Modelled estimates of hospitalisations attributable to respiratory syncytial virus and influenza in Australia, 2009-2017

Background

Respiratory syncytial virus (RSV) and influenza are important causes of disease in children and adults. In Australia, information on the burden of RSV in adults is particularly limited.

Methods

We used time series analysis to estimate respiratory, acute respiratory infection, pneumonia and influenza, and bronchiolitis hospitalisations attributable to RSV and influenza in Australia during 2009 through 2017. RSV and influenza‐coded hospitalisations in <5‐year‐olds were used as proxies for relative weekly viral activity.

Results

From 2009 to 2017, the estimated all‐age average annual rates of respiratory hospitalisations attributable to RSV and seasonal influenza (excluding 2009) were 54.8 (95% confidence interval [CI]: 20.1, 88.8) and 87.8 (95% CI: 74.5, 97.7) per 100,000, respectively. The highest estimated average annual RSV‐attributable respiratory hospitalisation rate per 100,000 was 464.2 (95% CI: 285.9, 641.2) in <5‐year‐olds. For seasonal influenza, it was 521.6 (95% CI: 420.9, 600.0) in persons aged ≥75 years. In ≥75‐year‐olds, modelled estimates were approximately eight and two times the coded estimates for RSV and seasonal influenza, respectively.

Conclusions

RSV and influenza are major causes of hospitalisation in young children and older adults in Australia, with morbidity underestimated by hospital diagnosis codes.

Inverse problem approach to regularized regression models with application to predicting recovery after stroke

Regression modelling is a powerful statistical tool often used in biomedical and clinical research. It could be formulated as an inverse problem that measures the discrepancy between the target outcome and the data produced by representation of the modelled predictors. This approach could simultaneously perform variable selection and coefficient estimation. We focus particularly on a linear regression issue, Y ∼ N ( X β , σ I n ) , where β ∈ R p is the parameter of interest and its components are the regression coefficients. The inverse problem finds an estimate for the parameter β , which is mapped by the linear operator ( L : β ⟶ X β ) to the observed outcome data Y = X β + ε . This problem could be conveyed by finding a solution in the affine subspaceL - 1 ( Y ) . However, in the presence of collinearity, high-dimensional data and high conditioning number of the related covariance matrix, the solution may not be unique, so the introduction of prior information to reduce the subsetL - 1 ( Y ) and regularize the inverse problem is needed. Informed by Huber's robust statistics framework, we propose an optimal regularizer to the regression problem. We compare results of the proposed method and other penalized regression regularization methods: ridge, lasso, adaptive-lasso and elastic-net under different strong hypothesis such as high conditioning number of the covariance matrix and high error amplitude, on both simulated and real data from the South London Stroke Register. The proposed approach can be extended to mixed regression models. Our inverse problem framework coupled with robust statistics methodology offer new insights in statistical regression and learning. It could open a new research development for model fitting and learning.

On the aggregation of published prognostic scores for causal inference in observational studies

As real world evidence on drug efficacy involves nonrandomized studies, statistical methods adjusting for confounding are needed. In this context, prognostic score (PGS) analysis has recently been proposed as a method for causal inference. It aims to restore balance across the different treatment groups by identifying subjects with a similar prognosis for a given reference exposure ("control"). This requires the development of a multivariable prognostic model in the control arm of the study sample, which is then extrapolated to the different treatment arms. Unfortunately, large cohorts for developing prognostic models are not always available. Prognostic models are therefore subject to a dilemma between overfitting and parsimony; the latter being prone to a violation of the assumption of no unmeasured confounders when important covariates are ignored. Although it is possible to limit overfitting by using penalization strategies, an alternative approach is to adopt evidence synthesis. Aggregating previously published prognostic models may improve the generalizability of PGS, while taking account of a large set of covariates-even when limited individual participant data are available. In this article, we extend a method for prediction model aggregation to PGS analysis in nonrandomized studies. We conduct extensive simulations to assess the validity of model aggregation, compared with other methods of PGS analysis for estimating marginal treatment effects. We show that aggregating existing PGS into a "meta-score" is robust to misspecification, even when elementary scores wrongfully omit confounders or focus on different outcomes. We illustrate our methods in a setting of treatments for asthma.

Using the Choice Sequence in Time Trade-Off as Discrete Choices: Do the Two Stories Match?

OBJECTIVES

The EQ-5D-5L valuation protocol recommends combining time trade-off (TTO) and discrete choice experiments (DCEs). DCEs that include a duration attribute (DCE_TTO) allow modeling on the quality-adjusted life-year scale. Because the choice sequence in a TTO can be construed as a series of DCE_TTO, we used data from a single TTO study to investigate the extent to which DCE values match TTO values when based on identical preferences.

METHODS

In a TTO design in which a fixed set of choices were administered without termination at preference indifference, 202 individuals each valued 10 EQ-5D health states. From identified indifference points, we estimated three sets of TTO values: (i) plotting means and (ii) applying censored regressions at -1 and 1. Using all strict preferences, we (iii) estimated DCE_TTO values with a logit model and a bootstrap procedure.

RESULTS

Estimated DCE_TTO and TTO values agreed well at the severe end of the quality-adjusted life-year scale, but with decreasing severity, DCE_TTO values were higher than TTO-values, with the difference peaking at 0.37 for the mildest health state. Left-censoring TTO values at -1 worsen the agreement for the worst health states and did not affect health states. Right censoring at 1 improved the agreement for mild states.

CONCLUSIONS

TTO and the DCE_TTO values estimated from the same preference data diverged, with increasing difference for milder health states. Although the values converged when applying censored regression at +1, we question the validity of this adjustment.

Development of an FTIR based chemometric model for the qualitative and quantitative evaluation of cane sugar as an added sugar adulterant in apple fruit juices

A Fourier Transform Infrared Spectroscopy based chemometric model was evaluated for the rapid identification and estimation of cane sugar as an added sugar adulterant in apple fruit juices. For all the ninety samples, spectra were acquired in the mid-infrared range (4000 cm^-1-400 cm^-1). The spectral analysis provided information regarding the distinctive variable region, which lies in the range of 1200cm^-1 to 900cm^-1, designated as fingerprint region for the carbohydrates. A specific peak in the fingerprint region was observed at 997cm^-1 in all the adulterated samples and was undetectable in pure samples. Based on different levels of cane sugar adulteration (5, 10, 15, and 20%), principal component analysis showed the clustering of samples and further helped us in compression of data by selecting wavenumbers with maximum variability based on the loading line plot. Supervised classification methods (SIMCA and LDA) were evaluated based on their classification efficiencies for a test set. Though SIMCA showed 100% classification efficiency (Raw data set), LDA was able to classify the test set with an accuracy of only 96.67% (Raw as well as Transformed data set) between pure and 5% adulterated samples. For the quantitative estimation, calibration models were developed using partial least square regression (PLS-R) and principal component regression method (PCR) methods. PLS-1^st derivative showed a maximum coefficient of determination (R²) with a value of 0.991 for calibration and 0.992 for prediction. The RMSECV, RMSEP, LOD and LOQ observed for PLS-1^st derivative model were 0.75% w/v, 0.61% w/v, 1.28%w/v and 3.88%w/v, respectively. The coefficient of variation as a measure of precision (repeatability) was also determined for all models, and it ranged from 0.23% to 1.83% (interday), and 0.25% to 1.43% (intraday).

Heavy Metal Soil Contamination Detection Using Combined Geochemistry and Field Spectroradiometry in the United Kingdom

Technological advances in hyperspectral remote sensing have been widely applied in heavy metal soil contamination studies, as they are able to provide assessments in a rapid and cost-effective way. The present work investigates the potential role of combining field and laboratory spectroradiometry with geochemical data of lead (Pb), zinc (Zn), copper (Cu) and cadmium (Cd) in quantifying and modelling heavy metal soil contamination (HMSC) for a floodplain site located in Wales, United Kingdom. The study objectives were to: (i) collect field- and lab-based spectra from contaminated soils by using ASD FieldSpec^® 3, where the spectrum varies between 350 and 2500 nm; (ii) build field- and lab-based spectral libraries; (iii) conduct geochemical analyses of Pb, Zn, Cu and Cd using atomic absorption spectrometer; (iv) identify the specific spectral regions associated to the modelling of HMSC; and (v) develop and validate heavy metal prediction models (HMPM) for the aforementioned contaminants, by considering their spectral features and concentrations in the soil. Herein, the field- and lab-based spectral features derived from 85 soil samples were used successfully to develop two spectral libraries, which along with the concentrations of Pb, Zn, Cu and Cd were combined to build eight HMPMs using stepwise multiple linear regression. The results showed, for the first time, the feasibility to predict HMSC in a highly contaminated floodplain site by combining soil geochemistry analyses and field spectroradiometry. The generated models help for mapping heavy metal concentrations over a huge area by using space-borne hyperspectral sensors. The results further demonstrated the feasibility of combining geochemistry analyses with filed spectroradiometric data to generate models that can predict heavy metal concentrations.

Can point-of-care testing shorten hospitalization length of stay? An exploratory investigation of infectious agents using regression modelling

This retrospective study investigates the potential benefits from the introduction of point-of-care tests for rapid diagnosis of infectious diseases. We analysed a sample of 441 hospitalized patients who had received a final diagnosis related to 18 pathogenic agents. These pathogens were mostly detected by standard tests but were also detectable by point-of-care testing. The length of hospital stay was partitioned into pre- and post-laboratory diagnosis stages. Regression analysis and elementary queueing theory were applied to estimate the impact of quick diagnosis on the mean length of stay and the utilization of healthcare resources. The analysis suggests that eliminating the pre-diagnosis times through point-of-care testing could shorten the mean length of hospital stay for infectious diseases by up to 34 per cent and result in an equal reduction in bed occupancy and other resources. Regression and other more sophisticated models can aid the financing decision-making of pilot point-of-care laboratories in healthcare systems.

Compressor map regression modelling based on partial least squares

In this work, two kinds of partial least squares modelling methods are applied to predict a compressor map: one uses a power function polynomial as the basis function (PLSO), and the other uses a trigonometric function polynomial (PLSN). To demonstrate the potential capabilities of PLSO and PLSN for a typical interpolated prediction and an extrapolated prediction, they are compared with two other classical data-driven modelling methods, namely the look-up table and artificial neural network (ANN). PLSO and PLSN are also compared with each other. The results show that PLSO and PLSN have a better prediction performance than the look-up table and the ANN, especially for the extrapolated prediction. The computational time is also decreased sharply. Compared with PLSO, PLSN is characterized by a higher prediction accuracy and shorter computational time than PLSO. It is expected that PLSN could save computational time and also improve the accuracy of a thermodynamic model of a diesel engine.

Reducing cost drivers in total joint arthroplasty: understanding patient readmission risk and supply cost

Introduction: Understanding and planning for the factors that impact supply cost and unplanned readmission risk for total joint arthroplasty (TJA) patients is helpful for hospitals at financial risk under bundled payments. Readmission and operating room supply costs are two of the biggest expenses. Methods: Logistic and linear regressions are used to measure the impacts of TJA patient attributes on readmission risk and supply costs, respectively. Results: Patients' health market segment and the number/type of comorbidity impacts 30/90-day readmission rates. Surgeon implant preference and type of surgery impact supply costs. Discharge location and two of the five health market segments increase the odds of 30-day readmission. Arrhythmia and lymphoma are the primary comorbidities that impact the odds of readmission at 90 days. Conclusions: Preoperatively identifying TJA patients likely to have large supply costs and higher readmission risk allows hospitals to invest in low-cost interventions to reduce risk and improve healthcare value.

Modelling the effects of standard prognostic factors in node-positive breast cancer. German Breast Cancer Study Group (GBSG)

Prognostic models that predict the clinical course of a breast cancer patient are important in oncology. We propose an approach to constructing such models based on fractional polynomials in which useful transformations of the continuous factors are determined. The idea may be applied with all types of regression model, including Cox regression, the method of choice for survival-time data. We analyse a prospective study of node-positive breast cancer. Seven standard prognostic factors--age, menopausal status, tumour size, tumour grade, number of positive lymph nodes, progesterone and oestrogen receptor concentrations--were investigated in 686 patients, of whom 299 had an event for recurrence-free survival and 171 died. We determine a final model with transformations of prognostic factors and compare it with the more traditional approaches using categorized variables or assuming a straight line relationship. We conclude that analysis using fractional polynomials can extract important prognostic information which the traditional approaches may miss.