Use of posterior predictive assessments to evaluate model fit in multilevel logistic regression

Associations between Johne's disease and fertility in UK dairy herds

The objective of this observational study was to quantify associations between Mycobacterium avium subspecies paratuberculosis (MAP) antibody status and a variety of fertility outcomes, in UK dairy cattle. Longitudinal milk recording, fertility and MAP antibody enzyme-linked immunosorbent assay (ELISA) milk test data were collated retrospectively from 121,762 lactations in 78 herds. Datasets were structured into appropriate units to suit outcomes and enable temporal association between current and future MAP status, and fertility measures. Current MAP status was categorised according to most recent status within 180 days, with time-related future MAP status assigned based on MAP antibody ELISA milk test data for each cow. Multilevel multivariable logistic regression models were used to evaluate associations between MAP status and 21-day pregnancy and submission rate and conception risk. Posterior predictions and cross-validation techniques were used to assess model fit and check model building assumptions. A negative association was found between risk of insemination (Odds Ratio [OR], 0.78; 95% Credible Interval [CI], 0.66-0.92) and conception occurring (OR, 0.65; CI, 0.5-0.84) and transition from negative to non-negative MAP test status in the next 30-90 days. A positive association was observed between risk of insemination (OR, 1.34; CI, 1.16-1.52) and conception occurring (OR, 1.26; CI, 1.11-1.43) and transition from negative to non-negative MAP test status in the next 90-180 days. Current positive MAP test status was negatively and positively associated with insemination (OR, 0.59; CI, 0.49-0.70) and conception risk (OR, 1.12; CI, 0.96-1.30), respectively. Herd managers will have had access to test results, declaring cows with past recent or multiple positive MAP antibody ELISA results not to be bred, negatively influencing insemination risk. Overall, these results demonstrate the temporal association between a positive MAP antibody ELISA result and dairy cow fertility outcomes, with particular variability prior to a positive MAP antibody ELISA result.

Appropriateness of the current parasitological control target for hookworm morbidity: A statistical analysis of individual-level data

BACKGROUND

Soil-transmitted helminths affect almost 2 billion people globally. Hookworm species contribute to most of the related morbidity. Hookworms mainly cause anaemia, due to blood loss at the site of the attachment of the adult worms to the human intestinal mucosa. The World Health Organization (WHO) aims to eliminate hookworm morbidity by 2030 through achieving a prevalence of moderate and heavy intensity (M&HI) infections below 2%. In this paper, we aim to assess the suitability of this threshold to reflect hookworm-attributable morbidity.

METHODOLOGY/PRINCIPAL FINDINGS

We developed a hierarchical statistical model to simulate individual haemoglobin concentrations in association with hookworm burdens, accounting for low haemoglobin values attributable to other causes. The model was fitted to individual-level data within a Bayesian framework. Then, we generated different endemicity settings corresponding to infection prevalence ranging from 10% to 90% (0% to 55% M&HI prevalence), using 1, 2 or 4 Kato-Katz slides. For each scenario, we estimated the prevalence of anaemia due to hookworm. Our results showed that on average, haemoglobin falls below the WHO threshold for anaemia when intensities are above 2000 eggs per gram of faeces. For the different simulated scenarios, the estimated prevalence of anaemia attributable to hookworm ranges from 0% to 30% (95%-PI: 24% - 36%) being mainly associated to the prevalence of M&HI infections. Simulations show that a 2% prevalence of M&HI infections in adults corresponds to a prevalence of hookworm-attributable anaemia lower than 1%.

CONCLUSIONS/SIGNIFICANCE

Our results support the use of the current WHO thresholds of 2% prevalence of M&HI as a proxy for hookworm morbidity. A single Kato-Katz slide may be sufficient to assess the achievement of the morbidity target. Further studies are needed to elucidate haemoglobin dynamics pre- and post- control, ideally using longitudinal data in adults and children.

Global Pediatric Peripheral Intravenous Catheter Practice and Performance: A Secondary Analysis of 4206 Catheters

PURPOSE

To describe worldwide characteristics, performance and risk factors of peripheral intravenous catheters (PIVCs), in pediatrics.

DESIGN

A secondary, subgroup analysis of pediatric (<18 years) data was undertaken, using a global, cross-sectional study of PIVCs. Practice characteristics included: demographic, diagnostic, utility, management, performance and resources. Multivariate regression identified complication risks factors.

RESULTS

Data from 4206 children in 278 hospitals across 47 countries. Most PIVCs (outside of Australia, New Zealand) were inserted by nurses (71%; n = 2950), with dedicated teams only common in North America (23.2%; n = 85). Large gauges (≤18G) were mostly used in South America, Europe and Africa. Regions predominantly placed 24G (49%; n = 2060) except in Australia and New Zealand, who more commonly placed 22G (38.7%; n = 192). The most common placement was the hand (51%; n = 2143), however North America, Australia and New Zealand frequently utilised the antecubital fossa (24.5%, n = 90; 21.4%; n = 106). Polyurethane dressings were most used (67.1%; 2822), and many were not clean, dry and intact (17.1%; n = 715). Over 8% of PIVCs were idle, with the highest rates in North America (21.2%; n = 78). PIVC local complication risk factors included: >2 years age (odds ratio [OR] > 1.58; 1.2-2.1); ambulance/emergency insertion (OR 1.65; 1.2-2.3); upper arm/antecubital placement (OR 1.44; 1.1-2.0); poor dressing integrity (OR 5.4; 4.2-6.9); and 24-72 h dwell (OR > 1.9; 1.3-2.6).

CONCLUSIONS

There is global inconsistency in pediatric PIVC practice, which may be causing harm.

CLINICAL IMPLICATIONS

Improvements in pediatric PIVC placement, dressings, and gauge selection are needed.

Bayesian regression models for the estimation of net cost of disease using aggregate data

Estimation of net costs attributed to a disease or other health condition is very important for health economists and policy makers. Skewness and heteroscedasticity are well-known characteristics for cost data, making linear models generally inappropriate and dictating the use of other types of models, such as gamma regression. Additional hurdles emerge when individual level data are not available. In this paper, we consider the latter case were data are only available at the aggregate level, containing means and standard deviations for different strata defined by a number of demographic and clinical factors. We summarize a number of methods that can be used for this estimation, and we propose a Bayesian approach that utilizes the sample stratum specific standard deviations as stochastic. We investigate the performance of two linear mixed models, comparing them with two proposed gamma regression mixed models, to analyze simulated data generated by gamma and log-normal distributions. Our proposed Bayesian approach seems to have significant advantages for net cost estimation when only aggregate data are available. The implemented gamma models do not seem to offer the expected benefits over the linear models; however, further investigation and refinement is needed.

Onchocerciasis: the pre-control association between prevalence of palpable nodules and skin microfilariae

BACKGROUND

The prospect of eliminating onchocerciasis from Africa by mass treatment with ivermectin has been rejuvenated following recent successes in foci in Mali, Nigeria and Senegal. Elimination prospects depend strongly on local transmission conditions and therefore on pre-control infection levels. Pre-control infection levels in Africa have been mapped largely by means of nodule palpation of adult males, a relatively crude method for detecting infection. We investigated how informative pre-control nodule prevalence data are for estimating the pre-control prevalence of microfilariae (mf) in the skin and discuss implications for assessing elimination prospects.

METHODS AND FINDINGS

We analyzed published data on pre-control nodule prevalence in males aged ≥ 20 years and mf prevalence in the population aged ≥ 5 years from 148 African villages. A meta-analysis was performed by means of Bayesian hierarchical multivariate logistic regression, accounting for measurement error in mf and nodule prevalence, bioclimatic zones, and other geographical variation. There was a strong positive correlation between nodule prevalence in adult males and mf prevalence in the general population. In the forest-savanna mosaic area, the pattern in nodule and mf prevalence differed significantly from that in the savanna or forest areas.

SIGNIFICANCE

We provide a tool to convert pre-control nodule prevalence in adult males to mf prevalence in the general population, allowing historical data to be interpreted in terms of elimination prospects and disease burden of onchocerciasis. Furthermore, we identified significant geographical variation in mf prevalence and nodule prevalence patterns warranting further investigation of geographical differences in transmission patterns of onchocerciasis.

Bayesian shared spatial-component models to combine and borrow strength across sparse disease surveillance sources

When analyzing the geographical variations of disease risk, one common problem is data sparseness. In such a setting, we investigate the possibility of using Bayesian shared spatial component models to strengthen inference and correct for any spatially structured sources of bias, when distinct data sources on one or more related diseases are available. Specifically, we apply our models to analyze the spatial variation of risk of two forms of scrapie infection affecting sheep in Wales (UK) using three surveillance sources on each disease. We first model each disease separately from the combined data sources and then extend our approach to jointly analyze diseases and data sources. We assess the predictive performances of several nested joint models through pseudo cross-validatory predictive model checks.

Statistical analysis of genetic interactions

Many common human diseases and complex traits are highly heritable and influenced by multiple genetic and environmental factors. Although genome-wide association studies (GWAS) have successfully identified many disease-associated variants, these genetic variants explain only a small proportion of the heritability of most complex diseases. Genetic interactions (gene-gene and gene-environment) substantially contribute to complex traits and diseases and could be one of the main sources of the missing heritability. This paper provides an overview of the available statistical methods and related computer software for identifying genetic interactions in animal and plant experimental crosses and human genetic association studies. The main discussion falls under the three broad issues in statistical analysis of genetic interactions: the definition, detection and interpretation of genetic interactions. Recently developed methods based on modern techniques for high-dimensional data are reviewed, including penalized likelihood approaches and hierarchical models; the relationships between these methods are also discussed. I conclude this review by highlighting some areas of future research.