Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Predictive”

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 739 records · Page 41Linked to original sources
PubMed2000

Can antenatal clinical and biochemical markers predict the development of severe preeclampsia?

OBJECTIVE: This study was undertaken to develop a multivariable clinical predictive rule for severe preeclampsia using second-trimester clinical factors and biochemical markers. STUDY DESIGN: We performed a retrospective cohort study of all pregnant patients with single gestations from 1995 through 1997 for whom we had complete follow-up data. Through medical record review we determined whether patients had severe preeclampsia develop according to American College of Obstetricians and Gynecologists criteria. Case patients with severe preeclampsia were compared with control subjects with respect to clinical data and multiple-marker screening test results. With potential predictive factors identified in the bivariate and stratified analyses both an explanatory logistic regression model and a clinical prediction rule were created. Patients were assigned a predictive score according to the presence or absence of predictive factors, and receiver operating characteristic analysis was used to determine the optimal score cutoff point for prediction of severe preeclampsia with maximal sensitivity. RESULTS: Among the 1998 patients we found 49 patients with severe preeclampsia (prevalence, 2.5%). After we controlled for confounding variables, case patients and control subjects had similar human chorionic gonadotropin and alpha-fetoprotein levels, and the only variables that remained significantly associated with severe preeclampsia were nulliparity (relative risk, 3.8; 95% confidence interval, 1.7-8.3), history of preeclampsia (relative risk, 5.0; 95% confidence interval, 1.7-17.2), elevated screening mean arterial pressure (relative risk, 3.5; 95% confidence interval, 1.7-7.2), and low unconjugated estriol concentration (relative risk, 1.7; 95% confidence interval, 0.9-3.4). Our predictive model for severe preeclampsia, which included only these 4 variables, had a sensitivity of 76% and a specificity of 46%. CONCLUSION: Even after incorporation of the strongest risk factors, our predictive model had only modest sensitivity and specificity for discrimination of patients at risk for development of severe preeclampsia. The addition of the human chorionic gonadotropin and alpha-fetoprotein biochemical markers did not enhance the model's predictive value for severe preeclampsia.

Adult↗
PubMed2000

Use of cervical ultrasonography in prediction of spontaneous preterm birth in twin gestations.

OBJECTIVE: This study was undertaken to compare various ultrasonographic cervical parameters with respect to ability to predict spontaneous preterm birth in twin gestations. STUDY DESIGN: This prospective study involved 131 women carrying twins who were longitudinally evaluated on 524 occasions between 15 and 28 weeks' gestation with transvaginal cervical ultrasonography and transfundal pressure. The following cervical parameters were obtained: funnel width and length, cervical length, percentage of funneling, and cervical index. Receiver operating characteristic curve analysis was used to determine the ultrasonographic cervical parameter evaluated at 15 to 20 weeks' gestation, 21 to 24 weeks' gestation, and 25 to 28 weeks' gestation that were best for prediction of spontaneous preterm birth at <28 weeks' gestation, <30 weeks' gestation, <32 weeks' gestation, and <34 weeks' gestation. RESULTS: The median gestational age at delivery was 36 weeks' gestation (range, 21-41 weeks' gestation). Receiver operating characteristic curve analysis indicted that a cervical length of < or =2.0 cm, regardless of gestational age category at cervical measurement, was at least as good as other ultrasonographic cervical parameters at predicting spontaneous preterm birth. Between 15 and 20 weeks' gestation a cervical length cutoff value of < or =2.0 cm had specificities of 97%, 98%, 99%, and 100% and negative predictive values of 99%, 98%, 95%, and 89% for delivery at <28, <30, <32, and <34 weeks' gestation, respectively. The positive predictive values for delivery at <32 and <34 weeks' gestation were 80% and 100%, respectively. Between 21 and 24 weeks' gestation a cervical length of < or =2.0 cm had specificities of 84%, 84%, 85%, and 86% and negative predictive values of 99%, 99%, 94%, and 87% for delivery at <28, <30, <32, and <34 weeks' gestation, respectively. Between 25 and 28 weeks' gestation cervical length had excellent negative predictive values of 99%, 98%, 95%, and 93% for delivery at <28, <30, <32, and <34 weeks' gestation, respectively. CONCLUSIONS: In twin gestations a cervical length of < or =2.0 cm measured between 15 and 28 weeks' gestation was at least as good as other ultrasonographic cervical parameters at predicting spontaneous preterm birth. The high specificities indicate that cervical length was better at predicting the absence than the presence of various degrees of spontaneous prematurity.

Cervix Uteri↗
PubMed2005

Prediction of oral clearance from in vitro metabolic data using recombinant CYPs: comparison among well-stirred, parallel-tube, distributed and dispersion models.

Intrinsic clearances (CLint-HLM-total) for the metabolism of NE-100, metoprolol, clarithromycin (CAM), lornoxicam and tenoxicam were predicted from in vitro data with recombinant cytochorme P450s (CYPs) using relative activity factor (RAF) and then compared with CLint-HLM observed in human liver microsomes (HLM). The predicted CLint-HLM-total correlated well with the observed CLint-HLM in HLM. When oral clearances (CLoral) of low-clearance drugs such as metoprolol, CAM, lornoxicam and tenoxicam were predicted from the in vitro data using four physiological models (well-stirred, parallel tube, distributed and dispersion models), the predicted CLoral corresponded well with the observed CLoralin vivo and were similar among the four models. For a high-clearance drug, the predicted CLoral of NE-100 in extensive CYP2D6 metabolizers (EMs) was substantially different between individual models, although the predicted CLoral in a poor metabolizer of CYP2D6 (PMs) was similar. The CLoral ratio of NE-100 between the EMs and the PMs predicted from the dispersion model, which leads to a reliable prediction for the high-clearance drug, was 48.4, but the ratio decreased depending on the increase of the NE-100 plasma concentration. The results suggest that the CLoral decrease in the EMs is caused by saturation of NE-100 metabolism mediated by CYP2D6 and is based on increases in plasma NE-100 concentrations dependent on the dose of NE-100. The study suggests that the RAF and the in vitro-in vivo scaling approaches are useful for predicting CLoral from in vitro data with recombinant CYPs without using HLM and hepatocytes.

Administration, Oral↗
PubMed2001

Prediction of drug content and hardness of intact tablets using artificial neural network and near-infrared spectroscopy.

The purpose of this study was to predict drug content and hardness of intact tablets using artificial neural networks (ANN) and near-infrared spectroscopy (NIRS). Tablets for the drug content study were compressed from mixtures of Avicel PH-101, 0.5% magnesium stearate, and varying concentrations (0%, 1%, 2%, 5%, 10%, 20%, and 40% w/w) of theophylline. Tablets for the hardness study were compressed from mixtures of Avicel PH-101 and 0.5% magnesium stearate at varying compression forces ranging from 0.4 to 1 ton. An Intact Analyzer was used to obtain near infrared spectra from the tablets with varying drug contents, whereas a Rapid Content Analyzer (RCA) was used to obtain spectral data from the tablets with varying hardness. Two sets of tablets from each batch (i.e., tablets with varying drug content and hardness) were randomly selected. One set of tablets was used to generate appropriate calibration models, while the other set was used as the unknown (test) set. A total of 10 ANN calibration models (5 each with 10 and 160 inputs at appropriate wavelengths) and five separate 4-factor partial least squares (PLS) calibration models were generated to predict drug contents of the test tablets from the spectral data. For the prediction of tablet hardness, two ANN calibration models (one each with 10 and 160 inputs) and two 4-factor PLS calibration models were generated and used to predict the hardness of test tablets. The PLS calibration models were generated using Vision software. Prediction of drug contents of test tablets using the ANN calibration models generated with 10 inputs was significantly better than the prediction obtained with the ANN calibration models with 160 inputs. For tablets with low drug concentrations (less than or equal to 2% w/w) prediction of drug content was better with either of the two ANN calibration models than with the PLS calibration models. However, prediction of drug contents of tablets with greater than or equal to 5% w/w drug was better with the PLS calibration models than with the ANN calibration models. Prediction of tablet hardness was better with the ANN calibration models generated with either 10 or 160 inputs than with the PLS calibration models. This work demonstrated that a well-trained ANN model is a powerful alternative technique for analysis of NIRS data. Moreover, the technique could be used in instances when the conventional modeling of data does not work adequately.

Calibration↗
PubMed1997

Combined use of dobutamine echocardiography and myocardial contrast echocardiography in predicting regional dysfunction recovery after coronary revascularization in patients with recent myocardial infarction.

BACKGROUND: Myocardial contrast echocardiography and dobutamine echocardiography have recently emerged as potentially useful clinical tools to detect reversible myocardial dysfunction. However, the relative accuracy of these two techniques in predicting regional wall motion improvement after coronary interventions is still unclear. The aim of the present study was to compare their diagnostic value in predicting functional recovery after coronary revascularization in patients with recent acute myocardial infarction. METHODS AND RESULTS: Twenty-four patients with acute myocardial infarction underwent myocardial contrast echocardiography and dobutamine echocardiography within 2 weeks of hospital admission. Infarct zone contrast score and wall motion score indexes were derived in each patient. Infarct-related artery revascularization was performed before hospital discharge in all selected patients. Resting echocardiography was repeated 3 months after revascularization, and regional function recovery was analysed. The degree of wall motion score improvement at 3-month follow-up and the percentage of positive responses to dobutamine echo were greater (P < 0.001 and P < 0.002, respectively) in patients with a higher baseline contrast score (> or = 0.50). Conversely, no significant changes were observed either during dobutamine echo or after revascularization in the group of patients without residual perfusion within the infarct area. Diagnostic agreement between both techniques in predicting reversible dysfunction was high (81% of segments). The sensitivity and negative predictive value in predicting functional outcome were 100% (95% confidence interval [CI], 87% to 100%) and 100% (95% CI, 93% to 100%) by contrast echo, and 85% (95% CI, 66% to 96%) and 93% (95% CI, 84% to 98%) by dobutamine echo. The specificity and positive predictive value were 90% (95% CI, 80% to 96%) and 81% (95% CI, 64% to 93%) by contrast echo, and 88% (95% CI, 78% to 95%) and 76% (95% CI, 58% to 90%) by dobutamine echo. The combination of myocardial contrast and dobutamine echocardiography positive responses improved specificity and positive predictive value in detecting functional recovery after revascularization to 100% (95% CI, 94% to 100%) and 100% (95% CI, 85% to 100%), respectively. However, the sensitivity and negative predictive value slightly decreased with the use of both methods (85% [95% CI, 66% to 96%)] and (93%[95% CI, 85% to 98%)], respectively. CONCLUSIONS: In patients with recent myocardial infarction, reversible dysfunction after coronary revascularization and the response to dobutamine infusion are strictly dependent on microvascular integrity. However, microvascular perfusion does not always imply functional recovery after coronary revascularization. The integration with dobutamine echo results seems particularly helpful to further improve myocardial contrast echo specificity and positive predictive values.

Adult↗
PubMed1997

The value of in vitro drug activity and pharmacokinetics in predicting the effectiveness of antimycobacterial therapy: a critical review.

Marked increases in case rates of drug-resistant tuberculosis and nontuberculous mycobacterial infections have brought renewed urgency to the development of new treatment regimens for mycobacterial infections. Preclinical data, such as in vitro measures of drug activity and pharmacokinetics, are used in the design of new treatment regimens. This review surveys the extensive published clinical experience concerning the treatment of drug-susceptible tuberculosis to evaluate the use of these preclinical measures in predicting clinical outcomes of antimycobacterial therapy. In vitro measures of drug activity predict the potency of a drug to prevent the emergence of resistance to other antimycobacterial drugs but do not predict the sterilizing activity of a drug or the activity of drug combinations. In vitro measures of drug activity do not allow reliable predictions of the level at which an organism should be considered resistant. Assays of drug penetration in tissues and activity against intracellular bacilli add modestly to the predictive value of in vitro measures of drug activity but still do not predict sterilizing activity. In contrast, animal models of tuberculosis have predicted relative drug potency (including sterilizing activity), the efficacy of multidrug regimens, and the duration of therapy needed. Despite pharmacokinetic parameters that would suggest the need for multiple doses per day, all of the first-line antituberculous drugs are active when given as infrequently as twice weekly. It is difficult to predict the efficacy of therapy for an intracellular pathogen that has the capacity for dormancy. Better in vitro models are needed, particularly ones that predict sterilizing activity.

Anti-Bacterial Agents↗
PubMed1994

What is the best way to determine oropharyngeal classification and mandibular space length to predict difficult laryngoscopy?

BACKGROUND: Previous studies have suggested that the degree of visibility of oropharyngeal structures (OP class) and mandibular space (MS) length can predict difficult laryngoscopy. However, those studies were either inconsistent or omit description of how to perform these tests with regard to body, head and tongue position, and the use of phonation, hyoid versus thyroid cartilage and inside versus outside of the mentum. The purpose of this investigation was to determine which method of testing best predicts difficult laryngoscopy. METHODS: In each of 213 consenting adults the OP class was determined in 24 method combinations: two body positions (sitting and supine), three head positions (neutral, sniff, and full extension), two tongue positions (in and out), and with and without phonation. In each patient MS length was measured in 24 method combinations: two body positions (sitting and supine), three head positions (neutral, sniff, and full extension), two distal end points (hyoid and thyroid cartilage), and two proximal end points (inside and outside of the mentum). In each patient the laryngoscopic grade was determined at the time of induction of anesthesia. We defined laryngoscopic grades III (n = 24) and 4 (n = 0) as difficult. The area under the receiver operating characteristic curve (ROC area) for each combination was used to compare the combinations and determine significant differences: ROC area = 0.5 implied a totally uninformative combination and ROC area = 1.0 a combination that predicted perfectly. Logistic regression analysis was used to calculate a predictor of difficult intubation that combined both OP class and MS length (the performance index). The performance index could then be used to calculate sensitivity, specificity, positive and negative predictive value, and probability of difficult intubation. RESULTS: The ROC areas for the different combinations used to assess OP class ranged from 0.78 to 0.94. The best combination was with the patient sitting, head in extension, tongue out, and with or without phonation. For MS length, the ROC areas ranged from 0.58 to 0.77; the best combination was the patient sitting, with the head in extension, with distance measured from the inside of the mentum to the thyroid cartilage. Combining the OP class and MS length (performance index = 2.5 X OP class - MS length in centimeters) significantly increased predictability of difficult intubation. At performance index = 0 and = 2, the probability of difficult intubation was 3.5% and 24%, respectively. With clinically relevant cutpoints for the performance index it was found that most difficult intubations could be predicted, but approximately half of those predicted to be difficult would in fact be easy. CONCLUSIONS: Based on the above ROC areas and ease of performing the test for the patient, we recommend that these tests be performed with patients in the sitting position, with the head in full extension, the tongue out, and with phonation, and with distance measured from the thyroid cartilage to inside of the mentum. Nevertheless, it is clear that these two tests, either used alone or in combination, will fail to predict a few difficult laryngoscopies and that they will predict difficult laryngoscopy in a significant number of patients in whom the trachea is easy to intubate.

Adult↗
PubMed2003

Use of somatosensory-evoked potentials and cognitive event-related potentials in predicting outcomes of patients with severe traumatic brain injury.

OBJECTIVE: This study was performed to evaluate the usefulness of somatosensory-evoked potentials (SEPs) and cognitive event-related potentials (ERPs) in predicting functional outcomes of severe traumatic brain injury patients. DESIGN: Prospective study of 22 patients with severe traumatic brain injury. Demographic information, Glasgow Coma Scale, and electrophysiologic measurements were recorded. Functional outcomes, as quantified by the Glasgow Outcome Scale-Extended, were obtained. RESULTS: Bilateral absence of median nerve SEP was strongly predictive of the worst functional outcome. The specificity and positive predictive value of absent SEP for predicting death or persistent vegetative state at 6 mo after traumatic brain injury were as high as 100%. If the definition of unfavorable outcome was expanded to include Glasgow Outcome Scale-Extended 1-4, absence of ERP was equivalent to the absence of SEP in specificity and positive predictive value. On the other hand, normal ERPs showed higher sensitivity and negative predictive value for prognosticating the best outcomes compared with normal SEPs. If the definition of favorable outcome was expanded to include Glasgow Outcome Scale-Extended 5-8, ERP was still superior to SEP for prognosticating good outcome. Interestingly, the highest sensitivity and negative predictive value for favorable outcomes were associated with the presence of any discernible waveform. CONCLUSIONS: Although median nerve SEP continues to make reliable prediction of ominous outcome in severe traumatic brain injury, the addition of the speech-evoked ERPs may be helpful in predicting favorable outcomes. The strength of the latter test seems to complement the weakness of the former.

Adolescent↗
PubMed1999

Prediction of "intent", "discrepancy with intent", and "discrepancy with nonintent" for the patient with chronic pain to return to work after treatment at a pain facility.

OBJECTIVE: We previously determined that "intent" to return to work post pain facility treatment is the strongest predictor for actual return to work. The purposes of the present study were the following: to identify variables predicting "intent"; to predict membership in the "discrepant with intent" group [those chronic pain patients (CPPs) who do intend to return to work but do not]; and to predict membership in the "discrepant with nonintent" group (those CPPs who do not intend to return to work but do). DESIGN: A total of 128 CPPs completed a series of rating scales and yes/no questions relating to their preinjury job perceptions and a question relating to "intent" to return to the same type of preinjury job post-pain facility treatment. These CPPs were part of a grant study for prediction of return to work, and therefore their work status was determined at 1, 3, 6, 12, 18, 24, and 30 months posttreatment. Preinjury job perceptions and other demographic variables were utilized using stepwise discriminant analysis to identify variables predicting "intent" and predicting membership in the "discrepant with intent" and "discrepant with nonintent" groups. SETTING: Pain facility (multidisciplinary pain center). PATIENTS: Consecutive low back pain CPPs, mean age 41.66+/-9.54 years, with the most frequent highest educational status being high school completion (54.7%) and 60.2% being worker compensation CPPs. RESULTS: "Intent" was predicted by (in decreasing order of probability) postinjury job availability variables, job characteristic variables, and a litigation variable. "Discrepant with intent" was predicted by (in decreasing order of probability) for the 1-month follow-up time point, postinjury job availability variables, pain variables, a litigation variable, and a function perception variable, and for the final follow-up time point, pain variables only. "Discrepant with nonintent" was predicted by (in order of decreasing probability) for the 1-month follow-up time point, a job availability variable, a demographic variable, and a functional perception variable, and for the final follow-up time point a pain variable and a job availability variable. The percentage of CPPs correctly classified by each of these analyses was as follows: "intent" 81.25%, "discrepant with intent" 87.01% (at 1-month follow-up) and 74.03% (final follow-up), "discrepant with nonintent" 92.16% (at 1-month follow-up) and 75.00% (final follow-up). CONCLUSIONS: CPPs intentions of returning to their preinjury jobs are mainly determined by job availability and job characteristic variables but surprisingly not by pain variables. However, the results with "discrepant with intent" and "discrepant with nonintent" groups indicate that actual return to work is determined by an interaction between job availability variables and pain variables with pain variables predominating for long-term outcome.

Adult↗
PubMed1998

Evaluation of acute physiology and chronic health evaluation III predictions of hospital mortality in an independent database.

OBJECTIVE: To assess the accuracy and validity of Acute Physiology and Chronic Health Evaluation (APACHE) III hospital mortality predictions in an independent sample of U.S. intensive care unit (ICU) admissions. DESIGN: Nonrandomized, observational, cohort study. SETTING: Two hundred eighty-five ICUs in 161 U.S. hospitals, including 65 members of the Council of Teaching Hospitals and 64 nonteaching hospitals. PATIENTS: A consecutive sample of 37,668 ICU admissions during 1993 to 1996; including 25,448 admissions at hospitals with >400 beds and 1,074 admissions at hospitals with <200 beds. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: We used demographic, clinical, and physiologic information recorded during ICU day 1 and the APACHE III equation to predict the probability of hospital mortality for each patient. We compared observed and predicted mortality for all admissions and across patient subgroups and assessed predictive accuracy using tests of discrimination and calibration. Aggregate hospital death rate was 12.35% and predicted hospital death rate was 12.27% (p =.541). The model discriminated between survivors and nonsurvivors well (area under receiver operating curve = 0.89). A calibration curve showed that the observed number of hospital deaths was close to the number of deaths predicted by the model, but when tested across deciles of risk, goodness-of-fit (Hosmer-Lemeshow statistic, chi-square = 48.71, 8 degrees of freedom, p< .0001) was not perfect. Observed and predicted hospital mortality rates were not significantly (p < .01) different for 55 (84.6%) of APACHE III's 65 specific ICU admission diagnoses and for 11 (84.6%) of the 13 residual organ system-related categories. The most frequent diagnoses with significant (p < .01) differences between observed and predicted hospital mortality rates included acute myocardial infarction, drug overdose, nonoperative head trauma, and nonoperative multiple trauma. CONCLUSIONS: APACHE III accurately predicted aggregate hospital mortality in an independent sample of U.S. ICU admissions. Further improvements in calibration can be achieved by more precise disease labeling, improved acquisition and weighting of neurologic abnormalities, adjustments that reflect changes in treatment outcomes over time, and a larger national database.

APACHE↗
PubMedDate not supplied

Do emergency nurses' triage decisions predict differences in admission or discharge diagnoses for acute coronary syndromes?

Coronary heart disease is the number 1 killer of adults in the United States, affecting 1 in 5 men and women. However, women are more likely than men to die after an acute coronary event and are less likely to receive prompt or aggressive treatment. Few studies have examined the role of emergency nurses' triage decisions in these disparities, even though nurses often determine initial patient priority and urgency status for emergency cardiac evaluation and treatment. The purpose of this prospective study was to examine if nurses' initial triage decisions could predict admission or discharge diagnoses for acute coronary syndromes (ACS). A total of 108 nurses' triage decisions made by 13 nurses were examined. There were no differences in nurses' triage decisions based on patient gender, race, or age. By multivariate analysis, chest pain, history of coronary heart disease, history of myocardial infarction, and smoking were predictive of an ACS decision. Overall, accuracy for predicting admission diagnosis was poor. Sensitivity and specificity were 57% and 59%, respectively, with a positive predictive value of 68% and a negative predictive value of 56%. It was similarly poor for predicting discharge diagnosis. Sensitivity and specificity for discharge diagnosis were 55% and 69%, respectively, with a positive predictive value of 17% and a negative predictive value of 93%. Findings indicate limitations in the ability of nurses' triage decisions to predict admission and discharge diagnoses for ACS.

Adult↗
PubMed2002

Pelvic ring disruptions: prediction of associated injuries, transfusion requirement, pelvic arteriography, complications, and mortality.

OBJECTIVE: To determine if age, fracture pattern, systolic blood pressure on arrival, base deficit, or the Revised Trauma Score is predictive of mortality, transfusion requirements, use of pelvic arteriography, later complications, or injuries associated with the pelvic ring disruption. STUDY DESIGN: Retrospective review of a prospectively collected database. METHODS: All closed pelvic ring disruptions seen between November 1, 1997 and November 30, 1999 were included. Predictive variables and outcome variables were recorded for each patient. Statistical analysis was used to determine if the above variables were predictive. RESULTS: Shock on arrival and the Revised Trauma Score were significantly associated with mortality, transfusion requirement, Injury Severity Score, and all the Abbreviated Injury Scores except the one for skin. In addition, the Revised Trauma Score was significantly associated with the use of pelvic arteriography and predicted more complications than did shock on arrival. Age was significantly associated with transfusion requirement, Injury Severity Score, the chest and skin Abbreviated Injury Scores, use of arteriography, and death. The mortality rate among patients who presented in shock was 57 percent. A Revised Trauma Score of less than 11 predicted mortality with a sensitivity and specificity of 58 percent and 92 percent, respectively. Shock on arrival predicted mortality with a sensitivity and specificity of 27 percent and 96 percent, respectively. Age greater than sixty years predicted mortality with a sensitivity and specificity of 26 percent and 91 percent, respectively. In our analysis of the fracture patterns, we were unable to demonstrate consistent, meaningful links between specific fracture classes and the outcome variables. CONCLUSIONS: Shock on arrival and the Revised Trauma Score are useful predictors of mortality and transfusion requirements, Injury Severity Score, and Abbreviated Injury Scores for the head and neck, face, chest, abdomen, and extremities. In addition, the Revised Trauma Score predicts the use of pelvic arteriography and later complications. Age predicted transfusion requirement, Injury Severity Score, the chest and skin Abbreviated Injury Scores, use of arteriography, and death.

Adolescent↗
PubMed2001

Sensitivity and positive predictive values of presurgical clinical diagnosis of excised benign and malignant skin tumors: a prospective study of 835 lesions in 778 patients.

This article reports on the sensitivity and positive predictive value of clinical diagnosis of benign and malignant skin tumors by expert plastic surgeons in an Israeli clinic. Most published reports have focused on the sensitivity of clinicians' diagnoses, a general measure of the physician's skill that does not predict the rate of accuracy of a physician's diagnoses. Our study of 835 lesions in 778 patients, one of the largest Israeli series, assesses the clinical diagnosis of malignant and benign skin tumors and is one of the few that provide information on the positive predictive value, the measure that is of interest to both physicians and patients. The majority of tumors were benign (56.8 percent), 31.6 percent were malignant, and 11.6 percent were premalignant. Among the 474 benign lesions, 46 percent were nevi. The most common nevi subclass was compound nevi (53 percent), 9 percent of the nevi were dysplastic, and 5 percent were blue nevi. The most common malignant tumor was basal cell carcinoma, accounting for 78 percent of malignant tumors. Although sensitivity for clinical diagnosis of malignancy was 91.3 percent, the positive predictive value for clinical diagnosis of malignancy was 71.3 percent. The sensitivity rate for clinically diagnosing premalignant tumors was 42.3 percent, whereas the positive predictive value for these diagnoses was higher (64.1 percent). The sensitivity rate for diagnosis of all benign lesions was 85.9 percent, and the positive predictive value was 94.2 percent. The sensitivity rate for diagnosis of all nevi was 87.6 percent, and the positive predictive value was 85.7 percent: i.e., only seven of the 218 pathologically proven diagnoses of nevi (3.2 percent) were falsely diagnosed as malignant lesions. Even more interestingly, five of the 223 clinical diagnoses of nevi (2.2 percent) were pathologically proven to be malignant melanomas, and seven were found to be premalignant lesions (3.1 percent). It was concluded that publications which report only on the sensitivity neglect to provide information of interest regarding the positive predictive value. Often, positive predictive value is qualitatively different from the sensitivity, and thus relying only on the sensitivity may lead to incorrect evaluation of a clinical judgment, which may result in erroneous surgical decisions.

Adolescent↗
PubMed1997

Multivariate regression modeling for the prediction of inflammation, systemic pressure, and end-organ function in severe sepsis.

The purpose of this study was to evaluate the feasibility of developing multivariate equations that predicted blood pressure and measured levels of end-organ function indicators quantitatively up to 72 h in advance in critically ill patients with severe sepsis. Data collected prospectively from 59 patients entered into two sequential placebo-controlled clinical trials of recombinant interleukin-1 receptor antagonist in severe sepsis and septic shock was analyzed retrospectively. A series of multivariate equations were developed to predict systemic pressure, coagulation, and vital organ function indicators quantitatively at 24, 48, and 72 h after the onset of severe sepsis. These equations used physiologic and clinical laboratory measurements, plus circulating levels of eicosanoids and cytokines obtained when severe sepsis criteria first were met, and end-organ function indicators measured 24, 48, and 72 h later. Multivariate predictive equations were developed for temperature, white blood cell count, mean arterial pressure (MAP), Pao2/FiO2 ratio, the Murray acute lung injury score, alanine and aspartate aminotransferases, prothrombin time, partial thromboplastin time, platelet count, serum creatinine, and Glasgow Coma Scale. The percentage of data variation explained by the equations ranged from 11.4% (MAP at 48 h) to 85.1% (platelet count at 24 h). Linear regression analysis of predicted values, obtained by entering baseline data from individual patients into the multivariate equations, versus observed results at 24, 48, and 72 h yielded regression coefficients ranging from .371 (MAP at 48 h) to .924 (platelet count at 24 h). Among patients without end-organ dysfunction at baseline, sensitivities for predicting values consistent with the onset of organ failure were > or = 88% in 21/27 (78%) of the predictive equations. Resolution of organ failure indicators present at baseline was predicted successfully in individual patients, with 20/27 (74%) specificities > or = 76%. In critically ill patients with severe sepsis, multivariate analysis of interactions among clinical observations, standard laboratory tests, and inflammatory response mediators produced equations that predicted systemic blood pressure and inflammatory and end-organ function indicators quantitatively up to 72 h in advance. Whether or not this methodology might be developed further to predict subclinically the onset and resolution of acute organ failure and shock in critically ill patients, and if it can be validated in a prospective trial will require further studies.

Adolescent↗
PubMed2000

Prospectively validated prediction of organ failure and hypotension in patients with septic shock: the Systemic Mediator Associated Response Test (SMART).

Conventional outcomes research provides only percentage risk categories that are not applicable to individual patients, and it predicts only mortality, utilization of resources and/or broad groupings of multiple organ system dysfunction. The purpose of the present study was to determine whether or not the Systemic Mediator Associated Response Test (SMART) methodology could identify interactions among demographics, physiologic parameters, standard hospital laboratory tests, and circulating cytokine concentrations to predict continuous and dichotomous dependent clinical variables, in advance, in individual patients with septic shock and to integrate these into prospectively validated models. Two hundred forty (240) patients with septic shock who were entered into the placebo arm of a multi-institutional clinical trial were randomly separated into a model building training cohort (n = 154) and a predictive cohort (n = 86), which was used to prospectively validate the prognostic models built upon the training cohort database. From baseline patient demographics; hospital laboratory tests; and plasma levels of interleukin-6, interleukin-8, and granulocyte colony stimulating factor, multiple regression models were developed that predicted clinically important continuous dependent variables quantitatively in individual patients. Multivariate stepwise logistic regression was utilized to develop models that prognosticated dichotomous dependent end points. At the completion of the modeling process, baseline data from individual patients in the predictive cohort was inserted into each multivariate model for each day. Prospective validation was accomplished by simple linear regression of individual predicted versus observed values for continuous dependent variables, and by establishing the Receiver Operator Characteristics Area Under the Curve (ROC AUC) for logistic regression models that predicted dichotomous end points. Through seven days, SMART quantitative predictions of selected physiologic and metabolic parameters were validated at r > 0.500 in 51%. Up to seven days after baseline, 31/49 (63%) SMART models for renal and liver function indicators were validated prospectively at the r > 0.700 level. For hematologic/coagulation models, 37/56 (66%) up to seven days had r > 0.900. Among dichotomous models, ROC AUC > 0.700 was achieved in 30/49 (61%) during the first week. SMART integration of demographics, bedside physiology, hospital laboratory tests, and circulating cytokines predicts organ failure and physiologic function indicators in individual patients with septic shock.

Area Under Curve↗
PubMed1996

Predicting endoscopic diagnosis in the dyspeptic patient: the value of clinical judgement.

OBJECTIVE: To compare the quality of chance-corrected clinical diagnosis in two groups of dyspeptic patients, using endoscopy as the diagnostic standard. DESIGN: Structured interview before endoscopy and clinical predictions of endoscopic diagnosis as either malignancy, peptic ulcer, oesophagitis or non-ulcer dyspepsia. The quality of the predictions was corrected for chance using iota-correction. Patients gave a provisional prediction of their own endoscopic diagnosis. SETTING: Two endoscopy units in Odense and Svendborg, Denmark. PATIENTS: Two groups of dyspeptic outpatients: (1) 1026 patients referred for open-access endoscopy and (2) 207 empirically managed patients randomly assigned to prompt endoscopy as part of a clinical trial. RESULTS: The overall diagnostic validity for all diagnoses was equal in the two groups of patients (57 and 59%) and was mainly accounted for by positive predictive values for non-ulcer dyspepsia of 75%. Elimination of random accuracy for non-ulcer dyspepsia showed a validity of only 23 and 21%. Patients with a major pathologic lesion (cancer, ulcer, complicated oesophagitis) were misclassified clinically as non-ulcer dyspepsia in 36 and 38% of cases. The sensitivity of a clinical prediction of ulcer was only 52 and 36%, despite positive predictive values of 34%, and most valid when corrected for chance in the group of patients referred for open-access endoscopy. The patients' provisional diagnoses had no predictive value. CONCLUSION: Clinical diagnosis in dyspepsia was unreliable as it misclassified one-third of patients with a major pathological lesion. Fifty percent of patients with ulcer were misclassified and that clinical diagnosis could only be confirmed in one-third of the cases. The chance-corrected validity of non-ulcer dyspepsia was only slightly better than chance. There was no predictive value of the patients' predictions of their own diagnosis.

Adult↗
PubMed2006

Optimizing early prediction for antipsychotic response in schizophrenia.

OBJECTIVE: Researchers, by studying first-generation antipsychotics, have established an early prediction model, which had a favorable specificity but a low sensitivity. This study aims to optimize early prediction of treatment response for schizophrenia using a novel statistic method that can be done even under the Microsoft Excel system of a personal computer. METHODS: One hundred twenty-three inpatients with acutely exacerbated schizophrenia were given optimal therapy of risperidone, a commonly used second-generation antipsychotic agent. Response was defined as a reduction of 20% or more in the Positive and Negative Syndrome Scale total score. We applied the generalized estimating equation method's logistic regression to establish an early prediction model based on the treatment results of the first and the second weeks. RESULTS: The proposed method correctly predicted nonresponse at 4 and 6 weeks in 80.8% and 81.8% of the patients, respectively. The method also identified responder at 4 and 6 weeks in 80.0% and 82.8%, respectively. The predictive powers (or correct prediction rates) at 4 and 6 weeks were 80.3% and 82.4%, respectively. In addition, the results based on the responses in Positive and Negative Syndrome Scale scores were slightly better than those in Brief Psychiatric Rating Scale scores. CONCLUSIONS: Using the first 2 weeks' treatment results to predict the fourth or sixth week's treatment response is acceptable in terms of specificity, sensitivity, and predictive power. Further studies are needed. Moreover, whether this model could be applied to establish a prediction system for other psychotropics, such as antidepressants, also deserves research.

Adolescent↗
PubMed2004

Can we improve the prediction of stone-free status after extracorporeal shock wave lithotripsy for ureteral stones? A neural network or a statistical model?

PURPOSE: We evaluated whether an artificial neural network (ANN) can improve the prediction of stone-free status after extracorporeal shock wave lithotripsy (ESWL) (Dornier Medical Systems, Inc., Marietta, Georgia) for ureteral stones compared to a logistic regression (LR) model. MATERIALS AND METHODS: Between February 1989 and December 1998, 984 patients with ureteral stones, including 780 males and 204 females with a mean age +/- SD of 40.85 +/- 10.33 years, were treated with ESWL. Stone-free status at 3 months was determined by urinary tract plain x-ray and excretory urography. Of all patients 919 (93.3%) were free of stones. The impact of 10 factors on stone-free status was studied using an LR model and ANN. These factors were patient age and sex, renal anatomy, stone location, side, number, length and width, whether stones were de novo or recurrent, and stent use. An LR model was constructed and ANN was trained on 688 randomly selected patients (70%) to predict stone-free status at 3 months. The 10 factors were used as covariates in the LR model and as input parameters to ANN. Performance of the trained net and developed logistic model was evaluated in the remaining 296 patients (30%), who served as the test set. The sensitivity (percent of correctly predicted stone-free cases), specificity (percent of correctly predicted nonstonefree cases), positive predictive value, overall accuracy and average classification rate of the 2 techniques were compared. Relevant variables influencing the construction of the 2 models were compared. RESULTS: Evaluating the performance of the LR and ANN models on the test set revealed a sensitivity of 100% and 77.9%, a specificity of 0.0% and 75%, a positive predictive value of 93.2% and 97.2%, an overall accuracy of 93.2% and 77.7%, and an average classification rate of 50% and 76.5%, respectively. LR failed to predict any nonstone free cases. LR and ANN identified stone location and stent use as important factors in determining the outcome, while ANN also identified stone length and width as influential factors. CONCLUSIONS: ANN and LR could predict adequately those who would be stone-free after ESWL for ureteral stones. The neural network has a higher ability to predict those who fail to respond to ESWL.

Female↗