Clinician decisions and computers

Computer clinical decision support that automates personalized clinical care: a challenging but needed healthcare delivery strategy

How to deliver best care in various clinical settings remains a vexing problem. All pertinent healthcare-related questions have not, cannot, and will not be addressable with costly time- and resource-consuming controlled clinical trials. At present, evidence-based guidelines can address only a small fraction of the types of care that clinicians deliver. Furthermore, underserved areas rarely can access state-of-the-art evidence-based guidelines in real-time, and often lack the wherewithal to implement advanced guidelines. Care providers in such settings frequently do not have sufficient training to undertake advanced guideline implementation. Nevertheless, in advanced modern healthcare delivery environments, use of eActions (validated clinical decision support systems) could help overcome the cognitive limitations of overburdened clinicians. Widespread use of eActions will require surmounting current healthcare technical and cultural barriers and installing clinical evidence/data curation systems. The authors expect that increased numbers of evidence-based guidelines will result from future comparative effectiveness clinical research carried out during routine healthcare delivery within learning healthcare systems.

Enabling a learning healthcare system with automated computer protocols that produce replicable and personalized clinician actions

Clinical decision-making is based on knowledge, expertise, and authority, with clinicians approving almost every intervention-the starting point for delivery of "All the right care, but only the right care," an unachieved healthcare quality improvement goal. Unaided clinicians suffer from human cognitive limitations and biases when decisions are based only on their training, expertise, and experience. Electronic health records (EHRs) could improve healthcare with robust decision-support tools that reduce unwarranted variation of clinician decisions and actions. Current EHRs, focused on results review, documentation, and accounting, are awkward, time-consuming, and contribute to clinician stress and burnout. Decision-support tools could reduce clinician burden and enable replicable clinician decisions and actions that personalize patient care. Most current clinical decision-support tools or aids lack detail and neither reduce burden nor enable replicable actions. Clinicians must provide subjective interpretation and missing logic, thus introducing personal biases and mindless, unwarranted, variation from evidence-based practice. Replicability occurs when different clinicians, with the same patient information and context, come to the same decision and action. We propose a feasible subset of therapeutic decision-support tools based on credible clinical outcome evidence: computer protocols leading to replicable clinician actions (eActions). eActions enable different clinicians to make consistent decisions and actions when faced with the same patient input data. eActions embrace good everyday decision-making informed by evidence, experience, EHR data, and individual patient status. eActions can reduce unwarranted variation, increase quality of clinical care and research, reduce EHR noise, and could enable a learning healthcare system.

Iliad: Moving Medical Decision-Making into New Frontiers

If the essence of clinical practice is a process of sequential problem-solving, whereby a physician works with a patient to formulate a series of decisions about diagnostic treatment, then it would naturally follow that the essence of medical education should evolve around the training of would-be clinicians in the difficult art of diagnostic decisionmaking. Yet this is not often the case.

Prior convictions

Large, randomized clinical trials ("megatrials") are key drivers of modern cardiovascular practice, since they are cited frequently as the authoritative foundation for evidence-based management policies. Nevertheless, fundamental limitations in the conventional approach to statistical hypothesis testing undermine the scientific basis of the conclusions drawn from these trials. This review describes the conventional approach to statistical inference, highlights its limitations, and proposes an alternative approach based on Bayes' theorem. Despite its inherent subjectivity, the Bayesian approach possesses a number of practical advantages over the conventional approach: 1). it allows the explicit integration of previous knowledge with new empirical data; 2). it avoids the inevitable misinterpretations of p values derived from megatrial populations; and 3). it replaces the misleading p value with a summary statistic having a natural, clinically relevant interpretation-the probability that the study hypothesis is true given the observations. This posterior probability thereby quantifies the likelihood of various magnitudes of therapeutic benefit rather than the single null magnitude to which the p value refers, and it lends itself to graphical sensitivity analyses with respect to its underlying assumptions. Accordingly, the Bayesian approach should be employed more widely in the design, analysis, and interpretation of clinical megatrials.

Risk stratification in coronary artery disease: implications for stabilization and prevention

Noninvasive nuclear imaging techniques, including dual-isotope myocardial perfusion single-photon emission computed tomography (SPECT), have been employed in the development of strategies for diagnosis and risk stratification of patients with suspected or known coronary artery disease. These risk-stratification strategies are based on studies in which known outcome has been linked to diagnostic and prognostic information provided by myocardial perfusion SPECT. This article describes a validated dual-isotope exercise protocol for assessment of perfusion and function and reviews the evidence on which a cost-effective risk management strategy is based.

Prior restraint: a Bayesian perspective on the optimization of technology utilization for diagnosis of coronary artery disease

In conclusion, at least 1/3 of patients with suspected coronary artery disease are inappropriately referred for scintigraphic diagnostic testing from a Bayesian such as those described in this report, may be a powerful mechanism for encouraging more appropriate technology utilization while simultaneously controlling costs, and are thereby deserving of a formal prospective demonstration trial. However, since only half the patients currently being tested are referred for diagnostic purposes, analogous strategies must be developed with respect to prognostic and therapeutic evaluation.

Computerized surgical databases: introduction to design and function

There has been an enormous increase in the variety and volume of information encountered in surgical practice, either as measurable clinical variables or laboratory research data over the last three decades. Despite its form and origin, this information remains the basis of both daily clinical decision making and analytical research. Inferences drawn from past experience will modify an individual's approach to disease and treatment. However, when the demands of information processing cannot be met, mischief begins and quality of care declines. Modern computers offer an extraordinarily powerful method of processing the large volumes of medical data that are acquired, and provide techniques for analysis that would have been impossible, and often inconceivable, without computers. The applications of computer technology to surgical data management range from such simple and repetitive tasks as practice administration and accounting to elegant statistical and image analysis. This paper outlines the utility of computerized data management in clinical surgery and surgical research, and describes techniques for designing and implementing a customized surgical database system.

Development and validation of a logistic regression-derived algorithm for estimating the incremental probability of coronary artery disease before and after exercise testing

OBJECTIVES

Our goals were to develop and validate a multivariate algorithm for estimating the incremental probability of the presence of coronary artery disease.

BACKGROUND

Multivariate methods, including logistic regression analysis, have been extensively applied to diagnostic exercise testing. However, few previous studies have included both an incremental design and external validation.

METHODS

A retrospective collection of clinical, exercise test and catheterization data was performed involving four U.S. referral medical centers. All patients had no prior history of coronary disease and had undergone coronary angiography < or = 3 months after exercise stress testing. An algorithm was developed in one center (590 patients with a 41% prevalence of coronary artery disease) with the use of logistic regression analysis and was validated in the other three centers (1,234 patients, 70% prevalence). The algorithm incorporated pretest variables (age, gender, symptoms, diabetes, cholesterol), exercise electrocardiographic (ECG) variables (mm of ST segment depression, ST slope, peak heart rate, metabolic equivalents [METs], exercise angina) and one thallium variable. Discrimination was measured with receiver operating characteristic curve analysis. Calibration (that is, reliability) was assessed from a comparison of probability estimates and the actual prevalence of disease.

RESULTS

The overall incremental receiver operating characteristic curve areas for the validation group were pretest, -0.738 +/- 0.016; postexercise ECG, 0.78 (SE 0.017); and postthallium, 0.82 (SE 0.016); p < 0.01 for both increments. Within the three validation institutions, the institution with a disease prevalence closest to that of the derivation institution had the best incremental receiver operating characteristic curve areas. There was a stepwise incremental improvement in calibration especially from exercise ECG to thallium testing.

CONCLUSIONS

An incremental multivariate algorithm derived in one center reliably estimated disease probability in patients from three other centers. The incremental value of testing was best demonstrated when the derivation and validation groups had a similar disease prevalence. This algorithm may be useful in decision making that relates to the diagnosis of coronary disease.

A method to describe physician decision thresholds and its application in examining the diagnosis of coronary artery disease based on exercise treadmill testing

The authors developed a method that utilizes logistic regression analysis to 1) calculate the disease probability with confidence intervals at which any specified proportion of physicians reaches a clinical decision, 2) statistically test whether factors other than disease probability affect this clinical decision, and 3) statistically test whether physician decision making in relation to disease probability varies by other factors. They apply the method to analyze the relationship between disease probability and the proportion of physicians who diagnosed coronary artery disease (CAD) in 127 consecutive subjects who completed the treadmill exercise tolerance test (ETT) at two hospitals. Twenty-five percent of the physicians decided that CAD was possible or definite at a post-ETT disease probability of 0.24 (95% CL= 0.07-0.35); 50% at 0.54 (95% CL = 0.43-0.70); and 75% at 0.82 (95% CL = 0.67-1.0). Multivariate logistic regression analysis revealed three factors significantly and independently related to the diagnosis of CAD: post-ETT disease probability, positive ETT result, and cigarette smoking. The proportion of physicians who reached a diagnosis of CAD did not differ by hospital setting (VA versus university), level of training (attending versus housestaff/fellow), or diagnosing service (cardiology versus other internal medicine). It is concluded that factors other than disease probability may affect physician diagnostic decisions.

Comparison of logistic regression and Bayesian-based algorithms to estimate posttest probability in patients with suspected coronary artery disease undergoing exercise ECG

Two multivariate methods, a logistic regression-derived algorithm and a Bayesian independence-assuming method (CADENZA), were compared concerning their abilities to estimate posttest probability of coronary disease in patients with suspected coronary disease. All patients underwent exercise testing within 3 months prior to coronary angiography. Coronary disease was defined as the presence of one or more vessels with greater than or equal to 50% luminal diameter narrowing. A group of 300 patients (disease prevalence = 37%) was used to derive the algorithm. Another group of 950 patients was used to validate the algorithm and compare it to CADENZA. Seven variables (age, sex, symptoms, diabetes, mm ST depression, ST slope, and peak heart rate) were used to generate posttest probabilities for each method. The receiver operating characteristic curve area for the logistic regression method (0.81 +/- 0.01) was significantly higher than CADENZA (0.75 +/- 0.01; p less than 0.05). There was, however, no difference in the calibration of the two methods. When given equivalent variable information, the logistic regression algorithm had better discrimination than CADENZA for estimating the probability of coronary disease following exercise electrocardiography.

Multivariate analysis to simplify the differential diagnosis of broad complex tachycardia

Univariate analysis has identified several criteria that aid the differential diagnosis of broad complex tachycardia. In this study of 102 consecutive patients multivariate analysis was performed to identify which of 15 clinical and 11 electrocardiographic variables were independent predictors of ventricular tachycardia. These were shown to be a history of myocardial infarction, the QRS waveforms in leads aVF and V1, and a change in axis from sinus rhythm to tachycardia of more than 40 degrees. If none of the criteria was met, the diagnosis was almost certainly supraventricular tachycardia. If one criterion was met the diagnosis was probably supraventricular tachycardia. If two criteria were met then the diagnosis was probably ventricular tachycardia. If three or four criteria were met, the diagnosis was almost certainly ventricular tachycardia. The predictive accuracy was 93%. This was increased to 95% by including two other criteria--definite independent P wave activity and ventricular extrasystoles with the same QRS configuration as that in tachycardia. These criteria were not included in the multivariate analysis because though they were 100% specific they were seldom seen. These four criteria can be used as simple rules in determining the origin of a broad complex tachycardia.

International application of a new probability algorithm for the diagnosis of coronary artery disease

A new discriminant function model for estimating probabilities of angiographic coronary disease was tested for reliability and clinical utility in 3 patient test groups. This model, derived from the clinical and noninvasive test results of 303 patients undergoing angiography at the Cleveland Clinic in Cleveland, Ohio, was applied to a group of 425 patients undergoing angiography at the Hungarian Institute of Cardiology in Budapest, Hungary (disease prevalence 38%); 200 patients undergoing angiography at the Veterans Administration Medical Center in Long Beach, California (disease prevalence 75%); and 143 such patients from the University Hospitals in Zurich and Basel, Switzerland (disease prevalence 84%). The probabilities that resulted from the application of the Cleveland algorithm were compared with those derived by applying a Bayesian algorithm derived from published medical studies called CADENZA to the same 3 patient test groups. Both algorithms overpredicted the probability of disease at the Hungarian and American centers. Overprediction was more pronounced with the use of CADENZA (average overestimation 16 vs 10% and 11 vs 5%, p less than 0.001). In the Swiss group, the discriminant function underestimated (by 7%) and CADENZA slightly overestimated (by 2%) disease probability. Clinical utility, assessed as the percentage of patients correctly classified, was modestly superior for the new discriminant function as compared with CADENZA in the Hungarian group and similar in the American and Swiss groups. It was concluded that coronary disease probabilities derived from discriminant functions are reliable and clinically useful when applied to patients with chest pain syndromes and intermediate disease prevalence.

Computers in medicine: liability issues for physicians

Physicians routinely use computers to store, access, and retrieve medical information. As computer use becomes even more widespread in medicine, failure to utilize information systems may be seen as a violation of professional custom and lead to findings of professional liability. Even when a technology is not widespread, failure to incorporate it into medical practice may give rise to liability if the technology is accessible to the physician and reduces risk to the patient. Improvement in the availability of medical information sources imposes a greater burden on the physician to keep current and to obtain informed consent from patients. To routinely perform computer-assisted literature searches for informed consent and diagnosis is 'good medicine'. Clinical and diagnostic applications of computer technology now include computer-assisted decision making with the aid of sophisticated databases. Although such systems will expand the knowledge base and competence of physicians, malfunctioning software raises a major liability question. Also, complex computer-driven technology is used in direct patient care. Defective or improperly used hardware or software can lead to patient injury, thus raising additional complicated questions of professional liability and product liability.

The clinical use of nuclear exercise tests

In the mid-1970s, after decades of clinical use, the utility of electrocardiographic exercise testing for the evaluation of patients with suspected ischemic heart disease was critically examined and questioned. Concurrent with this critique, two sophisticated, more expensive and powerful "nuclear" exercise tests were introduced sequentially for clinical use: myocardial perfusion imaging with thallium-201 and radionuclide ventriculography with technetium 99m. The published indications for the two tests are similar, and both have been shown to offer advantages over ECG stress testing in selected populations. However, few data are available regarding the comparative utility of thallium versus ventriculographic imaging. As part of a prospective study to assess the efficacy of cardiovascular nuclear medicine studies, we undertook the present analysis to assess the clinical evolution of these tests and to elucidate factors responsible for clinicians' choice for the often competing examinations. The study examined 213 consecutive patient referrals for thallium scintigraphy and 183 referrals for ventriculography, ranging from patients with no symptoms or highly non-specific chest pain syndromes (21% of referrals) to patients with proven coronary disease (28% of the referrals). Twenty-one percent of patients were referred to confirm the clinical impression that the patient did not have coronary disease, 40% to confirm its presence, and 37% to determine its severity. Analyses were undertaken to determine the factors that dictated a preference for thallium scintigraphy rather than ventriculography; only the physician's intent in testing and level of training were significant predictors for a particular nuclear test.