Patients with Lemierre syndrome have a high risk of new thromboembolic complications, clinical sequelae and death: an analysis of 712 cases

BACKGROUND

Lemierre syndrome is characterized by head/neck vein thrombosis and septic embolism usually complicating an acute oropharyngeal bacterial infection in adolescents and young adults. We described the course of Lemierre syndrome in the contemporary era.

METHODS

In our individual-level analysis of 712 patients (2000-2017), we included cases described as Lemierre syndrome if these criteria were met: (i) primary site of bacterial infection in the head/neck; (ii) objectively confirmed local thrombotic complications or septic embolism. The study outcomes were new or recurrent venous thromboembolism or peripheral septic lesions, major bleeding, all-cause death and clinical sequelae.

RESULTS

The median age was 21 (Q1-Q3: 17-33) years, and 295 (41%) were female. At diagnosis, acute thrombosis of head/neck veins was detected in 597 (84%) patients, septic embolism in 582 (82%) and both in 468 (80%). After diagnosis and during in-hospital follow-up, new venous thromboembolism occurred in 34 (5.2%, 95% CI 3.8-7.2%) patients, new peripheral septic lesions became evident in 76 (11.7%; 9.4-14.3%). The rate of either was lower in patients who received anticoagulation (OR: 0.59; 0.36-0.94), higher in those with initial intracranial involvement (OR: 2.35; 1.45-3.80). Major bleeding occurred in 19 patients (2.9%; 1.9-4.5%), and 26 died (4.0%; 2.7-5.8%). Clinical sequelae were reported in 65 (10.4%, 8.2-13.0%) individuals, often consisting of cranial nerve palsy (n = 24) and orthopaedic limitations (n = 19).

CONCLUSIONS

Patients with Lemierre syndrome were characterized by a substantial risk of new thromboembolic complications and death. This risk was higher in the presence of initial intracranial involvement. One-tenth of survivors suffered major clinical sequelae.

Treatment Effect of Niaspan, a Controlled-release Niacin, in Patients with Hypercholesterolemia: A Placebo-controlled Trial

Background

The present study was designed to determine the efficacy and safety of Niaspan (Kos Pharmaceuticals, Inc, Hollywood, FL), a new controlled-release formulation of niacin, in the treatment of primary hyperlipidemia, the occurrence and severity of flushing events, and potential adverse effects, particularly hepatotoxicity.

Methods and Results

The study was conducted as a multicenter, randomized, double-blind, placebo-controlled, parallel comparison of Niaspan in doses of 1000 mg/day and 2000 mg/day, administered once a day at bedtime. One hundred twenty-two patients with low-density lipoprotein cholesterol levels > 4.14 mM/L (160 mg/dL) with dietary intervention and high-density lipoprotein cholesterol ≤ 1.81 mM/L (70 mg/dL) were randomized to one of three treatment groups: placebo, and 1000 mg/day or 2000 mg/day of Niaspan. Safety and efficacy measures included 12-hour serum fasting lipid and lipoprotein concentrations, serum analyte levels for major organ function, flushing diaries, and adverse event records. The placebo group demonstrated no significant changes in serum lipoprotein concentrations over the treatment period of 12 weeks, except for a slight 4% increase in high-density lipoprotein cholesterol. Niaspan significantly lowered low-density lipoprotein cholesterol levels by 6% and 14% for the 1000 mg/day and 2000 mg/day doses, respectively. High-density lipoprotein cholesterol levels rose significantly, with a 17% increase occurring at the 1000 mg/day dose and a 23% increase occurring at the 2000 mg/day dose. Niaspan (2000 mg/day) produced significant decreases of 27% and 29%, respectively, for serum lipoprotein(a) and triglyceride concentration. Although the incidence of flushing was significant, these episodes were generally well tolerated.

Conclusion

Niaspan administered in doses of 1000 mg/day and 2000 mg/day at bedtime were well tolerated with few side effects and produced favorable effects on the major circulating lipoproteins of patients with primary dyslipidemias as specified by the enrollment criteria.

Health-related quality of life and virologic outcomes in an HIV clinic

OBJECTIVE

The purpose of this study was to describe the relationship between viral load and health-related quality of life (HRQOL) in a cohort of persons with human immunodeficiency virus (HIV) infection.

DESIGN

We evaluated HRQOL measurements in a clinical cohort of HIV-positive patients recruited from a university-associated HIV primary care clinic. HRQOL instruments included the medical outcomes survey-short form-36(MOS-SF-36) from which mental and physical component summary scores (MCS and PCS) and subscale scores were calculated.

RESULTS

Significant negative associations were found between viral load and SF-36 PCS, physical functioning (PF), role-physical (RP), bodily pain (BP), general health (GH), role-emotional (RE), and vitality (VT). Similar negative associations were found between CD4 cell count and SF-36 summary and subscale scores, with the notable exception of bodily pain. Multivariate analyses controlling for the effects of CD4 cell count and other clinical variables indicated viral load as an independent predictor of SF-36 PCS, RP, BP and VT scores.

CONCLUSIONS

The relationship between viral load, a measure of HIV disease activity, and several dimensions of the SF-36, a patient-focused measure of HRQOL, appears to be strong and independent of CD4 cell count. These findings suggest that having a lower viral load positively impacts the quality of life of HIV-positive patients.

Relationship of hospital teaching status with quality of care and mortality for Medicare patients with acute MI

CONTEXT

Issues of cost and quality are gaining importance in the delivery of medical care, and whether quality of care is better in teaching vs nonteaching hospitals is an essential question in this current national debate.

OBJECTIVE

To examine the association of hospital teaching status with quality of care and mortality for fee-for-service Medicare patients with acute myocardial infarction (AMI).

DESIGN, SETTING, AND PATIENTS

Analysis of Cooperative Cardiovascular Project data for 114,411 Medicare patients from 4361 hospitals (22,354 patients from 439 major teaching hospitals, 22,493 patients from 455 minor teaching hospitals, and 69,564 patients from 3467 nonteaching hospitals) who had AMI between February 1994 and July 1995.

MAIN OUTCOME MEASURES

Administration of reperfusion therapy on admission, aspirin during hospitalization, and beta-blockers and angiotensin-converting enzyme inhibitors at discharge for patients meeting strict inclusion criteria; mortality at 30, 60, and 90 days and 2 years after admission.

RESULTS

Among major teaching, minor teaching, and nonteaching hospitals, respectively, administration rates for aspirin were 91.2%, 86.4%, and 81.4% (P<.001); for angiotensin-converting enzyme inhibitors, 63. 7%, 60.0%, and 58.0% (P<.001); for beta-blockers, 48.8%, 40.3%, and 36.4% (P<.001); and for reperfusion therapy, 55.5%, 58.9%, and 55.2% (P =.29). Differences in unadjusted 30-day, 60-day, 90-day, and 2-year mortality among hospitals were significant at P<.001 for all time periods, with a gradient of increasing mortality from major teaching to minor teaching to nonteaching hospitals. Mortality differences were attenuated by adjustment for patient characteristics and were almost eliminated by additional adjustment for receipt of therapy.

CONCLUSIONS

In this study of elderly patients with AMI, admission to a teaching hospital was associated with better quality of care based on 3 of 4 quality indicators and lower mortality. JAMA. 2000;284:1256-1262

What is heartburn worth? A cost-utility analysis of management strategies

OBJECTIVE

To determine the best treatment strategy for the management of patients presenting with symptoms consistent with uncomplicated heartburn.

METHODS

We performed a cost-utility analysis of 4 alternatives: empirical proton pump inhibitor, empirical histamine2-receptor antagonist, and diagnostic strategies consisting of either esophagogastroduodenoscopy (EGD) or an upper gastrointestinal series before treatment. The time horizon of the model was 1 year. The base case analysis assumed a cohort of otherwise healthy 45-year-old individuals in a primary care practice.

MAIN RESULTS

Empirical treatment with a proton pump inhibitor was projected to provide the greatest quality-adjusted survival for the cohort. Empirical treatment with a histamine2 receptor antagonist was projected to be the least costly of the alternatives. The marginal cost-effectiveness of using a proton pump inhibitor over a histamine2-receptor antagonist was approximately $10,400 per quality-adjusted life year (QALY) gained in the base case analysis and was less than $50,000 per QALY as long as the utility for heartburn was less than 0.95. Both diagnostic strategies were dominated by proton pump inhibitor alternative.

CONCLUSIONS

Empirical treatment seems to be the optimal initial management strategy for patients with heartburn, but the choice between a proton pump inhibitor or histamine2-receptor antagonist depends on the impact of heartburn on quality of life.

An intervention to increase physicians' use of adherence-enhancing strategies in managing hypercholesterolemic patients

PURPOSE

Patients' lack of adherence to medical regimens frustrates many practicing physicians. This study was conducted to determine the effectiveness of a combined continuing medical education intervention in increasing physicians' adherence-enhancing skills and improving hypercholesterolemic patients' health.

METHOD

A prospective, randomized, controlled trial was designed using a nested cohort of 28 community physicians throughout Alabama and 222 of their hypercholesterolemic outpatients. The intervention, carried out in 1998, consisted of three interactive case audio-conferences plus chart reminders. Physicians' learning was measured by unannounced standardized patients, and patients' health by serum cholesterol levels, weight, knowledge of hypercholesterolemia, self-reported dietary habits, and health status.

RESULTS

No significant difference was found in the numbers of physician adherence-enhancing strategies, although the number did increase within the treatment group. There were significant differences in the intervention group's patients' knowledge of cholesterol management (p = .008) and significant reductions in their self-reported consumption of dietary fats (p = .002). A significant difference was found in the serum cholesterol level of men in the intervention group nine months after the intervention (p = .02).

CONCLUSION

Combining a series of interactive case audio-conferences with chart reminders shows promise in increasing physicians' adherence-enhancing strategies. In chronic disease management, the problem of enhancing adherence remains complex.

The art and science of searching MEDLINE to answer clinical questions. Finding the right number of articles

The current medical environment makes information retrieval a matter of practical importance for clinicians. Many avenues present themselves to the clinician, but here we focus on MEDLINE by summarizing the current state of the art and providing an innovative approach for skill enhancement. Because new search engines appear rapidly, we focus on generic principles that can be easily adapted to various systems, even those not yet available. We propose an idealized classification system for the results of a MEDLINE search. Type A searches produce a few articles of high quality that are directly focused on the immediate question. Type B searches yield a large number of articles, some more relevant than others. Type C searches produce few or no articles, and those that are located are not germane. Providing that relevant, high-quality articles do exist, type B and C searches may often be improved with attention to search technique. Problems stem from poor recall and poor precision. The most daunting task lies in achieving the balance between too few and too many articles. By providing a theoretical framework and several practical examples, we prepare the searcher to overcome the following barriers: a) failure to begin with a well-built question; b) failure to use the Medical Subject Headings; c) failure to leverage the relationship between recall and precision; and d) failure to apply proper limits to the search. Thought and practice will increase the utility and enjoyment of searching MEDLINE.

The association of physician attitudes about uncertainty and risk taking with resource use in a Medicare HMO

PURPOSE

To explore the association between the attitudes of primary care physicians toward uncertainty and risk taking, as measured by a validated survey, with resource use in a Medicare HMO.

DESIGN

All primary-care internists (n=20) in a large, multi-specialty clinic were surveyed to measure their attitudes about uncertainty and risk taking using three previously developed scales. Results were linked with administrative data for 792 consecutive patients in a recently created Medicare HMO. The patients' index visits occurred between April 1, 1995, and November 30, 1995.

ANALYSIS

Charges stemming from several claim types (primary care and subspecialty physician, laboratory, radiology, and ambulatory procedures) in the 30 days following the index visit were summed. The physician scales were dichotomized at the median to seek unadjusted associations with charges. Generalized estimation equations were used to account for the correlation of charges resulting from patients' being nested within physicians and adjusted for physician characteristics (age, sex, years in practice) and patient characteristics (age, sex, comorbidity).

MAIN RESULTS

The physician response rate was 90%. Most physicians (90%) were male. The mean age of the patients was 74 years, and 69% were female. The mean cost (+/-SD) per patient was $621.61+/-1,737.31. From the unadjusted analysis, high "anxiety due to uncertainty" was associated with higher patient charges ($197.85 vs $158.21, p=0.01). From the multivariable analysis, each standard deviation increase in "anxiety due to uncertainty" (3.5 points) corresponded to a 17% increase in mean charges (p < 0.01) and each similar increase in "reluctance to disclose uncertainty to patients" (1.92 points) corresponded to a 12% increase (p=0.03). However, increasing "reluctance to disclose mistakes to physicians" and increasing physician risk-taking propensity were associated with decreased total charges [-10% per standard deviation (1.34 points), p=0.02, and -8% per standard deviation (3.26 points), p=0.02, respectively].

CONCLUSION

Physician attitudes toward uncertainty were significantly associated with patient charges. Further investigation may improve prediction of patient-care charges, offer insight into the medical decision-making process, and perhaps clarify the relationship between cost, uncertainty, and quality of care.

An international comparison of physicians' judgments of outcome rates of cardiac procedures and attitudes toward risk, uncertainty, justifiability, and regret

OBJECTIVE

Compare U.K. and U.S. physicians' judgments of population probabilities of important outcomes of invasive cardiac procedures; and values held by them about risk, uncertainty, regret, and justifiability relevant to utilization of cardiac treatments.

DESIGN

Cross-sectional study.

SETTING

University hospital and VA medical center in the United States; two teaching hospitals in the United Kingdom.

PARTICIPANTS

171 housestaff and attendings at U.S. teaching hospitals; 51 physician trainees and consultants at U.K. hospitals.

MEASURES

Judgments of probabilities of severe complications and deaths due to Swan-Ganz catheterization, cardiac catheterization, percutaneous transluminal coronary angioplasty (PTCA), and coronary artery bypass grafting (CABG); judgments of malpractice risks for case vignettes; Nightingale's risk-aversion instrument; Gerrity's reaction-to-uncertainty instrument; questions about need to justify decisions; responses to case vignettes regarding regret.

RESULTS

The U.S. physicians judged rates of two bad outcomes of cardiac procedures (complications due to cardiac catheterization; death due to CABG) to be significantly higher (p < or = 0.01) than did the U.K. physicians (U.S. medians, 5 and 3.5, respectively; U.K. medians 3 and 2). The median ratio of (risk of malpractice suit I error of omission)/(risk of suit I error of commission) judged by U.K. physicians, 3, was significantly (p=0.0006) higher than that judged by U.S. physicians, 1.5. The U.K. physicians were less often risk-seeking in the context of possible losses than the U.S. physicians (odds ratio for practicing in the U.K. as a predictor of risk seeking 0.3, p=0.003). The U.K. physicians had significantly more discomfort with uncertainty than did the U.S. physicians, as reflected by higher scores on the stress scale (U.K. median 48, U.S. 42, p=0.0001) and the reluctance-to-disclose-uncertainty scale (U.K. 40, U.S. 37, p < 0.0001) of the Gerrity instrument. There was no clear international difference in perceived need to justify decisions, or in regret.

CONCLUSIONS

The results were not clearly consistent with the uncertainty hypothesis that international practice variation is due to differences in judged rates of outcomes of therapy or with the imperfect-agency hypothesis that practice variation is due to differences in physicians' personal values. The causes and implications of practice variations remain unclear.

Interpreting cost analyses of clinical interventions

OBJECTIVE

In the present era of cost containment, physicians need reliable data about specific interventions. The objectives of this study were to assist practitioners in interpretation of economic analyses and estimation of their own costs of implementing recommended interventions.

DATA SOURCES

MEDLINE search from 1966 through 1995 using the text words cost or expense and medical subject heading (MeSH) terms costs and cost analysis, cost control, cost of illness, cost savings, or cost-benefit analysis.

STUDY SELECTION

The 4 eligibility criteria were clinical trial with random assignment; health care quality improvement intervention tested; effects measured on the process or outcome of care; and cost calculation mentioned in the report.

DATA EXTRACTION

After independent abstraction and after consensus development, financial data were entered into a costing protocol to determine which costs related to the intervention were provided.

DATA SYNTHESIS

Of 181 articles, 97 (53.6%) included actual numbers on the costs of the intervention. Of 97 articles analyzed, the most frequently reported cost figures were in the category of operating expenses (direct cost, 61.9%; labor, 42.3%; and supplies, 32.0%). General overhead was not presented in 91 (93.8%) of the 97 studies. Only 14 (14.4%) of the 97 studies mentioned start-up costs. The text word $ in the abstract and the most useful MeSH index term of cost-benefit analysis appeared with nearly equal frequency in the articles that included actual cost data (37.1 % vs 35.1%). Two thirds of articles indexed with the MeSH term cost control did not include cost figures.

CONCLUSIONS

Statements regarding cost without substantiating data are made habitually in reports of clinical trials. In clinical trial reports presenting data on expenditures, start-up costs and general overhead are frequently disregarded. Practitioners can detect missing information by placing cost data in a standardized protocol. The costing protocol of this study can help bridge care delivery and economic analyses.

Physicians' judgments of the risks of cardiac procedures. Differences between cardiologists and other internists

OBJECTIVES

The authors compared judgments of the population risks of invasive cardiac procedures made by cardiologists and other internal medicine physicians. Our main hypotheses were that cardiologists' judgments would differ from those made by the other physicians and that cardiologists' judgments would be more accurate than those of other physicians.

METHODS

This was a cross-sectional survey of senior staff and physician-trainees at two teaching hospitals affiliated with a US medical school, Emergency Department physicians at a community hospital in the same metropolitan area, and senior staff and trainees at two teaching hospitals affiliated with a UK school. Judgments of the risks of severe morbidity and death due to Swan-Ganz catheterization, cardiac catheterization, percutaneous coronary angioplasty, and coronary artery bypass grafting were assessed.

RESULTS

Nineteen cardiologists judged the risks of severe morbidity due to all procedures and the risks of death due to all procedures except coronary artery bypass grafting to be significantly lower than did the 78 other internists. Cardiologists more frequently made accurate judgments of the rates of morbidity and death due to cardiac catheterization than did the other internists; other internists more frequently made accurate judgments for the rates of morbidity due to Swan-Ganz catheterization.

CONCLUSIONS

Disagreements about the risks of procedures may arise from a paucity of published data, or from an over-supply of confusing data.

Physicians' survival predictions for patients with acute congestive heart failure

BACKGROUND

Current guidelines suggest that patients with low likelihoods of survival may be excluded from intensive care. Patients with new or exacerbated congestive heart failure are frequently but not inevitably admitted to critical care units.

OBJECTIVE

To assess how well physicians could predict the probability of survival for acutely ill patients with congestive heart failure, and in particular how well they could identify patients with small chances of survival.

METHODS

This was a prospective cohort study done in the emergency departments of a university hospital, a Veterans Affairs medical center, and a community hospital. The study population was consecutive adults for whom new or exacerbated congestive heart failure, diagnosed clinically, was a major reason for the emergency department visit. Physicians caring for the study patients in the emergency departments recorded their judgments of the numeric probability that each patient would survive for 90 days and for 1 year. The patients vital status at 90 days and 1 year was ascertained by multiple means, including interview, chart review, and review of hospital and state databases.

RESULTS

By calibration curve analysis, the physicians underestimated survival probability at both 90 days and 1 year, particularly for patients they judged to have the lowest probabilities of survival. Their predictions had modest discriminating ability (receiver operating characteristic curve areas, 0.66 [SE = 0.020] for 90 days; 0.63 [SE = 0.017] for 1 year). The physicians identified only 15 patients they judged to have a 90-day survival probability of 10% or less, whose survival rate was actually 33.3%.

CONCLUSIONS

Physicians have great difficulty predicting survival for patients with acute congestive heart failure and cannot identify patients with poor chances of survival. Current triage guidelines that suggest patients with poor chances of survival may be excluded from critical care may be impractical or harmful.

Choice of long-term strategy for the management of patients with severe esophagitis: a cost-utility analysis

BACKGROUND & AIMS

Omeprazole has shown remarkable efficacy and safety in the treatment of patients with gastroesophageal reflux disease (GERD); similarly, laparoscopic techniques have allowed less morbidity in patients undergoing fundoplication procedures. Concerns about the long-term cost and safety of both strategies have prompted a debate of their role in long-term management of patients with severe erosive esophagitis.

METHODS

A cost-utility analysis was performed to compare two strategies: laparoscopic Nissen fundoplication (LNF) vs. omeprazole. A two-stage Markov model was used to obtain cost and efficacy estimates; all estimates were discounted at 3% per year. The time horizon was 5 years. Sensitivity analyses were performed on all relevant variables.

RESULTS

Both strategies were similarly effective (4.33 quality-adjusted life years per patient), with omeprazole less expensive than LNF ($6053 vs. $9482 per patient). At 10 years, LNF and omeprazole costs were similar. Efficacy estimates were extremely sensitive to changes in quality of life associated with postoperative symptoms and long-term use of medication.

CONCLUSIONS

Medical therapy is the preferred treatment strategy for most patients with severe erosive esophagitis. Individuals with a long life expectancy are good candidates for LNF if postoperative morbidity is low and GERD symptoms remain abated for many years.

Racial differences in the medical treatment of elderly Medicare patients with acute myocardial infarction

OBJECTIVE

To compare the use of medications in African-American and Caucasian elderly Medicare patients hospitalized with acute myocardial infarction (AMI) in Alabama.

DESIGN

Retrospective medical record review.

SETTING

All acute care hospitals in Alabama.

PATIENTS

All Medicare patients with a principal discharge diagnosis of AMI from June 1992 through February 1993. We excluded those patients less than 65 years of age and those of ethnicity other than African-American or Caucasian (N = 4,052).

MEASUREMENTS

We first performed a crude analysis using all cases to compare by race the use of thrombolysis, beta-adrenergic blockade, and aspirin in the setting of AMI. In addition, we developed a multivariable model with receipt of therapy as the outcome and demographics, severity of illness, comorbidity, and algorithm-determined candidacy for therapy as covariates. The algorithms, developed as part of the Cooperative Cardiovascular Project, were designed to identify an "ideal" pool of candidates for each therapy.

MAIN RESULTS

For all cases, 9.2% (95% confidence interval [CI] 6.8, 12.1) of African Americans received thrombolysis compared with 17.3% (95% CI 16.0, 18.6) of Caucasians. Approximately 16.4% of patients received beta-adrenergic blockade, and 45.1% received aspirin, both with no racial difference. By multivariate analysis, the adjusted odds ratio for African Americans receiving thrombolysis was 0.55 (95% CI 0.41, 0.76). The corresponding odds ratio was 1.25 (95% CI 0.99, 1.59) for beta-adrenergic blockade and 1.13 (95% CI 0.96, 1.37) for aspirin. African Americans presented later after the onset of chest pain, but the refusal rate of thrombolytic therapy did not differ.

CONCLUSIONS

According to this analysis, Alabama physicians used beta-adrenergic blockade and aspirin equivalently in African Americans and Caucasians. African Americans received thrombolysis less often according to the crude analysis. The multivariable analysis suggests less use of thrombolytics, even after adjusting for several covariates including indication by clinical algorithm. However, the small number of African-American patients deemed ideal candidates for thrombolysis attenuates the precision of this finding.

Impact of patient history on residents' evaluation on child sexual abuse

OBJECTIVE

To determine if historical information influences residents' interpretation of physical findings in sexually abused children.

METHODOLOGY

In a pediatric residency training program, all residents viewed 15 slides of children's genitalia (8 normal, 7 abnormal) with either a history specific for sexual abuse or one which was nonspecific. Three weeks later the same slides were viewed but with the alternate history scenario. The residents were asked if the physical findings were specific for sexual abuse.

RESULTS

Sixty-four percent of residents completed both surveys. Correct response rate did not vary by gender or year of training. Responses were most often correct when the slide and history were normal (87%). Responses were least accurate when normal historical information was presented with abnormal slides (49%). A logistic regression model demonstrated that residents were less accurate when history and physical did not agree (95% CI = .54- .78). Reexamination of the data using areas under the Receiver Operating Characteristic (ROC) curve confirmed that residents performed on a less accurate ROC curve when the slide and history were incongruent (p < .01).

CONCLUSION

Incongruency between patient history and physical exam findings negatively affected this group of residents' ability to discriminate between abuse and nonabuse findings.

Practice variation in the management of pharyngitis: the importance of variability in patients' clinical characteristics and in physicians' responses to them

The objective of this study was to assess whether geographic differences in antibiotic-prescribing rates for patients with pharyngitis could be explained by intersite differences in patients' clinical characteristics and in how physicians responded to these clinical cues when making decisions. As part of the initial phase of a prospective controlled trial to improve physicians' diagnostic ability, the authors enrolled cohorts of consecutive patients seen at staff-model--HMO student health services in Pennsylvania and Nebraska. Physicians' decisions whether to prescribe antibiotics for 310 consecutive patients presenting with pharyngitis to the former and 214 such patients presenting to the latter at the time of the initial visit were examined. There was a large discrepancy between the antibiotic-prescribing rates at the student health services in Pennsylvania, 106/310, 32.4%, and Nebraska, 156/214, 72.9%. The clinical variables significantly independently associated with treatment at both sites in a logistic regression model were fever, adjusted odds ratio = 2.1 (95% CI = 1.1, 3.8); exudates, 5.4 (2.8, 10); palatine petechiae, 6.5 (1.5, 28); rhinorrhea, 0.46, (0.25, 0.85); and high risk of complications, 3.8 (1.04, 14). There was a significant interaction between site and anterior cervical adenopathy, 5.5 (1.6, 19); and a borderline interaction between site and rhinorrhea, 2.4 (0.89, 6.7). Site was not a significant independent predictor of treatment, 1.8 (0.45, 6.6.). Practice variation was related to geographic differences in patients' clinical characteristics and in how physicians responded to these factors when prescribing antibiotics. How physicians weight patients' clinical characteristics when making decisions may be an important element of their "practice styles."

Evaluation of two ELISA's for detecting Chlamydia trachomatis from endocervical swabs

Two enzyme immunoassays (EIAs) detecting Chlamydia trachomatis from endocervical swabs, Syva MicroTrak (MT) and Abbott Chlamydiazyme (CZ), were compared with a tissue culture (TC) standard. Initially, 8% (100 of 1250) of specimens were TC positive, yielding sensitivities of 94% (94 of 100) for MT and 79% (79 of 100) for CZ with identical 98% specificities (1129 of 1150 for MT and 1130 of 1150 for CZ). Discrepant specimens were retested by both EIAs and assayed for elementary bodies (EBs) by a fluorescent antibody test. After discrepancy analysis, 9.5% (118) of 1240 patients were either TC or EB positive, yielding sensitivities of 94.1% for MT (111 of 118) and 79.7% for CZ (94 of 118) with identical specificities of 100% (1122 of 1122). These results indicate that the MT is significantly more sensitive (p less than 0.05, McNemar test) than CZ in detecting C. trachomatis from endocervical swabs.

Screening for drinking disorders in the elderly using the CAGE questionnaire

OBJECTIVE

To assess the performance of the CAGE questionnaire in identifying elderly medicine outpatients with drinking problems.

DESIGN

Cross-sectional design, with the alcohol module of the Diagnostic Interview Schedule as the criterion standard.

SETTING

The outpatient medical practice of an urban university teaching hospital.

PATIENTS

Consecutive patients 18 years or older who signed a consent form approved by the university's institutional review board. For this study, 323 patients greater than or equal to 60 years old.

MAIN OUTCOME MEASURES

Sensitivity, specificity, receiver operating characteristics (ROC) curve and positive predictive value for CAGE scores of 0-4 for patients 60 years or older.

RESULTS

Thirty-three percent of the sample group met study criteria for a history of drinking problems, including 63% of the male patients and 22% of the female patients. The sensitivity and specificity for a cut-off score of one for all patients was 86% and 78%, respectively, and 70% and 91% for a cut-off of two. The calculation of the area under the ROC curve was .86, and the positive predictive value of CAGE scores of 0-4 were 33%, 66%, 79%, 82%, and 94%, respectively. The predictive value for any score was higher in males than females, reflecting the higher prevalence of problems in the male population.

CONCLUSIONS

The CAGE can effectively discriminate elderly patients with a history of drinking problems from those without such a history. The chosen cut-off score should consider the prevalence of drinking problems in the population being tested.

Screening for alcohol abuse using CAGE scores and likelihood ratios

OBJECTIVE

To assess the performance of the CAGE (acronym referring to four questions, see below) questionnaire in discriminating between medicine outpatients with and without an alcohol abuse or dependence disorder.

DESIGN

A cross-sectional design of a sample of consecutive patients who received both the alcohol module of the diagnostic interview schedule and the CAGE (Cut down, Annoyed, Guilty, Eye-opener) screening questionnaire.

SETTING

The outpatient medical practice of an urban university teaching hospital.

PATIENTS

All patients 18 years or older who signed a consent form approved by the university's institutional review board.

MEASUREMENT

Calculation of the sensitivity, specificity, receiver operating characteristic (ROC) curve, and likelihood ratio for CAGE scores of 0 to 4.

RESULTS

Thirty-six percent of the sample group met criteria for a history of alcohol abuse or dependence. A CAGE score of 2 or more was associated with a sensitivity and specificity of 74% and 91%. The calculated area under the ROC curve was 0.89, whereas the likelihood ratios for CAGE scores of 0 to 4 were 0.14, 1.5, 4.5, 13, and 100, respectively. These ratios were associated with posterior probabilities for an abuse or dependence disorder of 7%, 46%, 72%, 88%, and 98%, respectively.

CONCLUSION

Clinicians can improve their ability to estimate a patient's risk for an alcohol abuse or dependence disorder using likelihood ratios for CAGE scores.

Signal detectability: the use of ROC curves and their analyses

Issues related to ROC curves are addressed. The original article on the subject by Lee Lusted, describing the "state of the art" 20 years ago, is reviewed. The concepts that Lusted addressed are then expanded, suggesting the current state of the art. New issues that have arisen with regard to ROC curves and their use in medicine are addressed. Finally, potential areas for future investigation are suggested.

The use of nurses to evaluate houseofficers' humanistic behavior

OBJECTIVES

1) To determine whether nurse evaluations of humanistic behavior discriminate between houseofficers in an internal medicine training program, and 2) to compare nurse and attending physician evaluations.

DESIGN

Prospective, six-month comparison of nurse and attending ratings of houseofficer humanistic behavior.

PROCEDURE

Using a six-item, Likert-scale humanistic behavior rating form, nurses and ward attendings evaluated 76 PGY-1, PGY-2, and PGY-3 houseofficers over a six-month period. Nurses and attendings voluntarily evaluated houseofficers on all inpatient units in both university and Veterans Administration teaching hospitals.

MEASUREMENTS AND MAIN RESULTS

Nurse ratings discriminated residents from one another throughout the six months of the study and over all units in both hospitals. Attending physician ratings were only moderately correlated with nurses' and were significantly more lenient. Exploratory analyses of the nursing evaluations revealed that female houseofficers received significantly more favorable evaluations than did men and that ward nurses were significantly more lenient than were critical care nurses. Nurse ratings did not differ by hospital, training year, or month of evaluation.

CONCLUSIONS

Nurses can provide information about humanistic behavior that will allow program directors to discriminate among different levels of houseofficer behavior. Information from nurses differs from that provided by attending physicians. Nurse ratings are affected by gender and by the type of unit from which they are obtained.

Clinical and microbiological evidence for endemic pharyngitis among adults due to group C streptococci

Group C beta-hemolytic streptococci cause rare epidemic outbreaks of pharyngitis, but their role in sporadic endemic pharyngitis has been uncertain. We addressed the question of whether non-group A beta-hemolytic streptococci are associated with endemic pharyngitis in two ways. First, we compared rates of isolation from throat swabs of group A, B, C, and G and ungrouped beta-hemolytic streptococci ("culture negative") in adult patients vs those rates in controls. Second, we collected in standardized form clinical indexes of patients with pharyngitis: signs and symptoms graded for severity, the examining physician's subjective estimate of the probability of streptococcal pharyngitis, a logistic regression score predicting streptococcal pharyngitis, and whether antibiotic therapy was prescribed. After collecting data and cultures on 1425 patients with sore throats and cultures on 284 controls, we found the following: group C streptococci were isolated significantly more frequently in patients with sore throats than in controls (6% vs 1.4%); four clinical signs and two symptoms distinguished group C-associated pharyngitis as more severe than culture-negative pharyngitis; and six clinical signs and one symptom distinguished group C-associated pharyngitis as less severe than group A pharyngitis. Physicians' subjective estimates, logistic regression scores, and antibiotic treatment all characterized group C-associated pharyngitis as more severe than culture-negative sore throats but less severe than group A pharyngitis. From these data we present the first definitive evidence that group C streptococci are associated with endemic pharyngitis, show that clinical presentation distinguishes a group of patients with group C-associated pharyngitis from populations with culture-negative sore throats and from those with group A pharyngitis. Physicians' response to that presentation merits consideration in the context of rapid group-specific diagnosis of streptococcal pharyngitis by group A antigen tests.

Risk factors for attempted suicide during adolescence

It is unknown whether adolescents can provide the information necessary to identify their risk for attempted suicide. The present study was designed to determine whether data collected directly from adolescents can be used to develop a simple model for differentiating suicidal from nonsuicidal adolescents. Patients aged 13 to 19 years hospitalized for medical complications of serious suicide attempts (n = 56) or for acute illnesses unrelated to injuries or ingestions (n = 248) completed self-administered questionnaires pertaining to psychosocial function, recent stress, alcohol and drug use, and health care use. Compared with ill adolescents, suicidal adolescents had poorer mental health, impulse control, family relationships, and school performance; higher 3-month stress scores and alcohol-use scores; and more use of 7 of 12 drugs (P less than .05). Compared with ill adolescents, suicidal adolescents were more likely to report previous suicide attempts (39% vs 10%, P less than .001) and previous mental health care (27% vs 8%, P less than .001) but were less likely to identify a primary care site (61% vs 87%, P less than .001). In a logistic regression model based on previous suicide attempts, previous mental health care, poor school performance, marijuana use, and dependence on the emergency room for primary care, 84% of the suicidal and 55% of the ill adolescents were correctly identified. If validated prospectively, these five self-administered questions may constitute a helpful screen for the rapid identification of suicidal adolescents.

Evaluating physicians' probabilistic judgments

Physicians increasingly are challenged to make probabilistic judgments quantitatively. Their ability to make such judgments may be directly linked to the quality of care they provide. Many methods are available to evaluate these judgments. Graphic means of assessment include the calibration curve, covariance graph, and receiver operating characteristic (ROC) curve. Statistical tools can measure the significance of departures from ideal calibration, and measure the area under ROC curve. Modeling the calibration curve using linear or logistic regression provides another method to assess probabilistic judgments, although these may be limited by failure of the data to meet the model's assumptions. Scoring rules provide indices of overall judgmental performance, although their reliability is difficult to gauge for small sample sizes. Decompositions of scoring rules separate judgmental performance into functional components. The authors provide preliminary guidelines for choosing methods for specific research in this area.

Rational decision making based on history: adult sore throats

Primary care physicians are often required to make preliminary evaluations based only on the patient's history, especially during telephone encounters about sore throats. The authors studied adults with sore throats to determine whether patients can be stratified into higher and lower risks of strep throat by history alone. They first obtained data from 517 patients seen in an emergency room. Providers graded symptoms on a four-point scale (absent, mild, moderate, or severe). Initial analyses showed that prediction based on history should include three variables: fever, difficulty in swallowing, and cough. For ease of computation, these were consolidated into one score, "history" (= fever history + difficulty in swallowing - cough). This score was used to develop a model that predicts the probability of infection with group A beta-hemolytic streptococcus, and the model's performance was tested in two additional patient groups. The predictive accuracy of the "history" score was confirmed in all patient groups, despite differences in providers and disease prevalences. Primary care physicians may use this model to help them make decisions in situations such as telephone encounters without using additional data.

Effects of low dose oral iodide supplementation on thyroid function in normal men

Previous studies have demonstrated that short-term oral iodide administration, in doses ranging from 1500 micrograms to 250 mg/day, has an inhibitory effect on thyroid hormone secretion in normal men. As iodide intake in the USA may be as high as 800 micrograms/d, we investigated the effects of very low dose iodide supplementation on thyroid function. Thirty normal men aged 22-40 years were randomly assigned to receive 500, 1500, and 4500 micrograms iodide/day for 2 weeks. Blood was obtained on days 1 and 15 for measurement of serum T4, T3, T3-charcoal uptake, TSH, protein-bound iodide (PBI) and total iodide, and 24 h urine samples were collected on these days for measurement of urinary iodide excretion. TRH tests were performed before and at the end of the period of iodide administration. Serum inorganic iodide was calculated by subtracting the PBI from the serum total iodide. We found significant dose-related increases in serum total and inorganic iodide concentrations, as well as urinary iodide excretion. The mean serum T4 concentration and free T4 index values decreased significantly at the 1500 micrograms/day and 4500 micrograms/day doses. No changes in T3-charcoal uptake or serum T3 concentration occurred at any dose. Administration of 500 micrograms iodide/day resulted in a significant increase (P less than 0.005) in the serum TSH response to TRH, and the two larger iodide doses resulted in increases in both basal and TRH-stimulated serum TSH concentrations.(ABSTRACT TRUNCATED AT 250 WORDS)

The adult respiratory distress syndrome. Cell populations and soluble mediators in the air spaces of patients at high risk

In order to better understand the modulation of polymorphonuclear neutrophil influx into the lung during the development of the adult respiratory distress syndrome (ARDS), we evaluated bronchoalveolar lavage fluids from control subjects (n = 9), patients at high risk of developing ARDS (n = 12), and patients with ARDS (n = 11) for cellular and protein content and capacity to promote neutrophil adhesion to tissue culture plastic. Analysis of the lavage fluids from high risk patients and patients with ARDS showed an 8- to 10-fold increase in the total number of cells, an increase in the percentage of neutrophils present (control subjects = 1 +/- 0.4%, high risk = 53 +/- 8%, ARDS = 70 +/- 7%), and a 10- to 40-fold increase in protein content. The adherence of normal neutrophils to plastic surfaces after pretreatment with either concentrated lavage fluid, ultrafiltrates of BALF, or plasma samples was determined to evaluate the neutrophil adherence-promoting activity of each. Lavage fluid from high risk patients and patients with ARDS promoted an approximate 3-fold increase in neutrophil adherence when compared with control lavage fluid. Neutrophil adhesion-promoting activity of the plasma and lavage filtrates (mw less than 500 daltons) was not significantly different from that of control subjects. The adherence-promoting activity found in ARDS lavage was stable at 56 degrees C for 30 min.(ABSTRACT TRUNCATED AT 250 WORDS)

What a difference a day makes: a decision analysis of adult streptococcal pharyngitis

With the availability of group A beta-hemolytic streptococcal (GABHS) antigen detection tests, the management of adult pharyngitis is being reassessed. A decision analytic model was developed which considered four strategies: immediate treatment, no treatment, performing a rapid antigen test, or obtaining a bacterial culture. Patient outcomes were expressed in "well" days, which were reduced by the "sick" days associated with adverse reactions to treatment or complications of GABHS infection. When immediate test results are available, testing is the optimal strategy for probabilities of GABHS between 1 and 49 per cent. This range includes almost all patients, using probability estimates based on clinical criteria. The absolute benefit of testing was 0.1 days. The major advantage of a rapid test is the avoidance of penicillin reactions. Variations in the symptomatic benefits of treatment had minimal effects on the analysis. The analysis supports the use of an antigen test for adult patients with pharyngitis.

Throat cultures and rapid tests for diagnosis of group A streptococcal pharyngitis

This article reviews the use of diagnostic tests to guide management of adults with sore throats. Pharyngitis due to group A beta-hemolytic streptococci represents the major diagnostic concern in these patients. Organisms other than group A streptococci can cause pharyngitis, but their clinical importance and their diagnostic tests have not yet been established. For many years, physicians have used routine throat cultures to diagnose group A streptococcal pharyngitis. Rapid tests have recently been introduced that detect the group A streptococcal antigen on throat swab specimens. Because both tests have high sensitivity and specificity, the choice of tests may depend on test turnaround time. Rapid tests should improve management by decreasing both short-term morbidity and inappropriate use of antibiotics.

Effect of medium and cultivation conditions on comparisons between latex agglutination and culture detection of group A streptococci

In the laboratory diagnosis of pharyngitis, results from latex agglutination tests (LAT) performed directly on throat swabs are often compared with the isolation of group A beta-hemolytic streptococci (GABHS) from simultaneously obtained swabs cultivated on a variety of media under different atmospheric conditions. In this study, results of an LAT, Directigen, were compared with those of two different media: sheep blood agar (SBA) and group A selective strep agar (ssA). Specimens inoculated on SBA were incubated in three different atmospheres: air, 3 to 5% CO2, and anaerobically. Those inoculated on ssA were incubated in 3 to 5% CO2 only. Isolation of GABHS was confirmed by coagglutination. The standard for true positivity was the isolation of GABHS from at least one of the simultaneous cultures. Comparisons were made with samples from 693 adult patients. GABHS was isolated on at least one of the three cultures in 143 patients, demonstrating an isolation rate of 20.6%. LAT exhibited a sensitivity of 95.1%. SBA incubated in air, in CO2, or anaerobically had sensitivities of 86.2, 85.9, and 93.7%, respectively. The ssA detected 99.3% of the positive specimens. Single SBA culture proved to be inferior to LAT and therefore was a poor standard for measuring LAT performance. Single ssA cultures demonstrated the greatest sensitivity in GABHS detection and therefore could serve as a standard for measuring LAT performance.

Underdocumentation of benzodiazepine prescriptions in a general medicine clinic

During a three-month period the authors reviewed the charts of patients prescribed benzodiazepine and non-benzodiazepine medications by 73 housestaff practicing in an ambulatory medical clinic. Compared with non-benzodiazepine prescriptions, benzodiazepine name (p less than 0.001), instructions (p less than 0.001), and targeted problems (p less than 0.0001) were significantly underrecorded. In 11% of the records reviewed there was no indication that a mood disorder had been identified or a benzodiazepine prescribed (p less than 0.0001). Problems targeted for benzodiazepine management were found less frequently in the records of elderly patients than in those of patients less than 65 years of age (p less than 0.05). The authors conclude that many houseofficers significantly underdocument the prescriptions they write for benzodiazepine medications and that this may be a marker of their regard for managing mood disorders with benzodiazepines.

Rapid diagnosis of streptococcal pharyngitis in adult emergency room patients

A rapid latex agglutination slide test for group A beta-hemolytic streptococcal throat infections was prospectively evaluated. Resident physicians, working in an adult non-acute emergency room, recorded clinical data and collected throat swabs from 729 adult patients with sore throats. Research assistants obtained throat swabs from 329 control patients. Sensitivity and specificity, compared with routine cultures, were 96% and 97%, respectively. Analyses of clinical predictions and of test results for control patients, however, suggest that this test may perform better than routine culture. The test provides a rapid, accurate, potentially useful alternative for diagnosing group A beta-hemolytic streptococcal pharyngitis in adults.

Measures of the value of a diagnostic test derived from stochastic thresholds

Previous indices for measuring the potential impact of a diagnostic test on a physician's management of a given patient were derived based on a fixed threshold model. The authors adapted these indices to a stochastic threshold model. In the stochastic threshold model the physician's probability of treating the patient is a function of the patient's probability of disease. From this model the authors derived the management value index (the expected effect that the test has on the physician's probability for treating the patient) and the utility value index (the expected benefit to the patient if the diagnostic test is used). Graphs of the indices versus the patient's probability of disease may be useful in teaching appropriate use of diagnostic tests.

Stochastic thresholds

Thresholds have traditionally been represented by a single number; the optimal management of the patient depends on whether his probability of disease is above or below this number. The concept of a threshold as a single number, however, inadequately represents the treatment approach of a group of physicians who do not have all the same threshold or a single physician who is uncertain about the exact value of the threshold. An alternative to a single valued threshold is to consider the threshold as having a probability distribution: for every probability that the patient has the disease there is a probability that the threshold is exceeded. This "stochastic" threshold model contains information about the uncertainty of the threshold estimation. Stochastic thresholds can be useful for testing the sensitivity of a management decision to the patient's probability of disease. They can also be used for comparing the standards of practice of individual physicians or comparing the practice of an individual physician with that of a group.

Assessing the responsiveness of functional scales to clinical change: an analogy to diagnostic test performance

One characteristic of newer health or functional status scales which has received little attention is their responsiveness over time to clinical change. In part, this is because methods for assessing this characteristic are crude and not well standardized. We suggest that scales be viewed as "diagnostic tests" for discriminating between improved and unimproved patients. With this perspective, one may construct receiver operating characteristic (ROC) curves describing a scale's ability to detect improvement (or failure to improve) using some external criterion. This method is illustrated using data from a study of acute low back pain, comparing the Sickness Impact Profile, its major subscales, and a shorter, more disease-specific scale. The results demonstrate an advantage of the ROC approach over simple pre- and post-treatment score comparisons in assessing scale responsiveness. They also suggest some advantage for a brief disease-specific scale over the lengthier "generic" SIP.

Transportability of a decision rule for the diagnosis of streptococcal pharyngitis

Use of existing decision rules could improve management of streptococcal pharyngitis, but the validity of such rules outside their original patient population is not known. We applied a four-item decision rule derived at the Medical College of Virginia, Richmond, to 516 patients at the University of Nebraska, Omaha, to test how accurately it would predict outcome of throat culture. After correction for differences in the prevalence of a positive culture (17% in Virginia, 26% in Nebraska), the rule closely predicted the frequency of positive cultures in five subgroups based on the presence of clinical findings. We conclude that this rule transported well to a different patient population and would have been useful in identifying patients with pharyngitis who had a high likelihood of throat cultures positive for group A streptococci.

A Visicalc program for estimating the area under a receiver operating characteristic (ROC) curve

The area under the ROC curve interests us as a method for analyzing discrimination or detectability. One can assess a diagnostic test or probability assessor with respect to its degree of discrimination. The area under the ROC curve gives us the probability of correctly identifying abnormal from normal in a forced-choice, two-alternative problem. Previous methods used for calculating the area involved maximum likelihood estimation on a mainframe or minicomputer. This paper demonstrates an adaptation of a recently published nonparametric method for estimating the area. This adaptation takes advantage of electronic spreadsheet software and may be used on most (if not all) microcomputers. The paper develops the construction of the program needed for the necessary calculations.

An evaluation of methods for estimating the area under the receiver operating characteristic (ROC) curve

The area under the receiver operating characteristic (ROC) curve serves as one means for evaluating the performance of diagnostic and predictive test systems. The most commonly used method for estimating the area under an ROC curve utilizes the maximum-likelihood-estimation technique, and a nonparametric method to calculate the area under an ROC curve was recently described. We compared the performance of these two methods. The results for the area under the ROC curve and the standard error of the estimate as calculated by each of the two methods exhibited high correlation. Generally, the nonparametric method yields lower area estimates than the maximum-likelihood-estimation technique. However, these differences generally were small, particularly with ROC curves derived from five or more cutoff points. Consistent results of hypothesis testing of the significance of differences between two ROC curves will be similar, regardless of which method is used, as long as one uses the same estimation technique on the two curves and as long as the two ROC curves being compared are of similar shape.

Development of the adult respiratory distress syndrome: progressive alteration of neutrophil chemotactic and secretory processes

Chemotaxis and lysosomal enzyme release in peripheral blood neutrophils taken from patients before, during, and after recovery from the adult respiratory distress syndrome (ARDS) were studied. This allowed for correlation of cellular function with changes in a patient's clinical status. It was found that neutrophils from 8 of 9 patients with the fully developed syndrome exhibited a profound depression of chemotaxis (63% depressed, P = 0.0001) and a fourfold elevation of basal lysosomal enzyme release relative to neutrophils from healthy controls (P = 0.0001). These findings of depression of chemotaxis and enhanced basal enzyme release were also detected in neutrophils taken from 7 of 11 patients in whom clinical risk factors (eg, sepsis, pneumonia) for the syndrome had developed. Following resolution of the adult respiratory distress syndrome, the above changes in neutrophil function resolved in the four patients studied during convalescence. Healthy neutrophils exposed to plasma samples (untreated or zymosan-activated) from control subjects and patients with ARDS could not be distinguished with respect to chemotaxis and enzyme secretion. It is concluded that patients in whom ARDS develops show profound but reversible changes in peripheral neutrophil activity which can be measured following the development of a clinical predisposition for the syndrome. Further, the presence of a humoral substance capable of promoting chemotaxis or enzyme secretion from healthy neutrophils in the untreated plasma of patients suffering from ARDS was not demonstrated. This suggests that alteration of neutrophil activity measured in patients with the fully developed syndrome may be cellular in origin.

Why physicians don't pursue abnormal laboratory tests: an investigation of hypercalcemia and the follow-up of abnormal test results

For unknown reasons, physicians often ignore unsolicited clinical data. This is thought to impair the quality of medical care and the efficacy of screening programs. To investigate this problem the authors followed 156 consecutive hypercalcemic patients for nine to 15 months. Twenty-eight were lost to follow-up, and the hypercalcemia was ignored in 26. Calcium tests were repeated for 102, and hypercalcemia was confirmed in 53. Of these, 39 were and 14 were not further investigated. Analysis by logistic regression revealed a highly significant relationship between the degree of hypercalcemia and the likelihood that calcium testing would be repeated or that abnormal levels would be further investigated. The authors conclude that, contrary to common opinion, when physicians ignore abnormal laboratory values they are making complex clinical judgments based on the degree of abnormality, the likelihood that further investigation will affect therapy, and the cost of the risk associated with further investigation. Evaluation and attempts to modify this behavior should take into account the complexity of these decisions.