ORIGINAL INVESTIGATION

HEALTH CARE REFORM

Comparative Effectiveness of 2 -!oc"ers
in H#pertensive $atients
Emily D. Parker, MPH, PhD; Karen L. Margolis, MD, MPH; Nicole K. Trower,
BS; Davi. !. Magi, MD, MPH; Hea"her M. Tavel, BS; S#san M. She""erly, MS;
P. Michael Ho, MD, PhD; Bi$ E. Swain, MS; Pa"rick !. %&'onnor, MD, MPH
Background: Randomized controlled trials have dem-
onstrated the efficacy of selected -blockers for prevent-ing
cardiovascular (CV) events in patients following myo-
cardial infarction (M) or with heart failure (!")# !owever$
the effectiveness of -blockers for preventing CV events in
patients with hypertension has been %ues-tioned recently$
but it is unclear whether this is a class effect#
Methods: &sing electronic medical record and health plan
data from the Cardiovascular Research 'etwork !y-
pertension Registry$ we compared incident M$ !"$ and
stroke in patients who were new -blocker users be-tween
())) and ())*# +atients had no history of CV dis-ease and
had not previously filled a prescription for a -blocker# Co,
proportional hazards regression was used to e,amine the
associations of atenolol and metoprolol tartrate with
incident CV events using both standard co-variate
ad-ustment (n . /() *01) and propensity score2 matching
methods (n . (( 34()#
Results: 5uring follow-up (median$ 4#( years)$ there were
34/0incidentM$3(0(incident!"$and3667incidentstroke
events# !azard ratios for M$ !"$ and stroke in metoprolol
tartrate users were )#** (*48 C$ )#*0-/#)()$ )#** (*48 C$
)#*6-/#)/)$ and )#** (*48 C$ )#*0-/#)()$ respectively# 9n
alternativeapproachusingpropensityscorematchingyielded
similar results in // /06 new metoprolol tartrate users$ who
were similar to // /06 new atenolol users with regard to
demographic and clinical characteristics#
Conclusions: :here were no statistically significant dif-
ferences in incident CV events between atenolol and meto-
prolol tartrate users with hypertension# ;arge registries
similar to the one used in this analysis may be useful for
addressing comparative effectiveness %uestions that are
unlikely to be resolved by randomized trials#
(rch )n"ern Me. *+,*;,-*.,/01,2+34
,2,*. P#5lishe online (#g#s" *-, *+,*.
oi1,+.,++,6archin"ernme.*+,*.2*-3
A%t&or Affi!iations'
!ealth+artners nstitute for
<ducation and Research$
Minneapolis$ Minnesota (5rs
+arker$ Margolis$ and
=>Connor and Ms :rower)?
nstitute for !ealth Research$
@aiser +ermanente Colorado$
5enver (5r Magid and Mss
:avel and Ahetterly)? 5enver
V9 Medical Center and 5ivision
of Cardiology$ Achool of
Medicine$ &niversity of
Colorado$ 5enver (5r !o)? and
5ivision of Research$ @aiser
+ermanente 'orthern California$
=akland (Mr Awain)#
I
' :!< :R<9:M<': =" !B+<R:<'-sion$ -blockers are widely used and are one of
the drug classes recommended as initial
treat-ment in hypertension guide-
lines based on reduction of morbidity and
mortality in placebo-controlled trials#
/-4
!owever$ following the publication of (
large trials that found that atenolol-based
regimens were less effective than other
antihypertensive drugs for preven-tion of
cardiovascular (CV) events in pa-tients
with hypertension$
6$0
the first-line status of
-blockers has increasingly been called into
%uestion#
3$1- //
9 recent meta-analysis
including these studies found that -blockers
were inferior to other agents primarily with
regard to stroke preven-tion$ but the authors
and editorialist pointed out that data on
-blockers other than atenolol were sparse
enough that it is unclear whether this
conclusion ap-plies to the entire -blocker
class#
*$/(
9 sec-
ond meta-analysis and editorial echoed
these findings and concerns#
//$/3
Cithin the drug class of -blockers$ there
are differences in pharmacokinetic
properties#
/7$/4
5ifferences in lipophilic-ity$
bioavailability$ and metabolism be-tween
atenolol and metoprolol tartrate may have
relevance for protecting the heart#
/)$//
See Invited Commentary
at end of article
5espite these differences$ it is unlikely
that they will be compared head to head
in a randomized controlled trial# :here-
fore$ we sought to compare the effective-
ness of ( commonly used -blockers$
using data from a hypertension registry
from 3 large integrated health care deliv-
ery systems# Ce compared the incidence
of myocardial infarction (M)$ stroke$ and
heart failure (!") in adult hyperten-
9RC! ':<R' M<5D V=; /0( ('=# /1)$ =C: 1$ ()/( CCC#9RC!':<R'M<5#C=M
()*+
,2*(2 American Me-ica! Association. A!! ri/&ts reserve-.
Downloaded From: http://archinte.jamanetwork.com/ on 04/29/2014
sive patients who were new users of atenolol and meto-
prolol tartrate#
METHO0S
ST102 SETTING AN0
REGISTR2 $O$1LATION
:his report is derived from the !ypertension Registry of the
Cardiovascular Research 'etwork (CVR')# :he registry in-
cludes all adult patients identified as having hypertension be-
tween ())) and ())* at 3 large integrated health care delivery
systemsE !ealth+artners of Minnesota$ @aiser +ermanente
Colo-rado$ and @aiser +ermanente 'orthern California#
<lectronic data on longitudinal blood pressure (F+)
measurements$ pre-scription drugs$ laboratory test results$
diagnoses$ and health care utilization were available from
electronic health records and administrative databases at all
sites# 5ata from each of the health plans were restructured into
a common$ standardized format with identical variable names$
definitions$ labels$ and coding#
Ce defined hypertension using criteria adapted from pre-vious
CVR' studies
/6-()
based on outpatient F+ readings$ di-agnostic
codes from outpatient and hospital records$ phar-macy
prescriptions$ and laboratory results# +atients entered the registry
on the date they first met / (or more) of the following criteriaE (/)
( consecutive elevated F+ measurements (ie$ sys-tolic F+ GAF+H
/7) mm !g andDor diastolic F+ G5F+H *) mm !g$ or /3)D1) mm
!g in the presence of diabetes melli-tus or chronic kidney disease
GC@5H)? (() ( diagnostic codes for hypertension ()n"erna"ional
'lassi7ica"ion o7 Diseases, Nin"h 8evision, 'linical Moi7ica"ion
G)'D494'MH code 7)/#,-7)4#,) recorded on separate dates? (3) /
diagnostic code for hyper-tension plus prescription for an
antihypertensive medication? or (7) / elevated F+ measurement
plus / diagnostic code for hypertension# Flood pressure readings
from emergency and ur-gent care settings were e,cluded because
they were found to be consistently higher than other ambulatory
measurements in the same patients in similar periods# :o confirm
that the al-gorithms designed to identify hypertensive patients
were valid and that the analytic data accurately reflected the
source data$ we conducted a review of 74) randomly selected
medical charts (/4) from each site)# Ce confirmed that
hypertension was in fact incident on the date assigned by the
algorithm in *68 of cases$ and agreement on F+ values between
the electronic da-tabase and medical chart records was *18#
VARIALES 1SE0 IN ANAL2SIS
77/#0$ 773#*$ 777#)$ and 777#()? and congestive !" ()'D494
'M diagnosis codes 7(1#,,$ 7)(#,,$ and 3*1#*/)# ncident M
()'D494'M code 7/)#,,)$ !"$ and stroke events were de-fined
using the primary )n"erna"ional 'lassi7ica"ion o7 Diseases,
Nin"h 8evision ()'D49) codes from a discharge from an inpa-
tient stay#
=ther comorbidities included in the analysis were diabetes
mellitus$ C@5$ and lipid disorders# 5iabetes was defined by
(/) ( outpatient diagnoses or / primary inpatient discharge
diag-nosis of diabetes mellitus ()'D494'M code (4)#,)? (() a
pre-scription for any antidiabetic medication other than
metfor-min or thiazolidinediones? (3) a prescription for
metformin or a thiazolidinedione plus a diagnosis of diabetes?
or (7) a he-moglobin 9/c value higher than 08 or ( fasting
plasma glu-cose values of /(6 mgDd; or higher (to convert to
millimoles per liter$ multiply by )#)444) on separate dates#
Chronic kid-ney disease was defined by (/) ( consecutive
serum creatinine values that yield estimated glomerular
filtration rates lower than 6) m;Dmin or (() an )n"erna"ional
'lassi7ica"ion o7 Diseases, Nin"h 8evision ()'D49) diagnostic
code for C@5 ()'D494'M codes 414#/-414#*)# ;ipid disorders
were identified by )'D494'M codes (0(#,#
ST102 $O$1LATION
Ce used a new user design$ which restricts the analysis to per-
sons under observation at the start of the current course of treat-
ment#
(/
:he study population included all patients /1 years or
older with hypertension during ())) through ())*$ who were
started on therapy with either atenolol or metoprolol tartrate after
the date of first diagnosis with no prior use of any -blocker for at
least /( months (n . /*3 /(3)# +revious use of any other class of
antihypertensive drug was not an e,clusion# +rescrip-tion
databases were searched as far back as /**6 or to health plan
enrollment if that occurred after /**6# =ther -blockers$ including
metoprolol succinate$ were not used fre%uently enough during the
years of the study to be included in the analysis# Ce e,cluded
pregnant women (n . 376)# n addition$ we e,cluded 76 1)*
patients who had evidence of CV5 before starting therapy with
atenolol or metoprolol tartrate# :hese e,clusions were based on
the previously described CV diagnosis codes as well as pro-cedure
codes for cardiac bypass surgery ('#rren" Proce#ral Ter4
minology G'PTH codes 334/)-334(3 and 33433-33436) and per-
cutaneous coronary interventions ('PT codes *(*1)-*(**6)# :o
e,clude patients with suspected CV5$ we also e,cluded (7 **)
patients with a visit to cardiology specialist within the year prior to
starting the -blocker therapy$ leaving /() *01 patients for this
analysis (Figure)#
STATISTICAL ANAL2SIS
+atient age and se, were available for all patients from mem-
bership databases# RaceDethnicity was obtained from outpa-
tient registration data$ hospital discharge records$ member sat-
isfaction surveys$ and other research survey data sets and was
available for 148 of cohort members# Aystolic F+s and 5F+s
measured within ( months prior to the initiation of a -blocker
therapy and appro,imately 6 months (I6) days) after the ini-
tiation of a -blocker therapy were included# +harmacy rec-ords
were used to identify dates of treatment with -blockers and
other antihypertensive drug classes used within *) days of
starting the -blocker therapy#
Cardiovascular disease (CV5) was identified using diagno-
ses and procedure codes from inpatient and ambulatory rec-
ords# :hese included ischemic heart disease ()'D494'M diag-
nosis codes 7/)#,-7/7#,,)? stroke ()'D494'M diagnosis codes
73)#,,-737#,,$ 736#,,$ 14(#)$ 14(#(# 14(#7$ and 143#))? pe-
ripheral vascular disease ()'D494'M diagnosis codes 77/#3-
9ll statistical analyses were completed using A9A version *#(
(A9A nstitute nc)# Faseline characteristics were compared be-
tween patients started on atenolol therapy vs metoprolol tar-trate
therapy using means and standard deviations for continu-ous
variables and percentages for categorical and binary variables#
Co, proportional hazards models were used to compare time with
outcome events between atenolol and metoprolol tar-trate# "ollow-
up time was computed in days from the day fol-lowing the first
dispensing of the new -blocker to the date of the first observed
outcome event$ termination of enrollment$ or 5ecember 3/$ ())*$
whichever occurred first# +atients who were lost to follow-up were
censored at the last point of con-tact# Multivariable models were
ad-usted for year of -blocker therapy initiation$ age$ se,$ number
of visits in the prior year$ AF+ at the start of -blocker therapy$ lipid
disorder$ diabetes mellitus$ C@5$ and use of other
antihypertensive medica-
9RC! ':<R' M<5D V=; /0( ('=# /1)$ =C: 1$ ()/( CCC#9RC!':<R'M<5#C=M
()*3
,2*(2 American Me-ica! Association. A!! ri/&ts reserve-.
Downloaded From: http://archinte.jamanetwork.com/ on 04/29/2014
193 123 New -blocker users with hypertension
18 years or older between 2001 and 2009
Exclusions:
46 809 History of CVD
24 990 Seen by cardiologist in last year
346 Pregnant women
120 978
Met inclusion criteria, included in multivariable regression analysis
91% Atenolol 9% Metoprolol tartrate
98 626 Unmatched
11 176 Atenolol users 11 176 Metoprolol tartrate users
Matched on
propensity score
22 352 Included in propensity scorematched analysis
Figure. Selection of patients for analyses. CVD indicates
cardiovascular disease.
tions# n a supplemental analysis of 61 11( patients in whom
we had follow-up F+ data$ we used linear regression to e,am-
ine the effect of atenolol and metoprolol tartrate on lowering
AF+ and 5F+ 6 months after the start of the -blocker therapy#
Fecause this is an observational study and patients were not
randomized to receive either treatment$ we also used alterna-tive
strategies to minimize confounding by indication# :o mini-mize
confounding by indication$ we also ran a conditional lo-gistic
regression matched on propensity score# 9 logistic model (which
included all the variables in Table 1 e,cept for 5F+) was used to
generate a propensity score for the probability of being prescribed
metoprolol tartrate# Ce then used a 4-digit greedy /E/ matching
algorithm
(3
to match metoprolol tartrate users to atenolol users
based on propensity score# 9fter con-ducting the propensity
matching$ there were ** 6(6 un-matched patients$ leaving (( 34(
matched patients for statis-tical analyses of adverse CV events and
/3 *)1 matched patients with 6-month follow-up F+s for the
analyses of F+ lowering# :he selection of patients for the analyses
is shown in the "igure# 9 second alternative strategy was to
conduct a sensitivity analy-sis e,cluding patients who had events
in the first /( months of follow-up so as to e,clude those with
CV5 not e,cluded by diagnosis codes or visits to a cardiologist
that could have had an impact on prescribing behavior#
RES1LTS
5uring the follow-up period (median$ 4#( years)$ there
were 34/0 incident Ms$ 3(0( incident !" hospitaliza-
tions$ and 3667 incident strokes# Multivariable Co, pro-
portional hazards regression yielded hazard ratios of )#**$
)#**$ )#**$ and )#*1 and narrow *48 confidence inter-vals
that included the null value for M$ !"$ stroke$ and any CV
event$ respectively (Table 2)# n the propen-sity score2
matched Co, proportional hazards models$ the hazard
ratios for M$ !"$ stroke$ and any CV event were virtually
identical to the multivariable results with nar-row *48
confidence intervals that included the null value (:able ()#
n sensitivity analyses e,cluding patients who had events in
the first /( months of follow-up$ the haz-ard ratios were
virtually unchanged (data not shown)#
<stimates and standard errors of the supplemental
analysis of the F+-lowering effects of the ( -blockers in
the subgroup with follow-up measures are given in Table
3# n multivariable analysis of new -blocker us-ers$ at
baseline there were statistically significant differ-ences
between atenolol and metoprolol tartrate users in AF+s
(/71#4 and /74#7 mm !g$ respectively? P #))/) and 5F+s
(17#( and 1(#4$ respectively? P #))/)# 9t the 6-month
follow-up$ AF+s were /30#7 and /30#4 mm !g in the
atenolol- and metoprolol tartrate-treated pa-tients$
respectively (P . #1()# 9t 6 months$ 5F+s were 00#3 and
00#0 mm !g in the atenolol- and metoprolol tartrate2
treated patients$ respectively (P . #))4)# :here was no sta-
tistically significant difference in change in AF+ and a
small but statistically significant difference in change in
5F+ (4#* and 4#4 mm !g for atenolol and metoprolol tar-
trate$ respectively P . #))4)# :he propensity score2
matched analysis of F+ lowering had similar results when
comparing new atenolol and metoprolol tartrate users in
AF+ (/77#( and /73#3 mm !g$ respectively? P . #))0) and
5F+ (1/#3 and 1)#( mm !g$ respectively? P #))/)# 9t the
6-month follow-up$ there were no statistically sig-nificant
differences between atenolol and metoprolol tar-trate users
in AF+ or 5F+# n the propensity-matched model$ the mean
F+ lowering was slightly greater in aten-olol vs metoprolol
tartrate users (0#0 and 6#0 mm !g$ re-spectively? P . #)()#
9tenolol lowered 5F+ slightly more than metoprolol
tartrate (7#0 and 3#7 mm !g$ respec-tively? P #))/)#
COMMENT
:he baseline characteristics for this cohort of new
-blocker users are given in :able /# 9 total of /() *01
patients without history of CV5 events from the CVR'
!ypertension Registry initiated treatment with either
aten-olol or metoprolol tartrate between ())) and ())*#
5ur-ing this period atenolol was used in appro,imately
/)-fold more patients than metoprolol tartrate# +atients
who filled a prescription for metoprolol tartrate tended
to be older$ have a government insurance payer$ and
have more ambulatory visits# Metoprolol tartrate users
had slightly lower AF+s and 5F+s at the start of
-blocker treat-ment$ were more likely to be using other
antihyperten-sive medications$ and more often had lipid
disorders$ dia-betes$ and C@5#
:he ob-ective of this study was to assess the compara-
tive effectiveness of ( -blockers$ atenolol and metopro-
lol tartrate$ in patients without a history of CV5# :o our
knowledge$ this study is among the first to address this
important clinical %uestion# n this retrospective cohort
study comparing patients initiating -blocker treat-ment
with either atenolol or metoprolol tartrate$ there were no
statistically significant differences in rates of in-cident
M$ !"$ or stroke after ad-usting for potential con-
founders# n addition$ there were no statistically signifi-
cant differences in AF+-lowering effects comparing
atenolol and metoprolol tartrate#
&ntil recently$ -blockers had been widely recom-
mended as first-line therapy for hypertension$
/-4
but many
of the trials supporting their use had given investigators
9RC! ':<R' M<5D V=; /0( ('=# /1)$ =C: 1$ ()/( CCC#9RC!':<R'M<5#C=M
()*4
,2*(2 American Me-ica! Association. A!! ri/&ts reserve-.
Downloaded From: http://archinte.jamanetwork.com/ on 04/29/2014
Table 1. Descriptive Characteristic of Patients Initiating -Blocker (BB) Use Between 2001 and 2009 (CVRN Hypertension Registry)
a
Total Matched on Propensity Score
(n = 120 978) (n = 22 352)
Atenolol Metoprolol Tartrate Atenolol Metoprolol Tartrate
Variable (n = 109 798) (n = 11 180) (n = 11 176) (n = 11 176)
Year of BB therapy initiation
2000 14 10 87 10
2001 11 6 7 6
2002 15 9 9 9
2003 15 10 10 10
2004 14 12 12 12
2005 12 15 16 15
2006 10 16 16 16
2007 8 14 14 14
2008 2 5 5 5
2009 2 4 4 4
Age, mean (SD), y 60.8 (13.0) 65.1 (13.6) 65.0 (13.4) 65.2 (13.6)
Age category, y
50 20 13 14 13
50-59 27 21 20 21
60-69 25 25 25 25
70-79 20 26 26 26
80 7 15 15 15
Male 43 44 43 44
Race/ethnicity
White 62 66 63 66
African American 10 11 12 11
Asian 9 7 8 8
Nonwhite Hispanic 1 1 1 1
Other/multiple/unknown 19 14 15 14
Median ambulatory visits in prior year 5.0 7.0 7.0 7.0
Insurance payer
Commercial 79 71 70 71
Government 21 29 28 29
SBP at start of BB therapy, mean (SD), mm Hg 148.5 (20.2) 144.0 (21.5) 144.8 (20.4) 144.0 (21.5)
DBP at start of BB therapy, mean (SD), mm Hg 84.5 (13.2) 82 (12.5) 82.1 (12.6) 80.9 (12.9)
Lipid disorder
b
32 42 47 42
Diabetes mellitus
c
21 31 30 30
Chronic kidney disease
d
12 28 28 28
ACE inhibitor or ARB use within 6 mo prior 36 46 47 46
to BB therapy initiation
CCB use within 6 mo prior to BB therapy initiation 14 23 17 23
Diuretic use within 6 mo prior to BB therapy initiation 51 49 54 49
Other antihypertensive medication use within 6 mo 7 13 9 13
prior to BB therapy initiation
Antihypertensive medications within 6 mo prior
to BB therapy initiation, No.
0 32 26 26 26
1 26 31 32 31
2 26 30 32 30
3 6 11 8 11
4 1 2 1 2
Abbreviations: ACE, angiotensin-converting enzyme; ARB, angiotensin receptor blocker; CCB, calcium channel blocker; CVRN, the
Cardiovascular Research Network.
aData are given as percentage of patients unless otherwise indicated. Percentages may not add to 100 because of rounding.
bLipid disorders were identified by International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM ) codes 272.x.
cDiabetes mellitus was defined by (1) 2 outpatient diagnoses or 1 primary inpatient discharge diagnosis of diabetes ( ICD-9-CM code 250.x); (2) prescription for
any antidiabetic medication other than metformin or thiazolidinediones; (3) prescription for metformin or a thiazolidinedione plus a diagnosis of diabetes; or
(4) hemoglobin A1c value higher than 7% or 2 fasting plasma glucose values of 126 mg/dL or higher (to convert to millimoles per liter, multiply by
0.0555)
on separate dates.
d
Chronic kidney disease was defined by (1) 2 consecutive serum creatinine values that yield estimated glomerular filtration rates lower than
60 mL/min when the Modification of Diet in Renal Disease equation
22
is applied or (2) an International Classification of Diseases, Ninth Revision
(ICD-9 ) diagnostic code for chronic kidney disease (ICD-9-CM codes 585.1-585.9).
the choice of using either a thiazide diuretic or -blocker
alone or in combination as Jconventional therapy#K :he
combination compared favorably against other antihy-
pertensive drugs classes for prevention of CV events#
/$(7
:he use of -blockers as a first-line therapy has recently
been challenged based on evidence of a weak effect on
stroke
(4
and the absence of an effect on coronary heart
disease
(4-(0
compared with placebo$ as well as inferiority
9RC! ':<R' M<5D V=; /0( ('=# /1)$ =C: 1$ ()/( CCC#9RC!':<R'M<5#C=M
()*5
,2*(2 American Me-ica! Association. A!! ri/&ts reserve-.
Downloaded From: http://archinte.jamanetwork.com/ on 04/29/2014
Table 2. Incident Cardiovascular (CV) Events Associated With Metoprolol Tartrate Compared With Atenolol
Multivariable
a
Cox Proportional Hazards Regression Propensity ScoreMatched
b
Cox Proportional Hazards Model
Variable No. of Events Person-years Hazard Ratio (95% CI) No. of Events Person-years Hazard Ratio (95% CI)
MI 3517 631 403 0.99 (0.97-1.01) 712 94 261 0.99 (0.97-1.02)
HF 3272 633 987 0.99 (0.97-1.01) 831 94 257 0.99 (0.96-1.01)
Stroke 3664 632 386 0.99 (0.97-1.01) 773 94 346 0.99 (0.97-1.02)
Any CV event 9353 616 028 0.98 (0.99-1.00) 2064 91 191 0.98 (0.95-1.00)
Abbreviations: HF, heart failure; MI, myocardial infarction.
aMultivariable model adjusted for year of -blocker therapy initiation, age, sex, number of visits in prior year, systolic blood pressure at start of -blocker therapy, lipid
disorder, diabetes mellitus, chronic kidney disease, and if using other antihypertensive medications.
bMatched on propensity score. Propensity score adjusted for year of -blocker therapy initiation, age, sex, number of visits in prior year, systolic blood
pressure at start of -blocker therapy, lipid disorder, diabetes, chronic kidney disease mellitus, and if using other antihypertensive medications.
Table 3. Comparison of Blood Pressure (BP)-Lowering Effects at 6 Months in New -Blocker Users
Multivariable
a
Linear Regression Propensity ScoreMatched
b
Linear Regression
BP Estimate (SE), mm Hg BP Estimate (SE), mm Hg
Atenolol Metoprolol Tartrate Atenolol Metoprolol Tartrate
Variable (n = 61 869) (n = 7013) P Value (n = 6907) (n = 7010) P Value
Baseline SBP and DBP
SBP 148.5 (0.25) 145.4 (0.33) .001 144.2(0.25) 143.3 (0.25) .007
DBP 84.2 (0.15) 82.5 (0.20) .001 81.3(0.15) 80.2 (0.15) .001
SBP and DBP at 6 mo after
-blocker therapy initiation
SBP 137.4 (0.23) 137.5 (0.30) .82 136.5(0.23) 136.6 (0.23) .85
DBP 77.3 (0.13) 77.7 (0.17) .005 76.6(0.13) 76.7 (0.13) .41
Change in SBP and DBP over
the 6 mo follow-up
SBP 9.8 (0.23) 9.8 (0.30) .82 7.7(0.29) 6.7 (0.29) .02
DBP 5.9 (0.13) 5.5 (0.17) .005 4.7(0.16) 3.4 (0.16) .001
Abbreviations: DBP, diastolic BP; SBP, systolic BP.
aMultivariable model adjusted for year of -blocker therapy initiation, age, sex, number of visits in prior year, SBP at start of -blocker therapy, lipid disorder, diabetes mellitus,
chronic kidney disease, and if using other antihypertensive medications.
bMatched on propensity score. Propensity score adjusted for year of -blocker therapy initiation, age, sex, number of visits in prior year, SBP at -blocker
therapy initiation, lipid disorder, diabetes mellitus, chronic kidney disease, and if using other antihypertensive medications.
compared with other treatments for total mortality$ coro-
nary heart disease$ and stroke#
6$0$(1
Meta-analyses and a
Cochrane review of recent trials that looked specifically
at -blockers used as monotherapy or as the first-line
drug in a stepped care approach concluded that the
evidence did not support use of -blockers as a first-line
therapy#
//
Fased on these findings$ recently issued
guidelines have relegated -blockers to third- or fourth-
line treatment for uncomplicated hypertension#
(*
-blockers differ in selectivity for the /- and (- and
-adrenergic receptors$ lipophilicity$ penetration across the
blood-brain barrier$ duration of action$ vasodilation
properties$ and type 3 antiarrhythmic activity#
/7$/4$3)
5if-
ferent types of -blockers may be indicated depending on
patient profiles and tolerances# Liven that most of the
evidence comes from trials where atenolol was the -blocker
used$
//
it is unclear if the observed effects of -blockers in
comparison with other antihypertensive medications are
due to properties of atenolol or the en-tire class of
-blockers# !owever$ there have been no trials comparing
the different subtypes of -blockers# Chile both atenolol and
metoprolol tartrate are both /-adrenergic receptors$ they
differ in lipophilicity$ bioavail-ability$ and
metabolism#
/)$//$3/
Metoprolol is lipid soluble
and tends to have highly variable bioavailability and a short
plasma half-life# n contrast$ atenolol is more water soluble$
shows less variance in bioavailability$ and has a longer
plasma half-life# 5espite these differences$ both drugs have
the effect of increasing vagal tone and causing a reduc-tion
in sympathetic outflow$ likely via peripheral -ad-renergic
blockade#
3($33
=ur findings that there are no dif-ferences
between atenolol and metoprolol tartrate in event rates and
effectiveness at F+ lowering in a cohort of adults without
prior CV events suggest that the unfavorable trial data with
atenolol may also apply to other -blockers#
9s with any observational study$ there are potential
limitations and caveats# Ce were unable to compare
aten-olol with any -blocker other than metoprolol
tartrate because of the low use of other agents in our
study popu-lation during the years of observation# :he
use of meto-prolol succinate$ a once-daily drug that may
have better adherence rates compared with twice-daily
metoprolol tartrate$ has been increasing owing to the
availability of generic versions in recent years$ but the
shift away from -blockers after ())0 may make
comparative effective-ness analyses more difficult#
Most importantly$ patients were not randomly as-signed
to treatment with either atenolol or metoprolol tar-
9RC! ':<R' M<5D V=; /0( ('=# /1)$ =C: 1$ ()/( CCC#9RC!':<R'M<5#C=M
()(*
,2*(2 American Me-ica! Association. A!! ri/&ts reserve-.
Downloaded From: http://archinte.jamanetwork.com/ on 04/29/2014
trate# :he decision on the part of the clinician to choose one
drug over another may be related to patient charac-teristics
associated with F+ control or CV risk or physi-cian
characteristics associated with differences in %uality of
care# :o reduce the potential bias related to confound-ing
by indication$ we took ( approachesE (/) we used a new
user design
(/$37
and restricted the sample to those pa-tients
with no evidence of diagnosed or suspected CV5
37
and (()
we used propensity score matching to ensure that patients
were comparable with regard to baseline covari-ates and
the probability of receiving each treatment#
5espite these robust methods$ no observational study
can rule out the impact of unmeasured confounding# f
unmeasured variables associated with poorer prognosis
were more common in patients prescribed metoprolol
tar-trate$ it could mask a beneficial effect of metoprolol#
Ce e,cluded patients who had seen a cardiologist in the
/( months prior to the initiation of -blocker therapy$ but
this strategy may have been insufficient to rule out sus-
pected CV5# !owever$ in a recent study using data from
one of the study sites$ we found no evidence of sus-
pected heart disease in audits of physician medical chart
notes in (7) patients lacking specific )'D49 codes for
heart disease (7/)-7/7 and 7()-7(*)#
34
=ther important
po-tential unmeasured confounders that are not available
in electronic medical records include behavioral or envi-
ronmental risk factors$ such as poor diet$ low level of
physical activity$ or e,posure to second-hand smoke$ al-
though we have no reason to believe that patients with
these risk factors would be more likely to be prescribed
metoprolol tartrate rather than atenolol#
n conclusion$ we found no differences in CV event
rates when comparing patients without a history of CV
events who were initiating treatment with either ateno-lol
or metoprolol tartrate# :hese findings suggest that hy-
pertension trial outcomes with atenolol may not relate to
unfavorable characteristics of this particular drug# :hese
results should be interpreted cautiously$ since there have
been no trials comparing these ( -blockers directly#
Accepte- for $%6!ication' Mune //$ ()/(#
$%6!is&e- On!ine' 9ugust (0$ ()/(# doiE/)#/))/
Darchinternmed#()/(#7(06
Correspon-ence' <mily 5# +arker$ M+!$ +h5$ !ealth-
+artners nstitute for <ducation and Research$ Fo, /4(7$
Mail Atop (////R$ Minneapolis$ M' 4477)-/4(7 (<mily
#5#+arkerN!ealthpartners#com)#
A%t&or Contri6%tions' S"#y conce:" an esign1 +arker$
Margolis$ and =>Connor# (c;#isi"ion o7 a"a1 Margolis$
:rower$ Magid$ :avel$ Ahetterly$ Awain$ and =>Connor#
(nalysis an in"er:re"a"ion o7 a"a1 +arker$ Margolis$ !o$
and =>Connor# Dra7"ing o7 "he man#scri:"1 +arker$ Mar-
golis$ :rower$ Awain$ and =>Connor# 'ri"ical revision o7
"he man#scri:" 7or im:or"an" in"ellec"#al con"en"1 +arker$
Magid$ :avel$ Ahetterly$ and !o# S"a"is"ical analysis1
+arker and :rower# %5"aine 7#ning1 Magid and
=>Connor# S"#y s#:ervision1 Margolis#
Financia! 0isc!os%re' 'one reported#
F%n-in/7S%pport' :his pro-ect was funded by grant
'!D '!;FD&/* !;)*//0* from the 'ational !eart$
;ung$ and Flood nstitute and subcontract to
!ealth+artners nstitute for <ducation and Research#
REFERENCES
1. Turnbull F; Blood Pressure Lowering Treatment Trialists
Collaboration. Effects of different blood-pressure-lowering regimens
on major cardiovascular events: results of prospectively-designed
overviews of randomised trials. Lancet. 2003; 362(9395):1527-1535.
2. Psaty BM, Smith NL, Siscovick DS, et al. Health outcomes
associated with an-tihypertensive therapies used as first-line agents:
a systematic review and meta-analysis. JAMA. 1997;277(9):739-745.
3. Messerli FH, Grossman E, Goldbourt U. Are -blockers efficacious as
first-line therapy for hypertension in the elderly? a systematic review.
JAMA. 1998;279 (23):1903-1907.
4. Mancia G, De Backer G, Dominiczak A, et al; The task force for the management
of arterial hypertension of the European Society of Hypertension; The task force
for the management of arterial hypertension of the European Society of Cardi-
ology. 2007 Guidelines for the management of arterial hypertension. Eur Heart J.
2007;28(12):1462-1536.
5. Chobanian AV, Bakris GL, Black HR, et al; National Heart, Lung, and Blood In-
stitute Joint National Committee on Prevention, Detection, Evaluation, and Treat-
ment of High Blood Pressure; National High Blood Pressure Education Program
Coordinating Committee. The Seventh Report of the Joint National Committee
on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: the
JNC 7 report. JAMA. 2003;289(19):2560-2572.
6. Dahlof B, Devereux RB, Kjeldsen SE, et al; LIFE Study Group.
Cardiovascular mor-bidity and mortality in the Losartan Intervention
For Endpoint reduction in hy-pertension study (LIFE): a randomised
trial against atenolol. Lancet. 2002; 359(9311):995-1003.
7. Dahlof B, Sever PS, Poulter NR, et al; ASCOT Investigators. Prevention
of car-diovascular events with an antihypertensive regimen of amlodipine
adding per-indopril as required versus atenolol adding
bendroflumethiazide as required, in the Anglo-Scandinavian Cardiac
Outcomes Trial-Blood Pressure Lowering Arm (ASCOT-BPLA): a
multicentre randomised controlled trial. Lancet. 2005;366 (9489):895-906.
8. Messerli FH, Beevers DG, Franklin SS, Pickering TG. beta-Blockers
in hypertension-the emperor has no clothes: an open letter to present
and prospective drafters of new guidelines for the treatment of
hypertension. Am J Hypertens. 2003; 16(10):870-873.
9. Lindholm LH, Carlberg B, Samuelsson O. Should beta blockers
remain first choice in the treatment of primary hypertension? a meta-
analysis. Lancet. 2005;366 (9496):1545-1553.
10. Ong HT. Beta blockers in hypertension and cardiovascular disease.
BMJ. 2007; 334(7600):946-949.
11. Wiysonge CS, Bradley H, Mayosi BM, et al. Beta-blockers for
hypertension. Coch-rane Database Syst Rev. 2007;(1):CD002003.
12. Beevers DG. The end of beta blockers for uncomplicated
hypertension? Lancet. 2005;366(9496):1510-1512.
13. Massie BM. Review: available evidence does not support the use of beta block-
ers as first line treatment for hypertension. Evid Based Med. 2007;12(4):112.
14. Reid JL. Optimal features of a new beta-blocker. Am Heart J.
1988;116(5, pt 2): 1400-1404.
15. Drayer DE. Lipophilicity, hydrophilicity, and the central nervous system
side ef-fects of beta blockers. Pharmacotherapy. 1987;7(4):87-91.
16. Ho PM, Zeng C, Tavel HM, et al. Trends in first-line therapy for
hypertension in the Cardiovascular Research Network Hypertension
Registry, 2002-2007. Arch Intern Med. 2010;170(10):912-913.
17. Magid DJ, Shetterly SM, Margolis KL, et al. Comparative effectiveness of angio-
tensin-converting enzyme inhibitors versus beta-blockers as second-line therapy
for hypertension. Circ Cardiovasc Qual Outcomes. 2010;3(5):453-458.
18. Schmittdiel J, Selby JV, Swain B, et al. Missed opportunities in cardiovascular
disease prevention? low rates of hypertension recognition for women at medi-
cine and obstetrics-gynecology clinics. Hypertension. 2011;57(4):717-722.
19. Selby JV, Lee J, Swain BE, et al. Trends in time to confirmation and recognition
of new-onset hypertension, 2002-2006. Hypertension. 2010;56(4):605-611.
20. Selby JV, Peng T, Karter AJ, et al. High rates of co-occurrence of hypertension,
elevated low-density lipoprotein cholesterol, and diabetes mellitus in a large
man-aged care population. Am J Manag Care. 2004;10(2, pt 2):163-170.
21. Ray WA. Evaluating medication effects outside of clinical trials: new-
user designs. Am J Epidemiol. 2003;158(9):915-920.
22. National Kidney Foundation. K/DOQI clinical practice guidelines for
chronic kid-ney disease: evaluation, classification, and stratification.
Am J Kidney Dis. 2002; 39(2)(suppl 1):S1-S266.
23. Parsons L. Reducing bias in a propensity score matched-pair sample using greedy matching
techniques. In: Proceedings of the Twenty-sixth Annual SAS Users Group International
Conference. Cary, NC: SAS Institute Inc; 2001:214-226.
9RC! ':<R' M<5D V=; /0( ('=# /1)$ =C: 1$ ()/( CCC#9RC!':<R'M<5#C=M
()((
,2*(2 American Me-ica! Association. A!! ri/&ts reserve-.
Downloaded From: http://archinte.jamanetwork.com/ on 04/29/2014
24. Neal B, MacMahon S, Chapman N; Blood Pressure Lowering Treatment
Trial-ists Collaboration. Effects of ACE inhibitors, calcium antagonists, and
other blood-pressure-lowering drugs: results of prospectively designed
overviews of ran-domised trials. Blood Pressure Lowering Treatment
Trialists Collaboration. Lancet. 2000;356(9246):1955-1964.
25. Coope J, Warrender TS. Randomised trial of treatment of
hypertension in el-derly patients in primary care. Br Med J (Clin Res
Ed). 1986;293(6555):1145-1151.
26. Medical Research Council Working Party. MRC trial of treatment of
mild hyper-tension: principal results. Br Med J (Clin Res Ed).
1985;291(6488):97-104.
27. The IPPPSH Collaborative Group. Cardiovascular risk and risk
factors in a ran-domized trial of treatment based on the beta-blocker
oxprenolol: the Interna-tional Prospective Primary Prevention Study in
Hypertension (IPPPSH). J Hypertens. 1985;3(4):379-392.
28. MRC Working Party. Medical Research Council trial of treatment of
hyper-tension in older adults: principal results. BMJ.
1992;304(6824):405-412.
29. National Institute for Health and Clinical Excellence. Hypertension: Clinical Man-
agement of Primary Hypertension in Adults. London, England: National Institute
for Health and Clinical Excellence; 2011. NICE Clinical Guideline 127.
30. Weber MA. The role of the new beta-blockers in treating cardiovascular
disease. Am J Hypertens. 2005;18(12, pt 2):169S-176S.
31. Wikstrand J, Kendall M. The role of beta receptor blockade in
preventing sudden death. Eur Heart J. 1992;13(suppl D):111-120.
32. Sandrone G, Mortara A, Torzillo D, La Rovere MT, Malliani A, Lombardi F.
Ef-fects of beta blockers (atenolol or metoprolol) on heart rate variability
after acute myocardial infarction. Am J Cardiol. 1994;74(4):340-345.
33. Tuininga YS, Crijns HJ, Brouwer J, et al. Evaluation of importance of
central ef-fects of atenolol and metoprolol measured by heart rate
variability during men-tal performance tasks, physical exercise, and daily
life in stable postinfarct patients. Circulation. 1995;92(12):3415-3423.
34. Psaty BM, Siscovick DS. Minimizing bias due to confounding by
indication in comparative effectiveness research: the importance of
restriction. JAMA. 2010; 304(8):897-898.
35. Kottke T, Baechler C, Parker E. The accuracy of heart disease
prevalence esti-mated from claims data compared to electronic
health record. Prev Chronic Dis. 2012;9:E141.
INVITE0 COMMENTAR2
O6servationa! Comparative Effectiveness St%-ies
of 0r%/ T&erapies
High-Quality Answers or Important Clinical Questions?
F
or the prevention of cardiovascular disease (CV5)$
-blockers are among the most widely used therapies#
Multiple clinical trials have
established their efficacy in preventing death after
myocardial infarction (M) and in treating congestive
heart failure (C!") due to systolic dysfunction#
/$(
-Flockers have also long been used to treat hyperten-
sion# 9lthough low-dose diuretics are the recom-mended
first-line agent for pharmacologic therapy for
uncomplicated high blood pressure$
3
several large trials
funded by the pharmaceutical industry have used
-blockers as the active-comparison control treatment$
7
and the results of these trials suggest that other therapies
are more effective than atenolol in pre-venting
cardiovascular events$ particularly stroke#
4$6
Fecause no
primary prevention trial among hyperten-sive patients
has compared atenolol head to head with other
-blockers$ their comparative effectiveness in this setting
remains unknown#
:o address this %uestion$ +arker and colleagues
0
conducted an observational study that compared the new
use of atenolol and metoprolol tartrate$ ( widely used
-blockers in the &nited Atates$ for the prevention of M$
stroke$ and C!" in patients with treated hyper-tension#
:his study was nested within the hypertension registry
of the Cardiovascular Research 'etwork (CVR')$
which includes all adult patients with hyper-tension
enrolled in 3 large integrated health care plans from
())) to ())*# Most -blocker use was in combi-nation
with other therapies$ and half of the study popu-lation
used diuretics within 6 months prior to starting a
-blocker# "or all outcomes$ the relative risk estimates
were null$ and the *48 confidence intervals e,cluded a
greater than (8 increased risk associated with metopro-
lol use compared with atenolol#
:his study has several strengths# :he validation of en-
try criteria in the hypertension registry and the use of elec-
tronic prescriptions records allowed for a new-user study
design$ which compares users of different treatments at a
similar point in the natural history of hypertensive dis-ease
and avoids some sources of bias that are common in studies
that include prevalent users of medications#
1
Fecause of the
careful use of restriction to e,clude per-sons with known
prevalent CV5 and even persons re-ferred to a cardiologist$
who may be more likely than non-referred patients to have
undocumented or suspected but undiagnosed CV5$ the
observed cardiovascular events likely reflect incident
disease#
:he authors used several analytic methods to mini-
mize confounding bias# n one set of analyses$ factors
as-sociated with both the choice of -blocker and the risk
of outcomes were ad-usted for# n another$ propensity
scores were used to make comparisons among a subset
of the study population with similar probabilities of
treat-ment based on known risk factors# "urthermore$
ateno-lol and metoprolol$ which are both cardioselective
/
-adrenergic receptor blockers$ have similar
pharmacologic properties and similar indications#
*
:he
relative risk es-timates from the ( analytic approaches
were similar$ and because of large sample sizes$ the *48
confidence inter-vals were narrow#
:he study by +arker and colleagues
0
also shares the
traditional and persistent weaknesses of observational
studies$ particularly those that rely on administrative data#
Aome potential confounding variables are not well cap-
tured by administrative codes$ and information on oth-
9RC! ':<R' M<5D V=; /0( ('=# /1)$ =C: 1$ ()/( CCC#9RC!':<R'M<5#C=M
()(2
,2*(2 American Me-ica! Association. A!! ri/&ts reserve-.
Downloaded From: http://archinte.jamanetwork.com/ on 04/29/2014