Adverse Outcomes of Underuse of β-Blockers in Elderly Survivors of Acute Myocardial Infarction

Objectives.\p=m-\To study determinants and adverse outcomes (mortality and re-hospitalization) of \g=b\-blocker underuse in elderly patients with myocardial infarction; and whether the relative risks (RRs) of survival associated with \g=b\-blocker use were comparable to those reported in the large randomized controlled trials (RCTs). Setting.\p=m-\New Jersey Medicare population. Design.\p=m-\Retrospective cohort design using linked Medicare and drug claims data from 1987 to 1992. Patients.\p=m-\Statewide cohort of 5332 elderly 30-day acute myocardial infarction (AMI) survivors with prescription drug coverage, of whom 3737 were eligible for \g=b\-blockers. Main Outcome Measures.\p=m-\\g=b\-Blocker and calcium channel blocker use in the first 90 days after discharge and mortality rates and cardiac hospital readmissions over the 2-year period after discharge, controlling for sociodemographic and base-line risk variables. Results.\p=m-\Only 21% of eligible patients received \g=b\-blockertherapy; this rate remained unchanged from 1987 to 1991. Patients were almost 3 times more likely to receive a new prescription for a calcium channel blocker than for a new \g=b\-blocker after their AMIs. Advanced age and calcium channel blocker use predicted under-use of \g=b\-blockers.Controlling for other predictors of survival, the mortality rate among \g=b\-blocker recipients was 43% less than that for nonrecipients (RR=0.57; 95% confidence interval [CI], 0.47-0.69). Effects on mortality were substantial in all age strata (65-74 years, 75-84 years,

Adverse Outcomes of Underuse of \g=b\-Blockers in Elderly Survivors of Acute Myocardial Infarction ß-BLOCKER prophylaxis after acute myocardial infarction (AMD is one of the most scientifically substantiated, cost-effective preventive medical ser¬ vices.1 Multiple randomized controlled trials (RCTs), involving over 20 000 pa¬ tients, have shown that ß-blocker use following AMI decreases cardiovascu¬ lar mortality and reinfarctions and in¬ creases the chances of survival by 20% to 40%.2"6This evidence has led national cardiology consensus committees to strongly recommend their use in eligible populations of AMI patients.7However, For editorial comment see 155.
few data exist on rates and determi¬ nants of prescription of ß-blockers, es¬ pecially in community settings and among elderly patients who have been underrepresented in RCTs.8Although 80% of all deaths due to AMI occur in the elderly, virtually no patients older than 75 years have been included in RCTs of ß-blockers.8"10In this cohort study, we linked several large adminis¬ trative databases on survival and use of inpatient and outpatient health care ser¬ vices to measure levels, determinants, and outcomes of prescribing ß-blockers among a community population of 5332 elderly 30-day survivors of AMI in New Jersey.
Although claims databases can iden¬ tify variations in practice patterns11•12 and evaluate quality-of-care effects of cost- containment policies,13,14 the validity of their use in assessing the outcomes of adherence to evidence-based practice rec¬ ommendations is unknown.1519To clarify the potential utility of claims-based out¬ comes research, trialists have recom- From the Department of Ambulatory Care and Pre- vention, Harvard Medical School and Harvard Pilgrim Health Care, Boston, Mass (Drs Soumerai and McLaughlin); Departments of Epidemiology and Bio- statistics, Harvard School of Public Health, Boston (Dr Spiegelman); Department of Epidemiology, Har- vard School of Public Health, Boston (Ms Hertzmark); Department of Medicine, Harvard Medical School and Brigham and Women's Hospital, Boston (Dr Thibault); and the Department of Medicine, University of Califor- nia, San Francisco (Dr Goldman).
An earlier version of this article was presented at the 1996 annual meeting of the Society for General Inter- nal Medicine, Washington, DC, May 4, 1996.Reprints: Stephen B. Soumerai, ScD, Department of Ambulatory Care and Prevention, Harvard Medical School and Harvard Pilgrim Health Care, 126 Brookline Ave, Suite 200, Boston, MA 02215.Downloaded From: http://jamanetwork.com/pdfaccess.ashx?url=/data/journals/jama/9868/ by a Harvard University User on 02/22/2017 mended that the results of such obser¬ vational studies should be compared with the results of RCTs of 6 technologies whose effects are unambiguous, includ¬ ing the effects of ß-blockade on survival after an AMI.20   In this study, we sought to answer the following specific questions: (1) What proportion of eligible elderly AMI pa¬ tients receive ß-blocker prophylaxis af¬ ter AMI? (2) Controlling for differences in risk status, are patient characteris¬ tics (age, sex, race, socioeconomic sta¬ tus [SES]) and use of alternative medi¬ cations (eg, calcium channel blockers) associated with receipt of ß-blockers in eligible patients? (3)Is the nonuse of ß-blockers among eligible patients as¬ sociated with increased mortality and rehospitalization for cardiovascular ill¬ ness following AMI, controlling for po¬ tentially confounding patient variables?(4) Are the relative mortality rates as¬ sociated with ß-blocker use obtained in this observational study comparable to those reported for elderly subgroups of large RCTs?

Data Sources
We linked 3 large, longitudinal data¬ bases: (1) New Jersey Medicare hospi¬ tal admissions (part A) and enrollment data for a 100% sample of AMI patients from 1986 to 1992; (2) New Jersey Med¬ icaid drug utilization and enrollment files for a 100% sample of Medicaid patients for the years 1986 to 1991; and (3) the New Jersey Program of Pharmacy As¬ sistance for the Aged and Disabled (PAAD) drug utilization data for non- Medicaid elderly for a 100% sample of enrollees from 1986 to 1991.
Using the Part A Medicare files, we determined the admission and discharge dates, primary and secondary diagnoses, and procedures associated with the first (index) AMI hospitalization (which oc¬ curred during [1986][1987][1988][1989][1990] and with all other hospital admissions in the year prior and 2 years following the index AMI. Mdicare enrollment (HISKEW [Health Insurance Skeletonized Write¬ off file]) data also included demographic characteristics (age, sex, race) as well as a record of survival and date of death up to 2 years following the index AMI.The reliability of these data for ascertaining mortality is well established.21Medicaid and PAAD drug claims data contained complete and reliable longi¬ tudinal histories of the dates and iden¬ tities of outpatient drug prescrip¬ tions,14•22•23 which were provided at no charge (or for a small co-payment) to New Jersey Medicare beneficiaries en¬ rolled in Medicaid or PAAD.Individual PAAD enrollees were ineligible for Med¬ icaid but nonetheless had incomes of less than $19000 in 1991.

Cohort Definition
We constructed a sample of Medicare patients with AMI enrolled in Medicaid or PAAD between 1986 and 1990, using inclusion and exclusion criteria almost identical to those in previous studies of the outcomes of AMI among Medicare populations.24Inclusion Criteria.-Weidentified el¬ derly persons (aged 65 years and older) discharged from a hospital from 1986 to 1990 with a principal diagnosis of AMI (International Classification ofDiseases, Ninth Revision [ICD-9] codes 410.0- 410.9).A recent study comparing such principal diagnoses with independently derived diagnoses obtained from hospital charts indicated high sensitivity (94%) and predictive value (92%).25We increased predictive value further using the exclu¬ sions described below.We defined the index AMI admission for each patient as the first AMI admission after January 1, 1987, but before January 1,19911, (AMIs in 1986 were used only to identify previous AMIs for the 1987 cohort).
Exclusion Criteria.-Weexcluded the following: (1) patients with end-stage re¬ nal disease or those residing outside New Jersey; (2) those hospitalized with an AMI in the 12 months preceding their index hospitalization; (3) patients who died dur¬ ing the incident admission or within 30 days of discharge (to ensure a minimum time window for measuring outpatient use of ß-blockers); (4) those not enrolled in either drug benefit program for at least 6 months before the index admission and at least 30 days after discharge; and (5) patients discharged alive whose length of stay for the index AMI admission was less than 5 days, indicating a possible mis¬ coding of AMI diagnosis.26

Contraindications to ß-Blocker Use
We defined our primary study group as AMI patients who met the inclusion and exclusion criteria described above, and had no measurable absolute or rela¬ tive contraindications to prophylactic use of ß-blockers.7Although it was impos¬ sible to identify all patients who had spe¬ cific contraindications to ß-blockade, we identified diagnoses and medications used before the index admission that repre¬ sented potential contraindications to ß-blocker use (for example, furosemide as an indicator ofsevere congestive heart failure [CHF], which was considered an absolute contraindication to ß-blocker use at the time).Therefore, we eliminated from analysis all persons with 1 or more prescriptions for furosemide in the 6 months before the index admission.We also considered a principal hospital diag¬ nosis of heart failure (ICD-9 code 428) in the prior year as a marker for severe CHF and an absolute contraindication to ß-blockers; this measure has been found to have moderately high sensitivity (85%) and positive predictive value (87%) in hos¬ pital claims data.26Since angiotensin-converting enzyme (ACE) inhibitors and di¬ goxin are also used for heart failure we included them as covariates in the analy¬ ses (Table 1).
Since asthma is also considered a con¬ traindication to ß-blockers,7 patients who used oral or inhaled bronchodilators (eg, theophylline) in the 6 months before the index AMI, or who had a principal or secondary hospital discharge diagnosis of asthma or chronic obstructive pul¬ monary disease before their AMI were excluded from our analysis.Since insulin- dependent diabetes is a relative contra¬ indication to ß-blockers, we eliminated from analysis all patients with any pre¬ scriptions of insulin during the 6 months before the index AMI.

Dependent Variables
For the analyses of predictors of ß-blocker use, the dependent variable was the time to first outpatient use of any ß-blocker following the index AMI within the first 90 days after discharge.We also conducted sensitivity analyses using a time window of 1 month after discharge as a more specific definition of prophy¬ lactic ß-blocker use.However, this defi¬ nition did not change any of our findings.The reference time for the analysis was the day of discharge from the index ad¬ mission; patients were followed until a ß-blocker was dispensed or the patient was censored from the analysis either through death, loss of eligibility, or end of the predefined time window.
In the second phase of analysis, we studied the relationship of ß-blocker ex¬ posure with 2 patient outcomes, adjust¬ ing for the clinically relevant and signifi¬ cant predictor variables described below.The primary outcome variable was mor¬ tality (time to death), because this is the most frequent measure used in RCTs of ß-blockers and can be measured reliably.

Independent Variables
Patient variables, such as age, sex, race, and Medicaid enrollment, have all been associated with differing levels of access to drug therapies and cardiac technolo¬ gies as well as survival2729; therefore, all analyses of ß-blocker use and outcomes adjusted for these variables (see Table 1).Because virtually all elderly Medicaid patients had yearly incomes below $6000 compared with PAAD recipient incomes of up to $19000 per year, this binary variable (Medicaid vs PAAD) also rep¬ resented a reliable indicator of relative poverty or SES.
Indicators of severity of illness (Table 1) included the number of hospital ad¬ missions in the year prior to the index hospitalization with a principal diagnosis of a cardiac condition (angina, ischemie heart disease, CHF), presence of princi¬ pal or secondary admission diagnoses for CHF, angina, other ischemie conditions, and other cardiovascular diseases (eg, es¬ sential hypertension); use of drugs in the 6 months before the AMI as markers for specific cardiovascular conditions (digoxin, ACE inhibitors, and antianginal agents); and use of ACE inhibitors in the 90 days after the index AMI (marker of possible CHF).We also included categorical vari¬ ables for use of ß-blockers and calcium channel blockers in the 6 months before the AMI, because they were likely to pre¬ dict use of these agents after the index AMI, as well as several control variables measuring the length of stay of the index admission, and whether the patient un¬ derwent revascularization.
Potential indicators of comorbidities are also included in Table 1.We mea¬ sured the number of noncardiac hospi¬ talizations (based on principal diagnoses) in the year before the index AMI, as well as the number of secondary diag¬ noses at the index admission.Also, 4 specific categories of secondary diag¬ noses (all cancers, chronic renal failure, cerebrovascular disease, and pneumo¬ nia) were included because they were significantly associated with a 40% or higher 2-year mortality rate in compari¬ son with all Medicare patients.We also constructed an index indicating the num¬ ber of different drug products taken in the 6 months before the AMI, because such variables have predicted adverse outcomes in our previous research.13Statistical

Analysis
We used Cox proportional hazards re¬ gression models to measure the effect and relative importance of patient char¬ acteristics on speed ofaccess to ß-blocker therapy following AMI.30The dependent variable in these analyses was the time- to-patient use of ß-blockers in the first 90 days after the index AMI.For all analy-ses, we included variables for age, sex, race, SES, and year of AMI regardless of their significance levels.All other risk adjustment and control variables were included in the final models only if they achieved a significance level of .10 or less in a step wise regression procedure.][32] The second phase of analyses esti¬ mated the effect of ß-blocker use on sur¬ vival times and time to new cardiac hos¬ pitalization, adjusting for the effect of any patient, severity, or comorbidity variables that independently predicted access to ß-blockers and/or health out¬ comes, using Cox proportional hazards models as described above.As before, we included in the analysis all AMI study patients eligible for ß-blocker treatment (ie, had no identifiable contraindications).We treated any use of ß-blockers, ACE inhibitors, or calcium channel blockers within the first 90 days following the index admission as a time-varying co- variate.Patients were censored due to death or disenrollment, but not for dis¬ continuation of ß-blocker therapy.
To ascertain the sensitivity of our pri¬ mary results to residual confounding, we conducted 2 additional sets of analyses to adjust more completely for potentially confounding variables.In the first ap¬ proach, which we termed the "saturated" model, we added to the list of candidate variables eligible for selection in step- wise regression all 2-way interactions of variables listed in Table 1.In addition, we used =.30 as the criterion for variable inclusion.This is a more conservative ap¬ proach to control for confounding, and has been shown in simulations to perform well in this regard.33In the second analy¬ sis, we used a propensity score approach:54 Propensity to exposure (in this case, ß-blocker use by 90 days) scores were developed using saturated Cox regres¬ sion models as described previously.Given all model covariate values, a propensity score was calculated for each study sub¬ ject and was used to adjust the estimated effect of ß-blocker use on mortality by entering this score as a single covariate in addition to ß-blocker use in the models for these outcomes.In addition, subjects were stratified by fertiles of propensity for ß-blocker use and stratum-specific es¬ timates of the effect of ß-blocker use on patient outcomes were obtained and com¬ pared, adjusted for propensity within strata.*Data expressed as No. (%) unless otherwise specified.Ml indicates myocardial infarction; ACE, angiotensln-converting enzyme; CABG, coronary artery bypass graft; and PTCA, percutaneous transluminal coronary angioplasty.
(/Moderately low-to middle-income individuals partici¬ pating in program of pharmaceutical assistance for the elderly (New Jersey Program of Pharmacy Assistance for the Aged and Disabled maximum income, $19 000 per year [1991]).
Secondary diagnosis only, because primary diagnosis of congestive heart failure resulted in ineligibility for study cohort.§Control variables for both severity and practice style.

Characteristics of Sample
Of the 5332 elderly patients who met study criteria (see above), 1595 patients (30%) had 1 or more absolute or relative contraindications to ß-blocker treatment: 8.5% with an asthma admission or bronchodilator use, 9% with insulin use, and 19% with possible severe CHF (principal hospital discharge diagnosis of CHF or use of furosemide).This resulted in a final study sample of 3737 AMI patients who were defined as eligible for ß-blocker therapy.
Eligible study patients (Table 1) had a median age of 77 years, slightly more than one half of the study subjects were women, 10% were nonwhite, and 4% had very low incomes (Medicaid) compared with 96% (non-Medicaid) with low to moderate incomes.
One quarter of the study patients had been admitted to a hospital in the year prior to their index AMI admission, most often with a noncardiac diagnosis (Table 1).Twenty-three percent to 27% of the cohort members experienced their in¬ dex AMI in each of the 4 years of ac¬ cession into the study (1987)(1988)(1989)(1990).
The high burden of illness of AMI in the elderly8,24 was reflected by a high rate ofadverse events in the 2 years of followup.Of the cohort members, all of whom had survived 30 days, 17% died during the first year after their AMI; after 2 years of follow-up, 27% had died.A total of 33% experienced a new hospital ad¬ mission with a principal diagnosis of AMI, angina, other ischemie heart disease, or CHF during the first year of follow-up.
Outpatient Use of ß-Blockers and

Alternate Medications
Only 21% of eligible study subjects received 1 or more prescriptions for a ß-blocker in the 90 days following dis¬ charge from the index AMI admission.This rate of ß-blocker use was essen¬ tially unchanged from the rate before the AMI (Table 1).The prevalences of ß-blocker use after the AMI in the 1987, 1988,1989, and 1990 cohorts were 20%, 18%, 20%, and 24%, respectively.
Use of calcium channel blockers in¬ creased immediately after the AMI from 23% of patients in the 6 months before the index admission to 49% of AMI sur¬ vivors during the first 90 days after hos¬ pital discharge.Among 2881 patients not receiving a calcium channel blocker before the AMI, 42% received 1 or more prescriptions from this drug class dur¬ ing the first 90 days after the AMI.In contrast, among 3084 patients not re¬ ceiving a ß-blocker before the AMI, only 15% were started on ß-blocker therapy during the 3 months after the AMI.Simi¬ larly, 73% of recipients of calcium chan¬ nel blockers before the AMI continued such therapy, compared with only 48% of recipients of ß-blockers before the AMI who continued receiving ß-blockade after discharge.The prevalences of use of calcium channel blockers in the 90 days after the AMI in the 1987, 1988, 1989, and 1990 cohorts were 50%, 48%, 43%, and 54%, respectively.

Predictors of ß-Blocker Use
Controlling for all other covariates in the proportional hazards model, patients aged 75 to 84 years and patients older includes all sociodemographic variables, severity, comorbidity, and control variables from Table 1 with P<.10 in stepwise model.CI indicates confidence interval; CHF, congestive heart failure; Ml, myocardial infarction; and ACE, angiotensin-converting enzyme.
(/Reference category.^Reference category: no use of each medication.¿Relative risk for each additional admission, diagnosis, or drug.UCancer, renal failure, pneumonia, or cerebrovascular disease.than 85 years were 14% and 44%, re¬ spectively, less likely than the "youngold" (ages 65-74 years) to receive a pre¬ scription for ß-blockers (RR=0.86 and 0.56; see Table 2).Sex, race, and SES were not independently associated with ß-blocker treatment.The substantially higher use of calcium channel blockers shortly after the index infarction was associated with a 36% reduction in the likelihood of ß-blocker use among eli¬ gible patients (RR=0.64;95% CI, 0.51- 0.80).

ß-Blocker Use and Mortality
Consistent with the RCT evidence, the receipt of a ß-blocker among patients con¬ sidered eligible for prophylactic ß-blockade after the AMI was strongly and in¬ dependently associated with a decreased mortality risk during 2 years of follow- up (Table 3).Controlling for all other pre¬ dictors of mortality, the adjusted rela¬ tive mortality rate among ß-blocker recipients was about 43% less than for nonrecipients (RR=0.57;95% CI, 0.47- 0.69).
The effect of ß-blocker use on mor¬ tality rates was consistent and substan¬ tial in all age strata (Figure 1).The find¬ ings of this study were also consistent with the results of the 2 large RCTs that included substantial numbers of elderly patients (Figure 2).
The estimated RR of death among ß-blocker recipients vs nonrecipients was unchanged when we used alternative methods to control more completely for possible confounding factors, such as ad¬ justment by the propensity to use ß-block¬ ers obtained from the multivariate Cox regression models (RR=0.61;95% CI, 0.50-0.74),34and by more highly saturated models that included all 2-way interac¬ tion terms with values less than or equal to .30.The adjusted relative mor¬ tality rate among ß-blocker recipients vs nonrecipients was constant across fer¬ tiles of propensity to receive ß-blockers and was also unaffected by exclusion of 623 additional patients with markers of CHF (any CHF diagnosis, use of digoxin with or without ACE inhibitors).
Using the estimated attributable mor¬ tality risk among those who did not re¬ ceive ß-blockers (43%, see Table 3), ap¬ proximately 381 of the 886 deaths occurring among these patients might have been avoided if they had been given ß-blockers.

Use of Calcium Channel Blockers and Mortality
Controlling for all predictors of mor¬ tality listed in Table 3, recipients of cal¬ cium channel blockers alone (n=1380 pa¬ tients eligible for ß-blockers) had nearly twice the risk of death as patients who "Controlling for all sociodemographic variables and the following severity, comorbidity, and control variables from Table 1 with P<.10 in stepwise model: pre-myocardial infarction (Ml) use of: digoxin, angiotensin-converting enzyme (ACE) inhibitors, calcium channel blockers; post-MI use of ACE Inhibitors; number of noncardiac admissions (prior year); number of secondary diagnoses at index admission; secondary diagnosis of cancer, renal failure, pneumonia, or cerebrovascular disease at index admission; number of different drugs (prior 6 months); coronary artery bypass graft during index admission; percutaneous transluminal coronan/ angioplasty during index admission; and length of stay of index admission.tReference category.[bottom]).Kaplan-Meier curves, evaluated at the average value of all model covariates (Table 3) at baseline within each stratum, were used to con¬ struct this figure (adjusted relative risk [RR]=0.50,95% confidence interval [CI], 0.36-0.72]for those aged 65-74 years; RR=0.56, 95% CI, 0.43-0.73for those aged 75-84 years; and RR=0.72, 95% CI, 0.47-1.11for those aged a85 years).-Relative risks (RRs) of death among ß-blocker recipients compared with nonrecipients among elderly subgroups of 2 large RCTs, and in 3 age strata (65-74 years, 75-84 years, and s85 years) of the New Jersey cohort.The adjusted RRs of death are plotted as solid squares; error bars represent 95% confidence intervals.Data from the ß-Blocker Heart Attack Trial (BHAT)6 and the Nor¬ wegian Multicenter Study (NMS).5, received only ß-blockers (n=334; RR = 1.98; 95% CI, 1.44-2.72).However, recipients of calcium channel blockers did not have a greater adjusted risk of death compared with patients not re¬ ceiving either ß-blockers or calcium channel blockers.Thus, the observed ad¬ verse outcomes associated with use of calcium channel blockers appeared to be due to the substitution of these agents for more effective ß-blockers rather than to intrinsic dangers of calcium channel blockers.
Although demographic characteris¬ tics, prior hospitalization, and comor¬ bidities among calcium channel blocker recipients and ß-blocker recipients were comparable, more calcium channel blocker recipients (19%) than ß-blocker recipients (10%) had markers of CHF.To control more completely for possible confounding due to CHF, we repeated the survival analyses of calcium channel blocker recipients after excluding all pa¬ tients with any markers of CHF.The resulting RRs were almost identical to those calculated for the entire sample, indicating that the severity and comor¬ bidity indicators used in our original model had adequately controlled for this type of confounding.ß-Blocker Use and New Hospital Admissions Controlling for the number of cardiac admissions in the year before the index AMI and all other significant severity and comorbidity variables (Table 1), the risk of rehospitalization among ß-blocker recipients was about 22% less than for nonrecipients (RR=0.78;95% CI, 0.67- 0.90).

COMMENT
The findings of this study suggest that substantial opportunities exist for in¬ creased use of an inexpensive preven¬ tive therapy for reducing morbidity and mortality among elderly AMI patients.
Advanced age, indicators of heart fail¬ ure, and use of calcium channel blockers soon after the index AMI were strong indicators of nonreceipt of ß-blocker therapy.Such underuse of ß-blockers was consistently associated with in¬ creased mortality and rehospitalization, even among 2345 patients older than 75 years, an age group that has been con¬ sistently excluded from RCTs of longterm ß-blockers after AMI.
Underuse of ß-blockers among sur¬ viving AMI patients in the community is of growing concern to both specialists and generalists.1'2'4•35Previous data on the prevalence of actual use of ß-block¬ ers in patients eligible for this therapy, while scant, are consistent with the find¬ ings of this study.9•10,36'37 Advanced age is a strong predictor of reduced use of ß-blockers in the acute phase of ill¬ ness,38,39 and was associated with lower use of these agents in outpatient set¬ tings in our study as well.Yet, para¬ doxically, the survival benefits of ß-blocker therapy appear to be at least as great among older patients as com¬ pared with the nonelderly.8 Given their substantial beneficial ef¬ fects, why is use of ß-blockers so low?Based on a large 2-state survey, 50% of generalists and 75% of cardiologists be¬ lieve that long-term ß-blocker therapy "definitely improves survival." 35Yet, lower actual ß-blocker prescribing, as determined in this and previous stud- ies,9' 36•37 suggests that clinicians may know the "right" answer to survey ques¬ tions regarding ß-blocker therapy, but continue to omit such agents from pa¬ tients' regimens.Barriers to ß-blocker therapy may include mistaken beliefs that these agents are harmful or less beneficial for patients with left ventricu¬ lar dysfunction or with diabetes (a rela¬ tive contraindication), and exaggerated Downloaded From: http://jamanetwork.com/pdfaccess.ashx?url=/data/journals/jama/9868/ by a Harvard University User on 02/22/2017 concerns regarding adverse effects on quality of life.1 For example, previous concerns regarding increased depres¬ sion, fatigue, and reduced libido have proven to be unsubstantiated if ß-selective agents are prescribed at the lowest effective doses.40,41Extensive market¬ ing of the newer calcium channel block¬ ers may have also contributed to clini¬ cians' negative attitudes toward ß-blockers.
This study has several limitations.First, the study data did not allow us to identify heart block, sinus bradycardia, and severe peripheral vascular disease, which represent absolute or relative contraindications to ß-blockade.Fortu¬ nately, however, we were able to ana¬ lyze clinical data from a recently pub¬ lished medical record study in 37 Minnesota hospitals,38 which indicated that after excluding patients with insulin- dependent diabetes, asthma, or chronic obstructive pulmonary disease, or pa¬ tients taking furosemide, only 6.7% of remaining elderly inpatients with AMI experienced any heart block, bra¬ dycardia, or severe peripheral vascular disease between the third day of admis¬ sion and hospital discharge.Thus, as¬ suming conservatively that 6.7% of the New Jersey cohort had these contrain¬ dications and did not receive ß-block¬ ers, the proportion of eligible patients receiving ß-blockers would have changed only marginally from 21% to 22.5%.
A second limitation is our inability to measure obesity, smoking, or other life¬ style risk factors that are not contained in administrative data.However, this limitation would not affect our results unless such factors are associated with ß-blocker use.Third, we could not iden¬ tify patients who may have received ß-blockers in the hospital, experienced an adverse drug reaction, and subse¬ quently discontinued therapy before re¬ ceiving an outpatient prescription.How¬ ever, only a small fraction of elderly AMI patients receive oral ß-blockers in hospitals42; only about 3% of patients are withdrawn from ß-blockers due to an adverse drug reaction during the first 2 weeks after the AMI5; and patients with outpatient drug coverage are un¬ likely to receive more than a 1-or 2-day supply of medications at discharge.Despite our attempts to control for confounding, it is possible that the higher risk associated with the use of calcium channel blockers reflects patients' higher intrinsic risks for CHF or other com¬ plications.In our cohort, patients re¬ ceiving calcium channel blockers alone fared much worse than patients who re¬ ceived ß-blockers alone.This higher mor¬ tality risk among calcium channel blocker recipients was very stable, even after multivariate adjustment for CHF and other risk factors; and this excess risk remained unchanged after excluding all patients with any markers of CHF.Our data add to the growing concern that calcium channel blockers should not be substituted for ß-blockers in circum¬ stances in which the latter have been proven to be effective.43However, this study was not designed to answer ques¬ tions regarding any independent adverse effects of calcium channel blockers, and ß-blocker eligible patients who instead received calcium channel blockers had adjusted outcomes similar to patients who received neither.
The similarity of our estimates of the effects of ß-blocker treatment on sur¬ vival to estimates from several large RCTs provides evidence that cohort studies using administrative databases may sometimes be useful in estimating the outcomes of guideline adherence.A previous study using clinical records at 1 teaching hospital was also able to rep¬ licate the results of 1 ß-blocker RCT by carefully defining cohorts at risk so that they resembled the patient samples re¬ cruited for the RCTs.44However, other studies of different drugs in surgery have observed uncontrollable selection biases that resulted in a greater likelihood for the retrospective comparisons to indi¬ cate treatment effectiveness when com¬ pared with the RCTs. 19We speculate that several characteristics ofour popu¬ lation, technology, and data sets may have increased the likelihood of obtain¬ ing valid findings.First, patients expe¬ riencing a new AMI are a well-defined population, and our exclusion criteria also reduced the likelihood of misclas- sification.In addition, there are only a few contraindications to ß-blocker treat¬ ment after AMI, and most ofthese could be measured in the database.Thus, a study of prophylactic use of ß-blockers is less subject to confounding by indi¬ cation than evaluation of other drug treatments that represent markers of serious illness and reduced survival (eg, ACE inhibitors following AMI).Finally, the relevant outcomes of ß-blocker therapy (mortality and rehospitalization) could be measured reliably in Medicare data.However, additional research is needed on similar well-defined popula¬ tions and other treatments with known effects before the validity and general- izability of such targeted outcomes re¬ search can be determined.
In summary, despite strong evidence demonstrating that use of ß-blockers fol¬ lowing AMI decreases morbidity and mortality, they are substantially under¬ used in the elderly.Our findings sug¬ gest that this underuse leads to mea¬ surable adverse outcomes, including a 43% excess risk of 2-year mortality and a 20% increase in rates of rehospitaliza¬ tion for cardiovascular disease.The ap¬ parent frequent substitution of calcium channel blockers for ß-blockers follow¬ ing AMI is also associated with an in¬ creased mortality risk.This analysis of a large cohort of typical community pa¬ tients provides strong support for ex¬ isting guidelines to prescribe ß-block¬ ers instead of calcium channel blockers as a routine preventive therapy for el¬ derly AMI patients.
This study was supported by grants ROÍ HS07631 and ROI HS06341 from the Agency for Health Care Policy and Research and by the Har¬ vard Pilgrim Health Care Foundation.

Table 2 .
-Independent Predictors of Trial of ß-Blockers Within 90 Days of Discharge (Index Admission) Among Eligible Patients in a Proportional Hazards Regression Model*

Table 3 .
-Association of Receipt of ß-Blocker Therapy With Mortality Over 2 Years of Follow-up, Control¬ ling for Demographic, Severity, and Comorbidity Variables From Proportional Hazards Regression Model All-Cause Mortality