Likelihood ratio: A powerful tool for incorporating the results of a diagnostic test into clinical decisionmaking

Validation of rules to predict emergent surgical intervention in pediatric trauma patients

BACKGROUND

Trauma centers use guidelines to determine when a trauma surgeon is needed in the emergency department (ED) on patient arrival. A decision rule from Loma Linda University identified patients with penetrating injury and tachycardia as requiring emergent surgical intervention. Our goal was to validate this rule and to compare it with the American College of Surgeons' Major Resuscitation Criteria (MRC).

STUDY DESIGN

We used data from 1993 through 2010 from 2 level 1 trauma centers in Denver, CO. Patient demographics, injury severity, times of ED arrival and surgical intervention, and all variables of the Loma Linda Rule and the MRC were obtained. The outcome, emergent intervention (defined as requiring operative intervention by a trauma surgeon within 1 hour of arrival to the ED or performance of cricothyroidotomy or thoracotomy in the ED), was confirmed using standardized abstraction. Sensitivities, specificities, and 95% confidence intervals were calculated.

RESULTS

There were 8,078 patients included, and 47 (0.6%) required emergent intervention. Of the 47 patients, the median age was 11 years (interquartile range [IQR] 7 to 14 years), 70% were male, 30% had penetrating mechanisms, and the median Injury Severity Score (ISS) was 25 (IQR 9 to 41). At the 2 institutions, the Loma Linda Rule had a sensitivity and specificity of 69% (95% CI 45% to 94%) and 76% (95% CI 69% to 83%), respectively, and the MRC had a sensitivity and specificity of 80% (95% CI 70% to 92%) and 81% (95% CI 77% to 85%), respectively.

CONCLUSIONS

Emergent surgical intervention is rare in the pediatric trauma population. Although precision of predictive accuracies of the Loma Linda Rule and MRC were limited by small numbers of outcomes, neither set of criteria appears to be sufficiently accurate to recommend their routine use.

The two-step Fagan's nomogram: ad hoc interpretation of a diagnostic test result without calculation

In 1975, Fagan published a nomogram to help practitioners determine, without the use of a calculator or computer, the probability of a patient truly having a condition of interest given a particular test result. Nomograms are very useful for bedside interpretations of test results, as no test is perfect. However, the practicality of Fagan's nomogram is limited by its use of the likelihood ratio (LR), a parameter not commonly reported in the evaluation studies of diagnostic tests. The LR reflects the direction and strength of evidence provided by a test result and can be computed from the conventional diagnostic sensitivity (DSe) and specificity (DSp) of the test. This initial computation is absent in Fagan's nomogram, making it impractical for routine use. We have seamlessly integrated the initial step to compute the LR and the resulting two-step nomogram allows the user to quickly interpret the outcome of a test. With the addition of the DSe and DSp, the nomogram, for the purposes of interpreting a dichotomous test result, is now complete. This tool is more accessible and flexible than the original, which will facilitate its use in routine evidence-based practice. The nomogram can be downloaded at: www.adelaide.edu.au/vetsci/research/pub_pop/2step-nomogram/.

Clinical test for diagnosis of patellofemoral pain syndrome: Systematic review with meta-analysis

The high incidence and diversity of factors attributed to the etiology of patellofemoral pain syndrome (PFPS) makes the diagnosis of this problem somewhat complex and susceptible to misinterpretation. Currently, there is not a defined set of procedures considered as ideal to diagnose PFPS. To investigate the diagnostic accuracy of clinical and functional tests used to diagnose PFPS through a systematic review. We searched relevant studies in the databases Medline, CINAHL, SPORTDiscus and Embase. The QUADAS score was used to assess the methodological quality of the eligible studies. We analyzed data that indicated the diagnostic properties of tests, such as sensibility, specificity, positive (LR+) and negative (LR-) likelihood ratio, and predictive values. The search identified 16,169 potential studies and five studies met the eligibility criteria. The 5 studies analyzed 25 tests intending to accurately diagnose PFPS. Two tests were analyzed in two studies and were possible to perform a meta-analysis. Within the five studies included, one study had high methodological quality, two studies had good methodological quality and two studies had low methodological quality. Two tests, the patellar tilt (LR+ = 5.4 and LR- = 0.6) and squatting (LR+ = 1.8 and LR- = 0.2), had values that show a trend for the diagnosis of PFPS (LR+ >5.0 and LR- <0.2), however their values do not represent clear evidence regarding diagnostic properties as suggested in the literature (LR+ >10 and LR- <0.1). Future diagnostic studies should focus on the sample homogeneity and standardization of tests analyzed so future systematic reviews can determine with more certainty the accuracy of the tests for diagnosis of PFPS.

Agreement of self-reported items and clinically assessed nerve root involvement (or sciatica) in a primary care setting

INTRODUCTION

We analysed baseline measures from an RCT involving adults with low back pain (LBP) with or without referred leg pain, to identify self-report items that best identified clinically determined nerve root involvement (sciatica).

METHODS

Potential indicators of nerve root involvement were gathered using a self-reported questionnaire. Participants underwent a standardised physical examination on the same day as questionnaire completion. Self-reported items were compared to a reference standard (clinical diagnosis) using sensitivity, specificity, predictive values, likelihood ratios (LRs), the area under the receiver operating characteristic curve and logistic regression. Two reference standards are presented: one based on a clinical diagnosis of nerve root problems and excluding possible/inconclusive cases (referred to as a confirmatory reference), and the other being inclusive of possible/inconclusive cases (referred to as an indicative reference).

RESULTS

Pain below knee was the best single item for diagnostic accuracy with an area under curve (AUC) of 0.67-0.68, which however is slightly less than the 'acceptable discrimination'. A cluster of three items, including distribution of pain below the knee, leg pain that is worse than back pain, and feeling of numbness or pins and needles in the leg, did improve discrimination to an 'acceptable' level with an AUC of 0.72-0.74 in relation to confirmatory and indicative references, respectively. However, the likelihood ratios from the models were reflective of a 'small' amount of discrimination.

CONCLUSION

In this primary care population seeking treatment for LBP with or without leg pain, we found no clear set of self-report items that accurately identified patients with nerve root pain. When accurate case definition is important, clinical assessment should be the method of choice for identifying LBP with possible nerve root involvement.

Validation of a clinical prediction rule to distinguish Lyme meningitis from aseptic meningitis

OBJECTIVES

The "Rule of 7's," a Lyme meningitis clinical prediction rule, classifies children at low risk for Lyme meningitis when each of the following 3 criteria are met: <7 days of headache, <70% cerebrospinal fluid (CSF) mononuclear cells, and absence of seventh or other cranial nerve palsy. The goal of this study was to test the performance of the Rule of 7's in a multicenter cohort of children with CSF pleocytosis.

METHODS

We performed a retrospective cohort study of children evaluated at 1 of 3 emergency departments located in Lyme disease-endemic areas with CSF pleocytosis and Lyme serology obtained. Lyme meningitis was defined using the Centers for Disease Control and Prevention criteria (either positive Lyme serology test result or an erythema migrans [EM] rash). We calculated the performance of the Rule of 7's in our overall study population and in children without physician-documented EM.

RESULTS

We identified 423 children, of whom 117 (28% [95% confidence interval (CI): 24%-32%]) had Lyme meningitis, 306 (72% [95% CI: 68%-76%]) had aseptic meningitis, and 0 (95% CI: 0%-1%) had bacterial meningitis. Of the 130 classified as low risk, 5 had Lyme meningitis (sensitivity, 112 of 117 [96% (95% CI: 90%-99%)]; specificity, 125 of 302 [41% (95% CI: 36%-47%)]). In the 390 children without EM, 3 of the 127 low-risk patients had Lyme meningitis (2% [95% CI: 0%-7%]).

CONCLUSIONS

Patients classified as low risk by using the Rule of 7's were unlikely to have Lyme meningitis and could be managed as outpatients while awaiting results of Lyme serology tests.

Signs and symptoms predicting influenza in children: a matched case-control analysis of prospectively collected clinical data

We aimed to determine whether there are signs or symptoms that could help clinicians to distinguish between influenza and other respiratory infections. The clinical data for this matched case-control analysis were derived from a 2-year prospective cohort study of respiratory infections among children aged≤13 years. At any signs of respiratory infection, the children were examined and nasal swabs were obtained for virologic analyses. Cases were 353 children with laboratory-confirmed influenza and controls were 353 children with respiratory symptoms who tested negative for influenza. Cases and controls were matched for gender, age, and timing of the visit. In the multivariate conditional logistic regression analyses, fever was the only sign that independently predicted influenza virus infection, with odds ratios ranging from 13.55 (95% confidence interval [CI], 6.90-26.63) to 50.10 (95% CI, 16.25-154.45), depending on the degree of fever. In all analyses, the predictive capability of fever increased with incremental elevations in the child's temperature. The likelihood ratio of fever≥40.0°C in predicting influenza was 6.00 (95% CI, 2.80-12.96). Among unselected children seen as outpatients during influenza outbreaks, fever is the only reliable predictor of influenza virus infection. The optimal use of influenza-specific antiviral drugs in children may require virologic confirmation.

Radiographic heart size and its rate of increase as tests for onset of congestive heart failure in Cavalier King Charles Spaniels with mitral valve regurgitation

BACKGROUND

In canine mitral regurgitation (MR) the rate of heart enlargement increases in the last year before congestive heart failure (CHF). Measurement of heart size and its rate of increase may be useful tests for CHF in MR.

OBJECTIVES

To determine the value of vertebral heart scale (VHS) and its rate of increase (∆VHS units/month) for diagnosing the presence and predicting the onset of CHF.

ANIMALS

Longitudinal study of 94 Cavalier King Charles Spaniels (CKCS).

METHODS

VHS was measured at intervals before CHF. ∆VHS/month was calculated from sequential pairs of VHS measurements and the interval between them. Diagnostic accuracy and utility were determined by the areas under receiver operating characteristic plots (AUROC), and likelihood ratios (LR).

RESULTS

AUROC for VHS at the onset of CHF was 0.93 (95% CI, 0.96-0.90), to predict CHF 1-12 months before CHF was 0.74 (95% CI, 0.81-0.66), and for ∆VHS/month at CHF was 0.98 (95% CI, 0.99-0.96). Interval LRs and their cutoff values for CHF were for VHS: 13 (95% CI, 20-7.3) at ≥12.7; 1.2 (95% CI, 2.0-0.68) between 12.7 and 12.0; 0.04 (95% CI, 0.18-0.01) at ≤12.0, and for ∆VHS/month: 15 (95% CI, 30-7.7) at ≥0.08; 0.72 (95% CI, 2.0-0.25) between 0.08 and 0.06; and 0.05 (95% CI, 0.13-0.02) at ≤0.06.

CONCLUSIONS AND CLINICAL IMPORTANCE

Under the conditions of this study, VHS and particularly ∆VHS/month are useful measurements for detecting onset of CHF in CKCS with MR.

Evidence-based diagnostics: adult septic arthritis

BACKGROUND

Acutely swollen or painful joints are common complaints in the emergency department (ED). Septic arthritis in adults is a challenging diagnosis, but prompt differentiation of a bacterial etiology is crucial to minimize morbidity and mortality.

OBJECTIVES

The objective was to perform a systematic review describing the diagnostic characteristics of history, physical examination, and bedside laboratory tests for nongonococcal septic arthritis. A secondary objective was to quantify test and treatment thresholds using derived estimates of sensitivity and specificity, as well as best-evidence diagnostic and treatment risks and anticipated benefits from appropriate therapy.

METHODS

Two electronic search engines (PUBMED and EMBASE) were used in conjunction with a selected bibliography and scientific abstract hand search. Inclusion criteria included adult trials of patients presenting with monoarticular complaints if they reported sufficient detail to reconstruct partial or complete 2 × 2 contingency tables for experimental diagnostic test characteristics using an acceptable criterion standard. Evidence was rated by two investigators using the Quality Assessment Tool for Diagnostic Accuracy Studies (QUADAS). When more than one similarly designed trial existed for a diagnostic test, meta-analysis was conducted using a random effects model. Interval likelihood ratios (LRs) were computed when possible. To illustrate one method to quantify theoretical points in the probability of disease whereby clinicians might cease testing altogether and either withhold treatment (test threshold) or initiate definitive therapy in lieu of further diagnostics (treatment threshold), an interactive spreadsheet was designed and sample calculations were provided based on research estimates of diagnostic accuracy, diagnostic risk, and therapeutic risk/benefits.

RESULTS

The prevalence of nongonococcal septic arthritis in ED patients with a single acutely painful joint is approximately 27% (95% confidence interval [CI] = 17% to 38%). With the exception of joint surgery (positive likelihood ratio [+LR] = 6.9) or skin infection overlying a prosthetic joint (+LR = 15.0), history, physical examination, and serum tests do not significantly alter posttest probability. Serum inflammatory markers such as white blood cell (WBC) counts, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) are not useful acutely. The interval LR for synovial white blood cell (sWBC) counts of 0 × 10(9)-25 × 10(9)/L was 0.33; for 25 × 10(9)-50 × 10(9)/L, 1.06; for 50 × 10(9)-100 × 10(9)/L, 3.59; and exceeding 100 × 10(9)/L, infinity. Synovial lactate may be useful to rule in or rule out the diagnosis of septic arthritis with a +LR ranging from 2.4 to infinity, and negative likelihood ratio (-LR) ranging from 0 to 0.46. Rapid polymerase chain reaction (PCR) of synovial fluid may identify the causative organism within 3 hours. Based on 56% sensitivity and 90% specificity for sWBC counts of >50 × 10(9)/L in conjunction with best-evidence estimates for diagnosis-related risk and treatment-related risk/benefit, the arthrocentesis test threshold is 5%, with a treatment threshold of 39%.

CONCLUSIONS

Recent joint surgery or cellulitis overlying a prosthetic hip or knee were the only findings on history or physical examination that significantly alter the probability of nongonococcal septic arthritis. Extreme values of sWBC (>50 × 10(9)/L) can increase, but not decrease, the probability of septic arthritis. Future ED-based diagnostic trials are needed to evaluate the role of clinical gestalt and the efficacy of nontraditional synovial markers such as lactate.

Validation and refinement of a rule to predict emergency intervention in adult trauma patients

STUDY OBJECTIVE

Trauma centers use "secondary triage" to determine the necessity of trauma surgeon involvement. A clinical decision rule, which includes penetrating injury, an initial systolic blood pressure less than 100 mm Hg, or an initial pulse rate greater than 100 beats/min, was developed to predict which trauma patients require emergency operative intervention or emergency procedural intervention (cricothyroidotomy or thoracotomy) in the emergency department. Our goal was to validate this rule in an adult trauma population and to compare it with the American College of Surgeons' major resuscitation criteria.

METHODS

We used Level I trauma center registry data from September 1, 1995, through November 30, 2008. Outcomes were confirmed with blinded abstractors. Sensitivity, specificity, and 95% confidence intervals (CIs) were calculated.

RESULTS

Our patient sample included 20,872 individuals. The median Injury Severity Score was 9 (interquartile range 4 to 16), 15.3% of patients had penetrating injuries, 13.5% had a systolic blood pressure less than 100 mm Hg, and 32.5% had a pulse rate greater than 100 beats/min. Emergency operative intervention or procedural intervention was required in 1,099 patients (5.3%; 95% CI 5.0% to 5.6%). The sensitivities and specificities of the rule and the major resuscitation criteria for predicting emergency operative intervention or emergency procedural intervention were 95.6% (95% CI 94.3% to 96.8%) and 56.1% (95% CI 55.4% to 56.8%) and 85.5% (95% CI 83.3% to 87.5%) and 80.9% (95% CI 80.3% to 81.4%), respectively.

CONCLUSION

This new rule was more sensitive for predicting the need for emergency operative intervention or emergency procedural intervention directly compared with the American College of Surgeons' major resuscitation criteria, which may improve the effectiveness and efficiency of trauma triage.

The Six-Item Screener and AD8 for the detection of cognitive impairment in geriatric emergency department patients

STUDY OBJECTIVE

We evaluate the diagnostic test characteristics of the Six-Item Screener and the AD8 to detect cognitive dysfunction in adults older than 65 years and using the emergency department (ED) for any reason.

METHODS

We conducted an observational cross-sectional cohort study at a single academic urban university-affiliated hospital. Subjects were consenting, non--critically ill, English-speaking adults older than 65 years and receiving care in the ED. We quantitatively assessed the diagnostic test characteristics of the Six-Item Screener and AD8 by using the Mini-Mental State Examination score less than 24 as the criterion standard for cognitive dysfunction.

RESULTS

The prevalence of cognitive dysfunction was 35%, but only 6% of charts noted a pre-existing deficit. The Six-Item Screener was superior to either the caregiver-administered AD8 or the patient-administered AD8 for the detection of cognitive dysfunction.

CONCLUSION

The Six-Item Screener was superior to the caregiver- or patient-administered AD8 to identify older adults at increased risk for occult cognitive dysfunction.

Four sensitive screening tools to detect cognitive dysfunction in geriatric emergency department patients: brief Alzheimer's Screen, Short Blessed Test, Ottawa 3DY, and the caregiver-completed AD8

BACKGROUND

Cognitive dysfunction, including dementia and delirium, is prevalent in geriatric emergency department (ED) patients, but often remains undetected. One barrier to reliable identification of acutely or chronically impaired cognitive function is the lack of an acceptable screening tool. While multiple brief screening instruments have been derived, ED validation trials have not previously demonstrated tools that are appropriately sensitive for clinical use.

OBJECTIVES

The primary objective was to evaluate and compare the Ottawa 3DY (O3DY), Brief Alzheimer's Screen (BAS), Short Blessed Test (SBT), and caregiver-completed AD8 (cAD8) diagnostic test performance for cognitive dysfunction in geriatric ED patients using the Mini Mental Status Exam (MMSE) as the criterion standard. A secondary objective was to assess the diagnostic accuracy for the cAD8 (which is an informant-based instrument) when used in combination with the other performance-based screening tools.

METHODS

In an observational cross-sectional cohort study at one urban academic university-affiliated medical center, trained research assistants (RAs) collected patients' responses on the Confusion Assessment Method for the Intensive Care Unit, BAS, and SBT. When available, reliable caregivers completed the cAD8. The MMSE was then obtained. The O3DY was reconstructed from elements of the MMSE and the BAS. Consenting subjects were non-critically ill, English-speaking adults over age 65 years, who had not received potentially sedating medications prior to or during cognitive testing. Using an MMSE score of ≤23 as the criterion standard for cognitive dysfunction, the sensitivity, specificity, likelihood ratios, and receiver operating characteristic (ROC) area under the curve (AUC) were computed. Venn diagrams were constructed to quantitatively compare the degree of overlap among positive test results between the performance-based instruments.

RESULTS

The prevalence of cognitive dysfunction for the 163 patients enrolled with complete data collection was 37%, including 5.5% with delirium. Dementia was self-reported in 3%. Caregivers were available to complete the cAD8 for 56% of patients. The SBT, BAS, and O3DY each demonstrated 95% sensitivity, compared with 83% sensitivity for the cAD8. The SBT had a superior specificity of 65%. No combination of instruments with the cAD8 significantly improved diagnostic accuracy. The SBT provided the optimal overlap with the MMSE.

CONCLUSIONS

The SBT, BAS, and O3DY are three brief performance-based screening instruments to identify geriatric patients with cognitive dysfunction more rapidly than the MMSE. Among these three instruments, the SBT provides the best diagnostic test characteristics and overlap with MMSE results. The addition of the cAD8 to the other instruments does not enhance diagnostic accuracy.

Non-invasive classification of severe sepsis and systemic inflammatory response syndrome using a nonlinear support vector machine: a preliminary study

Sepsis has been defined as the systemic response to infection in critically ill patients, with severe sepsis and septic shock representing increasingly severe stages of the same disease. Based on the non-invasive cardiovascular spectrum analysis, this paper presents a pilot study on the potential use of the nonlinear support vector machine (SVM) in the classification of the sepsis continuum into severe sepsis and systemic inflammatory response syndrome (SIRS) groups. 28 consecutive eligible patients attending the emergency department with presumptive diagnoses of sepsis syndrome have participated in this study. Through principal component analysis (PCA), the first three principal components were used to construct the SVM feature space. The SVM classifier with a fourth-order polynomial kernel was found to have a better overall performance compared with the other SVM classifiers, showing the following classification results: sensitivity = 94.44%, specificity = 62.50%, positive predictive value = 85.00%, negative predictive value = 83.33% and accuracy = 84.62%. Our classification results suggested that the combinatory use of cardiovascular spectrum analysis and the proposed SVM classification of autonomic neural activity is a potentially useful clinical tool to classify the sepsis continuum into two distinct pathological groups of varying sepsis severity.

Application of the Laboratory Risk Indicator in Necrotising Fasciitis (LRINEC) Score to Patients in a Tropical Tertiary Referral Centre

The objective of this study was to assess the applicability and usefulness of the Laboratory Risk Indicator in Necrotising Fasciitis (LRINEC) score in patients from our institution.

A retrospective analysis was undertaken of the case notes of all patients admitted to our facility between January 2002 and December 2005 with the admission diagnosis of necrotising fasciitis and the application of the LRINEC score upon the initial blood tests. The sensitivity, specificity and likelihood ratios were then calculated for patients with a LRINEC score of ≥6 compared with the findings of a surgical biopsy.

Twenty-eight patients were identified as having the admission diagnosis of necrotising fasciitis on the hospital database and were eligible for the study. Ten of these had biopsy-proven necrotising fasciitis.

With a cut-off score of ≥6, the LRINEC score had a sensitivity of 80%, specificity of 67%, a positive predictive value of 57% and a negative predictive value of 86% in distinguishing the patients with proven necrotising fasciitis from those with severe soft tissue infections. The likelihood ratio of a positive biopsy was 2.4 and the likelihood ratio of a negative biopsy was 0.3.

A cut-off value of ≥6 of the LRINEC score was not overly sensitive or specific, and the likelihood ratios suggest that, at this cut-off level, the LRINEC score would have only a very small effect on the post-test probability of the patients in the studied population having necrotising fasciitis.

Detection of Mycobacterium avium subspecies paratuberculosis: comparing fecal culture versus serum enzyme-linked immunosorbent assay and direct fecal polymerase chain reaction

Mycobacterium avium ssp. paratuberculosis (MAP) is the etiologic agent of Johne's disease in cattle. The disease causes diarrhea, reduced milk production, poor reproductivity, emaciation, and eventually death. Culture on Herrold's egg yolk agar is considered to be the definitive test for diagnosis of Johne's in cattle. This method has moderate sensitivity (30 to 50%) and is 100% specific; however, it can take up to 16 wk due to the slow growth of MAP. Currently, serum ELISA is used to screen herds for Johne's disease, but positive tests must be confirmed culturally or by PCR. The current research sought to evaluate an in-house direct fecal PCR procedure and directly compare it to ELISA using culture as the gold standard. Serum and fecal samples were collected from cows (n = 250) with unknown Johne's status. Fecal samples were processed for culture on Herrold's egg yolk agar and direct PCR. Serum samples were tested using the Parachek serum ELISA. Overall, 67/250 [26.8%, 95% confidence interval (CI) 21.4 to 32.8] animals were culturally confirmed to be shedding MAP. The PCR and ELISA detected 74/250 (29.6%, 95% CI 24 to 35.7) and 25/250 (10%, 95% CI 6.6 to 14.4), respectively. Culture and PCR were able to detect more positive animals than ELISA. Overall, direct fecal PCR was 70.2% sensitive and 85.3% specific when using culture as the gold standard. The ELISA method was 31.3% sensitive and 97.8% specific. When culture reported <10 cfu, the sensitivity and specificity of PCR and ELISA were 57.1 and 85.3%, and 4.8 and 97.8%, respectively. When culture reported 10 to <40 cfu, the sensitivity of PCR and ELISA were 75 and 50%, respectively. When culture reported > or =40 cfu, the sensitivity of PCR and ELISA were 100 and 88.2%, respectively. Specificity could not be calculated at these levels because there were no negative samples. The direct PCR outperformed the ELISA in detecting animals potentially infected with MAP and was not significantly different when compared with culture. The direct fecal PCR method described here provides faster results than traditional culture and is more sensitive than ELISA at detecting animals suspected of Johne's disease. These data support the use of PCR as an alternative method for screening herds for prevalence and diagnosis of Johne's disease.

Do the American College of Surgeons’ “Major Resuscitation” Trauma Triage Criteria Predict Emergency Operative Management?

STUDY OBJECTIVE

We wish to assess whether individual or collective American College of Surgeons' "major resuscitation" criteria accurately identify injured patients who receive emergency operative treatment.

METHODS

In this observational secondary registry analysis of 8,289 consecutive trauma team activations during a 7.5-year period, we evaluated the test performance of 5 American College of Surgeons' major criteria in predicting emergency (within 1 hour) operative management by general (for adults) or pediatric (for children) surgeons.

RESULTS

In adults, the individual major resuscitation criteria each predicted emergency operative management as follows (sorted from highest to lowest test performance): gunshot wounds to the neck or torso (likelihood ratio positive [LR+] 7.5; 95% confidence interval [CI] 6.2 to 9.1); confirmed hypotension (LR+ 5.3; 95% CI 4.0 to 7.1); interhospital transfers requiring blood transfusions (LR+ 4.6; 95% CI 2.6 to 8.2); respiratory compromise (LR+ 2.9; 95% CI 2.2 to 3.7), and Glasgow Coma Scale score less than 8 (LR+ 2.1; 95% CI 1.6 to 2.7). The collective strategy of using any of these 5 criteria exhibited a LR+ of 3.5 (95% CI 3.2 to 3.8), sensitivity 82% (95% CI 75% to 87%), and specificity 76% (95% CI 75% to 77%). Our findings in children were similar, but their precision was limited by the low baseline prevalence of emergency operative intervention.

CONCLUSION

These 5 American College of Surgeons-mandated major resuscitation criteria vary several-fold in their individual ability to predict emergency operative management and collectively exhibit modest test characteristics for this purpose. Selective use of these criteria or revisions thereof could result in more efficient secondary trauma triage. Our results do not support the existing obligatory use of these criteria to maintain American College of Surgeons trauma center certification.

Teaching Bayesian analysis to Emergency Medicine residents

Our objective was to determine if a brief didactic would improve Emergency Medicine (EM) resident performance at using a key evidence-based medicine (EBM) concept. We used a prospective, before and after, assessment of EM resident estimates of post-test pulmonary embolism (PE) probability for a defined pre-test probability, computed tomography (CT) and D-dimer results. The survey provided test sensitivity, and specificity for D-dimer and CT. Three months later, residents attended a brief didactic conference on how to use Fagan's Nomogram and likelihood ratios (LRs) to calculate post-test probability of disease. The accuracy of estimates of post-test PE probability was reassessed. The absolute percentage difference in resident estimates from the true post-test PE probabilities decreased from 14.5% (95% confidence interval [CI] 9.7%-19.9%) to 4.5% (95% CI 2.0-6.8%) after the educational intervention. This 10% effect size was statistically significant, p = 0.002. The study demonstrates the efficacy of the lecture method in teaching an EBM concept to EM residents.

Extreme thrombocytosis predicts Kawasaki disease in infants

Infants with Kawasaki disease are at high risk of developing life-threatening coronary complications, yet may elude timely diagnosis because they often lack the full complement of classic clinical features. We retrospectively studied 26,540 children 1 year of age or less who were evaluated at a tertiary care pediatric emergency department in whom a platelet count was performed. Among those infants with fever without a source identified, 8.5% with platelet counts of 800,000 cells/mm(3) or greater had Kawasaki disease compared to 0.4% with platelet counts of less than 800,000 cells/mm(3) (likelihood ratio for Kawasaki disease was 17 [95% confidence interval, 8-34]). Because many infants present atypically, Kawasaki disease should be considered in all children of 1 year or less with prolonged fever, extreme elevation of the platelet count, and no compelling alternative diagnosis.

Clinical experience did not reduce the variance in physicians' estimates of pretest probability in a cross-sectional survey

BACKGROUND AND OBJECTIVES

In light of the increasing popularity of the threshold approach in clinical decision-making, this study assesses the role of expertise in physicians' agreement in estimating the probability of disease in patients.

METHODS

A cross-sectional survey of physicians of different specialties, attending weekly staff meetings in four teaching hospitals in Jerusalem, Israel. An anonymous questionnaire describing three case scenarios of patients with chest pain was administered and participants were asked to estimate pretest probabilities of disease.

RESULTS

Eighty-six physicians (practicing cardiology, internal medicine, and family medicine, as well as general practitioners and internists) out of 125 approached (response rate 69%). The mean estimated probabilities were very similar for residents and specialists; however, the standard deviation was higher for specialists in all three cases: 20.7, 21.0, and 19.1 among specialists and 16.4, 20.5, and 14.9 among residents, respectively.

CONCLUSION

This study, based on case scenarios, did not find that medical expertise improved agreement among doctors when estimating the probability of disease in patients-despite the common belief that senior physicians should have smaller interobserver differences in probability estimates. The wide variation observed calls into question the applicability of the threshold approach.

Likelihood ratios increase diagnostic certainty in pulmonary embolism

Pulmonary embolism (PE) is a difficult diagnosis to confirm. The choice of tests has led to a myriad of algorithms. Diagnostic uncertainty can be quantified by the application of the tests' likelihood ratios (LR). Positive and negative LR enable the conversion of a pretest to a post-test probability, given a positive and negative test result, respectively. Thus, a pretest probability of <17% and a negative D-dimer with a negative LR of 0.05 (sensitivity 98%, specificity 40%) lead to a post-test probability of PE of <1%. Ventilation perfusion (V/Q) scans with a normal, very low, low, intermediate and high probability result have an LR of 0, 0.125, 0.25, 1 and 17, respectively. Also, patients with a V/Q scan result other than normal or high probability still have a post-test probability of PE from 3 to 65%. Positive and negative computed tomography pulmonary angiograms (CTPA) have an LR of 8.6 and 0.06, respectively (sensitivity 95%, specificity 89%). Patients with a high pretest probability and negative CTPA again still have a post-test probability of more than 10%. However, as the post-test probability after one test becomes the pretest probability for the next, test results used cumulatively progressively narrow the gap to a final diagnosis. The post-test probability after a D-dimer, V/Q scan, CTPA, leg ultrasound or pulmonary angiography, alone or in any combination or in any order, can be calculated using their LR. Use of LR thus assists in the precise interpretation of test results, such as in complex algorithms for PE.

World Health Organization grade 'a' motility and zona-binding test accurately predict IVF outcome for mild male factor and unexplained infertilities

BACKGROUND

The aim of this study was to determine the pronostic value of a sperm-zona pellucida (ZP) binding assay, combined with World Health Organization (WHO) grade 'a' sperm motility on the day of the IVF attempt, to predict sperm fertilizing ability in unexplained and moderate male factor infertilities.

METHODS

In total, 84 couples (64 unexplained infertility; 20 male factor) underwent both a sperm-ZP binding assay and an IVF attempt, irrespective of the test's result. The test was negative when grade 'a' motility was #5% and/or the ZP binding index was <0.7. Fertilization and pregnancy rates were related to the test's results.

RESULTS

Thirty-one patients had a negative test (group N) and 53 a positive test (group P). A difference was observed concerning the fertilization rate [median (range): 0 (0-75%) versus 50 (0-100%); P = 0.0001] and the number of cycles with fertilization rate <20% (65 versus 23%; P = 0.0002) between groups N and P respectively. In the group of unexplained and male factor infertilities, the test showed a sensitivity of 83 and 60%, specificity of 50 and 90%, positive predictive value of 76 and 86%, and negative predictive value of 61 and 69% respectively.

CONCLUSION

Sperm-ZP binding test, combined with WHO grade 'a' motility assessment, is relevant to prevent IVF fertilization failures in unexplained infertility and, most particularly, in moderate male factor infertility.

Refining clinical diagnosis with likelihood ratios

Likelihood ratios can refine clinical diagnosis on the basis of signs and symptoms; however, they are underused for patients' care. A likelihood ratio is the percentage of ill people with a given test result divided by the percentage of well individuals with the same result. Ideally, abnormal test results should be much more typical in ill individuals than in those who are well (high likelihood ratio) and normal test results should be most frequent in well people than in sick people (low likelihood ratio). Likelihood ratios near unity have little effect on decision-making; by contrast, high or low ratios can greatly shift the clinician's estimate of the probability of disease. Likelihood ratios can be calculated not only for dichotomous (positive or negative) tests but also for tests with multiple levels of results, such as creatine kinase or ventilation-perfusion scans. When combined with an accurate clinical diagnosis, likelihood ratios from ancillary tests improve diagnostic accuracy in a synergistic manner.

Validity of simple measurement to diagnose pupillary dilation

STUDY OBJECTIVE

The aim of the study was to compare the validity of 3 methods of detecting pupillary dilation: bright-light measurement, room-light measurement, and gestalt judgment.

METHODS

In reach volunteer, by random assignment, placebo was instilled in one eye and dilute phenylephrine in the other. Emergency care providers judged whether each pupil was dilated and measured it in bright light (>54,000 lux) and in room light (2700-5400 lux) while the other eye was covered. Test characteristics for measurement were determined according to published cut-points, and measurement methods were compared using receiver operating curve analysis.

RESULTS

There were 136 pupillary assessments-68 in placebo and 68 in phenylephrine eyes. Compared with gestalt judgment, bright-light measurement had higher specificity (0.94 vs 0.68) but lower sensitivity (0.43 vs 0.79). Bright-light measurement was more discriminating than room-light measurement.

CONCLUSION

Bright-light measurement has higher specificity, but lower sensitivity, than gestalt judgement, and is superior to room-light measurement.

Oxygen saturation as a predictor of prolonged, frequent bronchodilator therapy in children with acute asthma

OBJECTIVES

To examine if the initial oxygen saturation (SaO2) in the Emergency Department is a useful predictor of prolonged frequent bronchodilator therapy (FBT) in children with acute asthma.

STUDY DESIGN

Prospective cohort study of 273 children, 1 to 17 years of age, requiring systemic corticosteroids. Patients were categorized as needing FBT for >4 hours (n=166) versus >4 hours (n=107) and >12 hours (n=79) versus >12 hours (n=194). Multiple logistic regression determined the association between SaO2 and these outcomes.

RESULTS

Baseline SaO2 remains a significant independent predictor of FBT for >4 hours (OR=0.81) and >12 hours (OR=0.84); 91% of patients with SaO2 of 90% to 91% had FBT >4 hours and 80% of patients with SaO2 of < or =89% had FBT >12 hours. Children with SaO2 of < or =91% are 14.7 and 12.0 times more likely to require FBT for >4 hours and >12 hours, respectively, than those with SaO2 of 98% to 100%. The interval likelihood ratios for FBT >4 hours were 12.3 for SaO2 of < or =89%, 6.5 for 90% to 91%, but only 1.8 for 92% to 93%. The likelihood ratios for FBT >12 hours decreased from 9.8 for SaO2 of < or =89% to 3.5 for SaO2 of 90% to 91%.

CONCLUSIONS

SaO2 is a useful predictor of FBT >4 hours if it is < or =91% and of FBT >12 hours if it is < or =89%.

Usefulness and validity of diagnostic nasogastric aspiration in patients without hematemesis

STUDY OBJECTIVE

We estimate the test characteristics of nasogastric aspiration to diagnose upper gastrointestinal tract hemorrhage in patients without hematemesis.

METHODS

In this retrospective cohort study, medical records from patients admitted to 2 urban hospitals between 1997 and 2002 for gastrointestinal tract bleeding without hematemesis were reviewed. Positive nasogastric aspiration results were classified by the severity of hemorrhage, and negative results were classified by the presence or absence of bile. The reference standard for nasogastric aspiration was the source of bleeding-upper versus non--upper gastrointestinal tract--from the hospital discharge summary. Confidence intervals (CIs) for proportions and likelihood ratios (LRs) were calculated.

RESULTS

Of 333 eligible patients, 235 were offered nasogastric aspiration, and 220 accepted the test. Results of 220 attempts were distributed as follows: negative, 158 (72%), including 9 (4%) with bile; nasogastric aspiration aborted, 13 (6%); and positive, 49 (23%), including 4 (2%) that were strongly positive (> or =450 mL red blood). Test characteristics of nasogastric aspiration to detect upper gastrointestinal tract bleeding in 213 patients with a reference standard diagnosis were as follows: sensitivity 42% (95% CI 32% to 51%), specificity 91% (95% CI 83% to 95%), negative predictive value 64% (95% CI 56% to 71%), and positive predictive value 92% (95% CI 79% to 97%). The nasogastric aspiration accurately predicted the source of bleeding in 66% of patients (95% CI 59% to 72%). The likelihood ratio of a positive nasogastric aspiration was 11 (95% CI 4 to 30), and the likelihood ratio of a negative nasogastric aspiration was 0.6 (95% CI 0.5 to 0.7).

CONCLUSION

In patients without hematemesis, a positive nasogastric aspiration, seen in 23%, indicates probable upper gastrointestinal tract bleeding (LR+ 11), but a negative nasogastric aspiration, seen in 72%, provides little information (LR- 0.6).

Probabilistic reasoning and clinical decision-making: do doctors overestimate diagnostic probabilities?

BACKGROUND

The "threshold approach" is based on a physician's assessment of the likelihood of a disease expressed as a probability. The use of Bayes' theorem to calculate disease probability in patients with and without a particular characteristic, may be hampered by the presence of subadditivity (i.e. the sum of probabilities concerning a single case scenario exceeding 100%).

AIM

To assess the presence of subadditivity in physicians' estimations of probabilities and the degree of concordance among doctors in their probability assessments.

DESIGN

Prospective questionnaire.

METHODS

Residents and trained physicians in Family Medicine, Internal Medicine and Cardiology (n = 84) were asked to estimate the probability of each component of the differential diagnosis in a case scenario describing a patient with chest pain.

RESULTS

Subadditivity was exhibited in 65% of the participants. The total sum of probabilities given by each participant ranged from 44% to 290% (mean 137%). There was wide variability in the assignment of probabilities for each diagnostic possibility (SD 16-21%).

DISCUSSION

The finding of substantial subadditivity, coupled with the marked discordance in probability estimates, questions the applicability of the threshold approach. Physicians need guidance, explicit tools and formal training in probability estimation to optimize the use of this approach in clinical practice.

Evidence-based emergency medicine/skills for evidence-based emergency care. Interval likelihood ratios: another advantage for the evidence-based diagnostician

Emergency physicians are often confronted with making diagnostic decisions on the basis of a test result represented on a continuous scale. When the results of continuous data are expressed as binary outcomes using a single cutoff, loss of information and distortion may occur. In this setting, interval likelihood ratios provide a distinct advantage in interpretation over those based on a dichotomized sensitivity and specificity. Dividing the data into intervals uses more of the information contained in the data and allows the clinician to more appropriately interpret the test results and to make valid clinical decisions. This article illustrates the advantages of interval likelihood ratios with examples and demonstrates how to calculate them on the basis of different data formats. Authors and journals need to be encouraged to report the results of studies of performance of diagnostic tests using interval ranges rather than simple dichotomization when the tests involve continuous variables.

Reliability and validity of lumbosacral spine radiograph reading by chiropractors, chiropractic radiologists, and medical radiologists

STUDY DESIGN

A cross-sectional diagnostic study was conducted in two sessions.

OBJECTIVE

To determine and compare the reliability and validity of contraindications to chiropractic treatment (infections, malignancies, inflammatory spondylitis, and spondylolysis-listhesis) detected by chiropractors, chiropractic radiologists, and medical radiologists on plain lumbosacral radiographs.

SUMMARY OF BACKGROUND DATA

Plain radiography of the spine is an established part of chiropractic practice. Few studies have assessed the ability of chiropractors to read plain radiographs.

METHODS

Five chiropractors, three chiropractic radiologists and five medical radiologists read a set of 300 blinded lumbosacral radiographs, 50 of which showed an abnormality (prevalence, 16.7%), in two sessions. The results were expressed in terms of reliability (percentage and kappa) and validity (sensitivity and specificity).

RESULTS

The interobserver agreement in the first session showed generalized kappas of 0.44 for the chiropractors, 0.55 for the chiropractic radiologists, and 0.60 for the medical radiologists. The intraobserver agreement showed mean kappas of 0.58, 0.68, and 0.72, respectively. The difference between the chiropractic radiologists and medical radiologists was not significant. However, there was a difference between the chiropractors and the other professional groups. The mean sensitivity and specificity of the first round, respectively was 0.86 and 0.88 for the chiropractors, 0.90 and 0.84 for the chiropractic radiologists, and 0.84 and 0.92 for the medical radiologists. No differences in the sensitivities were found between the professional groups. The medical radiologists were more specific than the others.

CONCLUSIONS

Small differences with little clinical relevance were found. All the professional groups could adequately detect contraindications to chiropractic treatment on radiographs. For this indication, there is no reason to restrict interpretation of radiographs to medical radiologists. Good professional relationships between the professions are recommended to facilitate interprofessional consultation in case of doubt by the chiropractors.

Initial oxygen saturation as a predictor of admission in children presenting to the emergency department with acute asthma

BACKGROUND

Previous studies have looked at the relationship between initial oxygen saturation (SaO (2)) and the need for admission in children presenting with an acute asthma exacerbation. If initial SaO (2) value is indeed predictive of admission, then the admission process could be initiated sooner, and time spent in the emergency department could be potentially lessened.

STUDY OBJECTIVE

The objective of the current study was to examine whether initial room air SaO (2) in children presenting to the ED with acute asthma is a reliable predictor of hospital admission.

METHODS

This was a prospective multicenter study during 1997 and 1998 at 44 North American EDs as part of the Multicenter Airway Research Collaboration. Inclusion criteria were physician diagnosis of acute asthma and age between 2 and 17 years. The association between hospital admission and SaO (2) was examined by using logistic regression. Likelihood ratios were used to assess the diagnostic value of SaO (2).

RESULTS

Of the 1,184 children enrolled in the current study, 1,040 (88%) had a documented initial SaO (2) value on room air. The mean age of the cohort was 8+/-4 years, with a mean initial SaO (2) of 95%+/-4%. Overall, 241 (23%) children were admitted to the hospital. The mean SaO (2) value of children admitted to the hospital was 93%+/-5% versus 96%+/-3% for those not admitted (P <.001). The admission rate decreased with increasing SaO (2); 73% (30/41) of children with an SaO (2) value of 88% or less were admitted versus 8% (7/88) with an SaO (2) value of 100%. In the logistic regression model, children with an SaO (2) value of 88% or less were 32 (95% confidence interval 11 to 89) times more likely to be admitted than those with an SaO (2) value of 100%. The likelihood ratio for admission was 12 for children with an SaO (2) value of 88% or less (42/1,040) but decreased to 4.6 for children with an SaO (2) value of 91% or less (130/1,040) and 2.7 for children with an SaO (2) value of 94% or less (333/1,040).

CONCLUSION

This large, clinical multicenter study does not support earlier findings that SaO (2) alone is a clinically useful predictor of hospital admission in children who present to the ED with acute asthma.

Examining diagnostic tests: an evidence-based perspective

Diagnosis is an important aspect of physical therapist practice. Selecting tests that will provide the most accurate information and evaluating the results appropriately are important clinical skills. Most of the discussion in physical therapy to date has centered on defining diagnosis, with considerably less attention paid to elucidating the diagnostic process. Determining the best diagnostic tests for use in clinical situations requires an ability to appraise evidence in the literature that describes the accuracy and interpretation of the results of testing. Important issues for judging studies of diagnostic tests are not widely disseminated or adhered to in the literature. Lack of awareness of these issues may lead to misinterpretation of the results. The application of evidence to clinical practice also requires an understanding of evidence and its use in decision making. The purpose of this article is to present an evidence-based perspective on the diagnostic process in physical therapy. Issues relevant to the appraisal of evidence regarding diagnostic tests and integration of the evidence into patient management are presented.

Emergency department ultrasonography in the evaluation of hypotensive and normotensive children with blunt abdominal trauma

PURPOSE

The aim of this study was to evaluate the accuracy of emergency department (ED) ultrasound scan in identifying which children with blunt torso trauma have intraperitoneal fluid associated with intraabdominal injuries (IAI).

METHODS

The authors conducted a prospective, observational study of children (< 16 years old) with blunt trauma who presented to a level 1 trauma center over a 29-month period and underwent abdominal ultrasound scan while in the ED. Ultrasound examinations were ordered at the discretion of the trauma surgeons or ED physicians caring for the patients, performed by trained sonographers, and interpreted at the time of the ultrasound. Ultrasound examinations were interpreted solely for the presence or absence of intraperitoneal fluid. Hypotension was defined as > or = 1 standard deviation below the age-adjusted mean. Patients underwent follow-up to identify those with intraperitoneal fluid and IAI.

RESULTS

A total of 224 pediatric blunt trauma patients had ultrasound scan performed and were enrolled. Thirty-three patients had IAI with intraperitoneal fluid, and ultrasound scan was positive in 27. The accuracy of abdominal ultrasound for detecting intraperitoneal fluid associated with IAI was sensitivity, 82% (95% confidence interval [CI] 65% to 93%); specificity, 95% (95% CI 91% to 97%); positive predictive value, 73% (95% CI 56% to 86%); and negative predictive value, 97% (95% CI 93% to 99%). In the 13 patients who were hypotensive, ultrasound scan correctly identified intraperitoneal fluid in all 7 patients (sensitivity 100%) with IAI, and hemoperitoneum and was negative in all 6 patients (specificity 100%) who did not have hemoperitoneum. Nine patients had IAI without intraperitoneal fluid, and ultrasound scan result was negative for fluid in all 9.

CONCLUSIONS

ED abdominal ultrasound scan used solely for the detection of intraperitoneal fluid in pediatric blunt trauma patients has a modest accuracy. Ultrasonography has the best test performance in those children who are hypotensive and should be obtained early in the ED evaluation of these patients.

Predictors of ventriculoperitoneal shunt pathology

BACKGROUND

Diagnosis of ventriculoperitoneal (VP) shunt pathology remains a dilemma in patients with nonspecific constitutional signs and symptoms. Eosinophilia has been described in association with shunt infection and malfunction. Our purpose was to further define the relationship of eosinophilia and shunt pathology and to determine other predictors of VP shunt infection and malfunction.

METHODS

Records of all patients admitted with a suspected VP shunt infection or malfunction were reviewed. The following data were abstracted: age; reason for and age at initial shunt placement; number of revisions; date of last revision; history of fever or vomiting; ventricular fluid cell count; differential and culture; complete blood count and differential; need for shunt revision or replacement; and use of antibiotics. After exclusion of patients admitted for initial shunt placement, the remainder were divided into three groups: those with shunt infection; those with shunt malfunction; and those without documented infection or malfunction.

RESULTS

Of 12 patients with shunt infection and 69 with shunt malfunction, 2 and 11, respectively, had eosinophilia defined as > or =5%. The presence of eosinophilia had a 96% positive predictive value for shunt pathology and raised the pretest probability of pathology from 84% to a post test probability of 96%. The combination of fever history and ventricular fluid neutrophils >10% had a 99% specificity for shunt infection, had a 93 and 95% positive and negative predictive value, respectively, and raised the pretest probability of infection from 12% to a posttest probability of 92%.

CONCLUSIONS

In patients suspected of having a VP shunt malfunction, the presence of > or =5% eosinophils in the ventricular fluid indicates shunt pathology. The combination of fever and ventricular fluid neutrophils > 10% is predictive of shunt infection.

Can urine clarity exclude the diagnosis of urinary tract infection?

OBJECTIVES

To determine the association of clear urine by visual inspection with the absence of significant bacteruria, and to compare it with standard urinalysis.

METHODS

The study was performed in the emergency department of Children's Hospital Medical Center, Cincinnati, Ohio. It was a prospective, convenience sample of children <21 years of age who had catheterized or midstream clean-catch urine specimen collected for culture. Clinical findings including the presence or absence of fever, abdominal pain, dysuria, frequency, and urgency were collected for each patient. Urine was visually assessed for clarity by 2 independent observers using a standardized technique. Standard laboratory urinalysis and microscopy were also performed on all specimens. A positive urine culture was defined as >/=10(4) colony-forming unit (CFU)/mL of a urinary pathogen if obtained by catheterization and >/=10(5) CFU/mL if obtained by midstream.

RESULTS

Samples were obtained from 159 patients ranging in age from 4 weeks to 19 years. Females comprised 77% of the patients. One hundred ten of the samples (69%) were clear to visual inspection. There were a total of 29 positive cultures; however, 3 were in children with clear urine. The finding of clear urine on visual inspection had a negative predictive value of 97.3%. These results were similar to those obtained with standard urinalysis.

CONCLUSION

Clear urine on visual inspection cannot completely eliminate the possibility that a child has a urinary tract infection. However, it is a reproducible test that offers the advantages of being simple, fast, and inexpensive. The finding of clear urine should be considered a reasonable and relatively effective bedside screen for the presence of a urinary tract infection.