Predictors Of Hospital Admission Research Articles

1303-P: Comparative Performance of a Recurrent Neural Network (RNN) and Logistic Regression (LR) Model to Predict Diabetic Ketoacidosis (DKA) among Youth Postdiagnosis with Type 1 Diabetes (T1D)

Preventing dangerous and costly episodes of DKA is a goal of diabetes care, but clinicians lack tools to predict DKA events. We sought to compare performance characteristics of an RNN model with that of a LR model to predict hospital admission for DKA in the next 180 days. We developed an RNN model using training (n=1453) and testing (n=1530) datasets restricted to youth <18 yo with T1D >30 days. In this sample, 6% (90/1530) of youth were admitted for DKA over 180 days. The model considered >500 features per quarter from discrete medical record data and free-text clinical documents. For comparative purposes, we developed an LR model using a testing dataset derived from the same sample (n=1388). The LR model included 10 evidence-based variables derived from critical review of the literature: number of previous DKAs (p<0.001), most recent A1c (p<0.001), age, gender, race, ethnicity, duration of T1D, days since most recent A1c, multiple daily injections vs. pump therapy, and public vs. private insurance. We used model output parameters to produce probability of DKA admission. RNN yielded an AUC=0.77 [95% CI: 0.71,0.83] while LR produced an AUC=0.79 [95% CI: 0.74,0.85]. We generated rank-ordered lists of the 90 youth with highest probability of DKA admission for both models and measured performance at different thresholds in rank-ordered lists. For ranks 1-5, RNN yielded precision (Pr)=100% (all 5 members on the list experienced a DKA admission) and LR yielded Pr=80%. For ranks 1-10, RNN yielded Pr=100% and LR yielded Pr=70%. For ranks 1-25, RNN yielded Pr=48% and LR yielded Pr=44%. An RNN model, in terms of precision, performs favorably relative to LR in predicting future DKA for a rank-ordered list of youth. Patient lists rank-ordered by DKA risk can be used by clinicians to identify youth with T1D for intensive intervention. Disclosure D.D. Williams: None. S. Dass: None. J. Bass: Employee; Self; Cyft, Inc. S.R. Patton: None. S.N. Mehta: None. R. McDonough: None. C. Mullaney: Employee; Self; Cyft. L. Davolio: Employee; Self; Cyft Inc. M.A. Clements: Consultant; Self; Glooko, Inc. Other Relationship; Self; Glooko, Inc. Funding The Leona M. and Harry B. Helmsley Charitable Trust (G-2017PG-T1D019)

Predicting hospital admission for older emergency department patients: Insights from machine learning

BackgroundEmergency departments (ED) are a portal of entry into the hospital and are uniquely positioned to influence the health care trajectories of older adults seeking medical attention. Older adults present to the ED with distinct needs and complex medical histories, which can make disposition planning more challenging. Machine learning (ML) approaches have been previously used to inform decision-making surrounding ED disposition in the general population. However, little is known about the performance and utility of ML methods in predicting hospital admission among older ED patients. We applied a series of ML algorithms to predict ED admission in older adults and discuss their clinical and policy implications. Materials and methodsWe analyzed the Canadian data from the interRAI multinational ED study, the largest prospective cohort study of older ED patients to date. The data included 2274 ED patients 75 years of age and older from eight ED sites across Canada between November 2009 and April 2012. Data were extracted from the interRAI ED Contact Assessment, with predictors including a series of geriatric syndromes, functional assessments, and baseline care needs. We applied a total of five ML algorithms. Models were trained, assessed, and analyzed using 10-fold cross-validation. The performance of predictive models was measured using the area under the receiver operating characteristic curve (AUC). We also report the accuracy, sensitivity, and specificity of each model to supplement performance interpretation. ResultsGradient boosted trees was the most accurate model to predict older ED patients who would require hospitalization (AUC = 0.80). The five most informative features include home intravenous therapy, time of ED presentation, a requirement for formal support services, independence in walking, and the presence of an unstable medical condition. ConclusionTo the best of our knowledge, this is the first study to predict hospital admission in older ED patients using a series of geriatric syndromes and functional assessments. We were able to predict hospital admission in older ED patients with good accuracy using the items available in the interRAI ED Contact Assessment. This information can be used to inform decision-making about ED disposition and may expedite admission processes and proactive discharge planning.

Medication Regimen Complexity Index Prediction of Adverse Drug Reaction-Related Hospital Admissions.

The relationship between the medication regimen complexity index (MRCI) and adverse drug reaction (ADR)-related hospital admissions has not yet specifically been investigated. To evaluate the MRCI and compare with medication count for prediction of ADR-related hospital admissions in older patients. This was a retrospective analysis of a prospectively collected convenience sample of 768 unplanned medical admissions of Australians aged 65 years old and older. The sample consisted of 115 (15.0%) ADR-related unplanned hospital admissions and 653 (85.0%) non-ADR-related unplanned medical admissions. The MRCI score was calculated from the medical records and analyzed to predict ADR-related hospital admissions. The cohort had a median age of 81 years, 5 comorbidities, and 11 medications, with a slight majority of women. The MRCI score was not significantly different in patients who had ADR-related admissions compared with other medical admissions-38.5 versus 34.0, respectively; Wilcoxon Rank Sum test W = 33 522; P = 0.067. The medication count was significantly different between the ADR-related admissions compared with other medical admissions: 12 versus 10; W = 32 508; P = 0.021. However, the medication count was not a strong predictor of ADR-related admissions; unadjusted odds ratio = 1.044; 95% CI = 1.006-1.084. The MRCI score did not discriminate between ADR-related admissions and other medical admissions despite taking time to calculate with potential for inconsistent application. Medication count is more readily applicable with marginally greater relevance in this cohort; however, both measures do not appear to be useful when used alone for clinicians to identify patients at risk of ADRs.

Predicting Hospital Admission and Prolonged Length of Stay in Older Adults in the Emergency Department: The PRO-AGE Scoring System

We developed prediction models for hospital admission and prolonged length of stay in older adults admitted from the emergency department (ED). This was a retrospective cohort study of patients aged 70 years or older who visited a geriatric ED in Brazil (N=5,025 visits). We randomly allocated participants to derivation and validation samples in a 2:1 ratio. We then selected 21 variables based on their clinical relevance and generated models to predict the following outcomes: hospital admission and prolonged length of stay, defined as the upper tertile of hospital stay. We used backward stepwise logistic regressions to select our final predictors and developed risk scoring systems based on the relative values of their β coefficients. Overall, 57% of the participants were women, 31% were hospitalized, and 1% died in the hospital. The upper tertile of hospital stay was greater than 7 days. Hospital admission was best predicted by a model including male sex, aged 90 years or older, hospitalization in the previous 6 months, weight loss greater than or equal to 5% in the previous year, acute mental alteration, and acute functional decline. The prediction of prolonged length of stay retained the same variables, except male sex, which was substituted for fatigue. The final scoring system reached areas under the receiver operating characteristic curve of 0.74 for hospital admission and 0.79 for prolonged length of stay, and their accuracies were confirmed in the validation models. The PRO-AGE scoring system predicted hospital admission and prolonged length of stay in older adults with good accuracy, using a simple approach and only 7 easily obtained clinical variables.

The Admit-AF risk score: A clinical risk score for predicting hospital admissions in patients with atrial fibrillation.

To develop and externally validate a risk score for all-cause hospital admissions in patients with atrial fibrillation. We used a prospective cohort of 2387 patients with established atrial fibrillation as derivation cohort. Independent risk factors were selected from a broad range of variables using the least absolute shrinkage and selection operator method fit to a Cox model. The risk score was validated in a separate prospective cohort of 1300 atrial fibrillation patients. The incidence of all-cause hospital admission was 19.1 per 100 person-years in the derivation cohort and it was 26.1 per 100 person-years in the validation cohort. The most important predictors for admission were age (75-79 years: adjusted hazard ratio (aHR), 1.34; 95% confidence interval (CI), 1.01-1.78; 80-84 years: aHR, 1.50; 95% CI, 1.11-2.03; ≥85 years: aHR, 1.88; 95% CI, 1.36-2.62), prior pulmonary vein isolation (aHR, 0.72; 95% CI, 0.58-0.88), hypertension (aHR, 1.16; 95% CI, 0.99-1.36), diabetes (aHR, 1.38; 95% CI, 1.17-1.62), coronary heart disease (aHR, 1.17; 95% CI, 1.02-1.36), prior stroke/transient ischaemic attack (aHR, 1.26; 95% CI, 1.18-1.47), heart failure (aHR, 1.19; 95% CI, 1.03-1.39), peripheral artery disease (aHR, 1.35; 95% CI, 1.08-1.67), cancer (aHR, 1.33; 95% CI, 1.12-1.57), renal failure (aHR, 1.17; 95% CI, 0.99-1.37) and previous falls (aHR, 1.40; 95% CI, 1.13-1.74). A risk score with these variables was well calibrated, and achieved a C-index of 0.64 in the derivation and 0.59 in the validation cohort. Multiple risk factors were associated with hospital admissions in atrial fibrillation patients. This prediction tool selects high-risk patients who may benefit from preventive interventions.

Clinically useful prediction of hospital admissions in an older population

BackgroundThe healthcare for older adults is insufficient in many countries, not designed to meet their needs and is often described as disorganized and reactive. Prediction of older persons at risk of admission to hospital may be one important way for the future healthcare system to act proactively when meeting increasing needs for care. Therefore, we wanted to develop and test a clinically useful model for predicting hospital admissions of older persons based on routine healthcare data.MethodsWe used the healthcare data on 40,728 persons, 75–109 years of age to predict hospital in-ward care in a prospective cohort. Multivariable logistic regression was used to identify significant factors predictive of unplanned hospital admission. Model fitting was accomplished using forward selection. The accuracy of the prediction model was expressed as area under the receiver operating characteristic (ROC) curve, AUC.ResultsThe prediction model consisting of 38 variables exhibited a good discriminative accuracy for unplanned hospital admissions over the following 12 months (AUC 0.69 [95% confidence interval, CI 0.68–0.70]) and was validated on external datasets. Clinically relevant proportions of predicted cases of 40 or 45% resulted in sensitivities of 62 and 66%, respectively. The corresponding positive predicted values (PPV) was 31 and 29%, respectively.ConclusionA prediction model based on routine administrative healthcare data from older persons can be used to find patients at risk of admission to hospital. Identifying the risk population can enable proactive intervention for older patients with as-yet unknown needs for healthcare.

The Application of Deep Learning in Cancer Prognosis Prediction.

Deep learning has been applied to many areas in health care, including imaging diagnosis, digital pathology, prediction of hospital admission, drug design, classification of cancer and stromal cells, doctor assistance, etc. Cancer prognosis is to estimate the fate of cancer, probabilities of cancer recurrence and progression, and to provide survival estimation to the patients. The accuracy of cancer prognosis prediction will greatly benefit clinical management of cancer patients. The improvement of biomedical translational research and the application of advanced statistical analysis and machine learning methods are the driving forces to improve cancer prognosis prediction. Recent years, there is a significant increase of computational power and rapid advancement in the technology of artificial intelligence, particularly in deep learning. In addition, the cost reduction in large scale next-generation sequencing, and the availability of such data through open source databases (e.g., TCGA and GEO databases) offer us opportunities to possibly build more powerful and accurate models to predict cancer prognosis more accurately. In this review, we reviewed the most recent published works that used deep learning to build models for cancer prognosis prediction. Deep learning has been suggested to be a more generic model, requires less data engineering, and achieves more accurate prediction when working with large amounts of data. The application of deep learning in cancer prognosis has been shown to be equivalent or better than current approaches, such as Cox-PH. With the burst of multi-omics data, including genomics data, transcriptomics data and clinical information in cancer studies, we believe that deep learning would potentially improve cancer prognosis.

Serum Copeptin levels in the emergency department predict major clinical outcomes in adult trauma patients

BackgroundEarly prognostication in trauma patients is challenging, but particularly important. We wanted to explore the ability of copeptin, the C-terminal fragment of arginine vasopressin, to identify major trauma, defined as Injury Severity Score (ISS) > 15, in a heterogeneous cohort of trauma patients and to compare its performances with lactate. We also evaluated copeptin performance in predicting other clinical outcomes: mortality, hospital admission, blood transfusion, emergency surgery, and Intensive Care Unit (ICU) admission.MethodsThis single center, pragmatic, prospective observational study was conducted at Arcispedale Santa Maria Nuova, a level II trauma center in Reggio Emilia, Italy. Copeptin determination was obtained on Emergency Department (ED) arrival, together with venous lactate. Different outcomes were measured including ISS, Revised Trauma Score (RTS), hospital and ICU admission, blood transfusion, emergency surgery, and mortality.ResultsOne hundred and twenty five adult trauma patients admitted to the ED between June 2017 and March 2018. Copeptin showed a good ability to identify patients with ISS > 15 (AUC 0.819). Similar good performances were recorded also in predicting other outcomes. Copeptin was significantly superior to lactate in identifying patients with ISS > 15 (P 0.0015), and in predicting hospital admission (P 0.0002) and blood transfusion (P 0.016). Comparable results were observed in a subgroup of patients with RTS 7.84.ConclusionsIn a heterogeneous group of trauma patients, a single copeptin determination at the time of ED admission proved to be an accurate biomarker, statistically superior to lactate for the identification of major trauma, hospital admission, and blood transfusion, while no statistical difference was observed for ICU admission and emergency surgery. These results, if confirmed, may support a role for copeptin during early management of trauma patients.

Police involvement, characteristics and outcomes of place of safety referrals in the Scottish Highlands.

To characterise police involvement with those detained under place of safety legislation and determine factors associated with admission to hospital. Place of safety referrals over a 1-year period were identified retrospectively and evaluated. Place of safety legislation is generally used with regard to concerns about suicide. Individuals are often removed from high-risk areas and referrals to police are frequently initiated by individuals themselves. A diagnosis of mental illness or personality disorder predicted hospital admission. Presence of senior nursing staff at assessment, but not the seniority of the doctor, was associated with discharge. Closer multiagency working is required as police are currently being recruited to fill a void between mental health services and the population they serve. Junior doctors require more senior support in making complex, and often risky, emergency management decisions with this population.

Alcohol use predicts emergency psychiatric unit admission for non-fatal suicidal behaviour in the Western Cape (South Africa): a case–control study

Objective: We aimed to describe patterns of substance use among patients admitted to an emergency psychiatric unit for non-fatal suicidal behaviour (NFSB) or another psychiatric reason and establish whether there were significant differences in patterns of substance use between the two groups.Methods: We employed a case–control design (N = 50) and collected data about participants' substance use in Cape Town, South Africa. Data were analysed using Chi-square and Mann–Whitney tests, factor analysis, and logistic regressions.Results: Prevalence of lifetime Alcohol Use Disorder (AUD) was 60% in the NFSB group and 28% in the control group. 12% of the NFSB group and 20% of the control group had a lifetime Tobacco Use Disorder. Prevalence of lifetime illicit Substance Use Disorder was 44% in the NFSB group and 60% in the control group. Hospital admission for NFSB was associated with: any past 24-hour alcohol use; quantity of past 24-hour alcohol use; quantity of past-month alcohol use; lifetime AUD; past 12-month AUD; and current AUD; and was not associated with the use of any other substances (p<.05). Past 12-month AUD was the best predictor of hospital admission for NFSB, controlling for, respectively, any past 24-hour alcohol use (aOR = 13.33, p = .023) and quantity of past 24-hour alcohol use (aOR = 9.01, p = .022)Conclusions: Patients admitted to emergency psychiatric units for NFSB have increased needs for the treatment of AUDs compared to patients admitted for another psychiatric emergency. Findings support calls for interventions to prevent NFSB among psychiatric patients with a history of AUD.Key pointsRates of substance use among patients admitted to emergency psychiatric units in South Africa were high compared to the general population.Hazardous alcohol use was uniquely associated with hospital admission for non-fatal suicidal behaviour compared to another psychiatric emergency.Tobacco use and illicit substance use were not associated with hospital admission for non-fatal suicidal behaviour compared to another psychiatric emergency.The association between hazardous alcohol use and hospital admission for non-fatal suicidal behaviour did not appear to be affected by demographic variables.Patients admitted to hospital for non-fatal suicidal behaviour have increased needs for the treatment of alcohol use disorders compared to other psychiatric patients.

Simulation modeling for predicting hospital admissions and bed utilisation

Demographic research of world population shows that societies are ageing. The ongoing changes in the population structure will require appropriate quantitative and qualitative adjustments in health services to meet the needs of society. Simulation methods turn out to be helpful in these kinds of analyses. In this paper, the authors present a case study on using discrete event simulation (DES) to support decision-making in the field of hospital bed management in the light of demographic changes. The case study was elaborated for one of the Polish district hospitals. A discrete simulation model was built to simulate admissions to two hospital wards: pediatric and geriatric. A series of experiments was carried out as based on real data extracted from the hospital database and forecasted demographic trends elaborated by The Central Statistical Office (CSO). The influence of demographic changes on hospital admissions in the chosen age-gender cohorts was explored, examining different variants of hospital bed availability. The results of the experiments show that demographic trends significantly influence healthcare admission and bed utilisation. The reduction in the number of admissions to the pediatric ward by about 6% results in a change in average bed utilisation from 57.90% to 54.06%. With about 12% more admissions to the geriatric ward, the change is from 68.88% to 75.59%.

Emergency hospital admission prediction using data mining techniques

Congestion in Emergency Departments (ED's) greatly affects the patients alongside negative outcomes. In this way it gets critical to discover creative strategies to improve patient stream and avoid stuffing. This should be possible with the assistance of various AI systems to foresee the emergency clinic affirmations from ED. This will help in asset arranging of clinic too. The paper comprises of overview of the various strategies utilized for expectation. The paper likewise comprises the usage aftereffect of three distinct strategies for the AI procedures alongside its outcome’s examination.

The 5-Factor Modified Frailty Index Predicts Complications, Hospital Admission, and Mortality Following Arthroscopic Rotator Cuff Repair

The purpose of this study is to evaluate the utility of the modified frailty index-5 (mFI-5) as a predictor for postoperative complications in patients undergoing arthroscopic rotator cuff repair (RCR). The National Surgical Quality Improvement Program database was queried for patients undergoing arthroscopic RCR between 2006 and 2016. The mFI-5, a 5-factor score comprising comorbid diabetes, hypertension, congestive heart failure, chronic obstructive pulmonary disease, and functional status limiting independence, was calculated for each patient. Multivariate logistic regression models were used to evaluate the mFI-5 score as a predictor for complications including medical complications, surgical-site infections, hospital admission, discharge to a facility, and mortality. In total, 24,477 patients met criteria for inclusion. The mFI-5 was a strong predictor for medical complications (P < .001), hospital admission (P < .001), length of stay (P= .007), and discharge to a facility (P= .001) but not surgical-site infections (P= .153). For each point increase in mFI-5 score, the risk for a medical complication increased by 66%, readmission by 52%, and adverse discharge by 45%. However, of all the measured complications, the mFI-5 was the strongest predictor for mortality, with the risk more than doubling for each increase in mFI-5 point (odds ratio 2.66, P= .025). The mFI-5 is a sensitive tool for predicting life-threatening medical complications, hospital admission, increased length of stay, adverse discharge, and mortality following arthroscopic RCR. The 5 comorbidities comprising the mFI-5 are easily obtained through the patient history, making it a practical clinical tool for identifying high-risk patients, informing preoperative counseling, and improving value-based health care. Level III, prognostic.

A validation of machine learning-based risk scores in the prehospital setting.

The triage of patients in prehospital care is a difficult task, and improved risk assessment tools are needed both at the dispatch center and on the ambulance to differentiate between low- and high-risk patients. This study validates a machine learning-based approach to generating risk scores based on hospital outcomes using routinely collected prehospital data. Dispatch, ambulance, and hospital data were collected in one Swedish region from 2016-2017. Dispatch center and ambulance records were used to develop gradient boosting models predicting hospital admission, critical care (defined as admission to an intensive care unit or in-hospital mortality), and two-day mortality. Composite risk scores were generated based on the models and compared to National Early Warning Scores (NEWS) and actual dispatched priorities in a prospectively gathered dataset from 2018. A total of 38203 patients were included from 2016-2018. Concordance indexes (or areas under the receiver operating characteristics curve) for dispatched priorities ranged from 0.51-0.66, while those for NEWS ranged from 0.66-0.85. Concordance ranged from 0.70-0.79 for risk scores based only on dispatch data, and 0.79-0.89 for risk scores including ambulance data. Dispatch data-based risk scores consistently outperformed dispatched priorities in predicting hospital outcomes, while models including ambulance data also consistently outperformed NEWS. Model performance in the prospective test dataset was similar to that found using cross-validation, and calibration was comparable to that of NEWS. Machine learning-based risk scores outperformed a widely-used rule-based triage algorithm and human prioritization decisions in predicting hospital outcomes. Performance was robust in a prospectively gathered dataset, and scores demonstrated adequate calibration. Future research should explore the robustness of these methods when applied to other settings, establish appropriate outcome measures for use in determining the need for prehospital care, and investigate the clinical impact of interventions based on these methods.

The effectiveness of a national early warning score as a triage tool for activating a rapid response system in an outpatient setting: A retrospective cohort study.

Rapid response system (RRS) efficacy and national early warning score (NEWS) performances have largely been reported in inpatient settings, with few such reports undertaken in outpatient settings.This study aimed to investigate NEWS validity in predicting poor clinical outcomes among outpatients who had activated the RRS using single-parameter criteria.A single-center retrospective cohort studyFrom April 1, 2014 to November 30, 2017 in an urban 350-bed referral hospital in JapanWe collected patient characteristics such as activation triggers, interventions, arrival times, dispositions, final diagnoses, and patient outcomes. Poor clinical outcomes were defined as unplanned intensive care unit transfers or deaths within 24 hours. Correlations between the NEWS and clinical outcomes at the time of deterioration and disposition were analyzed.Among 31 outpatients, the NEWS value decreased significantly after a medical emergency team intervention (median, 8 vs 4, P < .001). The difference in the NEWS at the time of deterioration and at disposition was significantly less in patients with poor clinical outcomes (median 3 vs 1.5, P = .03). The area under the curve (AUC) for the NEWS high-risk patient group at the time of deterioration for predicting hospital admission was 0.85 (95% confidence interval [CI], 0.67-1.0), while the AUC for the NEWS high-risk patient group at disposition for predicting poor clinical outcomes was 0.83 (95% CI, 0.62-1.0).The difference between the NEWS at the time of deterioration and at disposition might usefully predict admissions and poor clinical outcomes in RRS outpatient settings.

286 Predictors and Outcomes of Older Persons Attending the Emergency Department of a Large Acute Dublin Teaching Hospital

Abstract Background An emergency department (ED) visit is often considered a sentinel event for an older person due to elevated risks of deleterious outcomes. This prospective cohort study explored factors associated with disposition post ED index visit and unscheduled 30 day revisit. Methods Consecutive ED attendees aged ≥70 years, and Manchester Triage System 3-5, in a 6-month period, were eligible for inclusion. Participants underwent an interdisciplinary assessment by a Home FIRsT (Frailty Intervention &amp; Response Team) member. Logistic regression analyses were performed to identify factors predictive of admission post ED index visit and unscheduled revisit. Predictive validity of frailty instruments were performed using receiver operating characteristic (ROC) curves Results 1,156 ED attendances were included - 59% were female; median age was 80 years; 66% were discharged home post index visit; 17.8% had an unscheduled ED revisit within 30 days. Age and sex did not predict hospital admission. Hospital admission post ED visit was imperfectly predicted by mild-moderate frailty: Clinical Frailty Scale 5-6 (Odds Ratio (OR) 1.83 (95% CI 1.11 – 3.04), p=0.019); ‘Think Frailty’ 3 (OR 1.75 (95% CI 1.07 – 2.85), p=0.025) and 4 (OR 2.32 (95% CI 1.16 – 4.63), p=0.017). Paradoxically, higher frailty scores were not predictive. Cognitive impairment (4AT 1-3) similarly predicted admission (OR 1.62 (95% 1.16 – 2.27, p=0.005), while delirium (4AT ≥4) was the strongest predictor of admission (OR 5.87 (95% CI 3.17 – 10.88, p&lt;0.001). ROC curves of both frailty scales showed moderate ‘diagnostic accuracy’ for admission, but less so for unscheduled revisits. Conclusion Home FIRsT operationalised the assessment of frailty and delirium in the ED. Delirium is a strong predictor of admission post ED index visit, more than frailty status. Older persons have a high rate of 30-day unscheduled revisits; however, it is difficult to produce models from patient information available at index visit that can reliably predict unscheduled revisits.

Occupational lifting predicts hospital admission due to low back pain in a cohort of airport baggage handlers

PurposeTo examine if occupational lifting assessed as cumulative years as a baggage handler is associated with first-time hospital diagnosis or treatment for low back disorders.MethodsThis study is based on the Copenhagen Airport Cohort consisting of male baggage handlers performing heavy lifting every day and a reference group of unskilled men from the greater Copenhagen area during the period 1990–2012. We followed the cohort in the National Patient Register and Civil Registration System to obtain information on diagnoses, surgery, mortality, and migration. The outcomes were first-time hospital diagnosis or surgery for (1) lumbar disc herniation or (2) low back pain (LBP).ResultsBaggage handlers (N = 3473) had a higher incidence rate of LBP, but not of lumbar disc herniation, compared to the reference group (N = 65,702). Baggage handlers with longer employment had a higher incidence of LBP compared to baggage handlers with shorter employment. The linear association of cumulative years as a baggage handler on LBP was significantly increased with an incidence rate ratio of 1.16 (95% CI 1.07–1.25) for a 5-year increase of employment as baggage handler.ConclusionsIn this large cohort study, we found an increased incidence of LBP among baggage handlers compared to the reference group with indications of a dose–response relationship between years of employment and the outcome. For baggage handlers working on the apron, the incidence was particularly increased before introduction of technical lifting equipment, suggesting that preventive measures to reduce cumulative work load may have a positive effect.

Comparison of Machine Learning Techniques for Prediction of Hospitalization in Heart Failure Patients.

The present study aims to compare the performance of eight Machine Learning Techniques (MLTs) in the prediction of hospitalization among patients with heart failure, using data from the Gestione Integrata dello Scompenso Cardiaco (GISC) study. The GISC project is an ongoing study that takes place in the region of Puglia, Southern Italy. Patients with a diagnosis of heart failure are enrolled in a long-term assistance program that includes the adoption of an online platform for data sharing between general practitioners and cardiologists working in hospitals and community health districts. Logistic regression, generalized linear model net (GLMN), classification and regression tree, random forest, adaboost, logitboost, support vector machine, and neural networks were applied to evaluate the feasibility of such techniques in predicting hospitalization of 380 patients enrolled in the GISC study, using data about demographic characteristics, medical history, and clinical characteristics of each patient. The MLTs were compared both without and with missing data imputation. Overall, models trained without missing data imputation showed higher predictive performances. The GLMN showed better performance in predicting hospitalization than the other MLTs, with an average accuracy, positive predictive value and negative predictive value of 81.2%, 87.5%, and 75%, respectively. Present findings suggest that MLTs may represent a promising opportunity to predict hospital admission of heart failure patients by exploiting health care information generated by the contact of such patients with the health care system.

Resistin as a predictor of cardiovascular hospital admissions and renal deterioration in diabetic patients with chronic kidney disease

BackgroundHigh resistin levels have been associated with cardiovascular disease (CVD). Cardiovascular hospitalizations are common, especially in diabetic and renal impaired patients. The purpose of this study is to determine the role of serum resistin as a predictor of cardiovascular hospitalizations in type 2 diabetic patients with mild to moderate chronic kidney disease (CKD). MethodsWe conducted a prospective, observational study. 78 diabetic patients with mild to moderate CKD and no previous CVD were included. The population was divided in two groups: G-1 with cardiovascular related admission (n = 13) and G-2 without cardiovascular related admission (n = 65). A Student's t-test was conducted to determine correlations between laboratory findings and hospitalization. We used logistic regression to assess predictors of cardiovascular events requiring hospitalization and Cox regression to identify predictors of end-stage renal disease (ESRD). ResultseGFR, albumin, HbA1c, phosphorous, PTH, IR, CRP, resistin and active vitamin D, were related to cardiovascular admissions. In a multivariate regression model, resistin (OR = 2.074, p = 0.047) was an independent predictor of cardiovascular hospitalization. Cox regression showed that resistin (HR = 1.931, p = 0.031) and UACr (HR = 1.151, p = 0.048) were also independent predictors of renal disease progression. ConclusionResistin demonstrated to be valuable in predicting hospital admissions and progression to ESRD.

Prediction of hospital admissions and grade 3-4 toxicities in cancer patients 70years old and older receiving chemotherapy.

To predict chemotherapy toxicity and hospitalisations in elderly patients using clinical and laboratory parameters. Records of cancer patients 70years old or older who received adjuvant chemotherapy or first-line chemotherapy for a cancer in a single centre were reviewed. Factors associated with hospitalisations, grade 3-4 toxicities and dose reductions during treatment were evaluated. A total of 275 patients included in the study. Most patients (53.8%) were 70 to 75years old. One hundred and five patients (38.2%) had a hospital admission during or within a month after their chemotherapy treatment. The only factor associated with admissions in the multivariate analysis was ECOG performance status (PS) >1 (p=.008, odds ratio 2.66, 95% CI: 1.28-5.53) and hypoalbuminaemia approached significance. Grade 3 and 4 toxicities were associated with a lower creatinine clearance in the multivariate analysis (p=.01, odds ratio 0.98, 95% CI: 0.97-1.0), and dose reductions were associated with metastatic stage (p=.03, odds ratio 1.88, 95% CI: 1.05-3.35). A combined index with all four parameters was associated with all three outcomes of interest. ECOG PS, stage, albumin and creatinine clearance may be predictive of hospital admissions, grade 3-4 toxicities and dose reduction rates in cancer patients 70years old and older receiving chemotherapy.