The clinical utility of the leukocyte differential in emergency medicine.

Leukocyte differentials from 468 emergency room patients were assessed for clinical value by determining their associations with diagnosis, disposition, therapy, and prognosis. The test efficiency of an elevated band count as an indicator of infectious or inflammatory disease is 86%. However, all but 2 of the 99 patients in this disease category had additional indicators of inflammation, including elevated temperatures and/or white blood cell (WBC) counts. The band count lacks utility beyond this limited function. The remaining parameters of the differential count correlate poorly with all diagnostic subsets. The use of antibiotics correlates well with fever and WBC count (r = 0.95) and less well with the differential count, bands (r = 0.85), and granulocytes (r = 0.5). Elevations in the total WBC count and the band count are each associated with an increased likelihood of hospitalization. However, in the absence of leukocytosis, an elevated band count was instrumental in suggesting admission for only one patient. The patient's outcome correlates poorly with the total WBC and differential count. It is concluded that most leukocyte differentials performed for emergency room patients provide information that is no more clinically significant than that obtained from the medical history, physical examination, and absolute leukocyte count.

Failure to validate a predictive model for refusal of care to emergency-department patients.

OBJECTIVE: To determine whether previously developed triage criteria for refusal of care to patients presenting to an emergency department (ED) with nonurgent problems could be validated for an independent patient population. METHODS: A convenience sample of 534 adults presenting to a municipal hospital ED between July 1, 1992, and October 15, 1992, who met preestablished criteria for refusal of care were entered into a prospective, observational, cohort study. The single target outcome variable was hospitalization. In order to optimize the criteria's performance, both the triage nurse and the physician caring for the patient had to agree that all criteria for "refusal of care" were specifically met. No patient was refused care, nor was a patient's management or disposition interfered with in any way by the investigators. All patients were followed until hospital admission or release from the ED. RESULTS: Six (1.1%) of 534 patients (95% CI 0.4-2.4) who met the criteria for refusal of care were hospitalized. This represents a greater than 50-fold difference in incidence of hospitalization when compared with that found by other investigators, who reported that only 0.02% (95% CI 0.0004-0.04) of those patients who were refused care subsequently required hospitalization (p < 10 (-7)). CONCLUSION: The authors were unable to validate a previously developed predictive model for refusal of care to patients presenting to an ED. Refusal of care to selected ED patients based on current guidelines is not a viable solution to overcrowding. Alternative strategies must be sought.

Methodology-dependent variation in documentation of outcome predictors in out-of-hospital cardiac arrest.

OBJECTIVE: To identify variation in outcome predictor documentation in out-of-hospital cardiac arrest associated with two different methods of data collection: concurrent questioning of personnel following a resuscitation attempt and archival report review. METHODS: All patients > or = 18 years old who had out-of-hospital cardiac arrests, verified using the New York City 911 telephone system, between October 1, 1990, and April 1, 1991, were eligible for inclusion. The authors reviewed the first 200 cases of presumed primary cardiac arrest involving a resuscitation attempt among 3,243 consecutive ambulance call reports for cardiac arrest occurring during the study period. This archival data set was compared with data for the same 200 cases gathered through direct interview of field personnel by trained paramedics. The two data sets had been compiled independently by different individuals, using the same data collection instrument, which conformed to the Utstein template. RESULTS: Comparison of the data obtained from ambulance records with the data obtained from interviews of prehospital personnel revealed several areas of variance. Of note was a significantly lower proportion of bystander-witnessed ventricular fibrillation (VF) in the data set gathered from written reports (7% vs 18%; 95% CI for the difference 4-18%; p = 0.001). CONCLUSION: differences in methods of collection of out-of-hospital cardiac arrest data are associated with a more than twofold variation in the reported incidences of witnessed cardiac arrests manifesting as VF. Methodology-dependent variation in this important "denominator" may produce substantially different estimates of survival within the same cohort of patients.

Domestic violence: diurnal mismatch between need and availability of services.

OBJECTIVE: To determine whether a diurnal discordance exists between need and availability of services for victims of domestic violence. METHODS: A consecutive sample of women presenting to a municipal hospital ED with physical injuries suspected to be related to domestic violence were entered into a registry. Date and time of presentation and perceived need for services information were collected from all patients who answered affirmatively a screening question for domestic violence and whose conditions did not preclude administration of the data collection instrument. The Social Service Departments of all of the 53 911-receiving hospitals in New York City were contacted to ascertain availability of social services for victims of domestic violence by time of day. RESULTS: Twenty-eight of 32 (88%; 95% CI: 71%, 97%) victims of domestic violence presented to the ED during hours other than weekday 9 AM to 5 PM. Of these, 63% desired counseling, 32% lacked a safe place to go, and 82% had children. Of those who had children, 48% were concerned for the children's safety. In-hospital social services were universally available weekday daytime (9 AM to 5 PM) but were available in only 11% of hospital (95% CI: 4%, 23%) at other times. CONCLUSION: Approximately nine of ten victims of domestic violence presented to the ED during hours when only about one hospital in ten can provide the special services these patients require. A marked diurnal mismatch appears to exist between availability of domestic violence services in New York City and the need for these services as measured by a representative sample drawn from an ED population.

The effect of soft cervical collars on persistent neck pain in patients with whiplash injury.

OBJECTIVE: To assess the efficacy of soft cervical collars in the early management of whiplash-injury-related pain. METHODS: A controlled, clinical trial was conducted in an urban ED. Adults with neck pain following automobile crashes indicated their initial degrees of pain on a visual analog scale. Patients with cervical spine fractures or subluxation, focal neurologic deficits, or other major distracting injuries were excluded. Patients were assigned to receive a soft cervical collar or no collar based on their medical record numbers. Pain at > or = 6 weeks postinjury was coded as none, better, same, or worse, and analyzed as 3 dichotomous outcomes: recovered (pain = none); improved (pain = none or better); and deteriorated (pain = worse). RESULTS: Of 250 patients enrolled, 196 (78%) were available for follow-up. Of these patients, 104 (53%) were assigned to the soft cervical collar group, and 92 (47%) to the control group. These groups were similar in age, gender, seat position in the car, seat belt use, and initial pain score. Pain persisted at > or = 6 weeks in 122 (62%) patients. The groups showed no difference in follow-up pain category (p = 0.59). There was no significant difference between the 2 groups in complete recovery (p = 0.34), improvement (p = 0.34), or deterioration (p = 0.60). The study had a power of 80% to detect an absolute difference of at least 20% in recovery, 17% in improvement, and 7% in deterioration (2-tailed, alpha = 0.05). CONCLUSIONS: Most patients with whiplash injuries have persistent pain for at least 6 weeks. Soft cervical collars do not influence the duration or degree of persistent pain.

Variations in the electrocardiograms of young adults: are revised criteria for thrombolysis needed?

OBJECTIVE: The evaluation of chest pain in young adults has changed with the recognition of cocaine-induced myocardial ischemia. The high frequency of abnormal electrocardiograms (56-84%) in the cocaine-user population is largely due to "normal" electrocardiographic variants (early repolarization). The authors sought to determine the frequency of these "normal" variants in a young population, and whether these findings can be confused with acute ischemia. METHODS: A prospective convenience sample of subjects aged 18 to 35 without known heart disease was interviewed and had 12-lead electrocardiographic tracings performed. An emergency physician (physician 1) and a cardiologist (physician 2) read the tracings while blinded to patient history, age, and race. When the physicians disagreed, another physician adjudicated the diagnosis. RESULTS: Four hundred fourteen subjects (127 black, 175 Hispanic, and 112 Caucasian) were enrolled. Overall, 154 tracings (37%) were normal, 245 (59%) were abnormal but nondiagnostic for ischemia, and 15 (4%) were consistent with ischemia. Frequencies of repolarization "abnormalities" as determined by physicians 1 and 2, respectively, were: blacks, 32%, 51%; Hispanics, 26%, 35%; Caucasians, 17%, 27%; chi-squared, p = 0.02 and 0.0004. Patients with ischemic electrocardiograms according to physician 1 had a high frequency of repolarization "abnormalities" according to physician 2, and vice versa (100%, 61%). Electrocardiographic criteria for thrombolytic use per physician 2 were present in 31 patients (7%): blacks, 9%; Hispanics, 10%; and Caucasians, 2%; chi-squared, p = 0.03.

Cocaine-associated chest pain: one-year follow-up.

OBJECTIVE: To determine the one-year mortality and incidence of myocardial infarction (MI) post-hospital discharge or ED release for patients with cocaine-associated chest pain. METHODS: A prospective, observational study of an inception cohort of consecutive patients who presented to one of four municipal hospital EDs with cocaine-associated chest pain. Patients were followed for one year from the end of the enrollment period. Main outcome parameters were the one-year actuarial survival and the frequency of nonfatal MI. RESULTS: Mortality data were available for all 203 patients at a mean of 408 days. Additional clinical information was available for 185 patients (91%). There were six deaths (one-year actuarial survival 98%; 95% CI, 95-100%); none from MI. Nonfatal MI occurred in two patients (1%; 95% CI, 0-2%). Continued cocaine use was common (60%; 95% CI, 52-68%) and was associated with recurrent chest pain (75% vs 31%, p < 0.0001). No MI or death was reported for patients who claimed to have ceased cocaine use. CONCLUSIONS: Patients who presented with cocaine-associated chest pain commonly continued to use cocaine after discharge. Urgent evaluation of coronary anatomy or cardiac stress tests may not be necessary for patients for whom MI is ruled out and who do not have recurrent potentially ischemic pain. The subsequent risk for MI and death in this group appears to be low. Intervention strategies should emphasize cessation of cocaine use.

Prospective multicenter evaluation of cocaine-associated chest pain. Cocaine Associated Chest Pain (COCHPA) Study Group.

OBJECTIVE: To describe a large cohort of patients who had chest pain following cocaine use, and to determine the incidence of and clinical characteristics predictive for myocardial infarction in this group of patients. METHODS: A prospective observational cohort study of consecutive patients with cocaine-associated chest pain was conducted in six municipal hospital emergency departments (EDs). Demographic variables, drug abuse patterns, medical histories, chest pain characteristics, ECG results, and laboratory data were recorded. Myocardial infarction was the primary endpoint. RESULTS: Fourteen of 246 patients (5.7%; 95% confidence interval [CI], 2.7-8.7%) had myocardial infarction, as diagnosed by elevated CK-MB isoenzyme levels. There were two deaths (0.8%). The patients had a median age of 33 years. The majority were male (71.5%), non-white (83.3%), cigarette smokers (83.3%) who used cocaine regularly. Chest pain began a median of 60 minutes after cocaine use and persisted for a median of 120 minutes. Chest pain was most frequently described as substernal (71.3%) and pressure-like (46.7%). Shortness of breath (59.3%) and diaphoresis (38.6%) were common. There was no clinical difference between patients who had myocardial infarctions and those who did not. Twelve patients had arrhythmias and four had congestive heart failure. All cases requiring intervention were evident upon presentation. An ECG revealing ischemia or infarction had a sensitivity of 35.7% for predicting a myocardial infarction. The specificity, positive predictive value, and negative predictive value of the ECGs were 89.9%, 17.9%, and 95.8%, respectively. CONCLUSIONS: Myocardial infarction in patients who have cocaine-associated chest pain is not uncommon. No clinical parameter available to the physician can adequately identify patients at very low risk for myocardial infarction. Therefore, all patients with cocaine-associated chest pain should be evaluated for myocardial infarction.

Clinical criteria for the detection of pneumonia in adults: guidelines for ordering chest roentgenograms in the emergency department.

Adults presenting to an emergency department with acute respiratory illness were studied prospectively in an effort to identify sensitive clinical criteria for the diagnosis of pneumonia. Of 308 patients studied, 118 (38%) had definite or equivocal infiltrates and were considered to have pneumonia. No single symptom or sign was reliably predictive of pneumonia. Cough was the most common symptom in patients with pneumonia (86%), but was equally common in those with other respiratory illness. Fever was absent in 36 patients with pneumonia (31%). Abnormal findings on lung examination, that is, rales, rhonchi, decreased breath sounds, wheezes, altered fremitus, egophony, and percussion dullness, were each found in fewer than half of the patients with pneumonia. Twenty-six patients (22%) with a completely normal chest examination had pneumonia. Abnormal vital signs (temperature greater than 37.8 degrees C (100 degrees F), pulse greater than 100/min, or respirations greater than 20/min) were 97% sensitive for the detection of pneumonia. These criteria retained their sensitivity when films were subjected to a second, blinded interpretation by a senior radiologist. We conclude that restricting chest roentgenograms to patients with at least one abnormal vital sign will detect almost all radiographically demonstrable pneumonia in adult emergency department patients.

Circadian variation in the frequency of myocardial infarction and death associated with acute pulmonary edema.

The records of 103 patients hospitalized for acute pulmonary edema were reviewed to determine the relationship between short-term outcome and time of presentation to the emergency department. Although only 17% of the patients arrived in the emergency department during the early afternoon hours, half of the deaths in the study occurred in this group. Patients presenting between noon and four PM had a significantly higher incidence of acute myocardial infarction (76% v. 28%, p = 0.03) and death (47% v. 9% p = 0.03) compared with patients presenting at other times. Differences in the pathophysiology of daytime versus nocturnal acute pulmonary edema may account for some of the variation in outcome.

"Abnormal" electrocardiograms in patients with cocaine-associated chest pain are due to "normal" variants.

"Abnormal" electrocardiograms are found in 56% to 84% of patients with cocaine-associated chest pain. This study was designed to assess whether these findings can be explained by "normal" variations in young patients' electrocardiograms. This cross-sectional study was conducted in a municipal hospital emergency department and walk-in clinic. History and results of an electrocardiogram for consecutive patients with cocaine-associated chest pain, aged 18 to 35 years, were compared to normal controls matched for age, race, and gender. Electrocardiograms underwent detailed analysis by two physicians blinded to both the study protocol and the hypothesis. Interphysician concordance for electrocardiographic diagnosis was substantial. There were 112 patients enrolled, 56 in each group. There was no significant difference found in the mean frequency of electrocardiographic diagnoses between the cocaine-associated chest pain patients and controls. The early repolarization variant was common. In conclusion, "normal" variations (J point and ST segment elevations) account for many of the "abnormal" electrocardiograms observed in young patients with cocaine-associated chest pain. Further study is needed to define the prevalence of these "normal" variations, and to determine if standard electrocardiographic criteria for thrombolysis apply to young patients.

A Computerized Alert Screening for Severe Sepsis in Emergency Department Patients Increases Lactate Testing but does not Improve Inpatient Mortality.

OBJECTIVE: This study tested the hypothesis that lactate testing in ED sepsis patients could be increased using a computer alert that automatically recognizes systemic inflammatory response syndrome (SIRS) criteria and recommends lactate testing in cases of sepsis defined as ≥2 SIRS criteria plus physician suspicion of infection. Secondary outcomes included the effect of the alert on lactate testing among admitted sepsis patients, the proportion of admitted patients with lactate ≥4.0 mmol/L identified and the in-patient mortality difference before and after alert implementation. METHODS: After a 6 month pre-alert phase, a computer alert was implemented that computed and displayed abnormal vital signs and white blood cell counts for all patients with >2 SIRS criteria and recommended testing lactate if an infection was suspected. Data for admitted patients was collected electronically on consecutive patients meeting sepsis criteria for 6 months before and 6 months after implementation of the alert. RESULTS: There were a total of 5,796 subjects enrolled. Among all septic patients, lactate testing increased from 5.2% in the pre-alert phase to 12.7% in the alert phase, a 7.5% (95% CI 6.0 to 9.0%) absolute increase in lactate testing, p<0.001. Among the 1,798 admitted patients with sepsis, lactate testing increased from 15.3% to 34.2%, an 18.9% (95% CI 15.0 to 22.8%) absolute increase, p<0.001. Among admitted patients with sepsis, there was a 1.9% (95% CI 0.03 to 3.8%, p = 0.05) increase in absolute number of patients with elevated lactate levels identified and a 0.5% (95% CI -1.6 to 2.6%, p=0.64) decrease in mortality. CONCLUSION: The proportion of ED patients who had lactate tested and the number of admitted patients identified with a lactate level ≥4.0 mmol/L improved significantly after the implementation of a computer alert identifying sepsis patients with >2 SIRS criteria while mortality among admitted sepsis patients remained unchanged.

Outcome of out-of-hospital cardiac arrest in New York City. The Pre-Hospital Arrest Survival Evaluation (PHASE) Study.

OBJECTIVE: To determine survival from out-of-hospital cardiac arrest in New York City and to compare this with other urban, suburban, and rural areas. DESIGN: Observational cohort study. SETTING: New York City. PARTICIPANTS: Consecutive out-of-hospital cardiac arrests occurring between October 1, 1990, and April 1, 1991. INTERVENTION: Trained paramedics performed immediate postarrest interviews with care providers, using a standardized questionnaire. MAIN OUTCOME MEASURES: Entry criteria, elapsed time intervals, and nodal events conformed to Utstein recommendations. The single target end point was death or discharge home. RESULTS: Of 3243 consecutive cardiac arrests on which resuscitation was attempted, 2329 (72%) met entry criteria as primary cardiac events. Overall survival was 1.4% (99% confidence interval [CI], 0.9% to 2.3%). No patients were lost to follow-up. Survival from witnessed ventricular fibrillation was 5.3% (99% CI, 2.9% to 8.8%). Using survival from witnessed ventricular fibrillation for intersystem comparison, our survival rate was similar to that of Chicago, Ill (4.0%; 99% CI, 1.9% to 7.5%; P = .41), the only other large city on which data were available. However, it was significantly lower than that reported from midsized urban/suburban areas (33.0%; 99% CI, 30.4% to 35.6%; P < .0001) and suburban/rural areas (12.6%; 99% CI, 8.9% to 16.3%; P < .0001). Survival rate among arrests occurring after arrival of emergency medical services personnel (8.5%; 99% CI, 4.7% to 14.0%) was comparable with Chicago (6.6%; 99% CI, 3.3% to 11.5%; P = .41) but markedly lower than King County, Washington (36%; 99% CI, 28.6% to 43.8%; P < .0001). CONCLUSIONS: Survival from out-of-hospital cardiac arrest in New York City was poor. This was partly attributable to lengthy elapsed time intervals at every step in the chain of survival. However, examination of survival among arrests occurring after emergency medical services arrival suggests that other features may predispose residents of large cities to higher cardiac arrest mortality than individuals living in more suburban or rural settings. Since half the US population resides in large metropolitan areas, this represents a public health problem of considerable magnitude.

Clinical use of the erythrocyte sedimentation rate in the evaluation of febrile intravenous drug users.

STUDY OBJECTIVE: To investigate the use of the erythrocyte sedimentation rate in the evaluation of febrile IV drug users. DESIGN: Prospective observational cohort study. SETTING: Municipal hospital emergency department. TYPE OF PARTICIPANTS: One hundred six IV drug users aged 18 years or older, with rectal temperatures of 37.8 C or more. INTERVENTIONS: Clinical and laboratory variables were obtained by trained research assistants, using a standardized data collection instrument. RESULTS: The erythrocyte sedimentation rate was the only variable consistently associated with illness severity in both the univariate and multivariable analyses (P < .0001). At an erythrocyte sedimentation rate of 100 mm/hr or more, the test had a specificity of 96% (95% confidence interval, 81% to 100%). In contrast, the erythrocyte sedimentation rate displayed a relatively poor sensitivity at low values (88% [95% confidence interval, 77% to 95%] at an erythrocyte sedimentation rate less than 20). CONCLUSION: An erythrocyte sedimentation rate of 100 or more should be regarded as a marker for serious illness in IV drug users with fever, but a "normal" erythrocyte sedimentation rate of less than 20 does not reliably exclude the presence of serious disease in this patient population.

Identification of serious illness in febrile adults.

To determine whether age and other readily obtainable clinical and laboratory variables could be used to predict illness severity in febrile adults, data were collected on 39 patients presenting to an emergency department (ED) with rectal temperature > or = 37.8 degrees C (100 degrees F). Serious illness was defined as (1) need for emergency surgery; (2) intubation; (3) hypotension requiring treatment; (4) bacteremia requiring antibiotics; or (5) death. Six variables were associated with serious illness in the univariate analysis. In a stepwise logistic regression model, only age (P < .0001) and leukocyte count (P < .002) were independently associated with serious illness. Optimal partitioning of these two variables showed that febrile adults younger than 50 years of age with leukocyte counts of less than 15 E9/L have a 5% incidence of serious illness (95% confidence interval [CI], 3% to 8%). In contrast, those who are > or = 50 years of age with leukocyte counts > or = 15 E9/L have a 36% incidence of serious illness (95% CI, 22% to 52%). Patients in this latter category should be carefully examined and considered for hospitalization before concluding that they may be safely discharged from the ED.

Cost effectiveness of an accurate and rapid assay for serum human chorionic gonadotropin in suspected ectopic pregnancy.

The cost effectiveness and clinical utility of a simple, rapid, and accurate pregnancy test in the evaluation of suspected ectopic gestation were studied. The introduction of this qualitative serum assay for human chorionic gonadotropin into our outpatient department during a 1-month period was associated with a significant decrease in culdocenteses (p less than 0.001), ultrasound examinations (p less than 0.025), and hospital admissions (p less than 0.01), with a net projected institutional reduction in health care costs of +123,000 annually.

Methodology for data collection to study prehospital cardiac arrest in New York City: the PHASE methodology. PreHospital Arrest Survival Evaluation Group.

STUDY OBJECTIVE: To describe an effective methodology for the investigation of prehospital cardiac arrest in large cities. DESIGN: Observational cohort study. SETTING: New York City emergency medical services system. PARTICIPANTS: All cardiac arrests dispatched by the 911 system between October 1, 1990, and March 31, 1991. INTERVENTIONS: Trained paramedics performed immediate postarrest interviews with prehospital and hospital care providers using a standardized data collection instrument. RESULTS: Of 3,239 consecutive, confirmed cardiac arrests in which resuscitation was attempted, 2,329 (72%) were of cardiac etiology. Information was sought for 15 of the 17 core events and times recommended by the Utstein Consensus Conference Data were obtained in more than 98% of cases for all except one of these core events and times. One core time yielded data in 96% of cases. All patients were followed until death or discharge home. None were lost to follow-up. CONCLUSION: Concurrent, interactive acquisition of prehospital cardiac arrest data in a large urban setting captured over 98% of the core data recommended for completion of the Utstein template. This methodology may be a suitable means of investigating prehospital cardiac arrest in large cities.

Cardiac arrest witnessed by prehospital personnel: intersystem variation in initial rhythm as a basis for a proposed extension of the Utstein recommendations.

STUDY OBJECTIVE: To test the hypothesis that intersystem variation in initial rhythm among EMS-witnessed arrests is of sufficient magnitude to warrant standardization of survival by creation of an Utstein-style denominator of EMS-witnessed ventricular fibrillation (VF). METHODS: We conducted a planned subset analysis of a prospective observational cohort study of consecutive EMS-witnessed adult cardiac arrests occurring in New York City and meeting Utstein entry criteria. The primary outcome measure was intersystem variation in frequency of EMS-witnessed VF in New York City compared with that in other EMS systems. Secondary outcome measures were variations in survival after EMS-witnessed VF arrests and overall survival after all EMS-witnessed arrests. RESULTS: Intersystem variation showed a threefold difference in the frequency of EMS-witnessed VF (24% in New York City versus 77% in Scotland; 99% confidence interval [CI] for 53% difference, 43% to 63%; P < 10(-7), a twofold difference in survival after EMS-witnessed VF (25% in NYC versus 48% in King County, WA; 99% CI for 23% difference, 6% to 39%; P < .002), and a fourfold difference in survival after all EMS-witnessed arrests (9% in New York City versus 35% in King County; 99% CI for 26% difference, 18% to 34%; P < 10(-7). CONCLUSION: The marked variation in frequency of initial rhythm in EMS-witnessed arrests suggests that a modified Utstein denominator of EMS-witnessed VF would facilitate more uniform intersystem comparison of survival in this unique cohort. However, even after adjustment for initial rhythm, large residual intersystem survival differences remain unexplained.

Effectiveness of bystander cardiopulmonary resuscitation and survival following out-of-hospital cardiac arrest.

OBJECTIVE: To examine the independent relationship between effectiveness of bystander cardiopulmonary resuscitation (CPR) and survival following out-of-hospital cardiac arrest. DESIGN: Prospective observational cohort. SETTING: New York City. PARTICIPANTS: A total of 2071 consecutive out-of-hospital cardiac arrests meeting Utstein criteria. INTERVENTION: Trained prehospital personnel assessed the quality of bystander CPR on arrival at the scene. Satisfactory execution of CPR required performance of both adequate compressions and ventilations in conformity with current American Heart Association guidelines. MAIN OUTCOME MEASURE: Adjusted association between CPR effectiveness and survival. Survival was defined as discharge from hospital to home. RESULTS: Outcome was determined on all members of the inception cohort--none were lost to follow-up. When the association between bystander CPR and survival was adjusted for effectiveness of CPR in the parent data set (N = 2071), only effective CPR was retained in the logistic model (adjusted odds ratio [OR] = 5.7; 95% confidence interval [CI], 2.7 to 12.2; P < .001). Of the subset of 662 individuals (32%) who received bystander CPR, 305 (46%) had it performed effectively. Of these, 4.6% (14/305) survived vs 1.4% (5/357) of those with ineffective CPR (OR = 3.4; 95% CI, 1.1 to 12.1; P < .02). After adjustment for witness status, initial rhythm, interval from collapse to CPR, and interval from collapse to advanced life support, effective CPR remained independently associated with improved survival (adjusted OR = 3.9; 95% CI, 1.1 to 14.0; P < .04). CONCLUSION: The association between bystander CPR and survival in out-of-hospital cardiac arrest appears to be confounded by CPR quality. Effective CPR is independently associated with a quantitatively and statistically significant improvement in survival.