Prospective validation and comparison of clinical prediction models for early trauma care: A multicentre cohort study

ObjectivesTo evaluate the profile of non-urgent patients triaged ‘green’, as part of a triage trial in the emergency department (ED) of a secondary care hospital in India. The secondary aim...

Background: Malaria is caused by parasites of genus Plasmodium. It remains a major public health concern around the world. Though various diagnostic tools are available, there is an urgent need to use a more sensitive diagnostic method for early diagnosis to prevent unwanted outcomes. Objectives: (i) To assess the prevalence of malaria in a tertiary care hospital in Mumbai (ii) To detect and speciate Plasmodium by Peripheral Blood Smear (PBS), Rapid Diagnostic Test (RDT) and real-time Polymerase Chain Reaction (PCR) and to compare their performance characteristics. Methodology: A hospital-based diagnostic study was carried over a period of 18 months. A total of 550 non-duplicated blood samples from clinically suspected cases of malaria were collected and subjected to three tests- PBS by Field’s staining, RDT and PCR to detect Plasmodium species. The agreement and the differences between the three tests were analyzed and the statistical significance was assessed using Chi-square test. Results: Out of the 550 samples, 166 (30.2%) were positive for malaria by either of the three tests, of which 92 (55.42%), 119 (71.69%) and 161 (96.99%) samples were positive by PBS, RDT and PCR, respectively. P. vivax was the predominant Plasmodium species. The most significant finding was PCR detected mixed infections in 8.43% of cases whereas PBS and RDT could detect 0% and 1.20% of mixed infection, respectively. Considering PBS as gold standard, sensitivity of RDT and PCR was 100% and 97.83% and specificity was 94.10% and 84.50% respectively. Conclusions: In this study, 30% of the population was infected with malaria with P. vivax being the predominant Plasmodium species. PCR helped identify more mono and mixed infections than conventional methods and would be a helpful adjunct for malaria diagnosis in tertiary care setup.

<p><strong>Background:</strong> The disorder of the inner ear related to abnormally increased levels of endolymph in the membranous labyrinth is called Meniere’s disease. It is well studied that the prevalence of Meniere’s disease varies widely across regions. It was found to be 13% to 42% around the world in various places. It is important to know the incidence in India to know the population suffering and to plan better treatment strategies to cater to rehabilitation strategies. Hence, the current study was planned to estimate a load of Meniere’s disease in a tertiary care hospital in Mumbai.</p><p><strong>Methods:</strong> A retrospective analysis of the case reports was undertaken. As per American academy of otolaryngology-head and neck surgery (AAO-HNS, 2020), the patients were separated as definite and probable Meniere’s disease.</p><p><strong>Results:</strong> The descriptive analysis revealed that 6 in 1000 patients reporting for hearing evaluation in a tertiary care hospital are suffering from Meniere’s disease. There was a female preponderance found in the patients with a gender ratio of 1.85:1.</p><p><strong>Conclusions:</strong> The incidence of Meniere’s disease in a tertiary care hospital in Mumbai was found to be 0.61% of all the patients tested for hearing impairment.</p>

Introduction: Hepatitis B is a vaccine preventable disease caused by Hepatitis B virus which is a partially double stranded virus. Sharp injuries are the most common occupational injuries among healthcare workers (HCW). HCW are prone to sharp injuries during work, and infection after exposure. Materials and Methods: Anti-HBs titres were measured in 231 HCW in a tertiary care hospital in Mumbai. Result: Eighty six males and 145 females participated in the study. Out of these 41 were vaccine non compliant; 37 were partially immunized and 4 were not immunized. 32 out of 231 individuals tested had titres below 10 mIU/ml (10 mIU/ml considered to be the minimal protective titre). Among the individuals who had titres <10 mIU/ml, 17 were medical assistants, 10 were nursing officers and 5 were doctors. Out of these there were 04 who had not received any HepB vaccination in the past. The rest had received either complete or partial immunization. All those who had completed primary vaccination series and had titres <10 mIU/ml (n = 14) had completed primary vaccination more than 5 years ago. Conclusions: In our study 82.2% were vaccine compliant and 27.8% were vaccine non compliant. Out of 231 individuals 32 (13.85%) had titres <10 mIU/ml. This study highlights the need for HCW to undergo anti-HBs titres post primary vaccination and after 5 years of primary vaccination to identify susceptible individuals. The HCW with non protective titres should be give Hep B booster and antibody levels should be tested to confirm protective titres.

Background: End-stage kidney disease is associated with poor prognosis. Health care professionals must be prepared to address end-of-life issues and identify those at high risk for dying. A 6-month mortality prediction model for patients on dialysis derived in the United States is used but has not been externally validated. Aim: We aimed to assess the external validity and clinical utility in an independent cohort in Canada. Design: We examined the performance of the published 6-month mortality prediction model, using discrimination, calibration, and decision curve analyses. Setting/participants: Data were derived from a cohort of 374 prevalent dialysis patients in two regions of British Columbia, Canada, which included serum albumin, age, peripheral vascular disease, dementia, and answers to the “the surprise question” (“Would I be surprised if this patient died within the next year?”). Results: The observed mortality in the validation cohort was 11.5% at 6 months. The prediction model had reasonable discrimination (c-stat = 0.70) but poor calibration (calibration-in-the-large = −0.53 (95% confidence interval: −0.88, −0.18); calibration slope = 0.57 (95% confidence interval: 0.31, 0.83)) in our data. Decision curve analysis showed the model only has added value in guiding clinical decision in a small range of threshold probabilities: 8%–20%. Conclusion: Despite reasonable discrimination, the prediction model has poor calibration in this external study cohort; thus, it may have limited clinical utility in settings outside of where it was derived. Decision curve analysis clarifies limitations in clinical utility not apparent by receiver operating characteristic curve analysis. This study highlights the importance of external validation of prediction models prior to routine use in clinical practice.

To realize the potential of precision medicine, predictive models should be integrated within the framework of decision analysis, such as the decision curve analysis (DCA). To date, its application has required individual patient data (IPD) that are often unavailable. Performing DCA using aggregate data without requiring IPD may advance the goals of precision medicine. We present a statistical framework demonstrating that DCA can be conducted by using only the mean and standard deviation (SD) from the raw probabilities of the predictive model. We tested our theoretical framework by performing extensive simulations and comparing the aggregate-based DCA with IPD DCA. The latter was conducted using IPD from four predictive models that employed logistic regression, Coxor competing risk time-to-event modeling including(a) statins for primary prevention of cardiovascular disease (n = 4859), (b) hospice referral for terminally ill patients (n = 9104), (c) use of thromboprophylaxis for preventing venous thromboembolismin patients with cancer (n = 425)and (d) prevention of sinusoidal obstruction syndromeafter hematopoietic cell transplantation (SCT) (n = 80). Simulations assuming perfect calibration showed that regardless of which probability distributions informed the predictive models, the differences in DCA were negligible. Similarly, for the adequately powered models, the results of DCA based on the summary data were similar to IPD-derived DCA. The inherent instability of the predictive models, based on the smaller sample sizes, resulted in a somewhat larger discrepancy between aggregate and IPD-based DCA. DCA informed by adequately powered and well-calibrated models using only summary statistical estimates (mean and SD) approximates well models using IPD. Use of aggregate data will facilitate broader integration of predictive with decision modeling toward the goals of individualized decision-making.

This exploratory study reports the prevalence of lipohypertrophy (LH) in insulin-treated type 2 diabetes patients attending a private tertiary hospital in Mumbai, India, and explores the factors associated with this condition. A total of 96 patients attending the outpatient department, completed an interviewer-administered questionnaire followed by an examination of insulin injection sites for the presence of LH by a trained diabetes nurse educator. Nearly 68% of the respondents had LH. Most patients (82%) had LH on at least two injection sites, with the frontal areas of the right and left side of the abdomen being the most common sites. Despite a high prevalence of LH, the majority of patients were unaware that they had the condition (77%) and could not identify the causes (74%). Half of the patients reported awareness about the different sites for injecting insulin, injection techniques, and the need for rotating injection sites. Only 26% always rotated injection sites, while 16% changed needles more than half of the time in the week prior to the interview. Bivariate analysis found higher number of years with diabetes and on insulin and a higher insulin dose per day as significant factors (p value < 0.05). Logistic regression showed a strong relationship between the presence of LH and the number of years on insulin (p value < 0.05). The number of years on insulin for patients with and without LH averaged 8.2 and 3.7, respectively. Findings highlight a gap in patients’ knowledge and management of LH, and call for a hospital-wide intervention encompassing patient education and provider sensitization.

The Incidence, Aetiology and Antimicrobial Susceptibility of Central Line-Associated Bloodstream Infections in Intensive Care Unit Patients at a Private Tertiary Care Hospital in Mumbai, India

BackgroundA predictive nomogram model was established for the prognosis of cervical intraepithelial neoplasia 2 (CIN2).MethodsThis study was based on the research data of CIN2 obtained from the Shanxi CIN cohort study from 2019 to 2022. We conducted a cross-sectional analysis of 512 female patients with CIN2. Subsequently, the least absolute shrinkage and selection operator (LASSO) regression, along with univariate and multivariate regression analyses, were conducted to identify five risk factors associated with CIN2 prognosis. These factors include age at first sexual activity, ThinPrep cytologic test (TCT) results, Human papillomavirus (HPV) infection type, lesion area detected by colposcopy, and acetowhitening thickness. A predictive model was constructed employing R software. Receiver operating characteristic (ROC) curve and resampling methods were employed to evaluate the predictive model in terms of accuracy and calibration. Decision curve analysis (DCA) was performed to assess its clinical application value.ResultsWomen with CIN2 (n = 512) aged 19–65 were included in the study; after 6 months of follow-up, 185 showed lesion regression, and 336 showed lesion persistence or progression. The factors for the predictive model included age of sexual activity (P = 0.005), multiple sexual partners (P = 0.076), TCT results (P < 0.0001), HPV infection (P = 0.0025), lesion area (P < 0.0001), and the thickness of acetic acid stain (P < 0.0001). Subsequent ROC curve analysis showed the respective sensitivity and specificity of the predictive model to be 0.827 and 0.708. Finally, DCA, used to assess the predictive value of the 5-factor CIN2 regression predictive model, was higher than the combined predictive model of HPV and TCT.ConclusionThe study could successfully establish a model for predicting the pathological regression status of CIN2 patients.

Veno-arterial extracorporeal membrane oxygenation (VA-ECMO) is a critical life support technology for severely ill patients. Despite its benefits, patients face high costs and significant mortality risks. To improve clinical decision-making, this study aims to develop a non-invasive, efficient artificial intelligence (AI)-enabled model to predict the risk of mortality within 28 days post-weaning from VA-ECMO. A multicenter, retrospective cohort study was conducted across five hospitals in China, including all the patients who received VA-ECMO support between January 2020 and January 2024. Based on the innovatively selected 25 easily obtainable patient examination features as potentially relevant, this study involved developing ten predictive models using both classical and advanced machine learning techniques. The model’s performance is evaluated using various statistical metrics and the optimal predictive model are identified. Feature correlations are analyzed using Pearson correlation coefficients, and SHapley Additive exPlanations (SHAP) are employed to interpret feature importance. Decision curve analysis is used to evaluate the clinical utility of the predictive models. The study included 225 patients, with 66 patients from one hospital forming the training cohort. Three validation cohorts were used: internal validation with 16 patients from the training hospital and external validation with 30 and 60 patients from the other 4 hospitals. The random forest model emerged as the best predictor of 28-day mortality, achieving an AUROC of 1.00 in the training cohort and 1.00, 0.97, and 0.93 in the three validation cohorts, respectively. Despite the limited training data, the developed model, eCMoML, demonstrated high accuracy, generalizability and reliability. The model will be available online for immediate use by clinicians. The eCMoML model, validated in a multicenter cohort study, offers a rapid, stable, and accurate tool for predicting 28-day mortality post-VA-ECMO weaning. It has the potential to significantly enhance clinical decision-making, helping doctors better assess patient prognosis, optimize treatment plans, and improve survival rates.

ObjectiveTo develop and validate models to predict which endometriosis patients are likely to experience pain reduction following therapeutic laparoscopy using intraoperative findings and patient characteristics.DesignA retrospective secondary data analysis with patient workshops.SettingAnalysis of a UK nationwide, specialist centre, surgical database (British Society for Gynaecological Endoscopy, BSGE) (2013–2019, N=9171) and two research databases, MEDAL (2011–2013, N=667) and LUNA (1998–2005, N=592) for exploratory analyses and external validation.ParticipantsDatabase patients had laparoscopically confirmed (BSGE) or clinically suspected endometriosis (MEDAL, LUNA) and ranged from 16 to 65 years. Patient workshops included UK-wide endometriosis patients from the community, secondary care doctors and endometriosis nurses.Primary and secondary outcome measuresFollowing model development and internal validation, primary outcome measures included model performance statistics for discrimination (C-statistic) and calibration (calibration slope and calibration-in-the-large) for pain-improvement models for each of the five clinically meaningful pain domains. Secondary outcome measures included performance statistics for externally validated models and net benefit (using decision curve analysis).ResultsFollowing internal validation for dyspareunia (pain during sexual intercourse), non-cyclical pelvic pain (NPP), dyschezia (painful defecation) and quality of life our models showed good discrimination ability with C-statistics of 0.768, 0.750, 0.808 and 0.792, respectively. Significant increases in odds of pain relief were associated with trying to conceive for less than 18 months, any treated endometriosis of the ovary or uterosacral ligament or hysterectomy at the time of laparoscopy. For those models for which sufficient data were available to do external validation, dyspareunia and NPP showed good ability to predict pain reduction following surgery with C-statistics of 0.759 and 0.741, respectively, but after external validation only the model for dyspareunia good discriminatory ability (C-statistic=0.718). Despite this, decision curve analysis indicated some net benefit for all externally validated models.ConclusionsClinical prediction models can help identify women who will experience pain reduction after therapeutic laparoscopy, but more work is required to externally validate the current models. Removal of ovarian and utero-sacral ligament endometriosis appears to convey pain relief after surgery, whereas removal of superficial peritoneal endometriosis does not.

Clinical prediction rules could help identify children at risk of slowed growth after an episode of diarrheal illness, and these rules may be generalizable to all children, regardless of diarrhea status.

Background and purpose — External validation of machine learning (ML) prediction models is an essential step before clinical application. We assessed the proportion, performance, and transparent reporting of externally validated ML prediction models in orthopedic surgery, using the Transparent Reporting for Individual Prognosis or Diagnosis (TRIPOD) guidelines. Material and methods — We performed a systematic search using synonyms for every orthopedic specialty, ML, and external validation. The proportion was determined by using 59 ML prediction models with only internal validation in orthopedic surgical outcome published up until June 18, 2020, previously identified by our group. Model performance was evaluated using discrimination, calibration, and decision-curve analysis. The TRIPOD guidelines assessed transparent reporting. Results — We included 18 studies externally validating 10 different ML prediction models of the 59 available ML models after screening 4,682 studies. All external validations identified in this review retained good discrimination. Other key performance measures were provided in only 3 studies, rendering overall performance evaluation difficult. The overall median TRIPOD completeness was 61% (IQR 43–89), with 6 items being reported in less than 4/18 of the studies. Interpretation — Most current predictive ML models are not externally validated. The 18 available external validation studies were characterized by incomplete reporting of performance measures, limiting a transparent examination of model performance. Further prospective studies are needed to validate or refute the myriad of predictive ML models in orthopedics while adhering to existing guidelines. This ensures clinicians can take full advantage of validated and clinically implementable ML decision tools.

Skeletal muscle is an important organ strongly associated with prognosis in critically ill patients. 60-day mortality represents a key endpoint for evaluating the transitional phase from ICU survival to functional recovery, yet the association between ultrasound-derive muscle indices at ICU admission and 60-day mortality remains unclear. This dual-center, prospective observational study was performed from January to December 2024. Four ultrasound indices of muscle quantity [thickness and cross-sectional area of the rectus femoris (RF-TH and RF-CSA), thickness of the vastus intermedius (VI-TH), thickness of the quadriceps femoris (QF-TH)], and one index of quality (tibialis anterior pennation angle, TA-PA) were assessed within 24 h of ICU admission. The primary endpoint was 60-day all-cause mortality. To identify predictive factors, multivariable logistic regression analysis was employed. Additionally, a predictive nomogram model was developed. A total of 247 critically ill adult patients were included, with a median age of 61.0 years (51.0-72.0), including 171 males (69.20%). During 60-day follow-up, 53 patients (21.50%) died. Compared with the survivors, the non-survivors exhibited significantly lower RF-TH (0.54 [0.43-0.70] vs. 0.64 [0.48-0.85]cm; P = 0.022), VI-TH (0.64 [0.47-0.86] vs. 0.77 [0.64-1.04]cm; P = 0.002), QF-TH (1.32 [0.96-1.63] vs. 1.48 [1.24-1.97]cm; P =0.005) and TA-PA (7.08[5.63-8.48] vs. 7.94[6.62-9.32]; P =0.002). Among them, TA-PA was identified as an independent predictor of 60-day mortality (β = -0.216, P = 0.035). TA-PA, in conjunction with APACHE II score and prealbumin level, constructed a predictive nomogram, with a consistency index (C-index) of 0.785 (95%CI: 0.728-0.834). Calibration assessed by Spiegelhalter Z-test showed P > 0.05 indicating adequate calibration for the predicted and observed models. Decision curve analysis (DCA) confirmed that the nomogram prediction model had good clinical benefits. Internal validation demonstrated stable model performance, with a concordance statistic (C-index) of 0.785. Compared to the ROC curve based solely on APACHE II score, the nomogram exhibited significantly higher area under the curve (AUC) (0.785, 95%CI: 0.728-0.834 vs. 0.726, 95%CI: 0.666-0.781, P = 0.023). Lower TA-PA at ICU admission is an independent predictor of 60-day mortality. Incorporating this ultrasound-derived muscle quality index with APACHE II score and prealbumin level may improve early risk stratification in critically ill patients.

Prediction model for asymptomatic carotid atherosclerosis using retinal microvascular intelligent analysis: A retrospective study