Impact of Computer-Generated Personalized Goals on HbA1c

Clinical Care/Education/Nutrition

ORIGINAL ARTICLE

Impact of Computer-Generated
Personalized Goals on HbA1c

CLARESA S. LEVETAN, MD1,2 DAVID C. ROBBINS, MD2 do not know their HbA1c levels and target
KAREN R. DAWN RN, CDE1,2 ROBERT E. RATNER, MD1,2 goal. Numerous studies underscore the
opportunities missed by physicians for
OBJECTIVE — The public is increasingly aware of the importance of HbA1c testing, yet the providing diabetes counseling aimed at
vast majority of patients with diabetes do not know their HbA1c status or goal. We set forth to optimizing glycemic control (8 –18). For
evaluate the impact of a system that provides uniquely formatted and personalized reports of example, among large managed care or-
diabetes status and goals on changes in HbA1c levels. ganizations in which 92% of patients per-
form self blood glucose monitoring, less
RESEARCH DESIGN AND METHODS — A total of 150 patients with diabetes were than one-third had heard of the “A-
randomized to receive either standard care or intervention inclusive of a computer-generated One-C” test (16,17).
11” ϫ 17” color poster depicting an individual’s HbA1c status and goals along with personalized
steps to aid in goal achievement. All patients enrolled received diabetes education during the 3 Traditional diabetes self-management
months before enrollment. HbA1c was performed at baseline and 6 months. training programs have had limited effi-
cacy on glycemic control when evaluated
RESULTS — At baseline, there were no significant differences between patient groups in 6 months after the intervention (18 –25).
terms of age, sex, education level, race, and HbA1c or lipid levels. Among patients with baseline Conventional methods of communicating
HbA1c Ն7.0%, there was an 8.6% (0.77% absolute) reduction in HbA1c among control subjects health messages to patients via brochures,
compared with a 17.0% (1.69% absolute) decline in the intervention group (P ϭ 0.032). There videos, and booklets are also of limited
were no differences between the control and intervention groups with respect to the frequency value, and there are no standardized ed-
of patients experiencing any decline in HbA1c (63 vs. 69%, P ϭ 0.87); among these patients ucational materials demonstrating effi-
experiencing a decline, the most substantial reductions were seen with the control group, which cacy in improving diabetes outcomes
had a 13.3% (1.15% absolute) decline compared with the intervention patients, who reduced (26 –28). Because physicians have less
their HbA1c by 24.2% (2.26% absolute reduction; P ϭ 0.0048). At study close, 77% of the time to see more patients, and preventive
patients had their poster displayed on their refrigerator. services are almost nonexistent in most
practices, creative solutions are required
CONCLUSIONS — This unique and personalized computer-generated intervention re- to address the realities of modern health
sulted in HbA1c lowering comparable to that of hypoglycemic agents. care. We designed a computer program
that produces unique, customized com-
Diabetes Care 25:2– 8, 2002 puter-generated tools that provide pa-
tients with their diabetes status, goals,
W ritten goals and objectives lay the their health goals. For example, people and steps to meet these goals. We set forth
foundation for achieving success who know their health goals and believe to evaluate whether personalized and
in most disciplines, including that these goals are within their control uniquely delivered laboratory results
business, science, and education. Written are more likely to have improved out- along with written goals might facilitate
contracts between health educators and comes and engage in self-care behaviors, HbA1c lowering.
patients have resulted in improved out- including exercise and weight loss pro-
comes by shifting the locus of control grams (1–15). RESEARCH DESIGN AND
from the health care provider to the pa- METHODS — Identification and en-
tient (1–3). These principles have not typ- Despite the successful efforts of nu- rollment was initiated among 150 pa-
ically been incorporated into medical merous national organizations in raising tients with diabetes completing an
school curricula, nor are physicians ex- public awareness of the role of HbA1c in American Diabetes Association (ADA)-
posed to innovative modes of communi- the development of diabetes-related com- recognized diabetes education program
cation that may aid patients in achieving plications, most patients with the disease during the 3-month period before study
have never heard of the term HbA1c and enrollment. All patients were enrolled be-
tween October 1998 and April 1999.
●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●
Exclusion criteria
From the 1Department of Internal Medicine, Division of Endocrinology, Washington Hospital Center, Patients giving a history of renal insuffi-
Washington, DC; and the 2MedStar Research Institute, Washington, DC. ciency with a creatinine level Ͼ1.5 mg/dl,
women who were pregnant at the time or
Address correspondence and reprint requests to Claresa Levetan, MD, MedStar Clinical Research Center, planning a pregnancy during the study
650 Pennsylvania Ave., SE, Suite 50, Washington, DC 20003-4393. E-mail: [email protected]. period, and patients using insulin pumps
were excluded. Patients who could not
Received for publication 29 May 2001 and accepted in revised form 4 October 2001. read were excluded from the study. To
R.E.R holds stock in Roche Diagnosics. avoid spurious HbA1c results, patients
Abbreviations: ADA, American Diabetes Association; CDE, certified diabetes educator.
A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion
factors for many substances.

2 DIABETES CARE, VOLUME 25, NUMBER 1, JANUARY 2002

Levetan and Associates

Table 1—Tools received by intervention group versus control subjects line HbA1c, lipid, and blood pressure
status, with room to document subse-
Intervention Standard care quent values. For the duration of the
study, each patient was sent one postcard
Patient Physician Patient Physician (Fig. 3) per month that emphasized the
relation between HbA1c and diabetes-
Poster ߛ ߛ related complications and provided an ac-
Wallet card ߛ tion step for lowering HbA1c (Table 1).
Monthly postcards ߛ
Color chart report After receipt of the poster and person-
Traditional lab report ߛ alized wallet card, intervention patients
ߛ received one phone call from a health ed-
ucator to discuss their personalized
who received a blood transfusion within vided the names of all physicians to poster. This structured phone call lasted
the past 30 days and those with an whom they would want their laboratory no more than 10 min and strictly focused
underlying illness, such as malignancy or results sent, and standard laboratory re- discussion on the patient’s report. There
a condition that was expected to impact ports were sent to all of these physicians. were no other educational, nutritional, or
their survival over the next 6 months, exercise interventions provided. The
were also excluded. Standard care (control) health educator told patients to discuss
Patients in the control group received issues and questions regarding medica-
Protocol usual diabetes healthcare advice provided tion and dosing with their primary physi-
All patients were told they were entering by their physician during the study pe- cian.
an educational study designed to evaluate riod. Other than the initial interview by a
the impact of diabetes educational tools CDE, no additional diabetes educational In addition to a traditional laboratory
on outcomes. Patients who met enroll- materials were provided (Table 1). report form, physicians whose patients
ment criteria and agreed to participate were randomized to the intervention re-
were asked to read and sign an informed Intervention ceived a unique color report (Fig. 1) that
consent document. A certified diabetes Each patient in the intervention group re- was similar to the poster that their patient
educator (CDE) obtained each patient’s ceived a computer-generated customized received but was designed for the pa-
baseline demographic data, weight, and report presented as an 11” ϫ 17” lami- tient’s medical record. In addition to a
blood pressure level, as well as a list of nated color poster backed with magnets, graphic display of their HbA1c, both the
current medications. Additionally, labo- with a bulleted list of personalized goals patient’s and physician’s personalized
ratory tests including HbA1c, direct LDL and recommended steps for achieving the report included information on the pa-
cholesterol, HDL cholesterol, and spot goals (Fig. 1). The individual’s report was tient’s blood pressure, lipid, and micro-
urine for microalbumin were performed generated from a Microsoft Access– based albumin status; the date for the patient’s
on all patients. Baseline HbA1c level was decision support system that collected pa- next dilated retinal examination; and bul-
not a study exclusion. tient information from the enrollment leted suggestions on management of their
questionnaire and matched it against a patient based on the following: ADA’s
Each participant completed a brief knowledge base of established diabetes, 1998 Clinical Practice Recommenda-
patient questionnaire. It included the pa- cardiovascular, nutrition, and exercise tions, the Kaiser Permanente of Mid-
tient’s report of diet, exercise, smoking guidelines. Atlantic Region Clinical Guidelines to the
habits, frequency of foot examinations, Management of Diabetes, the National
and the date of the most recent dilated eye There was no subjective interpreta- Cholesterol Education Program, Healthy
examination. The questionnaire provided tion from the personal interview. The People 2000, the National Institutes of
multiple options, and participants were questionnaire asked participants to pro- Health Consensus Statement on Physical
asked to select the one best choice. In- vide the names of family members, pets, Activity and Cardiovascular Health, and
cluded was a question of whether or not or friends who exercise, cook, or share the Departments of Agriculture and
patients had heard of the term “hemoglo- time with the participant, and these Health and Human Services’ Dietary
bin A1c” or the “A-One-C” blood test. All names were included in the personalized Guidelines for Americans (29 –34). The
patients who were not aware of the test reports. The posters included discussion tailored suggestions were tested among
were told about the test and its impor- points for patients to mention to their patients and physicians before the study
tance in evaluating diabetes control. physicians. For example, for a patient for their appropriateness.
who was treated with a submaximal dose
After the initial interview, we ran- of a glucose-lowering medication but had Follow-up
domly assigned patients to standard care not achieved their HbA1c goal, their ac- At 6 months after enrollment, all patients
or the experimental intervention. All tion plan would include a recommenda- received a follow-up letter and question-
baseline patient data and questionnaire tion to talk to their physician about naire. Follow-up appointments occurred
responses were entered into a relational optimizing their glucose-lowering medi- at the MedStar Clinical Research Center.
database and algorithm. Of the 150 pa- cation. The report included both the ge- At the close of the study, all patients and
tients enrolled in the study, 75 were ran- neric and trade names of medications. their physicians received a letter with
domly assigned to the control group and Patients also received a personalized wal- their baseline and follow-up results.
were assigned 75 to the intervention arm let card (Fig. 2) that included their base-
using block randomization. Patients pro-

DIABETES CARE, VOLUME 25, NUMBER 1, JANUARY 2002 3

Computer-generated personalized goals impact HbA1c

Figure 1—Personalized 11” ϫ 17” poster, laminated and backed with magnets for patients to place on their refrigerator.
4 DIABETES CARE, VOLUME 25, NUMBER 1, JANUARY 2002

Levetan and Associates

dianapolis, IN). The laboratory used
Abel-Kendall analyzed serum, purchased
from Northwest Lipid Research Labora-
tory (Seattle, WA), as the calibrator. Con-
trols included those supplied by Roche
and Northwest Lipid Research Center
(NWLRC). The laboratory participates in
the NWLRC lipid quality assurance pro-
gram (Cholesterol Standardization Certi-
fication). LDL cholesterol was measured
directly in fresh plasma using reagents
from Sigma Diagnostics (St. Louis, MO).
Urine microalbumin testing was per-
formed using Micral urine test strips pro-
vided by Roche.

Figure 2—Personalized wallet card identifying baseline status, with room for documenting fol- Statistical methods
low-up HbA1c tests and blood pressure status.
Power calculations. We anticipated a
Laboratory measurements 12-h fast. Direct LDL and HDL choles- 10% change in HbA1c levels (ϳ0.8% ab-
terol levels were measured enzymatically solute reduction) during the 6-month
The Penn Medical Laboratory (Washing- on a Hitachi 717 autoanalyzer using re- study period among intervention pa-
agents supplied by Roche Diagnostics (In- tients, assuming equal variation between
ton, DC) performed all HbA1c and choles- groups (estimated SD 1.9) (35). A sample
terol profiles in serum collected after a size of 63 per group was needed to reach
statistical significance, using a Student’s t
test. We assumed a dropout rate of 10%
from this urban, minority population
during the 6-month study period, yield-

Figure 3—One of five monthly postcards sent to patients in the intervention group. The postcards emphasized the importance of achieving HbA1c
goals and urged patients to receive follow-up cholesterol testing.

DIABETES CARE, VOLUME 25, NUMBER 1, JANUARY 2002 5

Computer-generated personalized goals impact HbA1c

Table 2—Patient characteristics at baseline between the intervention and control

groups with respect to baseline age, sex,

Variable Control Intervention P* education level, race, baseline choles-
terol levels, and comorbidities (Table 2).

n 64 64 Patients were similar with respect to base-

Male sex (%) 30 35 0.71 line HbA1c and LDL and HDL cholesterol.
African-American (%) 83 89 0.32 More than half of the patients in each

Age (years) 60 57 0.25 group reported a history of hypertension,

Height (in)† 66 66 0.97 and 75% (49 control and 45 intervention

Weight (lb)† 197 197 0.62 patients) had a baseline HbA1c Ն7.0%.
Systolic BP (mmHg)† 143 142 0.85 At the 6-month follow-up, there were

Diastolic BP (mmHg)† 83 83 0.80 no significant differences in outcomes

Education (% no college) 44 47 0.62 within or between groups with respect to

Duration of diabetes (years)‡ 3 5.5 0.10 weight, systolic or diastolic blood pres-

HDL cholesterol (mg/dl)† 41 42 0.57 sure, or lipids (Table 3). There was also no

Microalbuminuria (% with 30–300 ␮g) 61 58 0.72 difference between the control and inter-

LDL cholesterol (mg/dl)† 116 115 0.98 vention groups with respect to the per-

HbA1c (%) 8.39 Ϯ 2.03 8.85 Ϯ 2.48 0.25 centage of patients in each group who

Home glucose monitoring (%) 89 87 0.88 experienced a decline in HbA1c (63 vs.
Heard of “A-One-C” test (%) 52 42 0.47 69%; P ϭ 0.87).

Data are means and means Ϯ SD, unless otherwise indicated. *P values Ͻ0.05 were considered significant; At the study close, the intervention

†median with upper and lower bound for median; ‡Fisher’s Exact (2-tailed); all others report 95% CI about patients had a significant reduction in
the mean. BP, blood pressure.
HbA1c compared with control subjects
(Table 4). Among patients with a baseline

ing a total recruitment size of 146 to com- was one death in the control group that HbA1c Ն7.0%, there was an 8.6% (0.77%
plete the study (37). These estimates were was attributed to cardiovascular disease. absolute) reduction in HbA1c among con-
conservative because they assumed an Three patients in the control group devel- trol subjects and a 17.0% (1.69% abso-

analysis would be univariate. The analysis oped chronic debilitating syndromes lute) decline in the intervention group
used a t2 test (a multivariate t test). The (e.g., cancer) and were dropped from the (P ϭ 0.032).

experimental-wise error rate was set at study, two lab specimens were lost, and In both the control and intervention

0.05 (the test-based ␣ was 0.025), and the five patients declined follow-up after ini- groups, the most sizable and significant

type II error rate was set at 0.2. tial enrollment. reductions in HbA1c were noted among

Analysis of results. The major outcome Among the intervention group, there the subgroup of patients who lowered

variable was HbA1c. Differences between was one death, two lost laboratory speci- their HbA1c during the study period and
pre- and posttreatment intervention peri- mens, and seven patients who declined were classified as responders. Responders

ods and the control group at baseline and follow-up. One patient developed a in the control group experienced a de-

follow-up were assessed using parametric chronic debilitating illness requiring cline of 13.3% (1.15% absolute) as com-

(Student’s t tests), nonparametric (medi- chronic corticosteroid therapy and was pared with intervention responders, who

an tests), and contingency table analyses dropped from the study. We report the had a decline of 24.2% (a 2.26% absolute

(Fisher’s exact tests) to detect the differ- data on the 128 remaining patients in the reduction; P ϭ 0.0048) during the study

ence in demographics and laboratory as- intervention and control group, all of period. Altogether, 61% of responders

signment variables between the patient whom completed the final questionnaire who were not at the goal at baseline in the

study groups. The changes observed and returned for follow-up HbA1c and intervention group and 38% of control
patients achieved an HbA1c of Յ7% (P ϭ
within each cohort were evaluated for sig- cholesterol testing.

nificant differences between the pre- and There were no significant differences 0.05) by study end.

postintervention periods using a two-

tailed paired t test. The significance of the

difference between the treatment and Table 3—Changes from baseline at follow-up

control groups was evaluated by repeated

measures of analysis of variance that Control ⌬ from Intervention ⌬
tested for changes between the two Variable baseline from baseline
groups from the pre- to postintervention

periods while controlling for the different n 64 64

baseline values on the outcomes of inter- Weight (lb) ϩ1.0 ϩ1.54

est. A P value Ͻ0.05 was considered sig- Systolic BP (mmHg) Ϫ4 Ϫ4

nificant. Diastolic BP (mmHg) Ϫ5 Ϫ4

HDL cholesterol (mg/dl) ϩ3 ϩ3

RESULTS — Of the randomized pa- LDL cholesterol (mg/dl) Ϫ7 Ϫ5

tients, 85.3% completed the study and Heard of “A-One-C” test (%) Ϫ5.1 ϩ14.3

were evaluated in the final analysis. There BP, blood pressure.

6 DIABETES CARE, VOLUME 25, NUMBER 1, JANUARY 2002

Levetan and Associates

Table 4—Changes in HbA1c from baseline to make these tools available to physician
practices and managed care populations;
HbA1c Control Intervention this system may serve as an adjunct to tra-
ditional diabetes self-management training,
n 64 64 with potential HbA1c lowering comparable
Baseline (%) 8.39 Ϯ 2.03 8.85 Ϯ 2.48 to that of oral agents.
End of study (%) 7.79 Ϯ 1.91 7.78 Ϯ 2.22
Change from baseline (%) Ϫ0.6 (P Ͼ 0.05) Ϫ1.08 (P ϭ 0.013)* Although we demonstrated that per-
sonalized empowerment tools could po-
Data are means Ϯ SD. *P value within group comparison. tentially have a significant impact on
short-term HbA1c outcomes, further
There were no significant changes in We are particularly impressed by the abil- study is necessary to determine the long-
weight during the study period seen in ity of the intervention to lower HbA1c in a term implications of personalized em-
either the control or intervention patients predominately minority population, powerment tools, such as the ones we
(Table 3). The intervention was equally demonstrating the cultural sensitivity of designed on diabetes-related outcomes.
effective among patients who had only the intervention. The visual nature of the As we develop a strategic health plan for
grade school or high school education intervention may also have contributed to the 21st century, the critical research that
compared with those with a college edu- the success of the intervention in a popu- identifies the genetic, physiological, and
cation. At the study close, 77% of all pa- lation with less than a college education. environmental determinants of disease
tients in the intervention group reported must also be accompanied by clinical re-
that their poster remained displayed on The Industrial Revolution taught that search that evaluates how scientific ad-
their refrigerator. Patients in the interven- performance and production among as- vances can best be translated into
tion group reported greater changes in sembly line workers were enhanced when practical steps that patients can use to im-
their diabetes medications; they were also individuals knew their goals and were prove their health.
more likely to talk to their physicians given feedback on their own production
about checking their HbA1c level and rate (37). These same theories of enhanc- Acknowledgments — Funding for this study
were knowledgeable of the HbA1c test, ing task performance by involving pa- was provided by an unrestricted educational
but these self-reported outcomes trended tients directly with their diabetes goals grant from Roche Diagnostics (Indianapolis,
toward, but did not achieve, statistical sig- were used in this study. Unlike providing IN) and by MedStar Research Institute.
nificance. generalized knowledge on the subject of
diabetes, we allowed patients and their The authors thank Jarita Odei and Laura
CONCLUSIONS — Putting preven- families to have a benchmark of their per- Want for their assistance with the study and
tion into practice is one of the three major sonal diabetes status and their goal. Ellen Shair for her thoughtful review and ed-
goals of Healthy People 2000, the national iting of the manuscript. We thank Dr. Barbara
health promotion and disease prevention The poster was not thrown away or Howard and the Hochberger and Mihm fami-
objectives of the U.S. Department of filed away in a drawer, and most of the lies for their love, support, and understanding
Health and Human Services (30). Based patients kept the poster up on their refrig- throughout the study. The authors appreciate
on the simplistic hypothesis that knowing erator for the duration of the study. All the support of Roche Diagnostics for this clin-
ones own glycemic status and goals could patients were recruited from those who ical study.
potentially improve performance, the received diabetes education during a
study set forth to put prevention into 3-month period before recruitment, and References
practice among patients with diabetes by thus our patient population was suffi- 1. Schlenk EA, Boehm S: Behaviors in type II
developing and testing the impact of com- ciently motivated to voluntarily attend an
puter-generated personalized empower- ADA-recognized self-management pro- diabetes during contingency contracting.
ment tools that were designed to lower gram, which may explain why the major- Appl Nurs Res 11:77– 83, 1998
HbA1c concentrations and that required ity of patients in both groups experienced 2. Becker MH, Maiman LA: Strategies for en-
no effort on the part of the health care a reduction in HbA1c. hancing patient compliance. J Community
provider. Health 6:113–135, 1980
It was the objective of the researchers 3. Wallston BS, Wallston KA: Locus of con-
Similar to many other studies, we to help patients feel that good health is trol and health: a review of the literature.
found that diabetes educational interven- within their ability to achieve, i.e., to Health Educ Monogr 6:107–117, 1978
tions do not result in a decline in HbA1c strengthen their internal locus of control 4. Wallhagen MI, Strawbridge WJ, Kaplan
among all patients (19). After an ADA- (38,39). We embraced the concepts of Dr. GA: Impact of internal health locus of
recognized education course, Ͼ60% of Elliott P. Joslin that “the person with dia- control on health outcomes for older men
the patients experienced a reduction of betes who knows the most about their and women: a longitudinal perspective.
their HbA1c level during the study period, disease, lives the longest” (40). The inves- Gerontologist 34:299 –306, 1994
yet even greater improvement was seen tigators believe that the greatest impact 5. Oberle K: A decade of research in locus of
among patients in the intervention group. resulted from a patient seeing his or her control. What have we learned? J Adv Nurs
We attribute these findings to the con- own diabetes status and goal. 16:800 – 806, 1991
stant visual reminder to patients and their 6. Cohen S, Herbert TB: Health psychology:
families of their diabetes status and goals. The software algorithm uses laboratory psychological factors and physical disease
data and patient data derived from a short from the perspective of human psycho-
questionnaire that can be self-administered. neuroimmunology. Annu Rev Psychol 47:
The investigators have developed a system 113–142, 1996

DIABETES CARE, VOLUME 25, NUMBER 1, JANUARY 2002 7

Computer-generated personalized goals impact HbA1c

7. Adler N, Matthews K: Health psychology. 19. Weiner JP, Parente ST, Garnick DW, 50212)
Why do some people get sick and some Fowles J, Lawthers AG, Palmer RH: Vari- 30. American Diabetes Association: Clinical
stay well? Annu Rev Psychol 45:229 –259, ation in office-based quality: a claims-
1994 based profile of care provided to Medicare Practice Recommendations 1999. Diabe-
patients with diabetes. JAMA 273: 1503– tes Care 22 (Suppl. 1):S1–S114, 1999
8. Sallis JF, Pinski RB, Grossman RM, Patter- 1508, 1995 31. Kaiser Permanente: Kaiser Permanente
son TL, Nader PR: The development of Treatment Guidelines for Diabetes. Estab-
self-efficacy scales for health-related diet 20. Streja DA, Rabkin SW: Factors associated lished for Kaiser Permanente physicians
and exercise behaviors. Health Educ Res with implementation of preventive care and health care providers in the Mid-At-
3:283–292, 1988 measures in patients with diabetes melli- lantic States Region. Effective date: March
tus. Arch Intern Med 159:294 –302, 1999 1998.
9. Marcus BH, Owen N: Motivational readi- 32. Expert Panel on Detection, Evaluation,
ness, self-efficacy and decision-making 21. Helseth LD, Susman JL, Crabtree BF, and Treatment of High Blood Cholesterol
for exercise. J Appl Social Psychology 22:3– O’Connor PJ: Primary care physicians’ in Adults: Summary of the second report
16, 1992 perceptions of diabetes management: a of the National Cholesterol Education
balancing act. J Fam Pract 48:37– 42, 1999 Program (NCEP) expert panel on detec-
10. Kaplan RM, Atkins CJ, Reinsch S: Specific tion, evaluation, and treatment of high
efficacy expectations mediate exercise 22. Payne TH, Gabella BA, Michael SL, Young blood cholesterol on adults (Adult Treat-
compliance in patients with COPD. WF, Pickard J, Hofeldt FD, Fan F, Strom- ment Panel II). JAMA 269:3015–3023,
Health Psychol 3:223–242, 1984 berg JS, Hamman RF: Preventive care in 1993
diabetes mellitus: current practice in ur- 33. National Institutes of Health Consensus
11. Holman HR, Lorig K: Perceived self-effi- ban health-care system. Diabetes Care 12: Statements, Consensus Development
cacy in self-management of chronic dis- 745–747, 1989 Program: Physical Activity and Cardiovas-
ease. In Self-Efficacy: Thought Control of cular Health. Vol. 13, no. 3. Bethesda, MD,
Action. R. Schwarzer, Ed. Washington, 23. Tuttleman M, Lipsett L, Harris MI: Atti- National Institutes of Health, 1995
DC, Hemisphere, 1984, pp. 305–323 tudes and behaviors of primary care phy- 34. U.S. Departments of Agriculture and
sicians regarding tight control of blood Health and Human Services: Nutrition and
12. Godin G, Valois P, Lepage L: The pattern glucose in IDDM patients. Diabetes Care Your Health: Dietary Guidelines for Ameri-
of influence of perceived behavioral con- 15:765–772, 1993 cans. 4th ed. Washington, DC, U.S. De-
trol upon exercising behavior: an applica- partments of Agriculture and Health and
tion of Ajzen’s Theory of Planned 24. Larme AC, Pugh JA: Attitudes of primary Human Services, 1995
Behavior. J Behav Med 16:81–102, 1993 care providers toward diabetes: barriers to 35. Larsen ML, Horder M, Mogensen EF:
guideline implementation. Diabetes Care Effect of long-term monitoring of glyco-
13. Garcia AW, King AC: Predicting long- 21:1391–1396, 1998 sylated hemoglobin levels in insulin de-
term adherence to aerobic exercise: a pendent diabetes. NEJM 323:1021–1025,
comparison of two models. J Sport Exerc 25. Ornstein SM, Garr DR, Jenkins RG, Rust 1990
Psychol 13:394 – 410, 1991 PF, Zemp L, Arnon A: Compliance with 36. Kaul L, Nidiry JJ: Management of obesity
five health promotion recommendations in low-income African-Americans. J Natl
14. Clark MM, Abrams DB, Niaura RS, Eaton in a university-based family practice. J Med Assoc 91:139 –143, 1999
CA, Rossi JS: Self-efficacy in weight man- Fam Pract 29:163–168, 1989 37. Richmond BM, Farmer RN: In Manage-
agement. J Consult Clin Psychol 59:739 – ment and Organizations. Random House,
744, 1991 26. Mead VP, Rhyne RL, Wiese WH, Lambert New York, 1975, p. 22–25
L, Skipper B: Impact of environmental pa- 38. McCann S, Weinman J: Empowering the
15. Bernier M, Avard J: Self-efficacy, outcome tient education on preventive medicine patient in the consultation: a pilot study.
and attrition in a weight reduction pro- practices. J Fam Pract 40:363–369, 1995 Patient Educ Couns 27:227–234, 1996
gram. Cognitive Ther Res 10:319 –338, 39. Bennett P, Norman P, Moore L, Murphy
1986 27. Joos SK, Hickam DH, Gordon GH, Baker S, Tudor-Smith C: Health locus of control
LH: Effects of a physician communication and value for health in smokers and non-
16. Diabetes-specific preventive-care prac- intervention on patient care outcomes. smokers. Health Psychol 16:179 –182, 1997
tices among adults in a managed-care J Gen Intern Med 11:147–155, 1996 40. Education: A Treatment for Diabetes. In
population—Colorado, Behavioral Risk Joslin’s Diabetes Mellitus. 12th ed. Krall LP,
Factor Surveillance System, 1995. MMWR 28. Dollahite J, Thompson C, McNew R: Ed. Philadelphia, Lea & Febiger, 1985,
Morb Mortal Wkly Rep 46:1018 –1023, Readability of printed sources of diet and p. 465
1997 health information. Patient Edu Couns 27:
123–134, 1996
17. Trends in the prevalence and incidence of
self-reported diabetes mellitus—United 29. U.S. Department of Health and Human
States, 1980 –1994. MMWR Morb Mortal Services, Public Health Service: Healthy
Wkly Rep 46:1023–1027, 1997 People 2000: National Health Promotion
and Disease Prevention Objectives. Wash-
18. Clement S: Diabetes self-management ington, DC, U.S. Dept of Health and Hu-
education (Review). Diabetes Care 18: man Services, 1991 (publ. no. PHS-91–
1204 –1214, 1995

8 DIABETES CARE, VOLUME 25, NUMBER 1, JANUARY 2002