“Act on Threes” Paradigm for Treatment Intensification of Type 2 Diabetes in Managed Care: Results of a Randomized Controlled Study with an Educational Intervention Targeting Improved Glycemic Control

BACKGROUND: Clinical inertia, which has been defined as the recognition of a problem with a patient’s management but failing to act, is a concern in type 2 diabetes (T2D) because it places the patient at risk of diabetes-related complications. Despite managed care organizations making significant investment in this area, little is known about the impact of educational programs aimed at aligning patients and their physicians with diabetes guidelines and thus overcoming clinical inertia. OBJECTIVE: To assess the impact of an educational intervention specifically designed to align patients and their physicians with 2012 American Diabetes Association (ADA) guidelines on glycated hemoglobin (A1c) testing frequency and insulin initiation. METHODS: The “Act on Threes” educational intervention was a 12-month, randomized controlled prospective study that included Medicare Advantage patients aged 18-85 years with T2D, who received ≥ 3 oral antidiabetes drugs (OADs) and/or had A1c not at goal and/or had no recent A1c evaluation over 12 months, as identified through the analysis of administrative claims data (May 1, 2011-April 30, 2013) from the Humana database. Identified patients were randomized 3:1 to receive the Act on Threes educational intervention in conjunction with standard care (intervention group) or standard care alone (control group). For the educational intervention, patients and physicians were simultaneously mailed general and targeted information aimed at aligning them to 3 vital aspects of A1c control: timely measurement of A1c every 3 months; timely treatment intensification to meet A1c goals with treatment intensification every 3 months if A1c is not at goal; and insulin initiation when appropriate, including patients receiving ≥ 3 OADs with A1c not at goal. Control patients were only enrolled if the treating physician was not involved in the care of any patients in the intervention group. The primary outcome measures were A1c testing frequency based on the ADA standard for compliance of ≥ 2 tests per year and insulin initiation in the 12-month postintervention period. A1c levels were evaluated for the subgroup of patients with available A1c measurements in the pre- and postintervention periods. Descriptive statistics were used to analyze differences between the intervention and control groups. Multiple logistic regression analysis was used to identify determinants of insulin initiation in the full study cohort. RESULTS: 6,243 patients (mean age 70 years; 43.5% female) were identified: 4,555 were randomized to the intervention group and 1,688 to the control group. The percentage of patients with ≥ 2 A1c tests per year was not significantly different postintervention for patients in the intervention and control groups (47.7% vs. 46.8%, respectively; P = 0.995). Intriguingly, the frequency of A1c testing increased significantly from pre- to postintervention in the intervention and control groups. Change in A1c level from pre- to postintervention was also similar for the 2 groups (P = 0.240). A similar percentage of patients in the intervention and control groups initiated insulin during the postintervention period (6.3% vs. 7.6%, respectively; P = 0.059). CONCLUSIONS: This randomized controlled study demonstrated that, compared with standard care, the Act on Threes educational intervention combined with standard care did not result in any significant differences in the frequency of A1c testing or in the initiation of insulin in patients with T2D. These findings are in contrast to uncontrolled comparative studies showing significant improvements in outcomes postintervention and reinforce the importance of study design in evaluating the effectiveness of educational programs.

OBJECTIVE: To assess the impact of an educational intervention specifically designed to align patients and their physicians with 2012 American Diabetes Association (ADA) guidelines on glycated hemoglobin (A1c) testing frequency and insulin initiation.
METHODS: The "Act on Threes" educational intervention was a 12-month, randomized controlled prospective study that included Medicare Advantage patients aged 18-85 years with T2D, who received ≥ 3 oral antidiabetes drugs (OADs) and/or had A1c not at goal and/or had no recent A1c evaluation over 12 months, as identified through the analysis of administrative claims data (May 1, 2011-April 30, 2013) from the Humana database. Identified patients were randomized 3:1 to receive the Act on Threes educational intervention in conjunction with standard care (intervention group) or standard care alone (control group). For the educational intervention, patients and physicians were simultaneously mailed general and targeted information aimed at aligning them to 3 vital aspects of A1c control: timely measurement of A1c every 3 months; timely treatment intensification to meet A1c goals with treatment intensification every 3 months if A1c is not at goal; and insulin initiation when appropriate, including patients receiving ≥ 3 OADs with A1c not at goal. Control patients were only enrolled if the treating physician was not involved in the care of any patients in the intervention group. The primary outcome measures were A1c testing frequency based on the ADA standard for compliance of ≥ 2 tests per year and insulin initiation in the 12-month postintervention period. A1c levels were evaluated for the subgroup of patients with available A1c measurements in the pre-and postintervention periods. Descriptive statistics were used to analyze differences between the intervention and control groups. Multiple logistic regression analysis was used to identify determinants of insulin initiation in the full study cohort. RESULTS: 6,243 patients (mean age 70 years; 43.5% female) were identified: 4,555 were randomized to the intervention group and 1,688 to the control group. The percentage of patients with ≥ 2 A1c tests per year was not significantly different postintervention for patients in the intervention and control groups (47.7% vs. 46.8%, respectively; P = 0.995). Intriguingly, the frequency of A1c testing increased significantly from pre-to postintervention in the intervention and control groups. Change in A1c level from pre-to postintervention was also similar for the 2 groups (P = 0.240). A similar percentage of patients in the intervention and control groups initiated insulin during the postintervention period (6.3% vs. 7.6%, respectively; P = 0.059).

R E S E A R C H
• A low proportion of patients with type 2 diabetes (T2D) experience care that aligns with the American Diabetes Association (ADA) position statement regarding glycated hemoglobin (A1c) testing frequency; the average patient has an A1c level > 7.0% for approximately 10 years and > 8.0% for approximately 5 years before insulin initiation. • Positive effects on diabetes-related outcomes have been reported from studies that evaluated interventions aimed at improving diabetes management delivered in the form of educational mail-outs that targeted patients and physicians. • The majority of intervention studies used a pre-versus postintervention comparison without the inclusion of control groups.

What is already known about this subject
• This randomized controlled study showed that, compared with standard care, an educational intervention program such as "Act on Threes" did not result in a significant difference in treatment intensification with insulin and glycemic control in high-risk patients with T2D, despite an increase in A1c testing frequency in both cohorts during the study period. • An intervention that would increase awareness among physicians and patients of the long-term outcomes of an A1c level > 8.0% and delayed insulin initiation may be better able to help overcome clinical inertia and promote insulin initiation in alignment with the current ADA recommendations.
the impact on outcomes in those exposed versus those not exposed to the interventions. 11,12 The objective of the "Act on Threes" educational intervention was to align patients with T2D and their physicians with the 2012 ADA position statement on T2D management 5 and to assess 3 vital aspects of A1c control: the timely measurement of A1c levels every 3 months; timely treatment intensification to meet A1c goals, with treatment intensified every 3 months if A1c is not at goal; and insulin initiation when appropriate, including in patients already receiving ≥ 3 oral antidiabetes drugs (OADs) with A1c not at goal. Unlike previous studies, the Act on Threes intervention used a rigorous, randomized design with the objective of evaluating the impact of the educational intervention on A1c testing frequency and insulin initiation in patients with T2D who had not received A1c testing or had suboptimal control of their A1c (≥ 8.0%) according to the 2012 ADA position statement. 5 The secondary objective was to identify factors that predict insulin initiation in patients with T2D enrolled in the study.

■■ Methods Study Design
This was a 12-month, prospective, randomized, and interventional study of adult patients with T2D that evaluated the effects of the Act on Threes educational intervention on A1c testing frequency and insulin initiation through the analysis of administrative claims data from the Humana database. Humana is a for-profit health insurance company with more than 13 million U.S.-based customers. Following retrospective identification of patients with high-risk status using medical, pharmacy, laboratory, and enrollment indices of A1c screening frequency; number of noninsulin antihyperglycemic medications; and/or A1c test results as per ADA criteria for A1c testing frequency and antidiabetic treatment modifications 8 (identification period May 1, 2011-April 30, 2013), patients were randomized 3:1 to receive either the Act on Threes educational intervention in conjunction with standard care (intervention group) or standard care alone (control group; Figure 1). Controls were eligible only if the treating physician was a different physician from that of any intervention patient. The full study period included a 12-month pre-intervention (baseline) period and a 12-month postintervention (follow-up) period, with the index date being the date of the most recent diagnosis of T2D during the identification period.

Patients and Physicians
High-risk Medicare Advantage with prescription drug coverage (MAPD) members were eligible if they were aged 18-85 years with ≥ 1 inpatient or 2 physician visits dated ≥ 30 days apart with a primary or secondary diagnosis of T2D using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes 250.x0 or 250.x2. Patients were O ptimal glycemic control is critical in order to reduce the risk of diabetes-related morbidity and complications, as well as to reduce medical costs and health care utilization. However, optimal management of patients with type 2 diabetes (T2D) is a significant clinical challenge for any general practitioner or specialist physician. 1,2 Despite extensive research demonstrating the clinical benefits of glycemic control-and, notably, the benefits of early insulin initiation 3,4 -at least 60% of U.S. patients with T2D enrolled in commercial or Medicaid health maintenance organization plans who are appropriate candidates for stringent glycated hemoglobin (A1c) goals fail to attain the American Diabetes Association (ADA) recommended A1c goal of 7.0% or less. 1,5,6 Even considering the recent emphasis of the ADA on patient-centered care and the appropriateness of less stringent A1c targets such as in older patients with comorbidities, an estimated 54.5% of Medicaid patients, 41.1% of patients on a commercial insurance plan, and 34.5% of Medicare patients fail to achieve an A1c goal of 8.0% or less. 5,6 For patients with good glycemic control and who meet treatment goals, A1c testing is recommended twice a year, with quarterly testing for those not at goal and whose therapy has changed. 7 Yet, in a large U.S. cohort of adult patients with T2D, just 7% were adherent for 1 year to ADA guidelines for A1c testing frequency, and therapy was modified in accordance with guidelines in just 39%. 8 Another study indicated that from T2D diagnosis, the average patient spent approximately 10 years with an A1c level > 7.0% and approximately 5 years with an A1c level > 8.0% before initiating insulin. 9 Patient and physician concerns, particularly in relation to insulin initiation, contribute to delays in treatment intensification in patients with T2D 10 ; this clinical inertia may put patients at risk of diabetes-related complications. 1,2 Managed care organizations and the pharmaceutical industry invest heavily in physician-and patient-targeted educational outreach programs to overcome clinical inertia and patients' reluctance to initiate insulin, but little is currently known about the impact of such programs. Positive results from studies that have evaluated preversus postintervention outcomes of educational initiatives that targeted patients and physicians via mail-outs have encouraged similar interventions in managed care. However, these studies did not employ a randomized controlled study design to evaluate • The lack of effect of the Act on Threes educational intervention challenges the validity of other studies of educational interventions that have shown significant changes in diabetes-related outcomes based on a pre-to postintervention design without the use of appropriate control groups.

What this study adds (continued)
"Act on Threes" Paradigm for Treatment Intensification of Type 2 Diabetes in Managed Care:

Results of a Randomized Controlled Study with an Educational Intervention Targeting Improved Glycemic Control
Patients selected using indicators 1-5 N = 9,159  "Act on Threes" Paradigm for Treatment Intensification of Type 2 Diabetes in Managed Care: Results of a Randomized Controlled Study with an Educational Intervention Targeting Improved Glycemic Control required to have continuous enrollment in the MAPD plan with medical and pharmacy coverage for ≥ 12 months before (baseline) the most recent T2D diagnosis (index date) and current enrollment at the time of identification. Additionally, patients were required to meet 1 of the following high-risk selection indicators: (a) have no A1c evaluations in the 12 months before the index date; (b) concurrent use of ≥ 3 noninsulin antihyperglycemic agents without insulin at any time in the 12 month period before the index date, but without evidence of a medical claim for an A1c test; (c) at least 1 A1c value ≥ 8.0% that occurred ≥ 6 months before the index date, without any A1c evaluations within the previous 6 months; (d) taking ≥ 3 noninsulin antihyperglycemic agents and having ≥ 1 A1c value ≥ 8.0% in the 12 months before the index date; (e) at least 1 A1c value ≥ 8.0% in the 12 months before the index date with no subsequent change in therapy over the course of 3 A1c measures ≥ 8.0% regardless of current therapy. Patients with a prescription claim for insulin in the pre-intervention period were excluded. Patient selection was fully anonymized and compliant with Health Insurance Portability and Accountability Act regulations. Physicians targeted for the intervention were those involved in the treatment of the selected patients.

Act on Threes Intervention
The Act on Threes educational intervention had a 2-pronged approach that incorporated general and targeted T2D educational materials and cover letters, which were simultaneously mailed to patients and their treating physicians twice (see  Appendices A-D, available in online article). The educational materials were concise and used language at a level considered appropriate for patient and physician groups. The materials were developed in association with Humana's in-house clinical task force and communications team and were consistent with ADA clinical guideline recommendations for A1c testing frequency and treatment modification in patients with T2D. General T2D educational materials included patient and physician brochures (Appendices B and D, available in online article). A simple brochure encouraged patients to see their doctors for an A1c test every 3 months until A1c is at goal, then at least every 6 months to maintain goal A1c; talk to their doctors about their medications every 3 months if A1c is not at goal; and talk to their doctors about using insulin if they are on ≥ 3 diabetes medications, and A1c is not at goal. Another brochure was sent to physicians encouraging them to measure their patients' A1c every 3 months; intensify treatment every 3 months for patients not at goal A1c; and consider starting insulin for patients using ≥ 3 noninsulin antihyperglycemic drugs. Targeted educational materials were sent to physicians in the form of patient-specific profiles, outlining the reason why the patient was classified as high risk, together with relevant diabetes-related information such as comorbidities, hospitalizations, medication use, and laboratory A1c results (Appendix E, available in online article). The profiles were presented with 1 patient per page so the data could be easily incorporated into the patient's chart. Follow-up letters were sent to physicians and patients 3 months after the initial mailing. The control group received standard care and did not receive any messaging as part of the study.

Outcome Measures
The primary outcome measures were A1c testing frequency, as recorded in the medical claims, and based on the ADA standard for compliance of ≥ 2 tests per year and insulin initiation in the postintervention period, as recorded in the pharmacy claims. A1c testing frequency was measured in the entire study cohort, which was composed of patients remaining on the same treatment regimen and patients who changed treatment regimen (including those who initiated insulin therapy). A1c levels were examined for the subgroup of patients with available A1c measurements in the pre-and postintervention period.
Act on Threes campaign-related measures during the postintervention period were secondary outcomes and included the following: the number of T2D patients who had ≥ 1 A1c measure; the number of patients who had A1c measures every 3 months; the number of patients who initiated insulin; the number of patients who switched to insulin; and the number of patients with any change in treatment. Other secondary outcome measures were the frequency and proportion of patients who discontinued any OAD used in the 30 days before switching to insulin, with discontinuation defined as a 90-day gap in receipt of medication and including refills, and diabetes therapy intensification, defined as the number of class-level OADs added during the postintervention period in addition to those received in the pre-intervention period. Hypoglycemic events in any setting were identified by ICD-9-CM codes 250.8, 251.0, 251.1, and 251.2.

Statistical Analyses
Descriptive statistics were used to analyze differences in demographic and clinical characteristics between the intervention and control groups. Differences in the proportions of patients with ≥ 2 A1c tests in the pre-and postintervention periods were evaluated for each group using the McNemar test; the Student's t-test was used to evaluate between-group differences. Bivariate comparisons of outcome measures in the intervention and control groups were made using Χ 2 tests to evaluate for statistical significance of differences in proportions of patients. Multiple logistic regression analysis of the entire study cohort was used to assess predictors of insulin initiation in the post hoc analysis. All analyses of data were conducted using SAS software, version 9.1 (SAS Institute, Cary, NC).
The a priori alpha level for all inferential analyses was set at 0.05, and all statistical tests were 2-tailed. For the secondary objective of examining predictors of insulin initiation, an analysis of the data was conducted to evaluate insulin initiation in patients with T2D who had available A1c measures in the pre-and postintervention periods. During the 12-month postintervention period, insulin initiation was examined in  "Act on Threes" Paradigm for Treatment Intensification of Type 2 Diabetes in Managed Care:

Primary and Secondary Outcomes
The percentage of patients with ≥ 2 A1c tests per year was significantly higher postintervention compared with preintervention in the intervention and control groups (P < 0.001; Figure 2). However, when the pre-to postintervention change was compared for intervention and control patients, there was no significant difference between the groups in the proportion of patients who received ≥ 2 A1c tests (P = 0.995). There was also the entire cohort using pharmacy claims, and its determinants were identified from patients' pre-intervention demographic and clinical characteristics.

■■ Results Patient Demographics and Baseline Characteristics
Patient attrition is summarized in Figure 1. A total of 6,243 patients with a mean age of 70 years (43.5% female) were identified in this study; 4,555 patients were randomized to the intervention group and 1,688 patients to the control group. Patients in the intervention group had a significantly higher mean A1c level at baseline compared with the control group (8.66% vs. 8.53%, respectively; P = 0.043). Otherwise, patient demographics and baseline characteristics were similar for the 2 groups (Table 1).  "Act on Threes" Paradigm for Treatment Intensification of Type 2 Diabetes in Managed Care: Results of a Randomized Controlled Study with an Educational Intervention Targeting Improved Glycemic Control 49.6%; P < 0.001) and treatment with a glucagon-like peptide-1 receptor agonist (GLP-1 RA; 8.2% vs. 4.5%; P = 0.015). During the pre-intervention period, patients who initiated insulin had more outpatient visits than those who did not initiate insulin (mean = 11.1 vs. 9.6 per patient; P < 0.001) and were more likely to have an A1c level of ≥ 9.0% (38.6% vs. 17.8%; P < 0.001). Table 3 summarizes the findings of logistic regression analyses used to examine predictors of insulin initiation in this population. The Act on Threes educational intervention had no impact on insulin initiation in either the unadjusted or adjusted models. Significant predictors of insulin initiation in the final adjusted model were Deyo-Charlson Comorbidity Index score (P < 0.001); use of ≥ 3 OADs (P = 0.050); GLP-1 RA use (P = 0.029); and A1c of 8.0%-8.99% (P = 0.029), 9.0%-9.99% (P < 0.001), and ≥ 10% (P < 0.001). A1c < 7.0% significantly reduced the likelihood of insulin initiation in the adjusted model (P < 0.001).

■■ Discussion
Act on Threes was a novel study with a rigorous randomized design aimed at prospectively evaluating the effects of a targeted educational program directed to patients with high-risk T2D and their treating physicians. Act on Threes specifically targeted patients with either no A1c evaluations or those with A1c ≥ 8.0% and, to our knowledge, is the first randomized study of patients with T2D that compared the effects of a targeted educational program between the intervention and control groups in addition to comparing pre-and postintervention results. The main findings of this study were that the Act on Threes intervention, compared with standard care, did not result in any significant differences in the frequency of A1c testing or in the initiation of insulin among high-risk patients with T2D. Also, the Act on Threes intervention did not result in any differences between the intervention and control groups in terms of campaign-related measures. However, in the intervention and control groups, A1c testing was conducted significantly more frequently postintervention compared with pre-intervention.
Although it is difficult to speculate on the reasons for an increase in A1c testing frequency in the control group, there is confidence in our finding of a lack of a between-group effect because no treating physician was involved in the care of intervention and control group patients. Therefore, control group patients would not have been directed to interventionbased measures. We speculate that the absence of an intervention effect was a result of the higher cost of care associated with additional physician visits and A1c testing in a group of Medicare patients assigned to the intervention group. The intervention group, and particularly those on insulin, would be more likely than control patients to experience the Medicare no difference in the A1c level when comparing the pre-to postintervention change for patients in the intervention and control groups (P = 0.240).
The proportion of patients who initiated insulin therapy postintervention was similar for the intervention and control groups (6.3% vs. 7.6%, respectively; P = 0.059). When patients were stratified according to predefined high-risk selection indicators, there were no statistically significant differences in Act on Threes campaign-related measures between the study groups (P > 0.05; Figure 3). Diabetes therapy intensification during the 12-month postintervention period was comparable for the study groups, with most patients not having their OAD therapy intensified during the postintervention period (Table 2).

Characteristics and Predictors of Insulin Initiation
Determinants of insulin initiation were examined in a subgroup of 2,042 patients with available A1c measures in the preand postintervention periods. Of these, 220 (10.8%) patients initiated insulin during the postintervention period. Compared with patients who did not initiate insulin (n = 1,822), patients who initiated insulin (n = 220) had a higher mean Deyo-Charlson Comorbidity Index score (3.46 vs. 2.77; P < 0.001) and a higher prevalence of diabetic nephropathy (11.8% vs. 5.7%; P < 0.001) and hypoglycemia (8.6% vs. 5.4%; P = 0.049) during the pre-intervention period. Compared with patients who did not initiate insulin, a higher proportion of patients who initiated insulin were receiving ≥ 3 OADs (62.3% vs.

Predictors of Insulin Initiation Among Patients with at Least 1 A1c Lab Result (N = 2,042)
"Act on Threes" Paradigm for Treatment Intensification of Type 2 Diabetes in Managed Care: Results of a Randomized Controlled Study with an Educational Intervention Targeting Improved Glycemic Control coverage gap or "donut hole," which would require patient out-of-pocket expenses to fund the increased level of care. The changes seen pre-and postintervention are consistent with other studies reporting the results of educational interventions. A recent study examining the number of patients with diabetes who received retinal screening after a printed, mailed educational intervention was sent to physicians found no effect for intervention. 13 Another study investigating the impact of point-of-care A1c testing on provider compliance and A1c levels in a primary care setting reported a significant difference in pre-to postintervention A1c levels (mean A1c 6 months pre-intervention: 8.1%; 6 months postintervention: 7.7%), but unlike the present study, it did not include a control group. 12 Indeed, the lack of effect between the intervention and control groups in the present study challenges the validity of other studies that have shown significant changes in diabetes-related outcomes based on pre-to postintervention design. 11,12 A cluster randomized controlled trial evaluating the effects of peer support on T2D-related outcomes similarly found no impact of the intervention relative to a control group, but unlike the present study, this study did not employ a pre-versus postintervention comparison. 14 Another possible reason for the lack of intervention effect in the present study relates to resistance encountered from physicians during the educational intervention. A total of 122 calls from providers were received in response to the mailings for reasons that included reporting that a patient was no longer seen in the practice and that the patient/physician pairing listed on the profile was inaccurate (38.5%); calls were also received to provide updated information on the patient's A1c testing frequency or results and to relay information about a positive change in care (32.0%). A few providers also called to request that they not be sent these types of communications and that reading such materials would be "wasting their time," as they did not require help caring for their patients. Although few in number (n = 3; 2.5%), these calls may represent a widely held but unexpressed sentiment that any intervention that distracts physicians from their current focus or approach is unhelpful and time consuming. This resistance, if it is widespread, is of potential concern because it suggests that the delivery of patient-centered care may be a step too far for some physicians; it may also be a contributor to clinical inertia in the case of insulin initiation. Reasons for inertia in the management of T2D previously identified include patients being reluctant to add new medications, especially insulin, to already complex treatment regimens, as well as a lack of familiarity with insulin therapy initiation and failure to intensify on the part of the primary care physician. 15,16 Yet, there is a need to further investigate the reasons for inertia in insulin initiation among the physician community. Possible resistance from the physician aside, it is important to acknowledge that this study made no attempt to measure patient or physician engagement with the educational materials. Thus, it is possible that the physician participation rate was sufficiently low as to affect the observable outcomes and may better explain the absence of an intervention effect.
The Act on Threes initiative tried to influence insulin initiation in a positive way; however, only a small proportion of patients initiated insulin therapy during this study. This outcome suggests that clinical inertia, in conjunction with patient and/or physician reluctance to initiate insulin treatment, remains a problem in the managed care setting. The findings of this study in relation to the use of an educational intervention emphasize that simply increasing A1c testing frequency does not promote treatment intensification. This highlights the complexity of clinical inertia and shows that multiple approaches may be required to address clinical inertia in the managed care setting. The A1c threshold of 9.0% seemed to be a much stronger motivator for treatment intensification, with a 2-to almost 4-fold greater likelihood of insulin initiation associated with an A1c of 9.0% or higher. Other studies have previously reported an association between increasing A1c level and decreasing clinical inertia. [17][18][19] Other significant predictors of insulin initiation in the present study were the Deyo-Charlson Comorbidity Index score, which is a proxy for comorbidities, and use of ≥ 3 OADs or a GLP-1 RA during the pre-intervention period. A potential reason why the use of a GLP-1 RA predicted treatment intensification with insulin may be that such patients are already experienced with injectables and therefore more willing to initiate insulin.
The development and availability of new classes of OADs may further increase the length of time to insulin initiation. In addition to raising the awareness of ADA-recommended A1c testing, an intervention that would increase awareness among physicians and patients regarding the effect on long-term outcomes of an A1c level > 8.0% and delayed insulin initiation may be better able to overcome some of the inertia and promote insulin initiation at lower A1c thresholds.

Limitations
The Act on Threes educational intervention included Medicare T2D patients from the Humana administrative claims database only. Therefore, the results may not be representative of, or generalizable to, all patients with T2D. Further, it is not known what other concomitant interventions the patients received that were not related to this study. For example, managed care organizations are focused on improving quality measures and providing other educational initiatives, with such initiatives potentially affecting the pre-and postintervention outcomes in this study. Although it is unlikely that the 2 arms were affected "Act on Threes" Paradigm for Treatment Intensification of Type 2 Diabetes in Managed Care: Results of a Randomized Controlled Study with an Educational Intervention Targeting Improved Glycemic Control unequally, any impact of such initiatives may have "shadowed" the overall effect of the intervention. As already indicated, no efforts were made to measure patient or physician engagement, which may have been inadequate to achieve an intervention effect. Furthermore, although the educational materials conveyed simple messaging of the ADA, no external testing or validation of the materials was conducted to determine their utility for patients and physicians alike. Finally, personalized A1c goals may be more appropriate because they consider factors such as patient age and health status with the aim of balancing the benefits of glycemic control with its possible risks. 20 In this study, the less aggressive treatment goals may have affected patient and physician responses to educational intervention.

■■ Conclusions
The Act on Threes study is the first randomized study comparing pre-and postintervention results in an interventional and a control group. Our findings did not demonstrate an overall significant effect for the educational intervention on outcomes of A1c testing and the campaign measures. Importantly, these findings highlight a major drawback of numerous previous studies that used a pre-versus postintervention comparison without a control group. Given the significant postintervention increase in A1c testing observed in the intervention and control groups, our findings indicate the potential of such studies to generate flawed conclusions with the risk of inappropriately influencing the management of diabetes. Indeed, had we relied on a similar study design, we would have mistakenly concluded that the intervention resulted in a significant improvement in the primary outcome of percentage of patients with ≥ 2 A1c tests per year. As it was, a similar improvement in the control group negated any intervention effect, leaving us to speculate why.
We suggest that future studies are warranted to determine the effect of educational interventions in aligning physicians and their T2D patients with ADA recommendations. The use of pre-versus postintervention comparisons remains a valid method for determining the effect of an intervention and is particularly relevant for database studies. However, such studies should also use a randomized design that incorporates a control group and focus on a broader segment of the patient population in keeping with the possibility that cost restraints associated with the Medicare model detrimentally affect patient care and the ability of educational interventions to achieve desirable outcomes. Further profiling of determinants of insulin intensification may be helpful in designing and targeting appropriate interventions for patients who are likely to benefit most.