P. J. Solomon

Tympanic temperature measurements: Are they reliable in the critically ill? A clinical study of measures of agreement*

Objective: Accurate measurement of temperature is vital in the intensive care setting. A prospective trial was performed to compare the accuracy of tympanic, urinary, and axillary temperatures with that of pulmonary artery (PA) core temperature measurements. Design: A total of 110 patients were enrolled in a prospective observational cohort study. Setting: Multidisciplinary intensive care unit of a university teaching hospital. Patients: The cohort was (mean ± sd) 65 ± 16 yrs of age, Acute Physiology and Chronic Health Evaluation (APACHE) II score was 25 ± 9, 58% of the patients were men, and 76% were mechanically ventilated. The accuracy of tympanic (averaged over both ears), axillary (averaged over both sides), and urinary temperatures was referenced (as mean difference, &Dgr; degrees centigrade) to PA temperatures as standard in 6,703 recordings. Lin concordance correlation (pc) and Bland–Altman 95% limits of agreement (degrees centigrade) described the relationship between paired measurements. Regression analysis (linear mixed model) assessed covariate confounding with respect to temperature modes and reliability formulated as an intraclass correlation coefficient. Measurements and Main Results: Concordance of PA temperatures with tympanic, urinary, and axillary was 0.77, 0.92, and 0.83, respectively. Compared with PA temperatures, &Dgr; (limits of agreement) were 0.36°C (−0.56°C, 1.28°C), −0.05°C (−0.69°C, 0.59°C), and 0.30°C (−0.42°C, 1.01°C) for tympanic, urinary, and axillary temperatures, respectively. Temperature measurement mode effect, estimated via regression analysis, was consistent with concordance and &Dgr; (PA vs. urinary, p = .98). Patient age (p = .03), sedation score (p = .0001), and dialysis (p = .0001) had modest negative relations with temperature; quadratic relationships were identified with adrenaline and dobutamine. No interactions with particular temperature modes were identified (p ≥ .12 for all comparisons) and no relationship was identified with either mean arterial pressure or APACHE II score (p ≥ .64). The average temperature mode intraclass correlation coefficient for test–retest reliability was 0.72. Conclusion: Agreement of tympanic with pulmonary temperature was inferior to that of urinary temperature, which, on overall assessment, seemed more likely to reflect PA core temperature.

Mortality and length-of-stay outcomes, 1993-2003, in the binational Australian and New Zealand intensive care adult patient database*

Objective:Intensive care unit (ICU) outcomes have been the subject of controversy. The objective was to model hospital mortality and ICU length-of-stay time-change of patients recorded in the Australian and New Zealand Intensive Care Society adult patient database. Design:Retrospective, cohort study of prospectively collected data on index patient admissions. Setting:Australian and New Zealand ICUs, 1993–2003. Patients:The Australian and New Zealand Intensive Care Society adult patient database, which contains data for 223,129 patients. Interventions:None. Measurements and Main Results:Hospital mortality and ICU length of stay were modeled using logistic and linear regression, respectively, with determination (80%) and validation (20%) data sets. Model adequacy was assessed by discrimination (receiver operating characteristic curve area, AZ) and calibration (Hosmer-Lemeshow Ĉ) for mortality and R2 for length of stay. Predictor variables included patient demographics, severity score, surgical and ventilation status, ICU categories, and geographical locality. The data set comprised 223,129 patients: Their mean (sd) age was 59.2 (18.9) yrs, 41.7% were female, their mean (sd) Acute Physiology and Chronic Health Evaluation (APACHE) III score was 53 (31), they had 16.1% overall mortality rate, and 45.7% were mechanically ventilated. ICU length of stay was 3.6 (5.6) days. AZ, Ĉ statistic, and R2 for developmental and validation model data sets were 0.88, 17.64 (p = .02), and 0.18; and 0.88, 12.32 (p = .26), and 0.18, respectively. Variables with mortality impact (p ≤ .001) were age (odds ratio [OR] 1.023), gender (OR 1.16; males vs. females), APACHE III score (OR 1.06), mechanical ventilation (OR 1.66), and surgical status (elective, OR 0.17; emergency, OR 0.47; compared with nonsurgical). ICU level and locality had significant mortality-time effects. Similar variables were found to predict length of stay. Risk-adjusted mortality declined, during 1993–2003, from 0.19 (95% confidence interval 0.17–0.21) to 0.15 (0.13–0.16) and similarly for ventilated patients: 0.26 (0.24–0.29) to 0.23 (0.21–0.25). Predicted mean ICU length of stay (days) demonstrated minimal overall time-change: 3.4 (2.2) in 1993 to 3.5 (2.7) in 2003, peaking at 3.7 (2.4) in 2000. Conclusions:Overall hospital mortality rate in patients admitted to Australian and New Zealand ICUs decreased 4% over 11 yrs. A similar trend occurred for mechanically ventilated patients. Length of stay changed minimally over this period.

Sun exposure in Australian adolescents

BACKGROUND Little information is available on the sun-related behavior of teenagers despite the considerable resources spent to decrease sun exposure in this age group. OBJECTIVE Our purpose was to describe the sun exposure behavior of Australian adolescents and define characteristics that predict use of sun protection. METHODS Cross-sectional study of a random sample of 972 school students 13 to 15 years of age from three different locations in Australia (two urban and one rural) using a diary to document sun exposure and sun protection on two consecutive weekends. RESULTS More than 80% of adolescent boys in each place and more than 60% of adolescent girls in both of the large cities spent more than 2 hours outdoors during the peak ultraviolet (UV) periods on each weekend. Neither sunscreen nor hats were used for more than half the time spent in the sun; however, shirts were worn most of the time. Students who wore hats were more likely to be boys (odds ratios [OR] = 2.2, 95% confidence intervals [CI] 1.40 to 3.44) and live in the rural region (OR = 4.6, 95% CI 2.36 to 9.04). Students who used sunscreen tended to have skin that sunburned easily (OR = 3.2, 95% CI 1.27 to 7.88) and score highly on the knowledge questions (OR = 2.9, 95% CI 1.46-5.69). This model was not a good predictor of behavior on a subsequent weekend, possibly because behavior was highly variable, with 35% to 50% of students changing their pattern of protection use from one weekend to the next. CONCLUSION Adolescents spend long periods on summer weekends in the sun and do not follow recommended sun protection guidelines. The high variability of sun-related behavior makes modeling and consequent development of education programs a difficult task.

Accommodating change due to treatment in the method of back projection for estimating HIV infection incidence.

This note shows how the method of back projection, which is being widely applied to predict the incidence of HIV infection, can be extended to incorporate distributional changes due to a treatment effect, such as zidovudine (commonly known as AZT). By way of example we consider one of the approaches to back projection and apply the method to some Australian data.

Genetic regulators of myelopoiesis and leukemic signaling identified by gene profiling and linear modeling

Mechanisms controlling the balance between proliferation and self‐renewal versus growth suppression and differentiation during normal and leukemic myelopoiesis are not understood. We have used the bi‐potent FDB1 myeloid cell line model, which is responsive to myelopoietic cytokines and activated mutants of the granulocyte macrophage‐colony stimulating factor (GM‐CSF) receptor, having differential signaling and leukemogenic activity. This model is suited to large‐scale gene‐profiling, and we have used a factorial time‐course design to generate a substantial and powerful data set. Linear modeling was used to identify gene‐expression changes associated with continued proliferation, differentiation, or leukemic receptor signaling. We focused on the changing transcription factor profile, defined a set of novel genes with potential to regulate myeloid growth and differentiation, and demonstrated that the FDB1 cell line model is responsive to forced expression of oncogenes identified in this study. We also identified gene‐expression changes associated specifically with the leukemic GM‐CSF receptor mutant, V449E. Signaling from this receptor mutant down‐regulates CCAAT/enhancer‐binding protein α (C/EBPα) target genes and generates changes characteristic of a specific acute myeloid leukemia signature, defined previously by gene‐expression profiling and associated with C/EBPα mutations.

A review of statistical estimators for risk-adjusted length of stay: analysis of the Australian and new Zealand intensive care adult patient data-base, 2008–2009

BackgroundFor the analysis of length-of-stay (LOS) data, which is characteristically right-skewed, a number of statistical estimators have been proposed as alternatives to the traditional ordinary least squares (OLS) regression with log dependent variable.MethodsUsing a cohort of patients identified in the Australian and New Zealand Intensive Care Society Adult Patient Database, 2008–2009, 12 different methods were used for estimation of intensive care (ICU) length of stay. These encompassed risk-adjusted regression analysis of firstly: log LOS using OLS, linear mixed model [LMM], treatment effects, skew-normal and skew-t models; and secondly: unmodified (raw) LOS via OLS, generalised linear models [GLMs] with log-link and 4 different distributions [Poisson, gamma, negative binomial and inverse-Gaussian], extended estimating equations [EEE] and a finite mixture model including a gamma distribution. A fixed covariate list and ICU-site clustering with robust variance were utilised for model fitting with split-sample determination (80%) and validation (20%) data sets, and model simulation was undertaken to establish over-fitting (Copas test). Indices of model specification using Bayesian information criterion [BIC: lower values preferred] and residual analysis as well as predictive performance (R2, concordance correlation coefficient (CCC), mean absolute error [MAE]) were established for each estimator.ResultsThe data-set consisted of 111663 patients from 131 ICUs; with mean(SD) age 60.6(18.8) years, 43.0% were female, 40.7% were mechanically ventilated and ICU mortality was 7.8%. ICU length-of-stay was 3.4(5.1) (median 1.8, range (0.17-60)) days and demonstrated marked kurtosis and right skew (29.4 and 4.4 respectively). BIC showed considerable spread, from a maximum of 509801 (OLS-raw scale) to a minimum of 210286 (LMM). R2 ranged from 0.22 (LMM) to 0.17 and the CCC from 0.334 (LMM) to 0.149, with MAE 2.2-2.4. Superior residual behaviour was established for the log-scale estimators. There was a general tendency for over-prediction (negative residuals) and for over-fitting, the exception being the GLM negative binomial estimator. The mean-variance function was best approximated by a quadratic function, consistent with log-scale estimation; the link function was estimated (EEE) as 0.152(0.019, 0.285), consistent with a fractional-root function.ConclusionsFor ICU length of stay, log-scale estimation, in particular the LMM, appeared to be the most consistently performing estimator(s). Neither the GLM variants nor the skew-regression estimators dominated.

Mortality and intensive care volume in ventilated patients from 1995 to 2009 in the Australian and New Zealand binational adult patient intensive care database

Objectives:The mortality outcome of mechanical ventilation, a key intervention in the critically ill, has been variously reported to be determined by intensive care patient volume. We determined the volume-(mortality)-outcome relationship of mechanically ventilated patients whose records were contributed to the Australian and New Zealand Intensive Care Society Adult Patient Database. Design, Setting, and Participants:Retrospective cohort study of 208,810 index patient admissions from 136 Australian and New Zealand intensive care units in the same number of hospitals over the course of 1995–2009. Measurements and Main Results:The patient–volume effect on hospital mortality, overall and at the level of patient (nonsurgical, elective surgical, and emergency surgical) and intensive care unit (rural/regional, metropolitan, tertiary, and private) descriptors, was determined by random-effects logistic regression adjusting for illness severity and demographic and geographical predictors. Annualized patient volume was modeled both as a categorical (deciles) and, with calendar year, a continuous variable using fractional polynomials. The patients were of mean age of 59 yrs (SD, 19 yrs), Acute Physiology and Chronic Health Evaluation III score 66 (32), and 39.4% female, with a hospital mortality of 22.4%. Overall and at both the patient and intensive care unit descriptor levels, no progressive decline in mortality was demonstrated across the annual patient volume range (12–932). Over the whole database, mortality odds ratio for the last volume decile (801–932 patients) was 1.26 (95% confidence interval, 1.06–1.50; p = .009) compared with the first volume decile (12–101 patients). Calendar year mortality decreases were evident (odds ratio, 0.96; 95% confidence interval, 0.94–0.98; p = .0001). Using fractional polynomials, modest curvilinear mortality increases (range, 5%–8%) across the volume range were noted over the whole database for nonsurgical patients and at the tertiary intensive care unit level. Conclusion:No inverse volume-(mortality)-outcome relationship was apparent for ventilated patients in the Australian and New Zealand Intensive Care Society database. Mechanisms for mortality increments with patient volume were not identified but warrant further study.

The inheritance of height in a Finnish population

Although height is widely cited as the classic polygenic trait, there have been few large-scale studies of its inheritance since that of Pearson and Lee 80 years ago (Pearson and Lee 1903). Values of heritability derived in many standard genetics texts are based on those data. The data of Rissanen and Nikkila (1977) on 2869 individuals in 392 three-generation families provide an opportunity to make comparisons in a contemporary European population. We examine here height differences between regions, sexes and generations, and consider the form of the population distribution. Analysing the data within the classic polygenic framework we obtain a much higher spouse correlation in height than shown by earlier studies, a smaller heritability, and only a small effect of common-sib environment or genetic dominance deviations.

Parameter orthogonality in mixed regression models for survival data

The implications of parameter orthogonality for the robustness of survival regression models are considered. The question of which of the proportional hazards or the accelerated life families of models would be more appropriate for analysis is usually ignored, and the proportional hazards family is applied, particularly in medicine, for convenience. Accelerated life models have conventionally been used in reliability applications. We propose a one-parameter family mixture survival model which includes both the accelerated life and the proportional hazards models. By orthogonalizing relative to the mixture parameter, we can show that, for small effects of the covariates, the regression parameters under the alternative families agree to within a constant. This recovers a known misspecification result. We use notions of parameter orthogonality to explore robustness to other types of misspecification including misspecified base-line hazards. The results hold in the presence of censoring. We also study the important question of when proportionality matters.

Gene profiling for determining pluripotent genes in a time course microarray experiment

In microarray experiments, it is often of interest to identify genes which have a prespecified gene expression profile with respect to time. Methods available in the literature are, however, typically not stringent enough in identifying such genes, particularly when the profile requires equivalence of gene expression levels at certain time points. In this paper, the authors introduce a new methodology, called gene profiling, that uses simultaneous differential and equivalent gene expression level testing to rank genes according to a prespecified gene expression profile. Gene profiling treats the vector of true gene expression levels as a linear combination of appropriate vectors, for example, vectors that give the required criteria for the profile. This gene profile model is fitted to the data, and the resulting parameter estimates are summarized in a single test statistic that is then used to rank the genes. The theoretical underpinnings of gene profiling (equivalence testing, intersection-union tests) are discussed in this paper, and the gene profiling methodology is applied to our motivating stem-cell experiment.