Clinical Inertia in a Randomized Trial of Telemedicine-Based Chronic Disease Management: Lessons Learned

Abstract

Background:

Treatment nonadherence and clinical inertia perpetuate poor cardiovascular disease (CVD) risk factor control. Telemedicine interventions may counter both treatment nonadherence and clinical inertia.

Introduction:

We explored why a telemedicine intervention designed to reduce treatment nonadherence and clinical inertia did not improve CVD risk factor control, despite enhancing treatment adherence versus usual care.

Methods:

In this analysis of a randomized trial, we studied recipients of the 12-month telemedicine intervention. This intervention comprised two nurse-administered components: (1) monthly self-management education targeting improved treatment adherence; and (2) quarterly medication management facilitation designed to support treatment intensification by primary care (thereby reducing clinical inertia). For each medication management facilitation encounter, we ascertained whether patients met treatment goals, and if not, whether primary care recommended treatment intensification following the encounter. We assessed disease control associated with encounters, where intensification was/was not recommended.

Results:

We examined 455 encounters across 182 intervention recipients (100% African Americans with type 2 diabetes). Even after accounting for valid reasons for deferring intensification (e.g., treatment nonadherence), intensification was not recommended in 67.5% of encounters in which hemoglobin A1c was above goal, 72.5% in which systolic blood pressure was above goal, and 73.9% in which low-density lipoprotein cholesterol was above goal. In each disease state, treatment intensification was more likely with poorer control.

Conclusions:

Despite enhancing treatment adherence, this intervention was unsuccessful in countering clinical inertia, likely explaining its lack of effect on CVD risk factors. We identify several lessons learned that may benefit investigators and healthcare systems.

Introduction

For individuals with diabetes mellitus, the risk of dying from cardiovascular disease (CVD) is approximately twice that of nondiabetic patients.¹ Because control of risk factors, such as hypertension, dyslipidemia, and diabetes reduces CVD complications,^2

–5 achieving recommended disease treatment targets is critical.⁶ However, despite the importance of CVD risk reduction in diabetes, evidence-based goals remain elusive.⁷

Two factors often contribute to suboptimal CVD risk factor control. One is treatment nonadherence, which may be influenced by psychosocial, socioeconomic, and other patient factors.⁸ Another common contributor to inadequate risk factor control is provider failure to initiate or intensify therapy when indicated, or “clinical inertia.”^9,10 Clinical inertia contributes to CVD morbidity and mortality in up to 80% of cases.¹¹

A variety of factors may underlie clinical inertia in chronic disease management. Providers may not recognize evidence-based treatment goals, or competing demands may require prioritization over chronic disease management.⁹ Providers may have uncertainty regarding a patient's “true” level of control; for example, they may lack access to self-monitored blood glucose (SMBG) data, or may suspect that a clinic-based blood pressure (BP) value does not reflect the home BP.^12,13 In some cases, providers may reasonably hesitate to intensify therapy due to suspected patient treatment nonadherence or uncertainty about a patient's current medications.¹⁴

The Cholesterol, Hypertension, and Glucose Education (CHANGE) study sought to improve CVD risk factor control in African Americans with type 2 diabetes by reducing treatment nonadherence and mitigating factors that contribute to clinical inertia.¹⁵ We randomized patients to receive usual care with or without a nurse-administered telemedicine intervention; the CHANGE intervention combined monthly self-management education to support medication adherence and quarterly medication management facilitation to support treatment intensification. Based on a validated measure,¹⁶ intervention patients experienced improved self-reported medication adherence at 12 months compared with the usual care group (odds ratio for improved adherence 4.4, p = 0.0008),¹⁵ but did not exhibit relative improvement in diabetes, hypertension, or dyslipidemia outcomes versus usual care.

The purpose of this analysis was to explore treatment intensification patterns in the CHANGE intervention group, to understand why outcomes did not improve despite better self-reported treatment adherence versus usual care. We hypothesized that, despite its medication management facilitation component, the intervention did not promote high rates of treatment intensification (so failed to overcome clinical inertia).

Materials and Methods

Participants in the CHANGE study (ClinicalTrials.gov NCT00815789) received care at two Durham, NC clinics, where primary care providers (PCPs) serve large numbers of African American patients. The Duke University Institutional Review Board approved this study.

Subjects

Study inclusion criteria were age ≥18 years, self-reported black/African American race, ≥1 PCP visit in the past year, a type 2 diabetes diagnosis code within 3 years, and ≥1 hemoglobin A1c (HbA1c) measurement in the past year. Individuals were excluded for dementia, psychosis, or metastatic cancer; dialysis dependence; hospitalization for stroke, myocardial infarction, or coronary revascularization within three months; pregnancy, expected pregnancy, or breastfeeding; nursing home residence; lack of telephone access; severely impaired speech/vision; or inability to speak English.

Eligible patients received a letter that provided instructions for opting out of participation. A research assistant screened interested patients by phone and (if appropriate) arranged an in-person meeting. In this study, patients provided informed consent and underwent a baseline interview. Participants were then randomized to receive the intervention or usual care using a computer-generated, blocked sequence stratified by clinic site. A blinded staff member sealed randomization assignments within sequentially numbered, opaque, identical envelopes, and a research assistant revealed group assignments to participants.

Change Intervention

The 12-month CHANGE intervention included two components: (1) module-based self-management education targeting improved adherence to medication taking and other healthful behaviors, and (2) medication management facilitation, geared toward enhancing treatment intensification by addressing common factors underlying clinical inertia (e.g., uncertainty about a patient's treatment adherence, current medications, or “true” level of control, and inability to manage treatment changes and follow-up during appointments due to competing demands). Intervention nurses administered both components, and communicated remotely with patients and PCPs. The CHANGE approach was based on prior studies^17,18; detail regarding the intervention has been published.^15,19

The intervention nurses delivered self-management education modules during monthly phone encounters according to a schedule. Module material addressed three separate domains that underlie adherence to self-management behaviors: (1) disease management (including knowledge, self-monitoring, and medication use); (2) psychosocial determinants of disease control (including depression, memory, and social support); and (3) tailored behavior change (customized based on baseline patient assessment, could include diet, exercise, smoking cessation, and others). Interventionists delivered modules using software that provided scripts tailored to the patient's current stage of change.²⁰

For the intervention's medication management facilitation component, nurse interventionists contacted PCPs through secure electronic communication after completing the 3-, 6-, and 9-month telephone encounters with patients. The nurses relayed a summary report describing the patient's: (1) self-reported medication adherence (assessed for each medication using a single item: “How are you taking [medication]? Is that the way your doctor instructed you to take it?”); (2) any discrepancies with prescribed medication regimens; (3) CVD risk factor data, including last available HbA1c, clinic BP, and low-density lipoprotein cholesterol (LDL-C), and any available SMBG and BP data; and (4) willingness to accept medication changes. Nurses did not recommend medication changes, but encouraged PCPs to adjust medications as indicated, and offered to facilitate changes by communicating with patients and arranging clinic follow-up or laboratory testing. Each quarterly nurse-PCP contact generated a summary note in the electronic health record (EHR). If a patient's 3-, 6-, or 9-month telephone encounter could not be completed, interventionists did not initiate PCP contact for that period.

Treatment Intensification Measures

To determine how frequently appropriate treatment intensification occurred during the study, we examined all quarterly medication management facilitation encounter notes for each intervention arm patient. We evaluated data at the encounter level for each CVD risk factor of interest (diabetes, hypertension, and dyslipidemia). For each encounter, two investigators (D.E.O. and M.J.C.) abstracted data into a Microsoft Excel database. Abstracted information included patient study identifier, encounter month (3, 6, or 9), clinic site, whether the patient was willing to change medications (through a single-item, yes/no question asked during each encounter), disease control data (using the most recent pre-encounter HbA1c, BP, and LDL-C value available in the EHR), and whether a treatment change was made. Separate investigators overread these abstractions (M.J.C. for D.E.O. and A.B.B. for M.J.C.) to assure accuracy, and additionally abstracted whether the patient reported medication adherence during each encounter (yes/no).

After excluding encounters during which patients expressed unwillingness to change medications, we classified each encounter as either “at goal” or “not at goal” with regard to diabetes, hypertension, and dyslipidemia (making an individual determination for each CVD risk factor). In defining “goal” control, we utilized the American Diabetes Association guidelines from 2009 to 2010, when the intervention was delivered (HbA1c <7.0%, BP <130/80 mm Hg, and LDL-C <100 mg/dL).^21,22 When a patient was “at goal” for a particular CVD risk factor at a given encounter, we excluded that encounter from further analysis for that risk factor (because treatment intensification would be unwarranted with goal control).

For encounters in which a patient was “not at goal” for a given CVD risk factor, we examined actions taken by the patient's provider in response to the medication management facilitation note. We coded each encounter as either “intensification recommended” or “intensification not recommended” for each “not-at-goal” risk factor. Additionally, because medication nonadherence may be a valid reason for deferring treatment intensification, we reviewed each encounter note for evidence of patient-reported nonadherence. In summary, this process categorized each “not-at-goal” encounter as either “intensification recommended,” “intensification not recommended,” or “nonadherent,” with regard to each individual CVD risk factor.

Additional Measures

Demographic and medical data were collected at study baseline. Patients completed a baseline survey assessing health behaviors, health literacy,²³ and self-reported medication adherence,¹⁶ which was repeated by telephone at 12 months.

Outcomes

Coprimary study outcomes were HbA1c, systolic BP, and LDL-C, ascertained using clinic-based measurements from the EHR. We included all available measurements for each outcome within a specific window before and after the 12-month intervention period. This window covered 90 days before baseline through 90 days after study end for HbA1c, 30 days before baseline through 30 days after study end for systolic BP, and 90 days before baseline through 180 days after study end for LDL-C. For each encounter, the associated HbA1c, systolic BP, or LDL-C measurement was the most recent value available in the EHR.

Statistical Analyses

For each CVD risk factor, we descriptively characterized disease control parameters associated with the “not-at-goal” encounters. Specifically, we determined the mean and standard deviation for the relevant parameter associated with the “intensification recommended” and “intensification not recommended” encounters for each CVD risk factor (HbA1c for diabetes, systolic BP for hypertension, and LDL-C for dyslipidemia). All analyses used Microsoft Excel (version 15.20, 2016).

Results

Of 359 enrolled patients, 182 were randomized to receive the CHANGE study intervention. As Table 1 indicates, the intervention group comprised African American patients with type 2 diabetes with a mean age of 56. A majority of patients were female, unmarried, and hypertensive. While 70% of patients completed ≥12 years of schooling, 48% had low health literacy; 35% had an income of <$10,000 per year. Slightly more than half (51%) used insulin to treat their diabetes. At baseline, patients' mean HbA1c and BP were above contemporaneous goals, but mean LDL-C was at goal.

Table 1.

Baseline Characteristics for the Cholesterol, Hypertension, and Glucose Education Study Intervention Arm (n = 182)

VARIABLE	%
Demographics
Age in years, mean (SD)	56 (12)
Female	69
African American race	100
Married	35
Completed <12 years of schooling	30
Low health literacy (REALM score ≤60)	48
Employed	40
Income <$10,000 per year	35
Current smoker	19
Insurance
Private insurance/managed care	40
Medicare	41
Medicaid	15
Uninsured/Worker's compensation	4
Medical history
Diabetes mellitus, type 2	100
Hypertension	94
Coronary artery disease	31
Chronic kidney disease	11
Congestive heart failure	18
Atrial fibrillation	8
Medication use
Number of diabetes agents, mean (SD)	1.6 (0.9)
Patients using insulin	51
Number of antihypertensive agents, mean (SD)	2.6 (1.5)
Number of cholesterol-lowering agents, mean (SD)	0.8 (0.6)
Disease control^a
HbA1c (%), mean (SD)	8.3 (2.1)
Systolic BP (mm Hg), mean (SD)	138.0 (21.7)
LDL-C (mg/dL), mean (SD)	99.7 (39.0)

Patients with missing baseline data: HbA1c n = 155, systolic BP n = 0, LDL-C n = 12.

BP, blood pressure; CHANGE, Cholesterol, Hypertension, and Glucose Education; HbA1c, hemoglobin A1c; LDL-C, low-density lipoprotein cholesterol; REALM, rapid estimate of adult literacy in medicine; SD, standard deviation.

Rates of Medication Intensification

As outlined in the Figure 1, the 182 intervention group patients generated 455 encounters over the 12-month intervention period. In 31, the patient declined medication adjustment, so we excluded these encounters from the analysis. We examined treatment intensification patterns for the remaining 424 encounters.

Fig. 1.

Flow diagram depicting analysis of medication management facilitation encounters in the CHANGE intervention group. BP, blood pressure; CHANGE, Cholesterol, Hypertension, and Glucose Education; HbA1c, hemoglobin A1c; LDL-C, low-density lipoprotein cholesterol.

For diabetes, the patient's last HbA1c was at goal in 172 of 424 encounters, leaving 252 encounters during which the patient had suboptimal diabetes control. Among these 252 encounters, PCPs recommended treatment intensification following 64 (25.4%) encounters. Evidence of nonadherence was present in 18 of 252 encounters (7.1%), which we considered an acceptable reason for not intensifying treatment. Thus, for 170 of the 252 eligible encounters (67.5%), we identified inappropriate nonintensification of diabetes treatment despite recent suboptimal control.

For hypertension, the patient's last systolic BP measurement was at goal in 202 of 424 encounters, leaving 222 in which the patient had suboptimal control. Among these 222 encounters, PCPs recommended treatment intensification following 45 (20.3%) encounters. Evidence of nonadherence was present in 16 of 222 encounters (7.2%). Thus, for 161 of the 222 eligible encounters (72.5%), we identified inappropriate nonintensification of hypertension treatment despite recent suboptimal control.

For dyslipidemia, the patient's last LDL-C was at goal in 266 of 424 encounters and LDL-C was missing in 5, leaving 153 encounters in which the patient had suboptimal lipid control. Among these 153 encounters, PCPs recommended treatment intensification following 31 (20.3%) encounters. Evidence of nonadherence was present in 9 of 153 encounters (5.9%). Thus, for 113 of the 153 eligible encounters (73.9%), we identified inappropriate nonintensification of dyslipidemia treatment despite recent suboptimal control.

Disease Control Associated with Intensified and Nonintensified Encounters

To explore how the degree of CVD risk factor control during encounters related to treatment intensification decisions, we examined HbA1c, systolic BP, and LDL-C in the “not-at-goal” encounters based on whether or not PCPs ultimately recommended treatment intensification (Table 2). For each CVD risk factor, PCPs appeared more likely to recommend treatment intensification with poorer control. The mean (SD) HbA1c associated with encounters in which diabetes treatment intensification was recommended was 9.7% (1.9) versus 8.9% (1.9) for nonintensified encounters. The mean systolic BP for encounters, where antihypertensive treatment intensification was recommended was 149.1 mm Hg (14.2) versus 144.8 mm Hg (15.2) for nonintensified encounters. Finally, the mean LDL-C associated with encounters in which lipid treatment intensification was recommended was 138.4 mg/dL (26.1) versus 132.0 mg/dL (32.6) for nonintensified encounters.

Table 2.

Analysis of Mean Disease Control in “Not-at-Goal” Cholesterol, Hypertension, and Glucose Education Study Encounters With and Without Treatment Intensification

	DIABETES ENCOUNTERS		HYPERTENSION ENCOUNTERS		DYSLIPIDEMIA ENCOUNTERS
	n	MEAN HBA1C (SD) IN%	n	MEAN SBP (SD) IN mm Hg	n	MEAN LDL-C (SD) IN mg/dL
Intensified	64	9.7 (1.9)	45	149.1 (14.2)	31	138.4 (26.1)
Not intensified	170	8.9 (1.9)	161	144.8 (15.2)	112	132.0 (32.6)

SBP, systolic blood pressure.

Discussion

By combining self-management education and medication management facilitation, the CHANGE intervention targeted two factors that commonly underlie suboptimal CVD risk factor control: treatment nonadherence and clinical inertia. Despite enhancing self-reported medication adherence, the CHANGE intervention failed to improve diabetes, hypertension, and dyslipidemia control among African Americans with diabetes. With the current analysis, we explored this disconnect.

Even after accounting for potentially valid reasons for not recommending treatment intensification, such as patient preference and known medication nonadherence, PCPs frequently did not recommend treatment intensification. Nonintensification of diabetes medications occurred in 67.5% of encounters in which HbA1c was above goal, 72.5% of encounters in which systolic BP was above goal, and 73.9% in which LDL-C was above goal. These rates suggest that CHANGE's medication management facilitation approaches did not counter clinical inertia effectively, despite its intention to address barriers to treatment intensification, such as uncertainty about treatment adherence, current medications, or “true” level of control, and competing demands.

Given the role of clinical inertia in perpetuating poor chronic disease control,⁹ appropriate treatment intensification is important to effective care. Solely behavioral health services interventions typically have a relatively modest clinical impact; one meta-analysis showed that behavioral interventions for type 2 diabetes reduce HbA1c by 0.44%.²⁴ To assure maximum impact, health services interventions should therefore support treatment intensification. The suboptimal rates of treatment intensification we saw in this analysis help explain the intervention's lack of impact on HbA1c, systolic BP, and LDL-C. Our findings suggest that without adequate treatment intensification, the self-management education-based approaches used in the CHANGE intervention were insufficient to impact HbA1c, BP, and LDL-C, even though self-reported adherence improved. Given our cohort, effective treatment intensification may be particularly important for African American populations. This present analysis is consistent with prior reviews of culturally tailored self-management education in minority racial groups with type 2 diabetes, which show minimal effect on HbA1c at 12 months,^25,26 and no significant effect on BP or LDL-C.²⁵

While this analysis suggests that suboptimal treatment intensification contributed to CHANGE's lack of impact, the question of why our approach was not more effective warrants further consideration. There are multiple potential reasons for the low rates of treatment intensification in this study. To begin, our approach may not have accounted for all relevant factors underlying clinical inertia. For example, although our interventionists summarized all recent disease control data, the available information may not have sufficiently relieved PCP uncertainty about patients' true control.¹² Powers et al. reported that hypertension control cannot be reliably classified based on single BP values; in such cases, uncertainty can be reduced only by considering several measurements.¹³ Alternatively, persistent clinical inertia may have related to a lack of comfort with medication management. Insulin-treated patients pose particular management challenges,²⁷ which may have prevented some PCPs from intensifying therapy even when they recognized the need to do so. Still another potential cause for residual clinical inertia is that PCPs may have remained vulnerable to competing demands. We had hoped that providing easy access to data and help implementing changes would facilitate intensification; however, PCPs may have perceived the interventionists' electronic communications as yet another item on an already full list. We previously reported that PCPs did not respond to 24% of encounter communications,¹⁵ which supports competing demands as a challenge.

A different potential contributor to our approach's insufficient effect is that nurses couched within PCPs' clinics did not deliver the intervention. We chose a centralized model because: (1) use of experienced nurse interventionists to deliver self-management interventions has been effective in our group's studies^28,29; and (2) we were concerned about the effect of high staff turnover at the study clinics. However, a lack of familiarity with our nurse interventionists may have lessened PCP comfort with the intervention.

While we can theorize regarding contributors to clinical inertia in the CHANGE study, this analysis was not designed to identify specific causal factors. Our findings therefore call for further exploration of causes for insufficient treatment intensification in telemedicine interventions, with the goal of guiding intervention redesign to mitigate these barriers. Of note, data suggest that empowering interventionists to modify medications without waiting for PCP approval may enhance intervention potency in diabetes.³⁰ Use of such independent medication managers could compound the challenges of intervention implementation and scaling,³¹ but may represent one plausible strategy for circumventing potential contributors to insufficient treatment intensification.

Importantly, compared with the “not-at-goal” encounters, where intensification was recommended, we found that PCPs appeared less likely to intensify treatment among patients with lesser degrees of poor control. Such decisions may at times represent good judgment rather than clinical inertia, because poor control does not always recur when PCPs choose not to intensify therapy for mildly aberrant values.³²

Limitations

In addition to the limitations discussed above, we recruited an exclusively African American population from two clinics within a single academic health system, which may affect the generalizability of our results. Our patient population had relatively good baseline disease control, which may likewise limit generalizability to more poorly controlled populations. Although the treatment goals we employed for the CHANGE study differ from current standards, it is likely that the potential issues and challenges we have identified remain relevant regardless of specific targets. Finally, use of a more convenient communication platform (e.g., EHR-integrated messaging, which was unavailable during the study) may have facilitated treatment intensification.

Conclusions

This analysis demonstrates that a telemedicine intervention targeting clinical inertia in CVD risk factor management did not result in high rates of treatment intensification. Our findings clarify why the intervention did not improve clinical outcomes despite its effect on treatment adherence. We have identified several factors that may explain why our approach did not overcome clinical inertia, which should serve as lessons learned for health services investigators and healthcare systems.

Footnotes

Acknowledgments

This work was supported by grants from the Robert Wood Johnson Foundation, the Kate B. Reynolds Foundation, and the Durham Veterans Affairs (VA) Center for Health Services Research in Primary Care (COIN CIN 13-410). M.J.C. is supported by a VA Health Services Research and Development Career Development Award (CDA 13-261). H.B.B. is supported by a VA Health Services Research and Development Research Career Scientist award (RCS 08-027). The views expressed in this article are the authors' and do not necessarily represent the Department of Veterans Affairs.

Disclosure Statement

No competing financial interests exist.

References

Preis

, Hwang

, Coady

, et al. Trends in all-cause and cardiovascular disease mortality among women and men with and without diabetes mellitus in the Framingham Heart Study, 1950 to 2005. Circulation, 2009; 119:1728–1735.

James

, Oparil

, Carter

, et al. 2014 evidence-based guideline for the management of high blood pressure in adults: Report from the panel members appointed to the Eighth Joint National Committee (JNC 8). JAMA, 2014; 311:507–520.

LaRosa

, He

, Vupputuri

. Effect of statins on risk of coronary disease: A meta-analysis of randomized controlled trials. JAMA, 1999; 282:2340–2346.

Holman

, Paul

, Bethel

, Matthews

, Neil

. 10-year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med, 2008; 359:1577–1589.

Ray

, Seshasai

, Wijesuriya

, et al. Effect of intensive control of glucose on cardiovascular outcomes and death in patients with diabetes mellitus: A meta-analysis of randomized controlled trials. Lancet, 2009; 373:1765–1772.

Fox

, Coady

, Sorlie

, et al. Trends in cardiovascular complications of diabetes. JAMA, 2004; 292:2495–2499.

Stark Casagrande

, Fradkin

, Saydah

, et al. The prevalence of meeting A1C, blood pressure, and LDL goals among people with diabetes, 1988–2010. Diabetes Care, 2013; 36:2271–2279.

Bosworth

, Oddone

. A model of psychosocial and cultural antecedents of blood pressure control. J Natl Med Assoc, 2002; 94:236–248.

Phillips

, Branch

, Cook

, et al. Clinical inertia. Ann Intern Med, 2001; 135:825–834.

10.

Khunti

, Wolden

, Thorsted

, Andersen

, Davies

. Clinical inertia in people with type 2 diabetes: A retrospective cohort study of more than 80,000 people. Diabetes Care, 2013; 36:3411–3417.

11.

O'Connor

, Sperl-Hillen

, Johnson

, et al. Clinical inertia and outpatient medical errors. Advances in patient safety: from research to implementation, volume 2: concepts and methodology. Vol 2 (of 4). Rockville, MD: Agency for Healthcare Research and Quality, 2005:293–308.

12.

Safford

, Shewchuk

, Qu

, et al. Reasons for not intensifying medications: Differentiating “clinical inertia” from appropriate care. J Gen Intern Med, 2007; 22:1648–1655.

13.

Kerr

, Zikmund-Fisher

, Klamerus

, Subramanian

, Hogan

, Hofer

. The role of clinical uncertainty in treatment decisions for diabetic patients with uncontrolled blood pressure. Ann Intern Med, 2008; 148:717–727.

14.

Powers

, Olsen

, Smith

, Woolson

, Bosworth

, Oddone

. Measuring blood pressure for decision making and quality reporting: Where and how many measures?. Ann Intern Med, 2011; 21; 154:781–788.

15.

Crowley

, Powers

, Olsen

, et al. The Cholesterol, Hypertension, And Glucose Education (CHANGE) study: Results from a randomized controlled trial in African Americans with diabetes. Am Heart J, 2013; 166:179–186.

16.

Morisky

, Green

, Levine

. Concurrent and predictive validity of a self-reported measure of medication adherence. Med Care, 1986; 24:67–74.

17.

Bosworth

, Olsen

, Grubber

, et al. Two self-management interventions to improve hypertension control: A randomized trial. Ann Intern Med, 2009; 151:687–695.

18.

Bosworth

, Powers

, Olsen

, et al. Home blood pressure management and improved blood pressure control: Results from a randomized controlled trial. Arch Intern Med, 2011; 171:1173–1180.

19.

Powers

, King

, Ali

, et al. The Cholesterol, Hypertension, and Glucose Education (CHANGE) study for African Americans with diabetes: Study design and methodology. Am Heart J, 2009; 158:342–348.

20.

Prochaska

, Velicer

. The transtheoretical model of health behavior change. Am J Health Promot, 1997; 12:38–48.

21.

Parker

, Baker

, Williams

, Nurss

. The test of functional health literacy in adults: A new instrument for measuring patients' literacy skills. J Gen Intern Med, 1995; 10:537–541.

22.

American Diabetes Association. Standards of medical care in diabetes—2009. Diabetes Care, 2009; 32 Suppl 1:S13–S61.

23.

American Diabetes Association. Standards of medical care in diabetes—2010. Diabetes Care, 2010; 33 Suppl 1:S11–S61.

24.

Health Quality Ontario. Behavioural interventions for type 2 diabetes: An evidence-based analysis. Ont Health Technol Assess Ser, 2009; 9:1–45.

25.

Creamer

, Attridge

, Ramsden

, Cannings-John

, Hawthorne

. Culturally appropriate health education for Type 2 diabetes in ethnic minority groups: An updated Cochrane Review of randomized controlled trials. Diabet Med, 2016; 33:169–183.

26.

Hawthorne

, Robles

, Cannings-John

, et al. Culturally appropriate health education for type 2 diabetes mellitus in ethnic minority groups. Diabet Med, 2010; 27:613–623.

27.

Khunti

, Nikolajsen

, Thorsted

, Andersen

, Davies

, Paul

. Clinical inertia with regard to intensifying therapy in people with type 2 diabetes treated with basal insulin. Diabetes Obes Metab, 2016; 18:401–409.

28.

Jackson

, Oddone

, Olsen

, Powers

, Grubber

, McCant

, Bosworth

. Racial differences in the effect of a telephone-delivered hypertension disease management program. J Gen Intern Med, 2012; 27:1682–1689.

29.

Frosch

, Uy

, Ochoa

, et al. Evaluation of a behavior support intervention for patients with poorly controlled diabetes. Arch Intern Med, 2011; 171:2011–2017.

30.

Pimouguet

, Le Goff

, Thiébaut

, Dartigues

, Helmer

. Effectiveness of disease-management programs for improving diabetes care: A meta-analysis. CMAJ, 2011; 183:E115–E127.

31.

Zullig

, Gellad

, Moaddeb

, et al. Improving diabetes medication adherence: Successful, scalable interventions. Patient Prefer Adherence, 2015; 9:139–149.

32.

Crowley

, Smith

, Olsen

, et al. Treatment intensification in a hypertension telemanagement trial: Clinical inertia or good clinical judgment?. Hypertension, 2011; 58:552–558.