Predicting the hyperglycemic crisis death (PHD) score: a new decision rule for emergency and critical care

Using artificial intelligence to predict adverse outcomes in emergency department patients with hyperglycemic crises in real time

BACKGROUND

Hyperglycemic crises are associated with high morbidity and mortality. Previous studies have proposed methods to predict adverse outcomes of patients in hyperglycemic crises; however, artificial intelligence (AI) has never been used to predict adverse outcomes. We implemented an AI model integrated with the hospital information system (HIS) to clarify whether AI could predict adverse outcomes.

METHODS

We included 2,666 patients with hyperglycemic crises from emergency departments (ED) between 2009 and 2018. The patients were randomized into a 70%/30% split for AI model training and testing. Twenty-two feature variables from the electronic medical records were collected. The performance of the multilayer perceptron (MLP), logistic regression, random forest, Light Gradient Boosting Machine (LightGBM), support vector machine (SVM), and K-nearest neighbor (KNN) algorithms was compared. We selected the best algorithm to construct an AI model to predict sepsis or septic shock, intensive care unit (ICU) admission, and all-cause mortality within 1 month. The outcomes between the non-AI and AI groups were compared after implementing the HIS and predicting the hyperglycemic crisis death (PHD) score.

RESULTS

The MLP had the best performance in predicting the three adverse outcomes, compared with the random forest, logistic regression, SVM, KNN, and LightGBM models. The areas under the curves (AUCs) using the MLP model were 0.852 for sepsis or septic shock, 0.743 for ICU admission, and 0.796 for all-cause mortality. Furthermore, we integrated the AI predictive model with the HIS to assist decision making in real time. No significant differences in ICU admission or all-cause mortality were detected between the non-AI and AI groups. The AI model performed better than the PHD score for predicting all-cause mortality (AUC 0.796 vs. 0.693).

CONCLUSIONS

A real-time AI predictive model is a promising method for predicting adverse outcomes in ED patients with hyperglycemic crises. Further studies recruiting more patients are warranted.

Mortality prediction in patients with hyperglycaemic crisis using explainable machine learning: a prospective, multicentre study based on tertiary hospitals

BACKGROUND

Experiencing a hyperglycaemic crisis is associated with a short- and long-term increased risk of mortality. We aimed to develop an explainable machine learning model for predicting 3-year mortality and providing individualized risk factor assessment of patients with hyperglycaemic crisis after admission.

METHODS

Based on five representative machine learning algorithms, we trained prediction models on data from patients with hyperglycaemic crisis admitted to two tertiary hospitals between 2016 and 2020. The models were internally validated by tenfold cross-validation and externally validated using previously unseen data from two other tertiary hospitals. A SHapley Additive exPlanations algorithm was used to interpret the predictions of the best performing model, and the relative importance of the features in the model was compared with the traditional statistical test results.

RESULTS

A total of 337 patients with hyperglycaemic crisis were enrolled in the study, 3-year mortality was 13.6% (46 patients). 257 patients were used to train the models, and 80 patients were used for model validation. The Light Gradient Boosting Machine model performed best across testing cohorts (area under the ROC curve 0.89 [95% CI 0.77-0.97]). Advanced age, higher blood glucose and blood urea nitrogen were the three most important predictors for increased mortality.

CONCLUSION

The developed explainable model can provide estimates of the mortality and visual contribution of the features to the prediction for an individual patient with hyperglycaemic crisis. Advanced age, metabolic disorders, and impaired renal and cardiac function were important factors that predicted non-survival.

TRIAL REGISTRATION NUMBER

ChiCTR1800015981, 2018/05/04.

Using neutrophil to lymphocyte ratio to predict discharge among geriatric patients with influenza infection in emergency department

The neutrophil-to-lymphocyte ratio (NLR) is used to predict the prognosis of various diseases, such as coronavirus disease 2019, community-acquired pneumonia, bacteremia, and endocarditis. However, NLR has never been reported to predict patient discharge in geriatric patients with influenza infection. This retrospective case-control study enrolled geriatric patients (≥65 years) with influenza virus infection who visited the emergency department of a medical center between January 01, 2010 and December 31, 2015. Demographic data, vital signs, past histories, influenza subtypes, outcomes, and disposition were analyzed. The optimal NLR cut-off value to predict patient discharge was determined using the Youden index. We also evaluated the accuracy of NLR in predicting patient discharge using logistic regression and receiver operating characteristic analysis. The study included 409 geriatric patients in the emergency department with a mean age of 79.5 years and an approximately equal sex ratio. NLR was significantly lower in the discharged group than in the nondischarged group (5.8 ± 3.7 vs 9.7 ± 8.4). Logistic regression revealed that patients with NLR ≤ 6.5 predicted discharge with an odds ratio of 3.62. The Hosmer-Lemeshow goodness-of-fit test was calculated as 0.36, and the adjusted area under the receiver operating characteristic was 0.75. The negative predictive value of NLR ≤ 6.5, to predict patient discharge, was 91.8%. NLR ≤ 6.5 is a simple and easy-to-obtain laboratory tool to guide the physicians to discharge geriatric patients with influenza infection in the crowded emergency department.

National Early Warning Score for predicting intensive care unit admission among elderly patients with influenza infections in the emergency department: an effective disposition tool during the influenza season

OBJECTIVE

During the influenza epidemic season, the fragile elderlies are not only susceptible to influenza infections, but are also more likely to develop severe symptoms and syndromes. Such circumstances may pose a significant burden to the medical resources especially in the emergency department (ED). Disposition of the elderly patients with influenza infections to either the ward or intensive care unit (ICU) accurately is therefore a crucial issue.

STUDY DESIGN

Retrospective cohort study.

SETTING AND PARTICIPANTS

Elderly patients (≥65 years) with influenza visiting the ED of a medical centre between 1 January 2010 and 31 December 2015.

PRIMARY OUTCOME MEASURES

Demographic data, vital signs, medical history, subtype of influenza, national early warning score (NEWS) and outcomes (mortality) were analysed. We investigated the ability of NEWS to predict ICU admission via logistic regression and the receiver operating characteristic (ROC) analysis.

RESULTS

We included 409 geriatric patients in the ED with a mean age of 79.5 years and approximately equal sex ratio. The mean NEWS ±SD was 3.4±2.9, and NEWS ≥8 was reported in 11.0% of the total patients. Logistic regression revealed that NEWS ≥8 predicted ICU admission with an OR of 5.37 (95% CI 2.61 to 11.04). The Hosmer-Lemeshow goodness-of-fit test was calculated as 0.95, and the adjusted area under the ROC was 0.72. An NEWS ≥8 is associated with ICU-admission and may help to triage elderly patients with influenza infections during the influenza epidemic season.

CONCLUSION

The high specificity of NEWS ≥8 to predict ICU admission in elderly patients with influenza infection during the epidemic season may avoid unnecessary ICU admissions and ensure proper medical resource allocation.

Consensus on predictors of clinical deterioration in emergency departments: A Delphi process study

AIM

The study aim was to determine relevance and applicability of generic predictors of clinical deterioration in emergency departments based on consensus among clinicians.

METHODS

Thirty-three predictors of clinical deterioration identified from literature were assessed in a modified two-stage Delphi-process. Sixty-eight clinicians (physicians and nurses) participated in the first round and 48 in the second round; all treating hospitalized patients in Danish emergency departments, some with pre-hospital experience. The panel rated the predictors for relevance (relevant marker of clinical deterioration) and applicability (change in clinical presentation over time, generic in nature and possible to detect bedside). They rated their level of agreement on a 9-point Likert scale and were also invited to propose additional generic predictors between the rounds. New predictors suggested by more than one clinician were included in the second round along with non-consensus predictors from the first round. Final decisions of non-consensus predictors after second round were made by a research group and an impartial physician.

RESULTS

The Delphi-process resulted in 19 clinically relevant and applicable predictors based on vital signs and parameters (respiratory rate, saturation, dyspnoea, systolic blood pressure, pulse rate, abnormal electrocardiogram, altered mental state and temperature), biochemical tests (serum c-reactive protein, serum bicarbonate, serum lactate, serum pH, serum potassium, glucose, leucocyte counts and serum haemoglobin), objective clinical observations (skin conditions) and subjective clinical observations (pain reported as new or escalating, and relatives' concerns).

CONCLUSION

The Delphi-process led to consensus of 19 potential predictors of clinical deterioration widely accepted as relevant and applicable in emergency departments.

Risk of Dementia in Diabetic Patients with Hyperglycemic Crisis: A Nationwide Taiwanese Population-Based Cohort Study

BACKGROUND

A hyperglycemic crisis episode (HCE) signifies poor control of diabetes and may increase the risk of dementia via microvascular and macrovascular injuries.

OBJECTIVES

We conducted this study to clarify this issue, which remains unclear.

METHODS

Using the Taiwan National Health Insurance Database, we identified 9,466 diabetic patients with HCE and the identical number of diabetic patients without HCE who were matched by age and sex for this nationwide population-based cohort study. The risk of dementia was compared between the 2 cohorts by following up until 2014. Investigation of independent predictors of dementia was also done.

RESULTS

In the overall analysis, the risk of dementia between the 2 cohorts was not different. However, stratified analyses showed that patients with HCE had a higher risk of subsequent dementia in the age subgroup of 45-54 and 55-64 years (adjusted odds ratio [AOR]: 2.4, 95% confidence interval [CI]: 1.6-3.6, and AOR: 1.2, 95% CI: 1.0-1.5, respectively). In the overall analysis, older age, female sex, ≥3 HCEs, hypertension, hyperlipidemia, depression, cerebrovascular disease, Parkinson's disease, and head injury were independent predictors.

CONCLUSIONS

HCE increased the risk of dementia in diabetic patients aged 45-64 years. Dementia was predicted by ≥3 HCEs. Prevention of recurrent HCE, control of comorbidities, and close follow-up of cognitive decline and dementia are suggested in patients with HCE.

Validation of Decision Groups in Patients with Dengue Fever: A Study during 2015 Outbreak in Taiwan

The management of dengue fever (DF) has been suggested to be categorized into decision groups A, B, and C; however, its usefulness in predicting mortality is still unclear, and hence we conducted this study to clarify this issue. We conducted a study by recruiting 2,358 patients with DF from the 2015 outbreak in the Chi-Mei Medical Center. Demographic data, vital signs, clinical symptoms and signs, coexisting morbidities, laboratory data, decision groups categorized according to World Health Organization for clinical management of dengue in 2012, and 30-day mortality rates were included for analysis. The overall 30-day mortality rate was 1.4%. The 30-day mortality rates in decision groups A, B, and C were 0%, 0.5%, and 46.2%, respectively. Compared with Group A, there was a higher mortality risk in Group C (odds ratio [OR]: 1,480, 95% confidence interval [CI]: 195-11,200). The area under the curve of the variable of Group C was excellent (OR: 0.92, 95% CI: 0.85-0.99). The sensitivity, specificity, positive predictive value, and negative predictive value for predicting 30-day mortality in Group C were 88.2%, 98.5%, 46.2%, and 99.8%, respectively. This study showed that decision Group C has a good predictive value for 30-day mortality. Further studies including validation in other nations are warranted.

Risk of diabetes mellitus in physicians: a nationwide study in Taiwan

Background

The heavy workload of physicians in Taiwan may contribute to poor lifestyles and increased risk for diabetes mellitus (DM). We conducted this study to determine the risk for DM among physicians in Taiwan.

Methods

We used the Taiwan National Health Insurance Research Database to identify 28,440 physicians and 56,880 comparisons (general population) matched at a ratio of 1:2 by age and sex. Participants who had been diagnosed with DM before 2007 were excluded. We compared the risk for DM between physicians and comparisons by following up since 2007 to 2013. Comparisons among physician subgroups were also performed.

Results

After adjustment for hypertension, hyperlipidemia, hyperuricemia, coronary artery disease, congestive heart failure, hyperthyroidism, hypothyroidism, and polycystic ovary syndrome, physicians had a lower risk for DM than the comparisons (adjusted odds ratio [AOR]: 0.75; 95% confidence interval [CI]: 0.68–0.82). In comparisons among physicians, emergency physicians (AOR: 2.21; 95% CI: 1.44–3.40) and surgeons (AOR: 1.26; 95% CI: 1.05–1.52) had a higher risk for DM than other specialists.

Conclusions

This study found that physicians have a lower risk for DM than the general population and emergency physicians and surgeons have a higher risk for DM than other specialists. Thus, more attention should be paid to the occupational health of these doctors.

Utilization of systemic inflammatory response syndrome criteria in predicting mortality among geriatric patients with influenza in the emergency department

BACKGROUND

Systemic Inflammatory Response Syndrome (SIRS) criteria are often used to evaluate the risk of sepsis and to identify in-hospital mortality among patients with suspected infection. However, utilization of the SIRS criteria in mortality prediction among geriatric patients with influenza in the emergency department (ED) remains unclear. Therefore, we conducted a research to delineate this issue.

METHODS

This is a retrospective case-control study including geriatric patients (age ≥ 65 years) with influenza, who presented to the ED of a medical center between January 1, 2010 and December 31, 2015. Vital signs, past history, subtype of influenza, demographic data, and outcomes were collected from all patients and analyzed. We calculated the accuracy for predicting 30-days mortality using the SIRS criteria. We also performed covariate adjustment of the area under the receiver operating characteristic curve (AUROC) via regression modeling.

RESULTS

We recruited a total of 409 geriatric patients in the ED, with mean age 79.5 years and an equal sex ratio. The mean SIRS criteria score was 1.9 ± 1.1. The result of a Hosmer-Lemeshow goodness-of-fit test was 0.34 for SIRS criteria. SIRS criteria score ≥ 3 showed better mortality prediction, with odds ratio (OR) 3.37 (95% confidence interval (CI), 1.05-10.73); SIRS score ≥ 2 showed no statistical significance, with p = 0.85 (OR, 1.15; 95% CI, 0.28-4.69). SIRS score ≥ 3 had acceptable 30-days mortality discrimination, with AUROC 0.77 (95% CI, 0.68-0.87) after adjustment. SIRS score ≥ 3 also had a notable negative predictive value of 0.97 (95% CI, 0.94-0.99).

CONCLUSION

The presence of a higher number of SIRS criteria (≥ 3) showed greater accuracy for predicting mortality among geriatric patients with influenza.

Quick-SOFA score to predict mortality among geriatric patients with influenza in the emergency department

The quick sequential organ failure assessment (qSOFA) score is widely used to assess the risk of sepsis and predict in-hospital mortality in patients with suspected infection. However, its ability to predict mortality among geriatric patients with influenza in the emergency department (ED) remains unclear. Therefore, this study was conducted to delineate this issue.A retrospective case-control study was conducted on geriatric patients (age ≥65 years) with influenza who visited the ED of a medical center between January 01, 2010, and December 31, 2015. Demographic data, vital signs, past histories, influenza subtypes, and treatment outcomes were included in the analysis. We assessed the accuracy of the qSOFA score in predicting 30-day mortality via logistic regression. Covariate adjustment of the area under the receiver operating characteristic curve (AUROC) via regression modeling was performed too.In total, 409 geriatric ED patients with mean age of 79.5 years and nearly equal sex ratio were recruited. The mean qSOFA score was 0.55 ± 0.7. The Hosmer-Lemeshow goodness-of-fit test was 0.79 for qSOFA score. Patients with qSOFA score of ≥2 (odds ratio, 4.21; 95% confidence interval [CI], 1.56-11.40) had increased in-hospital mortality. qSOFA score of ≥2 also had excellent in-hospital mortality discrimination with an adjusted AUROC of 0.81 (95% CI, 0.71-0.90). A qSOFA of ≥2 had prominent specificity of 0.89 (95% CI, 0.86-0.92).An increase in qSOFA score of 2 greatly predicts mortality in geriatric patients with influenza.

Characteristics of hyperglycemic crises in an adult population in a teaching hospital in Colombia

Background

Hyperglycemic crisis are the most serious forms of acute decompensation of diabetes mellitus and require urgent medical attention. The epidemiological data of these conditions in Latin America are scarce and in Colombia unknown, that is why we decided to describe the clinical characteristics and factors associated with the mortality of adults who presented with hyperglycemic crises in a teaching hospital in Colombia.

Materials and methods

Retrospective cohort study of all episodes of hyperglycemic crisis treated in Pablo Tobón Uribe Hospital in a three-year period.

Results

The records of 2233 hospitalization episodes related to diabetes mellitus were review, the prevalence of hyperglycemic crises was 2%, half of the events were diabetic ketoacidosis and 57% of the events occurred in people with type 2 diabetes mellitus, 32% of the events were precipitated by an infection and 27% by and inadequate therapy. The average hospital length of stay was 14 ± 3 days and the mortality rate 2.27%.

Conclusions

In a teaching hospital in Latin America hyperglycemic crises are common, with diabetic ketoacidosis being the most frequent, and in a significant number of cases may be preventable. The hospital length of stay in our population is longer than reported in the literature.

Geriatric influenza death (GID) score: a new tool for predicting mortality in older people with influenza in the emergency department

Although influenza may cause death in the geriatric population, the best method for predicting mortality in this population is still unclear. We retrospectively recruited older people (≥65 yr) with influenza visiting the emergency department (ED) of a medical center between January 1, 2010, and December 31, 2015. We performed univariate and multivariate logistic regression to identify independent mortality predictors and then developed a prediction score. Four hundred nine older ED patients with a nearly equal sex ratio were recruited. Five independent mortality predictors were identified: severe coma (Glasgow Coma Scale score ≤8), past histories of cancer and coronary artery disease, elevated C-reactive protein levels (>10 mg/dl), and bandemia (>10% band cells). We divided the patients into three mortality risk and disposition groups: (1) low risk (1.1%; 95% confidence interval [CI], 0.5-3.0%); (2) moderate risk (16.7%; 95% CI, 9.3-28.0%); and (3) high risk (40%; 95% CI, 19.8-64.2%). The area under the receiver operating characteristic curve and the Hosmer-Lemeshow goodness of fit of the GID score were 0.86 and 0.578, respectively. The GID score is an efficient and simple tool for predicting mortality in older ED patients with influenza. Further studies are warranted to validate its use.

Shock index predicted mortality in geriatric patients with influenza in the emergency department

BACKGROUND

The shock index is a rapid and simple tool used to predict mortality in patients with acute illnesses including sepsis, multiple trauma, and postpartum hemorrhage. However, its ability to predict mortality in geriatric patients with influenza in the emergency department (ED) remains unclear. This study was conducted to clarify this issue.

METHODS

We conducted a retrospective case-control study, recruiting geriatric patients (≥ 65 years) with influenza visiting the ED of a medical center between January 01, 2010 and December 31, 2015. Demographic data, vital signs, shock index, past histories, subtypes of influenza, and outcomes were included for the analysis. We investigated the association between shock index ≥1 and 30-day mortality.

RESULTS

In total, 409 geriatric ED patients with mean age of 79.5 years and nearly equal sex ratio were recruited. The mean shock index ± standard deviation was 0.7 ± 0.22 and shock index ≥1 was accounted for in 7.1% of the total patients. Logistic regression showed that shock index ≥1 predicted mortality (odds ratio: 6.80; 95% confidence interval: 2.39-19.39). The area under the receiver operating characteristic was 0.62 and the result of the Hosmer-Lemeshow goodness-of-fit test was 0.23. The sensitivity, specificity, positive predictive value, and negative predictive value of a shock index ≥1 were 30.0%, 94.1%, 20.0%, and 96.4%.

CONCLUSIONS

A shock index ≥1 has a high specificity, negative predictive value, and good reliability to predict 30-day mortality in geriatric ED patients with influenza.

Association of hyperglycemic crisis with an increased risk of end-stage renal disease: A nationwide population-based cohort study

BACKGROUND

A hyperglycemic crisis episode (HCE) is associated with poor management of diabetes, which is a risk factor for end-stage renal disease (ESRD); however, the association between an HCE and ESRD has not been clarified. We conducted a nationwide population-based cohort study with the purpose of delineating this issue.

METHODS

We identified 9208 diabetic patients with an HCE and an identical number of diabetic patients with matched age, sex, and index date without an HCE between 2000 and 2002. A comparison of the risk of ESRD between the diabetic patients with and without an HCE was achieved by a follow-up until 2014.

RESULTS

A Cox proportional hazard regression analysis showed that the diabetic patients with an HCE were at a higher risk of ESRD than those without an HCE (the adjusted hazard ratio [AHR]: 1.47; 95% confidence interval [CI]: 1.34-1.62) by adjusting for renal disease, hypertension, hyperlipidemia, coronary artery disease, hyperuricemia, anemia, chronic obstructive pulmonary disease, liver disease, malignancy, connective tissue disease, non-steroid anti-inflammatory drug use, and monthly income. The increased risk of ESRD was more prominent in the age subgroup of 15-25 years (AHR: 4.91; 95% CI: 1.92-12.56); 25-35 years (AHR: 2.42; 95% CI: 1.51-3.86); 35-45 years (AHR: 3.01; 95% CI: 2.21-4.09); and 45-55 years (AHR: 1.75; 95% CI: 1.41-2.19).

CONCLUSIONS

An HCE was associated with an increased risk of ESRD, especially in the younger diabetic patients (15-55 years). A close follow-up for the control of diabetes and for monitoring renal function is proposed.

Validation of Predicting Hyperglycemic Crisis Death Score: A Risk Stratification Tool for Appropriate Disposition of Hyperglycemic Crisis Patients from the Emergency Department

CONTEXT

Predicting hyperglycemic crisis death (PHD) score is a simple, rapid tool with six independent mortality predictors to calculate 30-day mortality and appropriately dispose patients to Intensive Care Unit (ICU) or ward.

AIMS

This study aimed at validating the efficiency of PHD score as a decision rule for prognosticating 30-day mortality and classifying hyperglycemic crisis patients for appropriate disposition from the emergency department (ED).

MATERIALS AND METHODS

This is a prospective, observational study done in the ED of a teaching hospital over 14 months. All patients aged >18 years and who met the criteria of hyperglycemic crisis were enrolled. Thirty-day mortality of six independent predictors was the primary end point. Using PHD, risk scores were calculated and patients were disposed as per physician's clinical judgment. Finally, the treating physician's decision and PHD score disposition were compared and the efficiency of PHD in predicting 30-day mortality was analyzed. Multiple logistic regression models were used for analysis. Receiver operating characteristic curve was drawn, and area under the curve along with sensitivity, specificity, positive predictive value, and negative predictive value was analyzed. P < 0.05 was considered statistically significant.

RESULTS

A total of 133 patients were included. On applying PHD score, 69, 39, and 25 patients were in the low-, intermediate-, and high-risk groups, respectively, with a mortality rate of 5.8%, 20.5%, and 56%, respectively. On comparing physician disposition with PHD score, an increasing mortality was noticed in ICU, and PHD showed equal weight in risk stratification and appropriate disposition of patients.

CONCLUSION

In adult patients with hyperglycemic crisis, PHD score is validated as a straightforward, prompt tool for predicting 30-day mortality and aids in disposition. The mortality rate in the PHD score Model II was similar to the physician's clinical decision.

Shift to community-onset Clostridium difficile infection in the national Veterans Health Administration, 2003-2014

BACKGROUND

Clostridium difficile infection (CDI) occurs frequently in inpatient settings; however, community-onset cases have been reported more frequently in recent years. This study evaluated hospital-onset and community-onset CDI in the national Veterans Health Administration (VHA) population over a 12-year period.

METHODS

This was a retrospective cohort study of all adult VHA beneficiaries with CDI between October 1, 2002, and September 30, 2014. Data were obtained from the Veterans Affairs Informatics and Computing Infrastructure. CDI was categorized into community-associated CDI (CA-CDI); community-onset, health care facility-associated CDI; and health care facility-onset CDI (HCFO-CDI). Each type was described longitudinally and was assessed as an independent risk factor for health outcomes using multivariable logistic regression.

RESULTS

Overall, 30,326 patients with a first CDI episode were included. HCFO-CDI was the predominant type (60.2%), followed by CO-HCFA-CDI (20.6%) and CA-CDI (19.2%). The proportion of patients with HCFO-CDI decreased from 73.5% during fiscal year 2003 to 53.2% during fiscal year 2014, whereas CA-CDI increased from 8.3% to 26.7%. HCFO-CDI was a positive predictor of severe CDI (odds ratio [OR], 1.71; 95% confidence interval [CI], 1.59-1.84) and 30-day mortality (OR, 1.46; 95% CI, 1.32-1.61), but a negative predictor of 60-day recurrence (OR, 0.41; 95% CI, 0.37-0.46).

CONCLUSIONS

HCFO-CDI was the predominant CDI type. The proportion of patients with CA-CDI increased and HCFO-CDI decreased in recent years. Patients with HCFO-CDI experienced higher rates of severe CDI and mortality.

National epidemiology of initial and recurrent Clostridium difficile infection in the Veterans Health Administration from 2003 to 2014

Introduction

Prior studies demonstrated marked increases in Clostridium difficile infection (CDI) in the United States (U.S.) in recent years. The objective of this study was to describe the epidemiology of initial and recurrent CDI in a national Veterans Health Administration (VHA) cohort over a 12-year period.

Methods

This was a retrospective cohort study of all adult VHA beneficiaries with CDI (ICD-9-CM code 008.45) plus a positive CDI stool test between October 1, 2002 and September 30, 2014. Data were obtained from the VA Informatics and Computing Infrastructure. Recurrence was defined as a second ICD-9-CM code plus a new course of CDI therapy following a minimum three-day gap after the initial therapy was completed. CDI incidence and outcomes were presented descriptively and longitudinally.

Results

Overall, 30,326 patients met study inclusion criteria. CDI incidence increased from FY 2003 (1.6 per 10,000) to FY 2013 (5.1 per 10,000). Thereafter, CDI incidence decreased through FY 2014 (4.6 per 10,000). A total of 5,011 patients (17%) experienced a first recurrence and, of those, 1,713 (34%) experienced a second recurrence. Recurrence incidence increased 10-fold over the study period, from (0.1 per 10,000) in FY 2003, to (1.0 per 10,000) in FY 2014. Overall, 30-day mortality and median hospital length of stay (LOS) decreased among initial episodes over the study period. Mortality was higher for initial episodes (21%) compared to first recurrences (11%) and second recurrences (7%). Median hospital LOS was longer for first episodes (13 days) compared to first (9 days) and second recurrences (8 days).

Conclusions

Initial and recurrent CDI episodes increased among veterans over a 12-year period. Outcomes, such as mortality and hospital LOS improved in recent years; both of these outcomes are worse for initial CDI episodes than recurrent episodes.

Predicting the mortality in geriatric patients with dengue fever

Geriatric patients have high mortality for dengue fever (DF); however, there is no adequate method to predict mortality in geriatric patients. Therefore, we conducted this study to develop a tool in an attempt to address this issue.We conducted a retrospective case-control study in a tertiary medical center during the DF outbreak in Taiwan in 2015. All the geriatric patients (aged ≥65 years) who visited the study hospital between September 1, 2015, and December 31, 2015, were recruited into this study. Variables included demographic data, vital signs, symptoms and signs, comorbidities, living status, laboratory data, and 30-day mortality. We investigated independent mortality predictors by univariate analysis and multivariate logistic regression analysis and then combined these predictors to predict the mortality.A total of 627 geriatric DF patients were recruited, with a mortality rate of 4.3% (27 deaths and 600 survivals). The following 4 independent mortality predictors were identified: severe coma [Glasgow Coma Scale: ≤8; adjusted odds ratio (AOR): 11.36; 95% confidence interval (CI): 1.89-68.19], bedridden (AOR: 10.46; 95% CI: 1.58-69.16), severe hepatitis (aspartate aminotransferase >1000 U/L; AOR: 96.08; 95% CI: 14.11-654.40), and renal failure (serum creatinine >2 mg/dL; AOR: 6.03; 95% CI: 1.50-24.24). When we combined the predictors, we found that the sensitivity, specificity, positive predictive value, and negative predictive value for patients with 1 or more predictors were 70.37%, 88.17%, 21.11%, and 98.51%, respectively. For patients with 2 or more predictors, the respective values were 33.33%, 99.44%, 57.14%, and 98.51%.We developed a new method to help decision making. Among geriatric patients with none of the predictors, the survival rate was 98.51%, and among those with 2 or more predictors, the mortality rate was 57.14%. This method is simple and useful, especially in an outbreak.

Effects of Arsenic in Drinking Water on Risk of Hepatitis or Cirrhosis in Persons With and Without Chronic Viral Hepatitis

BACKGROUND & AIMS

Arsenic in drinking water is associated with hepatomegaly and death from liver cancer. However, confounding factors related to liver diseases have not been carefully studied. We examined associations between exposure of arsenic in drinking water and risk of hepatitis and cirrhosis, and the interaction with chronic viral hepatitis, in people living in the Lanyang Basin of northeastern Taiwan, where well water has an arsenic content that ranges from undetectable to 3590 μg/L.

METHODS

We tested blood samples from 4387 people who lived in arseniasis-endemic areas in northeastern Taiwan from 1991 through 1994 for hepatitis B virus DNA, hepatitis B surface antigen (HBsAg), and antibodies against hepatitis C virus (anti-HCV). We measured arsenic concentrations in well water and collected information on residents' histories of major chronic diseases. Reports of chronic hepatitis or cirrhosis were ascertained using the Taiwan National Health Insurance database. Reports of liver cancer were ascertained using the Taiwan National Cancer Registry.

RESULTS

Prevalence odds ratios in the overall study population for chronic hepatitis or cirrhosis for well water arsenic concentrations of ≤10 μg/L were 1.00 (reference), 0.93 for 10.1-49.9 μg/L (95% confidence interval [CI], 0.57-1.52), 1.24 for 50.0-99.9 μg/L (95% CI, 0.68-2.23), 0.98 for 100.0-299.9 (95% CI, 0.52-1.85), and 1.86 for ≥300.0 μg/L (95% CI, 1.08-3.20). Increasing levels of arsenic in drinking water were associated with increasing prevalence of chronic hepatitis or cirrhosis in residents who were seronegative for HBsAg and seronegative for anti-HCV, but not for seropositive for either HBsAg or anti-HCV. In individuals who were seropositive for HBsAg or anti-HCV, we observed an inverse association between hepatitis or cirrhosis and consumption of water with levels of arsenic ≥100.0 μg/L. Among participants who were seropositive for HBsAg or anti-HCV, consumption of water with levels of arsenic ≥100.0 μg/L was associated with a reduced risk of liver cancer (multivariate-adjusted hazard ratio, 0.29; 95% CI, 0.09-0.95; P < .05). A higher proportion of individuals exposed to cumulative arsenic level >14,000 μg/L ×year were carriers of inactive hepatitis B virus (DNA <10,000 copies/mL) and were positive for HBsAg (60%) than individuals exposed to water below this arsenic level (35%).

CONCLUSIONS

Concentrations of arsenic concentration in drinking water ≥300.0 μg/L significantly increase risk of hepatitis or cirrhosis in people without chronic viral hepatitis. However, in people with chronic viral hepatitis, levels of arsenic ≥100.0 μg/L in drinking water significantly reduce the risk of chronic hepatitis or cirrhosis.

Subsequent mortality after hyperglycemic crisis episode in the non-elderly: a national population-based cohort study

Hyperglycemic crisis episodes (HCEs)-diabetic ketoacidosis and the hyperosmolar hyperglycemic state-are the most serious acute metabolic complications of diabetes. We aimed to investigate the subsequent mortality after HCE in the non-elderly diabetic which is still unclear. This retrospective national population-based cohort study reviewed, in Taiwan's National Health Insurance Research Database, data from 23,079 non-elder patients (≤65 years) with new-onset diabetes between 2000 and 2002: 7693 patients with HCE and 15,386 patients without HCE (1:2). Both groups were compared, and follow-up prognoses were done until 2011. One thousand eighty-five (14.1%) patients with HCE and 725 (4.71%) patients without HCE died (P < 0.0001) during follow-up. Incidence rate ratios (IRR) of mortality were 3.24 times higher in patients with HCE than in patients without HCE (P < 0.0001). Individual analysis of diabetic ketoacidosis and hyperosmolar hyperglycemic state also showed the similar result with combination of both. After stratification by age, mortality was significant higher in the middle age (40-64 years) [IRR 3.29; 95% confidence interval (CI) 2.98-3.64] and young adult (18-39 years) (IRR 3.91; 95% CI 3.28-4.66), but not in the pediatric subgroup (<18 years) (IRR 1.28; 95% CI 0.21-7.64). The mortality risk was highest in the first month (IRR 54.43; 95% CI 27.98-105.89), and still high after 8 years (IRR 2.05; 95% CI 1.55-2.71). After adjusting for age, gender, and selected comorbidities, the mortality hazard ratio for patients with HCE was still four times higher than for patients without HCE. Moreover, older age, male gender, stroke, cancer, chronic obstructive pulmonary disease, congestive heart failure, and liver disease were independent mortality predictors. HCE significantly increases the subsequent mortality risk in the non-elderly with diabetes. Strategies for prevention and control of comorbidities are needed as soon as possible.

Long-term Mortality Risk After Hyperglycemic Crisis Episodes in Geriatric Patients With Diabetes: A National Population-Based Cohort Study

OBJECTIVE

Hyperglycemic crisis is one of the most serious diabetes-related complications. The increase in the prevalence of diabetes in the geriatric population leads to a large disease burden, but previous studies of geriatric hyperglycemic crisis were focused on acute hyperglycemic crisis episode (HCE). This study aimed to delineate the long-term mortality risk after HCE.

RESEARCH DESIGN AND METHODS

This retrospective national population-based cohort study reviewed, in Taiwan's National Health Insurance Research Database, data from 13,551 geriatric patients with new-onset diabetes between 2000 and 2002, including 4,517 with HCE (case subjects) (ICD-9 code 250.1 or 250.2) and 9,034 without HCE (control subjects). The groups were compared and followed until 2011.

RESULTS

One thousand six hundred thirty-four (36.17%) case and 1,692 (18.73%) control subjects died (P < 0.0001) during follow-up. Incidence rate ratios (IRRs) of death were 2.82 times higher in case subjects (P < 0.0001). The mortality risk was highest in the first month (IRR 26.56; 95% CI 17.97-39.27) and remained higher until 4-6 years after the HCE (IRR 1.49; 95% CI 1.23-1.81). After adjustment for age, sex, selected comorbidities, and monthly income, the mortality hazard ratio was still 2.848 and 4.525 times higher in case subjects with one episode and two or more episodes of hyperglycemic crisis, respectively. Older age, male sex, renal disease, stroke, cancer, chronic obstructive pulmonary disease, and congestive heart failure were independent mortality predictors.

CONCLUSIONS

Patients with diabetes had a higher mortality risk after HCE during the first 6 years of follow-up. Referral for proper education, better access to medical care, effective communication with a health care provider, and control of comorbidities should be done immediately after HCE.

Clinical characteristics of hyperglycemic crises in patients without a history of diabetes

Aims/Introduction

Hyperglycemic crises without a history of diabetes have not been well studied. We compared the clinical characteristics of patients with and without a history of diabetes, and evaluated the glycated hemoglobin levels.

Materials and Methods

Consecutive adult patients (aged >18 years) visiting the emergency department (ED) between January 2004 and December 2010 were enrolled if they met the criteria for a hyperglycemic crisis. Patients were separated into those without and those with a history of diabetes. The 30-day mortality was the primary end-point.

Results

We enrolled 295 patients who made 330 visits to the ED. Patients without a history of diabetes made up 24.5% (81/330) of the hyperglycemic crises. Patients without a history of diabetes were more prone than patients with a history of diabetes to be younger and male, and to have better consciousness and renal function, more significant diabetic signs and symptoms (e.g., thirst, polydipsia, polyuria and bodyweight loss), higher blood sugar, and less opportunity of infection and mortality. Most of the patients (93.8%, 76/81) had glycated hemoglobin of ≥6.5%.

Conclusions

The present study delineates the clinical characteristics of patients with hyperglycemic crises, but without a history of diabetes. Most patients had glycated hemoglobin ≥6.5%, which raises the argument of using this biomarker for routine screening of diabetes.

Cancer history, bandemia, and serum creatinine are independent mortality predictors in patients with infection-precipitated hyperglycemic crises

BACKGROUND

Infection is the most common precipitating factor and cause of death in patients with hyperglycemic crises. Treating infection-precipitated hyperglycemic crises includes using empiric antibiotics early; correcting dehydration, hyperglycemia, and electrolyte imbalances; and frequent monitoring. Intensive care unit admission, broad-spectrum antibiotics, and even novel therapy for infection may be beneficial for patients with a high risk of mortality. However, these management options are costly and not beneficial for every patient. Selecting high-risk patients who would most likely benefit is more appropriate. We investigated the independent mortality predictors of patients with infection-precipitated hyperglycemic crises to facilitate clinical decision making.

METHODS

This study was conducted in a university-affiliated medical center. Consecutive adult patients (> 18 years old) visiting the Emergency Department between January 2004 and December 2010 were enrolled when they met the criteria of an infection-precipitated hyperglycemic crisis. Thirty-day mortality was the primary endpoint.

RESULTS

One hundred forty-two patients were enrolled. The infection source did not predict mortality. The presenting variables that were independently associated with 30-day mortality in a multiple logistic regression model were cancer history (odds ratio [OR], 7.4; 95% confidence interval [CI], 2.4-23.2), bandemia (OR, 7.0; 95% CI, 1.6-30.3), and serum creatinine (OR, 1.4; 95% CI, 1.1-1.8). The common sources of infection were the lower respiratory tract (30.3%), urinary tract (49.3%), skin or soft tissue (12.0%), and intra-abdominal (6.3%).

CONCLUSIONS

Cancer history, bandemia, and serum creatinine level are three independent mortality predictors for patients with infection-precipitated hyperglycemic crises. These predictors are both readily available and valuable for physicians making decisions about risk stratification, treatment, and disposition.