Hosmer-Lemeshow Test Statistic Research Articles

Empowering Maternal Choice: Exploring Factors Influencing Early Postpartum Contraceptive Adoption Intention Among Pregnant Women in Northeast Ethiopia.

Despite progress in access to family planning services in many sub-Saharan African countries in recent decades, advances in early postpartum contraceptive adoption remain low, and the unmet need for early postpartum contraceptives is high. According to the Ethiopia Demographic and Health Survey report, early postpartum modern contraceptive method uptake is still unacceptably low in Ethiopia. This study aimed to determine the magnitude of intention to adopt early postpartum modern contraceptive methods and its associated factors among pregnant women in Dessie and Kombolcha town zones, northeast Ethiopia. A community-based cross-sectional study was deployed from 15 January-15 February 2023, in the Dessie and Kombolcha zones, northeast Ethiopia, among pregnant women. The study involved 780 pregnant women using the cluster sampling technique. A census was conducted in 20 randomly selected clusters to identify eligible pregnant women. Actual data were collected home-to-home in the community through face-to-face interviews. Data were collected by Open Data Kit (ODK) and exported to STATA 17 for analysis. A multivariable logistic regression analysis was performed, and the goodness of the model was checked by Hosmer-Lemeshow's test statistic and rock curve. An adjusted odds ratio with a 95% confidence interval and p-value < 0.05 was considered statistically significant. The study revealed that 49.6% of pregnant women lack autonomy and 50% lack knowledge about early postpartum contraception, with participants' wealth index status ranging from rich (36.6%) to poor (33.2%). The study found that 75.8% of pregnant women intended to adopt early postpartum modern contraceptive methods early after childbirth. After controlling the potential confounders, mother's age (AOR = 6.2 [2.6-14.6], birth interval (AOR = 2.5 [1.6-3.7]), have paid work (AOR = 1.9 [1.3-2.8]), health facility from home (AOR = 2.6 [1.5-4.4]), last delivery Place (AOR = 2.4 [1.1-5.7]), knowledge on (AOR = 1.5 [1.1-2.1]), and antenatal care follow-up (AOR = 1.9 [1.2-3.3]) were significant associated factors of intention to uptake early postpartum modern contraceptive methods among pregnant women. The study found that 75% of the participants had the intention to adopt contraceptive methods during the early postpartum period. Identified factors influencing this intention were age, birth interval, women's employment status, area of residence, distance to health facilities, last delivery place, knowledge of early postpartum modern contraception, gravidity, and antenatal care follow-up. These findings highlight the need for targeted interventions to address these factors, framing the intended users and enabling access to early adoption of postpartum contraceptive methods.

Abstract 15409: Analysis of Discrimination and Calibration of Two Cardiovascular Risk Scores in Patients With Non-Alcoholic Steatohepatitis Shows a Need for Improved Accuracy of These Models

Background: CV risk can be estimated by risk scores such as the Framingham and the Pooled Cohort Equations (PCE). The applicability of these tools in present day NASH (non-alcoholic steatohepatitis) cohorts, individuals with increased morbidity and mortality from CVD, was evaluated to assess their relevance. Hypothesis: The Null Hypothesis is that there is no difference between the predicted CV risk as estimated by the Framingham or PCE risk scores, and the actual CV events that occur in a NAFLD (non-alcoholic fatty liver disease)/ NASH cohort. Aim: To assess the prognostic performance of the Framingham and PCE risk scores in a NAFLD/NASH cohort. Methods: The study population were US patients in the observational TARGET-NASH study with NAFLD/NASH aged 18 years or older. Patients with any CV history at or prior to index were excluded. Patients’ five-year CV risk was estimated using recalibrated Framingham and PCE models and compared with observed CV events. Model discrimination and calibration were assessed using the area under the receiver operator curve (AUROC) and Hosmer-Lemeshow test statistic, respectively. Results: 980 and 587 patients in the Framingham and PCE cohorts, respectively, had all data needed for calculation of CV risk, of which 274 and 169 had at least five years of follow-up. Framingham five-year predicted CV risk was significantly greater among patients who did vs did not experience a CV event (13.7% [SD 9.0] vs 10.5% [SD 8.8]; p=0.02). No statistically significant difference was observed in the PCE analysis. The AUROC was 0.62 (95% CI 0.51, 0.73) for Framingham and 0.58 (95% CI 0.44, 0.72) for PCE at five years, with worst predictive performance among the subgroup of cirrhotic NASH patients for both risk equations. Analyses of model calibration revealed a statistically significant lack of calibration for both tools at five years (Framingham: χ 2 =25.58, p=0.001; PCE: χ 2 =33.86, p&lt;0.001). Conclusion: Framingham and PCE may have poor predictive accuracy for CVD risk in NAFLD/NASH cohorts. This tended to be most apparent among the upper decile of predicted risk, with predicted values far exceeding the observed. Lack of ten-year follow-up and unmeasured risk factors may also explain some of the residual overestimation not accounted for in our models.

Prediction of histopathologic grades of myxofibrosarcoma with radiomics based on magnetic resonance imaging.

To develop a radiomics-based model from preoperative magnetic resonance imaging (MRI) for predicting the histopathological grades of myxofibrosarcoma. This retrospective study included 54 patients. The tumors were classified into high-grade and low-grade myxofibrosarcoma. The tumor size, signal intensity heterogeneity, margin, and surrounding tissue were evaluated on MRI. Using the least absolute shrinkage and selection operator (LASSO) algorithms, 1037 radiomics features were obtained from fat-suppressed T2-weighted images (T2WI), and a radiomics signature was established. Using multivariable logistic regression analysis, three models were built to predict the histopathologic grade of myxofibrosarcoma. A radiomics nomogram represents the integrative model. The three models' performance was evaluated using the receiver operating characteristics (ROC) and calibration curves. The high-grade myxofibrosarcoma had greater depth (P = 0.027), more frequent heterogeneous signal intensity at T2WI (P = 0.015), and tail sign (P = 0.014) than the low-grade tumor. The area under curve (AUC) of these conventional MRI features models was 0.648, 0.656, and 0.668, respectively. Seven radiomic features were selected by LASSO to construct the radiomics signature model, with an AUC of 0.791. The AUC of the integrative model based on radiomics signature and conventional MRI features was 0.875. The integrative model's calibration curve and insignificant Hosmer-Lemeshow test statistic (P = 0.606) revealed good calibration. An integrative model using radiomics signature and three conventional MRI features can preoperatively predict low- or high-grade myxofibrosarcoma.

NTCP Modeling for High-Grade Temporal Radionecroses in a Large Cohort of Patients Receiving Pencil Beam Scanning Proton Therapy for Skull Base and Head and Neck Tumors

To develop a normal tissue complication probability model including clinical and dosimetric parameters for high-grade temporal lobe radionecroses (TRN) after pencil beam scanning proton therapy. We included data on 299 patients with skull base and head and neck tumors treated with pencil-beam scan proton therapy, with a total dose of ≥60 GyRBE (relative biological effectiveness) from May 2004 to November 2018. We considered 9 clinical and 27 dosimetric parameters for the structure-wise modeling of high-grade (grade ≥2) TRN. After eliminating strongly cross-correlated variables, we generated logistic regression models using least absolute shrinkage and selection operator regression. We performed bootstrapping to assess parameter selection robustness and evaluated model performance via cross-correlation by assessing the area under the curve of receiver operating characteristic curves and calibration with a Hosmer-Lemeshow test statistic. After a median radiologic follow-up of 51.5 months (range, 4-190), 27 patients (9%) developed grade ≥2 TRN. Eleven patients had bitemporal necrosis, resulting in 38 events in 598 temporal lobes for structure-wise analysis. During our bootstrapping analysis, we found that the highest selection frequency was for prescription dose, followed by age, V40Gy (%), hypertension, and dose to at least 1 cc (D1cc) (Gy) in the temporal lobe. During our cross-validation, we found that age*prescription-dose*D1cc (Gy)*hypertension was superior in all described test statistics. We built full cohort structure-wise and patient-wise models with maximum area under the curve of receiver operating characteristic curves of 0.79 (structure-wise) and 0.76 (patient-wise). While developing a logistic regression normal tissue complication probability model to predict grade ≥2 TRN, the best fit was found for the model containing age, prescription dose, D1cc (Gy), and hypertensive blood pressure as risk factors. External validation will be the next step to improve generalizability and potential introduction into clinical routine.

Multiple Logistic Regression Model for Determinants of Injectable Contraceptive Uptake Among Women of Reproductive Age in Kenya

Objective: The recent increase in the uptake of injectable contraceptives has occurred at the expense of the other modern contraceptive methods but the knowledge gap still exists on modeling dynamics and determinants associated with the use of the injectable. This study sought to model for injectable contraceptive usage to bridge the knowledge gap on the use of injectable contraceptives among women of childbearing age in Kenya. Materials and methods: Analytical cross-sectional study design was adopted. Secondary data for women collected during the (Performance Monitoring for Action) PMA2020 survey was used. PMA2020 survey used multistage stratified sampling with urban-rural representation. To establish the factors associated with the uptake of injectable contraceptives, a multiple logistic regression model was fitted using Stata version 13 and R version 3.5.3 statistical software. Hosmer-Lemeshow Test statistic was used to evaluate the goodness of model fit in predicting injectable contraceptive usage. Results: Multivariable analysis showed that women with post-primary/vocational levels of education were 54% less likely to use an injectable contraceptive compared to those who had no education at all. Hosmer-Lemeshow (HL) goodness of fit test statistic indicated that the model was a good fit for prediction. Education, marital status, wealth quintile, place of residence and number of births were significant predictors of the injectable contraceptive uptake among women of reproductive age in Kenya. Conclusion: The findings of this study will inform the design of targeted interventions aimed at addressing the increasing demand for injectable devices among women of reproductive age in Kenya.

A nomogram based on easily obtainable parameters for distal deep venous thrombosis in patients after acute stroke

ObjectivesTo develop and validate a nomogram to predict the probability of distal deep venous thrombosis (DVT) within first 14 days of stroke onset in patients by using easily obtainable parameters. MethodsThis is a retrospective study. The presence of distal DVT was evaluated using ultrasonography within the first 14 days. Data were randomly assigned to either a modelling data set or a validation data set. Univariable and multivariate logistic regression analysis was used to determine risk scores to predict distal DVT in the modelling data set, and nomogram and calibration curve were constructed by R project. ResultsA total of 1620 patients with acute stroke were enrolled in the study. The multivariate analysis revealed that the old age, female gender, haemorrhagic stroke, coronary heart disease, lower limb weakness, a low serum albumin level, and a high D-dimer level are highly predictive of 14-day risk of distal DVT. The AUC of the nomogram to predict the 14-day risk of distal DVT was 0.785 (95% CI, 0.742–0.827) and 0.813 (0.766–0.860) for the modelling cohort and external validation cohort, respectively. Moreover, the calibration of the nomogram showed a nonsignificant Hosmer-Lemeshow test statistic in the modelling (P = 0.876) and validation (P = 0.802) sets. With respect to decision curve analyses, the nomogram exhibited preferable net benefit gains than the staging system across a wide range of threshold probabilities. ConclusionThe established nomogram displayed a superior performance in terms of predictive accuracy, discrimination capability, and clinical utility, may be helpful for clinicians to identify high-risk groups of distal DVT.

Development and Validation of a Nomogram for Lower Extremity Deep Venous Thrombosis in Patients after Acute Stroke

ObjectivesTo develope and validate a nomogram to predict the probability of deep venous thrombosis (DVT) in patients after acute stroke during the first 14 days with clinical features and easily obtainable biochemical parameters. MethodsThis is a single-center prospective cohort study. The potential predictive variables for DVT at baseline were collected, and the presence of DVT was evaluated using ultrasonography within the first 14 days. Data were randomly assigned to either a modeling data set or a validation data set. Univariable and Multivariate logistic regression analysis was used to develop risk scores to predict DVT in the modeling data set and the area under the receiver operating characteristic curve to validate the score in the test data set, and nomogram and calibration curve were constructed by R project. ResultsA total of 1651 patients with acute stroke were enrolled in the study. The overall incidence of DVT after acute stroke within two weeks was 14.4%. Multivariable analysis detected older age (≥65 years),female gender, hemorrhagic stroke, malignancy, lower limb muscle strength<3 grade, Albumin<40 g·L−1 and D-dimer>0.5 mg·L−1 were highly predictive of 14-day risk of DVT. The AUC of the nomogram with these above-mentioned independent risk factors to predict the 14-day risk of DVT was 0.756 (95% CI, 0.712-0.812) and 0.811 (95%CI, 0.762-0.859) for the modeling cohort and external validation cohort, respectively. Moreover, the calibration of the nomogram showed a nonsignificant Hosmer-Lemeshow test statistic in the modeling (P = 0.250) and validation sets (P = 0.995). With respect to decision curve analyses, the nomogram exhibited preferable net benefit gains than the staging system across a wide range of threshold probabilities. ConclusionThis nomogram had a good performance in predictive accuracy, discrimination capability, and clinical utility, which was helpful for clinicians to identify high-risk groups of DVT and formulate relevant prevention and treatment measures.

Predicting 30-day readmission after congenital heart surgery across the lifespan.

Hospital readmission is an important driver of costs among patients with CHD. We assessed predictors of 30-day rehospitalisation following cardiac surgery in CHD patients across the lifespan. This was a retrospective analysis of 981 patients with CHD who had cardiac surgery between January 2011 and December 2012. A multivariate logistic regression model was used to identify demographic, clinical, and surgical predictors of 30-day readmission. Receiver operating curves derived from multivariate logistic modelling were utilised to discriminate between patients who were readmitted and not-readmitted at 30 days. Model goodness of fit was assessed using the Hosmer-Lemeshow test statistic. Readmission in the 30 days following congenital heart surgery is common (14.0%). Among 981 patients risk factors associated with increased odds of 30-day readmission after congenital heart surgery through multivariate analysis included a history of previous cardiac surgery (p < 0.001), longer post-operative length of stay (p < 0.001), as well as nutritional (p < 0.001), haematologic (p < 0.02), and endocrine (p = 0.04) co-morbidities. Patients who underwent septal defect repair had reduced odds of readmission (p < 0.001), as did children (p = 0.04) and adult (p = 0.005) patients relative to neonates. Risk factors for readmission include a history of cardiac surgery, longer length of stay, and co-morbid conditions. This information may serve to guide efforts to prevent readmission and inform resource allocation in the transition of care to the outpatient setting. This study also demonstrated the feasibility of linking a national subspecialty registry to a clinical and administrative data repository to follow longitudinal outcomes of interest.

MRI-based radiomics nomogram to predict synchronous liver metastasis in primary rectal cancer patients.

At the time of diagnosis, approximately 15%‐20% of patients with rectal cancer (RC) presented synchronous liver metastasis (SLM), which is the most common cause of death in patients with RC. Therefore, preoperative, noninvasive, and accurate prediction of SLM is crucial for personalized treatment strategies. Recently, radiomics has been considered as an advanced image analysis method to evaluate the neoplastic heterogeneity with respect to diagnosis of the tumor and prediction of prognosis. In this study, a total of 1409 radiomics features were extracted for each volume of interest (VOI) from high‐resolution T2WI images of the primary RC. Subsequently, five optimal radiomics features were selected based on the training set using the least absolute shrinkage and selection operator (LASSO) method to construct the radiomics signature. In addition, radiomics signature combined with carcinoembryonic antigen (CEA) and carbohydrate antigen 19‐9 (CA19‐9) was included in the multifactor logistic regression to construct the nomogram model. It showed an optimal predictive performance in the validation set as compared to that in the radiomics model. The favorable calibration of the radiomics nomogram showed a nonsignificant Hosmer‐Lemeshow test statistic (P > .05). The decision curve analysis (DCA) showed that the radiomics nomogram is clinically superior to the radiomics model. Therefore, the nomogram amalgamating the radiomics signature and clinical risk factors serve as an effective quantitative approach to predict the SLM of primary RC.

Radiomics model based on preoperative gadoxetic acid-enhanced MRI for predicting liver failure.

BACKGROUNDPostoperative liver failure is the most severe complication in cirrhotic patients with hepatocellular carcinoma (HCC) after major hepatectomy. Current available clinical indexes predicting postoperative residual liver function are not sufficiently accurate.AIMTo determine a radiomics model based on preoperative gadoxetic acid-enhanced magnetic resonance imaging for predicting liver failure in cirrhotic patients with HCC after major hepatectomy.METHODSFor this retrospective study, a radiomics-based model was developed based on preoperative hepatobiliary phase gadoxetic acid-enhanced magnetic resonance images in 101 patients with HCC between June 2012 and June 2018. Sixty-one radiomic features were extracted from hepatobiliary phase images and selected by the least absolute shrinkage and selection operator method to construct a radiomics signature. A clinical prediction model, and radiomics-based model incorporating significant clinical indexes and radiomics signature were built using multivariable logistic regression analysis. The integrated radiomics-based model was presented as a radiomics nomogram. The performances of clinical prediction model, radiomics signature, and radiomics-based model for predicting post-operative liver failure were determined using receiver operating characteristics curve, calibration curve, and decision curve analyses.RESULTSFive radiomics features from hepatobiliary phase images were selected to construct the radiomics signature. The clinical prediction model, radiomics signature, and radiomics-based model incorporating indocyanine green clearance rate at 15 min and radiomics signature showed favorable performance for predicting postoperative liver failure (area under the curve: 0.809-0.894). The radiomics-based model achieved the highest performance for predicting liver failure (area under the curve: 0.894; 95%CI: 0.823-0.964). The integrated discrimination improvement analysis showed a significant improvement in the accuracy of liver failure prediction when radiomics signature was added to the clinical prediction model (integrated discrimination improvement = 0.117, P = 0.002). The calibration curve and an insignificant Hosmer-Lemeshow test statistic (P = 0.841) demonstrated good calibration of the radiomics-based model. The decision curve analysis showed that patients would benefit more from a radiomics-based prediction model than from a clinical prediction model and radiomics signature alone.CONCLUSIONA radiomics-based model of preoperative gadoxetic acid–enhanced MRI can be used to predict liver failure in cirrhotic patients with HCC after major hepatectomy.

Multi-cohort modeling strategies for scalable globally accessible prostate cancer risk tools

BackgroundOnline clinical risk prediction tools built on data from multiple cohorts are increasingly being utilized for contemporary doctor-patient decision-making and validation. This report outlines a comprehensive data science strategy for building such tools with application to the Prostate Biopsy Collaborative Group prostate cancer risk prediction tool.MethodsWe created models for high-grade prostate cancer risk using six established risk factors. The data comprised 8492 prostate biopsies collected from ten institutions, 2 in Europe and 8 across North America. We calculated area under the receiver operating characteristic curve (AUC) for discrimination, the Hosmer-Lemeshow test statistic (HLS) for calibration and the clinical net benefit at risk threshold 15%. We implemented several internal cross-validation schemes to assess the influence of modeling method and individual cohort on validation performance.ResultsHigh-grade disease prevalence ranged from 18% in Zurich (1863 biopsies) to 39% in UT Health San Antonio (899 biopsies). Visualization revealed outliers in terms of risk factors, including San Juan VA (51% abnormal digital rectal exam), Durham VA (63% African American), and Zurich (2.8% family history). Exclusion of any cohort did not significantly affect the AUC or HLS, nor did the choice of prediction model (pooled, random-effects, meta-analysis). Excluding the lowest-prevalence Zurich cohort from training sets did not statistically significantly change the validation metrics for any of the individual cohorts, except for Sunnybrook, where the effect on the AUC was minimal. Therefore the final multivariable logistic model was built by pooling the data from all cohorts using logistic regression. Higher prostate-specific antigen and age, abnormal digital rectal exam, African ancestry and a family history of prostate cancer increased risk of high-grade prostate cancer, while a history of a prior negative prostate biopsy decreased risk (all p-values < 0.004).ConclusionsWe have outlined a multi-cohort model-building internal validation strategy for developing globally accessible and scalable risk prediction tools.

A genetic programming approach to development of clinical prediction models: A case study in symptomatic cardiovascular disease.

BackgroundGenetic programming (GP) is an evolutionary computing methodology capable of identifying complex, non-linear patterns in large data sets. Despite the potential advantages of GP over more typical, frequentist statistical approach methods, its applications to survival analyses are rare, at best. The aim of this study was to determine the utility of GP for the automatic development of clinical prediction models.MethodsWe compared GP against the commonly used Cox regression technique in terms of the development and performance of a cardiovascular risk score using data from the SMART study, a prospective cohort study of patients with symptomatic cardiovascular disease. The composite endpoint was cardiovascular death, non-fatal stroke, and myocardial infarction. A total of 3,873 patients aged 19–82 years were enrolled in the study 1996–2006. The cohort was split 70:30 into derivation and validation sets. The derivation set was used for development of both GP and Cox regression models. These models were then used to predict the discrete hazards at t = 1, 3, and 5 years. The predictive ability of both models was evaluated in terms of their risk discrimination and calibration using the validation set.ResultsThe discrimination of both models was comparable. At time points t = 1, 3, and 5 years the C-index was 0.59, 0.69, 0.64 and 0.66, 0.70, 0.70 for the GP and Cox regression models respectively. At the same time points, the calibration of both models, which was assessed using calibration plots and the generalization of the Hosmer-Lemeshow test statistic, was also comparable, but with the Cox model being better calibrated to the validation data.ConclusionUsing empirical data, we demonstrated that a prediction model developed automatically by GP has predictive ability comparable to that of manually tuned Cox regression. The GP model was more complex, but it was developed in a fully automated way and comprised fewer covariates. Furthermore, it did not require the expertise normally needed for its derivation, thereby alleviating the knowledge elicitation bottleneck. Overall, GP demonstrated considerable potential as a method for the automated development of clinical prediction models for diagnostic and prognostic purposes.

Semen analysis and prediction of natural conception.

Do two semen analyses predict natural conception better than a single semen analysis and will adding the results of repeated semen analyses to a prediction model for natural pregnancy improve predictions? A second semen analysis does not add helpful information for predicting natural conception compared with using the results of a single semen analysis and addition of the second analysis to a prediction model for natural conception did not improve predictions. A major problem with semen analyses is the large variability of results within an individual. High-quality evidence is lacking on how many semen analyses need to be performed during the fertility workup to achieve an accurate prediction of conception. We conducted a prospective cohort study of 897 consecutive couples presenting with subfertility in two university hospitals in the period 2002-2004 in the Netherlands. The laboratories scored sperm parameters according to the 1999 WHO criteria. Sperm concentration was counted and motility was assessed in a Makler counting chamber at a magnification of ×200. All assessments were performed by trained laboratory technicians. Follow-up started at the completion of the infertility workup and ended after 12 months. Primary end-point was natural conception resulting in an ongoing pregnancy. We constructed models for three strategies for the prediction of natural conception, using univariable and multivariable Cox hazard regression analyses. We evaluated the performance of the three strategies by comparing goodness-of-fit, discrimination and calibration. First, we analysed the semen parameters only. Secondly, we analysed the semen parameters in addition to the multivariable Hunault prediction model. Of the 897 couples, 132 (15%) achieved a pregnancy by natural conception. Using the results of a single semen analysis only, the calculated probabilities of natural conception within 12 months across the study population ranged from 0.12 to 0.38, with a median of 0.16 (IQR: 0.16-0.17). Using the results of two semen analyses did not lead to a better goodness-of-fit. Discriminative capacity was rather poor, with an area under the ROC curve (AUC) ranging from 0.51 to 0.56. Using the Hosmer-Lemeshow test statistic we found no signs of poor calibration. Using the results of two semen analyses in combination with the Hunault model did not significantly increase goodness-of-fit compared with using a single semen analysis. The Hunault model with the addition of the semen parameters fitted the data significantly better than the Hunault model itself (difference in -2 Log likelihood: 13; 3 df; P = 0.002). Using the Hosmer-Lemeshow test statistic we found no signs of poor calibration. The academic setting possibly explains the relatively low natural conception rates, with only 15% achieving a natural conception within 1 year. Men with azoospermia were excluded. Performing more than one semen analysis will not increase the prognostic power of the test in clinical practice. Adding the first semen analysis to the Hunault model for the prediction of natural conception improved performance significantly compared with using the Hunault model alone. External validation, in other populations, should follow to confirm our conclusions, and to evaluate the generalizability or transportability of the extended Hunault model. No external funding was involved in this study. None of the authors has any conflict of interest to declare.

Decline in usually visiting the dentist for a problem in Australia, 1994 to 2010: an age–period–cohort analysis

AbstractObjectivesTo determine the extent to which age, period and cohort factors have contributed to variation in problem‐oriented dental visiting over time.MethodsData were obtained from four National Dental Telephone Interview Surveys of Australian residents aged 5 years and over conducted in 1994, 1999, 2004 and 2010 (response rates 48–72%). The trend in the percentage of persons usually visiting the dentist for a problem was analysed by means of a standard cohort table and by a nested models framework for age–period–cohort analyses.ResultsThe percentage of persons usually visiting the dentist for a problem generally decreased over the periods examined (from 42.5% in 1994 to 31.5% in 2010). Problem visiting tended to be lower for children and adolescents compared with adult age groups at each point in time. Model fit tests revealed that the age–period–cohort model provided the best fit for the data (Hosmer–Lemeshow test statistic = 5.3; d.f. = 8; P‐value = 0.72), indicating that both period and cohort factors were influential in problem visiting.ConclusionThis study found similar, consistent stories for both the age–period and age–cohort models, with usually visiting for a problem tending to be higher in older age groups and older cohorts. Problem visiting tended to decline over time for most age groups and most age cohorts. Understanding patterns of dental service use over time can be used to help inform service delivery policies that promote and facilitate appropriate use of dental services.

Performance of the UKPDS Outcomes Model for Prediction of Myocardial Infarction and Stroke in the ADDITION-Europe Trial Cohort

ObjectivesWe assessed the performance of the UK Prospective Diabetes Study (UKPDS) outcomes model in predicting the risk of myocardial infarction (MI) and stroke in the Anglo-Danish-Dutch Study of Intensive Treatment in People with Screen Detected Diabetes in Primary Care (ADDITION-Europe) a trial cohort of patients with screen-detected type 2 diabetes from the United Kingdom, Denmark, and The Netherlands. MethodsWe estimated the 5-year accumulated risk of MI and stroke for 2899 screen-detected people with type 2 diabetes by using the UKPDS outcomes model (version 1.3). We compared the predicted and actual risks by country and by intervention group (routine care; intensive multifactorial treatment). We assessed discrimination and goodness of fit by using area under receiver operating characteristic curves and the Hosmer-Lemeshow chi-square test. Multiple imputations were used to overcome missing data. ResultsThe UKPDS outcomes model overestimated the risk of MI and stroke. Mean predicted/actual ratios of 5-year accumulated risk were 2.31 for MI in the routine care group and 3.97 in the intensive multifactorial treatment group and 1.59 and 1.48 for stroke, respectively. The differences in absolute risk between the intervention groups were underestimated for MI (observed vs. predicted: 0.0127 vs. 0.0009) and slightly overestimated for stroke (−0.0013 vs. −0.0004). The area under the receiver operating characteristic curve was 0.72 (95% confidence interval 0.66–0.79) for MI and 0.70 (95% confidence interval 0.64–0.77) for stroke. The Hosmer-Lemeshow test statistic was nonsignificant in all groups. The model performed better in absolute risk prediction in Denmark and the United Kingdom than in The Netherlands. ConclusionsThe UKPDS outcomes model has moderate discriminatory ability in the ADDITION-Europe trial cohort but overestimated absolute risk. The model may need updating for cardiovascular disease risk prediction in contemporary diabetes populations where patients may be diagnosed earlier in the disease trajectory and in whom cardiovascular risk is therefore lower.

Refining the trauma triage algorithm at an Australian major trauma centre: derivation and internal validation of a triage risk score.

To derive and internally validate a clinical prediction rule for trauma triage. Ambulance presentations requiring trauma team activation between 2007 and 2011 at a single inner city major trauma centre were analysed. The primary outcome was major trauma, defined as Injury Severity Score >15, intensive care unit admission or in-hospital death. Demographic details, vital signs on arrival at hospital, mechanism of injury and injured body regions were used in the modelling process. Multivariable logistic regression was used on a randomly selected derivation sample. Receiver operating characteristic (ROC) analysis and Hosmer-Lemeshow tests were used to assess the discrimination and calibration of the derived model. The model was further tested using bootstrapping cross-validation. A total of 3027 patients were identified. Predictors selected for the prediction model were age ≥65years (OR 1.58, 95%CI 1.08-2.32, p=0.02), abnormal vital signs (OR 3.72, 95%CI 2.64-5.25), Glasgow Coma Scale score ≤13 (OR 14, 95%CI 9.23-23.34 p<0.001), penetrating injury (OR 5.13, 95%CI 2.76-9.54, p<0.001), multiregion injury (OR 4.72 95%CI 3.45-6.46, p<0.001), falls (OR 1.51 95%CI 1.06-2.15, p=0.02) and motor vehicle crashes (OR 0.56, 95%CI 0.35-0.90, p=0.02). The ROC area under the curve (AUC) for the final model was 0.85 (95%CI 0.83-0.87) with a Hosmer-Lemeshow test statistic p=0.83. Bootstrapping cross-validation demonstrated an identical AUC. We have derived and internally validated a trauma risk prediction rule using trauma registry data. This may assist with the formulation of revised local and regional trauma triage protocols. External validation is required before implementation.

Deriving a prediction rule for short stay admission in trauma patients admitted at a major trauma centre in Australia

IntroductionThe aim of this study was to derive and internally validate a prediction rule for short stay admissions (SSAs) in trauma patients admitted to a major trauma centre.MethodsA retrospective study...

Calibration of EuroSCORE II

We read with great interest the recent development and internal validation of EuroSCORE II [1] in the anticipation of greater methodological rigour and reporting than for the original study describing the development of EuroSCORE I [2]. In general, we feel that there has been an improvement in reporting, however, like others we find the description of the internal validation of EuroSCORE II somewhat confusing, and subsequent responses have highlighted flaws in evaluating the calibration of EuroSCORE II [3, 4]. In particular, it is unclear, given that the authors randomly split the cohort into development and internal validation data sets, why they included an undefined portion of the development data set in calculating the Hosmer–Lemeshow test statistic, and did not just use the validation data set. Furthermore, it is disappointing that a Hosmer–Lemeshow test giving a P-value that only just exceeded the Holy Grail of 0.05 was taken to indicate good calibration; it is far from convincing evidence that EuroSCORE II is well-calibrated. Disappointingly, recent correspondence bringing this to the attention of interested readers committed the cardinal sin of associating statistical significance (P = 0.05) with clinical significance [3]. We urge authors to be clear when describing the results of a statistical test, particularly those without an associated effect measure, and refrain from inferring clinical importance on the basis of the magnitude of a single P-value. Similarly, a second correspondent who highlighted flaws of the Hosmer–Lemeshow analysis made another disappointing statement, namely that ‘statistical results are either significant or non-significant, black or white, there is no grey in statistics’ [4]. Such a statement is misinformed and discouraging; statistics is ’entirely’ about this grey zone. The foundations of statistics lie in probability theory where one is solely concerned with capturing uncertainty. Where variability in measurements exists, we can never be entirely confident that our results are not merely an artefact of the data. Furthermore, describing statistical results as significant or non-significant is highly uninformative and should be avoided in preference to measures of effect including accounting for uncertainty by reporting appropriate confidence intervals. The Hosmer–Lemeshow test has limited usefulness and provides no information on any overor under-prediction of a risk prediction model [5]. It would be more useful for authors to present calibration plots, and essential that they restrict this analysis to the validation data set only. Such plots provide a highly informative graphical display of calibration, which can be accompanied by estimates of the calibration slope with a confidence interval (a slope of 1 with intercept 0 indicates perfect discrimination) [6]. In the light of the unimpressive calibration results of EuroSCORE II, calibration plots (including examining calibration by key prognostic factors such as age) would provide an insight into where the model is miscalibrated.

Assessing the applicability of scoring systems for predicting postoperative nausea and vomiting

We have validated two scoring systems for predicting postoperative nausea and vomiting, derived by Apfel et al. and Koivuranta et al. from 1388 adult inpatients undergoing a wide range of surgical procedures. The predictive accuracy of the scoring systems was evaluated in terms of the ability to discriminate between patients with and without postoperative nausea and vomiting (discrimination) and agreement between observed and predicted outcomes (calibration). Discrimination and calibration were less than expected based on previous reports, with both scoring systems providing risk predictions that were too extreme. The area under the ROC curve was 0.63 for Apfel et al.'s scoring system and 0.66 for Koivuranta et al.'s scoring system. Neither of the scoring systems provided a risk threshold for administering anti-emetic prophylaxis that yielded satisfying results in terms of predictive values, sensitivity and specificity. Hence, in their original forms, the scoring systems do not guarantee accurate prediction of the risk of postoperative nausea and vomiting in other patient populations. Koivuranta et al.'s scoring system appears to be more robust across different populations.