Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “CLASSIFICATION”

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 1,027 records · Page 57Linked to original sources
PubMed2002

Primary cutaneous lymphomas: applicability of current classification schemes (European Organization for Research and Treatment of Cancer, World Health Organization) based on clinicopathologic features observed in a large group of patients.

Classification of primary cutaneous lymphomas (PCLs) is the subject of ongoing controversy. Based on a series of 556 patients, the applicability of the European Organization for Research and Treatment of Cancer (EORTC) classification for PCLs was assessed and compared to the proposed World Health Organization (WHO) classification of hematologic malignancies. The large majority of patients could be properly classified according to the scheme proposed by the EORTC. Comparison of estimated 5-year survival for specific diagnostic categories of PCLs demonstrated nearly complete concordance of the present results with those of the EORTC study for most of the indolent cutaneous T-cell lymphomas and cutaneous B-cell lymphomas, whereas differences were found for mycosis fungoides-associated follicular mucinosis and Sezary syndrome. A few patients with newly described entities (CD8(+) epidermotropic cytotoxic T-cell lymphoma, primary cutaneous natural killer/T-cell lymphoma) could not be classified according to the EORTC scheme. Comparison of the EORTC with the WHO classification showed that the EORTC scheme allows a more precise categorization of the patients, especially for cutaneous B-cell lymphoma. In conclusion, the study confirmed that the EORTC classification allows a better management of patients with PCL. Small amendments to that classification should be carried out to account for recently described entities and to unify some of the diagnostic categories.

Adolescent↗
PubMed2004

Boosting accuracy of automated classification of fluorescence microscope images for location proteomics.

BACKGROUND: Detailed knowledge of the subcellular location of each expressed protein is critical to a full understanding of its function. Fluorescence microscopy, in combination with methods for fluorescent tagging, is the most suitable current method for proteome-wide determination of subcellular location. Previous work has shown that neural network classifiers can distinguish all major protein subcellular location patterns in both 2D and 3D fluorescence microscope images. Building on these results, we evaluate here new classifiers and features to improve the recognition of protein subcellular location patterns in both 2D and 3D fluorescence microscope images. RESULTS: We report here a thorough comparison of the performance on this problem of eight different state-of-the-art classification methods, including neural networks, support vector machines with linear, polynomial, radial basis, and exponential radial basis kernel functions, and ensemble methods such as AdaBoost, Bagging, and Mixtures-of-Experts. Ten-fold cross validation was used to evaluate each classifier with various parameters on different Subcellular Location Feature sets representing both 2D and 3D fluorescence microscope images, including new feature sets incorporating features derived from Gabor and Daubechies wavelet transforms. After optimal parameters were chosen for each of the eight classifiers, optimal majority-voting ensemble classifiers were formed for each feature set. Comparison of results for each image for all eight classifiers permits estimation of the lower bound classification error rate for each subcellular pattern, which we interpret to reflect the fraction of cells whose patterns are distorted by mitosis, cell death or acquisition errors. Overall, we obtained statistically significant improvements in classification accuracy over the best previously published results, with the overall error rate being reduced by one-third to one-half and with the average accuracy for single 2D images being higher than 90% for the first time. In particular, the classification accuracy for the easily confused endomembrane compartments (endoplasmic reticulum, Golgi, endosomes, lysosomes) was improved by 5-15%. We achieved further improvements when classification was conducted on image sets rather than on individual cell images. CONCLUSIONS: The availability of accurate, fast, automated classification systems for protein location patterns in conjunction with high throughput fluorescence microscope imaging techniques enables a new subfield of proteomics, location proteomics. The accuracy and sensitivity of this approach represents an important alternative to low-resolution assignments by curation or sequence-based prediction.

Cell Line, Tumor↗
PubMed2006

Gene selection and classification of microarray data using random forest.

BACKGROUND: Selection of relevant genes for sample classification is a common task in most gene expression studies, where researchers try to identify the smallest possible set of genes that can still achieve good predictive performance (for instance, for future use with diagnostic purposes in clinical practice). Many gene selection approaches use univariate (gene-by-gene) rankings of gene relevance and arbitrary thresholds to select the number of genes, can only be applied to two-class problems, and use gene selection ranking criteria unrelated to the classification algorithm. In contrast, random forest is a classification algorithm well suited for microarray data: it shows excellent performance even when most predictive variables are noise, can be used when the number of variables is much larger than the number of observations and in problems involving more than two classes, and returns measures of variable importance. Thus, it is important to understand the performance of random forest with microarray data and its possible use for gene selection. RESULTS: We investigate the use of random forest for classification of microarray data (including multi-class problems) and propose a new method of gene selection in classification problems based on random forest. Using simulated and nine microarray data sets we show that random forest has comparable performance to other classification methods, including DLDA, KNN, and SVM, and that the new gene selection procedure yields very small sets of genes (often smaller than alternative methods) while preserving predictive accuracy. CONCLUSION: Because of its performance and features, random forest and gene selection using random forest should probably become part of the "standard tool-box" of methods for class prediction and gene selection with microarray data.

Algorithms↗
PubMed2006

DNA molecule classification using feature primitives.

BACKGROUND: We present a novel strategy for classification of DNA molecules using measurements from an alpha-Hemolysin channel detector. The proposed approach provides excellent classification performance for five different DNA hairpins that differ in only one base-pair. For multi-class DNA classification problems, practitioners usually adopt approaches that use decision trees consisting of binary classifiers. Finding the best tree topology requires exploring all possible tree topologies and is computationally prohibitive. We propose a computational framework based on feature primitives that eliminates the need of a decision tree of binary classifiers. In the first phase, we generate a pool of weak features from nanopore blockade current measurements by using HMM analysis, principal component analysis and various wavelet filters. In the next phase, feature selection is performed using AdaBoost. AdaBoost provides an ensemble of weak learners of various types learned from feature primitives. RESULTS AND CONCLUSION: We show that our technique, despite its inherent simplicity, provides a performance comparable to recent multi-class DNA molecule classification results. Unlike the approach presented by Winters-Hilt et al., where weaker data is dropped to obtain better classification, the proposed approach provides comparable classification accuracy without any need for rejection of weak data. A weakness of this approach, on the other hand, is the very "hands-on" tuning and feature selection that is required to obtain good generalization. Simply put, this method obtains a more informed set of features and provides better results for that reason. The strength of this approach appears to be in its ability to identify strong features, an area where further results are actively being sought.

Computational Biology↗
PubMed2006

Combined applications of fine needle aspiration cytology and flow cytometric immunphenotyping for diagnosis and classification of non Hodgkin lymphoma.

AIMS AND OBJECTIVES: In this present study we have evaluated the feasibility of sub-classification of non-Hodgkin's lymphoma (NHL) cases according to World Health Organization's (WHO) classification on fine needle aspiration cytology (FNAC) material along with flow cytometric immunotyping (FCI) as an adjunct. MATERIALS AND METHODS: In this five years study, only cases suggested or confirmed as NHL by FNAC were selected and FCI was performed with a complete panel of antibodies (CD3, CD2, CD 4, CD5, CD8, CD7, CD10, CD19, CD20, CD23, CD45, kappa and lambda) by dual color flow cytometry. Both cytologic findings and FCI data were interpreted together to diagnose and sub-classify NHL according to WHO classification. Wherever possible the diagnoses were compared with cytology. RESULTS: There were total 48 cases included in this study. The cases were classified on FNAC as predominant small cells (12), mixed small and large cells (5) and large cells (26). In five cases a suggestion of NHL was offered on FNAC material and these cases were labeled as NHL not otherwise specified (NHL-NOS). Flow cytometry could be performed in 45 cases (93.8%) and in rest of the three cases the material was inadequate because of scanty blood mixed aspirate. Light chain restriction was demonstrated in 30 cases out of 40 cases of B-NHL (75%). There were 15 cases each of kappa and lambda light chain restriction in these 30 cases. With the help of combined FCI and FNAC, it was possible to sub-classify 38 cases of NHL (79%) according to WHO classification. Combined FNAC and FCI data helped to diagnose 9 cases of small lymphocytic lymphoma (SLL), 2 cases of mantle cell lymphoma (MCL), 4 cases of follicular lymphoma (FL), 17 cases of diffuse large B lymphoma (DLBL) and 6 cases of lymphoblastic lymphoma. Histopathology diagnosis was available in 31 cases of NHL out of which there were 14 recurrent and 17 cases of primary NHL. Out of 15 DLBL cases diagnosed on FCI and FNAC, histology confirmed 14 cases and one of these cases was diagnosed as Burkitt's lymphoma on histology. Cases of FL (4), SLL (3) and MCL (2) were well correlated with histopathology. Out of the five cases suggestive of NHL on cytology, histopathology was available in four cases. Histology diagnosis was given as DLBL (1), SLL (1), anaplastic large cell lymphoma (1) and FL transformed into large cell NHL (1). Considering histopathology as gold standard, diagnostic specificity of combined FNAC and FCI was 100% (31/31) and sensitivity in sub-classification was 83.8% (26/31). CONCLUSION: FNAC combined with FCI may be helpful in accurately sub-classifying NHL according to WHO classification. Many of the subtypes of NHL such as FL and MCL which were previously recognized as a pure morphologic entity can be diagnosed by combined use of FNAC and FCI. Other ancillary investigations such as chromosomal changes, cell proliferation markers etc. may be helpful in this aspect.

Journal Article↗
PubMed2006

Three-dimensional easy morphological (3-DEMO) classification of scoliosis, part I.

BACKGROUND: While scoliosis has, for a long time, been defined as a three-dimensional (3D) deformity, morphological classifications are confined to the two dimensions of radiographic assessments. The actually existing 3-D classification proposals have been developed in research laboratories and appear difficult to be understood by clinicians. AIM OF THE STUDY: The aim of this study was to use the results of a 3D evaluation to obtain a simple and clinically oriented morphological classification (3-DEMO) that might make it possible to distinguish among different populations of scoliotic patients. METHOD: We used a large database of evaluations obtained through an optoelectronic system (AUSCAN) that gives a 3D reconstruction of the spine. The horizontal view was used, with a spinal reference system (Top View). An expert clinician evaluated the morphological reconstruction of 149 pathological spines in order to find parameters that could be used for classificatory ends. These were verified in a mathematical way and through computer simulations: some parameters had to be excluded. Pathological data were compared with those of 20 normal volunteers. RESULTS: We found three classificatory parameters, which are fully described and discussed in this paper: Direction, the angle between spinal pathological and normal AP axis; Shift, the co-ordinates of the barycentre of the Top View ; Phase, the parameter describing the spatial evolution of the curve. Using these parameters it was possible to distinguish normal and pathological spines, to classify our population and to differentiate scoliotic patients with identical AP classification but different 3D behaviors. CONCLUSION: The 3-DEMO classification offers a new and simple way of viewing the spine through an auxiliary plane using a spinal reference system. Further studies are currently under way to compare this new system with the existing 3-D classifications, to obtain it using everyday clinical and x-rays data, and to develop a triage for clinical use.

Journal Article↗
PubMed2006

The three-dimensional easy morphological (3-DEMO) classification of scoliosis, part II: Repeatability.

BACKGROUND: In the first part of this study we proposed a new classification approach for spinal deformities (3-DEMO). To be valid, a classification needs to overcome the repeatability issue which is inherent both in the used classificatory system and in the measured object. AIM: The aim of this study is to present procedures and results obtained within the repeatability of 3-DEMO classification for scoliosis analysis. METHOD: We acquired the data of 100 pathological and 20 normal spines with an optoelectronic system (AUSCAN) and of two dummies with simulated spine deformity. On the obtained 3D reconstruction of the spine, we considered the coronal view with a spinal reference system (Top View) and its three related parameters, defined in part I, constituting the 3-DEMO classification. We calculated the repeatability coefficient for the subjects (two acquisitions for each subject with a time interval of 26 +/- 12 sec), whereas we evaluated the system measurement error calculating the standard deviation of 50 consecutive acquisitions for each dummy. RESULTS: Comparing the results of the two types of acquisition, it emerged that the main part of parameters variability was due to postural adjustments The proportion of agreement for the 3-DEMO parameters gives a k value above 0.8; almost 10% of patients changed classification because of postural adjustments, but none had a "mirror-like" variation nor a change in more of one parameter at a time Repeatability coefficient is lower than the previously calculated normative limits. DISCUSSION: The 3-DEMO classification has a high repeatability when evaluated with an optoelectronic system such as the AUSCAN System, whose systematic error is very low. This means that the implied physiological phenomenon is consistent and overcomes the postural variability inherent in the measured object (normal or pathological subject).

Journal Article↗
PubMed2004

An Australian casemix classification for palliative care: technical development and results.

OBJECTIVES: To develop a palliative care casemix classification for use in all settings including hospital, hospice and home-based care. SAMPLE: 3866 palliative care patients who, in a three-month period, had 4596 episodes of care provided by 58 palliative care services in Australia and New Zealand. METHOD: A detailed clinical and service utilization profile was collected on each patient with staff time and other resources measured on a daily basis. Each day of care was costed using actual cost data from each study site. Regression tree analysis was used to group episodes of care with similar costs and clinical characteristics. RESULTS: In the resulting classification, the Australian National Sub-acute and Non-acute Patient (AN-SNAP) Classification Version 1, the branch for classifying inpatient palliative care episodes (including hospice care) has 11 classes and explains 20.98% of the variance in inpatient palliative care phase costs using trimmed data. There are 22 classes in the ambulatory palliative care branch that explains 17.14% variation in ambulatory phase cost using trimmed data. DISCUSSION: The term 'subacute' is used in Australia to describe health care in which the goal--a change in functional status or improvement in quality of life--is a better predictor of the need for, and the cost of, care than the patient's underlying diagnosis. The results suggest that phase of care (stage of illness) is the best predictor of the cost of Australian palliative care. Other predictors of cost are functional status and age. In the ambulatory setting, symptom severity and the model of palliative care are also predictive of cost. These variables are used in the AN-SNAP Version 1 classification to create 33 palliative care classes. The classification has clinical meaning but the overall statistical performance is only moderate. The structure of the classification allows for it to be improved over time as models of palliative care service delivery develop.

Ambulatory Care↗
PubMed2001

Classification system for malformations of cortical development: update 2001.

The many recent discoveries concerning the molecular biologic bases of malformations of cortical development and the discovery of new such malformations have rendered previous classifications out of date. A revised classification of malformations of cortical development is proposed, based on the stage of development (cell proliferation, neuronal migration, cortical organization) at which cortical development was first affected. The categories have been created based on known developmental steps, known pathologic features, known genetics (when possible), and, when necessary, neuroimaging features. In many cases, the precise developmental and genetic features are uncertain, so classification was made based on known relationships among the genetics, pathologic features, and neuroimaging features. A major change since the prior classification has been the elimination of the separation between diffuse and focal/multifocal malformations, based on the recognition that the processes involved in these processes are not fundamentally different; the difference may merely reflect mosaicism, X inactivation, the influence of modifying genes, or suboptimal imaging. Another change is the listing of fewer specific disorders to reduce the need for revisions; more detail is added in other smaller tables that list specific malformations and malformation syndromes. This classification is useful to the practicing physician in that its framework allows a better conceptual understanding of the disorders, while the component of neuroimaging characteristics allows it to be applied to all patients without necessitating brain biopsy, as in pathology-based classifications.

Brain↗
PubMed2006

A new method for the classification of invasive cervical cancer screening histories.

OBJECTIVES: To examine the ability of existing classification systems to provide screening histories for invasive cervical cancers which can be used in the evaluation of the NHS Cervical Screening Programme (NHSCSP), and to provide the diagnostic route data item required for the National Cancer Data Set (NCDS). METHODS: The ability of existing classification systems to derive unique, consistent screening histories for a cohort of invasive cervical cancers diagnosed in the West Midlands region in the period 2000-03 was tested using two separate timelines for women on normal routine recall (usually 3 or 5 years) and those on early recall having had an inadequate, low-grade abnormal or negative smear. RESULTS: Neither of the existing classification systems was capable of adequately categorizing all invasive cervical cancers. An original classification system incorporating features from the existing systems was therefore developed. This system includes both a 'screening status' component that essentially describes the status of a woman's interaction with the NHSCSP at the time her cancer was diagnosed, and a 'screening history' component that describes the results of previous screening tests. CONCLUSIONS: National adoption of this new screening histories classification system would provide a detailed, consistent, nationally comparable screening history for all invasive cervical cancers which can be used in the national and regional evaluation of the NHSCSP and in local audit by clinical teams supplemented by histology and colposcopy data. The classification categories could be collapsed down to provide the diagnostic route data item required for the NCDS.

Adult↗
PubMed2001

Classification in anatomic pathology.

Classification is the activity that allows pathologists to arrange the bewildering morphologic manifestations of disease into comprehensible order Ideally, our precise diagnoses would each be based on some timeless biological law of nature, and these diagnoses would group together patients with identical clinical manifestations and responses to therapy. However, in spite of the amazing success of pathology classification, it is apparent to everyone that diagnostic disagreements are common, and that patients who exactly fit into a diagnostic category often have markedly different disease courses and responses to therapy. At this time, it is unclear how quickly the potential of genomic medicine will be translated into revolutionary changes in pathology classification. However, it is clear that the rules of classification are part of the foundation of diagnostic pathology and that there is a high likelihood that pathology classification will undergo substantial changes in the next few years. This article reviews the topic of classification for pathologists who will practice during these interesting times.

Cytodiagnosis↗
PubMed2003

[Sensitivity and specificity of overweight classification of adolescents, Brazil].

OBJECTIVE: To evaluate the prevalence, sensitivity and specificity of two risk classifications of obesity based on the body mass index (BMI). METHODS: Five-hundred and two adolescents, aged 12-18 years, participants of a health and nutrition survey conducted in 1996 in the city of Rio de Janeiro, Brazil, were evaluated. The study variables included: weight, stature, BMI, and subscapular skinfold, according to sex and age. The BMI classifications were compared to a fatness classification based on the 90th percentile of subscapular skinfold thickness in American adolescents. RESULTS: The prevalence of risk of obesity was higher when using the subscapular skinfold measurement (p<0.0001), compared to both BMI-based classifications, which showed similar values. Specificity was higher than sensitivity in both BMI-based classifications. The balance between sensitivity and specificity was close to the 70thBMI percentile for boys and girls below 14 years old. For boys older than 15 the cutoff value was the 50th percentile. CONCLUSION: Both BMI-based classifications were more suitable to identify adolescents without obesity, and their sensitivities were too low for tracking risk of obesity.

Adolescent↗
PubMed2004

Prognostic factors in differentiated thyroid carcinomas and their implications for current staging classifications.

Differentiated thyroid carcinomas (DTC) (papillary, follicular and follicular type of papillary) have a favourable prognosis, but a proportion of patients develop recurrences and eventually die of the disease. Various prognostic factors have been identified and been used to create the current staging classifications (AGES, AMES, MACIS, EORTC, UICC-TNM). We examined 499 DTC patients retrospectively to validate known prognostic factors that enable them to be recognised as having either a low or a high risk of death related to a recurrence of DTC, by reference to the current staging classifications. Sixty-nine of them (14%) had local or distant recurrences, the mean time to recurrence being 7.7 years. The 10-year disease-free survival rate was 80%, and the ten-year overall survival rate for the entire group was 91%, with a mean survival time of 8.7 years. Male gender, a follicular type of tumour, larger tumour size, extrathyroidal invasion outside the capsule and nodal metastases were all related to a higher incidence of tumour recurrence, and the follicular type of histology, age > 45 years, larger tumour size and local invasion entailed poorer survival. The AMES and to some extent the EORTC classification were not reproducible in this material, mainly because some prognostic variants were no longer encountered or were insufficient in number to allow reliable conclusions to be drawn. The MACIS staging classification leaves the definition of the intermediate and high risk groups too wide and is therefore not very reliable. Pooling of stages I and II improved the relevance of the TNM classification. All the current staging classifications are able to discern a low risk DTC group well. We achieved a highly accurate definition of risk in the present material using only two parameters, age (cut-off value 50 years) and extracapsular invasion of the thyroid gland.

Adenocarcinoma, Follicular↗
PubMed1989

A new epidemiologic and laboratory classification system for paralytic poliomyelitis cases.

An epidemiologic classification of paralytic poliomyelitis cases (ECPPC) has been in use in the United States since 1976. In 1985, this classification system was reviewed because of recent changes in the epidemiology of paralytic poliomyelitis and improved laboratory capability to definitively characterize poliovirus strains. An alternative classification system was devised, the epidemiologic and laboratory classification of paralytic polio cases (ELCPPC), that incorporated virus isolation and strain characterization with epidemiologic information. Reported paralytic poliomyelitis cases for 1980-86 were classified by both the ECPPC and the ELCPPC classification systems. The new ELCPPC system classified 91 per cent of the reported cases as vaccine-associated, while the ECPPC system classified only 71 per cent of the reported cases as vaccine-associated. The proposed classification system provides more specific and useful information particularly concerning vaccine-associated paralytic poliomyelitis.

Epidemiologic Methods↗
PubMed1998

Intraobserver and interobserver reliability of the classification of thoracic adolescent idiopathic scoliosis.

The system described by King et al. is the standard method for the classification of thoracic adolescent idiopathic scoliosis. Although it is widely used and referenced, its reliability and reproducibility among scoliosis surgeons are unknown. We used a scoliosis case-presentation format to examine the interobserver and intraobserver reliability of the classification of thoracic adolescent idiopathic scoliosis with the system of King et al. Eight active, current members of the Scoliosis Research Society reviewed twenty-seven full-length radiographs that had been made before operative correction of the scoliotic deformity. On the basis of these images, which included posteroanterior and lateral radiographs made with the patient standing as well as right and left forced-side-bending radiographs made with the patient supine, the reviewers assigned a type to each curve according to the classification system of King et al. Kappa coefficients were used to test statistical reliability. The mean interobserver reliability of the classification was only 64 per cent (range, 54 to 77 per cent) when the responses of seven of the reviewers were compared with those of one of the originators of the classification. The mean kappa coefficient was 0.49 (range, 0.27 to 0.73), which indicates poor reliability. When each reviewer's responses were compared with those of the other reviewers, the reliability was similarly poor (interobserver reliability, 55 per cent [range, 33 to 81 per cent] and mean kappa coefficient, 0.40 [range, 0.21 to 0.63]). Intraobserver reliability was evaluated in a trial in which five reviewers in a group setting were shown the same radiographs in a different order at two different viewings. Comparison of the results at the two viewings revealed a mean intraobserver reliability of 69 per cent (range, 56 to 85 per cent) and a mean kappa coefficient of 0.62 (range, 0.34 to 0.95), which indicates fair reliability. The current method of classification of adolescent idiopathic scoliosis does not appear to have sufficient intraobserver or interobserver reliability among scoliosis surgeons to portray curve types accurately. Thus, it may not help to guide treatment with use of modern spinal fixation methods.

Adolescent↗
PubMed1998

Interobserver reliability and intraobserver reproducibility of the system of King et al. for the classification of adolescent idiopathic scoliosis.

The classification of adolescent idiopathic scoliosis with use of the system of King et al. has become widely accepted since its introduction. The purpose of the present study was to establish the interobserver reliability and intraobserver reproducibility of this classification system. The preoperative radiographs of sixty-three patients who were managed operatively for adolescent idiopathic scoliosis were classified by five observers with the system of King et al. Interobserver reliability was assessed by comparison of the classification of the curves among the observers, and intraobserver reproducibility was evaluated by comparison of the classifications of each set of radiographs by each observer on two occasions three weeks apart. The median interobserver reliability kappa coefficient for the classification system of King et al. was 0.44 (range, 0.28 to 0.50), and the median intraobserver reproducibility kappa coefficient was 0.64 (range, 0.44 to 0.72). According to the definition of Landis and Koch, the classification system of King et al. is substantially reproducible but is only moderately reliable. However, according to the stricter definition of Svanholm et al., its reproducibility is only fair and its reliability is poor.

Adolescent↗
PubMed2006

Two and three-dimensional computed tomography for the classification and management of distal humeral fractures. Evaluation of reliability and diagnostic accuracy.

BACKGROUND: Complex fractures of the distal part of the humerus can be difficult to characterize on plain radiographs and two-dimensional computed tomography scans. We tested the hypothesis that three-dimensional reconstructions of computed tomography scans improve the reliability and accuracy of fracture characterization, classification, and treatment decisions. METHODS: Five independent observers evaluated thirty consecutive intra-articular fractures of the distal part of the humerus for the presence of five fracture characteristics: a fracture line in the coronal plane; articular comminution; metaphyseal comminution; the presence of separate, entirely articular fragments; and impaction of the articular surface. Fractures were also classified according to the AO/ASIF Comprehensive Classification of Fractures and the classification system of Mehne and Matta. Two rounds of evaluation were performed and then compared. Initially, a combination of plain radiographs and two-dimensional computed tomography scans (2D) were evaluated, and then, two weeks later, a combination of radiographs, two-dimensional computed tomography scans, and three-dimensional reconstructions of computed tomography scans (3D) were assessed. RESULTS: Three-dimensional computed tomography improved both the intraobserver and the interobserver reliability of the AO classification system and the Mehne and Matta classification system. Three-dimensional computed tomography reconstructions also improved the intraobserver agreement for all fracture characteristics, from moderate (average kappa [kappa2D] = 0.554) to substantial agreement (kappa3D = 0.793). The addition of three-dimensional images had limited influence on the interobserver reliability and diagnostic characteristics (sensitivity, specificity, and accuracy) for the recognition of specific fracture characteristics. Three-dimensional computed tomography images improved intraobserver agreement (kappa2D = 0.62 compared with kappa3D = 0.75) but not interobserver agreement (kappa2D = 0.24 compared with kappa3D = 0.28) for treatment decisions. CONCLUSIONS: Three-dimensional reconstructions improve the reliability, but not the accuracy, of fracture classification and characterization. The influence of three-dimensional computed tomography was much more notable for intraobserver comparisons than for interobserver comparisons, suggesting that different observers see different things in the scans-most likely a reflection of the training, knowledge, and experience of the observer with regard to these relatively uncommon and complex injuries.

Adult↗
PubMed2000

Risk classification systems for drug use during pregnancy: are they a reliable source of information?

BACKGROUND: In several countries, risk classification systems have been set up to summarise the sparse data on drug safety during pregnancy. However, these have resulted in ambiguous statements that are often difficult to interpret and use with accuracy when counselling patients on drug use in pregnancy. OBJECTIVES: The objective of this study was to compare and analyse the consistency between and the criteria for risk classification for medications used during pregnancy included in 3 widely used international risk classification systems. All 3 systems use categories based on risk factors to summarise the degree to which available clinical information has ruled out the risk to unborn offspring, balanced against the drug's potential benefit to the patient. METHODS: Drugs included in the risk classification systems from the US Food and Drug Administration (FDA), the Australian Drug Evaluation Committee (ADEC) and the Swedish Catalogue of Approved Drugs (FASS), were reviewed and compared on basis of the risk factor category to which they had been assigned. Agreement between the systems was calculated as the number of drugs common to all 3 and assigned to the same risk factor category. In addition, evidence on teratogenicity and adverse effects during pregnancy was retrieved using a MEDLINE search (from 1966 up to 1998) for common drugs classified as teratogenic. RESULTS: Differences in the allocation of drugs to different risk factor categories were found. Risk factor category allocation for 645 drugs classified by the FDA, 446 classified by ADEC and 527 classified by FASS was compared. Only 61 (26%) of the 236 drugs common to all 3 systems were placed in the same risk factor category. Analysis of studies on the safety of common drugs during pregnancy of drugs classified as X by the FDA indicated that the variability in category allocation was not only attributable to the different definitions for the categories, but also depended on how the available scientific literature was handled. CONCLUSIONS: Differences in category allocation for the same drug can be a source of great confusion among users of the classification systems as well as for those who require information regarding risk for drug use during pregnancy, and may limit the usefulness and reliability of risk classification systems.

Animals↗