ORIGINAL RESEARCH

Antihypertensive Drug Class Interactions and Risk for Incident

Diabetes: A Nested Case–Control Study
Rhonda M. Cooper-DeHoff, PharmD, MS; Steven T. Bird, PharmD, MS; Gregory A. Nichols, PhD; Joseph A. Delaney, PhD;
Almut G. Winterstein, PhD

Background-—We aimed to determine how single and combination antihypertensive therapy alters risk for diabetes mellitus
(DM).Thiazide diuretics (TD), b blockers (BB), and renin–angiotensin system blockers (RASB) impact DM risk while calcium channel
blockers (CCB) are neutral. DM risk associated with combinations is unclear.
Methods and Results-—We enrolled nondiabetic patients from Kaiser Permanente Northwest with a fasting plasma glucose (FPG)
<126 mg/dL between 1997 and 2010. DM cases were defined by a FPG ≥126 mg/dL, random plasma glucose ≥200 mg/dL,
HbA1c ≥7.0%, or new DM prescription (index date). We used incidence density sampling to match 10 controls per case on the date
of follow-up glucose test (to reduce detection bias), in addition to age and date of cohort entry. Exposure to antihypertensive class
was assessed during the 30 days prior to index date. Our cohort contained 134 967 patients and had 412 604 glucose tests
eligible for matching. A total of 9097 DM cases were matched to 90 495 controls (median age 51 years). Exposure to TD (OR 1.54,
95% CI 1.41 to 1.68) or BB (OR 1.19, 95% CI 1.11 to 1.28) was associated with an increased DM risk, while CCB and RASB
exposure was not. TD+BB combination resulted in the fully combined diabetogenic risk of both agents (OR 1.99, 95% CI 1.80 to
2.20; interaction OR 1.09, 95% CI 0.97 to 1.22). In contrast, combination of RASB with either TD or BB showed significant negative
interactions, resulting in a smaller DM risk than TD or BB monotherapy.
Conclusions-—Diabetogenic potential of combination therapy should be considered when prescribing antihypertensive therapy.
( J Am Heart Assoc. 2013;2:e000125 doi: 10.1161/JAHA.113.000125)
Key Words: b blockers • diabetes • diabetogenic • drug interactions • hypertension • RAS blockers • thiazide diuretics

T ype 2 diabetes mellitus (DM) is a major societal and

clinical concern. In 2009, 655 000 adults were admitted
to a hospital due to DM or related complications, and DM,
which is a leading cause of cardiovascular (CV) morbidity and
mortality worldwide, is associated with an estimated cost of
$174 billion annually.1,2 These issues have forced consider-
From the Departments of Pharmacotherapy and Translational Research (R.M.C.-D.)
and Pharmaceutical Outcomes and Policy (S.T.B., J.A.D., A.G.W.), College of
ation of controllable conditions that may predispose individ-
Pharmacy, University of Florida, Gainesville, FL; Division of Cardiovascular uals to the development of diabetes.
Medicine, College of Medicine, University of Florida, Gainesville, FL (R.M.C.-D.); Hypertension is a prevalent health challenge in the United
Department of Health and Human Services, Food and Drug Administration, Center
States, affecting 30% of adults and 5% to 10% of
for Drug Evaluation and Research, Office of Surveillance and Epidemiology,
Department of Epidemiology, Silver Spring, MD (S.T.B.); Kaiser Permanente Center adolescents.3–5 By 2025, it is anticipated to afflict 1.56
for Health Research, Portland, OR (G.A.N.); Department of Epidemiology, University billion adults worldwide.6 Hypertension prevalence is closely
of Washington, Seattle, WA (J.A.D.); Department of Epidemiology, College of Public linked with obesity, and both increase risk for DM.4,7 While
Health and Health Professions and College of Medicine, University of Florida,
Gainesville, FL (A.G.W.). lifestyle modification remains the first step in the hyperten-
This research was presented at the American Heart Association 2012 sion treatment process, most individuals will require one or
Scientific Sessions in Los Angeles, California. more antihypertensive drugs, which could influence blood
Accompanying Appendix S1 is available at http://jaha.ahajournals.org/ glucose and DM risk.8 Use of antihypertensive drugs is
content/2/3/e000125/suppl/DC1
common, and more than 472 million antihypertensive
Correspondence to: Rhonda M. Cooper-DeHoff, PharmD, MS, Department of
Pharmacotherapy and Translational Research, College of Pharmacy, University prescriptions were filled in US pharmacies in 2010.9,10
of Florida, 1600 SW Archer Road, PO Box 100486, Gainesville, FL 32611. E- Some analyses have suggested that concurrent use of
mail: dehoff@cop.ufl.edu thiazide diuretics (TD) and b blockers (BB) can increase risk
Received February 11, 2013; accepted May 14, 2013.
for DM, while renin–angiotensin system blockers (RASB),
ª 2013 The Authors. Published on behalf of the American Heart Association,
Inc., by Wiley-Blackwell. This is an Open Access article under the terms of the
including angiotensin-converting enzyme (ACE) inhibitors and
Creative Commons Attribution Noncommercial License, which permits use, angiotensin receptor blockers (ARBs), have been shown to
distribution and reproduction in any medium, provided the original work is reduce DM risk.11–13
properly cited and is not used for commercial purposes.

DOI: 10.1161/JAHA.113.000125 Journal of the American Heart Association 1

Downloaded from http://jaha.ahajournals.org/ by guest on March 28, 2015

Antihypertensive Classes and Incident Diabetes Cooper-DeHoff et al

ORIGINAL RESEARCH
Little is known about whether, and to what extent, medications, laboratory tests, and in- or outpatient visits with
coprescribing drugs from different antihypertensive classes a diabetes diagnosis [ICD9-CM 250.x]). Patients were required
can result in drug–drug interactions that might alter the to have 1 year of plan eligibility after cohort entry. Patients
glucose modifying effects of individual agents. Such drug– were censored at development of DM, cessation of KPNW
drug interactions could increase or decrease the DM risk, enrollment, or December 2010, whichever came first. This
having potential additive or multiplicative effects. Under- research was reviewed and approved by the institutional
standing the effects of combination antihypertensive medica- review and privacy boards at the University of Florida and
tion regimens on DM development is important, given KPNW.
the increasing number of individuals who require combina-
tion therapy.14 Therefore, our study objective was to assess
risk for incident DM with combination antihypertensive Diabetes Cases
therapy. For the purposes of this analysis, incident DM was defined as
a new FPG ≥126 mg/dL, random plasma glucose (RPG)
≥200 mg/dL, an HbA1c ≥7.0%, or a new DM prescription,
Methods where the index date of the cases was the first of these occur.
Kaiser Permanente Northwest (KPNW) is a group-model health Because guidelines for use of HbA1c ≥6.5% as a DM diagnostic
maintenance organization that provides integrated health care criterion were not in place in the United States until early in
to 475 000 members in the Portland, Oregon area.15 KPNW 2010,16 we used a cutoff of 7% in our study to better reflect
maintains electronic medical records (EMR) to document diagnostic practice during the study period. We also per-
clinical interactions between physicians and patients. The formed sensitivity analyses whereby DM cases were
EMR contains information on all inpatient and outpatient restricted to those with at least one subsequent positive
encounters, pharmacy dispensing data, and laboratory tests. DM test within 1 year after original DM diagnosis. This test
Diagnoses are coded in International Classification of Disease could include any of the following: a second FPG ≥126 mg/dL,
—9th Revision—Clinical Modification (ICD-9-CM) format, RPG ≥200 mg/dL, HbA1c ≥7.0%, or use of DM medications.
while pharmacy data are recorded as national drug code,
prescription order date, and days’ supply. Height, weight,
Matched Controls
smoking status, and BP are continuously collected during
routine physician care. To complete each patient visit, the To account for a potential DM diagnostic bias during follow-
clinician is required to enter between 1 and 20 diagnoses in up, the pool of eligible controls consisted of patients with
the EMR. The KPNW organization provides online medical negative DM tests (values below the diagnostic threshold),
guidelines to assist clinicians in patient management for most where the index date for controls was the date of the negative
conditions, including hypertension. These guidelines also DM test. Incidence-density sampling was used to select 10
recommend lipid screening for men aged ≥35 years and controls per case, matching on date of the DM test
women aged ≥45 years. Fasting plasma glucose (FPG) tests (6 weeks), age at DM test (5 years) and date of cohort
are routinely ordered in conjunction with these lipid screening entry (6 weeks).
panels.
Drug Exposure
Design The primary exposures of interest were the dispensing of
Despite attempts to standardized glucose assessment drug(s) within the BB, TD, RASB, and CCB antihypertensive
(obtained at the time of lipid assessment), physician suspicion classes. Exposure was defined as an active days’ supply in the
for DM could lead to more glucose testing and DM diagnoses 30 days prior to index date. Prescriptions filled at index date
in high-risk or health-service-seeking individuals (detection were excluded.
bias). To account for this, and to capture a population
undergoing active monitoring for DM risk, we used glucose
tests to develop a nested case–control study. Included Statistical Analysis
patients were between the ages of 35 and 65, were enrolled We used conditional logistic regression analysis to compute
in KPNW for least 18 months between January 1997 and ORs and 95% CIs to evaluate associations between incident
December 2010, and had both prescription and medical DM and drug exposure. We estimated both the main effects
coverage. Patients entered the study cohort at the first and the interactions between antihypertensive drug classes
negative FPG test (<126 mg/dL) following a 6-month look- using the same statistical model. The drug–drug interaction
back period without evidence for manifest diabetes (based on term quantifies the excess (or reduced) risk beyond what

DOI: 10.1161/JAHA.113.000125 Journal of the American Heart Association 2

Downloaded from http://jaha.ahajournals.org/ by guest on March 28, 2015

Antihypertensive Classes and Incident Diabetes Cooper-DeHoff et al

ORIGINAL RESEARCH
would have been predicted from the combination of the (fluvastatin, lovastatin, pravastatin, simvastatin, atorvastatin,
individual effects of each drug. ORs for dual-antihypertensive or rosuvastatin); and atypical antipsychotics and antidepres-
therapy were calculated using the log odds of the parameter sants (selective serotonin reuptake inhibitors, selective nor-
estimates and interaction terms and the covariate matrix (see epinephrine reuptake inhibitors, or tricyclic antidepressants).
Appendix S1 for equation). Dual therapy ORs represent the Multiple imputation was used to impute missing values.
total risk of the drug combination, which includes the effect of The percentages of missingness for covariates were: HDL
each drug individually as well as any excess (or reduced risk) 12.4%, LDL 15.8%, triglycerides 10.3%, blood pressure 5.4%,
due to the combination of drugs. smoking status 53.9%, and BMI 12.2%. All analyses were
We considered as covariates gender, baseline age and FPG, performed using SAS version 9.2, and P values <0.05 were
as well as smoking status, lipid levels (including HDL, LDL, considered statistically significant.
and triglycerides), systolic and diastolic blood pressure (BP),
body mass index (BMI) and CV disease history defined as
in- or outpatient visit with diagnosis of cerebrovascular Results
disease (ICD9-CM 430-438), myocardial infarction (ICD9-CM Our cohort included 134 967 nondiabetic patients, with
410, 412), congestive heart failure (ICD9-CM 428), coronary 412 604 glucose tests eligible for matching (Figure 1). From
artery disease (ICD9-CM 414), or peripheral vascular disease this cohort we identified 9097 patients who developed DM
(ICD9-CM 441, 443.9, 785.4, V43.4), all of which were (cases), which were matched to 90 495 controls. Cases
assessed during the 6 months preceding cohort entry. To tended to be male with higher baseline FPG, BP, BMI,
account for other medication use that could affect DM triglycerides and lower HDL, all characteristics consistent with
manifestation, we also included drug covariates defined an adverse metabolic phenotype (Table 1). Over 80% (7823)
during the 30 days prior to index date, including oral of cases were determined based on either FPG or RPG values.
corticosteroids (prednisone, dexamethasone, hydrocortisone, Of the 8207 DM cases with data available 1 year after DM
triamcinolone, methylprednisolone, or prednisolone); statins diagnosis, we were able to confirm diagnosis in 6258 (69%)

Patients enrolled in KPNW

N=475,000

Data Extraction Requirements

1. Enrolled in KPNW between January 1997 to
December 2010
2. FPG < 126mg/dl between ages 35-65
3. No evidence for DM during six months prior to FPG
Non diabetic patients in (by medication, ICD-9-CM, or laboratory tests)
KPNW dataset with 4. Greater than one year of enrollment post-FPG
FPG <126mg/dl 5. Medication and healthcare coverage
N=134,967
N=~340,000 Excluded

Glucose tests with non-

diabetic glucose levels
N=412,604

Diabetes Cases Matched Controls

N=9,097 N=90,495

Figure 1. Flow diagram of Kaiser Permanente Northwest (KPNW) members included in the analysis. FPG indicates fasting plasma glucose; DM,
diabetes mellitus; ICD-9-CM, International Classification of Disease—9th Revision—Clinical Modification.

DOI: 10.1161/JAHA.113.000125 Journal of the American Heart Association 3

Downloaded from http://jaha.ahajournals.org/ by guest on March 28, 2015

Antihypertensive Classes and Incident Diabetes Cooper-DeHoff et al

ORIGINAL RESEARCH
Table 1. Baseline Characteristics of Cases and Controls

Cases Controls

Patients, n 9097 90 495

Enrollment in KPNW system, y 4.5 (3.0) 4.5 (3.0)
Age, y
Median (IQR) 51 (45 to 56) 51 (45 to 56)
35 to 45, n (%) 2531 (27.0) 24 447 (27.8)
45 to 55, n (%) 4102 (45.0) 40 742 (45.1)
55 to 65, n (%) 2464 (28.0) 25 306 (27.1)
Male% 55.2 44.8
FPG, mg/dL 104 (12.5) 96 (11.2)
Blood pressure, mm Hg
Systolic 134 (16.5) 129 (16.8)
Diastolic 83 (9.8) 80 (10.0)
2
BMI, kg/m 34 (7.4) 30 (7.1)
Lipid panel, mg/dL
HDL 45 (12.8) 51 (12.2)
LDL 130 (36.9) 130 (38.0)
TRI 225 (224.4) 179 (179.3)
Ever smoke, n (%) 3548 (39.0) 34 841 (38.5)
Corticosteroid use*, n (%) 184 (2.0) 2423 (2.7)
Statin use*, n (%) 1791 (19.7) 20 377 (22.5)
†
CV disease , n (%) 515 (5.7) 6028 (6.7)
Diabetes diagnosis
FPG, n (%) 6523 (71.7)
FPG, mg/dL 155 (52.0)
RPG, n (%) 1300 (14.3)
RPG, mg/dL 281 (115.0)
New antidiabetic drug use, n (%) 655 (7.2)
HbA1c%, n (%) 619 (6.8)
HbA1c% 7.7 (1.1)

Values are mean (standard deviation) unless otherwise indicated. KPNW indicates Kaiser Permanente Northwest; IQR, intraquartile range; FPG, fasting plasma glucose; BMI, body mass
index; HDL, high-density lipoprotein; LDL, low-density lipoprotein; TRI, triglycerides; RPG, random plasma glucose; HbA1c, hemoglobin A1c.
*Corticosteroid use included any dispensed prescription for oral prednisone, dexamethasone, hydrocortisone, triamcinolone, methylprednisolone, or prednisolone; statin use included any
dispensed prescription for pravastatin, lovastatin, simvastatin, atorvastatin, rosuvastatin, or fluvastatin.
†
CV disease was defined as diagnosis of stroke, myocardial infarction, congestive heart failure, coronary artery disease, or peripheral vascular disease before cohort entry.

based on a second measured glucose value or HbA1c or were no drug–drug interactions that significantly increased
continued use of diabetes medication(s). DM risk beyond the expected aggregate effect of the single
Drug exposure, by antihypertensive drug class or class drug classes. The combination of TD+RASB or BB+RASB
combination, and risk for incident DM is summarized in resulted in significant interactions that had lower DM risk
Table 2 for cases and controls. Cases were more likely to be than would be observed from either the TD or BB alone. On
exposed to BB, TD, and RASB, overall and as part of a the other hand, the combination of TD+BB and TD+CCB
combination antihypertensive regimen. The top right quadrant resulted in the fully combined diabetogenic risk of each agent,
of Table 2 describes the risk for DM based on exposure to with adjusted ORs for DM of 1.99, 95% CI 1.80 to 2.20, and
each of the 4 drug classes individually. The lower right 1.52, 95% CI 1.28 to 1.82, respectively. A forest plot of the
quadrant of Table 2 describes the risk for DM for each of the risk for DM for each drug class, individually and in combina-
possible combinations of antihypertensive drug classes. There tion with the other classes is shown in Figure 2.

DOI: 10.1161/JAHA.113.000125 Journal of the American Heart Association 4

Downloaded from http://jaha.ahajournals.org/ by guest on March 28, 2015

Antihypertensive Classes and Incident Diabetes Cooper-DeHoff et al

ORIGINAL RESEARCH
Table 2. Individual and Combined Effects of the Antihypertensive Drug Classes on the Risk for Diabetes

OR (95% CI)

Drug Use Cases (n=9097) Controls (n=90 495) Crude Adjusted*

Individual
None† 50.4 58.7 1.00 1.00
TD 21.5 14.3 1.64 (1.56 to 1.73) 1.54 (1.41 to 1.68)
BB 28.5 23.3 1.32 (1.26 to 1.38) 1.19 (1.11 to 1.28)
CCB 8.2 7.7 1.07 (0.98 to 1.15) 1.07 (0.93 to 1.23)
RASB 22.2 20.7 1.10 (1.04 to 1.16) 0.99 (0.91 to 1.07)
Combination Adjusted Interaction Term*‡ Adjusted Effect for Class Combination
§
No combination 77.1 81.7 1.00 1.00
TD+BB 10.1 5.7 1.09 (0.97 to 1.22) 1.99 (1.80 to 2.20)
TD+CCB 3.0 2.1 0.93 (0.78 to 1.11) 1.52 (1.28 to 1.82)
TD+RASB 8.3 6.6 0.71 (0.63 to 0.80) 1.08 (0.97 to 1.20)
BB+CCB 3.6 3.6 0.85 (0.71 to 1.00) 1.07 (0.90 to 1.27)
BB+RASB 9.7 9.0 0.84 (0.74 to 0.94) 0.98 (0.89 to 1.09)
RASB+CCB 3.6 3.7 0.93 (0.79 to 1.10) 0.98 (0.83 to 1.15)

Values are percentages unless otherwise indicated. TD indicates thiazide diuretics; BB, beta-blockers; CCB, calcium channel blockers; RASB, renin–angiotensin system blockers; OR, odds
ratio; CI, confidence interval; FPG, fasting plasma glucose; HDL, high-density lipoprotein; LDL, low-density lipoprotein; BP, blood pressure; BMI, body mass index; CV, cardiovascular.
*Adjusted for potential confounders including gender, baseline age and FPG, as well as smoking status, lipid levels, including HDL, LDL, and triglycerides, systolic and diastolic BP, BMI,
glucose altering drug use (corticosteroids, antidepressants, antipsychotics, and statins) and CV disease.
†
Patients who were exposed to none of the drug classes; these patients constituted the reference group for the individual drug analysis.
‡
Estimated excess or reduced risk of exposure to the combination of the 2 drug classes beyond the risk associated with exposure to each drug class individually (the risks of the individual
drug classes appear in the top half of the table).
§
Patients who were exposed to none of the drug combinations; these patients constituted the reference group for the combination drug analysis.

In a sensitivity analysis restricted to DM cases with a incident diabetes as lowest for RASB (ACE inhibitors and
subsequent positive DM test, we observed similar results with ARBs), followed by CCBs, which appear neutral, and highest
slightly higher estimates for the diabetogenic risk of TD or BB for BB and TD.13 Our data from the KPNW population are
(Table 3). We also observed similar DM risk for combination consistent with the findings in this meta-analysis of RCTs,
of TD+BB, OR 2.10, 95% CI 1.86 to 2.37, and TD+CCB, OR suggesting that incidence of DM following exposure to
1.62, 95% CI 1.31 to 2.01. antihypertensive medications is generalizable to a much
broader, real-life population.
The onset of alterations in glucose after exposure to TD or
Discussion BB have been reported to occur within 9 weeks of treatment
To our knowledge, this is the first large, population-based initiation,17 and to continue with ongoing exposure.18 The
study (over 130 000 individuals and over 410 000 available European Guidelines on cardiovascular disease prevention in
glucose tests) to investigate associations between different clinical practice (Version 2012), for the first time explicitly
combinations of antihypertensive drugs on risk for DM. Our state that BB and TD are not recommended in hypertensive
results indicate that treatment of hypertension with the patients with multiple metabolic risk factors because of an
combination of TD+BB was associated with significantly increased risk for incident DM. This recommendation is
increased risk for DM, suggesting caution should be considered Class III (harmful), and is associated with the
exercised when prescribing TD+BB combination therapy in highest level of evidence (A), suggesting an increasing level of
individuals at risk for DM. Additionally, we found that awareness and significance of antihypertensive-related
interactions for combinations of TD or BB with a RASB were adverse metabolic effects.19 Our data, showing increased
negative with regard to risk for DM, suggesting RASB are risk with the TD + BB combination, are consistent with these
beneficial second line agents in those treated with a BB or recent guidelines.
TD. Importantly, these results remained consistent after Treatment with RASBs, including ramipril20 and valsar-
sensitivity analysis. tan,21 has also been investigated with regard to DM
A recent network meta-analysis of hypertension clinical prevention. While treatment with ramipril has been associated
trials ranked the association of antihypertensive agents with with a significant increase in regression to normoglycemia, it

DOI: 10.1161/JAHA.113.000125 Journal of the American Heart Association 5

Downloaded from http://jaha.ahajournals.org/ by guest on March 28, 2015

Antihypertensive Classes and Incident Diabetes Cooper-DeHoff et al

ORIGINAL RESEARCH
Figure 2. Adjusted odds ratio and 95% confidence interval for risk of diabetes among members in the Kaiser Permanente Northwest database
who were prescribed (A) thiazide diuretics (TD), alone or in combination; (B) b blockers (BB), alone or in combination; (C) calcium channel blockers
(CCB), alone or in combination; and (D) renin–angiotensin system blockers (RASB), alone or in combination. Members exposed to none of the drug
classes constitute the reference group.

has not been shown to significantly reduce risk for DM in tensive and Lipid Lowering Treatment to Prevent Heart Attack
patients with metabolic syndrome.20 In patients with impaired Trial (ALLHAT) study, which observed increased risk for
fasting glucose, valsartan was associated with a significant incident DM in patients treated with chlorthalidone,25 did not
14% decreased risk for diabetes.21 While the importance of observe excess morbidity or mortality associated with
DM prevention continues to be stressed,22 and screening incident diabetes.26 However, follow-up in the ALLHAT
high-risk populations is cost effective,23 DM prevention Extension Study may have been insufficient to adequately
guidelines and reviews do not give consideration to raising assess long-term adverse outcomes, and was not set up to
awareness of pharmacotherapy with known diabetogenic risk, assess microvascular complications. It is known that the
like TD and BB, or make recommendations for alteration of deadly, debilitating, and costly complications of DM do not
therapy where appropriate to reduce that risk. Our data found appear immediately after disease onset. Complications typ-
negative interactions in combination therapy including RASB ically emerge a decade or more later, and importantly, the
and either BB or TD, suggesting these drugs may offset some duration and extent of hyperglycemia predict complications.22
DM risk with BB or TD monotherapy, and that patients treated Follow-up in most clinical trials is insufficient to fully elucidate
TD+BB would benefit from routine DM screening. the long-term complications that may arise and additional
Whether DM that develops as a result of exposure to drugs research in this area is warranted. In the meantime, in the
with dysmetabolic effects has the same adverse conse- absence of solid clinical trial evidence, caution seems
quences as DM that develops from other etiologies has been warranted when equally effective therapeutic alternatives to
the subject of much debate.24 Most recently, the Antihyper- BB+TD combinations exist.

DOI: 10.1161/JAHA.113.000125 Journal of the American Heart Association 6

Downloaded from http://jaha.ahajournals.org/ by guest on March 28, 2015

Antihypertensive Classes and Incident Diabetes Cooper-DeHoff et al

ORIGINAL RESEARCH
Table 3. Individual Effects of the Antihypertensive Drug Classes on the Risk for Diabetes in Patients With at Least 2 Tests
Indicating Diagnosis of Diabetes

OR (95% CI)

Drug Use Cases (n=6258) Controls (n=62 216) Crude Adjusted*

Individual
None† 49.4 59.2 1.00 1.00
TD 22.1 14.0 1.73 (1.63 to 1.85) 1.64 (1.47 to 1.83)
BB 28.9 23.0 1.36 (1.28 to 1.44) 1.26 (1.15 to 1.37)
CCB 7.8 7.7 1.02 (0.93 to 1.13) 1.02 (0.85 to 1.21)
RASB 22.9 20.5 1.15 (1.08 to 1.22) 1.05 (0.95 to 1.16)
‡
Combination Adjusted Interaction Term* Adjusted Effect for Class Combination
§
No combination 76.5 81.9 1.00 1.00
TD+BB 10.2 5.7 1.02 (0.88 to 1.17) 2.10 (1.86 to 2.37)
TD+CCB 2.9 2.1 0.97 (0.78 to 1.21) 1.62 (1.31 to 2.01)
TD+RASB 8.5 6.5 0.66 (0.57 to 0.76) 1.13 (0.99 to 1.28)
BB+CCB 3.4 3.7 0.81 (0.65 to 0.99) 1.03 (0.83 to 1.27)
BB+RASB 9.8 8.9 0.80 (0.69 to 0.92) 1.05 (0.93 to 1.18)
RASB+CCB 3.4 3.7 0.93 (0.75 to 1.14) 0.99 (0.81 to 1.21)

Values are percentage unless otherwise indicated. TD indicates thiazide diuretics; BB, beta-blockers; CCB, calcium channel blockers; RASB, renin–angiotensin system blockers; OR, odds
ratio; CI, confidence interval; FPG, fasting plasma glucose; HDL, high-density lipoprotein; LDL, low-density lipoprotein; BP, blood pressure; BMI, body mass index; CV, cardiovascular.
*Adjusted for potential confounders including gender, baseline age and FPG, as well as smoking status, lipid levels, including HDL, LDL, and triglycerides, systolic and diastolic BP, BMI,
glucose altering drug use (corticosteroids, antidepressants, antipsychotics and statins) and CV disease.
†
Patients who were exposed to none of the drug classes; these patients constituted the reference group for the individual drug analysis.
‡
Estimated excess or reduced risk of exposure to the combination of the 2 drug classes beyond the risk associated with exposure to each drug class individually (the risks of the individual
drug classes appear in the top half of the table).
§
Patients who were exposed to none of the drug combinations; these patients constituted the reference group for the combination drug analysis.

Our study has several strengths. First, we used a clinical ethnicity were not systematically collected and thus are not
practice data set (KPNW), which is a large primary care reliably available in the KPNW dataset. It is estimated that only
database containing longitudinal data on patients’ medical a small percentage of the KPNW population are nonwhite, and
history, and thus we were able to adjust for several important thus our findings require replication in other race/ethnic
potential confounders, including baseline glucose blood groups. Although we had access to BMI, we did not have waist
pressure, cholesterol, smoking status and CV disease. circumference data, which precluded our ability to determine
Second, drug exposure was time varying as a result of the the presence of metabolic syndrome. However, our study did
risk set sampling method used to select controls. Finally, we replicate known associations between antihypertensive mono-
did sensitivity analyses, which, overall, produced results therapy and DM risk based on RCTs,13 and reassures the
consistent with those of the primary analysis. appropriateness in using the KPNW cohort to assess drug-
There are also some limitations worthy of mention. First, the induced DM, after accounting for available clinical character-
identification of diabetes may have been subject to some istics. Fourthly, use of pharmacy dispensing data to define drug
misclassification. We relied on a single positive glucose test to exposure does not allow for assessment of duration of drug
establish incident DM, which may not reflect permanent exposure, nor does it guarantee patient adherence; however,
dysglycemia. However, our sensitivity analysis, expanding the any resulting bias from drug exposure misclassification would
diagnostic requirement to include a subsequent test, con- likely result in a bias toward reduced associations between
firmed our results. Second, based on the nature of an drug exposure and DM onset. Finally, the state-of-the-art EMR
observational study design, residual confounding by indication and clinical decision support within KPNW, in addition to focus
and disease severity may be present. There could also be on a privately insured and predominantly white population, may
confounding by contraindication (eg, a physician refraining reduce study generalizability.
from prescribing TDs to a patient at a high risk of developing In conclusion, we found that antihypertensive regimens
diabetes), which could have biased the results toward the null. composed of either a TD or a BB were associated with
Third, black race and Hispanic ethnicity have been associated increased risk for development of DM in the KPNW popula-
with increased risk for diabetes,27 however, data on race/ tion. While the long-term implications of drug-associated

DOI: 10.1161/JAHA.113.000125 Journal of the American Heart Association 7

Downloaded from http://jaha.ahajournals.org/ by guest on March 28, 2015

Antihypertensive Classes and Incident Diabetes Cooper-DeHoff et al

ORIGINAL RESEARCH
diabetes are unclear, our observation that the DM risk for 8. Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL, Jones
DW, Materson BJ, Oparil S, Wright JT, Roccella EJ; Joint National Committee
TD+BB combinations was stacked, with the full diabetogenic on Prevention, Detection, Evaluation, and Treatment of High Blood
effect of each of the drugs being realized, suggests that Pressure; National Heart, Lung, and Blood Institute; National High Blood
Pressure Education Program Coordinating Committee. The seventh report
TD+BB combination should be avoided in cases where of the Joint National Committee on prevention, detection, evaluation,
and treatment of high blood pressure: the JNC 7 report. JAMA.
alternative combination regimens with similar BP lowering 2003;289:2560–2572.
efficacy are available. Conversely, treatment with a RASB in 9. 2010 Top 200 Generic Drugs by Total Prescriptions: 2011. http://drugtopics.
modernmedicine.com/. Accessed June 2, 2013.
combination with a BB or TD resulted in drug–drug interac-
10. 2010 Top 200 Brand Drugs by Total Prescription: 2011. http://drugtopics.
tions that were negative, suggesting RASB containing com- modernmedicine.com/. Accessed June 2, 2013.
bination may be preferred in those at increased risk for DM. 11. Bangalore S, Parkar S, Grossman E, Messerli FH. A meta-analysis of 94,492
Further research is needed to confirm our findings regarding patients with hypertension treated with beta blockers to determine the risk of
new-onset diabetes mellitus. Am J Cardiol. 2007;100:1254–1262.
association of antihypertensive combination therapy and DM 12. Taylor EN, Hu FB, Curhan GC. Antihypertensive medications and the risk of
risk. incident type 2 diabetes. Diabetes Care. 2006;29:1065–1070.
13. Elliott WJ, Meyer PM. Incident diabetes in clinical trials of antihypertensive
drugs: a network meta-analysis. Lancet. 2007;369:201–207.
14. Kit BK, Ogden CL, Flegal KM. Prescription medication use among normal
weight, overweight, and obese adults, united states, 2005–2008. Ann
Sources of Funding Epidemiol. 2012;22:112–119.
This study was supported, in part, by the NIH National Heart 15. Nichols GA, Hillier TA, Brown JB. Normal fasting plasma glucose and risk of
type 2 diabetes diagnosis. Am J Med. 2008;121:519–524.
Lung and Blood institute (K23HL086558) and NIH National 16. Association AD. Standards of medical care in diabetes—2010. Diabetes Care.
Institute of General Medicine Sciences (2U01 GM074492) 2010;33(suppl 1):S11–S61.
(Dr. Cooper-DeHoff). 17. Cooper-DeHoff RM, Wen S, Beitelshees AL, Zineh I, Gums JG, Turner ST,
Gong Y, Hall K, Parekh V, Chapman AB, Boerwinkle E, Johnson JA.
Impact of abdominal obesity on incidence of adverse metabolic effects
associated with antihypertensive medications. Hypertension. 2010;55:61–
68.
Disclosures
18. Lind L, Pollare T, Berne C, Lithell H. Long-term metabolic effects of
Dr Nichols reports that he is employed by Kaiser Permanente antihypertensive drugs. Am Heart J. 1994;128:1177–1183.

Northwest. Dr Bird reports that he is employed by the Food 19. European Guidelines on cardiovascular disease prevention in clinical practice
(version 2012): the Fifth Joint Task Force of the European Society of
and Drug Administration/Center for Drug Evaluation and Cardiology and other societies on cardiovascular disease prevention in clinical
practice (constituted by representatives of nine societies and by invited
Research (FDA/CDER). This study represents the views of the experts). Eur J Prev Cardiol. 2012;19:585–667.
authors and not those of FDA/CDER. This study received no 20. Investigators DT, Bosch J, Yusuf S, Gerstein HC, Pogue J, Sheridan P,
Dagenais G, Diaz R, Avezum A, Lanas F, Probstfield J, Fodor G, Holman RR.
funding from Kaiser Permanente Northwest or from the FDA. Effect of ramipril on the incidence of diabetes. N Engl J Med. 2006;
Drs Cooper-DeHoff, Winterstein, and Delaney report no 355:1551–1562.
conflicts of interest. 21. Group NS, McMurray JJ, Holman RR, Haffner SM, Bethel MA, Holzhauer B, Hua
TA, Belenkov Y, Boolell M, Buse JB, Buckley BM, Chacra AR, Chiang FT,
Charbonnel B, Chow CC, Davies MJ, Deedwania P, Diem P, Einhorn D, Fonseca
V, Fulcher GR, Gaciong Z, Gaztambide S, Giles T, Horton E, Ilkova H, Jenssen T,
References Kahn SE, Krum H, Laakso M, Leiter LA, Levitt NS, Mareev V, Martinez F,
Masson C, Mazzone T, Meaney E, Nesto R, Pan C, Prager R, Raptis SA, Rutten
1. Economic costs of diabetes in the U.S. in 2007 Diabetes Care. 2008;31:596– GE, Sandstroem H, Schaper F, Scheen A, Schmitz O, Sinay I, Soska V, Stender
615. S, Tamas G, Tognoni G, Tuomilehto J, Villamil AS, Vozar J, Califf RM. Effect of
2. Centers for Disease Control and Prevention. National Diabetes Fact Sheet: valsartan on the incidence of diabetes and cardiovascular events. N Engl J
National Estimates and General Information on Diabetes and Prediabetes in the Med. 2010;362:1477–1490.
United States, 2011. Atlanta, GA: U.S. Department of Health and Human 22. Fradkin JE, Roberts BT, Rodgers GP. What’s preventing US from preventing
Services, Centers for Disease Control and Prevention; 2011. type 2 diabetes? N Engl J Med. 2012;367:1177–1179.
3. Keenan NL, Rosendorf KA; Centers for Disease Control and Prevention. 23. Chatterjee R, Narayan KM, Lipscomb J, Jackson SL, Long Q, Zhu M, Phillips LS.
Prevalence of hypertension and controlled hypertension—United States, Screening for diabetes and prediabetes should be cost-saving in patients at
2005–2008. MMWR Surveill Summ. 2011;60(suppl):94–97. high risk. Diabetes Care. 2013; Feb 7 [Epub ahead of print] doi: 10.2337/
4. McNiece KL, Poffenbarger TS, Turner JL, Franco KD, Sorof JM, Portman RJ. dc12-1752.
Prevalence of hypertension and pre-hypertension among adolescents.
J Pediatr. 2007;150:640–644, 644.e641. 24. Karnes JH, Cooper-DeHoff RM. Antihypertensive medications: benefits of
blood pressure lowering and hazards of metabolic effects. Expert Rev
5. Guo F, He D, Zhang W, Walton RG. Trends in prevalence, awareness, Cardiovasc Ther. 2009;7:689–702.
management, and control of hypertension among United States adults, 1999
to 2010. J Am Coll Cardiol. 2012;60:599–606. American College of Cardiology 25. Major outcomes in high-risk hypertensive patients randomized to angiotensin-
Foundation. Published by Elsevier Inc; 2012. converting enzyme inhibitor or calcium channel blocker vs diuretic: the
antihypertensive and lipid-lowering treatment to prevent heart attack trial
6. Kearney PM, Whelton M, Reynolds K, Muntner P, Whelton PK, He J. Global (ALLHAT). JAMA. 2002;288:2981–2997.
burden of hypertension: analysis of worldwide data. Lancet. 2005;365:217–
223. 26. Barzilay JI, Davis BR, Pressel SL, Cutler JA, Einhorn PT, Black HR, Cushman WC,
Ford CE, Margolis KL, Moloo J, Oparil S, Piller LB, Simmons DL, Sweeney ME,
7. Roger VL, Go AS, Lloyd-Jones DM, Benjamin EJ, Berry JD, Borden WB, Bravata Whelton PK, Wong ND, Wright JT, Group ACR. Long-term effects of incident
DM, Dai S, Ford ES, Fox CS, Fullerton HJ, Gillespie C, Hailpern SM, Heit JA, diabetes mellitus on cardiovascular outcomes in people treated for hyperten-
Howard VJ, Kissela BM, Kittner SJ, Lackland DT, Lichtman JH, Lisabeth LD, sion: the ALLHAT diabetes extension study. Circ Cardiovasc Qual Outcomes.
Makuc DM, Marcus GM, Marelli A, Matchar DB, Moy CS, Mozaffarian D, 2012;5:153–162.
Mussolino ME, Nichol G, Paynter NP, Soliman EZ, Sorlie PD, Sotoodehnia N,
Turan TN, Virani SS, Wong ND, Woo D, Turner MB; American Heart Association 27. Ali M, Bullard K, Imperatore G, Barker L, Gregg E. Characteristics associated
Statistics Committee and Stroke Statistics Subcommittee. Heart disease and with poor glycemic control among adults with self-reported diagnosed
stroke statistics—2012 update: a report from the American Heart Association. diabetes—National Health and Nutrition Examination Survey, United States,
Circulation. 2012;125:e2–e220. 2007–2010. Morb Mortal Wkly Rep. 2012;61:32–37.

DOI: 10.1161/JAHA.113.000125 Journal of the American Heart Association 8

Downloaded from http://jaha.ahajournals.org/ by guest on March 28, 2015

 Antihypertensive Drug Class Interactions and Risk for Incident Diabetes: A Nested Case−
Control Study
Rhonda M. Cooper-DeHoff, Steven T. Bird, Gregory A. Nichols, Joseph A. Delaney and Almut G.
Winterstein

J Am Heart Assoc. 2013;2:e000125; originally published June 10, 2013;

doi: 10.1161/JAHA.113.000125
The Journal of the American Heart Association is published by the American Heart Association, 7272 Greenville Avenue,
Dallas, TX 75231
Online ISSN: 2047-9980

The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://jaha.ahajournals.org/content/2/3/e000125

Data Supplement (unedited) at:

http://jaha.ahajournals.org/content/suppl/2013/06/10/2.3.e000125.DC1.html

Subscriptions, Permissions, and Reprints: The Journal of the American Heart Association is an online only Open
Access publication. Visit the Journal at http://jaha.ahajournals.org for more information.

Downloaded from http://jaha.ahajournals.org/ by guest on March 28, 2015