Random Effects Modeling Approaches for Estimating ROC Curves from Repeated Ordinal Tests without a Gold Standard

Biometrics ◽  
2007 ◽  
Vol 63 (2) ◽  
pp. 593-602 ◽  
Author(s):  
Paul S. Albert
Keyword(s):  
Random Effects ◽  
Gold Standard ◽  
Roc Curves ◽  
Modeling Approaches ◽  
Random Effects Modeling ◽  
Without A Gold Standard

scholarly journals Screening for Intermediately Vancomycin-Susceptible and Vancomycin-Heteroresistant Staphylococcus aureus by Use of Vancomycin-Supplemented Brain Heart Infusion Agar Biplates: Defining Growth Interpretation Criteria Based on Gold Standard Confirmation

2015 ◽  
Vol 53 (11) ◽  
pp. 3543-3546 ◽  
Author(s):  
Riad Khatib ◽  
Kathleen Riederer ◽  
Mamta Sharma ◽  
Stephen Shemes ◽  
Sugantha P. Iyer ◽  
...  
Keyword(s):  
Staphylococcus Aureus ◽  
Gold Standard ◽  
Population Analysis ◽  
Area Under The Curve ◽  
Brain Heart Infusion ◽  
Roc Curves ◽  
Clinical Samples ◽  
Content Type ◽  
Interpretation Criteria ◽  
Profile Area

BHI agars supplemented with vancomycin 4 (BHI-V4) and 3 (BHI-V3) mg/liter have been proposed for screening vancomycin intermediately susceptibleStaphylococcus aureus(VISA) and heteroresistant (hVISA) phenotypes, respectively, but growth interpretation criteria have not been established. We reviewed the growth results (CFU) during population analysis profile-area under the curve (PAP-AUC) of consecutive methicillin-resistantStaphylococcus aureus(MRSA) blood isolates, which were saved intermittently between 1996 and 2012. CFU counts on BHI-V4 and BHI-V3 plates were stratified according to PAP-AUC interpretive criteria: <0.90 (susceptible [S-MRSA]), 0.90 to 1.3 (hVISA), and >1.3 (VISA). CFU cutoffs that best predict VISA and hVISA were determined with the use of receiver operating characteristic (ROC) curves. Mu3, Mu50, and methicillin-susceptibleS. aureus(MSSA) controls were included. We also prospectively evaluated manufacturer-made BHI-V3/BHI-V4 biplates for screening of 2010-2012 isolates. The PAP-AUC of 616 clinical samples was consistent with S-MRSA, hVISA, and VISA in 550 (89.3%), 48 (7.8%), and 18 (2.9%) instances, respectively. For VISA screening on BHI-V4, a cutoff of 2 CFU/droplet provided 100% sensitivity and 97.7% specificity. To distinguish VISA from hVISA, a cutoff of 16 CFU provided 83.3% sensitivity and 94.7% specificity; the specificity was lowered to 89.5% with a 12-CFU cutoff. For detecting hVISA/VISA on BHI-V3, a 2-CFU/droplet cutoff provided 98.5% sensitivity and 93.8% specificity. These results suggest that 2-CFU/droplet cutoffs on BHI-V4 and BHI-V3 best approximate VISA and hVISA gold standard confirmation, respectively, with minimal overlap in samples with borderline PAP-AUC. Simultaneous screening for VISA/hVISA on manufacturer-made BHI-V4/BHI-V3 biplates is easy to standardize and may reduce the requirement for PAP-AUC confirmation.

scholarly journals Problems in detecting misfit of latent class models in diagnostic research without a gold standard were shown

2016 ◽  
Vol 74 ◽  
pp. 158-166 ◽  
Author(s):  
Maarten van Smeden ◽  
Daniel L. Oberski ◽  
Johannes B. Reitsma ◽  
Jeroen K. Vermunt ◽  
Karel G.M. Moons ◽  
...  
Keyword(s):  
Gold Standard ◽  
Latent Class ◽  
Latent Class Models ◽  
Diagnostic Research ◽  
Class Models ◽  
Without A Gold Standard

Electrocautery Adenoidectomy Outcomes: A Metaanalysis

2008 ◽  
Vol 139 (2_suppl) ◽  
pp. P33-P34
Author(s):  
Jeremy T. Reed ◽  
Shankar K. Sridhara ◽  
Scott E Brietzke
Keyword(s):  
Sample Size ◽  
Random Effects ◽  
English Language ◽  
Operative Time ◽  
Inclusion Criteria ◽  
Subjective Success ◽  
Random Effects Modeling ◽  
Lost To Follow Up

Objective Review and assess the current published literature regarding clinical outcomes of suction electrocautery adenoidectomy (ECA) in pediatric patients. Methods The MEDLINE database was systematically reviewed for articles reporting on the use of ECA. Inclusion criteria included English language, sample size greater than 5, and presentation of extractable data regarding pediatric outcomes with ECA. Random effects modeling was used to estimate summary outcomes. Results 9 studies met the inclusion criteria. There were 2 level 1b studies, 2 level 3b studies, and 5 level 4 studies. The mean sample size was 276 patients with a grand mean age of 6.0 years. Random effects modeling of summary estimates of intra-operative hemorrhage (4.1 cc vs. 24.0 cc 95% CI of difference = 16.5–23.1, p<0.001) and operative time (10.0 minutes vs. 11.9 minutes 95% CI of difference=0.82–2.90, p<0.001) favored ECA vs. traditional curette adenoidectomy. Subjective success was reported in 95.0% (95% CI=92.7–97.3%, p<0.001) of ECA patients with a grand mean of 5.8 months of postoperative follow-up and a grand mean lost to follow-up rate of 23.2%. Adenoid regrowth was evaluated objectively (endoscopy or X-ray) in only 116 of 2,132 (5.4%) total patients with an observed regrowth rate of 2.8% (95% CI=0–5.5%, p=0.052) with 846 total person years of follow-up. Conclusions The preponderance of evidence favors ECA versus curette adenoidectomy in terms of decreased intraoperative hemorrhage and decreased operative time. Long-term outcomes data for ECA are scarce, despite the fact that the procedure is likely performed hundreds of times each day, but suggest a low regrowth rate.

scholarly journals Moment reconstruction and moment‐adjusted imputation when exposure is generated by a complex, nonlinear random effects modeling process

Biometrics ◽  
2016 ◽  
Vol 72 (4) ◽  
pp. 1369-1377
Author(s):  
Cornelis J. Potgieter ◽  
Rubin Wei ◽  
Victor Kipnis ◽  
Laurence S. Freedman ◽  
Raymond J. Carroll
Keyword(s):  
Random Effects ◽  
Modeling Process ◽  
Random Effects Modeling

scholarly journals Burnout in dental students: effectiveness of different methods

2013 ◽  
Vol 42 (5) ◽  
pp. 324-329 ◽  
Author(s):  
Juliana Alvares Duarte Bonini Campos ◽  
Paula Cristina Jordani ◽  
Miriane Lucindo Zucoloto ◽  
Fernanda Salloume Sampaio Bonafé ◽  
João Maroco
Keyword(s):  
Undergraduate Students ◽  
Gold Standard ◽  
Reliability And Validity ◽  
Roc Curves ◽  
Maslach Burnout Inventory ◽  
Dental Students ◽  
Burnout Syndrome ◽  
Screening Methods ◽  
Student Survey ◽  
Adequate Reliability

OBJECTIVE: This study was proposed to estimate the effectiveness of different screening methods of the Burnout Syndrome among dental students. MATERIAL AND METHOD: The Burnout Syndrome assessment was performed using the Oldenburg Inventory-Student survey (OLBI-SS) and the Copenhagen Inventory-Student survey (CBI‑SS). The Maslach Burnout Inventory-Student survey (MBI-SS) was used as the gold standard. The psychometric properties of the instruments were measured. The second-order hierarchical model was estimated to calculate the overall scores for OLBI-SS and CBI-SS, and ROC curves were constructed and the areas were estimated (AUROC). RESULT: A total of 235 undergraduate students participated in this study. The instruments showed an adequate reliability and validity; however three questions had to be removed from OLBI-SS and one from CBI-SS. The Exhaustion dimension of OLBI-SS, and Personal Burnout and Study related Burnout of CBI-SS presented a good discriminating capacity. CONCLUSION: CBI-SS showed higher discriminating capacity, than OLBI-SS, to identify the Burnout Syndrome (DAUROC=.172 [.103-.240]; p<.05).

Sign in / Sign up

Export Citation Format

Share Document