Вы находитесь на странице: 1из 7

Epidemiology/Health Services/Psychosocial Research


Is Self-Efficacy Associated With Diabetes

Self-Management Across Race/Ethnicity
and Health Literacy?
URMIMALA SARKAR, MD, MPH1 to help avoid diabetes-related morbidity
LAWRENCE FISHER, PHD2 and mortality. Self-management is a cor-
DEAN SCHILLINGER, MD3 nerstone of diabetes care, and it is be-
lieved that improving patient self-efficacy
is a critical pathway to improved self-
OBJECTIVE — Although prior research demonstrated that improving diabetes self-efficacy management.
can improve self-management behavior, little is known about the applicability of this research The concept of self-efficacy is based
across race/ethnicity and health literacy levels. We examined the relationship between diabetes on social cognitive theory, which de-
self-efficacy and self-management behavior in an urban, diverse, low-income population with a scribes the interaction between behav-
high prevalence of limited health literacy. ioral, personal, and environmental factors
RESEARCH DESIGN AND METHODS — We administered an oral questionnaire in
in health and chronic disease. The theory
Spanish and English to patients with type 2 diabetes at two primary care clinics at a public of self-efficacy proposes that patients’
hospital. We measured self-efficacy, health literacy, and self-management behaviors using es- confidence in their ability to perform
tablished instruments. We performed multivariate regressions to explore the associations be- health behaviors influences which behav-
tween self-efficacy and self-management, adjusting for clinical and demographic factors. We iors they will engage in (4 – 6). Because
tested for interactions between self-efficacy, race/ethnicity, and health literacy on self- diabetes self-management incorporates
management. behavioral, personal, and environmental
factors into daily performance of recom-
RESULTS — The study participants were ethnically diverse (18% Asian/Pacific Islander, 25% mended activities, the concept of self-
African American, 42% Latino/a, and 15% white), and 52% had limited health literacy (short
efficacy is relevant for improving self-
version of the Test of Functional Health Literacy in Adults score ⬍23). Diabetes self-efficacy was
associated with four of the five self-management domains (P ⬍ 0.01). After adjustment, with management. Among highly selected
each 10% increase in self-efficacy score, patients were more likely to report optimal diet (0.14 patients, self-efficacy has been shown to
day more per week), exercise (0.09 day more per week), self-monitoring of blood glucose (odds be important for appropriate self-
ratio 1.16), and foot care (1.22), but not medication adherence (1.10, P ⫽ 0.40). The associa- management for many chronic health
tions between self-efficacy and self-management were consistent across race/ethnicity and health conditions (7–10), and, in diabetes, the
literacy levels. research demonstrates mixed results for
interventions that attempt to improve
CONCLUSIONS — Self-efficacy was associated with self-management behaviors in this vul- self-management behavior through im-
nerable population, across both race/ethnicity and health literacy levels. However, the magni- proved self-efficacy (11–17).
tude of the associations suggests that, among diverse populations, further study of the
determinants of and barriers to self-management is warranted. Policy efforts should be focused
Although a few recent studies have
on expanding the reach of self-management interventions to include ethnically diverse popula- addressed selected racial/ethnic minority
tions across the spectrum of health literacy. populations (18,19), little is known about
the applicability of self-efficacy research
Diabetes Care 29:823– 829, 2006 to ethnically diverse and low-income pa-
tients with diabetes. In these populations,
access barriers (20), costs of treatment

ype 2 diabetes is one of the most and there is an urgent need to improve
common diseases in the U.S., affect- quality of care and lower rates of avoid- (21), and cultural beliefs (22) may be key
ing ⬎16 million individuals (1). Di- able complications for these populations determinants of self-management behav-
abetes disproportionately affects low- (3). Patients with diabetes are expected to ior. To the extent that these factors con-
income and racial/ethnic minorities (2), perform daily self-management activities tribute to high rates of failed attempts
and/or lack of modeling of successful be-
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●
haviors, they may also contribute to lower
From the 1Division of General Internal Medicine, Department of Medicine, University of California, San self-efficacy.
Francisco, San Francisco, California; the 2Department of Family and Community Medicine, University of
California, San Francisco, San Francisco, California; and the 3Primary Care Research Center, University of Within this patient population, indi-
California, San Francisco, San Francisco General Hospital, San Francisco, California. viduals with limited health literacy may
Address correspondence and reprint requests to Urmimala Sarkar, Division of General Internal Medicine, be especially vulnerable to these experi-
Box 1211, University of California, San Francisco, San Francisco, CA. 94143. E-mail: usarkar@ ences. A growing body of research dem-
Received for publication 29 August 2005 and accepted in revised form 23 December 2005.
onstrates that limited health literacy, a
Abbreviations: SMBG, self-monitoring of blood glucose; s-TOFHLA, short version of the Test of Func- prevalent problem in vulnerable popula-
tional Health Literacy in Adults. tions, is independently associated with
A table elsewhere in this issue shows conventional and Système International (SI) units and conversion poor self-rated health (23,24), higher uti-
factors for many substances. lization of services (25–28), fewer pre-
© 2006 by the American Diabetes Association.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby ventive services (29,30), and worse
marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. glycemic control and more diabetes com-


Self-efficacy among diverse patients

plications (31). Therefore, self-efficacy speaking patients who attended a clinic We asked each participant to report
may be a relevant determinant of self- appointment over a 6-month period of his or her race/ethnicity from the follow-
management behaviors among popula- time, between June and December 2000. ing choices: Asian or Pacific Islander,
tions with limited health literacy (32–35). After informed consent was obtained black or African American, Latino/a or
We sought to determine whether di- from each participant, an oral question- Hispanic, white/Anglo, Native American,
abetes self-efficacy was associated with naire was administered in English or multiethnic, or other. To measure health
recommended self-management behav- Spanish. Each part of the instrument was literacy, we used the abbreviated form of
iors in an urban, diverse population with translated into Spanish and back- the short version of the Test of Functional
a high prevalence of limited health liter- translated into English until concordance Health Literacy in Adults (s-TOFHLA),
acy. Further, we examined whether a re- in meaning was attained. The Human Spanish or English version (41). The ab-
lationship between self-efficacy and self- Subjects Committee of University of Cal- breviated s-TOFHLA is a 36-item timed
management varied by health literacy ifornia, San Francisco, approved the reading comprehension test that uses the
score or race/ethnicity. Results from this protocol. modified Cloze procedure; every fifth to
study could inform future interventions seventh word in a passage is omitted and
to improve diabetes outcomes among eth- Measures four multiple choice options are pro-
nically diverse patients and those with We adapted a previously published, vali- vided. The abbreviated s-TOFHLA con-
limited health literacy (3). dated diabetes self-efficacy scale (38) that tains two health care passages, the first
used eight items with 4-point Likert-type selected from instructions for preparation
RESEARCH DESIGN AND responses from “1 ⫽ not at all sure” to for an upper gastrointestinal tract radio-
METHODS — The methods for this “4 ⫽ very sure.” For each item patients graph series (Gunning-Fog Index read-
study have been described in more detail rated their confidence in their ability to ability grade 4.3) and the second from the
in previous publications (31,36). This perform a recommended self-care rou- patient’s “Rights and Responsibilities”
study of adults with type 2 diabetes was tine. These items addressed diabetes- section of a Medicaid application (Gun-
based in two primary care clinics at San specific domains such as confidence in ning-Fog Index readability grade 10.4).
Francisco General Hospital, staffed by self-monitoring of blood glucose (SMBG), The abbreviated s-TOFHLA is scored on a
University of California, San Francisco, as well as general health domains such as 0 –36 scale. Using established conven-
attending faculty and residents. At the confidence in ability to get medical atten- tion, we categorized patients as having in-
time of the study, the clinics did not have tion and take care of health. We summed adequate health literacy if the s-TOFHLA
a disease-management system in place, the responses to obtain an overall self- score was between 0 and 16, marginal
but there were diabetes educators avail- efficacy score, and for ease of interpreta- health literacy if it was between 17 and
able for individual patient consultations. tion, we transformed the score to a 100- 22, and adequate health literacy if it was
We identified potential research sub- point scale with a higher score between 23 and 36 (42).
jects using the hospital system clinical representing greater self-efficacy.
and administrative database. The data- To measure diabetes self-manage- Analysis
base contains laboratory, radiology, bill- ment, we used The Summary of Diabetes To assess the self-efficacy scale, we mea-
ing, use, and demographic information Self-Care Activities Questionnaire sured internal consistency–reliability by
for patients who used the San Francisco (39,40) that assesses the frequency with calculating the Cronbach ␣ (43) for the
city and county public health system in which a patient followed a diabetes rou- overall sample and for the four most fre-
the 3 years preceding the study. Patients tine over the prior 7 days in five domains: quent racial/ethnic groups. We omitted
were eligible if they were ⬎30 years, diet, exercise, SMBG, foot care, and med- Native American (n ⫽ 2), multiethnic
spoke English or Spanish, and had type 2 ication adherence. For diet and exercise, (n ⫽ 6), and other (n ⫽ 11) ethnicity cat-
diabetes, controlled or uncontrolled, with patients reported the number of days in egories from the stratified ␣ calculations
or without complications (all ICD-9 the past week that they followed the rec- because of the small number of respon-
codes of 250.X0 or 250.X2). Patients had ommended diet or exercised at least 20 dents. We also calculated the Cronbach ␣
to have at least two visits to the same phy- min, respectively. For foot care, we asked among those with adequate versus less-
sician in one of the participating clinics, participants how often they checked their than-adequate health literacy. In calculat-
the first visit within 12 months, and an- feet for cuts and sores and dichotomized ing the ␣ by health literacy, we grouped
other within 6 months, of the interview the answers into “daily” and “less than marginal health literacy participants in
date. We excluded patients with any doc- daily” based on American Diabetes Asso- the less-than-adequate literacy group be-
umented diagnosis of end-stage renal dis- ciation guidelines. Similarly, for SMBG, cause of the small number of marginal lit-
ease, psychotic disorder, dementia, or we asked patients how often they checked eracy participants, as prior investigators
blindness, because these conditions could their blood glucose level and dichoto- have done (44).
interfere with interview completion and mized the answers into “at least daily” and Because the self-management do-
accurate health literacy assessment (37). “less than daily” SMBG. For medication mains tend not to be correlated with each
To ensure that patients identified from the adherence, we asked patients how many other within individuals (40,45), we ana-
database reflected the inclusion and ex- of their diabetes pills they missed in the lyzed the relationship between self-
clusion criteria, we provided primary care last 7 days. Because the majority of pa- efficacy and each self-management
physicians (n ⫽ 89) with a list of their tients reported optimal medication ad- domain separately. First, we created uni-
eligible patients and asked them to indi- herence, we had few responses across the variate models for the association of self-
cate additional patients to be excluded. range of adherence. Therefore, we dichot- efficacy and each self-management
Bilingual trained interviewers en- omized responses into “perfect adher- outcome. Diet and exercise were contin-
rolled all eligible English- or Spanish- ence” or “less-than-perfect adherence.” uous variables; therefore, we performed


Sarkar, Fisher, and Schillinger

an unadjusted linear regression of a 10- literacy on the five self-management Table 1—Patient characteristics
point increase in self-efficacy score on the outcomes.
frequency of following a diabetic diet or Total
frequency of exercise in the prior week. RESULTS — Eight hundred fifty-eight
We used a univariate logistic regression to patients were identified by the San Fran- n 408
calculate an unadjusted odds ratio for the cisco General Hospital clinical database as Age (years) 58.1 ⫾ 11.4
association of self-efficacy score and per- potentially eligible for the study. Of these, Ethnicity
formance of recommended SMBG, foot 142 were subsequently ineligible because Asian/Pacific Islander 75 (18)
care, and medication adherence. their primary care physician informed us African American 100 (25)
Next, to address potential confound- that the patient was not in his or her panel Hispanic 165 (40)
ers, we adjusted the self-efficacy–self- (n ⫽ 10); did not have type 2 diabetes White/non-Hispanic 51 (12)
management models for other correlates (n ⫽ 25); did not speak English or Span- Native American 2 (0.5)
of self-management. To control for dis- ish fluently (n ⫽ 28); had moved out of Multiethnic 6 (1.5)
ease-related variability in self-manage- the area (n ⫽ 35); had a psychiatric con- Other 11 (3)
ment, we tested clinical characteristics dition, e.g., dementia, psychosis, or men- Household annual income 379 (93)
such as duration of diabetes, medication tal retardation (n ⫽ 23); had died (n ⫽ 1); ⬍$20,000
regimen, and presence of complications or was identified as ineligible by the phy- Years with diabetes 9.5 ⫾ 8.0
in multivariate models. To further refine sician (n ⫽ 20). Of the 716 remaining Treatment regimen
our model, we tested demographic char- eligible patients, 261 did not make a pri- Diet alone 23 (6)
acteristics such as sex and income as po- mary care visit during the enrollment pe- Oral hypoglycemic alone 223 (54)
tential covariates in a similar fashion. riod. All remaining 455 patients were Insulin alone 49 (12)
When factors were significantly associ- approached at the time of a clinic appoint- Insulin and oral hypoglycemic 113 (28)
ated with more than one self-manage- ment. Of these, 36 refused to participate Received diabetes education 318 (78)
ment domain, we included them in the or were excluded because they were too ill Health literacy
model. For consistency, we included the to participate (n ⫽ 9) or were acutely in- Inadequate (s-TOFHLA score 156 (38.25)
same covariates in the multivariate analy- toxicated (n ⫽ 2), and 6 were excluded 0–16)
sis of each domain of self-management. because they had poor visual acuity Marginal (s-TOFHLA score 54 (13.25)
Third, after obtaining a disease- (ⱖ20/50). Thus, 413 patients completed 17–22)
adjusted model for the self-efficacy–self- the questionnaire. For 408 of the 413 pa- Adequate (s-TOFHLA score 198 (48.5)
management associations, we forced race/ tients at least one HbA1c value was avail- 23–36)
ethnicity into the model to adjust for race/ able in the San Francisco General Language
ethnicity-associated confounders and to Hospital database; these patients com- Spanish 148 (36)
assess whether race/ethnicity was inde- prised our study sample. Patients who re- English 260 (64)
pendently associated with self- fused to participate and patients who Data are means ⫾ SD or n (%).
management. Specifically, we included were not interviewed by virtue of not at-
four groups: Asian/Pacific Islander, Afri- tending a clinic appointment during the
can American, Hispanic, and white/non- enrollment period were more likely than ported optimal self-management over the
Hispanic. We omitted Native American study subjects to be younger and male, prior week varied by domain: 33% re-
(n ⫽ 2), multiethnic (n ⫽ 6), and other but were not different in terms of race/ ported optimal diet adherence, 35% re-
(n ⫽ 11) ethnicity categories from the ethnicity or language. ported exercising 4 or more days in the
multivariate analysis because of the small The study participants (n ⫽ 408) prior week; 63% reported that they
number of respondents, which reduced were ethnically diverse. Seventy-five checked their feet daily for cuts and sores,
our sample by 19 respondents. (18%) were Asian/Pacific Islander, 100 54% performed daily SMBG, and 64% re-
Finally, we included health literacy (25%) were African American, 165 (40%) ported missing no medication doses in
score, as a continuous variable, in the were Hispanic, and 51 (12%) were white/ the prior 7 days.
multivariate models because of its poten- non-Hispanic. They had low income and In the univariate analysis, we found
tial to confound the self-efficacy–self- were predominantly uninsured or pub- an association between increasing self-
management associations (31). The final licly insured (Table 1; 198 (48.5%) had efficacy score and self-management with
multivariate models incorporated the adequate health literacy (s-TOFHLA regard to diet, exercise, SMBG, and foot
main predictor, self-efficacy, as well as score ⬎22), 54 (13.3%) marginal health care. We did not observe an association
significant diabetes-related factors, race/ literacy, and 156 (38.3%) inadequate between self-efficacy and medication ad-
ethnicity, and health literacy, on each self- health literacy. herence (Table 3, model A). When we ad-
management outcome. We assessed The mean self-efficacy score for the justed the univariate models for disease-
model fit for the linear regression models overall sample was 74 of 100 (SD 18). The related factors, the relationships between
by checking for normality of residuals, mean self-efficacy scores did not differ self-efficacy and the self-management
linearity of continuous variables, and ev- significantly across race/ethnicity or liter- outcomes did not change (Table 3, model
idence of violation of constant variance. acy levels (Table 2). The standardized B). Next, we adjusted for race/ethnicity as
Similarly, for the logistic models, we ap- Cronbach ␣ for the scale was 0.78, and well as disease characteristics, and the
plied the Hosmer-Lemeshow goodness- the scale had similar internal consistency– self-efficacy–self-management associa-
of-fit test. We also tested for two-way reliability across race/ethnicity and health tions persisted (Table 3, model C). Our
interactions between self-efficacy and literacy level (Table 2). final multivariate model included disease
race/ethnicity and self-efficacy and health The proportion of patients who re- characteristics, race/ethnicity, and health


Self-efficacy among diverse patients

Table 2—Self-efficacy scale performance plexity of regimen as well as system-

related factors such as costs and access are
Self-efficacy score Standardized Cronbach ␣ known determinants of medication ad-
herence (21,49) that may supersede self-
Overall 74 ⫾ 18 (16–100) 0.78 e f fi c a c y a m o n g t h i s p o p u l a t i o n .
Ethnicity Alternatively, investigators have noted
Asian/Pacific Islander 76 ⫾ 18 (29–100) 0.76 that self-report may not accurately mea-
African American 82 ⫾ 15 (29–100) 0.80 sure medication adherence (50), which
Hispanic 67 ⫾ 18 (16–100) 0.73 may have affected our results. Because
White/non-Hispanic 75 ⫾ 15 (33–100) 0.71 medication adherence has been shown to
Health literacy be associated with glycemic control
Inadequate/marginal 73 ⫾ 19 (16–100) 0.78 (51,52), more detailed studies should ad-
Adequate 74 ⫾ 17 (29–100) 0.76 dress barriers to medication adherence
Data are means ⫾ SD (range) unless otherwise indicated. across diverse populations.
Although the associations between
literacy score as covariates, and, again, pa- and across race/ethnicity and health liter- our measure of self-efficacy and self-
tients with a higher self-efficacy score acy, with internal consistency–reliability management in this study were consistent
were more likely to report optimal diet, scores within the accepted range for psy-
and statistically significant, the modest ef-
exercise, SMBG, and foot care but not chological measures (46). Even when ad-
fect sizes we found underscore the impor-
medication adherence (Table 3, model justed for strong clinical predictors of self-
tance of further study of self-efficacy and
D). Using the final multivariate model, management, such as insulin use and
self-management among vulnerable
with each 10% increase in self-efficacy duration of diabetes, the relationship be-
groups with diabetes. In disadvantaged
score, patients were more likely to report tween self-efficacy and diabetes self-
optimal diet (0.14 day more per week), management remained. Therefore, self- populations, a variety of experiences and
exercise (0.09 day more per week), SMBG efficacy is independently associated with barriers may undermine self-manage-
(increased odds of daily SMBG by 16%), disparate self-management behaviors. ment performance, including comorbid
and foot care (increased odds of daily foot Furthermore, our race/ethnicity-adjusted conditions such as depression or chronic
care by 22%). Neither sex nor low- analysis showed strikingly similar self- pain (53–55), patient-physician commu-
income status was associated with self- efficacy–self-management associations, nication problems (45,56), and economic
management (not shown). We did not suggesting that in our sample, race/ barriers such as the cost of glucose test
find significant interactions between self- ethnicity-related predictors of self- strips or medications (20,21,55,57).
efficacy and race/ethnicity or self-efficacy management function through Moreover, in low-income neighbor-
and health literacy on the self- mechanisms other than self-efficacy. hoods, external barriers, such as lack of
management outcomes, but we did see a We also investigated whether the re- safe space to exercise (58) and the scarce
trend toward improved medication ad- lationship between self-efficacy and self- availability of recommended fresh foods
herence with higher self-efficacy scores management behaviors was influenced by (59,60), may limit patients’ abilities to fol-
among African-American and white par- health literacy. Adjusting for health liter- low lifestyle recommendations. We can-
ticipants (P value for interaction 0.08). acy does not alter the self-efficacy–self- not determine whether such experiences
management associations: we conclude and barriers would affect self-manage-
CONCLUSIONS — We found that, that carefully designed self-management ment independent of self-efficacy. In
in our diverse sample, self-efficacy was interventions that target self-efficacy may addition, because the self-efficacy instru-
significantly associated with diet, exer- be effective in populations with limited ment we used does not specifically ad-
cise, SMBG, and foot care. When viewed health literacy, as suggested by recent dress such factors, our self-efficacy scores
in the context of long-term diabetes man- studies (35,44,47,48). may not detect the extent to which these
agement, these incremental differences in We did not find an association be- experiences or barriers may influence di-
self-management behaviors are clinically tween self-efficacy and medication adher- abetes self-efficacy, which may, in part,
significant. The diabetes self-efficacy scale ence. Other diabetes-related factors such explain the modest effect sizes we ob-
from this study performed well overall as adverse medication effects and com- served. Nevertheless, the consistency of

Table 3—Association between self-efficacy score and self-management outcomes

Model Diet coefficient* Exercise coefficient* SMBG OR* Foot care OR* Medication OR*
A. Unadjusted 0.16 (0.072–0.24) 0.11 (0.026–0.19) 1.18 (1.06–1.33) 1.23 (1.13–1.43) 1.04 (0.93–1.17), NS
B. Adjusted for diabetes factors† 0.14 (0.06–0.23) 0.09 (0.015–0.18) 1.16 (1.03–1.31) 1.22 (1.10–1.41) 1.10 (0.94–1.20), NS
C. Adjusted for above and race/ 0.15 (0.065–0.23) 0.09 (0.011–0.18) 1.15 (1.10–1.42) 1.24 (1.04–1.33) 1.05 (0.94–1.20), NS
D. Final model: adjusted for 0.16 (0.075–0.24) 0.10 (0.020–0.19) 1.14 (1.04–1.33) 1.27 (1.13–1.45) 1.08 (0.96–1.22), NS
above and health literacy§
Sequential models with addition of significant covariates are shown. All P values are ⬍0.05 except where indicated by NS. *Regression coefficients/odds ratios (ORs)
for a 10-point increase in the 0- to 100-point self-efficacy scale. †Adjusted for duration of diabetes and insulin use. ‡Adjusted for duration of diabetes, insulin use,
and race/ethnicity. §Adjusted for duration of diabetes, insulin use, race/ethnicity, and literacy score.


Sarkar, Fisher, and Schillinger

the self-efficacy–self-management associ- prospective studies. Finally, policy proving self-care among older patients
ations attests to the importance of self- should be focused on expanding the reach with type 2 diabetes: the “Sixty Some-
efficacy within the context of these other of diabetes self-management interven- thing” Study. Patient Educ Couns 19:61–
issues in vulnerable groups. tions to include racial/ethnically diverse 74, 1992
13. Hurley A, Shea C: Self-efficacy: strategy
Our study has several additional lim- populations across the spectrum of health
for enhancing diabetes self-care. Diabetes
itations. First, although a debate exists as literacy. Educ 18:146 –150, 1992
to the utility of disease-specific versus 14. Rubin R, Peyrot M, Saudek C: Effect of
global self-efficacy instruments (61), we diabetes education on self-care, metabolic
elected to use a disease-specific measure. Acknowledgments — U.S. was supported by control, and emotional well-being. Diabe-
Investigators have measured diabetes self- National Research Service Award grant 1 T32 tes Care 12:673– 679, 1989
efficacy with disparate instruments rang- HP19025. D.S. was supported by National In- 15. Maddigan S, Majumdar S, Guirguis L, Le-
ing from a single item to in-depth stitutes of Health Mentored Clinical Scientist wanczuk RZ, Lee TK, Toth EL, Johnson
Award K-23 RR16539-03. Electronic data JA: Improvements in patient-reported
cognitive interviews (62– 69). Because of were made available through the San Fran-
the variety of self-efficacy measures and outcomes associated with an intervention
cisco General Hospital General Clinical Re- to enhance quality of care for rural pa-
analytic strategies in the published litera- search Center grant M01RR00083-42. tients with type 2 diabetes. Diabetes Care
ture, we cannot accurately compare the The authors thank Frances Wang, MS, for 27:1306 –1312, 2004
effect sizes for the self-efficacy–self- her statistical advice. 16. Krichbaum K, Aarestad V, Buethe M:
management associations we found to Exploring the connection between self-
those found in other studies. Second, we efficacy and effective diabetes self-man-
measured diabetes self-management be- References agement. Diabetes Educ 29:653– 662,
haviors by self-report. Although self- 1. National Diabetes Fact Sheet. Atlanta, GA, 2003
report provides an imperfect estimate of Centers for Disease Control and Preven- 17. Glasgow R, Toobert D, Gilette C: Psycho-
health behavior, it represents the most tion, 2000 social barriers to diabetes self-manage-
common method of health behavior mea- 2. Cowie C, Eberhardt M: Sociodemo- ment and quality of life. Diabetes Spectrum
surement (69). Because we assessed each graphic characteristics of persons with di- 14:33– 41, 2001
abetes. In Diabetes in America. 2nd ed. 18. Lorig K, Ritter P, Jacquez A: Outcomes of
domain of self-management with a single border health Spanish/English chronic
Harris M, Cowie C, Reiber G, Boyko E,
item, our results should be interpreted Stern M, Bennett P, Eds. Washington, DC, disease self-management programs. Dia-
with caution. Although we cannot ex- U.S. Govt. Printing Office, 1995 betes Educ 31:401– 409, 2005
clude the possibility of differential mis- 3. Agency for Healthcare Research and Qual- 19. Skelly A, Marshall J, Haughey B, Davis P,
measurement (due to social desirability or ity: Fact sheet: diabetes disparities among Dunford R: Self efficacy and confidence as
culturally specific interpretations of ques- racial and ethnic minorities [article online], determinants of self-care practices in in-
tions by race/ethnicity or literacy), the 2004. Available from http://www.ahrq. ner-city, African-American women with
consistency of the relationships across gov/research/diabdisp.pdf. Accessed 20 non-insulin-dependent diabetes. Diabetes
these subgroups makes measurement bias October 2005 Educ 21:38 – 46, 1995
less likely. Finally, although we concep- 4. Bandura A: Self-Efficacy: The Exercise of 20. Karter A, Ferrara A, Darbinian J, Ackerson
Control. New York, W.H. Freeman, 1997 L, Selby J: Self-monitoring of blood glu-
tualized self-efficacy as a predictor of self-
5. Lorig K, Holman H: Self-management ed- cose: language and financial barriers in a
management based on prior intervention ucation: history, definition, outcomes, managed care population with diabetes.
studies (5,11,12,14 –16,18,62), given the and mechanisms. Ann Behav Med 26:1–7, Diabetes Care 23:477– 483, 2000
cross-sectional study design we cannot 2003 21. Piette J, Wagner T, Potter M, Schillinger
determine that the self-efficacy–self- 6. Bandura A: Self-efficacy: toward a unify- D: Health insurance status, cost-related
management relationship was causal. Be- ing theory of behavioral change. Psychol medication underuse, and outcomes
cause successful performance of self- Rev 84:191–215, 1977 among diabetes patients in three systems
management behavior could improve 7. McCaul K, Glasgow R, Schafer L: Diabetes of care. Med Care 42:102–109, 2004
self-efficacy over time, it is likely that regimen behaviors: predicting adherence. 22. Schorling J, Saunders J: Is “sugar” the
there is a reciprocal relationship between Med Care 25:868 – 881, 1987 same as diabetes? Diabetes Care 23:330 –
these constructs that requires further pro- 8. Aljasem L, Peyrot M, Wissow L, Rubin R: 334, 2000
The impact of barriers and self-efficacy on 23. Weiss B, Hart G, McGee D, D’Estelle S:
spective investigation. self-care behaviors in type 2 diabetes. Di- Health status of illiterate adults: relation
The consistency of the self-efficacy– abetes Educ 27:393– 404, 2001 between literacy and health status among
self-management relationships across 9. Bernal H, Woolley S, Schenaul J, Dickin- persons with low literacy skills. J Am
self-management domains in our sample son J: Correlates of self-efficacy in diabe- Board Fam Pract 5:257–264, 1992
suggests that self-efficacy constitutes a tes self-care among Hispanic adults with 24. Berkman N, DeWalt D, Pignone M, Sheri-
useful intervention target in vulnerable diabetes. Diabetes Educ 26:673– 680, dan S, Lohr K, Lux L, Sutton S, Swin-
populations. However, because of the nu- 2000 son T, Bonito A: Literacy and health
merous barriers to effective self- 10. Kavanagh D, Gooley S, Wilson P: Predic- outcomes: Evidence Report/Technology
management that these patients face, tion of adherence and control in diabetes. Assessment no. 87, 04-E007-1. [article
interventions must address self-efficacy J Behav Med 16:509 –523, 1993 online], 2004. Available from http://
11. Anderson R, Funnell M, Butler P, Arnold www.ahrq.gov/downloads/pub/evidence/
within the context of the patients’ envi- M, Fitzgerald J, Feste C: Patient empow- pdf/literacy.literacy.pdf. Accessed 20 Oc-
ronment. In addition, to increase the ef- erment: results of a randomized con- tober 2005
fectiveness of diabetes self-management trolled trial. Diabetes Care 18:943–949, 25. Baker D, Parker R, Williams M, Clark W,
interventions, the dimensions that con- 1995 Nurss J: The relationship of patient read-
tribute to self-efficacy and its develop- 12. Glasgow R, Toobert D, Hampson S, ing ability to self-reported health and use
ment over time should be explored in Brown J, Lewinsohn P, Donnelly J: Im- of health services. Am J Public Health 87:


Self-efficacy among diverse patients

1027–1030, 1997 self-management: the Summary of Diabe- betes Care 27:2800 –2805, 2004
26. Baker D, Parker R, Williams M, Clark W: tes Self-Care Activities Questionnaire. In 53. Krein SL, Heisler M, Piette JD, Makki F,
Health literacy and the risk of hospital ad- Handbook of Psychology and Diabetes. Clare Kerr EA: The effect of chronic pain on
mission [see comments]. J Gen Intern Med B, Ed. Amsterdam, the Netherlands, Har- diabetes patients’ self-management. Dia-
13:791–798, 1998 wood Academic Publishers, 1994. p. betes Care 28:65–70, 2005
27. Howard DH, Gazmararian J, Parker RM: 351–375 54. Regenstein M, Huang J, Schillinger D,
The impact of low health literacy on the 40. Toobert D, Hampson S, Glasgow R: The Lessler D, Reilly B, Brunt M, Johnson J,
medical costs of Medicare managed care Summary of Diabetes Self-Care Activities Piette JD: Caring for Patients with Diabetes
enrollees. Am J Med 118:371–377, 2005 Measure: results from 7 studies and a re- in Safety Net Hospitals and Health Systems:
28. Gordon MM, Hampson R, Capell HA, vised scale. Diabetes Care 23:943–950, What Patients Say About Their Care. Wash-
Madhok R: Illiteracy in rheumatoid ar- 2000 ington, DC, National Association of Pub-
thritis patients as determined by the 41. Baker D, Williams M, Parker R, Gazma- lic Hospitals, 2004
Rapid Estimate of Adult Literacy in Med- rarian J, Nurss J: Development of a brief 55. Jerant AF, von Friederichs-Fitzwater MM,
icine (REALM) score. Rheumatology (Ox- test to measure functional health literacy. Moore M: Patients’ perceived barriers to
ford) 41:750 –754, 2002 Patient Educ Couns 38:33– 42, 1999 active self-management of chronic condi-
29. Lindau ST, Tomori C, Lyons T, Langseth 42. Williams M, Parker R, Baker D, Parikh N, tions. Patient Educ Couns 57:300 –307,
L, Bennett CL, Garcia P: The association Pitkin K, Coates W: Inadequate func- 2005
of health literacy with cervical cancer pre- tional health literacy among patients at 56. Heisler M, Vijan S, Anderson RM, Ubel
vention knowledge and health behaviors two public hospitals [see comments]. PA, Bernstein SJ, Hofer TP: When do pa-
in a multiethnic cohort of women. Am J JAMA 274:1677–1682, 1995 tients and their physicians agree on dia-
Obstet Gynecol 186:938 –943, 2002 43. Cronbach L: Coefficient alpha and the in- betes treatment goals and strategies, and
30. Scott TL, Gazmararian JA, Williams MV, ternal structure of tests. Psychometrika 16: what difference does it make? J Gen Intern
Baker DW: Health literacy and preventive 297–334, 1951 Med 18:893–902, 2003
health care use among Medicare enrollees 44. Gerber B, Brodsky I, Lawless K, Smolin 57. Zgibor JC, Simmons D: Barriers to blood
in a managed care organization. Med Care LI, Arozullah AM, Smith EV, Berbaum glucose monitoring in a multiethnic com-
40:395– 404, 2002 ML. Heckerling PS, Eiser AR: Implemen- munity. Diabetes Care 25:1772–1777,
31. Schillinger D, Grumbach K, Piette J, tation and evaluation of a low-literacy di- 2002
Wang R, Osmond D, Daher C, Palacios J, abetes education computer multimedia 58. Dutton G, Johnson J, Whitehead D,
sullivan GD, Bindman AB: Association of application. Diabetes Care 28:1574 –1580, Bodenlos J, Brantley P: Barriers to physi-
health literacy with diabetes outcomes. 2005 cal activity among predominantly low-in-
JAMA 288:475– 482, 2002 45. Piette J, Schillinger D, Potter M, Heisler come African American patients with type
32. Nielsen-Bohlman L, Panzer A, Kindig D M: Dimensions of patient-provider com- 2 diabetes. Diabetes Care 28:1209 –1210,
(Eds.): Health Literacy: A Prescription to munication and diabetes self-care in an 2005
End Confusion. Washington, DC, National ethnically diverse population. J Gen Intern 59. Horowitz CR, Colson KA, Hebert PL, Lan-
Academies Press; 2004 Med 18:624 – 633, 2003 caster K: Barriers to buying healthy foods
33. Chesla C, Fisher L, Skaff M, Mullan J, Gil- 46. McDowell I, Newell C: Measuring Health: for people with diabetes: evidence of en-
liss C, Kanter R: Family predictors of dis- A Guide to Rating Scales and Questionnaires. vironmental disparities. Am J Public Health
ease management over one year in Latino 2nd ed. New York, Oxford University 94:1549 –1554, 2004
and European American patients with Press, 1996 60. Rose D, Richards R: Food store access and
type 2 diabetes. Fam Process 42:375–390, 47. Rothman R, Malone R, Bryant B, Horlen C, household fruit and vegetable use among
2003 DeWalt D, Pignone M: The relationship be- participants in the US Food Stamp Pro-
34. Fisher L, Chesla C, Chun K, Skaff MM, tween literacy and glycemic control in a gram. Public Health Nutr 7:1081–1088,
Mullan JT, Kanter RA, Gardiner PS: Pa- diabetes disease-management program. 2004
tient-appraised couple emotion manage- Diabetes Educ 30:263–273, 2004 61. Lorig K, Ritter P, Stewart A, Sobel DS,
ment and disease management among 48. Rothman RL, DeWalt DA, Malone R, Bry- Brown BW Jr, Bandura A, Gonzalez VM,
Chinese American patients with type 2 di- ant B, Shintani A, Crigler B, Weinberger Laurent DD, Holman HR: Chronic disease
abetes. J Fam Psychol 18:302–310, 2004 M, Pignone M: Influence of patient liter- self-management program: 2-year health
35. Kim S, Love F, Questberg D, Shea JA: acy on the effectiveness of a primary care- status and health utilization outcomes.
Association of health literacy with self- based diabetes disease management Med Care 39:1217–1223, 2001
management behavior in patients with di- program. JAMA 292:1711–1716, 2004 62. Anderson R, Funnell M, Fitzgerald J, Mar-
abetes. Diabetes Care 27:2980 –2982, 49. Rubin R: Adherence to pharmacologic rero D: The diabetes empowerment scale:
2004 therapy in patients with type 2 diabetes a measure of psychosocial self-efficacy.
36. Schillinger D, Bindman A, Wang F, Stew- mellitus. Am J Med 2005 118 (Suppl. 5A): Diabetes Care 23:739 –743, 2000
art A, Piette J: Functional health literacy 27S–34S, 2005 63. Hurley A: Insulin management diabetes
and the quality of physician-patient com- 50. Cook C, Wade W, Martin B, Perri MR. self-efficacy scale. In The Measure of Clin-
munication among diabetes patients. Pa- Concordance among three self-reported ical and Educational Nursing Outcomes,
tient Educ Couns 52:315–323, 2004 measures of medication adherence and Waltz F, Strickland O, Eds. New York,
37. Parker R, Baker D, Williams M, Nurss J: pharmacy refill records. J Am Pharm Assoc Springer, 1990, p. 28 – 44
The test of functional health literacy in 45:151–159, 2005 64. Connell C, Davis W, Gallant M, Sharpe P:
adults: a new instrument for measuring 51. Rhee M, Slocum W, Ziemer D, Culler SD, Impact of social support, social cognitive
patients’ literacy skills. J Gen Intern Med Cook DB, El-Kebbi IM, Gallina DL, Bar- variables, and perceived threat on depres-
10:537–541, 1995 nes C, Phillips LS: Patient adherence im- sion among adults with diabetes. Health
38. Skaff M, Mullan J, Fisher L, Chesla C: A proves glycemic control. Diabetes Educ Psychol 13:263–273, 1994
contextual model of control beliefs, be- 31:240 –250, 2005 65. Skyler J: Psychological issues in diabetes.
havior, and health: Latino and European 52. Pladevall M, Williams L, Potts L, Divine Diabetes Care 4:656 – 657, 1981
Americans with type 2 diabetes. Psychol G, Xi H, Lefata JE: Clinical outcomes and 66. Grossman H, Brink S, Hauser S: Self-effi-
Health 18:295–312, 2003 adherence to medications measured by cacy in adolescents with IDDM. Diabetes
39. Toobert D, Glasgow R: Assessing diabetes claims data inpatients with diabetes. Dia- Care 10:324 –329, 1987


Sarkar, Fisher, and Schillinger

67. Ikeda K, Aoki H, Saito K, Muramatsu Y, 68. Piette J, Weinberger M, McPhee S: The 69. Glasgow R, Ory M, Klesges L, Cifuentes
Suzuki T: Associations of blood glucose effect of automated calls with telephone M, Fernald D, Green L: Practical and rel-
control with self-efficacy and rated anxi- nurse follow-up on patient-centered out- evant self-report measures of patient
ety-depression in type 2 diabetes patients. comes of diabetes care. Med Care 38:218 – health behaviors for primary care re-
Psychol Rep 92:540 –544, 2003 230, 2000 search. Ann Fam Med 3:73– 81, 2005