Nested case-control and case-cohort methods of sampling from a cohort: a critical comparison

Plasma Metabolites and Life's Simple 7 in REGARDS

BACKGROUND

The American Heart Association's Life's Simple 7 (LS7) is a health metric that captures important factors associated with cardiovascular and cerebrovascular health. Previous studies highlight the potential of plasma metabolites to serve as a marker for lifestyle and health behavior that could be a target for stroke prevention. The objectives of this study were to identify metabolites that were associated with LS7 and incident ischemic stroke and mediate the relationship between the two.

METHODS

Targeted metabolomic profiling of 162 metabolites by liquid chromatography-tandem mass spectrometry was used to identify candidate metabolites in a stroke case-cohort nested within the REGARDS study (Reasons for Geographic and Racial Differences in Stroke). Weighted linear regression and weighted Cox proportional hazard models were used to identify metabolites that were associated with LS7 and incident ischemic stroke, respectively. Effect measures were based on a 1-SD change in metabolite level. Metabolite mediators were examined using inverse odds ratio weighting mediation analysis.

RESULTS

The study comprised 1075 ischemic stroke cases and 968 participants in the random cohort sample. Three out of 162 metabolites were associated with the overall LS7 score including guanosine (β, -0.46 [95% CI, -0.65 to -0.27]; P=2.87×10-6), cotinine (β, -0.49 [95% CI, -0.70 to -0.28]; P=7.74×10-6), and acetylneuraminic acid (β, -0.59 [95% CI, -0.77 to -0.42]; P=4.29×10-11). Guanosine (hazard ratio, 1.47 [95% CI, 1.31-1.65]; P=6.97×10-11), cotinine (hazard ratio, 1.30 [95% CI, 1.16-1.44]; P=2.09×10-6), and acetylneuraminic acid (hazard ratio, 1.29 [95% CI, 1.15-1.45]; P=9.24×10-6) were associated with incident ischemic stroke. The mediation analysis identified guanosine (27% mediation, indirect effect; P=0.002), cotinine (30% mediation, indirect effect; P=0.004), and acetylneurminic acid (22% mediation, indirect effect; P=0.041) partially mediated the relationship between LS7 and ischemic stroke.

CONCLUSIONS

We identified guanosine, cotinine, and acetylneuraminic acid that were associated with LS7, incident ischemic stroke, and mediated the relationship between LS7 and ischemic stroke.

Privacy-preserving analysis of time-to-event data under nested case-control sampling

Analyses of distributed data networks of rare diseases are constrained by legitimate privacy and ethical concerns. Analytical centers (e.g. research institutions) are thus confronted with the challenging task of obtaining data from recruiting sites that are often unable or unwilling to share personal records of participants. For time-to-event data, recently popularized disclosure techniques with privacy guarantees (e.g., Differentially Private Generative Adversarial Networks) are generally computationally expensive or inaccessible to applied researchers. To perform the widely used Cox proportional hazards regression, we propose an easy-to-implement privacy-preserving data analysis technique by pooling (i.e. aggregating) individual records of covariates at recruiting sites under the nested case-control sampling framework before sharing the pooled nested case-control subcohort. We show that the pooled hazard ratio estimators, under the pooled nested case-control subsamples from the contributing sites, are maximum likelihood estimators and provide consistent estimates of the individual level full cohort HRs. Furthermore, a sampling technique for generating pseudo-event times for individual subjects that constitute the pooled nested case-control subsamples is proposed. Our method is demonstrated using extensive simulations and analysis of the National Lung Screening Trial data. The utility of our proposed approach is compared to the gold standard (full cohort) and synthetic data generated using classification and regression trees. The proposed pooling technique performs to near-optimal levels comparable to full cohort analysis or synthetic data; the efficiency improves in rare event settings when more controls are matched on during nested case-control subcohort sampling.

The Childhood Acute Illness and Nutrition (CHAIN) network nested case-cohort study protocol: a multi-omics approach to understanding mortality among children in sub-Saharan Africa and South Asia

Introduction: Many acutely ill children in low- and middle-income settings have a high risk of mortality both during and after hospitalisation despite guideline-based care. Understanding the biological mechanisms underpinning mortality may suggest optimal pathways to target for interventions to further reduce mortality. The Childhood Acute Illness and Nutrition (CHAIN) Network ( www.chainnnetwork.org) Nested Case-Cohort Study (CNCC) aims to investigate biological mechanisms leading to inpatient and post-discharge mortality through an integrated multi-omic approach. Methods and analysis; The CNCC comprises a subset of participants from the CHAIN cohort (1278/3101 hospitalised participants, including 350 children who died and 658 survivors, and 270/1140 well community children of similar age and household location) from nine sites in six countries across sub-Saharan Africa and South Asia. Systemic proteome, metabolome, lipidome, lipopolysaccharides, haemoglobin variants, toxins, pathogens, intestinal microbiome and biomarkers of enteropathy will be determined. Computational systems biology analysis will include machine learning and multivariate predictive modelling with stacked generalization approaches accounting for the different characteristics of each biological modality. This systems approach is anticipated to yield mechanistic insights, show interactions and behaviours of the components of biological entities, and help develop interventions to reduce mortality among acutely ill children. Ethics and dissemination. The CHAIN Network cohort and CNCC was approved by institutional review boards of all partner sites. Results will be published in open access, peer reviewed scientific journals and presented to academic and policy stakeholders. Data will be made publicly available, including uploading to recognised omics databases. Trial registration NCT03208725.

Development of a clinical prediction instrument to estimate risk of initial violent injury

BACKGROUND

Interpersonal violent injury is a public health crisis, disproportionately affecting young people of color. We aimed to evaluate associations between sociobehavioral predictors and first-time violent injury, and to develop a predictive risk score for violent injury.

METHODS

We performed a retrospective case-cohort study of adolescents aged 12-18 years. Multivariable logistic regression was used to estimate associations between 35 candidate variables and interpersonal first-time violent injury resulting in an emergency department (ED) visit. Multiple imputation was used to account for missing values and a risk score was developed by multiplying regression coefficients by 10 to generate a composite tool to predict initial violent injury (IVI). Discrimination and calibration were assessed using 10-fold cross validation.

RESULTS

19,210 adolescents were included, 276 (1.4%) as victims of IVI. The final model, the Initial Violent Injury Risk Prediction Tool (IVI-RPT), included: age, fight within the prior year, trouble with the law, and alcohol use. IVI-RPT scores were categorized as: 0-7 (low risk), 8-16 (moderate), and 17-26 (high), and IVI prevalence was 0.8% (95% confidence interval [CI]: 0.6%, 0.9%), 2.5% (95% CI: 1.9%, 3.1%), and 5.3% (95% CI: 4.1%, 6.6%), respectively. The area under the receiver operating characteristic curve was 0.70 (95% CI: 0.66, 0.73), while the slope of the calibration curve was 1.1 (95% CI: 0.9, 1.2).

CONCLUSIONS

We developed a promising clinical prediction instrument, the IVI-RPT, that categorizes individuals into risk groups with increasing probabilities of violent injury. External validation of this tool is required prior to clinical practice implementation.

The Childhood Acute Illness and Nutrition (CHAIN) network nested case-cohort study protocol: a multi-omics approach to understanding mortality among children in sub-Saharan Africa and South Asia

Introduction: Many acutely ill children in low- and middle-income settings have a high risk of mortality both during and after hospitalisation despite guideline-based care. Understanding the biological mechanisms underpinning mortality may suggest optimal pathways to target for interventions to further reduce mortality. The Childhood Acute Illness and Nutrition (CHAIN) Network ( www.chainnnetwork.org) Nested Case-Cohort Study (CNCC) aims to investigate biological mechanisms leading to inpatient and post-discharge mortality through an integrated multi-omic approach. Methods and analysis; The CNCC comprises a subset of participants from the CHAIN cohort (1278/3101 hospitalised participants, including 350 children who died and 658 survivors, and 270/1140 well community children of similar age and household location) from nine sites in six countries across sub-Saharan Africa and South Asia. Systemic proteome, metabolome, lipidome, lipopolysaccharides, haemoglobin variants, toxins, pathogens, intestinal microbiome and biomarkers of enteropathy will be determined. Computational systems biology analysis will include machine learning and multivariate predictive modelling with stacked generalization approaches accounting for the different characteristics of each biological modality. This systems approach is anticipated to yield mechanistic insights, show interactions and behaviours of the components of biological entities, and help develop interventions to reduce mortality among acutely ill children. Ethics and dissemination. The CHAIN Network cohort and CNCC was approved by institutional review boards of all partner sites. Results will be published in open access, peer reviewed scientific journals and presented to academic and policy stakeholders. Data will be made publicly available, including uploading to recognised omics databases. Trial registration NCT03208725.

The Case for Case-Cohort: An Applied Epidemiologist's Guide to Reframing Case-Cohort Studies to Improve Usability and Flexibility

When research questions require the use of precious samples, expensive assays or equipment, or labor-intensive data collection or analysis, nested case-control or case-cohort sampling of observational cohort study participants can often reduce costs. These study designs have similar statistical precision for addressing a singular research question, but case-cohort studies have broader efficiency and superior flexibility. Despite this, case-cohort designs are comparatively underutilized in the epidemiologic literature. Recent advances in statistical methods and software have made analyses of case-cohort data easier to implement, and advances from casual inference, such as inverse probability of sampling weights, have allowed the case-cohort design to be used with a variety of target parameters and populations. To provide an accessible link to this technical literature, we give a conceptual overview of case-cohort study analysis with inverse probability of sampling weights. We show how this general analytic approach can be leveraged to more efficiently study subgroups of interest or disease subtypes or to examine associations independent of case status. A brief discussion of how this framework could be extended to incorporate other related methodologic applications further demonstrates the broad cost-effectiveness and adaptability of case-cohort methods for a variety of modern epidemiologic applications in resource-limited settings.

Competing risks regression models with covariates-adjusted censoring weight under the generalized case-cohort design

A generalized case-cohort design has been used when measuring exposures is expensive and events are not rare in the full cohort. This design collects expensive exposure information from a (stratified) randomly selected subset from the full cohort, called the subcohort, and a fraction of cases outside the subcohort. For the full cohort study with competing risks, He et al. (Scand J Stat 43:103-122, 2016) studied the non-stratified proportional subdistribution hazards model with covariate-dependent censoring to directly evaluate covariate effects on the cumulative incidence function. In this paper, we propose a stratified proportional subdistribution hazards model with covariate-adjusted censoring weights for competing risks data under the generalized case-cohort design. We consider a general class of weight functions to account for the generalized case-cohort design. Then, we derive the optimal weight function which minimizes the asymptotic variance of parameter estimates within the general class of weight functions. The proposed estimator is shown to be consistent and asymptotically normally distributed. The simulation studies show (i) the proposed estimator with covariate-adjusted weight is unbiased when the censoring distribution depends on covariates; and (ii) the proposed estimator with the optimal weight function gains parameter estimation efficiency. We apply the proposed method to stem cell transplantation and diabetes data sets.

Case-cohort design in hematopoietic cell transplant studies

SERIES EDITORS- NOTE

Imagine you and your colleagues have done 1000 transplants in persons with acute myeloid leukaemia (AML) in 1st remission. 5 percent of the 20 percent of recipients relapsing posttransplant have an isolated central nervous system relapse. You are curious and want to know whether there is anything special about this 5 percent, specifically whether this risk corelates with any pretransplant clinical and laboratory co-variates. You have extensive clinical data and some typical laboratory data on all 1000 but you suspect the culprit is mutation topography. What to do? Fortunately you have bio-banked DNA from the 1000. If resources and monies are not limiting you can do targeted or next generation sequencing on all 1000 DNA samples and off you go. However, most of us lack unlimited resources and monies. How can you sensibly and efficiently tackle this research problem? The answer is a case-cohort design study. In the typescript which follows Profs. Cai and Kim explain how to accomplish this. If you follow their advice you may need only to analyze samples from < 300 recipients rather than 1000 to test your hypothesis. They explain how to design such a study and provide references to estimate sample size.Sadly, their typescript will not tell you how to get funding for the study, whish poor devil who will have to write the protocol, worse, who will shepherd it though endless committees for approval and the like. Help on these issues is outside the scope of our statistics series. In this context we suggest advice from Woody Allen's article in the New Yorker: The Kugelmass Episode (April 24, 1977). When Prof. Kugelmass (English, City College) tells his analyst Dr. Mandel he has fallen in love with Emma Bovary who died of arsenic poisoning near Rouen, France 120 years earlier the analyst says: After all, I'm an analyst, not a magician. Kugelmass' reply: Then perhaps what I need is a magician and is off to Coney Island to find one. Good luck, the magician may still be there! (Note: This typescript is R-rated. It contains an equation.)Robert Peter Gale, Imperial College London, and Mei-Jie Zhang, Medical College of Wisconsin and CIBMTR.

Urinary concentrations of phenols, parabens, and triclocarban in relation to uterine leiomyomata incidence and growth

OBJECTIVE

To examine the association of urinary concentrations of phenols, parabens, and triclocarban with incidence and growth of uterine leiomyomata (UL; fibroids).

DESIGN

Case-cohort study, nested within the Study of Environment, Lifestyle, and Fibroids, a prospective cohort study.

SETTING

Clinic visits at baseline and every 20 months for 60 months.

PATIENT(S)

754 Black women aged 23-35 years residing in the Detroit, Michigan area (enrolled during 2010-2012).

INTERVENTION

None.

MAIN OUTCOME MEASURE(S)

At each study visit, women underwent transvaginal ultrasound for measurement of UL incidence and growth and provided urine specimens in which we quantified concentrations of seven phenols, four parabens, and triclocarban. We used Cox proportional hazards regression to estimate hazard ratios and 95% confidence intervals (CIs) characterizing the relation of urinary biomarker concentrations with UL incidence during the 60 months of follow-up. In a subset of UL detected and measured at multiple time points, we used linear regression to assess the associations between biomarker concentrations and UL growth.

RESULT(S)

Urinary biomarker concentrations were generally inversely associated with UL incidence, but the associations were weak and nonmonotonic. For example, hazard ratios comparing concentrations ≥90th with <50th percentile were 0.77 (95% CI: 0.46, 1.27) for bisphenol A, 0.72 (95% CI: 0.40, 1.28) for bisphenol S, and 0.76 (95% CI: 0.43, 1.33) for methylparaben. Biomarker concentrations were not strongly associated with UL growth.

CONCLUSION(S)

In this study of reproductive-aged Black women, urinary phenols, parabens, and triclocarban biomarkers were neither strongly nor consistently associated with UL incidence and growth.

Sampling strategies to evaluate the prognostic value of a new biomarker on a time-to-event end-point

BACKGROUND

The availability of large epidemiological or clinical data storing biological samples allow to study the prognostic value of novel biomarkers, but efficient designs are needed to select a subsample on which to measure them, for parsimony and economical reasons. Two-phase stratified sampling is a flexible approach to perform such sub-sampling, but literature on stratification variables to be used in the sampling and power evaluation is lacking especially for survival data.

METHODS

We compared the performance of different sampling designs to assess the prognostic value of a new biomarker on a time-to-event endpoint, applying a Cox model weighted by the inverse of the empirical inclusion probability.

RESULTS

Our simulation results suggest that case-control stratified (or post stratified) by a surrogate variable of the marker can yield higher performances than simple random, probability proportional to size, and case-control sampling. In the presence of high censoring rate, results showed an advantage of nested case-control and counter-matching designs in term of design effect, although the use of a fixed ratio between cases and controls might be disadvantageous. On real data on childhood acute lymphoblastic leukemia, we found that optimal sampling using pilot data is greatly efficient.

CONCLUSIONS

Our study suggests that, in our sample, case-control stratified by surrogate and nested case-control yield estimates and power comparable to estimates obtained in the full cohort while strongly decreasing the number of patients required. We recommend to plan the sample size and using sampling designs for exploration of novel biomarker in clinical cohort data.

Oral Leukoplakia and Risk of Progression to Oral Cancer: A Population-Based Cohort Study

BACKGROUND

The optimal clinical management of oral precancer remains uncertain. We investigated the natural history of oral leukoplakia, the most common oral precancerous lesion, to estimate the relative and absolute risks of progression to cancer, the predictive accuracy of a clinician's decision to biopsy a leukoplakia vis-à-vis progression, and histopathologic predictors of progression.

METHODS

We conducted a retrospective cohort study (1996-2012) of patients with oral leukoplakia (n = 4886), identified using electronic medical records within Kaiser Permanente Northern California. Among patients with leukoplakia who received a biopsy (n = 1888), we conducted a case-cohort study to investigate histopathologic predictors of progression. Analyses included indirect standardization and unweighted or weighted Cox regression.

RESULTS

Compared with the overall Kaiser Permanente Northern California population, oral cancer incidence was substantially elevated in oral leukoplakia patients (standardized incidence ratio = 40.8, 95% confidence interval [CI] = 34.8 to 47.6; n = 161 cancers over 22 582 person-years). Biopsied leukoplakias had a higher oral cancer risk compared with those that were not biopsied (adjusted hazard ratio = 2.38, 95% CI = 1.73 to 3.28). However, to identify a prevalent or incident oral cancer, the biopsy decision had low sensitivity (59.6%), low specificity (62.1%), and moderate positive-predictive value (5.1%). Risk of progression to oral cancer statistically significantly increased with the grade of dysplasia; 5-year competing risk-adjusted absolute risks were: leukoplakia overall = 3.3%, 95% CI = 2.7% to 3.9%; no dysplasia = 2.2%, 95% CI = 1.5% to 3.1%; mild-dysplasia = 11.9%, 95% CI = 7.1% to 18.1%; moderate-dysplasia = 8.7%, 95% CI = 3.2% to 17.9%; and severe dysplasia = 32.2%, 95% CI = 8.1%-60.0%. Yet 39.6% of cancers arose from biopsied leukoplakias without dysplasia.

CONCLUSIONS

The modest accuracy of the decision to biopsy a leukoplakia vis-à-vis presence or eventual development of oral cancer highlights the need for routine biopsy of all leukoplakias regardless of visual or clinical impression. Leukoplakia patients, particularly those with dysplasia, need to be closely monitored for signs of early cancer.

Group testing can improve the cost-efficiency of prospective-retrospective biomarker studies

BACKGROUND

Cancer treatment is increasingly dependent on biomarkers for prognostication and treatment selection. Potential biomarkers are frequently evaluated in prospective-retrospective studies in which biomarkers are measured retrospectively on archived specimens after completion of prospective clinical trials. In light of the high costs of some assays, random sampling designs have been proposed that measure biomarkers for a random sub-sample of subjects selected on the basis of observed outcome and possibly other variables. Compared with a standard design that measures biomarkers on all subjects, a random sampling design can be cost-efficient in the sense of reducing the cost of the study substantially while achieving a reasonable level of precision.

METHODS

For a biomarker that indicates the presence of some molecular alteration (e.g., mutation in a gene), we explore the use of a group testing strategy, which involves physically pooling specimens across subjects and assaying pooled samples for the presence of the molecular alteration of interest, for further improvement in cost-efficiency beyond random sampling. We propose simple and general approaches to estimating the prognostic and predictive values of biomarkers with group testing, and conduct simulation studies to validate the proposed estimation procedures and to assess the cost-efficiency of the group testing design in comparison to the standard and random sampling designs.

RESULTS

Simulation results show that the proposed estimation procedures perform well in realistic settings and that a group testing design can have considerably higher cost-efficiency than a random sampling design.

CONCLUSIONS

Group testing can be used to improve the cost-efficiency of biomarker studies.

Assessing risk model calibration with missing covariates

When validating a risk model in an independent cohort, some predictors may be missing for some subjects. Missingness can be unplanned or by design, as in case-cohort or nested case-control studies, in which some covariates are measured only in subsampled subjects. Weighting methods and imputation are used to handle missing data. We propose methods to increase the efficiency of weighting to assess calibration of a risk model (i.e. bias in model predictions), which is quantified by the ratio of the number of observed events, $\mathcal{O}$, to expected events, $\mathcal{E}$, computed from the model. We adjust known inverse probability weights by incorporating auxiliary information available for all cohort members. We use survey calibration that requires the weighted sum of the auxiliary statistics in the complete data subset to equal their sum in the full cohort. We show that a pseudo-risk estimate that approximates the actual risk value but uses only variables available for the entire cohort is an excellent auxiliary statistic to estimate $\mathcal{E}$. We derive analytic variance formulas for $\mathcal{O}/\mathcal{E}$ with adjusted weights. In simulations, weight adjustment with pseudo-risk was much more efficient than inverse probability weighting and yielded consistent estimates even when the pseudo-risk was a poor approximation. Multiple imputation was often efficient but yielded biased estimates when the imputation model was misspecified. Using these methods, we assessed calibration of an absolute risk model for second primary thyroid cancer in an independent cohort.

Brominated flame retardants and organochlorine pesticides and incidence of uterine leiomyomata: A prospective ultrasound study

Uterine leiomyomata (UL) are hormone-responsive benign neoplasms. Brominated flame retardants and organochlorine pesticides (OCPs) can disrupt hormones involved in UL etiology.

METHODS

The Study of Environmental, Lifestyle, and Fibroids is a Detroit-area prospective cohort of 1693 Black women 23-35 years of age. At baseline and approximately every 20 months for 5 years, women completed questionnaires and underwent transvaginal ultrasounds. Using a case-cohort study design, we selected 729 UL-free participants at baseline and analyzed baseline plasma samples for polybrominated diphenyl ethers (PBDEs), a polybrominated biphenyl ether (PBB-153), and OCPs. We used Cox proportional hazard models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs).

RESULTS

Compared with total PBDE plasma concentrations <50th percentile, adjusted HRs for the 50th-74th, 75th-89th, and ≥90th percentiles were 1.00 (95% CI = 0.68, 1.47), 1.04 (95% CI = 0.63, 1.68), and 0.85 (95% CI = 0.48, 1.50), respectively. HRs for PBB-153 plasma concentrations were generally similar to total PBDE plasma concentrations. Compared with total OCP plasma concentrations <50th percentile, HRs for the 50th-74th, 75th-89th, and ≥90th percentiles were 0.86 (95% CI = 0.57, 1.29), 0.73 (95% CI = 0.43, 1.22), and 0.58 (95% CI = 0.32, 1.04), respectively. HRs for individual PBDEs and OCPs were similar to their respective totals.

CONCLUSION

We found little support for an association between brominated flame retardant plasma concentrations and UL incidence, and some evidence of lower UL incidence with the highest OCP plasma concentrations.

Small-sample inference for cluster-based outcome-dependent sampling schemes in resource-limited settings: Investigating low birthweight in Rwanda

The neonatal mortality rate in Rwanda remains above the United Nations Sustainable Development Goal 3 target of 12 deaths per 1000 live births. As part of a larger effort to reduce preventable neonatal deaths in the country, we conducted a study to examine risk factors for low birthweight. The data were collected via a cost-efficient cluster-based outcome-dependent sampling (ODS) scheme wherein clusters of individuals (health centers) were selected on the basis of, in part, the outcome rate of the individuals. For a given data set collected via a cluster-based ODS scheme, estimation for a marginal model may proceed via inverse-probability-weighted generalized estimating equations, where the cluster-specific weights are the inverse probability of the health center's inclusion in the sample. In this paper, we provide a detailed treatment of the asymptotic properties of this estimator, together with an explicit expression for the asymptotic variance and a corresponding estimator. Furthermore, motivated by the study we conducted in Rwanda, we propose a number of small-sample bias corrections to both the point estimates and the standard error estimates. Through simulation, we show that applying these corrections when the number of clusters is small generally reduces the bias in the point estimates, and results in closer to nominal coverage. The proposed methods are applied to data from 18 health centers and 1 district hospital in Rwanda.

Hospitalization With Major Infection and Incidence of End-Stage Renal Disease: The Atherosclerosis Risk in Communities (ARIC) Study

OBJECTIVE

To evaluate whether the incidence of infectious diseases increases the long-term risk for incident end-stage renal disease (ESRD) in the general population.

PATIENTS AND METHODS

In 10,290 participants of the Atherosclerosis Risk in Communities Study who attended visit 4 (1996-1998), we evaluated the association of incident hospitalization with major infections (pneumonia, urinary tract infection, bloodstream infection, and cellulitis and osteomyelitis) with subsequent risk for ESRD through September 30, 2015. Hospitalization with major infection was entered into multivariable Cox models as a time-varying exposure to estimate the hazard ratios.

RESULTS

Mean age was 63 years, and of 10,290 individuals, 56% (n=5781) were women, 22% (n=2252) were black, and 7% (n=666) had an estimated glomerular filtration rate less than 60 mL/min/1.73 m2. During a median follow-up of 17.4 years, there were 2642 incident hospitalizations with major infection and 281 cases of ESRD (132 cases after hospitalization with major infection). The risk for ESRD was higher following major infection compared with while free of major infection (crude incidence rate, 10.9 vs 1.0 per 1000 person-years). In multivariable time-varying Cox analysis, hospitalization with major infection was associated with a 3.3-fold increased risk for ESRD (hazard ratio, 3.34; 95% CI, 2.56-4.37). The association was similar across pneumonia, urinary tract infection, bloodstream infection, and cellulitis and osteomyelitis, and remained significant across subgroups of age, sex, race, diabetes, history of cardiovascular disease, and chronic kidney disease.

CONCLUSION

Hospitalization with major infection was independently and robustly associated with subsequent risk for ESRD. Whether preventive approaches against infection have beneficial effects on kidney outcomes may deserve future investigations.

Incident Hospitalization with Major Cardiovascular Diseases and Subsequent Risk of ESKD: Implications for Cardiorenal Syndrome

BACKGROUND

Cardiorenal syndrome is a well known concept, bolstered by extensive investigations of CKD as a risk factor of cardiovascular disease. However, data on whether cardiovascular disease increases long-term risk of ESKD are sparse.

METHODS

We assessed the association of incident hospitalization with major cardiovascular diseases (heart failure, atrial fibrillation, coronary heart disease, and stroke) with subsequent risk of ESKD among individuals enrolled in the Atherosclerosis Risk in Communities study; the analysis included 9047 individuals without prevalent cardiovascular disease at their fourth study visit. Each relevant incident cardiovascular disease event was entered into multivariable Cox proportional hazard models as a time-varying exposure to estimate hazard ratios.

RESULTS

During a median follow-up of 17.5 years, there were 2598 cases of hospitalization with cardiovascular disease (heart failure, n=1269; atrial fibrillation, n=1337; coronary heart disease, n=696; and stroke, n=559) and 210 cases of incident ESKD. The incidence of major cardiovascular disease was associated with increased risk of ESKD, with the highest risk for heart failure (hazard ratio, 11.40; 95% confidence interval, 8.38 to 15.50), followed by coronary heart disease, atrial fibrillation, and stroke. When we analyzed heart failure with preserved ejection fraction and heart failure with reduced ejection fraction separately, the risk was nominally higher for heart failure with preserved ejection fraction.

CONCLUSIONS

Major incident cardiovascular disease events were associated with ESKD, independent of kidney risk factors. In particular, heart failure showed a very strong association with ESKD. Our findings highlight the importance of monitoring and managing kidney disease in patients with cardiovascular disease. The potentially distinct contribution to ESKD of heart failure with preserved versus reduced ejection fraction deserves future investigation.

Lipid metabolic networks, Mediterranean diet and cardiovascular disease in the PREDIMED trial

Background

Perturbed lipid metabolic pathways may play important roles in the development of cardiovascular disease (CVD). However, existing epidemiological studies have focused more on discovering individual lipid metabolites for CVD risk prediction rather than assessing metabolic pathways.

Methods

This study included a subcohort of 787 participants and all 230 incident CVD cases from the PREDIMED trial. Applying a network-based analytical method, we identified lipid subnetworks and clusters from a global network of 200 lipid metabolites and linked these subnetworks/clusters to CVD risk.

Results

Lipid metabolites with more double bonds clustered within one subnetwork, whereas lipid metabolites with fewer double bonds clustered within other subnetworks. We identified 10 lipid clusters that were divergently associated with CVD risk. The hazard ratios [HRs, 95% confidence interval (CI)] of CVD per a 1-standard deviation (SD) increment in cluster score were 1.39 (1.17-1.66) for the hydroxylated phosphatidylcholine (HPC) cluster and 1.24 (1.11-1.37) for a cluster that included diglycerides and a monoglyceride with stearic acyl chain. Every 1-SD increase in the score of cluster that included highly unsaturated phospholipids and cholesterol esters was associated with an HR for CVD of 0.81 (95% CI, 0.67-0.98). Despite a suggestion that MedDiet modified the association between a subnetwork that included most lipids with a high degree of unsaturation and CVD, changes in lipid subnetworks/clusters during the first-year follow-up were not significantly different between intervention groups.

Conclusions

The degree of unsaturation was a major determinant of the architecture of lipid metabolic network. Lipid clusters that strongly predicted CVD risk, such as the HPC cluster, warrant further functional investigations.

Opportunities and Challenges for Environmental Exposure Assessment in Population-Based Studies

A growing number and increasing diversity of factors are available for epidemiological studies. These measures provide new avenues for discovery and prevention, yet they also raise many challenges for adoption in epidemiological investigations. Here, we evaluate 1) designs to investigate diseases that consider heterogeneous and multidimensional indicators of exposure and behavior, 2) the implementation of numerous methods to capture indicators of exposure, and 3) the analytical methods required for discovery and validation. We find that case-control studies have provided insights into genetic susceptibility but are insufficient for characterizing complex effects of environmental factors on disease development. Prospective and two-phase designs are required but must balance extended data collection with follow-up of study participants. We discuss innovations in assessments including the microbiome; mass spectrometry and metabolomics; behavioral assessment; dietary, physical activity, and occupational exposure assessment; air pollution monitoring; and global positioning and individual sensors. We claim the the availability of extensive correlated data raises new challenges in disentangling specific exposures that influence cancer risk from among extensive and often correlated exposures. In conclusion, new high-dimensional exposure assessments offer many new opportunities for environmental assessment in cancer development. Cancer Epidemiol Biomarkers Prev; 26(9); 1370-80. ©2017 AACR.

Plasma Ceramides, Mediterranean Diet, and Incident Cardiovascular Disease in the PREDIMED Trial (Prevención con Dieta Mediterránea)

BACKGROUND

Although in vitro studies and investigations in animal models and small clinical populations have suggested that ceramides may represent an intermediate link between overnutrition and certain pathological mechanisms underlying cardiovascular disease (CVD), no prospective studies have investigated the association between plasma ceramides and risk of CVD.

METHODS

The study population consisted of 980 participants from the PREDIMED trial (Prevención con Dieta Mediterránea), including 230 incident cases of CVD and 787 randomly selected participants at baseline (including 37 overlapping cases) followed for ≤7.4 years. Participants were randomized to a Mediterranean diet supplemented with extra virgin olive oil, a Mediterranean diet supplemented with nuts, or a control diet. Plasma ceramide concentrations were measured on a liquid chromatography tandem mass spectrometry metabolomics platform. The primary outcome was a composite of nonfatal acute myocardial infarction, nonfatal stroke, or cardiovascular death. Hazard ratios were estimated with weighted Cox regression models using Barlow weights to account for the case-cohort design.

RESULTS

The multivariable hazard ratios (HR) and 95% confidence intervals (CIs) comparing the extreme quartiles of plasma concentrations of C16:0, C22:0, C24:0, and C24:1 ceramides were 2.39 (1.49-3.83, Ptrend<0.001), 1.91 (1.21-3.01, Ptrend=0.003), 1.97 (1.21-3.20, Ptrend=0.004), and 1.73 (1.09-2.74, Ptrend=0.011), respectively. The ceramide score, calculated as a weighted sum of concentrations of four ceramides, was associated with a 2.18-fold higher risk of CVD across extreme quartiles (HR, 2.18; 95% CI, 1.36-3.49; Ptrend<0.001). The association between baseline ceramide score and incident CVD varied significantly by treatment groups (Pinteraction=0.010). Participants with a higher ceramide score and assigned to either of the 2 active intervention arms of the trial showed similar CVD risk to those with a lower ceramide score, whereas participants with a higher ceramide score and assigned to the control arm presented significantly higher CVD risk. Changes in ceramide concentration were not significantly different between Mediterranean diet and control groups during the first year of follow-up.

CONCLUSIONS

Our study documented a novel positive association between baseline plasma ceramide concentrations and incident CVD. In addition, a Mediterranean dietary intervention may mitigate potential deleterious effects of elevated plasma ceramide concentrations on CVD.

CLINICAL TRIAL REGISTRATION

URL: http://www.isrctn.com. Unique identifier: ISRCTN35739639.

A case-cohort approach for multi-state models in hospital epidemiology

Analysing the determinants and consequences of hospital-acquired infections involves the evaluation of large cohorts. Infected patients in the cohort are often rare for specific pathogens, because most of the patients admitted to the hospital are discharged or die without such an infection. Death and discharge are competing events to acquiring an infection, because these individuals are no longer at risk of getting a hospital-acquired infection. Therefore, the data is best analysed with an extended survival model - the extended illness-death model. A common problem in cohort studies is the costly collection of covariate values. In order to provide efficient use of data from infected as well as uninfected patients, we propose a tailored case-cohort approach for the extended illness-death model. The basic idea of the case-cohort design is to only use a random sample of the full cohort, referred to as subcohort, and all cases, namely the infected patients. Thus, covariate values are only obtained for a small part of the full cohort. The method is based on existing and established methods and is used to perform regression analysis in adapted Cox proportional hazards models. We propose estimation of all cause-specific cumulative hazards and transition probabilities in an extended illness-death model based on case-cohort sampling. As an example, we apply the methodology to infection with a specific pathogen using a large cohort from Spanish hospital data. The obtained results of the case-cohort design are compared with the results in the full cohort to investigate the performance of the proposed method. Copyright © 2016 John Wiley & Sons, Ltd.

Feasibility of reusing time-matched controls in an overlapping cohort

The methods developed for secondary analysis of nested case-control data have been illustrated only in simplified settings in a common cohort and have not found their way into biostatistical practice. This paper demonstrates the feasibility of reusing prior nested case-control data in a realistic setting where a new outcome is available in an overlapping cohort where no new controls were gathered and where all data have been anonymised. Using basic information about the background cohort and sampling criteria, the new cases and prior data are "aligned" to identify the common underlying study base. With this study base, a Kaplan-Meier table of the prior outcome extracts the risk sets required to calculate the weights to assign to the controls to remove the sampling bias. A weighted Cox regression, implemented in standard statistical software, provides unbiased hazard ratios. Using the method to compare cases of contralateral breast cancer to available controls from a prior study of metastases, we identified a multifocal tumor as a risk factor that has not been reported previously. We examine the sensitivity of the method to an imperfect weighting scheme and discuss its merits and pitfalls to provide guidance for its use in medical research studies.

Efficient estimation for accelerated failure time model under case-cohort and nested case-control sampling

Case-cohort (Prentice, 1986) and nested case-control (Thomas, 1977) designs have been widely used as a cost-effective alternative to the full-cohort design. In this article, we propose an efficient likelihood-based estimation method for the accelerated failure time model under case-cohort and nested case-control designs. An EM algorithm is developed to maximize the likelihood function and a kernel smoothing technique is adopted to facilitate the estimation in the M-step of the EM algorithm. We show that the proposed estimators for the regression coefficients are consistent and asymptotically normal. The asymptotic variance of the estimators can be consistently estimated using an EM-aided numerical differentiation method. Simulation studies are conducted to evaluate the finite-sample performance of the estimators and an application to a Wilms tumor data set is also given to illustrate the methodology.

A full competing risk analysis of hospital-acquired infections can easily be performed by a case-cohort approach

OBJECTIVES

We provide a case-cohort approach and show that a full competing risk analysis is feasible even in a reduced data set. Competing events for hospital-acquired infections are death or discharge from the hospital because they preclude the observation of such infections.

STUDY DESIGN AND SETTING

Using surveillance data of 6,568 patient admissions (full cohort) from two Spanish intensive care units, we propose a case-cohort approach which uses only data from a random sample of the full cohort and all infected patients (the cases). We combine established methodology to study following measures: event-specific as well as subdistribution hazard ratios for all three events (infection, death, and discharge), cumulative hazards as well as incidence functions by risk factor, and also for all three events.

RESULTS

Compared with the values from the full cohort, all measures are well approximated with the case-cohort design. For the event of interest (infection), event-specific and subdistribution hazards can be estimated with the full efficiency of the case-cohort design. So, standard errors are only slightly increased, whereas the precision of estimated hazards of the competing events is inflated according to the size of the subcohort.

CONCLUSION

The case-cohort design provides an appropriate sampling design for studying hospital-acquired infections in a reduced data set. Potential effects of risk factors on the competing events (death and discharge) can be evaluated.

Using the whole cohort in the analysis of countermatched samples

We present a technique for using calibrated weights to incorporate whole-cohort information in the analysis of a countermatched sample. Following Samuelsen's approach for matched case-control sampling, we derive expressions for the marginal sampling probabilities, so that the data can be treated as an unequally-sampled case-cohort design. Pseudolikelihood estimating equations are used to find the estimates. The sampling weights can be calibrated, allowing all whole-cohort variables to be used in estimation; in contrast, the partial likelihood analysis makes use only of a single discrete surrogate for exposure. Using a survey-sampling approach rather than a martingale approach simplifies the theory; in particular, the sampling weights need not be a predictable process. Our simulation results show that pseudolikelihood estimation gives lower efficiency than partial likelihood estimation, but that the gain from calibration of weights can more than compensate for this loss. If there is a good surrogate for exposure, countermatched sampling still outperforms case-cohort and two-phase case-control sampling even when calibrated weights are used. Findings are illustrated with data from the National Wilms' Tumour Study and the Welsh nickel refinery workers study.

Augmented case-only designs for randomized clinical trials with failure time endpoints

Under suitable assumptions and by exploiting the independence between inherited genetic susceptibility and treatment assignment, the case-only design yields efficient estimates for subgroup treatment effects and gene-treatment interaction in a Cox model. However it cannot provide estimates of the genetic main effect and baseline hazards, that are necessary to compute the absolute disease risk. For two-arm, placebo-controlled trials with rare failure time endpoints, we consider augmenting the case-only design with random samples of controls from both arms, as in the classical case-cohort sampling scheme, or with a random sample of controls from the active treatment arm only. The latter design is motivated by vaccine trials for cost-effective use of resources and specimens so that host genetics and vaccine-induced immune responses can be studied simultaneously in a bigger set of participants. We show that these designs can identify all parameters in a Cox model and that the efficient case-only estimator can be incorporated in a two-step plug-in procedure. Results in simulations and a data example suggest that incorporating case-only estimators in the classical case-cohort design improves the precision of all estimated parameters; sampling controls only in the active treatment arm attains a similar level of efficiency.

Analysis of Clinical Cohort Data Using Nested Case-control and Case-cohort Sampling Designs. A Powerful and Economical Tool

BACKGROUND

Sampling from a large cohort in order to derive a subsample that would be sufficient for statistical analysis is a frequently used method for handling large data sets in epidemiological studies with limited resources for exposure measurement. For clinical studies however, when interest is in the influence of a potential risk factor, cohort studies are often the first choice with all individuals entering the analysis.

OBJECTIVES

Our aim is to close the gap between epidemiological and clinical studies with respect to design and power considerations. Schoenfeld's formula for the number of events required for a Cox' proportional hazards model is fundamental. Our objective is to compare the power of analyzing the full cohort and the power of a nested case-control and a case-cohort design.

METHODS

We compare formulas for power for sampling designs and cohort studies. In our data example we simultaneously apply a nested case-control design with a varying number of controls matched to each case, a case cohort design with varying subcohort size, a random subsample and a full cohort analysis. For each design we calculate the standard error for estimated regression coefficients and the mean number of distinct persons, for whom covariate information is required.

RESULTS

The formula for the power of a nested case-control design and the power of a case-cohort design is directly connected to the power of a cohort study using the well known Schoenfeld formula. The loss in precision of parameter estimates is relatively small compared to the saving in resources.

CONCLUSIONS

Nested case-control and case-cohort studies, but not random subsamples yield an attractive alternative for analyzing clinical studies in the situation of a low event rate. Power calculations can be conducted straightforwardly to quantify the loss of power compared to the savings in the num-ber of patients using a sampling design instead of analyzing the full cohort.

A new comparison of nested case-control and case-cohort designs and methods

Existing literature comparing statistical properties of nested case-control and case-cohort methods have become insufficient for present day epidemiologists. The literature has not reconciled conflicting conclusions about the standard methods. Moreover, a comparison including newly developed methods, such as inverse probability weighting methods, is needed. Two analytical methods for nested case-control studies and six methods for case-cohort studies using proportional hazards regression model were summarized and their statistical properties were compared. The answer to which design and method is more powerful was more nuanced than what was previously reported. For both nested case-control and case-cohort designs, inverse probability weighting methods were more powerful than the standard methods. However, the difference became negligible when the proportion of failure events was very low (<1%) in the full cohort. The comparison between two designs depended on the censoring types and incidence proportion: with random censoring, nested case-control designs coupled with the inverse probability weighting method yielded the highest statistical power among all methods for both designs. With fixed censoring times, there was little difference in efficiency between two designs when inverse probability weighting methods were used; however, the standard case-cohort methods were more powerful than the conditional logistic method for nested case-control designs. As the proportion of failure events in the full cohort became smaller (<10%), nested case-control methods outperformed all case-cohort methods and the choice of analytic methods within each design became less important. When the predictor of interest was binary, the standard case-cohort methods were often more powerful than the conditional logistic method for nested case-control designs.

Biomarker discovery study design for type 1 diabetes in The Environmental Determinants of Diabetes in the Young (TEDDY) study

AIMS

The Environmental Determinants of Diabetes in the Young planned biomarker discovery studies on longitudinal samples for persistent confirmed islet cell autoantibodies and type 1 diabetes using dietary biomarkers, metabolomics, microbiome/viral metagenomics and gene expression.

METHODS

This article describes the details of planning The Environmental Determinants of Diabetes in the Young biomarker discovery studies using a nested case-control design that was chosen as an alternative to the full cohort analysis. In the frame of a nested case-control design, it guides the choice of matching factors, selection of controls, preparation of external quality control samples and reduction of batch effects along with proper sample allocation.

RESULTS AND CONCLUSION

Our design is to reduce potential bias and retain study power while reducing the costs by limiting the numbers of samples requiring laboratory analyses. It also covers two primary end points (the occurrence of diabetes-related autoantibodies and the diagnosis of type 1 diabetes). The resulting list of case-control matched samples for each laboratory was augmented with external quality control samples.

Case-control studies in neurosurgery

OBJECT

Observational studies, such as cohort and case-control studies, are valuable instruments in evidence-based medicine. Case-control studies, in particular, are becoming increasingly popular in the neurosurgical literature due to their low cost and relative ease of execution; however, no one has yet systematically assessed these types of studies for quality in methodology and reporting.

METHODS

The authors performed a literature search using PubMed/MEDLINE to identify all studies that explicitly identified themselves as "case-control" and were published in the JNS Publishing Group journals (Journal of Neurosurgery, Journal of Neurosurgery: Pediatrics, Journal of Neurosurgery: Spine, and Neurosurgical Focus) or Neurosurgery. Each paper was evaluated for 22 descriptive variables and then categorized as having either met or missed the basic definition of a case-control study. All studies that evaluated risk factors for a well-defined outcome were considered true case-control studies. The authors sought to identify key features or phrases that were or were not predictive of a true case-control study. Those papers that satisfied the definition were further evaluated using the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist.

RESULTS

The search detected 67 papers that met the inclusion criteria, of which 32 (48%) represented true case-control studies. The frequency of true case-control studies has not changed with time. Use of odds ratios (ORs) and logistic regression (LR) analysis were strong positive predictors of true case-control studies (for odds ratios, OR 15.33 and 95% CI 4.52-51.97; for logistic regression analysis, OR 8.77 and 95% CI 2.69-28.56). Conversely, negative predictors included focus on a procedure/intervention (OR 0.35, 95% CI 0.13-0.998) and use of the word "outcome" in the Results section (OR 0.23, 95% CI 0.082-0.65). After exclusion of nested case-control studies, the negative correlation between focus on a procedure/intervention and true case-control studies was strengthened (OR 0.053, 95% CI 0.0064-0.44). There was a trend toward a negative association between the use of survival analysis or Kaplan-Meier curves and true case-control studies (OR 0.13, 95% CI 0.015-1.12). True case-control studies were no more likely than their counterparts to use a potential study design "expert" (OR 1.50, 95% CI 0.57-3.95). The overall average STROBE score was 72% (range 50-86%). Examples of reporting deficiencies were reporting of bias (28%), missing data (55%), and funding (44%).

CONCLUSIONS

The results of this analysis show that the majority of studies in the neurosurgical literature that identify themselves as "case-control" studies are, in fact, labeled incorrectly. Positive and negative predictors were identified. The authors provide several recommendations that may reverse the incorrect and inappropriate use of the term "case-control" and improve the quality of design and reporting of true case-control studies in neurosurgery.

Analysis of secondary outcomes in nested case-control study designs

One of the main perceived advantages of using a case-cohort design compared with a nested case-control design in an epidemiologic study is the ability to evaluate with the same subcohort outcomes other than the primary outcome of interest. In this paper, we show that valid inferences about secondary outcomes can also be achieved in nested case-control studies by using the inclusion probability weighting method in combination with an approximate jackknife standard error that can be computed using existing software. Simulation studies demonstrate that when the sample size is sufficient, this approach yields valid type 1 error and coverage rates for the analysis of secondary outcomes in nested case-control designs. Interestingly, the statistical power of the nested case-control design was comparable with that of the case-cohort design when the primary and secondary outcomes were positively correlated. The proposed method is illustrated with the data from a cohort in Cardiovascular Health Study to study the association of C-reactive protein levels and the incidence of congestive heart failure.

A prospective stratified case-cohort study on statins and multiple adverse events in Japan

Purpose

To assess the association between statins and diverse adverse events in Japanese population.

Methods

New users of statin who started statin after 6-month period of non-use were identified in 68 hospitals between January 2008 and July 2010. In addition to the random sample subcohort, we selected additional subcohort members to make the stratified sample subcohort have at least one patient in all subgroups stratified by each combination of statin and hospital. By abstraction from medical records, detailed information was obtained for all potential cases and pre-selected subcohort members. The event review committee consisting of 3 specialists judged whether possible cases met the definition of one of the adverse events of interest, and for adjudicated cases the committee further judged whether statin was a certain, probable or possible cause of the occurrence of the event. Adjusted for covariates including age, gender, status of “switcher”, use of high daily dose and comorbidities at baseline, hazard ratio (HR) was estimated by the Cox proportional hazards model with Barlow’s weighting method. Data were also analyzed by the method proposed by Breslow in 2009.

Results

A total of 6,877 new users of a statin were identified (median age: 66 years; males: 52%). The hazard ratios of increase in serum creatinine for atorvastatin and fluvastatin have wide confidence intervals, but both of the point estimates were around 2.5. Estimates of hazard ratios by the method of Barlow (1999) were similar to those by the method of Breslow (2009).

Conclusions

Use of statin was not associated with a significant increased risk for renal, liver and muscle events. However, the hazard ratio of increase in serum creatinine tended to be high with atorvastatin and fluvastatin to require further studies.

More efficient estimators for case-cohort studies

The case-cohort study design, used to reduce costs in large cohort studies, is a random sample of the entire cohort, named the subcohort, augmented with subjects having the disease of interest but not in the subcohort sample. When several diseases are of interest, several case-cohort studies may be conducted using the same subcohort, with each disease analyzed separately, ignoring the additional exposure measurements collected on subjects with the other diseases. This is not an efficient use of the data, and in this paper, we propose more efficient estimators. We consider both joint and separate analyses for the multiple diseases. We propose an estimating equation approach with a new weight function, and we establish the consistency and asymptotic normality of the resulting estimator. Simulation studies show that the proposed methods using all available information gain efficiency. We apply our proposed method to the data from the Busselton Health Study.

Using full-cohort data in nested case-control and case-cohort studies by multiple imputation

In many large prospective cohorts, expensive exposure measurements cannot be obtained for all individuals. Exposure-disease association studies are therefore often based on nested case-control or case-cohort studies in which complete information is obtained only for sampled individuals. However, in the full cohort, there may be a large amount of information on cheaply available covariates and possibly a surrogate of the main exposure(s), which typically goes unused. We view the nested case-control or case-cohort study plus the remainder of the cohort as a full-cohort study with missing data. Hence, we propose using multiple imputation (MI) to utilise information in the full cohort when data from the sub-studies are analysed. We use the fully observed data to fit the imputation models. We consider using approximate imputation models and also using rejection sampling to draw imputed values from the true distribution of the missing values given the observed data. Simulation studies show that using MI to utilise full-cohort information in the analysis of nested case-control and case-cohort studies can result in important gains in efficiency, particularly when a surrogate of the main exposure is available in the full cohort. In simulations, this method outperforms counter-matching in nested case-control studies and a weighted analysis for case-cohort studies, both of which use some full-cohort information. Approximate imputation models perform well except when there are interactions or non-linear terms in the outcome model, where imputation using rejection sampling works well.

In utero exposure to ischemic-hypoxic conditions and attention-deficit/hyperactivity disorder

OBJECTIVE

To examine the association between ischemic-hypoxic conditions (IHCs) and attention-deficit/hyperactivity disorder (ADHD) by gestational age and race/ethnicity.

METHODS

Nested case-control study using the Kaiser Permanente Southern California (KPSC) medical records. The study cohort included children aged 5 to 11 years who were delivered and cared for in the KPSC between 1995 and 2010 (N = 308,634). Case children had a diagnosis of ADHD and received ≥ 2 prescriptions specific to ADHD during the follow-up period. For each case, 5 control children were matched by age at diagnosis. Exposures were defined by using International Classification of Diseases, Ninth Revision codes. A conditional regression model was used to estimate adjusted odds ratios (ORs).

RESULTS

Among eligible children, 13,613 (4.3%) had a diagnosis of ADHD. Compared with control children, case children were more likely to be male and of white or African American race/ethnicity. Case children were more likely to be exposed to IHCs (OR = 1.16, 95% confidence interval [CI] 1.11-1.21). When stratified by gestational age, cases born at 28 to 33, 34 to 36, and 37 to 42 weeks of gestation, were more likely to be exposed to IHCs (ORs, 1.6 [95% CI 1.2-2.1], 1.2 [95% CI 1.1-1.3], and 1.1 [95% CI 1.0-1.2], respectively) compared with controls. IHC was associated with increased odds of ADHD across all race/ethnicity groups.

CONCLUSIONS

These findings suggest that IHCs, especially birth asphyxia, respiratory distress syndrome, and preeclampsia, are independently associated with ADHD. This association was strongest in preterm births.

Design and analysis issues in gene and environment studies

Both nurture (environmental) and nature (genetic factors) play an important role in human disease etiology. Traditionally, these effects have been thought of as independent. This perspective is ill informed for non-mendelian complex disorders which result as an interaction between genetics and environment. To understand health and disease we must study how nature and nurture interact. Recent advances in human genomics and high-throughput biotechnology make it possible to study large numbers of genetic markers and gene products simultaneously to explore their interactions with environment. The purpose of this review is to discuss design and analytic issues for gene-environment interaction studies in the "-omics" era, with a focus on environmental and genetic epidemiological studies. We present an expanded environmental genomic disease paradigm. We discuss several study design issues for gene-environmental interaction studies, including confounding and selection bias, measurement of exposures and genotypes. We discuss statistical issues in studying gene-environment interactions in different study designs, such as choices of statistical models, assumptions regarding biological factors, and power and sample size considerations, especially in genome-wide gene-environment studies. Future research directions are also discussed.

Making inferences from a case-control study: implications of sampling

In view of the importance of case-control studies to veterinary epidemiological research, this review article reinforces key aspects of the interpretation of results from these studies. The implications of design elements on inferences are first discussed, followed by a review of three types of case-control study differentiated from each other by the timing of the selection of controls. An hypothetical example of a history of aggression (exposure) and consequent relinquishment (outcome) of pet dogs is used to illustrate and explain the differences between the exposure odds ratio and the outcome odds ratio, and to demonstrate how to interpret exposure odds ratios derived from the three types of case-control study correctly.

Developing novel prognostic biomarkers for multivariate fracture risk prediction algorithms

Multivariate prediction algorithms such as FRAX® and QFractureScores provide an opportunity for new prognostic biomarkers to be developed and incorporated, potentially leading to better fracture prediction. As more research is conducted into these novel biomarkers, a number of factors need to be considered for their successful development for inclusion in these algorithms. In this review, we describe two well-known multivariate prediction algorithms for osteoporosis fracture risk applicable to the UK population, FRAX and QFractureScores, and comment on the current prognostic tools available for fracture risk; dual X-ray assessment, quantitative ultrasonography, and genomic/biochemical markers. We also highlight the factors that need to be considered in the development of new biomarkers. These factors include the requirement for prospective data, collected in new cohort studies or using archived samples; the need for adequate stability data to be provided; and the need for appropriate storage methods to be used when retrospective data are required. Area under the receiver operating characteristic curve measures have been found to have limited utility in assessing the impact of the addition of new risk factors on the predictive performance of multivariate algorithms. New performance evaluation measures, such as net reclassification index and integrated discrimination improvement, are increasingly important in the evaluation of the impact of the addition of new markers to multivariate algorithms, and these are also discussed.

Conventional case-cohort design and analysis for studies of interaction

BACKGROUND

The case-cohort study design has received significant methodological attention in the statistical and epidemiological literature but has not been used as widely as other cohort-based sampling designs, such as the nested case-control design. Despite its efficiency and practicality for a wide range of epidemiological study purposes, researchers may not yet be aware of the fact that the design can be analysed using standard software with only minor adjustments. Furthermore, although the large number of options for design and analysis of case-cohort studies may be daunting, they can be reduced to a few simple recommendations.

METHODS

We review conventional methods for the design and analysis of case-cohort studies and describe empirical comparisons based on a study of radiation, gene polymorphisms and cancer in the Japanese atomic bomb survivor cohort.

RESULTS

Stratified, as opposed to simple, random subcohort selection is recommended, especially for studies of gene-environment interaction, which are notorious for lacking statistical power. Methods based on the score-unbiased exact pseudo-likelihood (or its analogue with stratified case-cohort data) are recommended for use in conjunction with the asymptotic variance estimator.

CONCLUSIONS

We present an example of how to implement case-cohort analysis methods using SPSS, a popular statistical package that lacks some of the features necessary to directly adapt and implement published methods based on other software platforms. We also illustrate case-control analysis using Epicure, which provides greater risk-modelling flexibility than other software. Our conclusions and recommendations should help investigators to better understand and apply the case-cohort design in epidemiological research.