4.5 Article

Bayesian graphical models for regression on multiple data sets with different variables

Journal

BIOSTATISTICS
Volume 10, Issue 2, Pages 335-351

Publisher

OXFORD UNIV PRESS
DOI: 10.1093/biostatistics/kxn041

Keywords

Air pollution; Confounding; Data synthesis; Low birth weight; Multiple imputation

Funding

  1. Economic and Social Research Council [RES-576-25-5003]
  2. Department of Health [PHI/03/C1/045]
  3. Medical Research Council [U. 1052.00.008, U. 1052.00.001]
  4. MRC [MC_U105232027] Funding Source: UKRI
  5. Economic and Social Research Council [RES-576-25-5003] Funding Source: researchfish
  6. Medical Research Council [MC_U105232027] Funding Source: researchfish

Ask authors/readers for more resources

Routinely collected administrative data sets, such as national registers, aim to collect information on a limited number of variables for the whole population. In contrast, survey and cohort studies contain more detailed data from a sample of the population. This paper describes Bayesian graphical models for fitting a common regression model to a combination of data sets with different sets of covariates. The methods are applied to a study of low birth weight and air pollution in England and Wales using a combination of register, survey, and small-area aggregate data. We discuss issues such as multiple imputation of confounding variables missing in one data set, survey selection bias, and appropriate propagation of information between model components. From the register data, there appears to be an association between low birth weight and environmental exposure to NO(2), but after adjusting for confounding by ethnicity and maternal smoking by combining the register and survey data under our models, we find there is no significant association. However, NO(2) was associated with a small but significant reduction in birth weight, modeled as a continuous variable.

Authors

I am an author on this paper
Click your name to claim this paper and add it to your profile.

Reviews

Primary Rating

4.5
Not enough ratings

Secondary Ratings

Novelty
-
Significance
-
Scientific rigor
-
Rate this paper

Recommended

Article Health Care Sciences & Services

Reference Case Methods for Expert Elicitation in Health Care Decision Making

Laura Bojke, Marta O. Soares, Karl Claxton, Abigail Colson, Aimee Fox, Chris Jackson, Dina Jankovic, Alec Morton, Linda D. Sharples, Andrea Taylor

Summary: This article focuses on establishing reference case methods for expert elicitation to inform health care decision making. Methods choices for elicitation lack empirical evidence supporting recommendations, but reference case methods have been determined for health technology assessment. The reference methods prioritize experts with substantive knowledge, eliciting quantities directly observed by experts, and individual elicitation of beliefs. Further applied examples and experimental evidence comparing methods are recommended.

MEDICAL DECISION MAKING (2022)

Article Health Care Sciences & Services

Multilevel and Quasi Monte Carlo Methods for the Calculation of the Expected Value of Partial Perfect Information

Wei Fang, Zhenru Wang, Michael B. Giles, Chris H. Jackson, Nicky J. Welton, Christophe Andrieu, Howard Thom

Summary: This article explores the computational methods for estimating the expected value of partial perfect information (EVPPI), introducing the potential of quasi Monte Carlo (QMC) and multilevel Monte Carlo (MLMC) estimation techniques. The study found that QMC and MLMC offer substantial computational savings in situations where parameter sets are large and correlated, and EVPPI is large.

MEDICAL DECISION MAKING (2022)

Review Health Care Sciences & Services

Estimating a time-to-event distribution from right-truncated data in an epidemic: A review of methods

Shaun R. Seaman, Anne Presanis, Christopher Jackson

Summary: Time-to-event data can be right-truncated, leading to biased distribution towards shorter times in the sample compared to the population. This article reviews statistical methods to address this bias, particularly in the context of infectious disease epidemics like COVID-19. Issues of identifiability, estimation of time-to-event distribution, and the effects of covariates are discussed, with an illustration using data on individuals who died with coronavirus disease by April 5, 2020.

STATISTICAL METHODS IN MEDICAL RESEARCH (2022)

Article Microbiology

Improving local prevalence estimates of SARS-CoV-2 infections using a causal debiasing framework

George Nicholson, Brieuc Lehmann, Tullia Padellini, Koen B. Pouwels, Radka Jersakova, James Lomax, Ruairidh E. King, Ann-Marie Mallon, Peter J. Diggle, Sylvia Richardson, Marta Blangiardo, Chris Holmes

Summary: Global and national surveillance of SARS-CoV-2 epidemiology is based on targeted testing of symptomatic individuals, but these data are often biased. A causal framework described in the study provides debiased estimates by combining targeted test counts and randomized surveillance data.

NATURE MICROBIOLOGY (2022)

Article Medicine, General & Internal

Trends in outcomes following COVID-19 symptom onset in Milan: a cohort study

Christopher H. Jackson, Francesca Grosso, Kevin Kunzmann, Alice Corbella, Maria Gramegna, Marcello Tirani, Silvana Castaldi, Danilo Cereda, Daniela De Angelis, Anne Presanis

Summary: This study investigates the risks of hospital admission, death without hospital admission, and recovery without admission for people with symptomatic COVID-19 in Milan, Italy. The results show that the risk of death without hospital admission was higher during the severe burden on the healthcare system, especially for care home residents. The outcomes improved as the first wave of the pandemic waned, community healthcare resources were reinforced, and testing became more widely available.

BMJ OPEN (2022)

Article Public, Environmental & Occupational Health

Linkage of multiple electronic health record datasets using a 'spine linkage' approach compared with all 'pairwise linkages'

Helen A. Blake, Linda D. Sharples, Katie Harron, Jan H. van der Meulen, Kate Walker

Summary: The study compares the pairwise and spine linkage approaches using real-world data, and finds that the spine linkage method can be an efficient alternative with little difference compared to the pairwise linkage approach.

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY (2022)

Article Health Care Sciences & Services

A comparison of two frameworks for multi-state modelling, applied to outcomes after hospital admissions with COVID-19

Christopher H. Jackson, Brian D. M. Tom, Peter D. Kirwan, Sema Mandal, Shaun R. Seaman, Kevin Kunzmann, Anne M. Presanis, Daniela De Angelis

Summary: This study compares two multi-state modelling frameworks for analyzing hospital admission data of COVID-19 patients. The findings suggest that using 'cure-rate' models to define transition-specific hazards can better represent the low risk of certain groups for intensive care unit admission.

STATISTICAL METHODS IN MEDICAL RESEARCH (2022)

Article Health Care Sciences & Services

Methods of analysis for survival outcomes with time-updated mediators, with application to longitudinal disease registry data

Kamaryn T. Tanner, Linda D. Sharples, Rhian M. Daniel, Ruth H. Keogh

Summary: This study introduces two methods, path-specific effects and additive hazards model with exposure splitting, to address statistical challenges in mediation analysis for time-to-event outcomes and longitudinal observational data. Both methods are applied to observational datasets and found to produce unbiased estimates, although estimates may be biased if the data are measured infrequently.

STATISTICAL METHODS IN MEDICAL RESEARCH (2022)

Article Biology

Contrasting principal stratum and hypothetical strategy estimands in multi-period crossover trials with incomplete data

John N. S. Matthews, Sofia Bazakou, Robin Henderson, Linda D. Sharples

Summary: This paper focuses on complete case analysis of complete crossover designs, which can be used to estimate the difference between principal stratum strategy and hypothetical strategy estimands. The results show that the difference between these estimands exceeds 5% only when the trial is severely affected by dropouts or if the within-subject correlation is low.

BIOMETRICS (2023)

Article Endocrinology & Metabolism

Mediation of the total effect of cystic fibrosis-related diabetes on mortality: A UK Cystic Fibrosis Registry cohort study

Kamaryn T. Tanner, Rhian M. Daniel, Diana Bilton, Nicholas J. Simmonds, Linda D. Sharples, Ruth H. Keogh

Summary: This study investigated the potential mediating factors, including lung function, pulmonary exacerbations, and nutritional status, through which cystic fibrosis-related diabetes (CFRD) affects the composite outcome of mortality or transplant in adult cystic fibrosis (CF) patients. The results showed that pulmonary exacerbations were the strongest mediator of the effect, attributing to a significant proportion of the total effect. Other factors may also contribute to the influence of CFRD on patient survival.

DIABETIC MEDICINE (2022)

Article Immunology

Time-varying Comparison of All-cause Mortality After Liver Transplantation Between Recipients With and Without Hepatocellular Carcinoma: A Population-based Cohort Study Using the United Kingdom Liver Transplant Registry

Jyoti Sehjal, Linda D. Sharples, Ruth H. Keogh, Kate Walker, Andreas Prachalias, Nigel Heaton, Tommy Ivanics, Jan van der Meulen, David Wallace

Summary: This study reveals time-varying differences in the hazard of all-cause mortality after liver transplantation between hepatocellular carcinoma (HCC) recipients and non-HCC recipients. It also provides evidence that some HCC patients have extra-hepatic spread at the time of transplantation, which has implications for post-transplant surveillance protocols.

TRANSPLANTATION (2022)

Article Endocrinology & Metabolism

Multicenter Randomized Trial of Intermittently Scanned Continuous Glucose Monitoring Versus Self-Monitoring of Blood Glucose in Individuals With Type 2 Diabetes and Recent-Onset Acute Myocardial Infarction: Results of the LIBERATES Trial

Ramzi A. Ajjan, Simon R. Heller, Colin C. Everett, Armando Vargas-Palacios, Ruchi Higham, Linda Sharples, Diana A. Gorog, Alice Rogers, Catherine Reynolds, Catherine Fernandez, Pedro Rodrigues, Thozhukat Sathyapalan, Robert F. Storey, Deborah D. Stocken

Summary: This study analyzed the impact of modern glucose-monitoring strategies on glycemic and patient-related outcomes in individuals with type 2 diabetes (T2D) and recent myocardial infarction (MI). The results showed that compared with self-monitoring of blood glucose (SMBG), intermittently scanned continuous glucose monitoring (isCGM) marginally increased time in range and significantly reduced hypoglycemic exposure in T2D individuals with MI, while equally improving HbA(1c). These findings explain the cost effectiveness of isCGM.

DIABETES CARE (2023)

Article Medicine, General & Internal

Effectiveness of GRACE risk score in patients admitted to hospital with non-ST elevation acute coronary syndrome (UKGRIS): parallel group cluster randomised controlled trial

Chris P. Gale, Deborah Stocken, Suleman Aktaa, Catherine Reynolds, Rachael Gilberts, David Brieger, Kathryn Carruthers, Derek P. Chew, Shaun G. Goodman, Catherine Fernandez, Linda Sharples, Andrew Yan, Keith Fox

Summary: This study aimed to determine the effectiveness of risk stratification using the Global Registry of Acute Coronary Events (GRACE) risk score (GRS) for patients with suspected non-ST elevation acute coronary syndrome. The results showed that the use of GRS did not improve adherence to guideline recommended management or reduce cardiovascular events at 12 months.

BMJ-BRITISH MEDICAL JOURNAL (2023)

Article Medicine, General & Internal

Sociodemographic characteristics and longitudinal progression of multimorbidity: A multistate modelling analysis of a large primary care records dataset in England

Sida Chen, Tom Marshall, Christopher Jackson, Jennifer Cooper, Francesca Crowe, Krish Nirantharakumar, Catherine L. Saunders, Paul Kirk, Sylvia Richardson, Duncan Edwards, Simon Griffin, Christopher Yau, Jessica K. Barrett

Summary: Multimorbidity, characterized by the coexistence of multiple chronic conditions, is a rising public health concern. This study utilized electronic primary care records from England to examine the associations between sociodemographic characteristics and the rate of disease progression. The findings indicate strong associations between social deprivation, gender, and age with disease diagnosis, which tend to attenuate with an increase in the number of preexisting conditions.

PLOS MEDICINE (2023)

Letter Cardiac & Cardiovascular Systems

Reply to Garcia-Villarreal et al.

Samer A. M. Nashef, Linda D. Sharples

EUROPEAN JOURNAL OF CARDIO-THORACIC SURGERY (2022)

No Data Available