A probability graph describing the predictive value of a highly sensitive diagnostic test

Inventaires d'auto-évaluation de la psychopathologie chez l'adulte. 1repartie: inventaires multidimensionnels

Les inventaires d'auto-évaluation de la psychopathologie chez l'adulte ont connu un développement remarquable au cours des 2 dernières décennies, à la fois en recherche clinique et en épidémiologie; leurs avantages et leurs limitations sont considérés dans cette revue. Leur construction, leur adaptation éventuelle, l'évaluation de leurs qualités métrologiques et l'établissement de leur stabilité factorielle reposent sur des concepts mélhodologiques dont l'origine et les définitions sont brièvement rappelées. LesHopkins Symptom Check-Lists(HSCL) et leGeneral Health Questionnaire(GHQ) représentent actuellement les 2 grandes familles d'inventaires multidimensionnels. Parmi les différentes HSCL, la HSCL-58 est l'inventaire le plus couramment utilisé, notamment en psychopharmacologie, grâce à sa diffusion, à sa facilité d'emploi, à sa bonne validité externe et à sa structure factorielle si on se limite aux troubles anxiodépressifs non psychotiques. L'existence des 9 dimensions symptomatiques primaires de la SCL-90 est beaucoup moins bien établie, mais ce questionnaire a été utilisé pour des enquêtes épidémiologiques et des essais cliniques, globalement ou à travers certaines de ses dimensions. Plusieurs versions et plusieurs cotations du GHQ ont été produites à partir des 60 items initiaux (GHQ-30, 28, 20, 12). Ils ont fait l'objet de nombreuses études de validation, qui ont, dans l'ensemble, confirmé son intérêt. Le GHQ-28, le plus récent, produit d'analyses factorielles du GHQ-60, englobe 4 dimensions de 7 items chacune: anxiété et insomnie, dépression sévère, gêne sociale, plaintes somatiques. A partir de plusieurs exemples internationaux, le GHQ s'est avéré être un outil intéressant pour le dépistage des cas psychiatriques. En conséquence, la version française du GHQ-28 nous paraît devoir être largement diffusée.

Likelihood Ratios for the Emergency Physician

The concept of likelihood ratios was introduced more than 40 years ago, yet this powerful metric has still not seen wider application or discussion in the medical decision-making process. There is concern that clinicians-in-training are still being taught an oversimplified approach to diagnostic test performance and have limited exposure to likelihood ratios. Even for those familiar with likelihood ratios, they might perceive them as mathematically cumbersome in application, if not difficult to determine for a particular disease process. This article takes a conceptual approach to likelihood ratios and applies them to two clinical settings: 1) severe intracranial injury after minor head trauma and 2) suspected pulmonary embolism with shortness of breath. Likelihood ratios are the most appropriate metric for efficient rational clinical examination and can prevent unnecessary and wasteful treatments and procedures.

Accuracy of Alcohol Use Disorders Identification Test for detecting problem drinking in 18-35 year-olds in England: method comparison study

AIMS

To assess the accuracy of Alcohol Use Disorders Identification Test (AUDIT) scores for problem drinking in males and females aged 18-35 in England.

METHODS

A method comparison study with 420 primary care patients aged 18-35. Test measures were AUDIT and AUDIT-C. Reference standard measures were (a) Time-Line Follow-Back interview for hazardous drinking; World Mental Health Composite International Diagnostic Interview for (b) DSM-IV alcohol abuse, (c) DSM-IV alcohol dependence, (d) DSM-5 alcohol use disorders.

RESULTS

Area under the curve (AUC) was (a) 0.79 (95% CI 0.73-0.85; males) and 0.84 (0.79-0.88; females); (b) 0.62 (0.54-0.72; males) and 0.65 (0.57-0.72; females); (c) 0.77 (0.65-0.87; males) and 0.76 (0.67-0.74; females); (d) 0.70 (0.60-0.78; males) and 0.73 (CI 0.67-0.78; females). Identification of threshold cut-point scores from the AUC was not straightforward. Youden J statistic optimal cut-point scores varied by 4-6 AUDIT scale points for each outcome according to whether sensitivity or specificity were prioritized. Using Bayes' Theorem, the post-test probability of drinking problems changed as AUDIT score increased, according to the slope of the probability curve.

CONCLUSION

The full AUDIT scale showed good or very good accuracy for all outcome measures for males and females, except for alcohol abuse which had sufficient accuracy. In a screening scenario where sensitivity might be prioritized, the optimal cut-point is lower than established AUDIT cut-points of 8+ for men and 6+ for women. Bayes' Theorem to calculate individual probabilities for problem drinking offers an alternative to arbitrary cut-point threshold scores in screening and brief intervention programmes.

Informational odds ratio: a useful measure of epidemiologic association in environment exposure studies

The informational odds ratio (IOR) measures the post-exposure odds divided by the pre-exposure odds (ie, information gained after knowing exposure status). A desirable property of an adjusted ratio estimate is collapsibility (ie, the combined crude ratio will not change after adjusting for a variable that is not a confounder). Adjusted traditional odds ratios (TORs) are not collapsible. In contrast, Mantel-Haenszel adjusted IORs generally are collapsible. IORs are a useful measure of disease association in environmental case-referent studies, especially when the disease is common in the exposed and/or unexposed groups.

The interpretation of serum biochemistry test results in domestic animals

The interpretation of serum biochemistry tests used in the investigation of disease was reviewed. Different methods of calculating reference ranges for biochemical constituents of serum were discussed. Reference ranges in current laboratory use were analyzed statistically to produce means and coefficients of variation.

Predictive value of fetal pulmonary venous flow patterns in identifying the need for atrial septoplasty in the newborn with hypoplastic left ventricle

BACKGROUND

Pulmonary venous Doppler (PVD) flow patterns in the fetus with hypoplastic left heart syndrome (HLHS) have been correlated with restrictive interatrial communication or intact atrial septum (RAS) postnatally; however, the ability of PVD to identify the neonate requiring emergent atrial septoplasty (EAS) for severe left atrial hypertension and hypoxemia has not been critically evaluated. It was the purpose of this study to determine the predictive power of fetal PVD in identifying the need for EAS in newborns with HLHS and RAS.

METHODS AND RESULTS

Forty-one patients with fetal PVD flow analysis and postnatally confirmed HLHS were studied. Pulsed-wave assessment of PVD flow included S-, D-, and A-wave velocity, time-velocity integral (VTI) of forward and reverse flow, and S/D velocity and forward/reverse VTI ratio. Neonatal EAS was used as the primary clinical outcome variable. Receiver operating characteristic curves were used to determine cutpoints at which PVD indices best predicted EAS. Cutpoints were evaluated for clinical accuracy and usefulness by use of Bayesian analysis. Eight of 41 subjects underwent EAS. Need for EAS was most accurately predicted by forward/reverse VTI ratio <5 (sensitivity, 0.88, 95% CI, 0.49 to 0.99; specificity, 0.97, 95% CI, 0.82 to 0.99), which, when present, increases the posttest likelihood of EAS to 74%, assuming a pretest prevalence of 10%. Accuracy and usefulness of other PVD indices were affected by false-positive results.

CONCLUSIONS

In the fetus with HLHS, a PVD forward/reverse VTI ratio of <5 is the strongest predictor of the need for EAS in the newborn period. These observations should improve our ability to identify and expectantly manage the fetus with HLHS and RAS.

Error patterns of 3rd-year medical students on the cardiovascular physical examination

BACKGROUND

Recent research documents widespread deficits in the physical examination skills of practicing physicians.

PURPOSE

This study explored physical examination skills of 3rd-year medical students after completion of a course in physical diagnosis.

METHODS

Standardized patient physical examination checklist data were analyzed for a cohort of 2,038 medical students for a patient presenting with classic signs and symptoms of an acute myocardial infarction. A follow-up paper case and survey explored reasons underlying omissions.

RESULTS

Students systematically omitted 3 of 10 component maneuvers critical to the evaluation of a patient with shortness of breath and chest pain. The same pattern of omissions was observed across 8 medical schools and over 2 successive years. The paper case follow-up study ruled out time constraints and performance anxiety as the cause. Survey data revealed that students may omit a maneuver due to inability to recall pertinence (blood pressure in both arms) or difficulty discriminating findings (heart sounds at different locations), or because of inadequate technical mastery (percussion of the lungs).

CONCLUSIONS

These data suggest fundamental inadequacies in the current paradigm for teaching physical examination skills. Standardized patient checklist data can provide an informative window into the efficacy of teaching practices.

HLA markers and prediction of clinical course and outcome in rheumatoid arthritis

OBJECTIVE

To evaluate HLA markers as early prognostic factors for disease severity in rheumatoid arthritis (RA).

METHODS

HLA genotyping was carried out in a retrospective analysis of 66 RA patients and in a prospective study of 55 RA patients and 87 healthy controls using polymerase chain reaction-based methods for HLA-DRB1 specificities, DR4 alleles, and their linked DQB1 alleles, as well as HLA-B27. The clinical course of RA was assessed by clinical and radiologic scores. The impact of HLA markers was evaluated by epidemiologic means in addition to modeling using multiple logistic regression analysis.

RESULTS

Shared epitope-positive (HVR3+) DR4 alleles and the HVR3 amino acid cassette QKRAA were associated with RA in both longstanding (relative risk [RR] 3.34 and 3.19) and recent-onset (RR 2.1 and 2.37) RA. In longstanding RA, radiologic evidence of severe joint destruction (Larsen score > 1.62) was seen more often in HVR3 shared epitope-positive patients than in epitope-negative patients (odds ratio [OR] = 25.67, chi 2 = 13.59, P = 0.0003). Moreover, rank sum analysis of Larsen indices indicated significantly higher ranking for the presence of the RA-associated HVR3 cassettes (QKRAA, QRRAA) when expressed on a DR4 allele (P < 0.0001). In the prospective study, DR4-positive patients had a significantly increased risk (OR = 13.75, P = 0.00083) of developing bony erosions. In addition, HVR3 epitope-positive DR4-positive individuals had significantly higher Larsen indices than did epitope-negative patients (P = 0.0083). In particular, the presence of the HVR3 epitope on DR4 resulted in an increased a posteriori likelihood (0.91) of developing early erosive disease compared with an a priori risk of 0.62. Conversely, the likelihood decreased to a minimum of 0.35 when the HVR3 epitope was absent.

CONCLUSION

While the contribution of HLA typing to establishing the diagnosis of RA is limited, HLA-DR genotyping and DR4 subtype determination provide valuable markers for the prognosis of joint destruction in RA.

A public/private partnership in the quest for quality: development of cerebrovascular disease practice guidelines and review criteria

The development of Uniform Clinical Data Set (UCDS) algorithms for cerebrovascular disease can serve as a model for cooperation between government agencies and physician organizations in the development of clinical guidelines and the translation of practice guidelines into review criteria. UCDS is a computerized system of data collection and case finding developed by the Health Care Financing Administration (HCFA) to standardize the Peer Review Organization screening process for potential utilization and quality problems. Input on the cerebrovascular disease algorithms was provided through the American Medical Association UCDS Advisory Panel and the American Academy of Neurology Task Force on Hospital Utilization for Stroke. Physician organizations were successful in identifying problematic areas, improving the original algorithms, suggesting new issues for development, and deleting inappropriate or misdirected algorithms. Since HCFA modified the cerebrovascular disease algorithms in response to feedback, physician organizations have had an important impact on algorithm development. Unfortunately, the process for algorithm review was often complicated, inefficient, and slow. This process is reviewed, problems and constraints are discussed, and specific recommendations are given for improving the process.

The prevalence, six-month persistence, and predictive values of laboratory indicators of bacterial vaginosis (nonspecific vaginitis) in asymptomatic women

The natural course of signs and laboratory test findings indicative of bacterial vaginosis was followed in an observational noninterventive 6-month longitudinal study of 270 asymptomatic women. Only the minority of positive Gardnerella vaginalis cultures (5 of 33), wet mount clue cells (5 of 14), sniff tests (3 of 11), Papanicolaou smear clue cells (0 of 5), and discharge consistent with bacterial vaginosis (11 of 49) persisted in the absence of therapy. While these four laboratory parameters as well as chromatographic succinate/lactate ratios (performed only on the final visit) were abnormal significantly more often in patients with abnormal discharge than in those with normal discharge (p = 0.006, p less than 0.0001, p less than 0.0001, p = 0.0003, and p = 0.002, respectively), all were insensitive predictors of abnormal discharge with sensitivities ranging between 10.6% and 20.2% and abnormal test predictive values between 30.6% and 65.2%. We conclude that G. vaginalis represents indigenous flora in some normal women and that therapy is unwarranted for the incidental finding of a positive laboratory indicator of bacterial vaginosis in the patient without symptoms.

The predictive factor--a method to simplify Bayes' formula and its application to diagnostic procedures

According to Bayes' rule the predictive value (PV) of a diagnostic test (= probability of disease if the test is positive) depends on the prevalence of the disease (= a priori probability), the sensitivity (c1) and the specificity (c2) of the test. A new variable has been introduced, the predictive factor (c), which is calculated as follows: c = c1/(c1 +1 -c2). Since the PV only depends on this factor and on the prevalence, the calculation is much easier and a general graphical solution is possible. This simplification renders several additional advantages and facilitates the understanding of the dependence of PV on prevalence.

Significance of haemodynamic variables in coronary care unit for prediction of survival after acute myocardial infarction

In order to assess the value of haemodynamic monitoring in the coronary care unit for long term prognosis after recovery of an acute myocardial infarction, the records of two groups of consecutive patients were reviewed retrospectively. From 254 patients, 32 (13%) died in the hospital and nine patients had to be excluded from subsequent follow-up for various reasons. Four year mortality among the 213 patients who were discharged from the hospital and could be followed up was 26%. Of the haemodynamic variables measured on admission a high pulmonary capillary wedge pressure, exceeding 18 mmHg, and a low mixed venous oxygen saturation, less than 60%, were not only associated with a high hospital but also with a high four year mortality, whereas a low systolic blood pressure (less than 100 mmHg), an important prognosticator during admission to hospital, was only of minor significance thereafter. A negative value on admission of a specific index 0.24 X systolic blood pressure (mmHg) -0.217 X pulmonary capillary wedge pressure (mmHg)+0.234 X mixed venous oxygen saturation (%)-13.1 developed for the prediction of short term survival was also associated with a much higher four year mortality than a positive value. Low cardiac index on admission could be correlated with high mortality during the first two years after discharge, whereas only 9% of patients with a higher cardiac index died. Haemodynamic monitoring in the coronary care unit is thus not only relevant for the immediate prognosis, but a high mortality risk during hospital stay persists for several years after discharge.

Clinical decision-making with treadmill testing and thallium 201

Fifty patients were prospectively evaluated for myocardial ischemia utilizing treadmill testing and thallium-201 imaging. By coronary angiography, 43 had significant coronary stenosis and seven were normal. The sensitivity, specificity, accuracy and predictive value of treadmill testing alone (81 per cent, 71 per cent, 80 per cent and 95 per cent) did not statistically differ from that of thallium-201 imaging (70 per cent, 86 per cent, 72 per cent and 97 per cent). Combined treadmill testing and thallium-201 imaging (84 per cent, 71 per cent, 80 per cent and 98 per cent) did not significantly affect the results of treadmill testing alone. Thallium-201 imaging failed to identify a number of patients with high risk lesions. The high prevalence of disease, the presentation of typical angina, preselection bias, multiple lead monitoring and exclusion of patients with abnormalities on the resting electrocardiogram probably accounted for failure of thallium-201 imaging to improve the results obtained with treadmill testing. The use of thallium-201 imaging in certain subsets of patient (resting electrocardiographic abnormalities, nondiagnostic treadmill testing, atypical chest pain or asymptomatic patients with abnormalities on treadmill testing) may be of value. However, the use of thallium-201 imaging as a routine screening procedure for myocardial ischemia in patients with typical angina, without due consideration of the prevalence of the disease in the population, is not justified.

Plasma zinc in acute myocardial infarction. Diagnostic and prognostic implications

Zinc is a metal component of many important enzymes, and its availability controls the rate of synthesis of nucleic acids and protein. Serum zinc levels have been shown to fall after acute tissue injury, including myocardial infarction. The purpose of this clinical study was to examine the value of plasma zinc measurements in a coronary care unit. Studies were made in 188 patients: 88 with unequivocal myocardial infarction, 52 controls, and 48 in a borderline group. Patients with myocardial infarction showed a fall in plasma zinc within the first three days, whereas patients in the other two groups did not. The difference in the mean minimum zinc levels between the groups with and without infarction was highly significant. In patients with myocardial infarction there was good correlation between the minimum plasma zinc level and the peal value of plasma enzymes, and also with some clinical estimators of prognosis. A fall in plasma zinc is a reliable diagnostic test for acute myocardial infarction, and the extent of the fall has prognostic implications.

Primer on certain elements of medical decision making

The value of a diagnostic test lies in its ability to detect patients with disease (its sensitivity) and to exclude patients without disease (its specificity). For tests with binary outcomes, these measures are fixed. For tests with a continuous scale of values, various cutoff points can be selected to adjust the sensitivity and specificity of the test to conform with the physician's goals. Principles of statistical decision theory and information theory suggest technics for objectively determining these cutoff points, depending upon whether the physician is concerned with health costs, with financial costs, or with the information content of the test.