Selective inference for effect modification via the lasso

Neighborhood greenspace may attract new residents and lead to sociodemographic or housing cost changes. We estimated relationships between greenspace and gentrification-related changes in the 43 largest metropolitan statistical areas (MSAs) of the United States (US). We used the US National Land Cover and Brown University Longitudinal Tracts databases, as well as spatial lag models, to estimate census tract-level associations between percentage greenspace (years 1990, 2000) and subsequent changes (1990–2000, 2000–2010) in percentage college-educated, percentage working professional jobs, race/ethnic composition, household income, percentage living in poverty, household rent, and home value. We also investigated effect modification by racial/ethnic composition. We ran models for each MSA and time period and used random-effects meta-analyses to derive summary estimates for each period. Estimates were modest in magnitude and heterogeneous across MSAs. After adjusting for census-tract level population density in 1990, compared to tracts with low percentage greenspace in 1992 (defined as ≤50th percentile of the MSA-specific distribution in 1992), those with high percentage greenspace (defined as >75th percentile of the MSA-specific distribution) experienced higher 1990–2000 increases in percentage of the employed civilian aged 16+ population working professional jobs (β: 0.18, 95% confidence interval (CI): 0.11, 0.26) and in median household income (β: 0.23, 95% CI: 0.15, 0.31). Adjusted estimates for the 2000–2010 period were near the null. We did not observe evidence of effect modification by race/ethnic composition. We observed evidence of modest associations between greenspace and gentrification trends. Further research is needed to explore reasons for heterogeneity and to quantify health implications.

Download Full-text

Post-randomization Biomarker Effect Modification Analysis in an HIV Vaccine Clinical Trial

Journal of Causal Inference ◽

10.1515/jci-2019-0022 ◽

2020 ◽

Vol 8 (1) ◽

pp. 54-69

Author(s):

Peter B. Gilbert ◽

Bryan S. Blette ◽

Bryan E. Shepherd ◽

Michael G. Hudgens

Keyword(s):

Causal Effect ◽

Effect Modification ◽

Hiv Vaccine ◽

Strongly Correlated ◽

Principal Stratification ◽

Response Variable ◽

Average Causal Effect ◽

Intermediate Response ◽

Structural Risk ◽

Augmented Design

AbstractWhile the HVTN 505 trial showed no overall efficacy of the tested vaccine to prevent HIV infection over placebo, markers measuring immune response to vaccination were strongly correlated with infection. This finding generated the hypothesis that some marker-defined vaccinated subgroups were partially protected whereas others had their risk increased. This hypothesis can be assessed using the principal stratification framework (Frangakis and Rubin, 2002) for studying treatment effect modification by an intermediate response variable, using methods in the sub-field of principal surrogate (PS) analysis that studies multiple principal strata. Unfortunately, available methods for PS analysis require an augmented study design not available in HVTN 505, and make untestable structural risk assumptions, motivating a need for more robust PS methods. Fortunately, another sub-field of principal stratification, survivor average causal effect (SACE) analysis (Rubin, 2006) – which studies effects in a single principal stratum – provides many methods not requiring an augmented design and making fewer assumptions. We show how, for a binary intermediate response variable, methods developed for SACE analysis can be adapted to PS analysis, providing new and more robust PS methods. Application to HVTN 505 supports that the vaccine partially protected individuals with vaccine-induced T-cells expressing certain combinations of functions.

Download Full-text

Associations between smoking and blood-group, and the risk of dyslipidaemia amongst French women

Scientific Reports ◽

10.1038/s41598-021-94239-9 ◽

2021 ◽

Vol 11 (1) ◽

Author(s):

C. J. MacDonald ◽

A. L. Madika ◽

G. Severi ◽

A. Fournier ◽

M. C. Boutron-Ruault

Keyword(s):

Blood Group ◽

Smoking Status ◽

Effect Modification ◽

Blood Groups ◽

Cross Sectional ◽

Increased Risk ◽

Cardio Vascular Disease ◽

Never Smokers ◽

Cardio Vascular ◽

Incident Cases

AbstractDyslipidaemia is a major risk factor for cardio-vascular disease, as it promotes atherosclerosis. While cross-sectional studies have identified higher serum cholesterol amongst individuals with the A blood group, there is less evidence from prospective studies whether this translates into a higher risk of dyslipidaemia that requires treatment, nor if this genetic factor interacts with smoking status. This study aimed to prospectively determine potential associations between smoking, ABO blood groups, and risk of incident dyslipidaemia requiring treatment, and to assess associations over strata of blood ABO group. We assessed associations between blood ABO group, smoking and dyslipidaemia in 74,206 women participating in the E3N cohort. We included women who did not have cardiovascular disease at baseline. Logistic regression was used to determine associations between ABO group, smoking and prevalent dyslipidaemia at baseline. Cox proportional hazard models were then used to determine if blood ABO group and smoking were associated with the risk of incident dyslipidaemia, amongst women free of dyslipidaemia at baseline. At baseline 28,281 women with prevalent dyslipidaemia were identified. Compared to the O-blood group, the non-O blood group was associated higher odds of with prevalent dyslipidaemia (ORnon-O = 1.09 [1.06: 1.13]). Amongst the women free of dyslipidaemia at baseline, 6041 incident cases of treated dyslipidaemia were identified during 454,951 person-years of follow-up. The non-O blood groups were associated with an increased risk of dyslipidaemia when compared to the O-group (HRnon-O = 1.16 [1.11: 1.22]), specifically the A blood-group (HRA = 1.18 [1.12: 1.25]). Current smokers were associated with an increased risk of incident dyslipidaemia (HR smokers = 1.27 [1.16: 1.37]), compared to never-smokers. No evidence for effect modification between smoking and ABO blood group was observed (p-effect modification = 0.45), although the highest risk was observed among AB blood group women who smoked (HR = 1.76 [1.22: 2.55]). In conclusion, the non-O blood groups, specifically the A group were associated with an increased risk of dyslipidaemia. Current smokers were associated with a 30% increased risk of dyslipidaemia. These results could aid in personalised approaches to the prevention of cardiovascular risk-factors.

Download Full-text

Educational inequalities in obesity: a multilevel analysis of survey data from cities in Latin America

Public Health Nutrition ◽

10.1017/s1368980021002457 ◽

2021 ◽

pp. 1-9

Author(s):

Mónica Mazariegos ◽

Amy H Auchincloss ◽

Ariela Braverman-Bronstein ◽

María F Kroker-Lobos ◽

Manuel Ramírez-Zea ◽

...

Keyword(s):

Higher Education ◽

Economic Development ◽

Survey Data ◽

Latin American ◽

Effect Modification ◽

University Education ◽

Educational Inequalities ◽

Obesity Prevalence ◽

Socio Economic Development ◽

American Cities

Abstract Objective: Using newly harmonised individual-level data on health and socio-economic environments in Latin American cities (from the Salud Urbana en América Latina (SALURBAL) study), we assessed the association between obesity and education levels and explored potential effect modification of this association by city-level socio-economic development. Design: This cross-sectional study used survey data collected between 2002 and 2017. Absolute and relative educational inequalities in obesity (BMI ≥ 30 kg/m2, derived from measured weight and height) were calculated first. Then, a two-level mixed-effects logistic regression was run to test for effect modification of the education–obesity association by city-level socio-economic development. All analyses were stratified by sex. Setting: One hundred seventy-six Latin American cities within eight countries (Brazil, Chile, Colombia, Costa Rica, El Salvador, Guatemala, Mexico and Peru). Participants: 53 186 adults aged >18 years old. Results: Among women, 25 % were living with obesity and obesity was negatively associated with educational level (higher education–lower obesity) and this pattern was consistent across city-level socio-economic development. Among men, 18 % were living with obesity and there was a positive association between education and obesity (higher education–higher obesity) for men living in cities with lower levels of development, whereas for those living in cities with higher levels of development, the pattern was inverted and university education was protective of obesity. Conclusions: Among women, education was protective of obesity regardless, whereas among men, it was only protective in cities with higher levels of development. These divergent results suggest the need for sex- and city-specific interventions to reduce obesity prevalence and inequalities.

Download Full-text

Diabetes and endometrial cancer: effect modification by body weight, physical activity and hypertension

British Journal of Cancer ◽

10.1038/sj.bjc.6603933 ◽

2007 ◽

Vol 97 (7) ◽

pp. 995-998 ◽

Cited By ~ 39

Author(s):

E Lucenteforte ◽

C Bosetti ◽

R Talamini ◽

M Montella ◽

A Zucchetto ◽

...

Keyword(s):

Physical Activity ◽

Body Weight ◽

Endometrial Cancer ◽

Effect Modification

Download Full-text

Vitamin D status and risk of type 2 diabetes in the Norwegian HUNT cohort study: does family history or genetic predisposition modify the association?

BMJ Open Diabetes Research & Care ◽

10.1136/bmjdrc-2020-001948 ◽

2021 ◽

Vol 9 (1) ◽

pp. e001948

Author(s):

Marion Denos ◽

Xiao-Mei Mai ◽

Bjørn Olav Åsvold ◽

Elin Pettersen Sørgjerd ◽

Yue Chen ◽

...

Keyword(s):

Type 2 Diabetes ◽

Family History ◽

Genetic Predisposition ◽

Effect Modification ◽

P Value ◽

Family History Of Diabetes ◽

Increased Risk ◽

History Of

IntroductionWe sought to investigate the relationship between serum 25-hydroxyvitamin D (25(OH)D) level and the risk of type 2 diabetes mellitus (T2DM) in adults who participated in the Trøndelag Health Study (HUNT), and the possible effect modification by family history and genetic predisposition.Research design and methodsThis prospective study included 3574 diabetes-free adults at baseline who participated in the HUNT2 (1995–1997) and HUNT3 (2006–2008) surveys. Serum 25(OH)D levels were determined at baseline and classified as <50 and ≥50 nmol/L. Family history of diabetes was defined as self-reported diabetes among parents and siblings. A Polygenic Risk Score (PRS) for T2DM based on 166 single-nucleotide polymorphisms was generated. Incident T2DM was defined by self-report and/or non-fasting glucose levels greater than 11 mmol/L and serum glutamic acid decarboxylase antibody level of <0.08 antibody index at the follow-up. Multivariable logistic regression models were applied to calculate adjusted ORs with 95% CIs. Effect modification by family history or PRS was assessed by likelihood ratio test (LRT).ResultsOver 11 years of follow-up, 92 (2.6%) participants developed T2DM. A higher risk of incident T2DM was observed in participants with serum 25(OH)D level of<50 nmol/L compared with those of ≥50 nmol/L (OR 1.72, 95% CI 1.03 to 2.86). Level of 25(OH)D<50 nmol/L was associated with an increased risk of T2DM in adults without family history of diabetes (OR 3.87, 95% CI 1.62 to 9.24) but not in those with a family history (OR 0.72, 95% CI 0.32 to 1.62, p value for LRT=0.003). There was no effect modification by PRS (p value for LRT>0.23).ConclusionSerum 25(OH)D<50 nmol/L was associated with an increased risk of T2DM in Norwegian adults. The inverse association was modified by family history of diabetes but not by genetic predisposition to T2DM.

Download Full-text