Epidemiology/Health Services/Psychosocial Research
O R I G I N A L
A R T I C L E
Adequacy of Glycemic, Lipid, and Blood
Pressure Management for Patients With
Diabetes in a Managed Care Setting
SARAH J. BEATON, PHD1
SOMA S. NAG, PHD2
MARGARET J. GUNTER, PHD1
JEREMY M. GLEESON, MD3
SHIVA S. SAJJAN, PHD2
CHARLES M. ALEXANDER, MD2
OBJECTIVE — We conducted a retrospective study to evaluate the adequacy of glycemic,
lipid, and blood pressure (BP) management for diabetic patients in a managed care organization
(MCO).
RESEARCH DESIGN AND METHODS — Patients aged ⱖ18 years with diabetes (n ⫽
7,114) were retrospectively identified over a 2-year period from the MCO’s administrative
database based on the Health Plan Employer Data and Information Set 2000 selection criteria
using pharmacy, laboratory, and encounter data. Analyses examined demographics and percentages of patients tested and meeting American Diabetes Association goals for HbA1c, lipids, and
BP, both overall and for those receiving medication treatment versus no treatment.
RESULTS — Testing rates for A1C, LDL cholesterol, and BP were 77, 54, and 95%, respectively. The percentage of patients tested who were at goal were 37% for A1C, 23% for LDL
cholesterol, and 41% for systolic BP. Of the patients in our sample, 72% were treated for glycemic
control, 64% were treated for BP control, and only 28% were treated for lipid control. Of the
patients who received medication treatment, less than one-third were at goal for A1C (29%) and
LDL cholesterol (32%), whereas 40% were at goal for systolic BP.
CONCLUSIONS — We found that although a large percentage of diabetic patients were
tested for A1C, LDL cholesterol, and systolic BP, a much smaller percentage had reached their
respective goals. More aggressive glycemic, lipid, and BP management appears to be needed to
improve care for these patients.
Diabetes Care 27:694 – 698, 2004
C
ardiovascular disease is the major
cause of morbidity and mortality
among diabetic patients, accounting
for ⬃75% of hospitalizations and 70 –
80% of deaths (1,2). In fact, coronary
heart disease (CHD) is the leading cause
of death among diabetic patients, who
have a two- to fourfold higher risk of CHD
mortality and incidence of nonfatal CHD
events compared with patients without
diabetes (3).
Initiatives such as the Be Smart About
Your Heart: Control the ABCs of Diabetes
campaign (developed by the American
Diabetes Association [ADA], the U.S. Department of Health and Human Services,
and the National Diabetes Education Program) (4) and the ADA/American College
● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●
From the 1Lovelace Clinic Foundation, Albuquerque, New Mexico; 2Outcomes Research & Management,
U.S. Medical & Scientific Affairs, Merck, West Point, Pennsylvania; and the 3Lovelace Sandia Health System,
Albuquerque, New Mexico.
Address correspondence and reprint requests to Sarah J. Beaton, PhD, Lovelace Clinic Foundation, 2309
Renard Pl. SE, Suite 103, Albuquerque, NM 87106. E-mail: [email protected].
Received for publication 7 May 2003 and accepted in revised form 9 December 2003.
S.J.B. and M.J.G. have received grant/research support from Merck. J.M.G. has received speaking fees from
Merck.
Abbreviations: ADA, American Diabetes Association; BP, blood pressure; CHD, coronary heart disease;
HEDIS, Health Plan Employer Data and Information Set; MCO, managed care organization.
© 2004 by the American Diabetes Association.
694
of Cardiology’s Make the Link! Diabetes,
Heart Disease and Stroke initiative (5)
highlight the link between diabetes and
cardiovascular disease. Both of these initiatives focus on comprehensive diabetes
care and underscore the importance of
managing blood glucose (A1C), blood
pressure (BP), and cholesterol to targets
recommended by the ADA guidelines.
Few studies report treatment and/or
goal attainment rates for diabetic patients
with respect to these risk factors (6 –9).
Aside from these few studies, there is little
or no information on the management of
patients with diabetes, especially within
the managed care setting. Additionally,
there has been little exploration of the
possible differences in management by
race/ethnicity.
To address these issues, we conducted a retrospective study to evaluate
the adequacy of glycemic, lipid, and BP
management for diabetic patients within a
large staff model health maintenance organization. The primary objective of this
study was to examine A1C, cholesterol
(LDL, HDL, and triglycerides), and BP
(systolic and diastolic) testing and goal attainment rates, and to examine differences in goal attainment rates between
those diabetic patients receiving medication versus those not receiving medication. A secondary objective was to
examine potential differences by ethnicity
for testing, treatment, and goal attainment
for the risk factors mentioned above.
RESEARCH DESIGN AND
METHODS — The study population
included all members of a large staff
model managed care organization (MCO)
in New Mexico who were continuously
enrolled from 1 January 1999 through 31
December 2000 and who had a pharmacy
benefit (n ⫽ 96,926), including those in a
Medicare risk plan. Patients with diabetes
(n ⫽ 7,114) aged ⱖ18 years were retrospectively identified from the MCO’s electronic administrative database using an
adaptation of the National Committee for
Quality Assurance’s Health Plan EmDIABETES CARE, VOLUME 27, NUMBER 3, MARCH 2004
Beaton and Associates
ployer Data and Information Set (HEDIS)
2000 selection criteria. Patients were
identified through pharmacy data (those
with an insulin or oral hypoglycemic/
antihyperglycemic prescription fill;
2.1%), encounter data (two or more outpatient visits with a diabetes diagnosis
code [250.xx, 357.2, 362.0, 366.41, or
648.0] or one inpatient or emergency department encounter with a diabetes diagnosis code; 97.5%), and laboratory data
(A1C results; 0.4%) obtained from the
MCO’s administrative database. In addition, we also identified those patients with
a diagnosis of hyperlipidemia (272.0,
272.1, 272.2, or 272.0) and/or hypertension (401.0, 401.1, or 401.9).
Because BP is not available in the
MCO’s administrative data, and due to resource considerations, a medical chart review was conducted to obtain BP values
for a sample of 409 patients randomly
drawn from those with diabetes. This
sample size of 409 was based on the mandated HEDIS sample size for measures requiring medical record review.
This MCO’s health plan, like most
others, does not collect race/ethnicity information on its individual members, except in its Medicaid population. However,
the MCO has access to special software,
GUESS (Generally Useful Ethnicity
Search System), that assigns ethnicity
based on surname and has been validated
on the New Mexico population (10). This
software was developed in the 1970s at
the University of New Mexico, and although surnames are grouped into a variety of ethnic classes by the software, the
only reliable classifications for New Mexico are Hispanic and non-Hispanic. This
MCO’s population is predominantly Hispanic and non-Hispanic white with very
low proportions of African Americans and
Native Americans.
Table 1—Demographic characteristics for the health plan population and the diabetes sample
Analyses
Descriptive analyses included comparing
demographic profiles of the diabetes population with those of the total health plan
population. For the overall diabetic sample, we calculated the percentage of patients who were tested, treated, and at
goal for each of the risk factors over the
2-year study period and used ␹2 statistics
to determine significance.
For each primary risk factors (A1C,
LDL cholesterol, and systolic BP), we
identified individuals who 1) had at least
one medication fill during the study pe-
RESULTS — As expected, patients in
the diabetic sample were older and more
likely to be male than in the continuously
enrolled health plan population (Table 1).
There were significantly more Hispanics
in the diabetic population (46.1%) than
in the overall health plan population
(35.9%) (P ⬍ 0.001), validating the
higher prevalence rate for Hispanics
found in national studies.
A sample of 450 patients was randomly drawn from the diabetic sample of
7,114 patients with the stipulation that
both samples were matched for age range,
DIABETES CARE, VOLUME 27, NUMBER 3, MARCH 2004
n
Age (years)
Age range (years)
18–39
40–59
60–79
⬎80
Sex
Male
Female
Ethnicity
Hispanic
Native American
Non-Hispanic
Health plan
population
Diabetes sample
Chart review sample
for BP
96,926
49.5 ⫾ 17.5
7,114
61.7 ⫾ 13.6
409
64.1 ⫾ 12.5
27,665 (28.6)
42,017 (43.3)
22,603 (23.3)
4,641 (4.8)
400 (5.6)
2,665 (37.5)
3,423 (48.1)
626 (8.8)
14 (3.4)
136 (33.3)
216 (52.8)
43 (10.5)
44,685 (46.1)
52,241 (53.9)
3,649 (51.3)
3,465 (48.7)
207 (50.6)
202 (49.4)
34,839 (35.9)
1,244 (1.3)
60,843 (62.8)
3,279 (46.1)
102 (1.4)
3,733 (52.5)
176 (43.0)
9 (2.2)
224 (54.8)
Data are means ⫾ SD or frequency (percent).
riod (insulin or oral hypoglycemic agent,
lipid-lowering medication, or antihypertension medication for A1C, lipids, or systolic BP, respectively) or 2) had no
medication fill. We then examined treatment and goal attainment rates for these
subgroups with respect to glycemic, lipid,
and BP management. We also examined
differences in A1C, lipid, and BP treatment and goal attainment by ethnicity
(Hispanic versus non-Hispanic) for the
groups for each risk factor.
The goals used for this study were
those specified by the 2003 ADA guidelines (11): 1) A1C ⬍7%, 2) LDL cholesterol ⬍100 mg/dl, 3) HDL cholesterol
⬎45 mg/dl, 4) triglycerides ⬍150 mg/dl,
and 5) systolic BP ⬍130 mmHg and diastolic BP ⬍80 mmHg. We particularly focused on systolic BP as a risk factor
because research has shown systolic BP to
be more predictive of CHD death and cardiovascular disease risk (12).
sex, and ethnicity. Of this sample, 409
patients had medical charts available for
review. Table 1 shows the summary demographics. The distribution of demographic variables for the two samples was
not significantly different for sex and ethnicity; for age range the difference was
small but reached statistical significance
(P ⬍ 0.05).
Testing and medication rates
Over the 2-year study period, 77.4% (n ⫽
5,505) of the 7,114 diabetic patients were
tested at least once for A1C, 54.0% (n ⫽
3,845) were tested for LDL cholesterol,
and 94.9% of the chart review sample had
a documented BP reading (388 of 409).
We recognize that the standard of care for
patients with diabetes is to have these risk
factors measured at least annually, but we
wanted to ensure a sufficient time period
to allow for testing results. Of the 5,505
patients with an A1C test, only 683
(12.4%) did not have an A1C test during
the most recent year.
Table 2 shows the percentage of diabetic patients on medications; 72.3% for
glycemic control medications, 27.5% for
lipid-lowering medications, and 63.8%
for antihypertensive medications. Of
those diabetic patients who also had a hyperlipidemia diagnosis (2,093 of 7,114),
64.9% were taking a lipid-lowering medication (of the 5,021 diabetic patients
with no hyperlipidemia diagnosis, 11.9%
were taking a lipid-lowering medication).
Of the diabetic patients in the chart review sample who also had a hypertension
695
MCO management of patients with diabetes
Table 2—Medications prescribed for glycemic, lipid, and BP control
Frequency (percent)
Glycemic control agents (n ⫽ 7,114)
Diet only/no meds
Insulin/oral agent
Lipid-lowering agents (n ⫽ 7,114)
No meds
Lipid-lowering meds
Statin only
Non-statin only
Statin and non-statin
Antihypertensive agents (n ⫽ 409)
No meds
Antihypertensive meds
Monotherapy
Combination therapy
diagnosis (213 of 409), 86.9% were on
antihypertensive medications (of the 196
patients in this sample with no hypertension diagnosis, 38.8% were taking an antihypertensive medication).
Glycemic, lipid, and BP profiles
Table 3 shows the glycemic and lipid profiles for health plan members with diabetes (n ⫽ 7,114) who received at least one
glycemic or lipid test (with values available) within the study period. Although
the largest percentage of patients fell in
the optimal category for glycemics
(37.2% with A1C ⬍7%), 30.1% had an
A1C between 7– 8% (30.1%) and 32.7%
had an A1C ⬎8%. For lipid levels, only
22.5% were at goal (⬍100 mg/dl) for LDL
cholesterol; however, when the LDL cholesterol cut point was increased to 130
mg/dl, this proportion increased to
67.1%. About 37% had HDL cholesterol
⬎45 mg/dl, and 34% had triglyceride values ⬍150 mg/dl. Table 3 also includes BP
profiles for the sample of 409 diabetic patients included in the chart review with
BP readings available. Of the 388 patients
with BP values, 41.2% had a systolic BP
⬍130 mmHg and 54.1% had a diastolic
BP ⬍80 mmHg. There were 28.6% who
were at goal for both systolic and diastolic
BP (⬍130/⬍80 mmHg). A much larger
percentage of patients were at goal for diastolic than systolic BP, and this difference was even greater when comparing
patients with high-risk BP; 37.2% of systolic patients had BP ⱖ140 mmHg compared with 11.7% of diastolic patients
with BP ⱖ90 mmHg.
696
1,971 (27.7)
5,143 (72.3)
5,157 (72.5)
1,957 (27.5)
1,525 (21.4)
260 (3.7)
172 (2.4)
148 (36.2)
261 (63.8)
143 (34.9)
118 (28.9)
Treated versus untreated subgroups
Diabetic patients with an A1C, LDL cholesterol, or BP test were separated into
those either taking or not taking a medication for glycemic, lipid, and BP control.
Figure 1 shows that for glycemic control,
goal attainment rates were twice as high
for treated (59.7%) versus untreated
(29.1%) patients. In contrast, the reverse
relationship was shown for LDL cholesterol; almost twice as many treated
(31.5%) versus untreated (17.1%) patients were at goal. For systolic BP, there
was little difference between treated and
untreated patients (39.9 vs. 43.6%).
Differences across ethnicity
Because the percentage of Native-American
patients with diabetes was very low for
this MCO (1.4%) they were not included
in the ethnicity analysis. In examining differences by ethnicity, we found a signifi-
Table 3—Glycemic, lipid, and BP profiles for patients with diabetes
Frequency (percentage of those
tested with values available)
A1C (n ⫽ 7,114)
⬍7%
7–8%
8.1–9%
9.1–10%
⬎10%
LDL cholesterol (n ⫽ 7,114)
⬍100 mg/dl
100–129 mg/dl
130–159 mg/dl
160 mg/dl
HDL cholesterol (n ⫽ 7,114)
⬎45 mg/dl
40–45 mg/dl
35–39 mg/dl
⬍35 mg/dl
Triglycerides (n ⫽ 3,985)
⬍150 mg/dl
150–199 mg/dl
200–499 mg/dl
500 mg/dl
Systolic BP (n ⫽ 409)
⬍130 mmHg
130–139 mmHg
140–159 mmHg
160⫹ mmHg
Diastolic BP (n ⫽ 409)
⬍80 mmHg
80–89 mmHg
90–99 mmHg
100⫹ mmHg
5,505 (77.4)
2,048 (37.2)
1,655 (30.1)
793 (14.4)
456 (8.3)
553 (10.0)
3,845 (54.0)
866 (22.5)
1,716 (44.6)
917 (23.9)
346 (9.0)
4,878 (68.6)
1,805 (37.0)
994 (20.4)
869 (17.8)
1,210 (24.8)
3,985 (56.0)
1,348 (33.8)
862 (21.6)
1,616 (40.6)
159 (4.0)
388 (94.9)
160 (41.2)
76 (19.6)
104 (26.8)
48 (12.4)
388 (94.9)
210 (54.1)
130 (33.5)
35 (9.0)
13 (3.4)
DIABETES CARE, VOLUME 27, NUMBER 3, MARCH 2004
Beaton and Associates
Figure 1— Goal attainment rates among patients treated versus untreated for glycemic, lipid, and
BP control. Number of individuals with a test for A1C, n ⫽ 5,505; LDL cholesterol, n ⫽ 3,845;
systolic BP, n ⫽ 388. *Percentage of treated patients with a test who were at goal; **percentage
of untreated patients with a test who were at goal.
cantly lower percentage of Hispanics
(26.8%) were at goal for A1C than nonHispanics (30.4%; P ⬍ 0.001). A small
but significant difference in the same direction was found for goal attainment between Hispanics and non-Hispanics for
LDL cholesterol (11.5 and 12.8%, respectively). The same trend, although not significant, was found for systolic BP; 36.9%
of Hispanics were at goal vs. 40.2% nonHispanics.
Of those treated for glycemic control,
a similar percentage of Hispanic (22.0%)
and non-Hispanic patients (23.6%) had
an A1C ⬍7%. There were also no differences in goal attainment among those
who received lipid-lowering agents between Hispanic (22.5%) and nonHispanic patients (23.8%). Although BP
goal attainment was lower among Hispanics treated with antihypertensive medication (32.1%) compared with treated nonHispanics (41.0%), this difference was
not significant (P ⬍ 0.10).
Relationship between A1C, lipid,
and systolic BP levels
Finally, we examined the relationships
between glycemic level and lipid and BP
levels for those patients with both A1C
and lipid or BP scores. Of the 3,656 patients with both A1C and LDL cholesterol
scores, 37.7% were at goal (⬍7%) for
A1C and 24.3% were at goal (⬍100 mg/
dl) for LDL cholesterol (10.0% were at
goal for both A1C and LDL cholesterol).
However, over half of the patients in this
subgroup had high-risk scores for either
A1C (⬎8%) or LDL cholesterol (⬎130
mg/dl) (30.2 and 32.0% of patients, reDIABETES CARE, VOLUME 27, NUMBER 3, MARCH 2004
spectively), and 385 patients (11.3%) had
high-risk scores for both A1C and LDL
cholesterol.
The relationship between A1C and
HDL cholesterol was somewhat better
than with LDL cholesterol; 45.2% of 4,609
patients with both A1C and HDL cholesterol had high-risk scores for either A1C
(⬎8%) or HDL cholesterol (⬍35 mg/dl)
(8.3% had high-risk scores for both factors). For A1C and triglycerides, of the
3,784 patients with scores for both risk factors, 1,347 (35.6%) had high-risk scores
for either A1C or triglycerides (3.4% had
high-risk scores for both factors).
For the BP chart review sample of 409
patients, there were quite different distributions between systolic and diastolic
pressures and A1C. Of the 367 patients
with both A1C and BP scores, 43.3% had
either A1C or systolic high-risk scores
(6.8% had both A1C and systolic highrisk scores). But only 31.9% of the 367
patients with A1C and diastolic scores
had high-risk scores for either risk factor
(1.4% had both A1C and diastolic highrisk scores).
CONCLUSIONS — The results of
this study show that although overall testing and medication rates were fairly high
(with the exception of LDL cholesterol),
goal attainment was modest for all three
risk factors (glycemic, lipid, and systolic
BP control). In this study overall, a greater
proportion of patients were tested for
A1C than for LDL cholesterol. For BP,
⬃95% of the randomly selected sample of
patients had documentation of BP level.
With respect to goal attainment,
much lower rates were found: 37% for
A1C, 23% for LDL cholesterol, and 41%
for systolic BP. And even these poor goal
attainment rates are deceptively high
since the denominators do not include
those patients with no tests conducted
during the 2-year study period. Medication rates for glycemic control were almost three times higher when compared
with those for lipid control (72 vs. 28%),
and almost two-thirds of the chart review
sample were treated with BP medications.
Similarly, when examining a breakdown of those patients receiving versus
not receiving medication, differences
were found for goal attainment. Although
high rates of goal attainment were found
for A1C among untreated patients (60%)
and for systolic BP among treated and untreated patients (40 and 44%, respectively), the very low goal attainment rate
for LDL cholesterol with treated patients
(17%) was surprising. Reasons for the differences between treated and untreated
patients are not clear. Although it is likely
that patients who are the most “out of
control” may be the ones receiving medications, it is also true that those not on
medications may be better able to control
their risk factors through diet. Nevertheless, it is disappointing to find so many
patients on medications unable to attain
their glycemic, lipid, or BP goals.
Our data are somewhat better than
the data of Straka et al. (13), who reported
slightly lower lipid treatment (25%) and
goal attainment (21%) rates for patients
with diabetes and without CHD in a staff
model MCO from 1996 to 1998. However, Straka et al. did not examine treatment or goal attainment for glucose or BP
control.
A limitation to our findings, however,
is that data for this retrospective study
were restricted to the years 1999 through
2000. One would expect higher levels of
goal attainment at present, i.e., 3 years
later. For example, Karter et al. (14) found
a greater percentage of patients in 1996 –
1998 with high-risk A1C scores than our
study. A longitudinal study would be
needed to validate this expectation.
Additionally, we believe it is important to examine not just goal attainment
but also the distribution of values across
the range of for all the risk factors (Table
3) in order to better understand where the
challenges lie in managing patients with
diabetes. For example, although a larger
proportion of patients had HDL choles697
MCO management of patients with diabetes
terol ⬎45 ml/dl (37%) than LDL cholesterol ⬍100 ml/dl (20%), an even larger
difference existed when examining the
proportion with high-risk HDL cholesterol ⬍35 ml/dl (25%) than LDL cholesterol ⱖ160 ml/dl (8%). This suggests that
concentrating efforts on goal attainment
alone may create a similar “disconnect”
between simply reaching the ADA goal
versus providing good diabetes management for all patients, even those not yet at
goal.
Finally, a distinctive feature of our
study was that the diabetic sample included a significantly higher percentage
of Hispanics (46%) than the overall health
plan, which is consistent with the higher
diabetes prevalence rate found in many
parts of the U.S. Although the differences
between Hispanics and non-Hispanics
with respect to goal attainment for any of
the risk factors were relatively small, we
believe the statistical significance indicates that MCOs need to monitor differences in ethnicity.
In summary, these findings show
that, within the managed care setting, diabetic patients still continue to be untreated or undertreated for glycemic,
lipid, and BP management, which in turn
is associated with a low rate of goal attainment. This treatment gap was even greater
for lipid control compared with glycemic
and BP control in our study. Moreover,
despite much higher treatment rates for
glucose control, A1C goal attainment
698
among diabetic patients was nevertheless
modest. The results of our study provide
further evidence of the need to implement
aggressive glycemic, lipid, and BP management to improve goal attainment,
thereby contributing to a reduction in the
burden of cardiovascular disease among
patients with diabetes.
References
1. Goldberg RB, Capuzzi D: Lipid disorders
in type 1 and type 2 diabetes. Clin Lab Med
21:147–172, 2001
2. Wingard DL, Barrett-Connor E: Heart disease and diabetes. In Diabetes in America.
2nd ed. Bethesda, MD, National Diabetes
Data Group, National Institutes of Health,
1995, p. 429 – 448 (NIH publication no.
95-1468)
3. Haffner SM, Lehto S, Ronnemaa T,
Pyorala K, Laakso M: Mortality from coronary heart disease in subjects with type 2
diabetes and in nondiabetic subjects with
and without prior myocardial infarction.
N Engl J Med 339:229 –234, 1998
4. National Institute of Diabetes and Digestive and Kidney Diseases: As diabetes epidemic surges, HHS and ADA join forces
to fight heart disease, the leading cause of
death for people with diabetes [article online], 2001. Available from http://www.
niddk.nih.gov/welcome/releases/11-0101.htm. Accessed 1 May 2003
5. American Diabetes Association, American
College of Cardiology: Make the link! initiative: hypertension in diabetes. In Diabetes and Cardiovascular Disease Review.
Issue 2, 2002
6. Saaddine JB, Engelgau MM, Beckles GL,
Gregg EW, Thompson TJ, Venkat
Narayan KM: A diabetes report card for
the United States: quality of care in the
1990s. Ann Intern Med 136:565–574,
2002
7. Harris MI: Health care and health status
and outcomes for patients with type 2 diabetes. Diabetes Care 23:754 –758, 2000
8. Kell SH, Drass J, Barker Bausell R, Thomas
KA, Osborn MA, Gohdes D: Measures of
disease control in medicare beneficiaries
with diabetes mellitus. J Am Geriatr Soc
47:417– 422, 1999
9. Geiss LS, Rolka DB, Engelgau MM: Elevated blood pressure among U.S. adults
with diabetes, 1988 –1994. Am J Prev Med
22:42– 48, 2001
10. Buechley RA: A reproducible method of
counting persons of Spanish surname.
J Am Stat Assoc 56:88 –97, 1961
11. American Diabetes Association: Standards of medical care for patients with diabetes mellitus. Diabetes Care 26 (Suppl.
1):S33–S50, 2003
12. Deedwania JR: The changing face of hypertension: is systolic blood pressure the
final answer? Arch Intern Med 162:506 –
508, 2002
13. Straka RJ, Taheri R, Cooper SL, Tan AW,
Smith AC: Assessment of hypercholesterolemia control in a managed care organization. Pharmacotherapy 21:818 – 827,
2001
14. Karter AJ, Ferrara A, Liu JY, Moffet HH,
Ackerson LM, Selby JV: Ethnic disparities
in diabetic complications in an insured
population. JAMA 287:2519 –2527, 2002
DIABETES CARE, VOLUME 27, NUMBER 3, MARCH 2004