Predicting outcome after emergent cerclage using classification tree analysis.

We sought to develop a predictive model for gestational age at delivery after placement of an emergent cerclage in the second trimester. Data were obtained for women undergoing emergent cerclage in response to documented cervical change on physical examination at a university hospital between 1980 and 2000. Hierarchically optimal classification tree analysis (CTA) was used to predict delivery prior to 24 weeks, between 24 and 27 6/7 weeks, or after 27 6/7 weeks. One hundred sixteen women were available for analysis. Delivery prior to 24 weeks was best predicted by presence of prolapsed membranes and gestational age at cerclage placement; delivery between 24 and 27 6/7 weeks was best predicted by parity alone; delivery of at least 28 weeks was best predicted by cervical dilation and length, presence of prolapsed membranes, and parity. When choosing a single model to predict delivery at the three different gestational age periods, the last model yielded the most accurate results. CTA can be used to construct a predictive model for outcome after emergent cerclage that may be informative for both patients and physicians.

Derivation of a triage algorithm for chest radiography of community-acquired pneumonia patients in the emergency department.

BACKGROUND: Community-acquired pneumonia (CAP) accounts for 1.5 million emergency department (ED) patient visits in the United States each year. OBJECTIVES: To derive an algorithm for the ED triage setting that facilitates rapid and accurate ordering of chest radiography (CXR) for CAP. METHODS: The authors conducted an ED-based retrospective matched case-control study using 100 radiographic confirmed CAP cases and 100 radiographic confirmed influenzalike illness (ILI) controls. Sensitivities and specificities of characteristics assessed in the triage setting were measured to discriminate CAP from ILI. The authors then used classification tree analysis to derive an algorithm that maximizes sensitivity and specificity for detecting patients with CAP in the ED triage setting. RESULTS: Temperature greater than 100.4 degrees F (likelihood ratio = 4.39, 95% confidence interval [CI] = 2.04 to 9.45), heart rate greater than 110 beats/minute (likelihood ratio = 3.59, 95% CI = 1.82 to 7.10), and pulse oximetry less than 96% (likelihood ratio = 2.36, 95% CI = 1.32 to 4.20) were the strongest predictors of CAP. However, no single characteristic was adequately sensitive and specific to accurately discriminate CAP from ILI. A three-step algorithm (using optimum cut points for elevated temperature, tachycardia, and hypoxemia on room air pulse oximetry) was derived that is 70.8% sensitive (95% CI = 60.7% to 79.7%) and 79.1% specific (95% CI = 69.3% to 86.9%). CONCLUSIONS: No single characteristic adequately discriminates CAP from ILI, but a derived clinical algorithm may detect most radiographic confirmed CAP patients in the triage setting. Prospective assessment of this algorithm will be needed to determine its effects on the care of ED patients with suspected pneumonia.

Development and validation of a short-form, rapid estimate of adult literacy in medicine.

BACKGROUND: Although prior studies used the 66-item Rapid Estimate of Adult Literacy in Medicine (REALM instrument) for literacy assessment, researchers may require a shorter, validated instrument when designing interventions for clinical contexts. OBJECTIVE: To develop and validate a very brief literacy assessment tool, the REALM-Short Form (REALM-SF). PATIENTS: The model development, validation, and field testing validation samples included 1336, 164, and 50 patients, respectively. SETTING: General medicine and subspecialty clinics and medicine inpatient wards. DESIGN: For development and validation samples, indicator variables for REALM instrument items were evaluated as potential predictors of REALM instrument score by stepwise multiple regression analysis with subsequent bootstrap and confirmatory factor analysis of selected items. Pearson correlations compared REALM-SF and REALM instrument scores and kappa analyses compared grade level assignments. For the field testing validation sample, Pearson correlations compared Wide Range Achievement Test and REALM-SF scores. RESULTS: The REALM-SF included 7 items with stable model coefficients and 1 underlying linear factor. REALM-SF and REALM instrument scores were highly correlated in development (r = 0.95, P < 0.001) and validation (r = 0.94, P < 0.001) samples. There was excellent agreement between REALM-SF and REALM instrument grade-level assignments when dichotomized at the 6th grade (development: 97% agreement, K = 0.88, P < 0.001; validation: 88% agreement, K = 0.75, P < 0.001) and 8th grade levels (development: 94% agreement, K = 0.78, P < 0.001; validation: 84% agreement, K = 0.67, P < 0.001). REALM-SF and Wide Range Achievement Test scores were highly correlated (r = 0.83, P < 0.001) in field testing validation. CONCLUSIONS: The REALM-SF provides researchers a brief, validated instrument for assessing patient literacy in diverse research settings.

Developing indicators of inpatient adverse drug events through nonlinear analysis using administrative data.

BACKGROUND: Because of uniform availability, hospital administrative data are appealing for surveillance of adverse drug events (ADEs). Expert-generated surveillance rules that rely on the presence of International Classification of Diseases, 9th Revision Clinical Modification (ICD-9-CM) codes have limited accuracy. Rules based on nonlinear associations among all types of available administrative data may be more accurate. OBJECTIVES: By applying hierarchically optimal classification tree analysis (HOCTA) to administrative data, derive and validate surveillance rules for bleeding/anticoagulation problems and delirium/psychosis. RESEARCH DESIGN: Retrospective cohort design. SUBJECTS: A random sample of 3987 admissions drawn from all 41 Utah acute-care hospitals in 2001 and 2003. MEASURES: Professional nurse reviewers identified ADEs using implicit chart review. Pharmacists assigned Medical Dictionary for Regulatory Activities codes to ADE descriptions for identification of clinical groups of events. Hospitals provided patient demographic, admission, and ICD9-CM data. RESULTS: Incidence proportions were 0.8% for drug-induced bleeding/anticoagulation problems and 1.0% for drug-induced delirium/psychosis. The model for bleeding had very good discrimination and sensitivity at 0.87 and 86% and fair positive predictive value (PPV) at 12%. The model for delirium had excellent sensitivity at 94%, good discrimination at 0.83, but low PPV at 3%. Poisoning and adverse event codes designed for the targeted ADEs had low sensitivities and, when forced in, degraded model accuracy. CONCLUSIONS: Hierarchically optimal classification tree analysis is a promising method for rapidly developing clinically meaningful surveillance rules for administrative data. The resultant model for drug-induced bleeding and anticoagulation problems may be useful for retrospective ADE screening and rate estimation.

Discriminating inhalational anthrax from community-acquired pneumonia using chest radiograph findings and a clinical algorithm.

BACKGROUND: Limiting the effects of a large-scale bioterrorist anthrax attack will require rapid and accurate detection of the earliest victims. We undertook this study to improve physicians' ability to rapidly detect inhalational anthrax victims. METHODS: We conducted a case-control study to compare chest radiograph findings from 47 patients from historical inhalational anthrax cases and 188 community-acquired pneumonia control subjects. We then used classification tree analyses to derive an algorithm of chest radiograph findings and clinical characteristics that accurately and explicitly discriminated between inhalational anthrax and community-acquired pneumonia. RESULTS: Twenty-two of the 47 patients from historical inhalational anthrax cases (46.8%) had reported chest radiograph findings. All 22 case patients (100%) had mediastinal widening, pleural effusion, or both. However, 16 case patients (72.7%) also had infiltrates. In comparison, all 188 community-acquired control subjects had reported chest radiographs. Of these, 127 control subjects (67.6%) had infiltrates, 43 control subjects (22.9%) had pleural effusions, and 15 control subjects (8.0%) had mediastinal widening. A derived algorithm with three predictor variables (chest radiograph finding of mediastinal widening, altered mental status, and elevated hematocrit) is 100% sensitive (95% confidence interval [CI], 73.5 to 100) and 98.3% specific (95% CI, 95.1 to 99.6). The derivation process used 12 patients with inhalational anthrax and 177 control subjects with community-acquired pneumonia who had information available for all three variables. CONCLUSIONS: There are significant chest radiograph differences between inhalational anthrax and community-acquired pneumonia, but none of the chest radiograph findings are both highly sensitive and highly specific. The derived clinical algorithm can improve physicians' ability to discriminate inhalational anthrax from community-acquired pneumonia, but its utility is limited to previously healthy individuals and its accuracy may be limited by missing values.

Evolving risk for thromboembolism in spinal cord injury (SPIRATE Study).

OBJECTIVE: To compare the frequency of venous thromboembolism (VTE) in patients examined during 1992-1995 with those examined during 1999-2003. DESIGN: A comprehensive review of the charts of all patients admitted between late 1999 and early 2003 for rehabilitation after spinal cord injury. Only records, including evidence for objective testing for VTE (ultrasound, venography, lung scanning) were included, and patients having inferior vena cava filter placement or previous VTE were excluded. Analysis variables included type and location of spinal cord injury, American Spinal Injury Association classification, concomitant injuries, surgical procedures, complications, preexisting illnesses, and use of antithrombotic prophylaxis. Using univariate optimal discriminant analysis, data from the current group of patients were compared with a previous study of 243 subjects examined between 1992 and 1995. RESULTS: The current sample consisted of 76 persons with acute spinal cord injury, of whom six had VTE (7.9%). As compared with the frequency of VTE in the previous patient sample (21%), this represented a significant decrease (P<0.01). The major differences between the current and previous patient samples were a decrease in the use of unfractionated heparin (15.8% vs. 56.8%, P<0.0001) and an increase in the use of low molecular weight heparin (81.6% vs. 59.7%, P<0.0001). CONCLUSIONS: VTE has been a common and occasionally lethal complication in persons with spinal cord injury. The recent switch from unfractionated heparin to low molecular weight heparin for the prevention of VTE has coincided with a decrease in the frequency of this complication in patients with spinal cord injury.

Assessment of resident professionalism using high-fidelity simulation of ethical dilemmas.

OBJECTIVES: To examine the responses of emergency medicine residents (EMRs) to ethical dilemmas in high-fidelity patient simulations as a means of assessing resident professionalism. METHODS: This cross-sectional observational study included all EMRs at a three-year training program. Subjects were excluded if they were unable or unwilling to participate. Each resident subject participated in a simulated critical patient encounter during an Emergency Medicine Crisis Resource Management course. An ethical dilemma was introduced before the end of each simulated encounter. Resident responses to that dilemma were compared with a professional performance checklist evaluation. Multi-response permutation procedure analysis was used to compare performance measures between resident classes, with the a priori hypothesis that mean performance should increase as experience increases. RESULTS: Of the 30 potential subjects, 90% (27) participated. The remaining three residents were unavailable due to scheduling conflicts. It was observed that senior residents (second and third year) performed more checklist items than did first-year residents (p < 0.028 for each senior class). Omnibus comparison of mean critical actions completed across all three years was not statistically significant (p < 0.13). Residents performed a critical action with 100% uniformity across training years in only one ethical scenario ("Practicing Procedures on the Recently Dead"). Residents performed the fewest critical actions and overall checklist items for the "Patient Confidentiality" case. CONCLUSIONS: Senior residents had better overall performance than incoming interns, suggesting that professional behaviors are learned through some facet of residency training. Although limited by small sample size, the application of this performance-assessment tool showed the ability to discriminate between experienced and inexperienced EMRs with respect to a variety of aspects of professional competency. These findings suggest a need for improved resident education in areas of professionalism and ethics.

Spinal Cord Injury Risk Assessment for Thromboembolism (SPIRATE Study).

OBJECTIVE: To identify risk factors for venous thromboembolism (VTE) during the rehabilitation phase of spinal cord injury. DESIGN: A comprehensive review of the charts of all patients admitted between 1992 and 1995 for rehabilitation after spinal cord injury. Only records including evidence for objective testing for thromboembolism were included. The analysis variables included type and location of spinal cord injury, American Spinal Injury Association classification, concomitant injuries, surgical procedures, complications, preexisting illnesses, and use of antithrombotic prophylaxis. Hierarchically optimal classification tree analysis was employed to develop a nonlinear model for predicting deep vein thrombosis. RESULTS: The analysis sample consisted of these 243 persons with acute spinal cord injury, 51 of whom had venous thromboembolism, and eight of whom died. A three-variable model emerged that identified patient groups differing in relative likelihood of experiencing deep vein thrombosis. The highest likelihood group were patients with cancer over the age of 35 yrs. Women between the ages of 36 and 58 yrs without cancer were also at increased risk, as were cancer-free men with flaccid paralysis. CONCLUSIONS: Venous thromboembolism is more likely to develop in spinal cord injury patients who are older, obese, and have flaccid paralysis or cancer. These patients should receive vigorous prophylaxis against venous thromboembolism.