Hui and Walter's latent-class reference-free approach may be more useful in assessing agreement than diagnostic performance.

BACKGROUND AND OBJECTIVE: Hui and Walter developed a latent class approach to assess the accuracy of a diagnostic procedure when no reference test is available. Our objective was to compare sensitivity and specificity estimates obtained with this reference-free approach and standard approaches, and to examine how and why they differed on a computerized tomography (CT) scan case study. STUDY DESIGN AND SETTING: We compared two sets of sensitivity and specificity estimates from four radiologists independently assessing tumoral and lymph node extension of 85 lung cancer patients with preoperative thoracic CT scan, those obtained relative to pathology findings from surgical specimens (reference set), and those derived from Hui and Walter's approach. RESULTS: The two sets of estimates significantly and markedly differed from each other. From simulations, we found that small-sample bias in Hui and Walter's estimates could be a major factor in explaining this difference. Furthermore, errors in pathology findings could account for part of this difference. Finally, our analyses revealed that the latent classes may differ intrinsically from the reference classes as defined from pathology findings and may have a different interpretation. CONCLUSION: Diagnostic parameters estimated with respect to latent classes may be more useful in providing a complete assessment of interobserver agreement than in assessing diagnostic performance.

Interobserver and intraobserver reproducibility in the histopathology of follicular thyroid carcinoma.

We evaluated the interobserver and intraobserver reproducibility in the histopathology of follicular thyroid carcinoma (FTC). Forty-one anonymous FTC pathology slides were independently reviewed by 5 pathologists, and 31 of them were also evaluated twice by the same pathologist. A final consensus diagnosis (FCD) was made at the end of the study. Interobserver and intraobserver agreement was determined as the kappa statistic for qualitative data and intraclass correlation coefficient for quantitative data. The agreement between the 5 observers' initial diagnosis and the FCD was 0.69, 0.41, 0.35, 0.28 and 0.11, respectively, strongly suggesting a leadership phenomenon. The FCD classified 30 cases as malignant, including 24 cases diagnosed as FTC. There was unanimous agreement about 13 of the 24 FTCs. Diagnostic reproducibility was found to be acceptable for the nonminimally invasive FTC. Diagnostic discrepancies occurred in 57% of the seven cases classified as minimally invasive FTC by the FCD. FCD excluded malignancy in 11 cases including 6 atypical adenomas. Interobserver and intraobserver agreement for FTC diagnosis was 0.23 (standard error [SE], 0.04) and 0.68, respectively. Interobserver and intraobserver agreement for the presence of vascular invasion was 0.20 (SE, 0.04) and 0.51, respectively, contrasting with a moderate to substantial level of agreement when considering the number of vascular invasion. Interobserver and intraobserver agreement for nucleus optical clearing were slight and moderate, respectively. The importance of the study is the confirmation that diagnostic reproducibility of minimally invasive FTC is low and that this has clinical implications, and also implications for the design of studies into the treatment and outcome of FTC.

Prevalence and risk factors for carriage of methicillin-resistant Staphylococcus aureus at admission to the intensive care unit: results of a multicenter study.

BACKGROUND: Detection of methicillin-resistant Staphylococcus aureus (MRSA) carriers on admission to the intensive care unit (ICU) is an important component of strategies for controlling the spread of MRSA. METHODS: A prospective multicenter study was conducted in 14 French ICUs for 6 months. All patients were screened within 24 hours after admission, using nasal and cutaneous swabs In addition, clinical samples were obtained. Patient data collected on ICU admission included presence of immunosuppression; history of hospital stay, surgery, antimicrobial treatments, or previous colonization with MRSA; chronic health evaluation and McCabe scores; reason for admission; whether the patient was transferred from another ward; severity of illness; presence of skin lesions; and invasive procedures. Risk factors for MRSA carriage at ICU admission were estimated, and significantly associated variables were used to develop a predictive score for MRSA carriage. A cost-benefit analysis was then performed. RESULTS: Of the 2347 admissions with MRSA screening, 162 (6.9%; range, 3.7%-20.0% among ICUs) were positive for MRSA, of whom 54.3% were detected through screening specimens only. Of the 2310 first admissions (vs repeat admissions) to the ICU, 96 were newly identified MRSA carriers. Factors associated with MRSA carriage in the multivariate analysis were age older than 60 years, prolonged hospital stay in transferred patients, history of hospitalization or surgery, and presence of open skin lesions in directly admitted patients. Only universal screening detected MRSA carriage with acceptable sensitivity. A cost-benefit analysis confirmed that universal screening and preventive isolation were beneficial. CONCLUSIONS: The prevalence of MRSA carriage on admission to the ICU is high in this endemic setting. Screening for MRSA on admission is useful to identify the imported cases and should be performed in all ICU-admitted patients.

Preoperative chemotherapy followed by surgery compared with primary surgery in resectable stage I (except T1N0), II, and IIIa non-small-cell lung cancer.

PURPOSE: To evaluate whether preoperative chemotherapy (PCT) could improve survival in resectable stage I (except T1N0), II, and IIIA non-small-cell lung cancer (NSCLC). PATIENTS AND METHODS: A randomized trial compared PCT to primary surgery (PRS). PCT consisted of two cycles of mitomycin (6 mg/m(2), day 1), ifosfamide (1.5 g/m(2), days 1 to 3) and cisplatin (30 mg/m(2), days 1 to 3), and two additional postoperative cycles for responding patients. In both arms, patients with pT3 or pN2 disease received thoracic radiotherapy. RESULTS: Three hundred fifty-five eligible patients were randomized. Overall response to PCT was 64%. There were two preoperative toxic deaths. Postoperative mortality was 6.7% in the PCT arm and 4.5% in the PRS arm (P =.38). Median survival was 37 months (95% confidence interval [CI], 26.7 to 48.3) for PCT and 26.0 months (95% CI, 19.8 to 33.6) for PRS (P =.15). Survival differences between both arms increased from 3.8% (95% CI, 1.3% to 25.1%) at 1 year to 8.6% (95% CI, 2.64% to 24.4%) at 4 years. A quantitative interaction between N status and treatment was observed, with benefit confined to N0 to N1 disease (relative risk [RR], 0.68; 95% CI, 0.49 to 0.96; P =.027). After a nonsignificant excess of deaths during treatment, the effect of PCT was significantly favorable on survival (RR, 0.74; 95% CI, 0.56 to 0.99; P =.044). Disease-free survival time was significantly longer in the PCT arm (P =.033). CONCLUSION: Although impressive differences in median, 3-year, and 4-year survival were observed, they were not statistically significant, except for stage I and II disease.