ORIGINAL ARTICLE
Quality Improvement Strategies for HypertensionManagement
A Systematic Review
Judith M. E. Walsh, MD, MPH,* Kathryn M. McDonald, MM,† Kaveh G. Shojania, MD,‡Vandana Sundaram, MPH,†§ Smita Nayak, MD,†§ Robyn Lewis, MA,†Douglas K. Owens, MD, MS,†§ and Mary Kane Goldstein, MD, MS†¶
Background: Care remains suboptimal for many patients withhypertension.Purpose: The purpose of this study was to assess the effectivenessof quality improvement (QI) strategies in lowering blood pressure.Data Sources: MEDLINE, Cochrane databases, and article bibliog-raphies were searched for this study.Study Selection: Trials, controlled before–after studies, and inter-rupted time series evaluating QI interventions targeting hypertensioncontrol and reporting blood pressure outcomes were studied.Data Extraction: Two reviewers abstracted data and classified QIstrategies into categories: provider education, provider reminders,facilitated relay of clinical information, patient education, self-management, patient reminders, audit and feedback, team change, orfinancial incentives were extracted.Data Synthesis: Forty-four articles reporting 57 comparisons un-derwent quantitative analysis. Patients in the intervention groupsexperienced median reductions in systolic blood pressure (SBP) anddiastolic blood pressure (DBP) that were 4.5 mm Hg (interquartilerange �IQR�: 1.5 to 11.0) and 2.1 mm Hg (IQR: �0.2 to 5.0) greaterthan observed for control patients. Median increases in the percent-age of individuals achieving target goals for SBP and DBP were
16.2% (IQR: 10.3 to 32.2) and 6.0% (IQR: 1.5 to 17.5). Interven-tions that included team change as a QI strategy were associatedwith the largest reductions in blood pressure outcomes. All teamchange studies included assignment of some responsibilities to ahealth professional other than the patient’s physician.Limitations: Not all QI strategies have been assessed equally,which limits the power to compare differences in effects betweenstrategies.Conclusion: QI strategies are associated with improved hyperten-sion control. A focus on hypertension by someone in addition to thepatient’s physician was associated with substantial improvement.Future research should examine the contributions of individual QIstrategies and their relative costs.
Key Words: quality improvement, blood pressure, hypertension,guideline implementation
(Med Care 2006;44: 646–657)
Hypertension care in the United States often fails tocomply with evidence-based guidelines. From 1999 to
2000, only 68.9% of individuals with hypertension wereaware of their hypertension, and only 58.4% of hypertensiveindividuals were treated.1 Even when treated, blood pressuremay not be adequately controlled. In the year 2000, only51.5% of patients in managed care health plans had con-trolled blood pressure (systolic blood pressure �SBP� �140mm Hg and diastolic blood pressure �DBP� �90 mm Hg).2
The Joint National Committee on Prevention, Detec-tion, Evaluation, and Treatment of High Blood Pressurerecommends targets for blood pressure control.3 Even smallimprovements in blood pressure control can have majorpublic health impact: lowering DBP by only 2 mm Hg couldresult in a 6% reduction in the risk of coronary heart diseaseand a 15% reduction in the risk of stroke and transientischemic attacks.4 Similarly, reducing usual SBP by 2 mm Hgis associated with 10% lower stroke mortality and approxi-mately 7% lower mortality from vascular causes in a middle-aged population.5
Many quality improvement (QI) strategies have fo-cused on improving hypertension control. QI strategies can
From the *Division of General Internal Medicine, Department of Medicine,University of California, San Francisco, California; the †Center forPrimary Care and Outcomes Research, Stanford University, Stanford,California; the ‡Department of Medicine and the Ottawa Health ResearchInstitute, University of Ottawa, Ottawa, Canada; the §VA Palo AltoHealth Care System, Palo Alto, California; and the ¶Geriatrics ResearchEducation and Clinical Center, VA Palo Alto Health Care System, PaloAlto, California.
This work was performed by the Stanford–UCSF Evidence-based PracticeCenter under contract number 290-02-0017 to the Agency for HealthcareResearch and Quality; Rockville, MD. This work was also supported inpart by the Department of Veterans Affairs.
The authors of this article are responsible for its content. Views expressed arethose of the authors and not necessarily those of the Department ofVeterans Affairs. Statements in this manuscript should not be construedas endorsements by the Agency for Healthcare Research and Quality orthe U.S. Department of Health and Human Services of a particular drug,device, test, treatment, or other clinical service.
Reprints: Mary Kane Goldstein, MD, MS, VA Palo Alto Health CareSystem, GRECC 182B, 3801 Miranda Avenue, Palo Alto, CA 94304.E-mail: [email protected].
Copyright © 2006 by Lippincott Williams & WilkinsISSN: 0025-7079/06/4407-0646
Medical Care • Volume 44, Number 7, July 2006646
include one or several components and can target the pro-vider, the patient, the healthcare system, or any combinationof these.
As part of a series of evidence reports, Closing theQuality Gap, funded by the Agency for Healthcare Researchand Quality on improving health care for selected topicsidentified by the Institute of Medicine as meriting nationalpriority,6,7 we conducted a systematic review of the evidencesupporting QI strategies for hypertension control. We ad-dressed the following questions: Are QI programs effective inproducing clinically meaningful reductions in blood pres-sure? Which QI strategies are most effective at ensuring thatblood pressures are lowered?
METHODS
Definition of Quality Improvement StrategiesWe defined a QI strategy as an intervention aimed at
reducing the quality gap (the difference between healthcareprocesses or outcomes observed in practice and those potentiallyobtainable based on current professional knowledge) for a groupof patients representative of those encountered in routine prac-tice.6 We developed our taxonomy of QI strategies by modifyingseveral well-established classification systems.8–12 A systematicreview of disease management studies that combined QI strat-egies and targets classified interventions as: provider education,provider feedback, provider reminders, patient education, patientreminders, and patient financial incentives,12 whereas an alter-native taxonomy described in a review of interventions topromote immunization and cancer screening11 specified 3 di-mensions for characterizing QI strategies: type of strategy (eg,education), mediators of intervention (eg, involvement of topmanagement), and audience targeted (eg, patients, providers,healthcare delivery systems). We modified these taxonomies toreview hypertension management evidence. We classified inter-ventions as provider education (materials/instruction given toproviders regarding appropriate care for patients), provider re-minders (prompts given to providers to perform specific caretasks), provider audit and feedback (summary clinical perfor-mance information given to healthcare providers), facilitatedrelay of clinical data to providers (clinical information collecteddirectly from patients and relayed to provider in which the dataare not routinely collected during a patient visit, eg, transmissionof a patient’s home blood pressure measurements), patient edu-cation (materials/instructions issued to patients providing hyper-tension information), patient reminders (efforts directed towardpatients encouraging them to keep appointments or adhere tocare), promotion of self-management (access to resources ordevices that enhance patients’ ability to manage their condition,eg, providing home blood pressure monitoring kit), team change(creation of multidisciplinary team, addition of new teammember�s�, change of role�s�, case or disease management), andfinancial incentives/regulation or reimbursement changes.
Included Study DesignsWe included patient- and cluster-randomized trials,
quasi-randomized trials, controlled before–after studies, andinterrupted time-series studies.13 A quasi-randomized trialwas defined as containing at least 2 cohorts of patients
assembled prospectively using an arbitrary, but nonrandomallocation procedure (eg, even/odd medical record numbers).A controlled before–after study was defined as a contempo-raneous observation for cohorts differing primarily with re-spect to exposure to the QI intervention. Interrupted timeseries required to report outcomes from at least 3 time pointsin the pre- and postintervention periods.
Included OutcomesWe restricted our analysis to studies reporting measures
of hypertension control before and after the intervention:these included SBP or DBP or change in SBP or DBP and/orthe percentage of patients achieving SBP or DBP within atarget range.
Search StrategyWe searched the MEDLINE database through July 2003
using key words and medical subject headings for hypertensionand blood pressure combined with terms related to qualityimprovement (eg, total quality management, diffusion of inno-vation, disease management). Additional search terms focusedon identifying multifactorial interventions and targeted providereducation, audit and feedback, and reminder systems (Appendix1: Search Strategy). We also reviewed citations from the Co-chrane Effective Practice and Organisation of Care (EPOC)registry of QI strategies.
We included articles if they reported QI strategies forhypertension and assessed blood pressure outcomes. We ex-cluded articles focusing only on secondary hypertension orspecialized subpopulations (eg, hypertension in patients withalcoholism). We restricted our review to interventions targetingsome component of provider behavior or organizational change(ie, articles that evaluated patient education or self-managementby themselves were excluded because these interventions withpatients as the only target constituted a separate Institute ofMedicine priority area, eg, health literacy/self-management).7
We excluded articles published before 1980, and regarded a daterestriction as appropriate given changes in hypertension care andapproaches to QI over the past 2 decades.
We screened titles and abstracts for relevance. Atfull-text review, 2 independent reviewers abstracted key in-formation (eg, study design, reported outcomes) and conflictswere resolved by consensus.
AnalysesWe conducted 2 types of quantitative analyses: calculation
of net change in blood pressure and multivariate analyses.
Calculation of Net Change in Blood PressureWe calculated the net change in the blood pressure
outcome attributable to the intervention (int) for each in-cluded study, defined as:
Net � in BP � (Postint BP � Preint BP)Study group
� (Postint BP � Preint BP)Control group
To characterize the effect of a particular type of QIstrategy, we then calculated the median change in the outcomeachieved by studies in which the intervention included this
Medical Care • Volume 44, Number 7, July 2006 Evidence Synthesis of Quality Improvement Strategies for Hypertension
© 2006 Lippincott Williams & Wilkins 647
strategy.14 For example, we computed the median of net changein SBP for the QI strategy of provider education. To do this, wefirst calculated the net change in SBP (in mm Hg), using thepreviously defined formula, for all studies that used providereducation and measured SBP. We then calculated the medianvalue of the net change for these studies. We used median ratherthan mean change to avoid having skewed summary measuresbased on outlier studies with particularly large or small changesin outcome.
Because of the possibility of overestimation of medianchange resulting from preferential reporting of positive studies,particularly with smaller studies, we analyzed studies in terms ofsample sizes comparing median change among studies withsample sizes in the lower half with those in the upper half.
We reviewed studies for 5 key methodologic features:randomized allocation of intervention, adequate concealmentof allocation, providers blinded, patients blinded, and thepossibility of unit-of-analysis errors.
To evaluate the impact of study or intervention char-acteristics, we performed a nonparametric test for differencesin median change in blood pressure, the Mann-Whitney ranksum test. These analyses were possible only for mutuallyexclusive categories, eg, all interventions with a given QIstrategy versus all interventions without this strategy. Wewere unable to compare one strategy with another because ofthe frequent occurrence of several strategies within a givenstudy. We also had insufficient information to test for differ-ences in outcomes reported as the percentage of patientsachieving SBP or DBP within a target blood pressure range.
Multivariate AnalysesWe used a mixed model incorporating fixed and ran-
dom effects to predict the postintervention difference be-tween intervention and control group values for mean SBPand DBP controlling for study size and for the difference inmean SBP and DBP values before the intervention. In uni-variate analyses, other study features (eg, trial design, studyyear) did not have significant associations with interventionoutcomes so were not included in the models. Postinterven-tion standard deviations in the control and interventiongroups were pooled with weighting by sample size in eachgroup. This measure of within-study variability provided theresidual error in a mixed model with a random study effect:
Y � X� � Z� � e; � � N�0; G� e � N�0; R�,
where � is the fixed effect, � is the random effect, and e is theerror at the study level. As described elsewhere,15 using ProcMixed (SAS software, version 8.2; SAS Institute, Cary, NC)for meta-analysis requires reversing the roles of the within-study and between-study variations and then postprocessingthe output.
Accounting for Cluster EffectsWe anticipated that a substantial number of studies
would assign study groups to providers or clinics but collectdata on individual patients (ie, the studies would exhibit“clustering”). To avoid spurious precision in our analy-
ses,16–18 we calculated an effective sample size for eachstudy, defined as:
NEffective � (k * m)/(1 (m � 1) * ICC)
Here, “k” represents the number of clusters, “m” denotes thenumber of observations per cluster, and “ICC” is the intra-cluster coefficient. For studies without clustering, ICC � 0,and NEffective � k * m (ie, the reported sample size).18,19
Descriptive Analysis of Team Change StrategiesTo provide a richer analysis of the most common
strategy evaluated, we further described the characteristicsand outcomes of the team change studies. We grouped studiesbased on common characteristics (eg, setting, involvement ofadditional caregivers, use of home monitoring). To review allof these studies together, regardless of outcomes assessed(SBP, DBP, SBP range, and DBP range), we classifiedstudies into those with improvements in blood pressure con-trol for all outcomes assessed that were consistently greaterthan the median (“consistently greater”), those with medianchanges within 10% of the median (“consistently equiva-lent”), and those with median changes less than the median(“consistently less”). Where there were conflicts betweenclassifications among multiple outcomes (eg, 2 “less” out-comes and 1 “greater” outcome), we designated the studyresults as “mixed.”
RESULTSThe search (Fig. 1) yielded 3165 citations. 359 articles
were eligible for full-text review and 110 were fully ab-stracted. Of these, 47 included patient education or self-management only, and 17 reported outcomes other than bloodpressure (eg, physician adherence to guidelines). Two studiesdid not provide information before the intervention. Theremaining 44 studies included a total of 57 comparisons of QIstrategies (because several articles had more than one inter-vention).20–63 Thirty-four studies were randomized, con-trolled trials (including 45 comparisons), 5 were quasi-ran-domized, controlled studies (including 6 comparisons), and 5were controlled before–after studies (including 6 compari-sons). (A table with comparative details about each study isavailable on request to the authors.)
Types and Numbers of Quality ImprovementStrategies
The majority of articles described interventions consist-ing of more than one strategy with the median number of QIstrategies per comparison equal to 3. Only 17.5% (10 of 57)comparisons evaluated interventions using a single strategy.Among the individual strategies, team change and patienteducation (in combination with other strategies) were mostcommonly used. Team change was used in 36 of thecomparisons and patient education was used in 28 of thecomparisons. Facilitated relay of clinical data (n � 22) andprovider education (n � 20) strategies were common aswell.
Walsh et al Medical Care • Volume 44, Number 7, July 2006
© 2006 Lippincott Williams & Wilkins648
Effect of Quality Improvement Strategies onBlood Pressure Outcomes
The majority of studies showed a modest improvementin SBP and DBP associated with QI interventions. Strategiesthat included team change, patient education, facilitated relayof clinical data, or promotion of self-management had largermedian improvements in blood pressure outcomes (Table 1,by inspection, not tested for statistical significance).
The majority of studies used a combination of QIstrategies. For both SBP and DBP reductions, there was noclear pattern of increasing or decreasing effect as the numberof QI strategies increased.
In the multivariate model adjusting for sample size andfor differences in baseline blood pressure, the mean effectsacross all interventions were reductions in SBP of 4.2 mm Hg(95% CI 1.8, 6.6) and DBP of 1.9 mm Hg (95% CI 0.7, 3.1).The differences in postintervention SBP and DBP associatedwith each of the QI strategies are shown in Figure 2. Therewere too few studies assessing financial incentives to includethis intervention in the multivariate analyses. Team change wasthe only strategy that showed significant results for both SBPand DBP. Strategies that included patient education and self-
management were associated with a significant reduction inDBP but not SBP.
For all QI strategies, the median increase in the pro-portion of patients in target SBP range and DBP range was16.2% (interquartile range �IQR�: 10.3 to 32.2) and 6.0%(IQR: 1.5 to 17.5), respectively. Again, strategies that in-cluded team change, patient education, facilitated relay ofclinical data, or promotion of self-management had the larg-est improvements (by inspection, not tested for statisticalsignificance).
Effect of Team Change on Blood PressureOutcomes
Team change was the most commonly used strategy andwas assessed in 28 studies (36 comparisons). Overall, studiesincluding team change demonstrated a median improvement inSBP of 9.7 mm Hg (IQR: 4.2 to 14.0) compared with a 2.0 mmHg (IQR: 1.0 to 3.9) median improvement for studies without ateam change component (Mann-Whitney U test; P � 0.005).Studies including team change also showed median improve-ment in DBP of 4.2 mm Hg (IQR: 0.2 to 6.8) compared withstudies without a team change component (median improvementin DBP of 0.6 �IQR: �0.2 to 2.1� mm Hg for studies withoutteam change component; Mann-Whitney P � 0.025 for com-parisons with versus without team change. The percentage ofpatients with SBP and DBP in the target range improved whenteam change strategies were included (Table 1).
In multivariate analyses that corrected for differences inbaseline blood pressure values, the effects of interventionswith or without team change were similar to findings with themedian net change but attenuated. Studies including teamchange showed an improvement in SBP of 5.9 mm Hg (95%CI: 2.8 to 8.9; P � 0.0004), whereas studies without teamchange showed an improvement in SBP of 1.8 mm Hg (95%CI: �1.8 to 5.4; P � 0.32). For DBP, studies including teamchange showed a lowering of 3.1 mm Hg (95% CI: 1.7 to 4.6;P � 0.0001), whereas studies without team change showed alowering of 0.2 mm Hg (95% CI: �1.4 to 1.9; P � 0.78).
Descriptive Analysis of Team Change StudiesTwenty-seven of these comparisons were conducted in the
United States20,24,25,28,30–32,34,39,42–44,47,49,52,55,56,58,60,62,63 and9 outside of the United States.23,27,29,37,38,45,46,48 Interventionswere most commonly conducted in a clinic, either a generaloutpatient clinic (n � 14)20,24,27,28,30,32,39,43,44,47,55,56 or a hy-pertension clinic (n � 7).25,29,48,58,63 Seven comparisons wereconducted in academic settings,24,32,43,44,55,58 6 at the work-site,31,34,45,46 3 at pharmacies,23,37,52 and 3 in the community.42,60,62
Six interventions included some component of the interventionoccurring at the patient’s residence.30,38,42,49,62 Of the compar-isons conducted in clinics, 14 were conducted at a singlesite,24,25,30,32,43,47,48,55,56,58,63 2 included 2 or 3 sites,28,29 and 5included 11 or more sites.20,27,39,44
Studies conducted in pharmacies reported blood pres-sure reductions consistently greater (n � 2)37,52 than orequivalent (n � 1)23 to the median reduction observed for allstudies. Studies conducted in academic settings also tended tohave bigger blood pressure reductions (4 of 7 had consistentlygreater reductions24,43,58 and only 1 study44 had outcomes
FIGURE 1. Results of the literature search. The figure showsthe flow of the article reviews starting with the number ofstudies that were identified through the search strategy.
Medical Care • Volume 44, Number 7, July 2006 Evidence Synthesis of Quality Improvement Strategies for Hypertension
© 2006 Lippincott Williams & Wilkins 649
TAB
LE1.
Effe
ctof
Qua
lity
Imp
rove
men
tSt
rate
gies
onBl
ood
Pres
sure
Out
com
es
Typ
eof
Qua
lity
Impr
ovem
ent
Med
ian
Red
ucti
onin
Syst
olic
Blo
odP
ress
ure
(mm
Hg)
�int
erqu
arti
lera
nge�
*n
�nu
mbe
rof
com
pari
sons
†
Med
ian
Red
ucti
onin
Dia
stol
icB
lood
Pre
ssur
e(m
mH
g)�i
nter
quar
tile
rang
e�*
n�
num
ber
ofco
mpa
riso
ns†
Med
ian
Abs
olut
eIn
crea
sein
Pro
port
ion
ofP
atie
nts
Ach
ievi
ngSy
stol
icB
lood
Pre
ssur
ein
aC
erta
inR
ange
�int
erqu
arti
lera
nge�
*n
�nu
mbe
rof
com
pari
sons
†
Med
ian
Abs
olut
eIn
crea
sein
Pro
port
ion
ofP
atie
nts
Ach
ievi
ngD
iast
olic
Blo
odP
ress
ure
ina
Cer
tain
Ran
ge�i
nter
quar
tile
rang
e�*
n�
num
ber
ofco
mpa
riso
ns†
Pro
vide
rre
min
ders
1.2
�1.0
–1.9
�0.
3��
0.2–
1.7�
NA
4.5
�2.0
–7.0
�
n�
633,4
1,5
1,5
3,6
1n
�63
3,4
1,5
1,5
3,6
1n
�0
n�
261
Fac
ilit
ated
rela
yof
clin
ical
data
8.0
�2.5
–12.
3�1.
8��
0.1–
4.5�
25.1
�17.
0–34
.2�
2.0
�1.6
–5.0
�
n�
1624,2
5,2
8,3
2,3
6,3
7,4
4,4
8,5
0,5
2,5
7,6
1,6
2n
�18
24,2
5,2
8,3
2,3
6,3
7,4
0,4
4,4
6,4
8,5
0,5
2,5
7,6
1,6
2n
�42
3,2
8,3
4,5
2n
�54
0,5
9,6
1
Aud
itan
dfe
edba
ck1.
5�1
.2–1
.7�
0.6
�0.4
–1.0
��
3.5
��5.
7–�
1.4�
2.0
�1.7
–4.3
�
n�
341,6
1n
�44
0,4
1,6
1n
�23
9n
�64
0,5
6,5
9,6
1
Pro
vide
red
ucat
ion
3.3
�1.2
–5.4
�0.
6��
0.7–
3.4�
10.9
�1.4
–13.
1�3.
5�1
.7–1
1.3�
n�
1121,2
8,3
3,3
5,3
8,4
1,5
1,5
3,6
1n
�16
21,2
8,3
3,3
5,3
8,4
0,4
1,5
1,5
3,5
4,6
1,6
3n
�62
1,2
7,2
8,3
5,3
9n
�62
1,3
5,4
0,6
1
Pat
ient
educ
atio
n8.
1�3
.3–1
1.8�
3.8
�0.6
–6.7
�19
.2�1
1.4–
33.2
�17
.0�1
1.4–
24.5
�
n�
1820,2
4,2
5,2
8,2
9,3
2,3
5–38,
42,4
4,4
7,4
9,5
2,5
7,6
2
n�
2120,2
4,2
5,2
8,2
9,3
2,3
5–38,4
2,4
4,
47,4
9,5
2,5
4,5
7,6
0,6
2
n�
827–29,3
1,3
4,3
5,5
2n
�72
9,3
0,3
5,4
3,4
7,6
0
Pro
mot
ion
ofse
lf–
man
agem
ent
3.3
�2.6
–10.
1�2.
8�0
.4–6
.7�
13.4
9.4
�5.3
–11.
4�
n�
935,3
6,3
8,4
7,4
9,5
0,5
7,6
2n
�13
22,3
5,3
6,3
8,4
6,4
7,4
9,5
0,5
4,5
7,6
2n
�13
5n
�32
2,3
5,4
7
Pat
ient
rem
inde
rs3.
3�2
.3–4
.5�
0.4
��2.
4–5.
0�N
A2.
0�1
.1–9
.4�
n�
537,4
7,4
8,5
7n
�92
2,3
7,4
7,4
8,5
4,5
7,6
0n
�0
n�
522,4
3,4
7,5
6,6
0
Tea
mch
ange
9.7‡
�4.2
–14.
0�4.
2‡�0
.2–6
.8�
21.8
�9.0
–33.
8�17
.0�5
.7–2
4.5�
n�
2020,2
4,2
5,2
8,2
9,3
2,3
7,3
8,4
2,4
4,
47–49,5
2,5
5,5
8,6
2
n�
2420,2
4,2
5,2
8,2
9,3
2,3
7,3
8,4
2,4
4–49,
52,5
5,5
8,6
0,6
2,6
3
n�
1223,2
7–29,3
1,3
4,3
9,5
2,5
5n
�72
9,3
0,4
3,4
7,5
6,6
0
Fin
anci
alin
cent
ives
�13
.30.
0��
2.0–
2.5�
NA
4.2
��1.
1–9.
4�
n�
147
n�
326,4
7,6
0n
�0
n�
247,6
0
All
com
pari
sons
4.5
�1.5
–11.
0�2.
1��
0.2–
5.0�
16.2
�10.
3to
32.2
�6.
0�1
.5–1
7.5�
n�
3320,2
1,2
4,2
5,2
8,2
9,3
2,3
3,3
5–38,
41,4
2,4
4,4
7–53,5
5,5
7,5
8,6
1,6
2
n�
4320–22,2
4–26,2
8,2
9,3
2,3
3,3
5–38,4
0–42,
44–55,5
7,5
8,6
0–63
n�
1421,2
3,2
7–29,3
1,3
4,3
5,3
9,5
2,5
5n
�16
21,2
2,2
9,3
0,3
5,4
0,4
3,4
7,5
6,5
9–61
*Whe
nn
�2,
brac
kets
show
the
actu
alre
sult
sof
each
stud
yra
ther
than
inte
rpol
ated
inte
rqua
rtil
era
nge.
†T
henu
mbe
rof
cita
tion
sli
sted
diff
ers
from
the
nin
case
sw
here
ther
ew
ere
mul
tipl
eco
mpa
riso
nsfo
ra
stud
y.‡P
�0.
05fo
rM
ann-
Whi
tney
anal
yses
ofre
duct
ions
insy
stol
icbl
ood
pres
sure
and
dias
toli
cbl
ood
pres
sure
com
pari
ngst
udie
sw
ith
the
qual
ity
impr
ovem
ent
stra
tegy
wit
hth
ose
wit
hout
it.
No
com
para
ble
stat
isti
cal
anal
yses
wer
efe
asib
lefo
rpr
opor
tion
ofpa
tien
tsac
hiev
ing
ace
rtai
nsy
stol
icbl
ood
pres
sure
ordi
asto
lic
bloo
dpr
essu
rera
nge.
NA
indi
cate
sno
tap
plic
able
.
Walsh et al Medical Care • Volume 44, Number 7, July 2006
© 2006 Lippincott Williams & Wilkins650
FIGURE 2. A, Changes in systolic blood pressure associated with each quality improvement (QI) strategy adjusting for studysize and baseline differences in blood pressure. B, Changes in diastolic blood pressure associated with each QI strategy adjust-ing for study size and baseline differences in blood pressure. Forty-four studies included 57 comparisons. A total of 33 of thesecomparisons reported absolute changes in systolic blood pressure and 43 of these comparisons reported absolute changes indiastolic blood pressure. The x-axis shows each QI strategy and the y-axis shows the difference in the postintervention changein blood pressure between studies with and without a particular QI strategy (change in blood pressure in the interventiongroup minus change in blood pressure in the control group). “All comparisons” shows the estimate for all comparisons re-porting blood pressure outcomes regardless of the QI strategy included in the comparison. “Diffpre” refers to the baseline dif-ferences in blood pressure control between intervention and control groups. Each estimate represents the difference betweenthe reduction in blood pressure (systolic blood pressure for top panel and diastolic blood pressure for bottom panel) associ-ated with the presence of a particular QI strategy and the benefit observed in interventions without that strategy. The num-bers in parentheses indicate the number of studies contributing to the estimate �eg, in (A), 10 comparisons evaluated inter-ventions involving provider education). Of note, there were too few studies of financial incentives to include this strategy.Negative results reflect lower blood pressure values when a QI strategy is present compared with its absence. Positive resultsindicate that interventions with a component of the QI strategy in question produced smaller reductions in blood pressurethan did interventions without such a component. The estimates include adjustment for the effects of study size and baselinedifferences in blood pressure control between intervention and control groups.
Medical Care • Volume 44, Number 7, July 2006 Evidence Synthesis of Quality Improvement Strategies for Hypertension
© 2006 Lippincott Williams & Wilkins 651
that were consistently less than the median). Studies that wereconducted at multiple sites generally had changes that wereless than the median. There was no compelling evidence todistinguish studies performed at hypertension clinics fromstudies performed in other settings.
All of the team change interventions for hypertensioncare involved assigning some patient care responsibilities tosomeone other than the patient’s doctor. Pharmacists, nurses,physician assistants, and worksite physicians took on coordi-nation, counseling, and patient follow-up functions in manyof the studies. Sometimes, these individuals carried out allcommunication with the patient related to blood pres-sure control (“all communication transferred”; 18 compar-isons20,23,27,29,30,38,42–44,47,49,55,58,60,62); in other cases, theyworked with the doctor (“shared responsibility”; 7 compari-sons24,25,28,32,37,52,63), and in a few cases, they or othersprovided support in the form of prompts or education tophysicians who retained full responsibility for the interactionswith the patients (“doctor-focused”; 5 comparisons39,48,56).The remaining 6 comparisons31,34,45,46 involved some formof triage and monitoring at the worksite.
An example of the “all communications transferred”intervention involved a clinical pharmacist meeting with thepatient to make changes in prescribed drugs, adjust dosages,provide drug counseling, and assess adherence to treatmentregimen. Interventions designated as “shared responsibility”involved, for example, a pharmacist relaying evidence-basedtreatment recommendations to a patient’s doctors and provid-ing the patient with education on dietary and lifestyle modi-fication as well as information about drug side effects. Bloodpressure reductions for studies involving doctor-focused in-terventions such as a specialist physician reviewing patientrecords and recommending treatment changes were consis-tently less than the median improvement and in some cases (4of 6 comparisons) showed a reduction in blood pressure lessthan the control arm. Most of the worksite interventionsshowed large improvements in blood pressure outcomes, with5 of 6 comparisons greater than the median (ie, 25%, 38%,40%, and 42% improvement in net SBP range and 5.6-mm-Hg improvement in net DBP).
Other characteristics represented repeatedly in the teamchange interventions were home blood pressure monitoringor use of a standard protocol for adding drugs. Of the 5studies with stepped care protocols, 244,63 had blood pressurereductions consistently less than the median and the other320,29,60 had mixed results depending on outcome assessed. Incontrast, 4 of 5 studies with home monitoring30,46,49,62 wereconsistently greater (n � 330,49,62) than or consistently equiv-alent (n � 130) to the median.
Effect of Methodologic Features on BloodPressure Outcomes
In our assessment of methodologic features, we foundno significant differences in blood pressure outcomes acrossstudies with more rigorous versus less rigorous methodologicfeatures.
Effect of Study Size on Blood PressureOutcomes
Studies with smaller sample sizes generally reportedlarger reductions in SBP and DBP (Fig. 3). We considered thepossibility that the inclusion of clustered trials might haveconfounded the relationship between study size and magni-tude of median change. Because allocation occurs at the levelof the provider or clinic, cluster trials might be expected tohave larger numbers of patients. Among the included studies,the clustered trials included a median of 120 patients (IQR:52 to 249) and the nonclustered trials included a median of 67patients (IQR: 26 to 151). The clustered trials reported amedian reduction in SBP of 2.9 mm Hg, which was lowerthan that seen in the nonclustered trials (7.1 mm Hg; P �0.08). When we restricted our analyses to studies withoutclustering (eg, patient level randomized, controlled trials), theinverse relation between sample size and median changemagnitude remained.
DISCUSSIONQuality improvement strategies are associated with
improved control of hypertension. QI strategies generallyimproved SBP and the proportion of patients achieving targetSBP range and had a more modest effect on DBP and theproportion of patients achieving target DBP range. All of thestrategies assessed may be beneficial in terms of clinicallymeaningful reductions in blood pressure under some circum-stances and in varying combinations.
In general, team change had the largest effect on bothSBP and DBP outcomes regardless of study design or size. QIstrategies, including patient education and self-management,had a significant effect on DBP but not SBP, which may berelated to more studies assessing DBP outcomes or to the factthat until recently, the main focus of blood pressure manage-ment was on DBP.
A common feature of the team change studies wasassignment of some patient care responsibilities to someoneother than the patient’s physician.
There are many possible explanations for the success ofteam change in achieving blood pressure control. Such inter-ventions typically require administrative support, which maybe an important factor in the success of a QI strategy. Manystudies of team change include designation of specific staff toaddress hypertension, which may represent either an increasein staffing or a reallocation of staff effort to hypertension. Thefindings regarding team change are consistent with resultsfrom an observational study of the Veterans Affairs HealthCare System, which reported moderate improvements in ratesof blood pressure control after implementation of systemwidereengineering.64,65 A trial published after our search end dateconfirms the major impact that team change may exert onblood pressure outcomes. Among inner-city African Ameri-can men, interventions by a multidisciplinary team improvedblood pressure control.66
Assigning some of the responsibility for blood pressurecontrol to a healthcare professional other than the patient’sphysician was common to all of the team change studies.What makes team change work is unclear, but could include
Walsh et al Medical Care • Volume 44, Number 7, July 2006
© 2006 Lippincott Williams & Wilkins652
FIGURE 3. A, Changes in systolic blood pressure based on adjusted sample sizes. B, Changes in diastolic blood pressure basedon adjusted sample sizes. Forty-four studies included 57 comparisons. A total of 33 of these comparisons reported absolutechanges in systolic blood pressure and 43 of these comparisons reported absolute changes in diastolic blood pressure. Eachbar represents the reduction in blood pressure �(A) shows reduction in systolic blood pressure and (B) shows reduction in dia-stolic blood pressure� for each quality improvement strategy. The bars with stripes show the reduction in blood pressure forthose studies that had smaller adjusted sample sizes (in the lower half) and the bars with dots show the reduction in bloodpressure for those studies that had larger adjusted sample sizes (in the upper half). The numbers in parentheses indicate thenumber of studies in each half �eg, in (A) for studies reporting systolic blood pressure outcome, there were 16 studies thathad smaller adjusted sample sizes and 17 studies with larger sample sizes). The x-axis shows each quality improvement strat-egy and the y-axis shows the reduction in blood pressure (mm Hg).
Medical Care • Volume 44, Number 7, July 2006 Evidence Synthesis of Quality Improvement Strategies for Hypertension
© 2006 Lippincott Williams & Wilkins 653
use of nonphysician providers, having a strict algorithm orguideline for hypertension control, or having clearer assign-ments of roles and responsibilities for hypertension caremanagement. Regardless of the reason for success, teamchange was an effective QI strategy in the studies included inour review. A recent Cochrane review of interventions usedto improve control of blood pressure also concluded that theuse of healthcare professionals such as nurses or pharmacistsin managing blood pressure was a successful strategy thatdeserved further evaluation.67 This review, which was limitedto randomized, controlled studies only, included 7 nurse/pharmacist-led studies and 8 other organizational interven-tions related to team change. Thus, our review almost triplesthe comparative data on team change strategies. Assigningadditional staff time for a health professional to work directlywith patients may be expensive in the short term; however, arecent report on diabetes raises the possibility that suchinterventions can be economically attractive.68
Patient education appeared to be a successful strategyfor improving blood pressure control, but we evaluated itonly when it was used in conjunction with other QI strategies.Therefore, the impact of patient education per se may beoverestimated in the studies we included, and we cannotdetermine the impact of patient education as a solo strategy.Promotion of patient self-management of blood pressure hada modest effect on blood pressure outcomes, an effect con-sistent with a recent meta-analysis that showed that promo-tion of patient self-management of hypertension and diabetescould produce clinically important benefits.69 Interventionsthat included provider education had only modest effects inaccord with the recent Cochrane review.67 Provider behaviorappears to be difficult to change9; however, relatively inex-pensive interventions directed toward providers, particularlyreminders, although yielding smaller blood pressure changeson a per-patient basis, could effect greater overall benefit asa result of the large number of patients that could be reached.In contrast, a more labor-intensive intervention might resultin a larger benefit for each patient but would reach a smallernumber of patients and at greater cost. Alternatively, chang-ing provider behavior may not be a critical factor for accom-plishing improved blood pressure control. Perhaps providerbehavior should be targeted only in conjunction with othersystem changes.70
LimitationsOnly 10 of the included studies assessed a single QI
strategy; because most studies included more than one QIstrategy, we could not discern definitively which individualQI strategies had the greatest effects, and we could notdetermine whether certain combinations of individual QIstrategies were more “potent” than others. Not all QI strate-gies have been assessed equally, which limits the power todetect statistically or clinically significant differences. Im-provements in blood pressure control were smaller in largerstudies than in smaller studies. Large studies may be morelikely to be reported than small studies if the results arenegative, raising the concern that overall measures of theeffectiveness of QI strategies may be overestimated because
of publication bias or other factors (eg, setting) potentiallyassociated with study size.
CONCLUSIONQI strategies can result in clinically important reduc-
tions in blood pressure control. Smaller studies generallyreported larger median changes, suggesting some publicationbias or an unexplained confounder. It is possible that largerstudies involved more practices and physicians who were lessenthusiastic or engaged in the project. Studies that includedteam change reported greater improvements in blood pressurecontrol than did studies without these strategies, but theevidence does not definitively establish the superiority of anyindividual QI strategy. The multidisciplinary team approachto patient care is gaining popularity and is an integral com-ponent of the management of many chronic diseases.71–73
The success of the QI strategies that involve assigning somepatient care responsibilities to someone other than the phy-sician fits well with this team approach and should be inves-tigated further. QI strategies seem to be effective in a varietyof settings, but there is inadequate evidence to suggest tai-loring of particular QI strategies to specific settings. Inaddition, the cost-effectiveness of individual QI strategies forhypertension management should be a priority for futureresearch.
ACKNOWLEDGMENTSThe authors thank Amy Markowitz, Robert Wachter,
Jeremy Grimshaw, and the Cochrane Effective Practice andOrganisation of Care for their assistance on this study. Theauthors also thank Sheryl Davies, Jody Mechanic, Christo-pher Sharp, Melinda Henne, Bimal Shah, and Jo Kay Chanfor their assistance with data abstraction and Alan Bostromfor his assistance with statistical analysis.
REFERENCES1. Hajjar I, Kotchen TA. Trends in prevalence, awareness, treatment, and
control of hypertension in the United States, 1988–2000. JAMA. 2003;290:199–206.
2. The State of Managed Care Quality. National Committee for QualityAssurance; 2001.
3. Chobanian AV, Bakris GL, Black HR, et al. The Seventh Report of theJoint National Committee on Prevention, Detection, Evaluation, andTreatment of High Blood Pressure: the JNC 7 report. JAMA. 2003;289:2560–2572.
4. Cook NR, Cohen J, Hebert PR, et al. Implications of small reductions indiastolic blood pressure for primary prevention. Arch Intern Med.1995;155:701–709.
5. Lewington S, Clarke R, Qizilbash N, et al. Age-specific relevance ofusual blood pressure to vascular mortality: a meta-analysis of individualdata for one million adults in 61 prospective studies. Lancet. 2002;360:1903–1913.
6. Shojania KG, McDonald KM, Wachter RM, et al. Closing the QualityGap: A Critical Analysis of Quality Improvement Strategies, Volume1—Series Overview and Methodology. Technical Review 9 (ContractNo. 290-02-0017 to the Stanford University-UCSF Evidence-basedPractices Center). AHRQ Publication No. 04-0051-1. Rockville, MD:Agency for Health Research and Quality; August 2004.
7. Institute of Medicine. Crossing the Quality Chasm: A New HealthSystem for the 21st Century. Washington, DC: National Academy Press;2001.
8. Ellis P, Robinson P, Ciliska D, et al. Diffusion and Dissemination ofEvidence-based Cancer Control Interventions. Evidence Report/Tech-
Walsh et al Medical Care • Volume 44, Number 7, July 2006
© 2006 Lippincott Williams & Wilkins654
nology Assessment No. 79 (Prepared by the McMaster UniversityEvidence-based Practice Center under contract no. 290-97-0017).AHRQ Publication No. 03-E032. Rockville, MD: Agency for HealthcareResearch and Quality; May 2003.
9. Grimshaw JM, Shirran L, Thomas R, et al. Changing provider behavior:an overview of systematic reviews of interventions. Med Care. 2001;39:II-2–I-45.
10. Stafford RS. Alternative strategies for controlling rising cesarean sectionrates. JAMA. 1990;263:683–687.
11. Stone EG, Morton SC, Hulscher ME, et al. Interventions that increaseuse of adult immunization and cancer screening services: a meta-analysis. Ann Intern Med. 2002;136:641–651.
12. Weingarten SR, Henning JM, Badamgarav E, et al. Interventions used indisease management programmes for patients with chronic illness—which ones work? Meta-analysis of published reports �review�. BMJ.2002;325:925.
13. Cochrane Effective Practice and Organisation of Care Group. EffectivePractice and Organisation of Care Group, vol 2004. Institute of Popu-lation Health, University of Ottawa; 2004.
14. Grimshaw J, Thomas R, MacLennan G, et al. Effectiveness and effi-ciency of guideline dissemination and implementation strategies. HealthTechnol Assess. 2004;8:1–84.
15. Normand SL. Meta-analysis: formulating, evaluating, combining, andreporting. Stat Med. 1999;18:321–359.
16. Campbell M, Grimshaw J, Steen N. Sample size calculations for clusterrandomised trials. Changing Professional Practice in Europe Group (EUBIOMED II Concerted Action). J Health Serv Res Policy. 2000;5:12–16.
17. Donner A, Klar N. Issues in the meta-analysis of cluster randomizedtrials. Stat Med. 2002;21:2971–2980.
18. Kerry SM, Bland JM. The intracluster correlation coefficient in clusterrandomisation �review�. BMJ. 1998;316:1455.
19. Campbell MK, Mollison J, Grimshaw JM. Cluster trials in implemen-tation research: estimation of intracluster correlation coefficients andsample size. Stat Med. 2001;20:391–399.
20. Mortality after 10 1/2 years for hypertensive participants in the MultipleRisk Factor Intervention Trial. Circulation. 1990;82:1616–1628.
21. Avanzini F, Corsetti A, Maglione T, et al. Simple, shared guidelinesraise the quality of antihypertensive treatment in routine care. AmHeart J. 2002;144:726–732 �online�.
22. Becker LA, Glanz K, Sobel E, et al. A randomized trial of special packagingof antihypertensive medications. J Fam Pract. 1986;22:357–361.
23. Blenkinsopp A, Phelan M, Bourne J, et al. Extended adherence supportby community pharmacists for patients with hypertension: a randomisedcontrolled trial. International Journal of Pharmacy Practice. 2000;8:165–175.
24. Bogden PE, Abbott RD, Williamson P, et al. Comparing standard carewith a physician and pharmacist team approach for uncontrolled hyper-tension. J Gen Intern Med. 1998;13:740–745.
25. Borenstein JE, Graber G, Saltiel E, et al. Physician–pharmacist coman-agement of hypertension: a randomized, comparative trial. Pharmaco-therapy. 2003;23:209–216.
26. Brook RH, Ware JE Jr, Rogers WH, et al. Does free care improve adults’health? N Engl J Med. 1983;309:1426–1434.
27. Campbell NC, Ritchie LD, Thain J, et al. Secondary prevention incoronary heart disease: a randomised trial of nurse led clinics in primarycare. Heart. 1998;80:447–452.
28. Carter BL, Barnette DJ, Chrischilles E, et al. Evaluation of hypertensivepatients after care provided by community pharmacists in a rural setting.Pharmacotherapy. 1997;17:1274–1285.
29. Curzio JL, Rubin PC, Kennedy SS, et al. A comparison of the manage-ment of hypertensive patients by nurse practitioners compared withconventional hospital care. J Hum Hypertens. 1990;4:665–670.
30. Earp JA, Ory MG, Strogatz DS. The effects of family involvement andpractitioner home visits on the control of hypertension. Am J PublicHealth. 1982;72:1146–1154.
31. Erfurt JC, Foote A, Heirich MA. Worksite wellness programs: incre-mental comparison of screening and referral alone, health education,follow-up counseling, and plant organization. Am J Health Promot.1991;5:438–448.
32. Erickson SR, Slaughter R, Halapy H. Pharmacists’ ability to influenceoutcomes of hypertension therapy. Pharmacotherapy. 1997;17:140–147.
33. Evans CE, Haynes RB, Birkett NJ, et al. Does a mailed continuingeducation program improve physician performance? Results of a ran-domized trial in antihypertensive care. JAMA. 1986;255:501–504.
34. Foote A, Erfurt JC. Hypertension control at the work site. Comparisonof screening and referral alone, referral and follow-up, and on-sitetreatment. N Engl J Med. 1983;308:809–813.
35. Fortmann SP, Winkleby MA, Flora JA, et al. Effect of long-termcommunity health education on blood pressure and hypertension control.The Stanford Five-City Project. Am J Epidemiol. 1990;132:629–646.
36. Friedman RH, Kazis LE, Jette A, et al. A telecommunications system formonitoring and counseling patients with hypertension. Impact on med-ication adherence and blood pressure control. Am J Hypertens. 1996;9:285–292.
37. Garcao JA, Cabrita J. Evaluation of a pharmaceutical care program forhypertensive patients in rural Portugal. J Am Pharm Assoc (Wash).2002;42:858–864.
38. Garcia-Pena C, Thorogood M, Armstrong B, et al. Pragmatic random-ized trial of home visits by a nurse to elderly people with hypertensionin Mexico. Int J Epidemiol. 2001;30:1485–1491.
39. Goldberg HI, Wagner EH, Fihn SD, et al. A randomized controlled trialof CQI teams and academic detailing: can they alter compliance withguidelines? Jt Comm J Qual Improv. 1998;24:130–142.
40. Gullion DS, Tschann JM, Adamson TE, et al. Management of hyper-tension in private practices: a randomized controlled trial in continuingmedical education. J Contin Educ Health Prof. 1988;8:239–255.
41. Hetlevik I, Holmen J, Kruger O. Implementing clinical guidelines in thetreatment of hypertension in general practice. Evaluation of patientoutcome related to implementation of a computer-based clinical decisionsupport system. Scand J Prim Health Care. 1999;17:35–40.
42. Hill MN, Bone LR, Hilton SC, et al. A clinical trial to improve highblood pressure care in young urban black men: recruitment, follow-up,and outcomes. Am J Hypertens. 1999;12:548–554.
43. Hla KM, Feussner JR, Blessing-Feussner CL, et al. BP control. Improve-ment in a university medical clinic by use of a physician’s associate.Arch Intern Med. 1983;143:920–923.
44. Solomon DK. Clinical and economic outcomes in the hypertension andCOPD arms of a multicenter outcomes study. J Am Pharm Assoc.1998;38:574–585. JC CP:85.
45. Logan AG, Milne BJ, Achber C, et al. Cost-effectiveness of a worksitehypertension treatment program. Hypertension. 1981;3:211–218.
46. Logan AG, Milne BJ, Flanagan PT, et al. Clinical effectiveness andcost-effectiveness of monitoring blood pressure of hypertensive employ-ees at work. Hypertension. 1983;5:828–836.
47. McClellan WM, Craxton LC. Improved follow-up care of hypertensivepatients by a nurse practitioner in a rural clinic. J Rural Health.1985;1:34–41.
48. McGhee SM, McInnes GT, Hedley AJ, et al. Coordinating and stan-dardizing long-term care: evaluation of the west of Scotland shared-carescheme for hypertension. Br J Gen Pract. 1994;44:441–445.
49. Mehos BM, Saseen JJ, MacLaughlin EJ. Effect of pharmacist interven-tion and initiation of home blood pressure monitoring in patients withuncontrolled hypertension. Pharmacotherapy. 2000;20:1384–1389.
50. Midanik LT, Resnick B, Hurley LB, et al. Home blood pressuremonitoring for mild hypertensives. Public Health Rep. 1991;106:85–89.
51. Montgomery AA, Fahey T, Peters TJ, et al. Evaluation of computerbased clinical decision support system and risk chart for management ofhypertension in primary care: randomised controlled trial. BMJ. 2000;320:686–690.
52. Park JJ, Kelly P, Carter BL, et al. Comprehensive pharmaceutical care inthe chain setting. J Am Pharm Assoc (Wash). 1996;NS36:443–451.
53. Sanders KM, Satyvavolu A. Improving blood pressure control in diabe-tes: limitations of a clinical reminder in influencing physician behavior.J Contin Educ Health Prof. 2002;22:23–32.
54. Saunders LD, Irwig LM, Gear JS, et al. A randomized controlled trial ofcompliance improving strategies in Soweto hypertensives. Med Care.1991;29:669–678.
55. Schneider PJ, Larrimer JN, Visconti JA, et al. Role effectiveness of apharmacist in the maintenance of patients with hypertension and con-gestive heart failure. Contemp Pharm Pract. 1982;5:74–79.
56. Siscovick DS, Strogatz DS, Wagner EH, et al. Provider-oriented interven-tions and management of hypertension. Med Care. 1987;25:254–258.
Medical Care • Volume 44, Number 7, July 2006 Evidence Synthesis of Quality Improvement Strategies for Hypertension
© 2006 Lippincott Williams & Wilkins 655
57. Soghikian K, Casper SM, Fireman BH, et al. Home blood pressuremonitoring. Effect on use of medical services and medical care costs.Med Care. 1992;30:855–865.
58. Vivian EM. Improving blood pressure control in a pharmacist-managedhypertension clinic. Pharmacotherapy. 2002;22:1533–1540.
59. Winickoff RN, Wilner S, Neisuler R, et al. Limitations of providerinterventions in hypertension quality assurance. Am J Public Health.1985;75:43–46.
60. Five-year findings of the hypertension detection and follow-up program.II. Mortality by race-sex and age. Hypertension Detection and Follow-upProgram Cooperative Group. JAMA. 1979;242:2572–2577.
61. Dickinson JC, Warshaw GA, Gehlbach SH, et al. Improving hyperten-sion control: impact of computer feedback and physician education. MedCare. 1981;19:843–854.
62. Artinian NT, Washington OG, Templin TN. Effects of home telem-onitoring and community-based monitoring on blood pressure controlin urban African Americans: a pilot study. Heart Lung. 2001;30:191–199.
63. Krishan I, Davis CS, Nobrega FT, et al. The Mayo three-communityhypertension control program. IV. Five-year outcomes of intervention inentire communities. Mayo Clin Proc. 1981;56:3–10.
64. Kizer KW, Demakis JG, Feussner JR. Reinventing VA health care:systematizing quality improvement and quality innovation. Med Care.2000;38:I7–16.
65. Jha AK, Perlin JB, Kizer KW, et al. Effect of the transformation of the
Veterans Affairs health care system on the quality of care. N Engl J Med.2003;348:2218–2227.
66. Hill MN, Han HR, Dennison CR, et al. Hypertension care and control inunderserved urban African American men: behavioral and physiologicoutcomes at 36 months. Am J Hypertens. 2003;16:906–913.
67. Fahey T, Schroeder K, Ebrahim S. Interventions used to improve controlof blood pressure in patients with hypertension. Cochrane Database SystRev. 2005:CD005182.
68. Villagra VG, Ahmed T. Effectiveness of a disease management programfor patients with diabetes. Health Aff (Millwood). 2004;23:255–266.
69. Chodosh J, Morton SC, Mojica W, et al. Meta-analysis: chronic diseaseself-management programs for older adults. Ann Intern Med. 2005;143:427–438.
70. Goldstein MK, Lavori P, Coleman R, et al. Improving adherence toguidelines for hypertension drug prescribing: cluster-randomized con-trolled trial of general versus patient-specific recommendations. Am JManag Care. 2005;11:677–685.
71. Wagner EH, Glasgow RE, Davis C, et al. Quality improvement in chronicillness care: a collaborative approach. Jt Comm J Qual Improv. 2001;27:63–80.
72. Wagner EH, Austin BT, Davis C, et al. Improving chronic illness care:translating evidence into action. Health Aff (Millwood). 2001;20:64–78.
73. Casalino L, Gillies RR, Shortell SM, et al. External incentives, infor-mation technology, and organized processes to improve health carequality for patients with chronic diseases. JAMA. 2003;289:434–441.
Walsh et al Medical Care • Volume 44, Number 7, July 2006
© 2006 Lippincott Williams & Wilkins656
APP
END
IX1
Sear
chSt
rate
gySe
arch
Sear
chSt
ring
Cit
atio
ns†
1T
arge
tsQ
Ist
rate
gies
that
tend
tobe
mul
ti-f
acto
rial
usin
gre
leva
ntM
eSH
term
san
dti
tle
wor
d
Dis
ease
Man
agem
ent
�mh�
OR
Pat
ient
Car
eP
lann
ing
�mh�
OR
Pat
ient
-Cen
tere
dC
are
�mh�
OR
Pri
mar
yH
ealt
hC
are
�mh�
OR
Pro
gres
sive
Pat
ient
Car
e�m
h�O
RC
riti
cal
Pat
hway
s�m
h�O
RD
eliv
ery
ofH
ealt
hC
are,
Inte
grat
ed�m
h�O
RH
ealt
hS
ervi
ces
Acc
essi
bili
ty�m
h�O
RM
anag
edC
are
Pro
gram
s�m
h�O
RP
rodu
ctL
ine
Man
agem
ent
�mh�
OR
Pat
ient
Car
eT
eam
�mh�
OR
Pat
ient
-Cen
tere
dC
are
�mh�
OR
Beh
avio
rC
ontr
ol�m
h�O
RC
ouns
elin
g�m
h�O
RH
ealt
hP
rom
otio
n�m
h�O
RP
atie
ntC
ompl
ianc
e�m
h�O
RA
fter
-Hou
rsC
are
�mh�
OR
((co
ordi
nati
on�t
i�O
Rco
ordi
nate
d�t
i�O
RM
ulti
fact
oria
l�t
i�O
RM
ulti
fact
oria
l�t
i�O
RM
ulti
com
pone
nt�t
i�O
RM
ulti
com
pone
nt�t
i�O
Rm
ulti
disc
ipli
nary
�ti�
OR
mul
tidi
scip
lina
ry�t
i�O
Rin
terd
isci
plin
ary
�ti�
OR
inte
rdis
cipl
inar
y�t
i�O
Rin
tegr
ated
�ti�
OR
com
mun
ity-
base
d�t
i�O
Ror
gani
zed
�ti�
)A
ND
(car
e�t
i�O
Rap
proa
ch�t
i�O
Rin
terv
enti
on�t
i�O
Rst
rate
gy�t
i�O
Rst
rate
gies
�ti�
OR
man
agem
ent
�ti�
OR
man
agin
g�t
i�O
Rce
nter
*�t
i�O
Rcl
inic
*�ti�)
)O
RO
rgan
izat
ion
and
Adm
inis
trat
ion
�mh�
683,
000
2T
arge
tsT
QM
and
CQ
IT
otal
Qua
lity
Man
agem
ent
�mh�
OR
Qua
lity
cont
rol
�mh�
OR
TQ
M�t
i�O
RC
QI
�ti�
OR
(qua
lity
�ti�
AN
D(c
onti
nuou
s�t
i�O
Rto
tal
�ti�
)A
ND
(man
agem
ent
�ti�
OR
impr
ovem
ent
�ti�
))28
,087
3T
arge
tspr
ovid
ered
ucat
ion
Edu
cati
on,
Con
tinu
ing
�mh�
OR
(Edu
cati
on�t
i�A
ND
Con
tinu
ing
�ti�
AN
D(m
edic
al�t
i�O
Rpr
ofes
sion
al*
�ti�
OR
nurs
ing
�ti�
OR
phys
icia
n*�t
i�O
Rnu
rse*
�ti�
))O
R(o
utre
ach
�ti�
AN
D(v
isit
*�ti�
OR
educ
atio
nal
�ti�
)O
R(a
cade
mic
�ti�
AN
Dde
tail
ing
�ti�
))35
,276
4T
arge
tsdi
ffus
ion
ofin
nova
tion
Dif
fusi
onof
Inno
vati
on�m
h�O
R(D
iffu
sion
�ti�
AN
D(I
nnov
atio
n�t
i�O
Rte
chno
logy
�ti�
))48
895
Tar
gets
audi
tan
dfe
edba
ck,
rem
inde
rsy
stem
s,an
dfi
nanc
ial
ince
ntiv
es
Med
ical
audi
t�m
h�O
R((
Aud
it�t
i�O
Rfe
edba
ck�t
i�O
Rco
mpl
ianc
e�t
i�O
Rad
here
nce
�ti�
OR
trai
ning
�ti�
)A
ND
(im
prov
emen
t*�t
i�O
Rim
prov
ing
�ti�
OR
impr
oves
�ti�
OR
impr
ove
�ti�
OR
guid
elin
e*�t
i�O
Rpr
acti
ce*
�ti�
OR
med
ical
�ti�
OR
prov
ider
*�t
i�O
Rph
ysic
ian*
�ti�
OR
nurs
e*�t
i�O
Rcl
inic
ian*
�ti�
OR
prac
tice
guid
elin
es�m
h�O
Rac
adem
ic�t
i�O
Rvi
sit*
�ti�
))O
RR
emin
der
Sys
tem
s�m
h�O
RR
emin
der*
�ti�
OR
((fi
nanc
ial
�ti�
OR
econ
omic
�ti�
OR
phys
icia
n*�t
i�O
Rpa
tien
t*)
AN
Din
cent
ive*
�ti�
)O
RR
eim
burs
emen
tM
echa
nism
s�m
h�
36,8
52
6T
arge
tsin
form
atic
san
dte
lem
edic
ine
Med
ical
Info
rmat
ics
�mh�
OR
com
pute
r�t
i�O
R(d
ecis
ion
�ti�
AN
Dsu
ppor
t�t
i�)
OR
Tel
emed
icin
e�m
h�O
RT
elem
edic
ine
�ti�
OR
tele
com
mun
icat
ion*
�ti�
OR
Inte
rnet
�mh�
OR
web
�ti�
OR
mod
em�t
i�O
Rte
leph
one*
�ti�
OR
tele
phon
e�m
h�30
6,70
3
7C
ombi
nes
1–6
for
over
all
set
ofar
ticl
esre
lati
ngto
QI
1O
R2
OR
3O
R4
OR
5O
R6
988,
356
8C
ombi
nes
over
all
QI
sear
chke
yte
rms
for
arti
cles
invo
lvin
ghy
pert
ensi
on
7A
ND
(Hyp
erte
nsio
n�m
h�O
RH
yper
tens
ion
�ti�
OR
(blo
od�t
i�A
ND
pres
sure
�ti�
))75
74
9Id
enti
fies
subs
etof
8li
kely
toin
volv
eor
igin
alre
sear
chor
syst
emat
icre
view
s
8A
ND
(sys
tem
atic
revi
ewse
arch
stri
ng‡
OR
orig
inal
rese
arch
stri
ng§)
3698
109
Lim
itto
Eng
lish
3144
1110
Lim
itto
Pub
sinc
e19
8029
4212
11B
UT
NO
T(e
dito
rial
�pt�
OR
com
men
t�p
t�O
Rle
tter
�pt�
)28
4213
Add
itio
nal
yiel
dof
jour
nal
sear
ch(8
AN
DJo
urna
lS
earc
hS
trin
g¶)
BU
TN
OT
(9O
Red
itor
ial
�pt�
OR
com
men
t�p
t�O
Rle
tter
�pt�
)�L
imit
edto
Eng
lish
,19
80�
220
14A
ddit
iona
lyi
eld
ofau
thor
sear
ch(8
AN
Dau
thor
sear
ch� )
BU
TN
OT
(13
OR
edit
oria
l�p
t�O
Rco
mm
ent
�pt�
OR
lett
er�p
t�)
�Lim
ited
toE
ngli
sh,
1980
�29
TO
TA
L12
OR
13O
R14
3070
refe
renc
esto
tal
†N
umbe
rsof
cita
tion
sre
flec
tse
arch
resu
lts
from
July
8,20
03.
‡((
met
a-an
alys
is�p
t�O
Rm
eta-
anal
ysis
�tw
�O
Rm
eta-
anal
ysis
�tw
�)O
R((
revi
ew�p
t�O
Rgu
idel
ine
�pt�
OR
cons
ensu
s�t
i�O
Rgu
idel
ine*
�ti�
OR
lite
ratu
re�t
i�O
Rov
ervi
ew�t
i�O
Rre
view
�ti�
OR
Dec
isio
nS
uppo
rtT
echn
ique
s�m
h�)
AN
D((
Coc
hran
e�t
w�
OR
Med
line
�tw
�O
RC
INA
HL
�tw
�O
R(N
atio
nal
�tw
�A
ND
Lib
rary
�tw
�))
OR
(han
dsea
rch*
�tw
�O
Rse
arch
*�t
w�
OR
sear
chin
g�t
w�)
AN
D(h
and
�tw
�O
Rm
anua
l�t
w�
OR
elec
tron
ic�t
w�
OR
bibl
iogr
aphi
*�t
w�
OR
data
base
*O
R(C
ochr
ane
�tw
�O
RM
edli
ne�t
w�
OR
CIN
AH
L�t
w�
OR
(Nat
iona
l�t
w�
AN
DL
ibra
ry�t
w�)
))))
OR
((sy
nthe
sis
�ti�
OR
over
view
�ti�
OR
revi
ew�t
i�O
Rsu
rvey
�ti�
)A
ND
(sys
tem
atic
�ti�
OR
crit
ical
�ti�
OR
met
hodo
logi
c�t
i�O
Rqu
anti
tati
ve�t
i�O
Rqu
alit
ativ
e�t
i�O
Rli
tera
ture
�ti�
OR
evid
ence
�ti�
OR
evid
ence
-bas
ed�t
i�))
)B
UT
NO
T(c
ase
repo
rt�m
h�O
Rca
se*
�ti�
OR
repo
rt�t
i�O
Red
itor
ial
�pt�
OR
com
men
t�p
t�O
Rle
tter
�pt�
)�
38,8
65M
ED
LIN
Ere
cord
s.§R
ando
mis
ed�t
i�O
RR
ando
miz
ed�t
i�O
RC
ontr
olle
d�t
i�O
Rin
terv
enti
on�t
i�O
Rev
alua
tion
�ti�
OR
impa
ct�t
i�O
Ref
fect
iven
ess
�ti�
OR
Eva
luat
ion
�ti�
OR
Stu
dies
�ti�
OR
stud
y�t
i�C
ompa
rati
ve�t
i�O
RF
easi
bili
ty�t
i�O
RP
rogr
am�t
i�O
RD
esig
n�t
i�O
RC
lini
cal
Tri
al�p
t�O
RR
ando
miz
edC
ontr
olle
dT
rial
�pt�
OR
Epi
dem
iolo
gic
Stu
dies
�mh�
OR
Eva
luat
ion
Stu
dies
�mh�
OR
Com
para
tive
Stu
dy�m
h�O
RF
easi
bili
tyS
tudi
es�m
h�O
RIn
terv
enti
onS
tudi
es�m
h�O
RP
rogr
amE
valu
atio
n�m
h�O
RE
pide
mio
logi
cR
esea
rch
Des
ign
�mh�
—
2,55
1,48
6M
ED
LIN
Ere
cord
s.¶N
Eng
lJ
Med
�ta�
OR
JAM
A�t
a�O
RA
nnIn
tern
Med
�ta�
OR
Am
JM
ed�t
a�O
RA
rch
Inte
rnM
ed�t
a�O
RJ
Gen
Inte
rnM
ed�t
a�O
RB
MJ
�ta�
OR
Lan
cet
�ta�
OR
CM
AJ
�ta�
OR
Cli
nIn
vest
Med
�ta�
OR
Arc
hF
amM
ed�t
a�O
RJ
Fam
Pra
ct�t
a�O
RF
amP
ract
�ta�
OR
Ann
Med
�ta�
OR
Br
JG
enP
ract
�ta�
OR
JIn
tern
Med
�ta�
OR
Med
JA
ust
�ta�
OR
Sou
thM
edJ
�ta�
OR
Wes
tJ
Med
�ta�
OR
Aus
tN
ZJ
Med
�ta�
OR
Med
Car
e�t
a�O
RH
ealt
hS
erv
Res
�ta�
OR
Inqu
iry
�ta�
OR
Mil
bank
Q�t
a�O
RH
ealt
hA
ff(M
illw
ood)
�ta�
OR
Hea
lth
Car
eF
inan
cR
ev�t
a�O
RM
edC
are
Res
Rev
�ta�
OR
eff
clin
prac
t�t
a�O
Rev
alhe
alth
prof
�ta�
OR
JtC
omm
JQ
ual
Impr
ov�t
a�O
RQ
ual
Saf
Hea
lth
Car
e�t
a�O
RIn
tJ
Qua
lH
ealt
hC
are
�mh�
OR
Qua
lH
ealt
hC
are
�ta�
OR
Qua
lH
ealt
hR
es�t
a�O
RR
epM
edG
uide
lO
utco
mes
Res
�ta�
OR
Am
JM
anag
Car
e�t
a�O
RA
mJ
Med
Qua
l�t
a�O
RJ
Con
tin
Edu
cH
ealt
hP
rof
�ta�
OR
Pre
vM
ed�t
a�O
RA
mJ
Pre
vM
ed�t
a�O
RP
atie
ntE
duc
Cou
ns�t
a�O
RA
nnB
ehav
Med
�ta�
OR
JH
umH
yper
tens
�ta�
OR
Hyp
erte
nsio
n�t
a�O
RA
mJ.
� (Ber
wic
kD
�au�
OR
berl
owit
zd
�au�
OR
davi
sd
�au�
OR
kief
ec
�au�
OR
wag
ner
e�a
u�O
Rgl
asgo
wr
�au�
OR
bodd
enhe
imer
t�a
u�O
RH
ulsc
her
M�a
u�O
Rgr
olr
�au�
OR
grim
shaw
j�a
u�O
Rha
ynes
b�a
u�O
Rha
ynes
rb�a
u�O
Rsa
cket
td
�au�
OR
gold
berg
h�a
u�O
RH
irsc
hI
�au�
OR
nash
d�a
u�O
Rro
per
w�a
u�O
Rw
eing
arte
ns
�au�
)—
6401
ME
DL
INE
reco
rds.
Medical Care • Volume 44, Number 7, July 2006 Evidence Synthesis of Quality Improvement Strategies for Hypertension
© 2006 Lippincott Williams & Wilkins 657