Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “diagnostics”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,063 records · Page 59Linked to original sources
PubMed2003

[VII: Diagnostic trials: Simple measures of validity and reliability].

The evaluation of a new diagnostic or imaging device has to focus both on its diagnostic validity and reliability. Cohen's kappa coefficient can be used to assess the agreement of the new device's findings with those of an established diagnostic reference, it therefore presents a surrogate measure of validity. Additional measures of validity are sensitivity and specificity, which assess the order of agreement with the reference for positive and negative reference findings, respectively. The latter, however, can only be applied for binary reference findings. If discordant findings are observed, the McNemar test can be used to detect a diagnostic shift between diagnostic novum and reference. The estimation of reliability can be based on the independent parallel rating of a patient subset by two clinical readers; again the kappa coefficient measures the agreement of the parallel ratings and therefore represents a surrogate measure of reliability. The McNemar test can be used to detect an observer bias.

Diagnostic Imaging↗
PubMed2026

Artificial intelligence in molecular diagnostics for pandemic preparedness.

INTRODUCTION: Molecular diagnostics focusing on the detection and analysis of nucleic acids are indispensable tools for early pathogen identification, transmission monitoring, and genomic surveillance during pandemics. Recent technological advances have broadened the diagnostic landscape, incorporating PCR-based methods, isothermal amplification, high-CRISPR-based amplification detection, and sequencing. Despite their diagnostic potential, widespread implementation remains limited by high validation costs, time and logistical constraints, the need for specialized professional knowledge, and a lack of adaptability in resource-limited settings. Artificial intelligence (AI) is increasingly recognized as a promising but challenging approach, offering tools that streamline assay development, automate data interpretation, and optimize real-time diagnostic performance. AREAS COVERED: This review introduces recently published AI tools with potential to enhance the in-silico design validation process of oligonucleotides for molecular assays. These cover tools for initial assay design and optimization to validation and continuous assay updates. The limitations, including concerns regarding data accuracy, the lack of transparency in data processing ('black box' models), and unresolved licensing and regulatory issues, are highlighted for each tool and as expert opinion. EXPERT OPINION: Collectively, these challenges currently confine most AI-based approaches to research settings and prevent their routine implementation in clinical molecular diagnostics. Their widespread adoption depends on addressing remaining technical, regulatory, and practical challenges.

Humans↗
PubMed2025

Targeted next-generation sequencing for drug-resistant tuberculosis diagnosis: implementation considerations for bacterial load, regimen selection and diagnostic algorithm placement.

INTRODUCTION: Early and accurate diagnosis of drug-resistant tuberculosis (DR-TB) is essential for improving treatment outcomes. Phenotypic drug susceptibility testing (pDST) is comprehensive but slow, while rapid molecular assays provide resistance information for a limited number of drugs. Targeted next-generation sequencing (tNGS) offers the potential for broad and rapid resistance detection, but its integration into diagnostic algorithms has been hindered by uncertainty about its placement within existing workflows. METHODS: This study evaluated the extent to which two tNGS solutions-Deeplex Myc-TB (GenoScreen) and TB Drug Resistance Test (Oxford Nanopore Technologies, ONT)-provided interpretable drug resistance results that could inform regimen design, in comparison to other WHO-recommended molecular assays and pDST. Data were collected from three high-burden DR-TB settings under the Seq&Treat study. Sequencing success rates and drug resistance detection were analysed based on: (1) the initial Xpert MTB/RIF result (very low, low, medium, high), (2) resistance results for drugs in WHO-recommended regimens and (3) performance relative to other WHO-endorsed assays. The potential impact of different algorithms on the estimates was also considered. Key factors influencing successful tNGS adoption within diagnostic pathways were identified, leveraging insights from the Seq&Treat diagnostic accuracy study. RESULTS: Sequencing success rates were 88.5% (GenoScreen) and 93.1% (ONT) across 763 samples. While tNGS provided complete resistance data for 73%-86% of drugs in recommended regimens, pDST achieved 92%-93%. Both tNGS solutions matched or exceeded the sensitivity of WHO-recommended molecular assays. CONCLUSIONS: This study highlights the critical role of tNGS as a centralised tool for comprehensive drug resistance testing to inform DR-TB treatment decisions following initial screening assays. By complementing existing molecular tests with tNGS, diagnostic workflows can be optimised to ensure timely and comprehensive resistance detection. These findings support policy updates to integrate tNGS into global TB diagnostic algorithms. TRIAL REGISTRATION NUMBER: NCT04239326.

Humans↗
PubMed1997

Trends in diagnostic imaging for low back pain: has MR imaging been a substitute or add-on?

PURPOSE: To assess whether magnetic resonance (MR) imaging has replaced other diagnostic imaging modalities in the evaluation of persistent low back pain. MATERIALS AND METHODS: Data on diagnostic imaging use in 2,374 adult patients with persistent low back pain in 1987-1990 were analyzed. Multiple logistic regression was used to estimate the annual adjusted odds of lumbosacral spine radiography, MR imaging, unenhanced computed tomography (CT), or CT myelography use. The national cost of diagnostic imaging for persistent low back pain in 1990 relative to 1987 was estimated. RESULTS: The adjusted odds of performing MR imaging in 1990 relative to 1987 was 3.44 (95% confidence interval, 2.63, 4.51), which reflects an estimated increase from 22 studies per 100 enrollees in 1987 to 75 studies per 100 enrollees in 1990. Use of MR imaging in combination with radiography, unenhanced CT, or CT myelography increased. The additional national cost of diagnostic imaging for persistent low back pain in 1990 relative to 1987 was estimated at $70-$176 million. CONCLUSION: MR imaging was used primarily as an add-on rather than a substitute for other imaging modalities in the evaluation of persistent low back pain. Thus, the volume and cost of diagnostic imaging for persistent low back pain have increased.

Adult↗
PubMed2026

Mapping the de-implementation of traditional diagnostic tests in pediatric acute lymphoblastic leukemia.

INTRODUCTION: Advances in cancer diagnostics raise questions about when and how to de-implement traditional approaches; however, these processes remain poorly described. At St. Jude Children's Research Hospital (SJCRH), routine conventional cytogenetics for pediatric acute lymphoblastic leukemia (ALL) diagnosis was de-implemented in 2018 following adoption of clinical genomics. This study aimed to map this process to inform future diagnostic de-implementation initiatives. METHODS: Interviews were conducted with SJCRH staff involved or impacted by cytogenetics de-implementation. Data were analyzed using thematic and rapid qualitative analysis informed by the Consolidated Framework for Implementation Research. Member-checking was used to verify and refine process maps, which were subsequently reviewed by an external expert panel, representing diverse settings, through focus group discussions. RESULTS: Thirteen SJCRH clinicians participated. De-implementation was described as successful, with no negative impact on patient outcomes. Decision-making began with internal correlation studies that demonstrated superior diagnostic performance of clinical genomics. De-implementation was viewed as a natural evolution that improved molecular classification, resource allocation, and workflow efficiency. Perceived risks included loss of cytogenetics competency, delayed turnaround time, and career insecurity, all addressed institutionally. Lessons learned highlighted the importance of deliberate discussion about logic and evidence supporting de-implementation. Fifteen external experts offered suggestions to improve process map generalizability, highlighting institutional- and system-level considerations. CONCLUSION: De-implementation of cytogenetics in ALL in favor of clinical genomics was successful at SJCRH. This study offers an example of diagnostic de-implementation in cancer care and proposes a structured approach to guide future efforts. De-implementation should be considered alongside introduction of novel diagnostic approaches.

cancer diagnostics↗
PubMed2005

[Molecular diagnostics in endocrine diseases].

Molecular diagnostic techniques provide an unsurpassed opportunity to understand the pathophysiological basis of endocrine disorders. Diseases have been associated with mutations in almost every gene known to have a role in either the production or secretion of a hormone or the mediators of hormone signalling. Even though most of these mutations are rare and account for only a small fraction of endocrine diseases, molecular diagnostics offers a valuable tool for the clinician in these cases. The most common endocrine disorders such as autoimmune thyroiditis, type 2 diabetes mellitus, osteoporosis, growth disorders, and obesity have all major genetic components, but these are mostly unknown. In this review the clinical implications of molecular diagnostics are illustrated for some endocrine diseases: congenital adrenal hyperplasia, congenital hypothyroidism, thyroid hormone resistance, familial hypocalciuric hypercalcaemia, growth hormone deficiency and resistance, and monogenic obesity. Improved diagnostic specificity has direct implications for treatment and follow up in these syndromes. Molecular diagnostics in endocrine tumours and diabetes are presented in two other articles in this series.

Adrenogenital Syndrome↗
PubMedDate not supplied

Use of methodological standards in diagnostic test research. Getting better but still not good.

OBJECTIVE: To determine the frequency and temporal changes in application of seven accepted methodological standards for the evaluation of diagnostic tests. DATA SOURCES: A search of the MEDLINE database yielded 1302 articles about diagnostic test studies, during a 16-year secular interval, 1978 through 1993, in four prominent general medical journals. STUDY SELECTION: In the 112 eligible studies, the test was intended for clinical use, indexes of accuracy (sensitivity and specificity or likelihood ratios) were provided, and more than 10 patients were enrolled. DATA EXTRACTION: Although each study was critically reviewed by one primary observer, a subset was independently evaluated for interrater consistency. DATA SYNTHESIS: The percentage of studies that fulfilled criteria for each of the seven methodological standards are as follows: (1) specify spectrum of evaluated patients, 27%; (2) report test indexes for clinical subgroups, 8%; (3) avoid workup bias, 46%; (4) avoid review bias, 38%; (5) provide numerical precision for test indexes, 11%; (6) report frequency and management of indeterminate results when calculating test indexes, 22%; and (7) specify test reproducibility, 23%. Secular increases were found for six of the seven standards in ranges of use from 14% to 31% during 1978-1981 to 1990-1993. Nevertheless, only one standard, avoidance of workup bias, was fulfilled by more than 50% of studies in the most recent secular interval. CONCLUSIONS: These results indicate that most diagnostic tests are still inadequately appraised. The routine demand for methodological standards could raise the quality of diagnostic test information, and the careful predissemination evaluation of diagnostic tests could eliminate useless tests before they receive widespread application.

Bias↗
PubMed2005

Diagnostic error in internal medicine.

BACKGROUND: The goal of this study was to determine the relative contribution of system-related and cognitive components to diagnostic error and to develop a comprehensive working taxonomy. METHODS: One hundred cases of diagnostic error involving internists were identified through autopsy discrepancies, quality assurance activities, and voluntary reports. Each case was evaluated to identify system-related and cognitive factors underlying error using record reviews and, if possible, provider interviews. RESULTS: Ninety cases involved injury, including 33 deaths. The underlying contributions to error fell into 3 natural categories: "no fault," system-related, and cognitive. Seven cases reflected no-fault errors alone. In the remaining 93 cases, we identified 548 different system-related or cognitive factors (5.9 per case). System-related factors contributed to the diagnostic error in 65% of the cases and cognitive factors in 74%. The most common system-related factors involved problems with policies and procedures, inefficient processes, teamwork, and communication. The most common cognitive problems involved faulty synthesis. Premature closure, ie, the failure to continue considering reasonable alternatives after an initial diagnosis was reached, was the single most common cause. Other common causes included faulty context generation, misjudging the salience of findings, faulty perception, and errors arising from the use of heuristics. Faulty or inadequate knowledge was uncommon. CONCLUSIONS: Diagnostic error is commonly multifactorial in origin, typically involving both system-related and cognitive factors. The results identify the dominant problems that should be targeted for additional research and early reduction; they also further the development of a comprehensive taxonomy for classifying diagnostic errors.

Australia↗
PubMed1996

Variability in physicians' reported ordering and perceived reassurance value of diagnostic tests in children with 'growing pains'.

OBJECTIVES: To determine the variability in the reported ordering of tests and treatment and to determine physicians' perceptions of the reassurance value to families of diagnostic tests in children with "growing pains." DESIGN: Cross-sectional survey using a mailed questionnaire. SETTINGS: Primary care and referral practices in Toronto, Ontario. PARTICIPANTS: University-affiliated primary care pediatricians and family physicians were surveyed, as well as all pediatric orthopedic surgeons and pediatric rheumatologists in Ontario. Pediatric orthopedic surgeons and pediatric rheumatologists were combined into a single group. MAIN OUTCOME MEASURES: Frequency of office visits because of growing pains, frequency of diagnostic testing, management strategies for these children, and physicians' perceptions of the reassurance value of diagnostic tests. RESULTS: Of 205 eligible physicians, 181 (88.3%) responded. The median reported frequency of office visits because of growing pains was 1%. Compared with the other physician groups, family physicians were significantly more likely to order a determination of the hemoglobin level (P = .003), erythrocyte sedimentation rate (P = .01), white blood cell count (P = .01), and differential blood cell count (P = .003), but not imaging tests. Family physicians were also more likely to order diagnostic tests when they were under parental pressure to do so (P = .001) or for the child with repeated visits (P = .02). In total, 86% of pediatric orthopedic surgeons and pediatric rheumatologists, 95% of pediatricians, and 100% of family physicians perceived normal test results to be reassuring to parents. Treatment strategies were similar across the 3 physician groups. CONCLUSIONS: The frequency of diagnostic testing varied among physician groups. Virtually all physicians perceived normal test results to be reassuring to families.

Adult↗
PubMed1997

Evaluation of diagnostic peritoneal lavage in stable patients with gunshot wounds to the abdomen.

BACKGROUND: Diagnostic peritoneal lavage (DPL) is used to diagnose intra-abdominal injury in patients with stab wounds and blunt trauma. Because exploratory celiotomy is routinely performed on patients with gunshot wounds to the abdomen, DPL is rarely employed. However, several studies have questioned routine exploration and have drawn attention to the associated morbidity of negative celiotomy. Diagnostic peritoneal lavage is an easily performed and inexpensive test that may be useful in this situation. OBJECTIVE: To evaluate the performance of DPL in the diagnosis of intra-abdominal injury in hemodynamically stable patients with gunshot wounds to the abdomen. DESIGN: A prospective clinical trial. SETTING: Two urban trauma centers. PATIENTS: Patients with gunshot wounds to the abdomen and a systolic blood pressure of at least 90 mm Hg. INTERVENTIONS: Clinical predication of intra-abdominal injury in the emergency department and DPL performed in the operating room before the initiation of celiotomy. Injuries found during the celiotomy were recorded. MAIN OUTCOME MEASURES: The results of the clinical evaluation and DPL were compared with the findings of the celiotomy. RESULTS: Forty-four patients were enrolled into the study. Intra-abdominal injury was present in 32 (73%) of these patients. The senior surgery resident correctly predicted the presence of intra-abdominal injury in 36 (82%) of the patients (sensitivity = 90.0%, specificity = 58.3%, positive predictive value = 85.3%, negative predictive value = 63.6%, phi = 0.52, P < .01) in the emergency department before DPL and celiotomy were performed. Diagnostic peritoneal lavage correctly identified the presence or absence of intra-abdominal injury in 40 (91%) of the patients (positive predictive value = 96.7%, negative predictive value = 78.6%, phi = 0.79, P < .01). CONCLUSIONS: Clinical judgment is highly accurate in separating patients with tangential gunshot wounds to the abdomen from those with intra-abdominal injury but may miss patients with intra-abdominal hemorrhage. Diagnostic peritoneal lavage is highly predictive of the presence of intra-abdominal injury. The return of gross blood on aspiration or a lavage red blood cell count greater than 10 x 10(9)/L should prompt an urgent celiotomy. Missed injuries are rare and most likely to be bowel perforations. Diagnostic peritoneal lavage is an objective test that may augment clinical judgment in selecting hemodynamically stable patients with potential tangential gunshot wounds for observation and is especially useful in identifying intra-abdominal hemorrhage.

Abdomen↗
PubMed2000

Statistical methods for the evaluation of diagnostic measurements concerning paired organs.

Diagnostic measurements involving paired organs may be modelled using marginal or conditional approaches. Relevant information could be ignored by using marginal models with a mean structure not including information from the fellow organ. The conditional Rosner model can be generalized to a symmetric polychotomous logistic regression model. The rather strong assumptions of the Rosner model are relaxed in this symmetric model. The use of the symmetric model is justified by its relation to discriminant analysis. Additionally, the information of the diagnostic measurements from the fellow organ is taken into account explicitly. Furthermore, in case-control studies this approach corresponds to the sampling scheme. When the status of the fellow organ and the diagnostic measurement of this organ both are available, a surprising effect may be observed; pathological values of the diagnostic measurement from the fellow organ decrease the disease probability of the selected organ. As an example, four different diagnostic procedures from the Erlangen Glaucoma Registry are investigated.

Contrast Sensitivity↗
PubMedDate not supplied

Willingness to pay for diagnostic services. A new approach to modelling patient benefits in health care.

The paper investigates the benefit the patient derives from medical diagnosis. By considering explicitly the prospects with respect to both health and monetary consequences resulting from a decision taken by the physician, a fairly general approach to discuss diagnostic services is developed. The willingness to pay of the patient is taken to be measured by his compensating option price, evaluated with respect to the reference state without further diagnostic information. Of particular interest are conditions governing positivity of the patient benefit. Imposing additional restrictions upon individual preferences considerably simplifies the analysis by relying on a loss function. The final section discusses the role of the patient benefit as regards cost-benefit analysis of diagnostic services. If health insurance is available providing at least partial coverage, a positive willingness to pay of the patient net of diagnostic cost can be shown to give no clue as to whether utilization of a diagnostic service is beneficial to society in the sense of cost-benefit analysis.

Attitude to Health↗
PubMed2002

A new parametric method based on S-distributions for computing receiver operating characteristic curves for continuous diagnostic tests.

Receiver operating characteristic (ROC) curves provides a method for evaluating the performance of a diagnostic test. These curves represent the true positive ratio, that is, the true positives among those affected by the disease, as a function of the false positive ratio, that is, the false positives among the healthy, corresponding to each possible value of the diagnostic variable. When the diagnostic variable is continuous, the corresponding ROC curve is also continuous. However, estimation of such curve through the analysis of sample data yields a step-line, unless some assumption is made on the underlying distribution of the considered variable. Since the actual distribution of the diagnostic test is seldom known, it is difficult to select an appropriate distribution for practical use. Data transformation may offer a solution but also may introduce a distortion on the evaluation of the diagnostic test. In this paper we show that the distribution family known as the S-distribution can be used to solve this problem. The S-distribution is defined as a differential equation in which the dependent variable is the cumulative. This special form provides a highly flexible family of distributions that can be used as models for unknown distributions. It has been shown that classical statistical distributions can be represented accurately as S-distributions and that they occur in a definite subspace of the parameter space corresponding to the whole S-distribution family. Consequently, many other distributional forms that do not correspond to known distributions are provided by the S-distribution. This property can be used to model observed data for unknown distributions and is very useful in constructing parametric ROC curves in those cases. After fitting an S-distribution to the observed samples of diseased and healthy populations, ROC curve computation is straightforward. A ROC curve can be considered as the solution of a differential equation in which the dependent variable is the ratio of true positives and the independent variable is the ratio of false positives. This equation can be easily obtained from the S-distributions fitted to observed data. Using these results, we can compute pointwise confidence bands for the ROC curve and the corresponding area under the curve. We shall compare this approach with the empirical and the binormal methods for estimating a ROC curve to show that the S-distribution based method is a useful parametric procedure.

Area Under Curve↗
PubMedDate not supplied

Cost-effective diagnostic test sequencing.

Proper diagnostic test selection and logical sequencing can minimize cost without compromising care. This study analyzes the logic for sequencing tests either in series (one after another) or in parallel (simultaneously). A model is created using 2 diagnostic tests for 2 diseases. Tests are assumed to have perfect performance characteristics. Factors involved in cost-effective test sequencing include cost of each test, test performance characteristics, per diem hospital charge, and the clinician's prior probability of the suspected diagnosis based on history, physical examination, and previous laboratory data. At a given point of clinical suspicion, cost-effective sequencing strategy should shift. In general, a cheap test can effectively be used in parallel even at low diagnostic probabilities; an expensive test should, in general, be used in series only when there is a high prior probability of diagnosis. High per diem hospitalization costs--such as for a patient in the intensive care unit--favor in parallel testing. Clinical acumen and suspicion of diagnosis (prior probability) primarily direct cost-effective diagnostic test sequencing. Decision analysis of diagnostic test sequencing represents a technique for making cost-effective decisions for sequencing and, thereby, minimizing costs while achieving optimal patient care.

Cost-Benefit Analysis↗
PubMed1999

[The operational diagnostic approach of the borderline personality disorder. An interview of psychiatrists and psychologists according to their concept of the syndrome].

Despite the diagnostic criteria of the ICD-10 and DSM-IV, the term borderline has been used in a non-specific way for a long time. For our investigation we constructed a questionnaire, which contained the ICD-10 diagnostic criteria for the borderline personality disorder (BPD), plus the criteria for the other personality disorders and for the schizotype disorder. The two additional criteria for the BPD used by the DSM-IV were added. 800 psychiatrists and psychologists were asked to mark all criteria, which they thought were typical for the BPD. By using the operational diagnostic approach for ICD-10 and DSM-IV a disorder was diagnosed from each of the 162 received questionnaires based on the marked criteria. 92.0% of the questionnaires contained marked criteria for the diagnosis of the BPD by ICD-10 and 96.3% by DSM-IV. In addition each questionnaire contained up to eight other diagnosis. This is a sign for the difficulty to separate the BPD from other personality disorders or the schizotype disorder by using operationalized criteria. Looking at the lack of specificity of the operational diagnostic systems for personality disorders, e.g. for the BPD which has been discussed by many authors for years, modification of the diagnostic systems should be considered.

Adult↗
PubMed2003

A readers' guide to the interpretation of diagnostic test properties: clinical example of sepsis.

BACKGROUND: One of the most challenging practical and daily problems in intensive care medicine is the interpretation of the results from diagnostic tests. In neonatology and pediatric intensive care the early diagnosis of potentially life-threatening infections is a particularly important issue. FOCUS: A plethora of tests have been suggested to improve diagnostic decision making in the clinical setting of infection which is a clinical example used in this article. Several criteria that are critical to evidence-based appraisal of published data are often not adhered to during the study or in reporting. To enhance the critical appraisal on articles on diagnostic tests we discuss various measures of test accuracy: sensitivity, specificity, receiver operating characteristic curves, positive and negative predictive values, likelihood ratios, pretest probability, posttest probability, and diagnostic odds ratio. CONCLUSIONS: We suggest the following minimal requirements for reporting on the diagnostic accuracy of tests: a plot of the raw data, multilevel likelihood ratios, the area under the receiver operating characteristic curve, and the cutoff yielding the highest discriminative ability. For critical appraisal it is mandatory to report confidence intervals for each of these measures. Moreover, to allow comparison to the readers' patient population authors should provide data on study population characteristics, in particular on the spectrum of diseases and illness severity.

Bayes Theorem↗
PubMed2001

Diagnostic and therapeutic management of spinal arachnoid cysts.

BACKGROUND: The wide variety of intraspinal cystic lesions necessitates different elaborate diagnostic procedures to choose the right therapeutic management in symptomatic patients. Based on the case reports of seven patients with symptomatic spinal arachnoid cysts we discuss the aetiology, diagnostic procedures and therapeutic management of extra- and intradural spinal cysts. METHOD: All patients underwent MRI, Myelography and CT-Myelography during diagnostic evaluation. During surgery the cyst was resected and the communication between the cyst and the subarachnoid space was closed. FINDINGS: Two patients were identified with intradural, five with extradural spinal arachnoid cysts. Postoperative outcome was favourable in those patients without preoperative cord damage. INTERPRETATION: MRI is the diagnostic procedure of first choice because of its potential to demonstrate the exact localisation, extent and relationship of the arachnoid cyst to the spinal cord. Cord atrophy secondary to compression can be visualised and used for prediction of neurological outcome. Myelography and CT-Myelography (CTM) are still of diagnostic value since they might demonstrate the communication between the subarachnoid space and the cyst, which is important for surgical planning. The aim of surgical treatment is neural decompression and prevention of refilling of the cyst which is best accomplished by complete resection of the cyst and closure of the communication between cyst and subarachnoid space.

Adult↗
PubMed2005

Estimating and comparing diagnostic tests' accuracy when the gold standard is not binary.

RATIONALE AND OBJECTIVES: Investigators often need to assess the accuracies of diagnostic tests when the gold standard is not binary-scale. The objective of this article is to describe nonparametric estimators of diagnostic test accuracy when the gold standard is continuous, ordinal, and nominal scale. MATERIALS AND METHODS: A nonparametric method of estimating and comparing the area under receiver operating characteristic (ROC) curves, proposed by DeLong et al, is extended to situations in which the gold standard is not binary. Two examples illustrate the methods. RESULTS: Measures of diagnostic test accuracy, their variance, and tests for comparing two diagnostic tests' accuracies in paired designs are presented for situations in which the gold standard is continuous, ordinal, and nominal scale. These summary measures of diagnostic test accuracy are analogous in form and interpretation to the area under the ROC curve. CONCLUSION: Dichotomizing the outcomes of a gold standard so that traditional ROC methods can be applied can lead to bias. The methods described here are useful for assessing and comparing summary test accuracy when the gold standard is not binary scale. They have limitations similar to other summary indices.

Diagnostic Imaging↗