimodal intervention contributed the most to doing so. At least for blood pressure control, concurrent peer review visits were associated with improved control. Whether failure to detect improvements in the other outcomes represents lack of patient adherence to recommendations from the concurrent peer review clinician or simply that changes made during a single visit were too weak to improve outcomes is not clear.
Blood pressure–, lipid-, and glucose-lowering agents all have relatively flat-dose responses compared with the effect of adding a new agent.25–27 However, in contrast to hypertension management, where there are many classes of agents from which a new, well-tolerated agent can be added to an existing regimen to improve blood pressure control, there are fewer affordable options available to improve diabetes and or hyperlipidemia. Many diabetics have “maxed out” on inexpensive, oral agents.28 Improved diabetic control in this setting often requires use of an expensive brand-name hypoglycemic agent (eg, a dipeptidyl peptidase IV inhibitor or an incretin mimetic) or, more often, the introduction of injectable insulin. Yet, in not a single concurrent peer review visit did the clinician initiate insulin with a patient. For reasons we can only speculate about, clinician peers failed to use the single most effective glucose-lowering agent available: insulin. Similarly, increases in statin dose improve LDL-C levels only modestly.26 Achieving significant reductions in LDL-C levels among those on statins often requires a switch to a potent statin, eg, atorvastatin or rosuvastatin. Both of these statins are currently under patent protection and are much more expensive than older ones.29 Thus, in a practice setting where more than half the patients were insured through Medicaid and an additional 14% lacked known insurance, medication costs and reluctance to initiate an injectable drug may have constrained treatment options to a greater extent for diabetes and hyperlipidemia than for hypertension. Potentially, these constraints explain differences in outcomes.
These promising findings are best understood in the context of the practice setting and study design limitations. First, concurrent peer review visits were implemented in the context of a preexisting quality-improvement project whereby clinicians were receiving point-of-care reminders and performance feedback. Thus, any benefits from concurrent peer review were beyond those based on these existing interventions. Second, the project was undertaken within a fairly cohesive group of clinicians dedicated to quality improvement. It is not known whether other practices would embrace the concept of concurrent peer review visits to the same extent. Third, in order to maximize generalizability, we conducted an encouragement trial. The advantage to this approach is that our population was less selected than if we had first enrolled each patient individually. The primary disadvantage is that this approach introduces greater risk for selection bias. Thus, we cannot exclude the possibility that patients with uncontrolled blood pressure who responded to the invitation were more motivated than those who did not. Fourth, we cannot exclude the possibility that clinician learning extended to the control group, thus biasing results to the null. Last, we evaluated 3 separate primary outcomes that increase the risk for a chance finding. However, the finding in which significantly greater treatment intensification occurred during concurrent peer review visits is consistent with the hypothesized effect.
In conclusion, these findings suggest that concurrent peer review visits represent a promising intervention for improving cardiovascular-related care in FQHCs. Further study is needed to confirm these findings and determine the extent to which they are generalizable to other settings.
Funding/Support: Support was provided by The Robert Wood Johnson Foundation, Finding Answers Program.
We are indebted to our original study statistician, Sean Meldrum, who died during the project, and Adjuah van Keken for assisting with manuscript preparation.