Estimating Additive Interaction in Two-Stage Individual Participant Data Meta-Analysis.

Individual participant data (IPD) meta-analysis provides important opportunities to study interaction and effect modification for which individual studies often lack power. While previous meta-analyses have commonly focused on multiplicative interaction, additive interaction holds greater relevance for public health and may in certain contexts better reflect biological interaction. Methodological literature on interaction in IPD meta-analysis does not cover additive interaction for models including binary or time-to-event outcomes. We aimed to describe how the Relative Excess Risk due to Interaction (RERI) and other measures of additive interaction or effect modification can be validly estimated within two-stage IPD meta-analysis. First, we explain why direct pooling of study-level RERI estimates may lead to invalid results. Next, we propose a three-step procedure to estimate additive interaction: 1) estimate effects of both exposures and their product term on the outcome within each individual study; 2) pool study-specific estimates using multivariate meta-analysis; 3) estimate an overall RERI and 95% confidence interval based on the pooled effect estimates. We illustrate this procedure by investigating interaction between depression and smoking and risk of smoking-related cancers using data from the PSYchosocial factors and Cancer (PSY-CA) consortium. We discuss implications of this procedure, including the application in meta-analysis based on published data.

Additive effects of a family history of schizophrenia spectrum disorders and an environmental risk score for the outcome of patients with non-affective first-episode psychosis.

BACKGROUND: First-episode psychotic disorders comprise a heterogeneous phenotype with a complex etiology involving numerous common small-effect genetic variations and a wide range of environmental exposures. We examined whether a family of schizophrenia spectrum disorder (FH-Sz) interacts with an environmental risk score (ERS-Sz) regarding the outcome of patients with non-affective first episode psychosis (NAFEP). METHODS: We included 288 patients with NAFEP who were evaluated after discharge from an intensive 2-year program. We evaluated three outcome measures: symptomatic remission, psychosocial functioning, and personal recovery. We analyzed the main and joint associations of a FH-Sz and the ERS-Sz on the outcomes by using the relative excess risk due to interaction (RERI) approach. RESULTS: A FH-Sz showed a significant association with poor symptomatic remission and psychosocial functioning outcomes, although there was no significant interaction between a FH-Sz and the ERS-Sz on these outcomes. The ERS-Sz did not show a significant association with poor symptomatic remission and psychosocial functioning outcomes, even though the magnitude of the interaction between ERS-Sz and FH-Sz with the later outcome was moderate (RERI = 6.89, 95% confidence interval -16.03 to 29.81). There was no association between a FH-Sz and the ERS-Sz and personal recovery. CONCLUSIONS: Our results provide further empirical support regarding the contribution of FH-Sz to poor symptomatic remission and poor psychosocial functioning outcomes in patients with NAFEP.

Impact of PD-1 gene polymorphism and its interaction with tea drinking on susceptibility to tuberculosis.

The aim of this study was to explore the impact of polymorphism of PD-1 gene and its interaction with tea drinking on susceptibility to tuberculosis (TB). A total of 503 patients with TB and 494 controls were enrolled in this case-control study. Three single-nucleotide polymorphisms of PD-1 (rs7568402, rs2227982 and rs36084323) were genotyped and unconditional logistic regression analysis was used to identify the association between PD-1 polymorphism and TB, while marginal structural linear odds models were used to estimate the interactions. Genotypes GA (OR 1.434), AA (OR 1.891) and GA + AA (OR 1.493) at rs7568402 were more prevalent in the TB patients than in the controls (P < 0.05). The relative excess risk of interaction (RERI) between rs7568402 of PD-1 genes and tea drinking was -0.3856 (95% confidence interval -0.7920 to -0.0209, P < 0.05), which showed a negative interaction. However, the RERIs between tea drinking and both rs2227982 and rs36084323 of PD-1 genes were not statistically significant. Our data demonstrate that rs7568402 of PD-1 genes was associated with susceptibility to TB, and there was a significant negative interaction between rs7568402 and tea drinking. Therefore, preventive measures through promoting the consumption of tea should be emphasised in the high-risk populations.

Prognostic and Predictive Values and Statistical Interactions in the Era of Targeted Treatment.

The current era of targeted treatment has accelerated the interest in studying gene-treatment, gene-gene, and gene-environment interactions using statistical models in the health sciences. Interactions are incorporated into models as product terms of risk factors. The statistical significance of interactions is traditionally examined using a likelihood ratio test (LRT). Epidemiological and clinical studies also evaluate interactions in order to understand the prognostic and predictive values of genetic factors. However, it is not clear how different types and magnitudes of interaction effects are related to prognostic and predictive values. The contribution of interaction to prognostic values can be examined via improvements in the area under the receiver operating characteristic curve due to the inclusion of interaction terms in the model (ΔAUC). We develop a resampling based approach to test the significance of this improvement and show that it is equivalent to LRT. Predictive values provide insights into whether carriers of genetic factors benefit from specific treatment or preventive interventions relative to noncarriers, under some definition of treatment benefit. However, there is no unique definition of the term treatment benefit. We show that ΔAUC and relative excess risk due to interaction (RERI) measure predictive values under two specific definitions of treatment benefit. We investigate the properties of LRT, ΔAUC, and RERI using simulations. We illustrate these approaches using published melanoma data to understand the benefits of possible intervention on sun exposure in relation to the MC1R gene. The goal is to evaluate possible interventions on sun exposure in relation to MC1R.

Mediterranean diet and hepatocellular carcinoma.

BACKGROUND & AIMS: Hepatocellular carcinoma (HCC) has a very poor prognosis and any effort to identify additional risk factors, besides those already established, would be important for the prevention of the disease. Data on the role of diet on HCC risk are still controversial. METHODS: We have evaluated the association of adherence to the Mediterranean diet with HCC risk, as well as the interaction of this dietary pattern with chronic hepatitis infection, by combining two case-control studies undertaken in Italy and Greece, including overall 518 cases of HCC and 772 controls. Adherence to the traditional Mediterranean diet was assessed through the Mediterranean diet score (MDS), which ranges between 0 (lowest adherence) and 9 (highest adherence). Odds ratios (OR) for HCC were obtained through multiple logistic regression models, controlling for potentially confounding factors, including chronic infection with hepatitis B/C viruses. RESULTS: Compared to MDS of 0-3, the ORs for HCC were 0.66 (95% confidence interval (CI), 0.41-1.04) for MDS equal to 4 and 0.51 (95% CI, 0.34-0.75) for MDS ⩾ 5, with a significant trend (p<0.001). The detrimental effect of poor adherence to Mediterranean diet on HCC risk was disproportionally high among those chronically infected with hepatitis B and/or C viruses, with a suggestion of super-additive interaction, albeit statistically non-significant. CONCLUSIONS: Closer adherence to the Mediterranean diet appears to be protective against HCC. Our results also point to potential benefits from adhering to a Mediterranean dietary pattern for patients chronically infected with hepatitis viruses.

The Association between Autism Spectrum Disorder and Precocious Puberty: Considering Effect Modification by Sex and Neuropsychiatric Comorbidities.

Limited knowledge is available about the association between autistic spectrum disorder (ASD) and precocious puberty. Our study examined the association between the two medical conditions and effect modification by sex and neuropsychiatric comorbidities in a nationwide population. To compare the risk of precocious puberty between ASD and non-ASD cases, we conducted a Cox regression analysis using ASD as the exposure and time to precocious puberty as the outcome. We adjusted for sex, attention-deficit/hyperactivity disorder (ADHD), tic disorder, obsessive-compulsive disorder (OCD), anxiety disorder, intellectual disability, and epilepsy. We performed a moderation analysis to examine the potential moderating effects of sex and comorbidities. Patients with ASD were prone to have precocious puberty, with an adjusted hazard ratio (aHR) of 1.80 (95% CI: 1.61-2.01). For effect modification, sex, specifically females, moderated the association between ASD and precocious puberty, with a relative excess risk due to interaction (RERI) of 7.35 (95% CI 4.90-9.80). No significant effect modification was found for any of the comorbidities within the scope of additive effect modification. We found that patients with ASD were prone to precocious puberty, regardless of sex or comorbid neuropsychiatric disorders. Girls with ASD are at a particularly higher risk of developing precocious puberty.

Synergistic effects of family history of hepatocellular carcinoma and hepatitis B virus infection on risk for incident hepatocellular carcinoma.

BACKGROUND & AIMS: Little is known about the effects of family history of hepatocellular carcinoma (HCC) on hepatitis B progression or risk of HCC. We examined how family HCC history and presence or stage of hepatitis B virus (HBV) infection affect risk for HCC. METHODS: We performed a population-based cohort study of 22,472 participants from 7 townships in Taiwan who underwent evaluation for liver disease from 1991 through 1992. Those who received a first diagnosis of HCC from January 1, 1991, to December 31, 2008, were identified from the Taiwanese cancer registry. RESULTS: There were 374 cases of incident HCC over 362,268 person-years of follow-up evaluation. The cumulative risk of HCC in hepatitis B surface antigen (HBsAg)-seronegative patients without a family history of HCC was 0.62%, in those with a family history of HCC the cumulative risk was 0.65%, in HBsAg-seropositive patients without a family history of HCC the cumulative risk was 7.5%, and in HBsAg-seropositive patients with a family history of HCC the cumulative risk was 15.8% (P < .001). The multivariate-adjusted hazard ratio for HBsAg-seropositive individuals with family history, compared with HBsAg-seronegative individuals without a family history of HCC, was 32.33 (95% confidence interval, 20.8-50.3; P < .001). The relative excess risk owing to interaction was 19, the attributable proportion was 0.59, and the synergy index value was 2.54. These findings indicate synergy between family HCC history and HBsAg serostatus. The synergy between these factors remained significant in stratification analyses by HBeAg serostatus and serum level of HBV DNA. CONCLUSIONS: Family history of HCC multiplies the risk of HCC at each stage of HBV infection. Patients with a family history of HCC require more intensive management of HBV infection and surveillance for liver cancer.

Childhood blood eosinophils and symptoms of allergic disorders: a cross-sectional study in Southern China.

PURPOSE: The relationship between childhood blood eosinophils and subtypes of allergic diseases remains understudied. This study aimed to examine the associations between childhood blood eosinophils and subtypes of asthma, rhinitis and dermatitis, as well as the modifying effect of age. METHODS: We obtained concurrent blood cell counts and serum Immunoglobulin E (IgE) test results in 5026 children (0-13, years) from First Affiliated Hospital of Guangzhou Medical University from 2014 to 2019. Generalized additive models with multivariable adjustments were utilized to model the exposure-response relationship between eosinophils and allergic symptoms. The robustness of the association was assessed in two age categories (<6, 6-13 years). RESULTS: The association of eosinophils with allergic asthma/rhinitis was positively nonlinear, with a plateau at levels of Q4 (≥0.51, 109/L). Conversely, exposure-response curves between eosinophils and the risk of non-allergic asthma and rhinitis were negatively linear, and especially, became statistically significant when levels of eosinophils were larger than Q3 (≥0.30, 109/L). Compared with their counterparts, school-aged children (6-13, years) with a higher level of blood eosinophils (≥0.35, 109/L) were more likely to suffer from allergic asthma [relative excess risk due to interaction (RERI), 2.51; 95% CI, 1.24-3.78], allergic rhinitis (RERI, 2.79; 95% CI, 1.14-4.45) but not allergic dermatitis (RERI not significant). CONCLUSION: Higher eosinophil counts were associated with the increased risk of allergic subtype symptoms and the decreased risk of non-allergic subtypes in children. Moreover, the associations between eosinophils and allergic asthma/rhinitis were accentuated in the school-aged child. These findings may contribute to providing novel insights for clinical administration relevance of allergic-related symptoms.Key messages:There was a positively nonlinear association between childhood eosinophils and allergic asthma/rhinitis.Age modified the associations between eosinophils and allergy-related outcomes. The associations of eosinophil with allergic asthma/rhinitis accentuated in the school-aged child (6-13, years).

Detection of Synergistic Interaction on an Additive Scale Between Two Drugs on Abnormal Elevation of Serum Alanine Aminotransferase Using Machine-Learning Algorithms.

Drug-induced liver injury (DILI) is a common adverse drug reaction, with abnormal elevation of serum alanine aminotransferase (ALT). Several clinical studies have investigated whether a combination of two drugs alters the reporting frequency of DILI using traditional statistical methods such as multiple logistic regression (MLR), but this model may over-fit the data. This study aimed to detect a synergistic interaction between two drugs on the risk of abnormal elevation of serum ALT in Japanese adult patients using three machine-learning algorithms: MLR, logistic least absolute shrinkage and selection operator (LASSO) regression, and extreme gradient boosting (XGBoost) algorithms. A total of 58,413 patients were extracted from Nihon University School of Medicine's Clinical Data Warehouse and assigned to case (N = 4,152) and control (N = 54,261) groups. The MLR model over-fitted a training set. In the logistic LASSO regression model, three combinations showed relative excess risk due to interaction (RERI) for abnormal elevation of serum ALT: diclofenac and famotidine (RERI 2.427, 95% bootstrap confidence interval 1.226-11.003), acetaminophen and ambroxol (0.540, 0.087-4.625), and aspirin and cilostazol (0.188, 0.135-3.010). Moreover, diclofenac (adjusted odds ratio 1.319, 95% bootstrap confidence interval 1.189-2.821) and famotidine (1.643, 1.332-2.071) individually affected the risk of abnormal elevation of serum ALT. In the XGBoost model, not only the individual effects of diclofenac (feature importance 0.004) and famotidine (0.016), but also the interaction term (0.004) was included in important predictors. Although further study is needed, the combination of diclofenac and famotidine appears to increase the risk of abnormal elevation of serum ALT in the real world.

Excess Risk of Dental Caries from Higher Free Sugars Intake Combined with Low Exposure to Water Fluoridation.

The study aimed to quantify the excess risk of interaction between high free sugars (sugars) intake and lack of exposure to water fluoridation on child dental caries. Data from the Australian National Child Oral Health Study, a population-based survey of 24,664 children aged 5 to 14 y, were collected using parental questionnaires and oral epidemiological examinations by trained examiners. Information on socioeconomic status, dental health behaviors, and dental service use was used as covariates. The number of servings of sugars-containing foods and drinks consumed in a usual day was assessed as the main exposure, categorized into 5 groups. Residential history was used to calculate lifetime exposure to fluoridated water (LEFW), categorized as low (<25%), medium (25% to <75%), or high (75%-100%). Caries prevalence (dmfs/DMFS >0) and experience (dmfs/DMFS) in the primary (ages 5-10 y) and permanent (ages 8-14 y) dentitions were the main dependent variables. The association of sugars intake and LEFW with each outcome was estimated in multivariable log-Poisson regression models with robust standard error estimation, adjusted for covariates. The relative excess risk due to interaction (RERI) between sugars intake and LEFW was estimated. Strong positive gradients in all outcomes were observed across sugars intake groups. Relative to the lowest intake group, the 3 highest intake groups had significantly higher adjusted prevalence ratios for having caries and higher adjusted mean ratios of caries experience in both dentitions, after controlling for all covariates. LEFW strongly and consistently attenuated the effects of all levels of sugars intake on the outcomes. RERI estimates indicated that a combination of lack of exposure to fluoridated water and high sugars intake resulted in greater excess risk of primary and permanent caries than if there was no interaction. Evidently, children with high sugars intakes and low exposure to water fluoridation are at disproportionately higher risk of dental caries.

Moving from two- to multi-way interactions among binary risk factors on the additive scale.

Many studies have focused on investigating deviations from additive interaction of two dichotomous risk factors on a binary outcome. There is, however, a gap in the literature with respect to interactions on the additive scale of >2 risk factors. In this paper, we present an approach for examining deviations from additive interaction among three or more binary exposures. The relative excess risk due to interaction (RERI) is used as measure of additive interaction. First, we concentrate on three risk factors - we propose to decompose the total RERI to: the RERI owned to the joint presence of all three risk factors and the RERI of any two risk factors, given that the third is absent. We then extend this approach, to >3 binary risk factors. For illustration, we use a sample from data from the Greek EPIC cohort and we investigate the association with overall mortality of Mediterranean diet, body mass index , and smoking. Our formulae enable better interpretability of any evidence for deviations from additivity owned to more than two risk factors and provide simple ways of communicating such results from a public health perspective by attributing any excess relative risk to specific combinations of these factors. Abbreviations: BMI: Body Mass Index; ERR: excess relative risk; EPIC: European Prospective Investigation into Cancer and nutrition; MD: Mediterranean diet; RERI: relative excess risk due to interaction; RR: relative risk; TotRERI: total relative excess risk due to interaction.

Aspirin for primary prevention of stroke in traumatic cerebrovascular injury: association with increased risk of transfusion.

OBJECTIVEBlunt traumatic extracranial carotid or vertebral artery injury (i.e., traumatic cerebrovascular injury [TCVI]) occurs in 1%-2% of all blunt trauma admissions, carries a 10% risk of thromboembolic ischemic stroke, and accounts for up to 9600 strokes annually in the US. Screening CT angiograms (CTAs) of patients with trauma has become ubiquitous in recent years, and patients with initially asymptomatic TCVI are commonly treated with antiplatelet agents to prevent stroke. Prophylaxis with antiplatelets is thought to be safer than anticoagulation, which carries a significant risk of hemorrhage in patients with trauma. However, the risk of hemorrhagic complications due to antiplatelets has not been assessed in this population.METHODSThis is a retrospective cohort study of patients in whom a screening CTA was obtained after admission for blunt trauma at a Level 1 trauma center. Patients with CTAs indicating TCVI were treated routinely with 325 mg aspirin daily. The risk of transfusion > 24 hours after admission was compared according to CTA findings (CTA+ or CTA- for positive or negative findings, respectively) and aspirin treatment (ASA+ or ASA- for treatment or no treatment, respectively).RESULTSThe mean overall transfusion amount (number of units of packed red blood cells [PRBCs]) was 0.9 ± 2.1 for CTA+/ASA+ patients (n = 196) and 0.3 ± 1.60 for CTA-/ASA- patients (n = 2290) (p < 0.0001). In adjusted models, the overall relative risk (RR) of PRBC transfusion was 1.70 (1.32-2.20) for CTA+/ASA+ patients compared with CTA-/ASA- patients. Among age groups, participants whose ages were 50-69 years had the greatest significantly elevated RR (1.71, 95% CI 1.08-2.72) for CTA+/ASA+ patients compared with CTA-/ASA- patients.CONCLUSIONSTreatment with aspirin for the prevention of stroke in patients with initially asymptomatic TCVI carries a significantly increased risk of PRBC transfusion. Future studies are needed to determine if this risk is offset by a reduced risk of ischemic stroke.

Joint Effect of Hypertension and Elevated Serum Phosphorus on the Risk of Mortality in National Health and Nutrition Examination Survey-III.

BACKGROUND: Elevated serum phosphorus might aggravate the effect of hypertension on mortality. The objective of this study was to examine the joint effect of hypertension and serum phosphorus on the risk of mortality. METHODS AND RESULTS: A large prospective (n=15 833), population-based cohort of participants from the National Health and Nutritional Examination Survey III was examined to test potential synergism between hypertension, elevated serum phosphorus, and the risk of mortality. Interaction on additive scale and multiplicative scale was estimated. After a median follow-up of 14.3 years, 1691 cases of cardiovascular mortality and 3875 cases of all-cause mortality were identified. Interaction was observed between hypertension and elevated serum phosphorus on the additive scale for cardiovascular mortality (relative excess risk due to interaction, 0.99, 95% CI: 0.06; 1.92, adjusted for age, gender, race, and estimated glomerular filtration rate). No statistically significant interaction was found between hypertension and serum phosphorus for all-cause mortality on the additive scale. No significant interaction was detected on the multiplicative scale. In sensitivity analysis, excluding participants who died in first 2 years and adjustment for additional confounders resulted in essentially similar findings. CONCLUSIONS: The joint effect of hypertension and elevated serum phosphorus was larger than the sum of the independent effects on cardiovascular mortality but not on all-cause mortality. Future studies should investigate whether controlling elevated serum phosphorus in hypertensive individuals helps in prevention of extra risk of cardiovascular mortality.

Physical activity and dietary behavior in US adults and their combined influence on health.

OBJECTIVE: To examine the association between objectively measured physical activity and dietary behavior and their combined effect on health. PATIENTS AND METHODS: Data for this study were obtained from the 2003-2006 National Health and Nutrition Examination Survey cycles. The data were evaluated between September 9, 2012, and August 14, 2013. As part of the national survey, participants wore an accelerometer for 4 or more days to assess physical activity, blood samples were obtained to assess various biological markers, and interviews were conducted to assess dietary behavior. We selected a sample of 5211 participants and categorized them into 4 groups: (1) healthy diet and active, (2) unhealthy diet and active, (3) healthy diet and inactive, and (4) unhealthy diet and inactive. RESULTS: A total of 16.5% of participants (weighted proportions) were classified as consuming a healthy diet and being sufficiently active. After adjustments, participants were 32% more likely to consume a healthy diet if they met physical activity guidelines. For nearly all biomarkers, those who consumed a healthy diet and were sufficiently active had the most favorable biomarker levels. Compared with those who consumed a healthy diet and were active, participants who consumed an unhealthy diet and were inactive were 2.4 times more likely to have metabolic syndrome. CONCLUSION: Our findings indicate a relationship between objectively measured physical activity and dietary behavior and that participating in regular physical activity and eating a healthy diet are associated with better health outcomes when compared with diet or physical activity alone.