The "To Err is Human" report and the patient safety literature.

BACKGROUND: The "To Err is Human" report published by the Institute of Medicine (IOM) in 1999 called for a national effort to make health care safer. Although the report has been widely credited with spawning efforts to study and improve safety in health care, there has been limited objective assessment of its impact. We evaluated the effects of the IOM report on patient safety publications and research awards. METHODS: We searched MEDLINE to identify English language articles on patient safety and medical errors published between 1 November 1994 and 1 November 2004. Using interrupted time series analyses, changes in the number, type, and subject matter of patient safety publications were measured. We also examined federal (US only) funding of patient safety research awards for the fiscal years 1995-2004. RESULTS: A total of 5514 articles on patient safety and medical errors were published during the 10 year study period. The rate of patient safety publications increased from 59 to 164 articles per 100,000 MEDLINE publications (p<0.001) following the release of the IOM report. Increased rates of publication were observed for all types of patient safety articles. Publications of original research increased from an average of 24 to 41 articles per 100,000 MEDLINE publications after the release of the report (p<0.001), while patient safety research awards increased from 5 to 141 awards per 100,000 federally funded biomedical research awards (p<0.001). The most frequent subject of patient safety publications before the IOM report was malpractice (6% v 2%, p<0.001) while organizational culture was the most frequent subject (1% v 5%, p<0.001) after publication of the report. CONCLUSIONS: Publication of the report "To Err is Human" was associated with an increased number of patient safety publications and research awards. The report appears to have stimulated research and discussion about patient safety issues, but whether this will translate into safer patient care remains unknown.

Antidepressant use and the risk of breast cancer: a non-association.

OBJECTIVE: Animal studies have suggested that some antidepressant medications may act as breast tumor promoters and recent epidemiologic studies of antidepressant use and breast cancer outcomes in humans have also reported such associations. DESIGN: Retrospective cohort study. STUDY POPULATION: 38,273 women who filled a prescription for any of a number of antidepressants and 32,949 who filled a prescription for any other medication during 1989-1991; all were > or =20, enrolled in Medicare, Medicaid or the Pharmaceutical Assistance to the Aged and Disabled (PAAD) programs of New Jersey, and free of evidence of breast cancer. DATA COLLECTION: Antidepressant use was assessed over a period lasting up to 24 months. Subjects were followed for a maximum of 7.5 years; those who had a first diagnosis of breast cancer in the New Jersey Cancer Registry at least 3 months after their index date were considered incident breast cancer cases. Other covariates, including demographic, clinical, and health care utilization variables were also assessed. MAIN OUTCOME OF INTEREST: Adjusted hazard ratio (HR) of developing breast cancer, based on multivariable proportional hazards models. RESULTS: Use of antidepressants was unrelated to the development of breast cancer (adjusted HR = 1.04; 95% CI 0.87-1.25). No elevated risks were found for specific antidepressants, including agents found to be breast tumor promoters in animal studies, as well as drugs thought to be associated with breast cancer in prior epidemiologic studies. There was no suggestion that breast cancer risks were increased with more intensive use of antidepressants or that antidepressant use is associated with a more severe stage of cancer at diagnosis. CONCLUSIONS: Despite recent reports from much smaller epidemiologic studies or laboratory animals, these results provide reassurance that antidepressant use is not associated with the development of breast cancer.

Finding incident breast cancer cases through US claims data and a state cancer registry.

OBJECTIVE: With the increasing availability of automated health-care data, new methods are available to screen large populations for the presence of cancer diagnoses. However, it is crucial to evaluate how completely incident cancer cases can be ascertained using these data sources. METHODS: We used capture-recapture techniques to estimate the total number of incident breast cancer cases occurring within one state during a 3-year period. We then compared the ascertainment of these cases by the following two data sources: claims for breast cancer surgery recorded in Medicaid and Medicare data vs a cancer registry in the same state. RESULTS: Medicaid-Medicare breast cancer surgery claims identified 68% of the total estimated number of incident breast cancer cases while cancer registry data identified 78%. Case ascertainment improved markedly to 91% when both registry and Medicare-Medicaid data sources were used together. The sensitivity of ascertainment was lower for Medicaid-Medicare data among those aged under 65 and non-white; ascertainment was lower for the registry among women who were aged under 65, poor, and non-white. CONCLUSIONS: Combining health insurance claims data with a population-based cancer registry improved the identification of incident cases of breast cancer, and may be particularly useful among demographic groups found to be at highest risk of under-ascertainment such as younger women, the poor, and racial minorities.

Cost-effectiveness of pravastatin therapy for survivors of myocardial infarction with average cholesterol levels.

BACKGROUND: The objective of this study was to assess the cost-effectiveness of pravastatin therapy in survivors of myocardial infarction with average cholesterol levels. METHODS: We performed a cost-effectiveness analysis based on actual clinical, cost, and health-related quality-of-life data from the Cholesterol and Recurrent Events (CARE) trial. Survival and recurrent coronary heart disease events were modeled from trial data in Markov models, with the use of different assumptions regarding the long-term benefit of therapy. RESULTS: Pravastatin therapy increased quality-adjusted life expectancy at an incremental cost of $16,000 to $32,000 per quality-adjusted life-year gained. In subgroup analyses, the cost-effectiveness of pravastatin therapy was more favorable for patients >60 years of age and for patients with pretreatment low-density lipoprotein cholesterol levels >125 mg/dL. Results were sensitive to the cost of pravastatin and to assumptions about long-term survival benefits from pravastatin therapy. CONCLUSIONS: The cost-effectiveness of pravastatin therapy in survivors of myocardial infarction with average cholesterol levels compares favorably with other interventions.

Health care utilization among homeless adults prior to death.

This study characterizes health care utilization prior to death in a group of 558 homeless adults in Boston. In the year before death, 27 percent of decedents had no outpatient visits, emergency department visits, or hospitalizations except those during which death occurred. However, 21 percent of homeless decedents had a health care contact within one month of death, and 21 percent had six or more outpatient visits in the year before death. Injection drug users and persons with HIV infection were more likely to have had contact with the health care system. This study concludes that homeless persons may be underusing health care services even when they are at high risk of death. Because a subset of homeless persons had extensive health care contacts prior to death, opportunities to prevent deaths may have been missed, and some deaths may not have been preventable through medical intervention.

The mode selection trial (MOST) in sinus node dysfunction: design, rationale, and baseline characteristics of the first 1000 patients.

BACKGROUND: More than 200,000 permanent pacemakers will be implanted in the United States in 2000 at a cost of more than $2 billion. Sick sinus syndrome (SSS) will likely account for approximately half of all cases necessitating implantation. Pacemaker technology permits the selection of ventricular (single-chamber) or dual-chamber devices. However, clinical and outcomes data are inadequate to support a clear recommendation that one or the other type of device be used. METHODS: The Mode Selection Trial (MOST) is a single-blind study supported by the National Heart, Lung, and Blood Institute designed to enroll 2000 patients with SSS. All patients will receive a DDDR pacemaker programmed to VVIR or DDDR before implantation. The average time of follow-up will be 3 years. MOST has a >90% power to detect a 25% reduction in the primary end point-nonfatal stroke or total (all cause) mortality-in the DDDR-treated group. Secondary end points will include health-related quality of life and cost effectiveness, atrial fibrillation, and development of pacemaker syndrome. Prespecified subgroups for analysis will include women and the elderly. Enrollment was completed in October 1999, with a total of 2010 patients. RESULTS: The median age of the first 1000 enrolled patients is 74 years, with 25% of patients 80 years or older. Women comprise 49%, and 17% are nonwhite, predominantly black (13%). Before pacemaker implantation, 22% of patients reported a history of congestive heart failure, 11% coronary angioplasty, and 25% coronary bypass surgery. Supraventricular tachycardia including atrial fibrillation was present in 53% of patients. A prior stroke was reported by 12%. Antiarrhythmic therapy was in use in 18% of patients. CONCLUSIONS: MOST will fill the clinical need for carefully designed prospective studies to define the benefits of dual-chamber versus single-chamber ventricular pacing in patients with SSS. The MOST population is typical of the overall pacemaker population in the United States. Thus the final results of MOST should be clinically generalizable.

Strategies for improving comorbidity measures based on Medicare and Medicaid claims data.

Claims-based measures of comorbid illness severity have generally relied on the diagnoses listed for a single hospitalization. Unfortunately, such diagnostic information is often limited because patients have not been hospitalized during periods of interest, because of incomplete coding of diagnoses on claims forms, or because listed diagnoses represent complications of the hospitalization rather than pre-existing comorbid conditions. To address these limitations, we developed and tested four comorbidity index scores for patients with breast cancer, each based on different sources of health services claims from Medicare and Medicaid: hospitalization for breast cancer surgery; outpatient care prior to the hospitalization; other inpatient care prior to the hospitalization; and all sources combined. Varying the number and type of sources of diagnostic information yielded only very small improvements in the prediction of mortality at 1 and 3 years. Surprisingly, even simpler measures of comorbidity (crude number of diagnoses) and of prior health care utilization (total days spent in the hospital) performed at least as well in predicting mortality as did the more complex index scores which assigned points and weights for specific conditions. The greatest improvement in explanatory power was observed when another source of clinical information (cancer stage derived from a population-based cancer registry) was used to supplement claims information. Expanding the source of claims diagnoses and focusing on time periods prior to an index hospitalization are insufficient for substantially improving the explanatory power of claims-based comorbidity indices. Other improvements suggested by our results should include: increasing the completeness and accuracy of claims diagnoses; supplementing diagnoses with health care utilization information in claims data; and supplementing claims data with other sources of clinical information.

Negligent care and malpractice claiming behavior in Utah and Colorado.

BACKGROUND: Previous studies relating the incidence of negligent medical care to malpractice lawsuits in the United States may not be generalizable. These studies are based on data from 2 of the most populous states (California and New York), collected more than a decade ago, during volatile periods in the history of malpractice litigation. OBJECTIVES: The study objectives were (1) to calculate how frequently negligent and nonnegligent management of patients in Utah and Colorado in 1992 led to malpractice claims and (2) to understand the characteristics of victims of negligent care who do not or cannot obtain compensation for their injuries from the medical malpractice system. DESIGN: We linked medical malpractice claims data from Utah and Colorado with clinical data from a review of 14,700 medical records. We then analyzed characteristics of claimants and nonclaimants using evidence from their medical records about whether they had experienced a negligent adverse event. MEASURES: The study measures were negligent adverse events and medical malpractice claims. RESULTS: Eighteen patients from our study sample filed claims: 14 were made in the absence of discernible negligence and 10 were made in the absence of any adverse event. Of the patients who suffered negligent injury in our study sample, 97% did not sue. Compared with patients who did sue for negligence occurring in 1992, these nonclaimants were more likely to be Medicare recipients (odds ratio [OR], 3.5; 95% CI [CI], 1.3 to 9.6), Medicaid recipients (OR, 3.6; 95% CI, 1.4 to 9.0), > or =75 years of age (OR, 7.0; 95% CI, 1.7 to 29.6), and low income earners (OR, 1.9; 95% CI, 0.9 to 4.2) and to have suffered minor disability as a result of their injury (OR, 6.3; 95% CI, 2.7 to 14.9). CONCLUSIONS: The poor correlation between medical negligence and malpractice claims that was present in New York in 1984 is also present in Utah and Colorado in 1992. Paradoxically, the incidence of negligent adverse events exceeds the incidence of malpractice claims but when a physician is sued, there is a high probability that it will be for rendering nonnegligent care. The elderly and the poor are particularly likely to be among those who suffer negligence and do not sue, perhaps because their socioeconomic status inhibits opportunities to secure legal representation.

Nonurgent emergency department visits: the effect of having a regular doctor.

OBJECTIVES: The authors assess the association between having a regular doctor and presentation for nonurgent versus urgent emergency department visits while controlling for potential confounders such as sociodemographics, health status, and comorbidity. METHODS: A cross-sectional study was conducted in emergency departments of five urban teaching hospitals in the northeast. Adult patients presenting with chest pain, abdominal pain, or asthma (n = 1696; 88% of eligible) were studied. Patients completed a survey on presentation, reporting sociodemographics, health status, comorbid diseases, and relationship with a regular doctor. Urgency on presentation was assessed by chart review using explicit criteria. RESULTS: Of the 1,696 study participants, 852 (50%) presented with nonurgent complaints. In logistic regression analyses, absence of a relationship with a regular physician was an independent correlate of presentation for a nonurgent emergency department visit (odds ratio 1.6; 95% confidence interval 1.2, 2.2) when controlling for age, gender, marital status, health status, and comorbid diseases. Race, lack of insurance, and education were not associated with nonurgent use. CONCLUSIONS: Absence of a relationship with a regular doctor was correlated with use of the emergency department for selected nonurgent conditions when controlling for important potential confounders. Our study suggests that maintaining a relationship with a regular physician may reduce nonurgent use of the emergency department regardless of insurance status or health status.

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.

Using a computerized sign-out program to improve continuity of inpatient care and prevent adverse events.

BACKGROUND: Many medical injuries are preventable, but there are few reported successful strategies to prevent such injuries. Previous work identified coverage by house staff not primarily responsible for the patient (cross-coverage) as a significant correlate of risk for preventable adverse events. A four-month intervention--computerized sign-outs--was introduced in 1993 in an urban teaching hospital to improve continuity of care during cross-coverage and thereby reduce risk for preventable adverse events. MEASUREMENTS: A previously tested confidential self-report system was used to identify adverse events, which were defined as unexpected complications of medical therapy that resulted in increased length of stay or disability at discharge. A panel of three board-certified internists confirmed events and evaluated preventability based on case summaries. RESULTS: After the intervention, the rate of preventable adverse events among the 3,747 patients admitted to the medical service decreased from 1.7% to 1.2% (p < 0.10). Both univariate and multivariate analysis revealed no association between cross coverage and preventable adverse events after the intervention. In the baseline period, the odds ratio (OR) for a patient suffering a preventable adverse event during cross coverage was 5.2 (95% confidence interval [CI], 1.5-18.2; p = 0.01), but was no longer significant after the intervention (OR, 1.5; 95% CI, 0.2-9.0). CONCLUSION: House staff are willing participants in efforts to measure and improve the quality of health care systems. The intervention may have reduced the risk for medical injury associated with discontinuity of inpatients care. Four years after the end of the study, the computerized sign-out program remained an integral part of the computing support system for house staff and was widely used.

The effect of change of health insurance on access to care.

This study examines how changes in health insurance status affect patients and their care. Results show that, controlling for socioeconomic factors, condition, age, and urgency, patients who lost insurance and patients who changed insurance were more likely to delay seeking care within the four months after visiting an emergency department than people whose health insurance status did not change. Patients who lost coverage were more likely to report no primary care provider and were less likely to have recommended follow-up care within the four-month period. Loss of insurance also was associated with lower likelihood of vaccine use and check-ups in the prior year. The study confirms that a loss or change in health insurance in the prior year has a measurable effect on access to health care. The greatest impact was among patients who lost insurance, though patients who changed health plans also were more likely to delay seeking care than patients whose health insurance status did not change.

Health-related quality of life after elective surgery: measurement of longitudinal changes.

OBJECTIVE: To examine the responsiveness of the 36-Item Short Form Health Survey (SF-36) to clinical changes in three surgical groups and to study how health-related quality of life (HRQL) changes with time among patients who undergo total hip arthroplasty, thoracic surgery for treatment of non-small-cell lung cancer, or abdominal aortic aneurysm (AAA) repair. DESIGN: Prospective cohort study with serial evaluations of HRQL preoperatively and at 1, 6, and 12 months after surgery. SETTING: University tertiary care hospital. PATIENTS: Of 528 patients, more than 50 years of age, who were admitted for these elective procedures, 454 (86%) provided preoperative health status data and are members of the study cohort. At 12 months after surgery, 439 (93%) of the cohort was successfully contacted and 390 (90%) provided follow-up interviews. MEASUREMENTS AND MAIN RESULTS: The Medical Outcomes Study SF-36, the Specific Activity Scale, five validated health transition questions, and a 0 to 100 scale measure of global health were used to assess changes in health status at 1, 6, and 12 months after surgery. Change in health status as measured by the SF-36 demonstrated that physical function and role limitations due to physical health problems were worse 1 month after these three surgeries. However, by 6 months after surgery, most patients experienced significant gains in the majority of the dimensions of health, and these gains were sustained at 12 months after surgery. Longitudinal changes in the SF-36 were positively associated with responses to the five health transition questions, to changes on the Specific Activity Scale and global health rating question, and to clinical parameters for persons who had AAA repair. These findings indicate that the SF-36 has evidence of validity and is responsive to expected changes in HRQL after elective surgery for these procedures. CONCLUSIONS: For the total hip arthroplasty patients, responsiveness was greatest for the SF-36 scales that measure physical constructs. However, for the two other procedures and at various points of recovery, significant changes were observed for all eight subscales, suggesting that responsiveness was dependent on the type of surgery and the timing of follow-up, and that multidimensional measures are needed to fully capture changes in HRQL after surgery.

Does the computerized display of charges affect inpatient ancillary test utilization?

BACKGROUND: The computerized display of charges for ancillary tests in outpatients has been found to affect physician-ordering behavior, but this issue has not been studied in inpatients. OBJECTIVE: To assess whether the computerized display of charges for clinical laboratory or radiological tests affected physician-ordering behavior. PATIENTS AND METHODS: Two prospective controlled trials, randomized by patient, were performed. Each trial included all medical and surgical inpatients at 1 large teaching hospital during 4 and 7 months: 3536 intervention and 3554 control inpatients in the group with clinical laboratory tests, and 8728 intervention and 8653 control inpatients in the group with radiological tests. The intervention consisted of the computerized display of charges for tests at the time of ordering. MAIN OUTCOME MEASURES: The number of clinical laboratory and radiological tests ordered per admission and the charges for these tests. RESULTS: For the clinical laboratory tests, during a 4-month study period, patients in the intervention group had 4.5% fewer tests ordered, and the total charges for these tests were 4.2% lower, although neither difference was statistically significant. Compared with historical controls from the same 4-month period a year before, the charges for the tests per admission had decreased 13.3%, but the decrease was temporally correlated with a restriction of future ordering of tests, and not with the introduction of the display of charges. For the radiological tests, during a 7-month period, the intervention group had almost identical numbers of tests ordered and charges for these tests. CONCLUSIONS: The computerized display of charges had no statistically significant effect on the number of clinical laboratory tests or radiological procedures ordered or performed, although small trends were present for clinical laboratory tests. More intensive interventions may be needed to affect physician test utilization.

Epidemiology of sepsis syndrome in 8 academic medical centers.

CONTEXT: Sepsis syndrome is a leading cause of mortality in hospitalized patients. However, few studies have described the epidemiology of sepsis syndrome in a hospitalwide population. OBJECTIVE: To describe the epidemiology of sepsis syndrome in the tertiary care hospital setting. DESIGN: Prospective, multi-institutional, observational study including 5-month follow-up. SETTING: Eight academic tertiary care centers. METHODS: Each center monitored a weighted random sample of intensive care unit (ICU) patients, non-ICU patients who had blood cultures drawn, and all patients who received a novel therapeutic agent or who died in an emergency department or ICU. Sepsis syndrome was defined as the presence of either a positive blood culture or the combination of fever, tachypnea, tachycardia, clinically suspected infection, and any 1 of 7 confirmatory criteria. Estimates of total cases expected annually were extrapolated from the number of cases, the period of observation, and the sampling fraction. RESULTS: From January 4, 1993, to April 2, 1994, 12759 patients were monitored and 1342 episodes of sepsis syndrome were documented. The extrapolated, weighted estimate of hospitalwide incidence (mean+/-95% confidence limit) of sepsis syndrome was 2.0+/-0.16 cases per 100 admissions, or 2.8+/-0.17 per 1000 patient-days. The unadjusted attack rate for sepsis syndrome between individual centers differed by as much as 3-fold, but after adjustment for institutional differences in organ transplant populations, variation from the expected number of cases was reduced to 2-fold and was not statistically significant overall. Patients in ICUs accounted for 59% of total extrapolated cases, non-ICU patients with positive blood cultures for 11%, and non-ICU patients with negative blood cultures for 30%. Septic shock was present at onset of sepsis syndrome in 25% of patients. Bloodstream infection was documented in 28%, with gram-positive organisms being the most frequent isolates. Mortality was 34% at 28 days and 45% at 5 months. CONCLUSIONS: Sepsis syndrome is common in academic hospitals, although the overall rates vary considerably with the patient population. A substantial fraction of cases occur outside ICUs. An understanding of the hospitalwide epidemiology of sepsis syndrome is vital for rational planning and treatment of hospitalized patients with sepsis syndrome, especially as new and expensive therapeutic agents become available.

Objective assessment of sleep and alertness in medical house staff and the impact of protected time for sleep.

We studied 26 physicians in postgraduate medical training ("house staff") to objectively quantify their sleep, alertness, and psychomotor performance while working on call. This study provided precise data on the extent of sleep deprivation during a typical call night, the workload factors predictive of sleep loss, and the extent to which protected time for sleep within the call night can ameliorate sleep loss and consequent daytime sleepiness. We used ambulatory EEG recording equipment and a standardized computer-based performance test to monitor sleep and alertness over the course of a 36-hour call day. Comparisons were made between interns provided with 4 hours of protected time for sleep by a covering resident ("night-float") and interns without such coverage. As anticipated, we found evidence that hospital interns were severely sleep-deprived, to an extent even greater than prior behavioral observations have suggested. Interns in both conditions spent an average of less than 5 hours (295.4 minutes) in bed attempting to sleep and obtained an average of 3.67 hours (220.1 minutes) of sleep (range 37.4-358.4 minutes). Provision of the night-float for 4 hours did not significantly change total sleep time (TST) (212.8 minutes covered vs. 224.9 minutes uncovered), but sleep efficiency was significantly improved (86.5% vs. 70.3%; p = 0.001). Covered interns also obtained significantly more slow-wave sleep than the uncovered interns (65.4 minutes vs. 51.1 minutes; p = 0.05). However, measures of alertness and performance were not significantly different between the two groups and were only weakly related to TST. These data suggest that significant chronic sleep deprivation is relatively unaffected by sleep obtained in the hospital and that provision of protected time for sleep does not significantly improve TST.

Cost-effectiveness of prescription recommendations for cholesterol-lowering drugs: a survey of a representative sample of American cardiologists.

OBJECTIVES: We sought to determine the cost-effectiveness of the recommendations of cardiologists for the pharmacologic treatment of hypercholesterolemia. BACKGROUND: Despite the publication of guidelines such as the report of the National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults, little is known about the national prescribing practices of physicians and how they compare with the recommendations of cost-effectiveness analyses. METHODS: Under the auspices of the Cardiovascular Norms Committee of the American College of Cardiology, a nationally representative sample of cardiologists was surveyed, and their recommendations for the pharmacologic treatment of hypercholesterolemia were assessed to determine cost-effectiveness. RESULTS: The 346 responding cardiologists were reasonably representative of the membership of the American College of Cardiology. For the 12 hypothetical patients, the cardiologists recommended pharmacologic treatment more commonly in cases in which previously published studies estimated the treatment to be more cost-effective, although there was a tendency to recommend such treatment for primary prevention even when it was estimated to cost well over $100,000/year of life saved. CONCLUSIONS: These findings suggest that the cardiologists' pharmacologic recommendations for lowering lipids are correlated with published cost-effectiveness analyses. However, substantial variation in their recommendations remains, with somewhat less aggressive treatment for secondary prevention and more aggressive treatment for primary prevention than would be recommended on the basis of cost-effectiveness analyses.

Patient noncompliance with medical advice after the emergency department visit.

STUDY OBJECTIVE: To determine the correlates of compliance with follow-up appointments and prescription filling after an emergency department visit. METHODS: This prospective cohort study was undertaken as part of the Emergency Department Quality Study evaluation of five urban teaching hospital EDs in the northeastern United States. Of 2,757 eligible patients who presented with abdominal pain, asthma, chest pain, hand lacerations, head trauma, or first-trimester vaginal bleeding and were enrolled during 1-month period, 2,315 (84%) completed on-site baseline questionnaires. Information about diagnoses, socioeconomic status, discharge instructions, insurance status, and primary care was obtained from the on-site patient surveys and from reviews of medical records. A 76% random sample of patients who completed the questionnaire was generated, and 1,386 patients (79% of the sample) were interviewed by telephone approximately 10 days after their ED visit to determine compliance with follow-up appointments and prescription filling. RESULTS: Of the 1,386 patients interviewed at 10 days, 914 (66%) had been discharged from the ED, and 408 (45%) of those discharged recalled being advised to take a medication. Fifty of these patients (12%) reported that they did not obtain the medication. Significant independent correlates of not filling prescriptions were lack of insurance (odds ratio [OR], 2.4; 95% confidence interval [CI], 1.1 to 5.5) and dissatisfaction with discharge instructions (OR, 2.8; 95% CI, 1.2 to 6.4). Two hundred thirty-five (26%) of the discharged patients said they were given follow-up appointments and did not have an appointment pending at the time of the interview; 77 (33%) of these patients reported having missed their appointment. The only significant independent correlate of missing follow-up appointments was being given a telephone number to call instead of leaving the ED with an appointment scheduled (OR, 3.8; 95% CI, 1.7 to 8.8). CONCLUSION: Not having an appointment made before leaving the ED was an independent correlate of missing follow-up appointments. Lack of insurance and dissatisfaction with discharge instructions were independent correlates of not filling prescriptions.

Comparison of the Medical Outcomes Study Short-Form 36-Item Health Survey in black patients and white patients with acute chest pain.

Few data are available regarding the performance of the Medical Outcomes Study (MOS) Short-Form 36-Item Health Survey (SF-36) in black patients. In this article, the reliability and validity of the MOS SF-36 is compared in a population of black patients and white patients with acute chest pain. The MOS SF-36 was administered to 1,160 patients (31% black) who presented to the emergency department of an urban teaching hospital with acute chest pain from October 1990 to May 1992. In unadjusted analyses, black patients had significantly lower scores compared with white patients for several dimensions of the SF-36. Correlations among the eight subscales were similar, and the internal consistency of each of the eight subscales was excellent for both groups (Cronbach's coefficient alpha range .64 to .93). Each subscale had similar clinical and nonclinical correlates in black patients and white patients. In multivariate models, race was not a significant independent correlate of any of the eight subscales. Thus, the MOS SF-36 had similar reliability and validity in this population of black patients and white patients with acute chest pain who presented to an urban teaching hospital. If these findings are confirmed in other populations, they suggest that results from the MOS SF-36 may be interpreted similarly in black patients and white patients, after adjusting for clinical and sociodemographic data. Whether these findings are generalizable to other conditions and less acute settings requires further investigation.

Using patient reports to measure health care system performance.

We developed a self-administered patient questionnaire that asks for data concerning the time to receive services (access to care), communication between providers (coordination of care), and follow up after tests and treatment (continuity of care). From these data, we construct rates of performance about the clinical management systems that support provision of these services. Rates of system performance are calculated for indicators using patients' responses to survey questions. These indicators add the number of patients reporting a problem of those patients who have encountered a particular clinical management system. Information derived from 3000 patient questionnaires is matched with data abstracted from health care medical records. The sensitivity and specificity of patient reports are being evaluated for all indicators classified as gold standards for medical records. Indicators considered gold standard items for patient reports are matched for agreement with any information contained in the medical record. Also, patient characteristics associated with accurate reporting is to be assessed using multivariate logistic regression models.