who can help me?

Results of a Successful Telephonic Intervention to Improve Diabetes Control in Urban Adults A randomized trial

ELIZABETH A. WALKER, PHD, RN1

CELIA SHMUKLER, MD2

RALPH ULLMAN, MBA, MS2

EMELINDA BLANCO, MED3

MELISSA SCOLLAN-KOLIOPOULUS, EDD, ANP4

HILLEL W. COHEN, DRPH, MPH3

OBJECTIVE — To compare the effectiveness of a telephonic and a print intervention over 1 year to improve diabetes control in low-income urban adults.

RESEARCH DESIGN AND METHODS — A randomized trial in Spanish and English comparing a telephonic intervention implemented by health educators with a print intervention. Participants (N � 526) had an A1C �7.5% and were prescribed one or more oral agents. All were members of a union/employer jointly sponsored health benefit plan. Health coverage included medications. Primary outcomes were A1C and pharmacy claims data; secondary outcomes included self-report of two medication adherence measures and other self-care behaviors.

RESULTS — Participants were 62% black and 23% Hispanic; 77% were foreign born, and 42% had annual family incomes �$30 thousand. Baseline median A1C was 8.6% (interquartile range 8.0 –10.0). Insulin was also prescribed for 24% of participants. The telephone group had mean � SE decline in A1C of 0.23 � 0.11% over 1 year compared with a rise of 0.13 � 0.13% for the print group (P � 0.04). After adjusting for baseline A1C, sex, age, and insulin use, the difference in A1C was 0.40% (95% CI 0.10 – 0.70, P � 0.009). Change in medication adherence measured by claims data, but not by self-report measures, was significantly associated with change in A1C (P � 0.01). Improvement in medication adherence was associated (P � 0.005) with the telephonic intervention, but only among those not taking insulin. No diabetes self-care activities were significantly correlated with the change in A1C.

CONCLUSIONS — A 1-year tailored telephonic intervention implemented by health edu- cators was successful in significantly, albeit modestly, improving diabetes control compared with a print intervention in a low-income, insured, minority population.

Diabetes Care 34:2–7, 2011

I mproving glycemic control in type 2 diabetes significantly decreases the risk of serious chronic complications such

as retinopathy, neuropathy, and ne- phropathy, as shown by large-scale clini- cal trials from the last 2 decades (1,2). These studies, along with smaller trials, set the stage for evidence-based medical management of diabetes (3). Although ef-

fective therapies for management have been developed, treatment goals are often not reached— especially in lower income and minority populations (4)—and many individuals find it challenging to perform routine self-management (5). Critical re- views of the scientific literature on inter- ventions to improve glycemic control show promising results for improved pro-

cesses of care, such as screening for com- plications and laboratory tests, as well as for behavioral interventions and self- management training (6,7).

Evidence is emerging for the use of telephonic interventions to improve dia- betes self-care and health outcomes; stud- ies include use of automated calls with nurse follow-up (8) or calls implemented by individuals with graduate degrees (9). Telephonic interventions may enhance self-care adherence (10) by offering the opportunity to customize information to individuals under real-world conditions. Nonetheless, the efficacy of telephonic in- terventions in all populations and settings has not been established, and improve- ments in health outcomes for patients re- main challenging even with many new pharmaceutical agents becoming avail- able and combinations of type 2 diabetes medications becoming a standard of care.

As an adjunct to diabetes self- management education and medical care, a telephonic intervention by health edu- cators may provide the coaching and mo- tivation needed for individuals to perform diabetes self-management activities over time, especially medication adherence. The Improving Diabetes Outcomes (I DO) study aimed to evaluate the incre- mental effect of a tailored telephone inter- vention, in English and Spanish, on the mean A1C levels and medication adher- ence beyond that achieved with the mail- ing of print self-management materials. The population is insured, lower-income, mostly minority individuals who had health care and medication benefits cov- ered in full by their labor union/employer plan. However, the study protocol al- lowed only telephonic and print contact with participants so that individuals who might not have agreed to participate in more conventional in-person studies could take part. The main study outcomes were changes in A1C and medication ad- herence. The study also sought to deter- mine what demographic and behavioral factors might mediate the effect of the in- terventions. We now report the main re-

● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●

From the 1Department of Medicine and Endocrinology, Albert Einstein College of Medicine, Bronx, New York; 21199SEIU Benefit and Pension Funds, New York, New York; the 3Department of Epidemiology and Population Health, Albert Einstein College of Medicine, Bronx, New York; and the 4Department of Medicine, University of Medicine and Dentistry of New Jersey, Newark, New Jersey.

Corresponding author: Elizabeth A. Walker, [email protected]. Received 26 May 2010 and accepted 23 September 2010. DOI: 10.2337/dc10-1005. Clinical trial reg. no. NCT00179374, clinicaltrials.gov. © 2011 by the American Diabetes Association. Readers may use this article as long as the work is properly

cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons. org/licenses/by-nc-nd/3.0/ for details.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

See accompanying editorial, p. 240.

C l i n i c a l C a r e / E d u c a t i o n / N u t r i t i o n / P s y c h o s o c i a l R e s e a r c h O R I G I N A L A R T I C L E

2 DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 care.diabetesjournals.org

sults of this translational randomized controlled behavioral trial, including self- care data to more fully explain the results.

RESEARCH DESIGN AND METHODS — The I DO study is a ran- domized controlled behavioral interven- tion study comparing the effectiveness of a telephonic intervention with a print (ac- tive control) intervention. It was devel- oped at the Einstein Diabetes Research and Training Center in collaboration with a union/employer jointly sponsored health benefit plan (1199SEIU Benefit and Pension Funds). As previously de- scribed (11), eligible participants were adult (�30 years of age) members of the health care worker union Fund based in New York City. These Fund members in- clude current full-time health workers or their spouses. The majority of members are service and clerical workers in nursing homes or hospitals, and others work as home health attendants. The Fund pro- vides full coverage of prescription medi- cations, medical visits, hospitalizations, and laboratory tests. Eligible participants had to read and speak English or Spanish, with no evidence of cognitive impair- ment. Eligibility also included the pre- scription of at least one oral glucose- lowering agent (OGLA) in the year prior to enrollment. The eligible A1C was �7.5%, which is above the usual manage- ment goal of �7% (3), but would provide a margin for lowering the A1C in a tele- phonic intervention with no in-person contact without raising safety concerns.

The study protocol aimed to evaluate interventions among individuals who might face challenges in completing in-person diabetes self-management edu- cation programs. It did not include any face-to-face interactions. Oral informed consent and Health Insurance Portability and Accountability Act (HIPAA) authori- zation were obtained by telephone with approval of the institutional review board o f t h e A l b e r t E i n s t e i n C o l l e g e o f Medicine.

There was a two-step recruitment process. The Fund database was used to identify members who might be eligible, and they were telephoned by study staff. If a person seemed eligible and completed a screening questionnaire, oral informed consent was documented. The second step was the mailing and completion of an A1C capillary blood test kit. Individuals with lab results of A1C �7.5% were en- rolled and randomized using a computer-

ized randomization scheme to either the telephone or the print intervention group.

Interventions All telephone participants could receive up to 10 calls at 4- to 6-week intervals from their health educator over the 1-year inter- vention. Health educators were trained and supervised by a certified diabetes educator nurse. Calls were tailored to the participant- reported needs but focused primarily on diabetes medication adherence and secondarily on lifestyle changes through healthy eating and physical activity. Prob- lem solving (12), goal setting (13), commu- nication skills, and preplanning for medical visits were important elements in the inter- vention. The protocol was based on im- proving empowerment and self-efficacy (14) using social-ecological approaches (15). Health educators used a manual to guide the telephone call content, but partic- ipants were encouraged to choose topics for each call. See the online appendix supple- mentary Table A1 (available at http://care. diabetesjournals.org/cgi/content/full/dc10- 1005/DC1) for an example of a call log that both guided and documented implementa- tion of the intervention. All participants received selected high-quality self-man- agement materials by mail after random- ization. Only telephone participants were prompted by health educators to use these materials.

Measures The primary outcome was change in A1C, measured only at baseline and postinter- vention using mail-in kits with “filter paper” methodology (also called “dry- dot”) from a laboratory vendor, Home Healthcare Laboratory of America (“Lab- in-an-Envelope”) (16). This A1C test pro- cessed with a Roche analyzer had been approved by the National Glycosylation Standardization Program (17). Partici- pants were asked to call the health educa- tor to guide them through the blood sampling while using a spring-loaded lan- cet to draw blood from their fingertips and fill in one to three circles (1.2 cm diameter) on a special filter paper card. This card was then mailed directly to the laboratory in a prepaid envelope for anal- ysis. A1C values from the filter paper method have been reported to corre- spond to those obtained by conventional venous whole-blood samples (18,19). If insufficient blood was obtained for a valid result, another test kit was sent to participants.

Medication adherence measures Pharmacy claims (i.e., administrative) data from the Fund, including each OGLA prescription filled, its class, the date, number of pills dispensed, and number of pills per day, were used to cal- culate a medication possession ratio (MPR) for each participant. This type of measure of medication adherence has been used in many studies (20,21). For each class of OGLA taken by a participant within the previous year, the number of pill-days available from each filled pre- scription was calculated. For each partic- ipant, MPRs (number of days’ supply of pills dispensed in 1 year/365) for the 1 year prior to randomization (baseline) and 1 year post randomization (follow- up) were calculated (range 0 –1) for each OGLA class, and then an average of the class MPRs was used to denote separately the participant’s pre- and postinterven- tion MPR. The methods and rationale for this approach have been previously de- scribed (11). A recording of insulin use during the study year was categorized “ever” or “never” on the basis of prescrip- tion orders for any insulin product.

Other diabetes self-management be- haviors were collected by telephone at baseline and end of study. The four-item Morisky Self-Reported Medication- Taking Scale (22) was administered, and scores �2 were considered poor adher- ence to diabetes medications. The Sum- mary of Diabetes Self Care Activities (SDSCA) (23) scale was also adminis- tered, including a single medication ad- herence item: How many days in the most recent week were diabetes pills taken as prescribed? This was treated as a nonpara- metric continuous variable (0 –7 days) and categorized as adherent (7 days) or not. Other SDSCA survey items addressed healthy eating and physical activity and were analyzed similarly. Hours of TV watching per day were recorded in cate- gories (0, 1, 2, 3, 4, �4 h) and dichoto- mized as �2 or �2 h per day. Self- reported demographics including sex, age, race/ethnicity, work status, marital status, income, education, and birthplace were collected, as were other characteris- tics including self-reported height and weight for calculating BMI, years since di- abetes diagnosis, and insulin use in the previous year.

Statistical analysis The study outcomes, change in A1C (�A1C) and change in MPR (�MPR), were calculated as follow-up minus base-

Walker and Associates

care.diabetesjournals.org DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 3

line values (negative values represent a decline) and were assessed for normality assumptions. �MPR was also dichoto- mized as �20 percentage points (e.g., go- ing from 60 to 80%) because very small changes were not expected to have a meaningful impact on A1C. Changes in SDSCA during follow-up were also calcu- lated both as continuous variables (days) and categorized as improved, worsened, or remained the same. Tests of bivariate associations with study arm were per- formed similarly to the comparison of baseline characteristics. Analyses of �A1C and �MPR outcomes were always adjusted for baseline levels. Adjustments for potential confounders or mediators were performed using linear regression models for continuous outcomes and bi- nary logistic models for dichotomous out- comes. To test potential mediation, baseline MPR and �MPR �20% were added to the model predicting �A1C. The number of educator calls received by par- ticipants in the telephone group was used as a proxy for intensity of the interven- tion. Among those in the telephone group, the number of calls completed during the intervention (range 0 –10) was categorized as 0 –5, 6 – 8, and 9 –10, and these were entered into regression models as dummy variables with print group al- location as reference. A test for trend of the association of these call categories with �A1C was also performed. Baseline values of the outcome variables were available as an inclusion criterion prior to randomization, but not all participants provided follow-up data. Outcome anal- yses were performed for those with com- plete data with sensitivity analyses using two alternate imputation methods to sim- ulate intention-to-treat analyses. Imputa- tion for missing outcome data were carried out with STATA (version 11) mul- tiple imputation procedure based on a Bayesian paradigm pooling 100 repeated imputations taking into account baseline A1C, age, sex, insulin use, and baseline MPR. An alternate imputation used base- line A1C values for missing follow-up that in this study was the same as a last obser- vation carried forward (LOCF) approach (24). Those with missing outcome data were compared by study arm to assess as- sumptions of missing at random. Residu- als-based regression diagnostics were performed to check linear regression model assumptions, and first-order inter- actions of covariates with study arm were tested with interaction product terms while simultaneously adjusting for main

effects terms. Hosmer-Lemeshow test for goodness-of-fit was performed for binary logistic models and first-order interac- tions were assessed.

RESULTS — The study flow diagram is in online appendix as supplementary Figure A1; it shows the database recruit- ment pool of 8,083 adults with diabetes taking OGLAs. Of the 4,548 individuals assessed for eligibility, 4,021 were ex- cluded (ineligible 55%, refused 45%), and 527 individuals were randomized, with intention-to-treat analysis of 526 cases. Description of baseline characteris- tics by group and total is found in Table 1. Participants were mainly minority in

terms of race/ethnicity, and were lower- income, middle-aged, and foreign born. The median baseline A1C was 8.6% (in- terquartile range 8.0 –10).

Primary outcomes Among the 444 participants (84.4%) with follow-up A1C, the 228 in the telephone group exhibited a mean � SE decline in A1C of 0.23 � 0.11% over the study year compared with a rise of 0.13 � 0.13% for the 216 in the print group (P � 0.04). After adjusting for baseline A1C, sex, age, and insulin use, the difference in �A1C between telephone and print groups was 0.40% (95% CI 0.10 – 0.70, P � 0.009). There was no strong evidence for media-

Table 1—Participant characteristics at baseline

Telephone group Print group Total

n 262 264 526 Female (%) 68.3 65.9 67.1 Race/ethnicity (%)

Black 61.5 61.7 61.6 Hispanic 24.8 20.5 22.6 White 5.7 6.1 5.9 Other 8.0 11.7 9.9

Age (years) 55.7 � 7.4 55.4 � 7.2 55.5 � 7.3 Married (%) 59.2 63.6 61.4 Foreign born* (%) 75.2 78.4 76.8 Spanish preferred (%) 18.7 12.9 15.8 Duration of diabetes (years) 8.8 � 6.8 9.5 � 6.4 9.2 � 6.6 Duration of diabetes (%)

�6 years 37.0 30.7 33.8 6–10 years 33.2 34.8 34.0 �10 years 29.8 34.5 32.1

Employed full time (%) 73.3 74.6 74.0 Household income (%)

�$20,000 17.2 14.4 15.8 $20–29,000 26.7 26.5 26.6 $30–39,000 29.0 29.2 29.1 $40–49,000 10.7 9.1 9.9 �$50,000 16.4 20.8 18.6

Education (%) �8th grade 16.4 16.7 16.5 9–11th grade 11.8 10.2 11.0 HS or GED 36.3 28.8 32.5 Some college 22.1 26.1 24.1 �College 13.4 18.2 15.8

Self-reported insulin use (%) 21.0 25.0 23.0 Insulin Rx in last year (%) 23.3 24.6 24.0 �2 diabetes pill classes (%) 68.7 68.2 68.4 BMI (kg/m2) 31.8 � 6.2 30.7 � 6.0 31.2 � 6.1 A1C (%) 8.6 (8.0–9.6) 8.7 (8.0–10.2) 8.6 (8.0–10.0) Morisky scale �2 (%) 35.1 38.6 36.9 Report taking diabetes pills

�7 days per week (%) 27.9 25.4 26.6

Data are means � SD or median (interquartile range). *Foreign born does not include those born in Puerto Rico. GED, high school equivalency; HS, high school; Rx, prescription.

Improving diabetes outcomes study

4 DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 care.diabetesjournals.org

tion of the �A1C by �MPR. With regard to the �A1C outcome, no statistically sig- nificant first order interactions with inter- vention group were observed.

When �MPR was assessed as an out- come variable, whether as a continuous variable or as �20% improvement, statis- tically significant (P � 0.04 and 0.01, re- spectively) interactions of intervention with insulin use (n � 141, 26.8%) during the 12-month study period were ob- served. �MPR as a continuous variable was not significantly associated with the telephone intervention either among those taking (P � 0.23) or not taking (P � 0.39) insulin, whereas �MPR �20% was significantly associated (P � 0.005) with the telephone intervention after adjusting for baseline MPR, age, and sex among those not taking insulin, but not among those taking insulin (P � 0.28). Among those not taking insulin, there was a significant (P � 0.001) lin- ear trend with �MPR �20% for the numbers of intervention calls received. Significant associations with interven- tion calls compared with print were only observed for those receiving at least six telephone calls (Table 2).

Secondary outcomes Attempts were made to complete 10 phone calls over 12 months to telephone participants (mean � SD number of com- pleted calls was 7.9 � 2.1). Fewer phone calls resulted from participants being un- reachable or refusing the telephone call. Only 3% (n � 7) of participants had no phone calls even after much staff effort. Mean length of each call was 14.1 � 4.6 min. Having at least six completed phone calls was associated with significant im- provement in A1C (Fig. 1).

Table 2 highlights the differences be- tween those who took insulin and at least one OGLA and those who took only an OGLA. The telephone intervention was not associated with a change in medica- tion adherence (�20% MPR) if the regi- men included insulin. Despite the lack of a statistically significant association of �MPR with the intervention among the minority taking insulin and an OGLA, for the group as a whole the multivariable model provided evidence that the im- provement of MPR was a mediator of the intervention association with improved glycemic control.

From the SDSCA survey, two items showed significant improvement associ- ated with the telephone intervention: number of days per week following a

healthy eating plan and number of days with �30 min of exercise. The other items, as well as hours of TV watched per day, showed a direction toward im- provement associated with the telephone intervention, but not significantly so. However, none of the changes in SDSCA or TV watching were significantly corre- lated with �A1C. Although �MPR de- rived from pharmacy claims data were significantly (P � 0.01) associated with �A1C in the adjusted model, changes in the two self-report medication adherence

measures (number of days taking medica- tion as prescribed item from the SDSCA and the Morisky score) were not signifi- cantly associated with �A1C.

Missing values and analysis with imputation Of the 526 randomized participants, fol- low-up A1C values were not available for 15.6% (18.2% telephone, 13.0% print, P � 0.10). Of the 82 with missing values, 5 (2 deaths and 3 withdrawals) also had missing values for the follow-up MPR. For

Figure 1—Decline in A1C, expressed as median (interquartile range), per category of telephone intervention intensity (number of calls) compared with print group (no calls), estimated in a multiple linear regression model adjusting for baseline A1C, age, sex, insulin use, and improve- ment in MPR �20%.

Table 2—Adjusted odds ratios for change in MPR >20% stratified by insulin use during study

No insulin use (n � 385) Insulin use (n � 141)

OR (95% CI)* P OR (95% CI)* P

Print Reference Reference 0–5 Calls† 1.0 (0.4–2.8) 0.98 0.3 (0.0–2.8) 0.29 6–8 Calls 1.9 (1.0–3.5) 0.04 0.6 (0.2–2.1) 0.41 9–10 Calls 2.6 (1.4–4.6) 0.002 0.4 (0.3–2.2) 0.61 Call linear trend 0.001 0.88

Baseline MPR 0.04 (0.02–1.3) �0.001 0.01 (0.001–0.10) �0.001 Age (years) 1.0 (1.0–1.0) 0.81 0.9 (0.9–1.0) 0.03 Male 0.9 (0.6–1.6) 0.81 1.0 (0.4–2.6) �0.99 Telephone 2.0 (1.2–3.2) 0.005 0.6 (0.3–1.5) 0.28

*Odds ratio (OR) (95% CI) estimated with binary logistic regression models. †Call categories for the telephone intervention with print as reference. Linear trend is across the categories. Telephone gives the overall odds ratio (irrespective of number of calls) with print as reference, estimated in separate adjusted models.

Walker and Associates

care.diabetesjournals.org DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 5

all the baseline characteristics in Table 1, there were no statistically significant asso- ciations with those missing a follow-up A1C among the print group; there was a single significant association in the tele- phone group, with those missing an A1C being (mean � SE) 3.3 � 1.3 years younger than those not missing an A1C. Median baseline A1C was 0.6% higher for those missing in the print group (P � 0.07), and in the telephone group the dif- ference in median was 0.2% (P � 0.54). Using the multiple imputation approach, being in the telephone group compared with print group was significantly associ- ated with greater decline in A1C, whether adjusting only for baseline A1C or also adjusting for age, sex, insulin use, and �MPR (both P � 0.03). These significant associations were also seen (both P � 0.01) when the LOCF imputation ap- proach was used.

CONCLUSIONS — A tailored tele- phonic behavioral intervention imple- mented by health educators under the supervision of a certified diabetes educa- tor nurse was successful in significantly, albeit modestly, improving A1C com- pared with a print intervention. Greater intensity of the intervention (�6 calls) was associated with greater improvement in A1C.

A possible explanation for the re- ported differences in intervention effec- tiveness for medication adherence related to insulin use (Table 2) may be that being prescribed insulin in combination with an OGLA is a regimen complexity that re- duces adherence to the OGLAs. An alter- n a t i v e e x p l a n a t i o n m a y b e t h a t nonclinical health educators, though su- pervised by a nurse certified diabetes ed- ucator, may not have been as effective in medication adherence counseling for par- ticipants also on various insulin regimens as they were with those on OGLAs alone.

Only a few self-care activity changes on the SDSCA were significantly associ- ated with the intervention. It is possible, however, that there was an overall cumu- lative effect on glycemic control of small improvements in multiple self-care activ- ities, even if they were individually too small to show significant associations with the intervention.

Limitations The dry-dot methodology for the A1C measure completed by the participant and mailed to the laboratory had its own limitations, which were imposed by the

nature of the protocol to not require sub- ject visits to a lab or research center. This A1C methodology may contribute to greater measurement variability. In this randomized trial, it would not be ex- pected to introduce a differential bias; and if a nondifferential bias were introduced for the change in A1C, it would be more likely toward the null. Not all patients completed the end of protocol survey or final A1C assessment. Although the final A1C was unobtainable for 15.6% of par- ticipants, this may be expected because we had no in-person contact with them. However, this did not impact the �MPR outcomes that were available administra- tively for all but five participants. Further, we used two alternate methods of impu- tation for an intention-to-treat analysis and both were consistent with results for those with complete data. A modest num- ber of participants in the intervention group had fewer than six calls. The obser- vation that only those with �6 calls over 1 year had statistically significant though modest improvements in glycemic con- trol adds to our confidence that the mech- anism of the calls, and not type 1 error, was responsible for the difference be- tween the telephone and print group out- comes. However, those accepting more calls may be more amenable to change, which could possibly confound these results.

Strengths This study explored the comparative ef- fectiveness of two interventions in a low- er-income, urban population that was racially and ethnically diverse; the major- ity were lower-income immigrants work- ing in support of health-care systems. They were homogeneous, however, in that they did not have economic barriers to securing medications or medical visits b e c a u s e o f t h e i r u n i o n / e m p l o y e r - sponsored health benefits. The sample was drawn from those with evidence of difficulty managing their diabetes; they were individuals who often, because of life circumstances, are unlikely to volun- teer for a study requiring them to visit a research center. Therefore, a strength of this study is that we may have avoided selection bias.

Evidence supports diabetes self- management education having greater success in health outcomes when it is maintained over a longer period of time (25). A telephone intervention may be a convenient and feasible intervention to support those who have difficulty access-

ing diabetes self-management education. This intervention could be more success- ful in improving A1C if embedded in ei- ther provider or payer models, especially if synergistic with other targeted quality improvement initiatives. In the context of current related literature (5–9), this study provides a successful model of an inter- vention delivering self-management sup- port at lower cost than studies using licensed health professionals or more in- tensive interventions, such as in-person or those having greater frequency of con- tact. This study extends previous research because it focused on a population with known health disparities. Health educa- tors trained and supervised by a certified diabetes educator may promote and maintain self-management skills and pro- vide crucial support needed by individu- als managing their diabetes.

Acknowledgments — This study was sup- ported by NIH grants R18 DK62038 and DK020541.

No potential conflicts of interest relevant to this article were reported.

E.A.W. and H.W.C. contributed to every aspect of this article. C.S., M.S.-K., and R.U. contributed to study design, discussion, re- search data, and editing of the manuscript. E.B. contributed to the research data, dis- cussion, and review and editing of the manuscript.

Parts of this study were presented in oral form at the 69th Scientific Sessions of the American Diabetes Association, New Orleans, Louisiana, 5–9 June 2009.

We gratefully acknowledge the data man- agement contributions of Fionnuala King of the 1199SEIU Benefit and Pension Funds; our talented health educators, including Giovanna DiFrancesca, Kathleen McCabe, Gisela Mojica, Jennifer Case, Tara DeWitt, Gabriel Ferreira, Marlene Taveras, Samara Lipsky, Dr. Hollie Jones, and Hector Cariello; data management support from Maria Kalten and Jennifer Lukin, all staff from the Albert Einstein College of Medicine; and especially, the participants in New York City who volunteered for our study.

References 1. Intensive blood-glucose control with sul-

phonylureas or insulin compared with conventional treatment and risk of com- plications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet 1998;352: 837– 853

2. ADVANCE Collaborative Group, Patel A, MacMahon S, Chalmers J, Neal B, Billot L, Woodward M, Marre M, Cooper M, Glas- ziou P, Grobbee D, Hamet P, Harrap S, Heller S, Liu L, Mancia G, Mogensen CE,

Improving diabetes outcomes study

6 DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 care.diabetesjournals.org

Pan C, Poulter N, Rodgers A, Williams B, Bompoint S, de Galan BE, Joshi R, Travert F. Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med 2008;358:2560 – 2572

3. American Diabetes Association. Stan- dards of medical care in diabetes–2010. Diabetes Care 2010;33:S11–S61

4. Cheung BM, Ong KL, Cherny SS, Sham PC, Tso AW, Lam KS. Diabetes preva- lence and therapeutic target achievement in the United States, 1999 to 2006. Am J Med 2009;122:443– 453

5. Norris SL, Engelgau MM, Narayan KM. Effectiveness of self-management training in type 2 diabetes: a systematic review of randomized controlled trials. Diabetes Care 2001;24:561–587

6. Vermeire E, Wens J, Van Royen P, Biot Y, Hearnshaw H, Lindenmeyer A. Interven- tions for improving adherence to treat- ment recommendations in people with type 2 diabetes mellitus. Cochrane Data- base Syst Rev 2005;2:CD003638

7. Clark M. Diabetes self-management edu- cation: a review of published studies. Pri- mary Care Diabetes 2008;2:113–120

8. Piette JD, Weinberger M, Kraemer FB, McPhee SJ. Impact of automated calls with nurse follow-up on diabetes treat- ment outcomes in a Department of Veter- ans Affairs Health Care System: a randomized controlled trial. Diabetes Care 2001;24:202–208

9. Wolever RQ, Dreusicke M, Fikkan J, Hawkins TV, Yeung S, Wakefield J, Duda L, Flowers P, Cook C, Skinner E. Integra-

tive health coaching for patients with type 2 diabetes: a randomized clinical trial. Di- abetes Educ 2010;36:629 – 639

10. Walker EA, Schechter CB, Caban A, Basch CE. Telephone intervention to promote diabetic retinopathy screening among the urban poor. Am J Prev Med 2008; 34:185–191

11. Cohen HW, Shmukler C, Ullman R, Rivera CM, Walker EA. Measurements of medication adherence in diabetic patients with poorly controlled HbA(1c). Diabet Med 2010;27:210 –216

12. Hill-Briggs F. Problem solving in diabetes self-management: a model of chronic ill- ness self-management behavior. Ann Be- hav Med 2003;25:182–193

13. Bodenheimer T, Handley MA. Goal-set- ting for behavior change in primary care: an exploration and status report. Patient Educ Couns 2009;76:174 –180

14. Bandura A. Self-Efficacy: The Exercise of Control. New York, W.H. Freeman and Company, 1997

15. Stokols D. Translating social ecological theory into guidelines for community health promotion. Am J Health Promot 1996;10:282–298

16. Lab in an Envelope [Internet]. Home Health- care Laboratories of America. Availabe from http://www.hhla.com/products_services_ lab_in_an_envelope.html. Accessed 5 May 2010

17. National Glycosylation Standardization Program. Available from http://www. ngsp.org. Accessed 5 May 2010

18. Jeppsson JO, Jerntorp P, Almër LO, Pers- son R, Ekberg G, Sundkvist G. Capillary

blood on filter paper for determination of HbA1c by ion exchange chromatography. Diabetes Care 1996;19:142–145

19. Gay EC, Cruickshanks KJ, Chase HP, Klingensmith G, Hamman RF. Accuracy of a filter paper method for measuring gly- cosylated hemoglobin. Diabetes Care 1992;15:108 –110

20. Pladevall M, Williams LK, Potts LA, Di- vine G, Xi H, Lafata JE. Clinical outcomes and adherence to medications measured by claims data in patients with diabetes. Diabetes Care 2004;27:2800 –2805

21. Karve S, Cleves MA, Helm M, Hudson TJ, West DS, Martin BC. An empirical basis for standardizing adherence measures de- rived from administrative claims data among diabetic patients. Medical Care 2008;46:1125–1133

22. Morisky DE, Green LW, Levine DM. Con- current and predictive validity of a self- reported measure of medication adherence. Med Care 1986;24:67–74

23. Toobert DJ, Hampson SE, Glasgow RE. The summary of diabetes self-care activi- ties measure: results from 7 studies and a revised scale. Diabetes Care 2000;23: 943–950

24. StataCorp. Multiple-Imputation Reference Manual. Release 11. College Station, Texas, Stata Press, 2009

25. Funnell MM, Brown TL, Childs BP, Haas LB, Hosey GM, Jensen B, Maryniuk M, Peyrot M, Piette JD, Reader D, Siminerio LM, Weinger K, Weiss MA. National stan- dards for diabetes self-management edu- cation. Diabetes Care 2007;30:1630 – 1637

Walker and Associates

care.diabetesjournals.org DIABETES CARE, VOLUME 34, NUMBER 1, JANUARY 2011 7