Dataset Information

Fixed effects modelling for provider mortality outcomes: Analysis of the Australia and New Zealand Intensive Care Society (ANZICS) Adult Patient Data-base.

ABSTRACT: BACKGROUND:Risk adjusted mortality for intensive care units (ICU) is usually estimated via logistic regression. Random effects (RE) or hierarchical models have been advocated to estimate provider risk-adjusted mortality on the basis that standard estimators increase false outlier classification. The utility of fixed effects (FE) estimators (separate ICU-specific intercepts) has not been fully explored. METHODS:Using a cohort from the Australian and New Zealand Intensive Care Society Adult Patient Database, 2009-2010, the model fit of different logistic estimators (FE, random-intercept and random-coefficient) was characterised: Bayesian Information Criterion (BIC; lower values better), receiver-operator characteristic curve area (AUC) and Hosmer-Lemeshow (H-L) statistic. ICU standardised hospital mortality ratios (SMR) and 95%CI were compared between models. ICU site performance (FE), relative to the grand observation-weighted mean (GO-WM) on odds ratio (OR), risk ratio (RR) and probability scales were assessed using model-based average marginal effects (AME). RESULTS:The data set consisted of 145355 patients in 128 ICUs, years 2009 (47.5%) & 2010 (52.5%), with mean(SD) age 60.9(18.8) years, 56% male and ICU and hospital mortalities of 7.0% and 10.9% respectively. The FE model had a BIC = 64058, AUC = 0.90 and an H-L statistic P-value = 0.22. The best-fitting random-intercept model had a BIC = 64457, AUC = 0.90 and H-L statistic P-value = 0.32 and random-coefficient model, BIC = 64556, AUC = 0.90 and H-L statistic P-value = 0.28. Across ICUs and over years no outliers (SMR 95% CI excluding null-value = 1) were identified and no model difference in SMR spread or 95%CI span was demonstrated. Using AME (OR and RR scale), ICU site-specific estimates diverged from the GO-WM, and the effect spread decreased over calendar years. On the probability scale, a majority of ICUs demonstrated calendar year decrease, but in the for-profit sector, this trend was reversed. CONCLUSIONS:The FE estimator had model advantage compared with conventional RE models. Using AME, between and over-year ICU site-effects were easily characterised.

SUBMITTER: Moran JL

PROVIDER: S-EPMC4100889 | biostudies-literature | 2014

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Fixed effects modelling for provider mortality outcomes: Analysis of the Australia and New Zealand Intensive Care Society (ANZICS) Adult Patient Data-base.

Moran John L JL Solomon Patricia J PJ

PloS one 20140716 7

<h4>Background</h4>Risk adjusted mortality for intensive care units (ICU) is usually estimated via logistic regression. Random effects (RE) or hierarchical models have been advocated to estimate provider risk-adjusted mortality on the basis that standard estimators increase false outlier classification. The utility of fixed effects (FE) estimators (separate ICU-specific intercepts) has not been fully explored.<h4>Methods</h4>Using a cohort from the Australian and New Zealand Intensive Care Socie ...[more]

PMID: 25029164

Similar Datasets

Project description:Wide variations in blood glucose excursions in critically ill patients may influence adverse outcomes such as hospital mortality. However, whether blood glucose variability is independently associated with mortality or merely captures the excess risk attributable to hyperglycemic and hypoglycemic episodes is not established. We investigated whether blood glucose variability independently predicted hospital mortality in nonhyperglycemic critical care patients. Design:Retrospective, registry data analyses of outcomes. Setting:Large, binational registry (Australia and New Zealand Intensive Care Society Centre for Outcome and Resource Evaluation Adult Patient Database repository) of 176 ICUs across Australia and New Zealand. Patients:We used 10-year data on nonhyperglycemic patients registered in the Australia and New Zealand Intensive Care Society Centre for Outcome and Resource Evaluation Adult Patient Database repository (n = 290,966). Interventions:None. Measurements and Main Results:Glucose variability was captured using glucose width defined as the difference between highest and lowest blood glucose concentration within first 24 hours of ICU admission. We used hierarchical, mixed effects logistic regression models that accounted for ICU variation and several fixed-effects covariates. Glucose width was specifically and independently associated with hospital mortality. The association of blood glucose variability with mortality remained significant (odds ratio for highest vs lowest quartile of glucose, 1.43; 95% CI, 1.32-1.55; p < 0.001) even after adjusting for the baseline risk of mortality, midpoint blood glucose level, occurrence of hypoglycemia and inter-ICU variation. Mixed effects modeling showed that there was a statistically significant variation in this association across ICUs. Conclusions:Our study demonstrates that glucose variability is independently associated with hospital mortality in critically ill adult patients. Inclusion of correction for glucose variability in glycemic control protocols needs to be investigated in future studies.

Project description:This position statement, updated from the 2015 guidelines for managing Australian and New Zealand children/adolescents and adults with chronic suppurative lung disease (CSLD) and bronchiectasis, resulted from systematic literature searches by a multi-disciplinary team that included consumers. The main statements are: Diagnose CSLD and bronchiectasis early; this requires awareness of bronchiectasis symptoms and its co-existence with other respiratory diseases (e.g., asthma, chronic obstructive pulmonary disease). Confirm bronchiectasis with a chest computed-tomography scan, using age-appropriate protocols and criteria in children. Undertake a baseline panel of investigations. Assess baseline severity, and health impact, and develop individualized management plans that include a multi-disciplinary approach and coordinated care between healthcare providers. Employ intensive treatment to improve symptom control, reduce exacerbation frequency, preserve lung function, optimize quality-of-life and enhance survival. In children, treatment also aims to optimize lung growth and, when possible, reverse bronchiectasis. Individualize airway clearance techniques (ACTs) taught by respiratory physiotherapists, encourage regular exercise, optimize nutrition, avoid air pollutants and administer vaccines following national schedules. Treat exacerbations with 14-day antibiotic courses based upon lower airway culture results, local antibiotic susceptibility patterns, clinical severity and patient tolerance. Patients with severe exacerbations and/or not responding to outpatient therapy are hospitalized for further treatments, including intravenous antibiotics and intensive ACTs. Eradicate Pseudomonas aeruginosa when newly detected in lower airway cultures. Individualize therapy for long-term antibiotics, inhaled corticosteroids, bronchodilators and mucoactive agents. Ensure ongoing care with 6-monthly monitoring for complications and co-morbidities. Undertake optimal care of under-served peoples, and despite its challenges, delivering best-practice treatment remains the overriding aim.

Project description:ObjectiveIntensive care (ICU) beds are scarce and decision-making regarding admission is complex and multi-factorial. This study aimed to characterise differences in admission decision making between Australia and New Zealand and compare to previous data to establish changes over time.DesignOnline Survey.Setting and participantsAn online survey was distributed to Australian and New Zealand intensive care doctors measuring triage behaviours in the last week and responses to ICU triage scenarios.Main outcome measuresPerceived ICU admission behaviours.Results103 responses were obtained, 83(80.6%) from Australia and 97 (94.2%) from specialist intensivists. The median number of triage decisions and patients declined were 6-10 and 1-5 respectively. No difference was noted in the role of ICU bed capacity in decision making between Australia and New Zealand. Compared to Australian intensivists, New Zealand intensivists were less likely to admit a patient: with relapsed acute myeloid leukaemia (AML) and acute respiratory distress syndrome (ARDS)(p=0.03), with persistent vegetative state and community acquired (p=0.02) or iatrogenic (p=0.03) pneumonia. Compared to respondents in 2009 (n=238), 2023 respondents were more likely to admit a patient: with a severe intracranial bleed who may become braindead (p=0.005), with relapsed AML and ARDS (p=0.02), with stroke for palliative care (p<0.001); and less likely to admit a patient with persistent vegetative state and iatrogenic pneumonia (p=0.03). In a multivariable analysis, respondents from Australian compared to New Zealand and from 2023 compared to 2009 were more likely to indicate they would admit patients to the ICU in the scenarios described (p<0.001 for both comparisons).ConclusionsOur study suggests that New Zealand intensivists may apply more restrictive ICU admission criteria than Australian intensivists. Changes in attitudes to admission since 2009 may reflect increased awareness of the importance of facilitating organ donation and the role of ICU as providers of palliative care.

Project description:INTRODUCTION: Older age is associated with higher prevalence of chronic illness and functional impairment, contributing to an increased rate of hospitalization and admission to intensive care. The primary objective was to evaluate the rate, characteristics and outcomes of very old (age >or= 80 years) patients admitted to intensive care units (ICUs). METHODS: Retrospective analysis of prospectively collected data from the Australian New Zealand Intensive Care Society Adult Patient Database. Data were obtained for 120,123 adult admissions for >or= 24 hours across 57 ICUs from 1 January 2000 to 31 December 2005. RESULTS: A total of 15,640 very old patients (13.0%) were admitted during the study. These patients were more likely to be from a chronic care facility, had greater co-morbid illness, greater illness severity, and were less likely to receive mechanical ventilation. Crude ICU and hospital mortalities were higher (ICU: 12% vs. 8.2%, P < 0.001; hospital: 24.0% vs. 13%, P < 0.001). By multivariable analysis, age >/= 80 years was associated with higher ICU and hospital death compared with younger age strata (ICU: odds ratio (OR) = 2.7, 95% confidence interval (CI) = 2.4 to 3.0; hospital: OR = 5.4, 95% CI = 4.9 to 5.9). Factors associated with lower survival included admission from a chronic care facility, co-morbid illness, nonsurgical admission, greater illness severity, mechanical ventilation, and longer stay in the ICU. Those aged >or= 80 years were more likely to be discharged to rehabilitation/long-term care (12.3% vs. 4.9%, OR = 2.7, 95% CI = 2.6 to 2.9). The admission rates of very old patients increased by 5.6% per year. This potentially translates to a 72.4% increase in demand for ICU bed-days by 2015. CONCLUSIONS: The proportion of patients aged >or= 80 years admitted to intensive care in Australia and New Zealand is rapidly increasing. Although these patients have more co-morbid illness, are less likely to be discharged home, and have a greater mortality than younger patients, approximately 80% survive to hospital discharge. These data also imply a potential major increase in demand for ICU bed-days for very old patients within a decade.

Project description:Background and objectivesThe burden of infectious disease is high among kidney transplant recipients because of concomitant immunosuppression. In this study the incidence of infectious-related mortality and associated factors were evaluated.Design, setting, participants, & measurementsIn this registry-based retrospective, longitudinal cohort study, recipients of a first kidney transplant in Australia and New Zealand between 1997 and 2015 were included. Cumulative incidence of infectious-related mortality was estimated using competing risk regression (using noninfectious mortality as a competing risk event), and compared with age-matched, populated-based data using standardized incidence ratios.ResultsAmong 12,519 patients, (median age 46 years, 63% men, 15% diabetic, 6% Indigenous ethnicity), 2197 (18%) died, of whom 416 (19%) died from infection. The incidence of infection-related mortality during the study period (1997-2015) was 45.8 (95% confidence interval [95% CI], 41.6 to 50.4) per 10,000 patient-years. The incidence of infection-related mortality reduced from 53.1 (95% CI, 45.0 to 62.5) per 10,000 person-years in 1997-2000 to 43.9 (95% CI, 32.5 to 59.1) per 10,000 person-years in 2011-2015 (P<0.001) Compared with the age-matched general population, kidney transplant recipients had a markedly higher risk of infectious-related death (standardized incidence ratio, 7.8; 95% CI, 7.1 to 8.6). Infectious mortality was associated with older age (≥60 years adjusted subdistribution hazard ratio [SHR], 4.16; 95% CI, 2.15 to 8.05; reference 20-30 years), female sex (SHR, 1.62; 95% CI, 1.19 to 2.29), Indigenous ethnicity (SHR, 2.87; 95% CI, 1.84 to 4.46; reference white), earlier transplant era (2011-2015: SHR, 0.39; 95% CI, 0.20 to 0.76; reference 1997-2000), and use of T cell-depleting therapy (SHR, 2.43; 95% CI, 1.36 to 4.33). Live donor transplantation was associated with lower risk of infection-related mortality (SHR, 0.53; 95% CI, 0.37 to 0.76).ConclusionsInfection-related mortality in kidney transplant recipients is significantly higher than the general population, but has reduced over time. Risk factors include older age, female sex, Indigenous ethnicity, T cell-depleting therapy, and deceased donor transplantation.PodcastThis article contains a podcast at https://www.asn-online.org/media/podcast/CJASN/2019_08_27_CJN03200319.mp3.

Dataset Information

Fixed effects modelling for provider mortality outcomes: Analysis of the Australia and New Zealand Intensive Care Society (ANZICS) Adult Patient Data-base.

Publications

Fixed effects modelling for provider mortality outcomes: Analysis of the Australia and New Zealand Intensive Care Society (ANZICS) Adult Patient Data-base.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets