2. American Heart Journal
Volume 157, Number 3
Decker et al 557
Methods Table I. Frequency of risk factors recalled from instructions
Patient sample received at or since hospital discharge
Acute myocardial infarction patients (N = 2,498) were Recall
consecutively recruited between January 1, 2003, and June 28,
2004, into the Prospective Registry Evaluating Myocardial Risk factor Yes No
Infarction: Events and Recovery (PREMIER) health status
study.10 Of 10,911 patients screened, 3,953 were eligible; and Medication 1663 (88.3%) 221 (11.7%)
2,498 subsequently consented and were enrolled. This 19-center Diet 1067 (65.1%) 573 (34.9%)
national registry included baseline data of chart abstractions Whom to call 1031 (71.5%) 410 (28.5%)
(presentation, clinical comorbidities, in-hospital treatments, Cardiac rehabilitation 824 (78.5%) 226 (21.5%)
discharge medications, discharge instructions, etc) and inter- Exercise 745 (70.9%) 306 (29.1%)
views by trained data collectors within 24 to 72 hours of Smoking 457 (76.3%) 142 (23.7%)
Cholesterol therapy 252 (40.5%) 371 (59.5%)
admission. Each participating hospital obtained Institutional
Diabetes management 182 (65.2%) 97 (34.8%)
Research Board approval, and patients signed an informed Cholesterol check 126 (39.1%) 196 (60.9%)
consent form for baseline and follow-up interviews. Warfarin 99 (75%) 33 (25%)
Weight management 76 (49%) 79 (50.1%)
Outcomes assessment Weight loss 29 (48.3%) 31 (51.7%)
Patients' general health status was measured by the Short Form The individual risk factor must have been documented as provided to the patient before
(SF)–12 Physical Component Scale (PCS). A score of 50 reflects hospital discharge.
the population average, and a 10-point deviation represents 1
SD.11 Disease-specific health status was assessed with the Seattle and the 12-month follow-up interviews were used to assess
Angina Questionnaire (SAQ), a 19-item disease-specific ques- patients' health status outcomes using the SF-12 and SAQ.
tionnaire. The SAQ Angina Frequency and Quality of Life (QoL)
scales were used as outcomes in this study, with SAQ scores
Additional variables
ranging from 0 to 100, where higher scores represent fewer Patients were also asked, on the baseline interview, whether
symptoms and better quality of life.12-14 A mean difference of N5 they avoided obtaining medical care because of cost (yes/
points is considered clinically significant.12 no),15 about the prevalence of depressive symptoms (using the
Patients' health status recovery was quantified through Patient Health Questionnaire [PHQ] score16,17), and about their
1-month and 12-month telephone interviews conducted by an social support (using the Enhancing Recovery in Coronary
experienced, central call center. The 30-minute phone inter- Heart Disease Social Support Instrument [ESSI]18). The avoid-
views included questions about treatment after discharge ing care question was used as a proxy for reported income,
(including hospitalizations, diagnostic tests, procedures, medi- which was missing on 39% of the baseline patient interviews
cations, and outpatient visits) since their last study contact. because of sensitivity about answering the question, and has
Mortality was determined through the Social Security Adminis- been reported in the past as a predictor of poor outcomes.19
tration Death Master File. The PHQ assesses the presence of 9 depressive symptoms; and
The baseline case report form abstracted from the medical the severity index ranges from 0 to 27, with a PHQ score ≥10
record which of the 13 discharge instructions were docu- defined as moderate to severe depression, representing the
mented as being provided to the patient (exercise, medication minimum number of symptoms required for the diagnosis of
adherence, diet modification, smoking cessation, weight major depression.20 The ESSI is a 7-item questionnaire
monitoring and loss, follow-up plans, to call a physician for assessing patients' social network for support and assistance.21
recurrent symptoms, cardiac rehabilitation, cholesterol mon-
itoring, lipid therapy, diabetes management, and warfarin use). Statistical analysis
Afterward, at 1 month, patients were asked if they had The frequency with which patients recalled RFM advice
received instructions at, or since, discharge on any 1 of the was determined, along with the rate of very careful
13 RFM items and how well they had followed these adherence to the individual items. Descriptive demographic,
instructions. Responses included “very carefully,” “fairly well,” clinical, and treatment data for patients reporting adherence
“somewhat,” “not at all,” or “not able to do for other reasons.” to all RFM instructions provided, across the 4 adherence
To assess the degree to which an individual patient adhered to groups, were compared with Cochran-Armitage trend test for
RFM, we a priori defined adherence as the percentage of categorical data and analysis of variance trend tests for
relevant activities for which the patient reported “very continuous data.
carefully.” Only those RFMs that were relevant and documen- To identify the independent association of adherence at
ted at baseline for that patient were included in the 1 month on 12-month outcomes, multivariable analyses were
denominator (ie, only diabetic patients were included in the performed. All multivariable models included age, sex, white
assessment of diabetes management, only smokers were race, marital status, education Nhigh school, body mass index,
considered for the smoking cessation advice, etc). We then currently smoking, medical insurance, avoid care because of
summarized patients' reports of adherence into the following cost, ESSI social support score, depression (PHQ score ≥10),
4 classifications: poor (meaning that the patient adhered very history of diabetes, lung disease, hypercholesterolemia, con-
carefully to b49% of their RFMs; 0%-49%), partial (50%-74%), gestive heart failure, hypertension, prior MI, prior percuta-
careful (75%-99%), and very careful (100%). The 1-month neous coronary intervention, prior coronary artery bypass
responses were used to classify patients' reported adherence, graft, ST elevation MI, revascularization during hospitalization,
3. American Heart Journal
558 Decker et al March 2009
Table II. Baseline characteristics by category of reported adherence
Very careful (100%) Careful (75%-99%) Partial (50%-74%) Poor (0%-49%)
n = 393 n = 612 n = 677 n = 364 P value
Sociodemographic
Age (mean ± SD), y 64.6 ± 14 y 59.8 ± 11.6 y 60.1 ± 12.8 y 58.6y ± 12.4 y b.001
Gender (male) 257 (65.4%) 419 (68.5%) 464 (68.5%) 247 (67.9)% .477
Race (white) 305 (78%) 485 (79.4) 523 (77.6%) 266 (73.5%) .107
Married 233 (60.8%) 408 (67.4%) 429 (64.5%) 211 (58.8%) .384
Education (Nhigh school) 307 (80.8%) 496 (82.3%) 527 (78.7%) 289 (80.7%) .500
Low social support (BL)⁎ 41 (11%) 91 (15.4%) 95 (14.6%) 75 (21.4%) b.001
Self-reported economic burden
Avoid care because of cost 44 (11.6%) 109 (18.1%) 123 (18.5%) 75 (21%) .001
Payor:none/self-pay 29 (7.7%) 62 (10.6%) 85 (13.4%) 62 (18%) b.001
Clinical comorbidities
Final diagnosis:
STEMI 169 (43%) 310 (50.7%) 315 (46.5%) 153 (42%) .465
NSTEMI 224 (57%) 302 (49.3%) 362 (53.5%) 211 (58%) .465
Prior MI 74 (18.8%) 109 (17.8%) 145 (21.4%) 83 (22.8%) .062
Prior PCI 73 (18.6%) 95 (15.5%) 143 (21.1%) 69 (19%) .252
Prior CABG 54 (13.7%) 73 (11.9%) 91 (13.4%) 42 (11.5%) .612
Congestive heart failure 42 (10.7%) 55 (9%) 60 (8.9%) 35 (9.6%) .596
Depression (BL)† 70 (18.9%) 108 (18.6%) 141 (22.2%) 82 (23.6%) .043
Diabetes 97 (24.7%) 158 (25.8%) 177 (26.1%) 114 (31.3%) .057
Hypertension 242 (61.6%) 366 (59.8%) 433 (64%) 222 (61%) .638
Hypercholesterolemia 179 (45.5%) 335 (54.7%) 333 (49.2%) 173 (47.5%) .841
COPD 44 (11.2%) 51 (8.3%) 100 (14.8%) 49 (13.5%) .022
Current smoker 74 (19.2%) 193 (31.7%) 244 (36.2%) 161 (44.5%) b.001
Body mass index (kg/m2) 28.5 ± 6.0 29.3 ± 6.0 29.5 ± 6.5 30 ± 6.6 .001
Revascularization .856
PCI 119 (30.3%) 195 (31.9%) 235 (34.7%) 126 (34.6%)
CABG 43 (10.9%) 84 (13.7%) 68 (10%) 38 (10.4%)
Medical management 231 (58.8%) 333 (54.4%) 374 (55.2%) 200 (54.9%)
Nitrate medication (1 m) 89 (68.5%) 115 (63.5%) 151 (67.1%) 58 (48.3%) .008
β-Blocker medication (1 m) 264 (78.1%) 437 (80.3%) 482 (80.5%) 254 (77.9%) .997
SAQ Angina Y/N (BL) 192 (49.5%) 310 (50.8%) 366 (54.1%) 221 (60.9%) b.001
SAQ QoL (BL) 64 ± 22.7 64.5 ± 23.2 61.9 ± 23.4 61.5 ± 23.6 .052
SAQ QoL (1 y) 88.1 ± 16.5 85.7 ± 16.4 84 ± 18.4 82.1 ± 19.6 b.001
SF-12v2 PCS (BL) 43.9 ± 12.2 44.4 ± 12.2 42.8 ± 12.3 43.1 ± 12.8 .172
SF-12v2 PCS (1 y) 46 ± 11.5 46.5 ± 11.2 43.4 ± 12 43.8 ± 11.3 .001
SF-12v2 Mental Component Score (BL) 51.5 ± 11.4 50.2 ± 11.1 49.8 ± 11.4 47.6 ± 12 b.001
SF-12v2 Mental Component Score (1 y) 54.6 ± 8.4 53.7 ± 9.1 53.6 ± 9.4 52.1 ± 9.6 .001
BL, Baseline; STEMI, ST-segment elevation MI; NSTEMI, non–ST-segment elevation MI; PCI, percutaneous coronary intervention; CABG, coronary artery bypass grafting; COPD,
chronic obstructive pulmonary disease.
⁎ Based on the ESSI.
† Based on the PHQ.
β-blocker use at 1 month, nitrate use at 1 month, number of model.22 The SAQ QoL and SF-12 PCS were modeled using
risk factors applicable to each patient, and baseline health multivariable hierarchical linear regression.
status corresponding to each particular outcome. Lower One-year mortality and rehospitalization were modeled using
adherence categories were compared with the reference level multivariable proportional hazards regression stratified by site.
of 100% adherence for effect size reporting, and linear trend Restricted cubic spline terms were added to models for all
tests were used to calculate the P value for trend across all continuous variables to account for possible nonlinearity, and the
4 adherence groups for each outcome. hierarchical model structures used also accounted for correlation
Because of its skewed distribution (70% of patients of patients within site.23 Power was calculated for each outcome
reported no angina), SAQ Frequency scores were dichot- using PASS software (PASS 2008 version 08.0.05, www.ncss.
omized into any angina (scores b100) or no angina (scores = com). There was N80% power to detect a 5% difference for
100) and were modeled using multivariable hierarchical mortality and presence of angina between the lowest- and the
modified Poisson regression. Although typical analyses often highest-adherence group. There also was N80% power to detect a
use logistic regression to estimate adjusted odds ratios, these 5-point mean difference in the SAQ QoL and SF-12 PCS as well as
may not provide accurate representations of relative risks 80% power to detect a 10% difference in all-cause rehospitaliza-
when the outcomes are common. In this study, those events tion. All tests for statistical significance were 2-tailed with an α
that occurred in N25% of patients had their adjusted relative level of .05. Analyses were conducted using SAS software 9.1
risks estimated directly using a modified Poisson regression (SAS Institute, Cary, NC) and R version 2.1.1.24
4. American Heart Journal
Volume 157, Number 3
Decker et al 559
Missing data Figure 1
The primary analyses included patients who participated in
1-month follow-up interviews. Of the 2,498 patients enrolled
in PREMIER, 47 died before the 1-month assessment, 122
were contacted but refused an interview, 233 were lost to
follow-up, and 50 had incomplete RFM data. Thus, 2,046
patients had a 1-month follow-up interview that was
analyzed. Of these eligible patients, 84% provided 1-year
health status follow-up.
Missing information on one or more covariates was present
for 144 (7%) patients; 99 (4.8%) were missing N1 value. Missing
covariate data were assumed to be missing at random and were
imputed using multiple imputation methods to allow incor-
poration of all patients and to correctly account for uncertainty
due to missingness.23 The imputation model included the full
array of demographic, socioeconomic, patient history, treat-
ment, and all quality of life subscales.
To assess potential bias due to unavailable follow-up, we
created a nonparsimonious model of the propensity to be
missing a 1-year interview.25 For those patients who refused
1-year interviews or could not be contacted, propensity scores
were computed using logistic regression analyses to predict
their likelihood of unsuccessful follow-up. Predictor variables
included demographics, socioeconomic and lifestyle factors, “Very careful” reported adherence to individual RFM items.
clinical characteristics, vital signs and laboratory studies,
disease severity, baseline health status, medications, and acute
and nonacute treatments received during patients' initial AMI cholesterol management or weight loss recalled receiving
hospitalization. From these models, a probability of failure to these instructions.
complete an interview was calculated. The reciprocal of this
probability to complete an interview was then used as a Patient characteristics associated with RFM adherence
weight in the multivariable regression analyses to weight
Baseline characteristics of patients who recalled RFMs
those patients with available data with similar patient
and the percentage of reported adherence are reported in
characteristics as the patients who were lost to follow-up
more heavily. This method assesses potential observable bias Table II. Responses indicated that patients who very
from those lost to follow-up by overrepresenting the patient carefully adhered to their RFM were less likely to be current
type that is more likely to be lost to follow-up.25 The smokers (19% vs 45%), were older (64.6 vs 58.6 years), and
propensity weighting did not change the clinical interpreta- reported higher levels of social support (89% vs 79%) as
tion or significance of the results and were comparable with compared with those whose adherence scores were b50%
the primary data, suggesting little observable bias associated (P b .01 for all). Patients with greater adherence reported
with loss to follow-up. Accordingly, only the unweighted continuing their nitrate medication more often then the
primary data are reported. poorly adherent patients (68.5% vs 48.3%, trend P = .008).
A significant finding was that patients who reported
Funding for the PREMIER Registry was through CVTherapeu- avoiding care because of cost were less likely to report
tics, Palo Alto, CA.
adhering very carefully to RFM instructions.
Data collected on the frequency patients reported
Results adhering very carefully to RFMs demonstrated that most
Prevalence of RFM recall patients (82%) very carefully adhered to N50% of discharge
Overall, PREMIER patients were, on average, 61 years instructions given to them. Nineteen percent (393/2,046)
old, male (67%), and white (74%). The frequency of recall very carefully adhered to all instructions given.
of individual instructions they received at discharge or
since is reported in Table I. Eligible patients were those Prevalence of RFM adherence
with documentation at baseline as having received the Strong reported adherence at 1 month occurred most
individual RFM instruction. The most frequently docu- frequently with “taking medications as prescribed” and
mented discharge instruction was medication directions, “warfarin use” (94% and 86%, respectively) (Figure 1).
although only 88% (1,663/1,884) recalled receiving the Diet instructions was the second most commonly
instruction. The second most common RFM was diet documented instruction on the medical record, as
(n = 1,640), although only 65% recalled receiving this described above, with a 65.1% recall rate, but was
instruction during the 1-month interview. Less than half reported as being adhered to very carefully by only 51% of
of the patients who had received instructions about the patients. The least frequent RFM adhered to very
5. American Heart Journal
560 Decker et al March 2009
Table III. Summary of adjusted effect estimates of 12-month health status outcomes
Incidence of angina SAQ QOL SF-12 PCS
1-m reported very
careful adherence Trend Mean Trend Mean Trend
to RFMs RR P value difference P value difference P value
100% – .015 – .173 – .049
75%-99% 1.39 (0.85, 2.25) 0.10 (−2.40, 2.59) 0.43 (−1.11, 1.88)
50%-74% 1.53 (1.00, 2.33) −1.01 (−3.46, 1.43) −1.51 (−2.97, −0.05)
b50% 1.58 (1.05, 2.37) −1.62 (−4.40, 1.16) −1.08 (−2.76, 0.59)
All models included age, sex, white race, marital status, education Nhigh school, body mass index, currently smoking, medical insurance, avoid care because of cost, ESSI social
support score, depression (PHQ score ≥10), history of diabetes, lung disease, hypercholesterolemia, congestive heart failure, hypertension, prior MI, prior percutaneous coronary
intervention, prior coronary artery bypass graft, ST elevation MI, revascularization during hospitalization, β-blocker use at 1 month, nitrate use at 1 month, number of risk factors
applicable to each patient, and baseline health status corresponding to each particular outcome.
carefully was “losing weight” (43%) and cardiac rehabi- recommendations. Potential explanations might include
litation (33%). Incidentally, though, individuals who that the patients were preoccupied during discharge, the
reported they were “very carefully” adherent to cardiac patients were given written material that they could not
rehabilitation participation were also “very carefully” read or understand, the provision of instructions went to
adherent to the other RFMs that they were eligible for. family members, or no instructions were actually
presented despite documentation to the contrary.
Association of RFM adherence with health We found that patients who reported stronger adher-
status outcomes ence to RFM were more likely to not have angina 1 year
In multivariable models adjusting for sociodemo- after their AMI, although other health status outcomes
graphic characteristics, β-blocker and nitrate use at 1 were not found to be associated with RFM adherence.
month, clinical differences, and angina symptoms at 1 Although patients who continued their nitrate medication
month, patients who reported being b50% adherent were more often were also those who more closely adhered to
68% more likely to report angina at 1 year versus those RFMs, this minimally changed the multivariable model
with scores of 100% (relative risk [RR] 1.68, 95% CI 1.08- estimates, thus not explaining all the difference in
2.64, trend P = .01). The addition of depression severity reported angina. To date, we did not identify any studies
and social support to the multivariable model did not correlating patient adherence to RFM instructions and
attenuate our estimates; and thus, only the fully adjusted their 1-year angina incidence. For example, the Lifestyle
models are displayed in Table III (RR 1.58, 95% CI 1.05- Heart Trial demonstrated a correlation between intensive
2.37, trend P = .015). lifestyle change and the regression of coronary athero-
There was no independent effect of RFM reported sclerosis without assessing patient health status.2 The
adherence on quality of life, physical functioning, current study extends such work by examining adherence
rehospitalization, or mortality after adjusting for all in routine clinical care. Follow-up of patient health status
covariates. Although a small mean difference of 1.5 points for a greater length of time than 1 year would be important
(trend P = .049) for the SF-12 PCS was found for the to study, as this may not be sufficient time for potential
partially adherent group versus those who adhered very beneficial effect of RFM.
carefully to their RFMs, this would not be considered Several patient characteristics were observed to be
clinically meaningful. associated with reported lower adherence to RFM
instructions, including younger age and lower social
Discussion support. This latter finding is congruent with a study by
In light of the importance of risk factor modification on Conn et al26 that found that the presence of social support
secondary prevention after AMI, this study examined creates a significant difference in patients' behaviors
patients' recall of RFM instructions and their reported related to cardiovascular health. They demonstrated a
compliance with these recommendations. Our findings direct effect between social support and MI, a finding
do not necessarily support the impact RFM has on supported by our more contemporary investigation.
previously reported patient outcomes. This is the first Our study also confirms previous observations of
study, of which we are aware, to document the marked important patient characteristics and RFM adherence.
variation in the types of RFM recalled and adhered to by Previous cardiovascular studies have shown that the cost
AMI patients. We found that there were many RFMs that of medications and related health care is one of the
patients did not recall receiving the instruction regarding, potential reasons for poor medication-taking behavior.27
such as cholesterol monitoring and management. This Our study found that patients who avoid care because of
implies that, despite documentation, there is a deficiency cost also reported lower adherence to discharge
in recall that patients may not be “receiving” these instructions. We also validated previous observations
6. American Heart Journal
Volume 157, Number 3
Decker et al 561
that current smokers and those with significant depres- References
sive symptoms have poor adherence to RFM instruc- 1. Aldana SG, Whitmer WR, Greenlaw R, et al. Cardiovas-
tions, although they did not strongly influence the cular risk reductions associated with aggressive lifestyle
multivariable model.26,28 modification and cardiac rehabilitation. Heart Lung 2003;32:
All RFMs are not of equal importance. Perhaps, the RFM 374-82.
2. Ornish D, Scherwitz LW, Billings JH, et al. Intensive lifestyle
with the strongest association with improved outcomes is
changes for reversal of coronary heart disease. JAMA 1998;280:
cardiac rehabilitation, a method of simultaneous sup-
2001-7.
porting many RFMs. Cardiac rehabilitation serves as a 3. American Heart Association. Available at: http://americanheart.
source of education and reenforces crucial life-changing org/downloadable/heart/1140534985281Statsupdate06book.
habits for improved survival after AMI.29-31 In fact, the pdf. Last accessed September 18, 2006.
evidence is so strong in support of cardiac rehabilitation 4. Makaryus AN, Friedman EA. Patients' understanding of their
that it has recently been endorsed as a performance treatment plans and diagnosis at discharge. Mayo Clin Proc 2005;
measure of quality among patients recovering from 80:991-4.
AMI.31 Prior research has shown a survival benefit from 5. Hayes KS. Literacy for health information of adult patients and
caregivers in a rural emergency department. Clin Excell Nurse Pract
cardiac rehabilitation and improvement in quality of life
2000;4:35-40.
with decreased frequency of angina.32,33 6. Sdringola S, Nakagawa K, Nakagawa Y, et al. Combined intense
Our findings should be considered in light of several lifestyle and pharmacologic lipid treatment further reduce coronary
potential limitations. First, only 84% of patients partici- events and myocardial perfusion abnormalities compared with usual-
pated in the 1-month follow-up. Although extensive care cholesterol-lowering drugs in coronary artery disease. J Am Coll
propensity-based models for incomplete follow-up sug- Cardiol 2003;41:263-72.
gested no observable bias secondary to patients being 7. Smith Jr SC, Allen J, Blair SN, et al. AHA/ACC guidelines for
lost to follow-up, residual confounding cannot be secondary prevention for patients with coronary and other athero-
definitively excluded. A second potential concern is that sclerotic vascular disease: 2006 update: endorsed by the National
Heart, Lung, and Blood Institute. Circulation 2006;113:2363-72.
this is an observational registry that is exploratory in
8. Bradley EH, Herrin J, Elbel B, et al. Hospital quality for acute
nature and thus cannot definitively state that reported
myocardial infarction: correlation among process measures and
adherence to RFM will improve patient health status relationship with short-term mortality. JAMA 2006;296:72-8.
outcomes. The appropriateness of RFMs delivered during 9. Joint Commission. Available at: http://www.jointcommission.org.
hospitalization was not determined, only if it was Last accessed January 21, 2008.
documented. In addition, because the data were 10. Spertus JA, Peterson E, Rumsfeld JS, et al. The Prospective Registry
collected through interview and not direct observation, Evaluating Myocardial Infarction: Events and Recovery (PREMIER)—
patients may overreport their adherence to RFM prac- evaluating the impact of myocardial infarction on patient outcomes.
tices, although self-report is acceptable as an indirect Am Heart J 2006;151:589-97.
11. Ware Jr J, Kosinski M, Keller SD. A 12-item short-form health survey:
measure for many outcomes.34
construction of scales and preliminary tests of reliability and validity.
Med Care 1996;34:220-33.
12. Spertus JA, Winder JA, Dewhurst TA, et al. Development and
Clinical implications evaluation of the Seattle Angina Questionnaire: a new functional
We found that patient recall of RFMs that were status measure for coronary artery disease. J Am Coll Cardiol 1995;
documented as having been instructed is less than ideal, 25:333-41.
ranging from 39% to 88%. Patient's reported adherence to 13. Spertus JA, Winder JA, Dewhurst TA, et al. Monitoring the quality of life
the recalled risk factor was even more distressing. in patients with coronary artery disease. Am J Cardiol 1994;74:1240-4.
Variations in reported adherence to different RFM items 14. Spertus JA, Jones P, McDonell M, et al. Health status predicts long-
term outcome in outpatients with coronary disease. Circulation 2002;
were found, yet moderate or poor adherence to RFMs
106:43-9.
was associated with greater reported angina at 1 year.
15. Spertus J, Decker C, Woodman C, et al. Effect of difficulty affording
Changing lifelong habits and overcoming nonmodifiable health care on health status after coronary revascularization.
barriers such as age or gender requires motivation and Circulation 2005;111:2572-8.
encouragement along with individualized patient educa- 16. Spitzer RL, Kroenke K, Williams JB. Validation and utility of a self-
tion. More research is needed to explore if these findings report version of PRIME-MD: the PHQ primary care study. Primary
are seen in larger cardiac populations, as well as studying Care Evaluation of Mental Disorders. Patient Health Questionnaire.
methods to improve RFM processes for AMI patients, all JAMA 1999;282:1737-44.
targeted to increase adherence and optimize their 17. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief
depression severity measure. J Gen Intern Med 2001;16:606-13.
health status.
18. Enhancing recovery in coronary heart disease patients (ENRICHD):
study design and methods. The ENRICHD investigators. Am Heart J
2000;139:1-9.
Disclosures 19. Rahimi AR, Spertus JA, Reid KJ, et al. Financial barriers to health care
Dr Spertus owns the intellectual property rights to the and outcomes after acute myocardial infarction. JAMA 2007;297:
Seattle Angina Questionnaire. 1063-72.
7. American Heart Journal
562 Decker et al March 2009
20. American Psychiatric Association Committee on Nomenclature and 29. Ades PA. Cardiac rehabilitation and secondary prevention
Statistics. Diagnostic and statistical manual of mental disorders. 4th of coronary heart disease. N Engl J Med 2001;345:
ed Rev. Washington, DC: American Psychiatric Association; 1994. 892-902.
21. Vaglio Jr J, Conard M, Poston WS, et al. Testing the performance of 30. Williams MA, Ades PA, Hamm LF, et al. Clinical evidence for a health
the ENRICHD social support instrument in cardiac patients. Health benefit from cardiac rehabilitation: an update. Am Heart J 2006;
Qual Life Outcomes 2004;2:24. 152:835-41.
22. Zou G. A modified Poisson regression approach to prospective 31. Thomas RJ, King M, Lui K, et al. AACVPR/ACC/AHA 2007
studies with binary data. Am J Epidemiol 2004;159:702-6. performance measures on cardiac rehabilitation for referral to and
23. Harrell FE. Regression modeling strategies with applications to linear delivery of cardiac rehabilitation/secondary prevention services
models, logistic regression and survival analysis. New York: endorsed by the American College of Chest Physicians, American
Springer-Verlag; 2001. College of Sports Medicine, American Physical Therapy Associa-
24. R version 2.6.0 (R Development Core Team). R: a language and tion, Canadian Association of Cardiac Rehabilitation, European
environment for statistical computing. Vienna, Austria: R Foundation Association for Cardiovascular Prevention and Rehabilitation, Inter-
for Statistical Computing; 2006. American Heart Foundation, National Association of Clinical Nurse
25. Lunceford JK, Davidian M. Stratification and weighting via the Specialists, Preventive Cardiovascular Nurses Association, and the
propensity score in estimation of causal treatment effects: a Society of Thoracic Surgeons. J Am Coll Cardiol 2007;50:
comparative study. Stat Med 2004;23:2937-60. 1400-33.
26. Conn VS, Taylor SG, Abele PB. Myocardial infarction survivors: age 32. Sundararajan V, Bunker SJ, Begg S, et al. Attendance rates and
and gender differences in physical health, psychosocial state and outcomes of cardiac rehabilitation in Victoria, 1998. Med J Aust
regimen adherence. J Adv Nurs 1991;16:1026-34. 2004;180:268-71.
27. Burnier M. Medication adherence and persistence as the cornerstone of 33. Roman O, Gutierrez M, Luksic I, et al. Cardiac rehabilitation after
effective antihypertensive therapy. Am J Hypertens 2006;19:1190-6. acute myocardial infarction. 9-year controlled follow-up study.
28. Lavigne M, Rocher I, Steensma C, et al. The impact of smoking on Cardiology 1983;70:223-31.
adherence to treatment for latent tuberculosis infection. BMC Public 34. Osterberg L, Blaschke T. Adherence to medication. N Engl J Med
Health 2006;6:66. 2005;353:487-97.
The following article is an AHJ Online Exclusive.
Full text of this article is available at no charge at our website:
www.ahjonline.com
Coronary Artery Disease
Assessment of P2Y12 inhibition with the point-of-care device
VerifyNow P2Y12 in patients treated with prasugrel or clopidogrel
coadministered with aspirin
Christoph Varenhorst, MD, a Stefan James, MD, PhD, a David Erlinge, MD, PhD, b Oscar O. Braun, MD, PhD, b
¨
John T. Brandt, MD, c Kenneth J. Winters, MD, c Joseph A. Jakubowski, PhD, c Sylvia Olofsson, MSci, a
Lars Wallentin, MD, PhD, a Agneta Siegbahn, MD, PhD, d Uppsala and Lund, Sweden; and Indianapolis, IN
Background Variability in response to thienopyridines has led to Results Dose- and time-dependent inhibition of P2Y12 was evident
the development of point-of-care devices to assess adenosine diphosphate with VN-P2Y12. There was strong correlation with VN-P2Y12 and VASP or
(ADP)-induced platelet aggregation. These tests need to be evaluated in LTA for all treatments through a wide range of P2Y12 function. At high levels
comparison to reference measurements of P2Y12 function during different of P2Y12 inhibition, platelet function measured by VN-P2Y12 was maximally
thienopyridine treatments. inhibited and could not reflect further changes seen with VASP or LTA
methods. Correlation was also observed between exposure to clopidogrel's
Methods After a run-in on 75 mg aspirin, 110 subjects were
active metabolite and VN-P2Y12 during MD and LD, whereas it was
randomized to double-blind treatment with clopidogrel 600 mg loading
observed only with prasugrel MD.
dose (LD)/75 mg maintenance dose (MD) or prasugrel 60 mg LD/10 mg
MD. Antiplatelet effects were evaluated by VerifyNow P2Y12 (VN-P2Y12) Conclusion The VN-P2Y12 correlated strongly with inhibition of
device (Accumetrics, San Diego, CA), vasodilator-stimulated phosphopro- P2Y12 function, as measured with either VASP or LTA. VN-P2Y12 also
tein (VASP) phosphorylation assay, and light transmission aggregometry correlated to exposure to the active metabolite of prasugrel and clopidogrel
(LTA). Prasugrel's and clopidogrel's active metabolite concentration were up to levels associated with assumed saturation of the P2Y12 receptor.
also determined. (Am Heart J 2009;157:562.e1-562.e9.)