Author + information
- Received November 17, 1998
- Revision received May 17, 1999
- Accepted August 5, 1999
- Published online November 15, 1999.
- Elias A Iliadis, MDa,
- Lloyd W Klein, MD, FACP, FACCa,
- Betsy J Vandenberg, MSa,
- Diana Spokas, RNa,
- Tony Hursey, MPHa,
- Joseph E Parrillo, MD, FACC, FACPa and
- James E Calvin, MD, FACC, FRCPCa,* ()
- ↵*Reprint requests and correspondence: Dr. James E. Calvin, Section of Cardiology, Rush Presbyterian–St. Luke’s Medical Center, 1653 W. Congress Parkway, 214 Jones, Chicago, Illinois 60612-3864
To determine the influence of clinical practice guidelines on treatment patterns and clinical outcomes in unstable angina and the effectiveness of guideline reminders on implementing practice guidelines, two groups of medium and high risk patients with unstable angina were compared.
New guidelines have been published by the Agency for Health Care Policy and Research (AHCPR) for evaluating and managing patients with unstable angina. The impact of these guidelines to improve the quality of care has never been tested.
Group 1 included 338 consecutive medium or high risk patients admitted before publication of the AHCPR guidelines, and group 2 consisted of 181 consecutive similar risk patients admitted after institution of the AHCPR guideline reminders at this institution. Dissemination of clinical practice guidelines was ensured by a grand rounds lecture and by posting guideline reminders on all group 2 patients’ charts within 24 h of admission.
The two groups were similar in terms of most baseline characteristics, including hypercholesterolemia, diabetes, hypertension, smoking history, baseline ST segment depression and previous coronary artery bypass graft surgery. Group 1 patients were older (68 ± 13 vs. 63 ± 16 years, p = 0.001) and more frequently had a previous myocardial infarction (39% vs. 22%, p = 0.001). Group 2 patients more frequently required intravenous nitroglycerin to control the index episode of chest pain (43% vs. 34%, p = 0.003). Group 2 patients more frequently received aspirin (96% vs. 88%, p = 0.009) during admission and underwent coronary angiography (71% vs. 58%, p = 0.006). More importantly, group 2 patients received oral beta-blockers (p = 0.008), aspirin and coronary angiography (p = 0.001) earlier than group 1 patients and experienced recurrent angina (29% vs. 54%) and myocardial infarction or death less frequently (3% vs. 9%, p = 0.028).
In unstable angina, clinical practice guidelines were associated with greater use of aspirin and coronary angiography and greater use and earlier administration of beta-blockers. Variation in drug use over time was also reduced. Objective improvement in clinical outcome was also noted. Thus, practice guidelines improve the quality of care of patients with unstable angina.
Clinical practice guidelines have been proposed as an important means to improve “clinical effectiveness and quality of care” (1–3). Medical organizations, health care researchers, sponsors of health benefit plans and public officials have all expressed interest in practice guidelines to standardize treatment plans and to improve the cost effectiveness of the US health care system (4,5). The presumption for improved cost effectiveness is that clinical practice guidelines present evidence-based treatment options clearly and concisely. Physicians are then assumed to rapidly assimilate the best evidence, streamline their management and decision making and ultimately improve the effectiveness of care for their patients.
The use of guideline reminders to implement practice guidelines has previously been examined (6,7). Overhage and McDonald (6)examined the use of preventive medicine reminders given to physicians on daily rounds, suggesting preventive care for eligible patients, and demonstrated the ease with which guideline reminders can implement practice guidelines and that intervention must be directed at the physician responsible for point of care.
The Agency for Health Care Policy and Research (AHCPR) published clinical practice guidelines for the evaluation and diagnosis of unstable angina (8). A basic tenet of the recommended policy is to match utilization of resources (i.e., drugs, intensive care, procedures, costs) to the patient’s risk of adverse outcome. However, the best way to implement these guidelines in clinical practice is uncertain, and it was intentionally left by the AHCPR to the individual physicians and hospitals to determine how best to incorporate these concepts into practice. Thus, the current study was designed to assess whether the use of clinical practice guidelines alters the process of care in patients with unstable angina as defined by the type, frequency and timeliness of treatments received (so-called “process indicators”).
There were 484 patients with the diagnosis of unstable angina consecutively admitted to the Rush Presbyterian–St. Luke’s Medical Center in Chicago, Illinois, between June 1, 1992 (the beginning of the clinical data base) and April 1, 1994 (the month of publication of the AHCPR clinical practice guidelines). Of the 484 patients admitted with this diagnosis, 338 were considered by AHCPR definitions to have medium to high risk unstable angina and were defined as group 1—the control group. Between January 1, 1995 (the date of introduction of a revised data base) and March 30, 1996, 225 patients admitted with the diagnosis of unstable angina were admitted consecutively to the medical center. Of the 225 patients admitted during this period, 181 were identified as medium to high risk and were defined as group 2—the study group admitted during the implementation of guideline reminders.
Because of the lack of an emergency department (ED) observation area during these periods, all patients with this diagnosis were admitted to either the coronary care unit or the coronary step-down unit. The AHCPR guidelines defined patients as high risk for death or nonfatal myocardial infarction when one of the following was present: prolonged, ongoing (>20 min) rest pain, pulmonary edema, angina with new or worsening mitral regurgitation, rest pain with dynamic ST segment changes >1 mm, angina with an S3 gallop, rales or hypotension. We modified this slightly by adding another predictor—recent myocardial infarction (<14 days), which we have determined to be a very important predictor (9). The AHCPR guidelines defined patients as medium risk when no high risk features were present, but one of the following was present: resolved rest angina, angina with dynamic T wave changes, nocturnal angina, new-onset Canadian Cardiovascular Society class III or IV angina in the past two weeks, Q waves or ST segment depression >1 mm in multiple leads or age >65 years. Exclusion criteria included all patients with low risk chest pain syndromes as defined by the AHCPR guidelines, including increased angina of effort frequency, severity or duration or angina provoked at lower work thresholds, new-onset angina within two weeks to two months and a normal electrocardiogram (ECG). Patients with cardiac enzyme elevation <12 h after admission consistent with a non–Q wave myocardial infarction and those with chest pain syndromes believed by their attending physician to not be cardiac were also excluded.
The AHCPR guidelines’ intended purpose was to define optimal diagnostic and management strategies for patients with unstable angina. These included the use of intravenous (IV) nitroglycerin in the ED for ongoing ischemia, the use of IV heparin in the ED for all intermediate or high risk patients without contraindications and the use of aspirin and beta-blockers for all cardiac patients without contraindications. These treatments represent process indicators of the quality of care. In January 1995, a medical grand rounds was conducted to inform the medical house staff and attending physicians of the AHCPR guidelines on unstable angina.
Also, during the intervention period (1995), a double-sided two-page reminder was posted in the charts of all patients admitted with unstable angina to the coronary care unit or coronary care step down unit. Concurrently, a chest pain assessment form was developed and introduced into the ED. This form included prompts for indicated treatment, and emergency physicians were made aware of guideline assessment through this form. All patients were treated at the discretion of their attending physician.
The impact of clinical practice guidelines on clinical treatments was determined through evaluating treatments rendered before and after the institution of guideline reminders. Demographic variables, treatment options and diagnostic/therapeutic procedures performed were recorded for the two groups. These included age, gender, hypertension, diabetes mellitus, tobacco use, elevated cholesterol, previous medications, admission ECG and treatments rendered such as IV heparin, IV nitroglycerin, aspirin, IV or oral beta-blockers. Procedures incurred by the groups included stress testing, diagnostic angiography, angioplasty and coronary artery bypass graft surgery. Major cardiac complications were defined by the occurrence of either death or myocardial infarction after the first 12 h of admission during the hospital period. Acute myocardial infarction occurring as a complication of unstable angina was diagnosed if new Q waves developed after the first 24 h of hospital admission, the creatine kinase (CK) level exceeded 300 U/liter and the CK-MB fraction exceeded 0.05 after the first 12 h.
Data collection was performed by registered nurses trained in data collection and was standardized through a manual of operations that provided diagnostic definitions, how to identify tests, treatments and complications. The data were entered into a computerized data base directly by the data collection nurse, and data quality assurance was performed through electronic audits of selected missing or out-of-range values. Chart audit was performed on 25% of all charts for accuracy.
Univariate comparisons were made using the chi-square test for categoric variables and the Student ttest for continuous variables. The primary end points were frequency of use of aspirin, beta-blockers, heparin and IV nitrates, as well as the timeliness of their use. Secondary end points included coronary angiography and revascularization and major cardiac complications. To determine whether clinical confounders influenced the frequency and timeliness of management, and to adjust the effects of guidelines for these influences, multiple logistic regression analyses were performed using as dependent variables treatments and outcome variables found to be different between the two groups. Independent variables included guidelines group, all AHCPR predictors of complications, coronary risk factors and evidence of previous coronary artery disease. This analysis was performed in a forward stepwise fashion using SAS software using an F value of 4.0 and p <0.05 to add a variable. Goodness-of-fit was tested using −2 log of the likelihood. This result is summarized as the overall p value of the model. In addition, all variables entered in the stepwise regression model reduced the value of the Akaike Information Criterion. In this way, we attempted to adjust for important clinical confounders.
Table 1compared the baseline characteristics of groups 1 and 2. Although group 1 patients were older, the incidence of hypercholesterolemia, diabetes, family history, hypertension and smoking history were similar between the two groups. The groups also had similar frequencies of ST segment depression >1 mm and recent myocardial infarction. Remote myocardial infarction (>2 weeks) was more frequent in group 1, although the rates of previous revascularization and previous use of beta-blockers and calcium channel blockers were higher in group 2.
Treatment in the hospital (process indicators)
Medical treatments and interventions are summarized in Tables 2 and 3, respectively. ⇓⇓Group 2 patients received aspirin and IV nitroglycerin for the index pain more frequently than did group 1 patients. The use of IV beta-blockers was similar in both groups. Group 2 had a 7% increase in the use of heparin and a 16% increase in the use of oral beta-blockers as compared with group 1, although neither of these was statistically significant. This also included a 32% increase in a new beta-blocker usage (p = 0.099), which was also not statistically significant.
The timeliness of drug administration is demonstrated in Figure 1. The use of beta-blockers (p = 0.008, panel A) and aspirin (panel C, p = 0.090) was instituted earlier in group 2 patients. Intravenous heparin was unaffected. Also, the variation over the time of the study in the use of beta-blocker and aspirin (Fig. 2 and 3)⇓was reduced after guideline reminders (group 2) were introduced.
Although not a primary end point, coronary angiography was used more frequently in group 2 patients, whereas the rates of coronary artery bypass graft surgery and percutaneous transluminal coronary angioplasty were similar between the groups. However, coronary angiography was performed earlier in group 2 patients (Fig. 1, panel B, p = 0.001).
To determine if any confounders explained the differences in treatment between the two groups, multiple logistic regression analyses were performed for the following evaluations and treatments that we found to be different between the two groups: coronary angiography at any time during the hospital period, coronary angiography within 24 h, use of an oral beta-blocker within 24 h and use of aspirin and IV nitroglycerin on admission. These are summarized in Table 4. In each case the guideline group was associated with greater use of a treatment or test with odds ratios varying between 2.1 for an angiogram at any time during the hospital period to 2.66 for the use of aspirin. Clearly, other clinical factors were associated with a greater or earlier clinical use of a test or treatment. Family history, tobacco history, recent myocardial infarction (<14 days) and recurrent angina were associated with greater use of coronary angiography. Previous coronary artery bypass graft surgery and recurrent angina were associated with greater use of IV nitroglycerin at the time of hospital admission.
Patient outcomes are summarized in Table 5. Group 2 patients had a lower incidence of the combined end point of myocardial infarction or death (67% reduction), recurrent angina (46% reduction) and serious ventricular dysrhythmia (92% reduction). Table 6summarizes multiple logistic regression analysis for major cardiovascular outcomes. Two composite outcomes were used: 1) myocardial infarction, death, heart failure, ventricular tachycardia or ventricular fibrillation; and 2) any of the aforementioned composite outcomes or recurrent angina. Both composites were less frequent in the guideline group as compared with the control group, even after adjusting for other potential confounders. Other clinical variables associated with poor outcomes are also summarized in Table 6.
This study demonstrates that clinical practice guidelines can beneficially alter treatment patterns in unstable angina. Specifically, we determined that intensive medical treatment was more frequently and more uniformly offered after clinical practice guidelines were adopted and introduced to the medical staff by being posted on patients’ charts as a reminder. This pattern of intensive treatment suggests that patients with medium or higher risk unstable angina received better anti-anginal therapy at hospital admission, in accordance with AHCPR guidelines, than they did before publication of the guidelines. Furthermore, less variation in the use of medical treatment was also noted after the guidelines were introduced, demonstrating that the effect of our method of introduction of clinical practice guidelines occurred quickly and was sustained. Finally, the early and more frequent use of these treatments in this preliminary study was associated with fewer ischemic-related complications, showing for the first time that guidelines may impact clinical outcome in acute coronary syndromes. Although this finding requires further validation because of the small sample size, the potential for selection bias was addressed by multivariate analysis, which confirmed the effect of guidelines even after adjusting for other potential clinical confounders.
Industry definitions of quality of care focus on reducing variation in the processes of care to improve cost effectiveness (1). Medical practice guidelines are an attractive tool because, in the best of cases, they provide a coherent sequenced set of recommendations based on the best clinical evidence to achieve the best outcomes (3,10). However, one major limitation to date has been the failure to demonstrate the value of clinical practice guidelines on improving quality of practice or improving outcome. In fact, one meta-analysis has suggested that primary care guidelines do not improve outcome (11), and another study suggested that the guidelines might have contributed to extra cost, leading one expert (12)to conclude that “clinical practice guidelines were on trial.” However, a more recent meta-analysis that looked at a variety of interventions found not only positive effects of the guidelines but also that the guideline reminders, as used in this study, were one of the most effective ways to implement practice guidelines, resulting in significant practice changes and improved outcomes (13).
Although few clinical practice guidelines have been systematically evaluated for effectiveness in general, Weingarten et al. (14,15)did examine practice guidelines in the setting of patients being admitted to rule out myocardial infarction (many of whom had unstable angina). They examined the benefit, risk and costs of implementing a practice guideline for patients with chest pain syndromes who were considered low risk. Their results revealed that patients with chest pain who were considered low risk for a complication could be treated effectively with lower costs and without compromise in patient outcome or satisfaction. Notably, the impact and effectiveness of recent AHCPR guidelines on unstable angina evaluation and treatment have not been studied.
The current AHCPR guidelines on unstable angina (8)consists of over 100 pages of material, including over 100 recommendations and over 120 references, posing a significant challenge to health care deliverers to systematically implement the guidelines, particularly when it may affect over 200 physicians in a single center as it does in our institution. The approach we used was rather comprehensive and involved multiple strategies to enhance physician compliance, because previous reviews have suggested that a multimodality approach to implementation may be most effective (6,16). The use of a reminder system, in particular, was chosen because of ease of its implementation, its focus on providing useful information at the point of care and its respect for physician autonomy.
The primary end points used in this study were the proven treatments in unstable angina as recommended by the guidelines. These treatments serve as process indicators of quality of care. Because the guidelines also suggested that invasive and conservative strategies of evaluation were equivalent, less focus was placed on the type of strategy employed and more emphasis was placed on the timeliness of the chosen evaluation strategy. However, the AHCPR guidelines recommends coronary angiography in the presence of high risk features and recurrent angina (an inclusion criteria for this study). The use and choice of revascularization, which depended previously on coronary anatomy, were not a prime focus of this intervention.
The major limitations of the study are the absence of a concurrent comparison group and the nonrandomized nature of patient enrollment. To assess a difference in treatment patterns, separate periods of patient enrollment were necessary to avoid the “knowledge” bias, which may occur as physicians are repeatedly exposed to the guideline reminders. This knowledge bias may alter routine treatment patterns and subsequently greatly skew results. Several experts (12,16)have commented that the reference standard for a study such as ours, which is a behavioral one, is not necessarily a randomized clinical trial. Randomized, controlled trials would be particularly vulnerable to Hawthorne effects (i.e., the knowledge that one is in a trial can contaminate the experimental group), systematically exposing it to factors unrelated to guidelines that were not experienced by the control group (12).
In an attempt to address this issue, Hayward et al. (12)listed criteria for the quality of studies of clinical practice guidelines, besides randomization, most of which are fulfilled in this study. Furthermore, using a grading system proposed by Johnston et al. (17), our study design, despite not being randomized, would fall close to the median quality score of the 28 studies reviewed recently by that group. Furthermore, to adequately test the effectiveness of a practice guideline, one attempts to study an entire population rather than a small targeted group. This allows a determination of the amount of variation in treatment, which may exist in general, when treating such patients. Thus, study designs including separate control and study groups enrolled consecutively seemed best suited to optimize the effect that clinical guidelines have on treatment patterns. Furthermore, we used multiple logistic regression analysis to identify important clinical confounders independently influencing temporal changes in clinical practice and to assess their relative importance. Although some clinical confounders were identified, the use of guidelines more than doubled the odds of using important medical therapy and early coronary angiography, and was associated with fewer clinical events independent of any other identified clinical factor.
In summary, clinical practice guidelines in unstable angina, when implemented using an effective means, had a substantial impact on the evaluation and management of patients with unstable angina and were associated with better patient outcome.
- Agency for Health Care Policy and Research
- creatine kinase
- emergency department
- Received November 17, 1998.
- Revision received May 17, 1999.
- Accepted August 5, 1999.
- American College of Cardiology
- Eagle K.A,
- Lee T.H,
- Brennan T.A,
- et al.
- Gottlieb L,
- Margolis C,
- Schoenbaum S
- Braunwald E,
- Jones R,
- Mark D.B,
- et al.
- Worrall G.C,
- Freake D
- Hayward R