The challenging interpretation of instrumental variable estimates under monotonicity

Causal relationship between serum metabolites and idiopathic pulmonary fibrosis: Insights from a two-sample Mendelian randomization study

Background

Idiopathic pulmonary fibrosis (IPF) is an irreversible lung disease with unclear pathological mechanisms. In this study, we utilized bidirectional Mendelian randomization (MR) to analyze the relationship between serum metabolites and IPF, and conducted metabolic pathway analysis.

Aim

To determine the causal relationship between serum metabolites and IPF using MR analysis.

Methods

A two-sample MR analysis was conducted to evaluate the causal relationship between 824 serum metabolites and IPF. The inverse variance weighted (IVW) method was used to estimate the causal relationship between exposure and results. Sensitivity analysis was conducted using MR Egger, weighted median, and maximum likelihood to eliminate pleiotropy. Additionally, metabolic pathway analysis was conducted to identify potential metabolic pathways.

Results

We identified 12 serum metabolites (6 risks and 6 protective) associated with IPF from 824 metabolites. Among them, 11 were known and 1 was unknown. 1-Eicosatrienoylglycophorophospholine and 1-myristoylglycophorophospholine were bidirectional MR positive factors, with 1-myristoylglycophorophospholine being a risk factor (1.0013, 1.0097) and 1-eicosatrienoylglycophorine being a protective factor (0.9914, 0.9990). The four lipids (1-linoleoylglycerophoethanolamine*, total cholesterol in large high-density lipoprotein [HDL], cholesterol esters in very large HDL, and phospholipids in very large HDL) and one NA metabolite (degree of unsaturation) were included in the known hazardous metabolites. The known protective metabolites included three types of lipids (carnitine, 1-linoleoylglycerophoethanolamine*, and 1-eicosatrienoylglycerophophophorine), one amino acid (hypoxanthine), and two unknown metabolites (the ratio of omega-6 fatty acids to omega-3 fatty acids, and the ratio of photoshopids to total lipids ratio in chylomicrons and extremely large very low-density lipoprotein [VLDL]). Moreover, sn-Glycerol 3-phosphate and 1-Acyl-sn-glycero-3-phosphocline were found to be involved in the pathogenesis of IPF through metabolic pathways such as Glycerolide metabolism and Glycerophospholipid metabolism.

Conclusion

Our study identified 6 causal risks and 6 protective serum metabolites associated with IPF. Additionally, 2 metabolites were found to be involved in the pathogenesis of IPF through metabolic pathways, providing a new perspective for further understanding the metabolic pathway and the pathogenesis of IPF.

To treat or not to treat: a comparative effectiveness analysis of oral anticoagulant outcomes among U.S. nursing home residents with atrial fibrillation

BACKGROUND

Nursing home residents with atrial fibrillation are at high risk for ischemic stroke, but most are not treated with anticoagulants. This study compared the effectiveness and safety between oral anticoagulant (OAC) users and non-users.

METHODS

We conducted a new-user retrospective cohort study by using Minimum Data Set 3.0 assessments linked with Medicare claims. The participants were Medicare fee-for-service beneficiaries with atrial fibrillation residing in US nursing homes between 2011 and 2016, aged ≥ 65 years. The primary outcomes were occurrence of an ischemic stroke or systemic embolism (effectiveness), occurrence of intracranial or extracranial bleeding (safety) and net clinical outcome (effectiveness or safety outcomes). Secondary outcomes included total mortality and a net clinical and mortality outcome. Cox proportional hazards and Fine and Grey models estimated multivariable adjusted hazard ratios (aHRs) and sub-distribution hazard ratios (sHRs).

RESULTS

Outcome rates were low (effectiveness: OAC: 0.86; non-users: 1.73; safety: OAC: 2.26; non-users: 1.75 (per 100 person-years)). OAC use was associated with a lower rate of the effectiveness outcome (sHR: 0.69; 95% Confidence Interval (CI): 0.61-0.77), higher rates of the safety (sHR: 1.70; 95% CI: 1.58-1.84) and net clinical outcomes (sHR: 1.20; 95% CI: 1.13-1.28) lower rate of all-cause mortality outcome (sHR: 0.60; 95% CI: 0.59-0.61), and lower rate of the net clinical and mortality outcome (sHR: 0.60; 95% CI: 0.59-0.61). Warfarin users, but not DOAC users, had a higher rate of the net clinical outcome versus OAC non-users.

CONCLUSIONS

Our results support the benefits of treatment with OACs to prevent ischemic strokes and increase longevity, while highlighting the need to weigh apparent benefits against elevated risk for bleeding. Results were consistent with net favorability of DOACs versus warfarin.

Associations between special diet and incidence risk of osteoporosis: a Mendelian randomization study

Introduction

Osteoporosis is a prevalent challenge in clinical orthopedics, affecting a significant percentage of individuals aged 50 and above. The goal of this study was to comprehensively understand the relationships between a specialized dietary regimen and the risk of developing osteoporosis.

Methods

This study employed extensive genome-wide association study (GWAS) summary statistics derived from the UK Biobank. It encompassed 8 kinds of special diets and 7 datasets pertaining to osteoporosis and associated symptoms. The principal analytical approach employed was the inverse-variance weighted method. Additionally, sensitivity analysis was employed to elucidate the diverse multiplicity patterns observed in the final model.

Results

Our results showed that there is significant evidence that a gluten-free diet is associated with osteoporosis [odds ratio (OR): 1.080, 95% confidence interval (CI): 1.048-1.112, p = 4.23E-07)]. Furthermore, there exists a suggestive link between the three distinct dietary approaches and osteoporosis [(OR: 0.949, 95%CI: 0.929-0.970, p = 3.00E-06) for comprehensive consumption; (OR: 1.053, 95%CI: 1.018-1.089, p = 2.23E-03) for abstaining from wheat consumption; (OR: 1.036, 95%CI: 1.005-1.068, p = 1.97E-02) for abstaining from sugar consumption]. No additional correlation between the special dietary regimens and osteoporosis has been observed.

Conclusion

Our research has uncovered a notable correlation between a gluten-free diet and the occurrence of osteoporosis. Furthermore, it exerts a promoting influence on the onset of osteoporosis, which stands in direct contradiction to the therapeutic principles for Celiac Disease's complications. As such, a novel association among these three elements is postulated.

Influence of body mass index on health complains and life satisfaction

PURPOSE

This research work investigates the influence of children's weight status on well-being and school context in a sample of Spanish adolescences.

METHODS

The Spanish records from the 2013-14 Health Behaviour in School-Aged Children Survey are used, which gathers 9,565 adolescences aged 11, 13 and 15. Studies do not usually address the endogeneity of body mass index when analysing their effect on life satisfaction and health complaints, thus resulting in biased estimates. Considering the endogeneity of body mass index, we use the frequency of alcohol consumption as an instrumental variable in order to obtain consistent estimates of its influence.

RESULTS

The two-stage least squares estimation shows that children's body mass index has a significant negative influence on health complaints and it conditions the way children relate to each other at school. Likewise, results report significant influence on children's subjective well-being and their self-assessment of general health.

CONCLUSIONS

The results of this study provide compelling evidence that BMI plays a crucial role in shaping adolescents' well-being and their interactions with peers at school. These findings underscore the importance of addressing childhood overweight and promoting healthy body mass index levels. Furthermore, the study highlights the need for targeted policy interventions to combat the social stigma associated with being overweight, fostering a more inclusive and supportive school environment for all students.

Addressing unmeasured confounders in cohort studies: Instrumental variable method for a time-fixed exposure on an outcome trajectory

Instrumental variable methods, which handle unmeasured confounding by targeting the part of the exposure explained by an exogenous variable not subject to confounding, have gained much interest in observational studies. We consider the very frequent setting of estimating the unconfounded effect of an exposure measured at baseline on the subsequent trajectory of an outcome repeatedly measured over time. We didactically explain how to apply the instrumental variable method in such setting by adapting the two-stage classical methodology with (1) the prediction of the exposure according to the instrumental variable, (2) its inclusion into a mixed model to quantify the exposure association with the subsequent outcome trajectory, and (3) the computation of the estimated total variance. A simulation study illustrates the consequences of unmeasured confounding in classical analyses and the usefulness of the instrumental variable approach. The methodology is then applied to 6224 participants of the 3C cohort to estimate the association of type-2 diabetes with subsequent cognitive trajectory, using 42 genetic polymorphisms as instrumental variables. This contribution shows how to handle endogeneity when interested in repeated outcomes, along with a R implementation. However, it should still be used with caution as it relies on instrumental variable assumptions hardly testable in practice.

Effect of Duration of Education on Sexual Activity and the Mediating Role of Illness in Later Life: A Natural Experiment in English Schooling Reform

This study aimed to examine the effect of increases in the duration of education on sexual activity in later life using the instrumental variable approach. Cross-sectional data were obtained from the English Longitudinal Study of Ageing, Wave 6 (2012/2013) for the analysis. The effect of the duration of education on a binary variable for sexual activity in the past month was examined by two-stage least squares estimation using the 1947 schooling reform as the instrument. A total of 1493 participants were included. The 1947 schooling reform significantly extended the duration of education by a mean of 0.86 years (95% confidence interval (CI), 0.52-1.20; F = 24.70) for men and 0.81 years for women (95% CI, 0.54-1.09; F = 33.82). The second-stage regression showed that an additional year of education increased the probability of having sexual activity in the past month by 0.16 points (95% CI, 0.07-0.25) for men and decreased the probability of having sexual activity in the past month by 0.18 points (95% CI, - 0.30- - 0.05) for women. The causal mediation analysis revealed that long-standing illness mediated the mechanism between education and sexual activity, but the mediational effect was clearly observed only for men: the indirect effect for men was 0.10 (95% CI, 0.01-0.20; the proportion mediated, 78.1%) and for women was - 0.16 (95% CI, - 1.23-0.90; the proportion mediated, 97.8%), respectively. This study confirmed the causal effect of education on sexual activity. However, the effect for men and women was completely opposite.

Attributable mortality benefit of digoxin treatment in hypoplastic left heart syndrome after the Norwood operation: An instrumental variable-based analysis using data from the Pediatric Health Information Systems Database

BACKGROUND

Observational studies have demonstrated an association between the use of digoxin and reduced interstage mortality after Norwood operation for hypoplastic left heart syndrome (HLHS). Digoxin use has increased significantly but remains variable between different hospitals, independent of case-mix. Instrumental variable analyses have the potential to overcome unmeasured confounding, the major limitation of previous observational studies and to generate an estimate of the attributable benefit of treatment with digoxin.

METHODS

A cohort of neonates with HLHS born from January 1, 2007 to December 31, 2021 who underwent Norwood operation at Pediatric Health Information Systems Database hospitals and survived >14 days after operation were studied. Using hospital-specific, 6-month likelihood of administering digoxin as an instrumental variable, analyses adjusting for both unmeasured confounding (using the instrumental variable) and measured confounders with multivariable logistic regression were performed.

RESULTS

The study population included 5,148 subjects treated at 47 hospitals of which 63% were male and 46% non-Hispanic white. Of these, 44% (n = 2,184) were prescribed digoxin. Treatment with digoxin was associated with superior 1-year transplant-free survival in unadjusted analyses (85% vs 82%, P = .02). This survival benefit persisted in an instrumental-variable analysis (OR: 0.71, 95% CI: 0.54-0.94, P = .01), which can be converted to an absolute risk reduction of 5% (number needed to treat of 20).

CONCLUSIONS

In this observational study of patients with HLHS after Norwood using instrumental variable techniques, a significant benefit in 1-year transplant-free survival attributable to digoxin was demonstrated. In the absence of clinical trial data, this should encourage the use of digoxin in this vulnerable population.

Opportunities and challenges in using instrumental variables to study causal effects in nonrandomized stress and trauma research

OBJECTIVE

Researchers are often interested in assessing the causal effect of an exposure on an outcome when randomization is not ethical or feasible. Estimating causal effects by controlling for confounders can be unconvincing because important potential confounders remain unmeasured. Study designs leveraging instrumental variables (IVs) offer alternatives to confounder-control methods but are rarely used in stress and trauma research.

METHOD

We review the conceptual foundations and implementation of IV methods. We discuss strengths and limitations of IV approaches, contrasting with confounder-control methods, and illustrate the relevance of IVs for stress and trauma research.

RESULTS

IV approaches leverage an external or exogenous source of variation in the exposure. Instruments are variables that meet three conditions: relevance (variation in the IV is associated with variation in the chance of exposure), exclusion (the IV only affects the outcome through the exposure), and exchangeability (no unmeasured confounding of the IV-outcome relationship). Interpreting estimates from IV analyses requires an additional assumption, such as monotonicity (the instrument does not change the chance of exposure in different directions for any two individuals). Valid IVs circumvent the need to correctly identify, measure, and control for all confounders of the exposure-outcome relationship. The primary challenge is identifying a valid instrument.

CONCLUSIONS

IV approaches have strengths and weaknesses compared with confounder-control approaches. IVs offers a promising complementary study design to improve evidence about the causal effects of exposures on outcomes relevant to stress and trauma. Collaboration with scientists who are experienced with identifying and analyzing IVs will support this work. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Comparative effectiveness of decompressive craniectomy versus craniotomy for traumatic acute subdural hematoma (CENTER-TBI): an observational cohort study

Background

Limited evidence existed on the comparative effectiveness of decompressive craniectomy (DC) versus craniotomy for evacuation of traumatic acute subdural hematoma (ASDH) until the recently published randomised clinical trial RESCUE-ASDH. In this study, that ran concurrently, we aimed to determine current practice patterns and compare outcomes of primary DC versus craniotomy.

Methods

We conducted an analysis of centre treatment preference within the prospective, multicentre, observational Collaborative European NeuroTrauma Effectiveness Research in Traumatic Brain Injury (known as CENTER-TBI) and NeuroTraumatology Quality Registry (known as Net-QuRe) studies, which enrolled patients throughout Europe and Israel (2014-2020). We included patients with an ASDH who underwent acute neurosurgical evacuation. Patients with severe pre-existing neurological disorders were excluded. In an instrumental variable analysis, we compared outcomes between centres according to treatment preference, measured by the case-mix adjusted proportion DC per centre. The primary outcome was functional outcome rated by the 6-months Glasgow Outcome Scale Extended, estimated with ordinal regression as a common odds ratio (OR), adjusted for prespecified confounders. Variation in centre preference was quantified with the median odds ratio (MOR). CENTER-TBI is registered with ClinicalTrials.gov, number NCT02210221, and the Resource Identification Portal (Research Resource Identifier SCR_015582).

Findings

Between December 19, 2014 and December 17, 2017, 4559 patients with traumatic brain injury were enrolled in CENTER-TBI of whom 336 (7%) underwent acute surgery for ASDH evacuation; 91 (27%) underwent DC and 245 (63%) craniotomy. The proportion primary DC within total acute surgery cases ranged from 6 to 67% with an interquartile range (IQR) of 12-26% among 46 centres; the odds of receiving a DC for prognostically similar patients in one centre versus another randomly selected centre were trebled (adjusted median odds ratio 2.7, p < 0.0001). Higher centre preference for DC over craniotomy was not associated with better functional outcome (adjusted common odds ratio (OR) per 14% [IQR increase] more DC in a centre = 0.9 [95% CI 0.7-1.1], n = 200). Primary DC was associated with more follow-on surgeries and complications [secondary cranial surgery 27% vs. 18%; shunts 11 vs. 5%]; and similar odds of in-hospital mortality (adjusted OR per 14% IQR more primary DC 1.3 [95% CI (1.0-3.4), n = 200]).

Interpretation

We found substantial practice variation in the employment of DC over craniotomy for ASDH. This variation in treatment strategy did not result in different functional outcome. These findings suggest that primary DC should be restricted to salvageable patients in whom immediate replacement of the bone flap is not possible due to intraoperative brain swelling.

Funding

Hersenstichting Nederland for the Dutch NeuroTraumatology Quality Registry and the European Union Seventh Framework Program.

Comparison of intergenerational instrumental variable analyses of body mass index and mortality in UK Biobank

BACKGROUND

An increasing proportion of people have a body mass index (BMI) classified as overweight or obese and published studies disagree whether this will be beneficial or detrimental to health. We applied and evaluated two intergenerational instrumental variable methods to estimate the average causal effect of BMI on mortality in a cohort with many deaths: the parents of UK Biobank participants.

METHODS

In Cox regression models, parental BMI was instrumented by offspring BMI using an 'offspring as instrument' (OAI) estimation and by offspring BMI-related genetic variants in a 'proxy-genotype Mendelian randomization' (PGMR) estimation.

RESULTS

Complete-case analyses were performed in parents of 233 361 UK Biobank participants with full phenotypic, genotypic and covariate data. The PGMR method suggested that higher BMI increased mortality with hazard ratios per kg/m2 of 1.02 (95% CI: 1.01, 1.04) for mothers and 1.04 (95% CI: 1.02, 1.05) for fathers. The OAI method gave considerably higher estimates, which varied according to the parent-offspring pairing between 1.08 (95% CI: 1.06, 1.10; mother-son) and 1.23 (95% CI: 1.16, 1.29; father-daughter).

CONCLUSION

Both methods supported a causal role of higher BMI increasing mortality, although caution is required regarding the immediate causal interpretation of these exact values. Evidence of instrument invalidity from measured covariates was limited for the OAI method and minimal for the PGMR method. The methods are complementary for interrogating the average putative causal effects because the biases are expected to differ between them.

Associations between diet and incidence risk of lung cancer: A Mendelian randomization study

Background

Observational studies have revealed associations between diet and lung cancer. However, it is unclear whether the association is disturbed by confounding factors. We used a two-sample Mendelian randomization (MR) method to characterize the associations between diet and the lung cancer risk (including 3 subtypes: lung adenocarcinoma (LA), squamous cell lung carcinoma (SqCLC), and small cell lung cancer (SCLC)).

Materials and methods

Data on 20 diets were screened from the UK Biobank. Lung cancer data came from a large meta-analysis of 85,716 individuals. The inverse-variance weighted method was used as the main analysis. Sensitivity analysis was also used to explain the different multiplicity patterns of the final model.

Results

Our results showed significant evidence that 3 diets were associated with lung cancer [odds ratio (OR): 0.271, 95% confidence interval (CI): 0.150–0.488, p = 1.46 × 10−4, dried fruit; OR: 3.010, 95% CI: 1.608–5.632, p = 5.70 × 10−4, beer] and SqCLC (OR: 0.135, 95% CI: 0.062–0.293, p = 2.33 × 10−5, dried fruit; OR: 0.485, 95% CI: 0.328–0.717, p = 2.9 × 10−4, cheese). There were also suggestive correlations between 5 dietary intakes and lung cancer (OR: 0.441, 95% CI: 0.250–0.778, p = 0.008, cereal; OR: 2.267, 95% CI: 1.126–4.564, p = 0.022, beef), LA (OR: 0.494, 95% CI: 0.285–0.858, p = 0.012, dried fruit; OR: 3.536, 95% CI: 1.546–8.085, p = 0.003, beer) and SCLC (OR: 0.006, 95% CI: 0.000–0.222, p = 0.039, non-oily fish; OR: 0.239, 95% CI: 0.086–0.664, p = 0.006, dried fruit). No other association between diet and lung cancer was observed.

Conclusion

Our study preliminary found that cheese, dried fruit, and beer intake were significantly associated with the risk of lung cancer or its subtypes, while cereal, beef, and non-oily fish intake were suggestively associated with the risk of lung cancer or its subtypes. Well-designed prospective studies are still needed to confirm our findings in the future.

Duration of Education and Back Pain: Lessons From English Schooling Reforms

This study aimed to examine the associations of increases in the duration of education with back pain using the exogenous variation generated by the English schooling reforms of 1947 and 1972. We analyzed cross-sectional data derived from 9 waves (waves 1-9; 2002-2019) of the English Longitudinal Study of Ageing. An instrumental variables regression using 2-stage least squares with the 2-way cluster-robust standard error was used. The mean severity of back pain, measured using the Numerical Rating Scale, was used as the outcome. A total of 22,868 observations from 5,070 participants were included (the 1947 reform = 16,565 observations from 3,231 participants, mean age = 74.5 years; the 1972 reform = 6,303 observations from 1,839 participants, mean age = 59.3 years). The schooling reforms significantly extended years of school attendance by a mean of 0.57 years for the 1942 reform cohort and 0.66 years for 1972 reform cohort. For participants born within 5 years of the pivotal cohorts, an additional year of education decreased the severity of back pain by 0.78 points (95% confidence interval: 0.65, 0.92) for the 1972 reform cohort. Our finding underscores the importance of the length of education in the reduction of back pain in middle-aged individuals.

Circulating Isovalerylcarnitine and Lung Cancer Risk: Evidence from Mendelian Randomization and Prediagnostic Blood Measurements

BACKGROUND

Tobacco exposure causes 8 of 10 lung cancers, and identifying additional risk factors is challenging due to confounding introduced by smoking in traditional observational studies.

MATERIALS AND METHODS

We used Mendelian randomization (MR) to screen 207 metabolites for their role in lung cancer predisposition using independent genome-wide association studies (GWAS) of blood metabolite levels (n = 7,824) and lung cancer risk (n = 29,266 cases/56,450 controls). A nested case-control study (656 cases and 1,296 matched controls) was subsequently performed using prediagnostic blood samples to validate MR association with lung cancer incidence data from population-based cohorts (EPIC and NSHDS).

RESULTS

An MR-based scan of 207 circulating metabolites for lung cancer risk identified that blood isovalerylcarnitine (IVC) was associated with a decreased odds of lung cancer after accounting for multiple testing (log10-OR = 0.43; 95% CI, 0.29-0.63). Molar measurement of IVC in prediagnostic blood found similar results (log10-OR = 0.39; 95% CI, 0.21-0.72). Results were consistent across lung cancer subtypes.

CONCLUSIONS

Independent lines of evidence support an inverse association of elevated circulating IVC with lung cancer risk through a novel methodologic approach that integrates genetic and traditional epidemiology to efficiently identify novel cancer biomarkers.

IMPACT

Our results find compelling evidence in favor of a protective role for a circulating metabolite, IVC, in lung cancer etiology. From the treatment of a Mendelian disease, isovaleric acidemia, we know that circulating IVC is modifiable through a restricted protein diet or glycine and L-carnatine supplementation. IVC may represent a modifiable and inversely associated biomarker for lung cancer.

Exploring and mitigating potential bias when genetic instrumental variables are associated with multiple non-exposure traits in Mendelian randomization

With the increasing size and number of genome-wide association studies, individual single nucleotide polymorphisms are increasingly found to associate with multiple traits. Many different mechanisms could result in proposed genetic IVs for an exposure of interest being associated with multiple non-exposure traits, some of which could bias MR results. We describe and illustrate, through causal diagrams, a range of scenarios that could result in proposed IVs being related to non-exposure traits in MR studies. These associations could occur due to five scenarios: (i) confounding, (ii) vertical pleiotropy, (iii) horizontal pleiotropy, (iv) reverse causation and (v) selection bias. For each of these scenarios we outline steps that could be taken to explore the underlying mechanism and mitigate any resulting bias in the MR estimation. We recommend MR studies explore possible IV-non-exposure associations across a wider range of traits than is usually the case. We highlight the pros and cons of relying on sensitivity analyses without considering particular pleiotropic paths versus systematically exploring and controlling for potential pleiotropic or other biasing paths via known traits. We apply our recommendations to an illustrative example of the effect of maternal insomnia on offspring birthweight in UK Biobank.

Understanding the assumptions underlying Mendelian randomization

With the rapidly increasing availability of large genetic data sets in recent years, Mendelian Randomization (MR) has quickly gained popularity as a novel secondary analysis method. Leveraging genetic variants as instrumental variables, MR can be used to estimate the causal effects of one phenotype on another even when experimental research is not feasible, and therefore has the potential to be highly informative. It is dependent on strong assumptions however, often producing biased results if these are not met. It is therefore imperative that these assumptions are well-understood by researchers aiming to use MR, in order to evaluate their validity in the context of their analyses and data. The aim of this perspective is therefore to further elucidate these assumptions and the role they play in MR, as well as how different kinds of data can be used to further support them.

Mendelian Randomization: Concepts and Scope

Mendelian randomization (MR) is a method of studying the causal effects of modifiable exposures (i.e., potential risk factors) on health, social, and economic outcomes using genetic variants associated with the specific exposures of interest. MR provides a more robust understanding of the influence of these exposures on outcomes because germline genetic variants are randomly inherited from parents to offspring and, as a result, should not be related to potential confounding factors that influence exposure-outcome associations. The genetic variant can therefore be used as a tool to link the proposed risk factor and outcome, and to estimate this effect with less confounding and bias than conventional epidemiological approaches. We describe the scope of MR, highlighting the range of applications being made possible as genetic data sets and resources become larger and more freely available. We outline the MR approach in detail, covering concepts, assumptions, and estimation methods. We cover some common misconceptions, provide strategies for overcoming violation of assumptions, and discuss future prospects for extending the clinical applicability, methodological innovations, robustness, and generalizability of MR findings.

Contrasting approaches for addressing non-adherence in randomized controlled trials: An illustration from the REFLUX trial

BACKGROUND

In many randomized controlled trials, a substantial proportion of patients do not comply with the treatment protocol to which they have been randomly assigned. Randomized controlled trials are required to report results according to the intention-to-treat estimand, but recent methodological guidance recognizes the importance of estimating other causal quantities.

METHODS

This article outlines an analytical framework for randomized controlled trials with non-compliance. We apply the ICH E9 (R1) addendum and combine it with the potential outcomes framework to define key estimands, outline the major assumptions for identification of each estimand, and highlight the assumptions that cannot be verified from the randomized controlled trial data. We contrast the assumptions and estimates in a re-analysis of the REFLUX trial. We report alternative estimates for the effectiveness of receipt of laparoscopic surgery versus medical management for patients with gastro-intestinal reflux disease.

RESULTS

The article finds that adjusted as-treated and per-protocol estimates were similar in magnitude to those based intention-to-treat methods. Instrumental variable estimates of the complier average causal effect were larger, with wider confidence intervals.

CONCLUSION

We recommend that in randomized controlled trials with non-compliance, studies should outline which estimand is most relevant to the study context, evaluate key assumptions, and present estimates from a range of methods as a sensitivity analysis.

Instrumental variable analysis using offspring BMI in childhood as an indicator of parental BMI in relation to mortality

Childhood BMI shows associations with adult mortality, but these may be influenced by effects of ill health in childhood on BMI and later mortality. To avoid this, we used offspring childhood BMI as an instrumental variable (IV) for own BMI in relation to mortality and compared it with conventional associations of own childhood BMI and own mortality. We included 36,097 parent-offspring pairs with measured heights and weights from the Copenhagen School Health Records Register and register-based information on death. Hazard ratios (HR) were estimated using adjusted Cox regression models. For all-cause mortality, per zBMI at age 7 the conventional HR = 1.07 (95%CI: 1.04-1.09) in women and 1.02 (95%CI: 0.92-1.14) in men, whereas the IV HR = 1.23 (95%CI: 1.15-1.32) in women and 1.05 (95%CI: 0.94-1.17) in men. Per zBMI at age 13, the conventional HR = 1.11 (95%CI: 1.08-1.15) in women and 1.03 (95%CI: 0.99-1.06) in men, whereas the IV HR = 1.30 (95%CI: 1.19-1.42) in women and 1.15 (95%CI: 1.04-1.29) in men. Only conventional models showed indications of J-shaped associations. Our IV analyses suggest that there is a causal relationship between BMI and mortality that is positive at both high and low BMI values.

Genetically increased circulating FUT3 level leads to reduced risk of Idiopathic Pulmonary Fibrosis: a Mendelian Randomisation Study

Idiopathic pulmonary fibrosis (IPF) is a progressive, fatal fibrotic interstitial lung disease. Few circulating biomarkers have been identified to have causal effects on IPF.To identify candidate IPF-influencing circulating proteins, we undertook an efficient screen of circulating proteins by applying a two-sample Mendelian randomisation (MR) approach with existing publicly available data. For instruments we used genetic determinants of circulating proteins which reside cis to the encoded gene (cis-SNPs), identified by two genome-wide association studies (GWASs) in European individuals (3301 and 3200 subjects). We then applied MR methods to test if the levels of these circulating proteins influenced IPF susceptibility in the largest IPF GWAS (2668 cases and 8591 controls). We validated the MR results using colocalization analyses to ensure that both the circulating proteins and IPF shared a common genetic signal.MR analyses of 834 proteins found that a one sd increase in circulating FUT3 and FUT5 was associated with a reduced risk of IPF (OR: 0.81, 95%CI: 0.74-0.88, p=6.3×10^-7, and OR: 0.76, 95%CI: 0.68-0.86, p=1.1×10^-5). Sensitivity analyses including multiple-cis SNPs provided similar estimates both for FUT3 (inverse variance weighted [IVW] OR: 0.84, 95%CI: 0.78-0.91, p=9.8×10^-6, MR-Egger OR: 0.69, 95%CI: 0.50-0.97, p=0.03) and FUT5 (IVW OR: 0.84, 95%CI: 0.77-0.92, p=1.4×10^-4, MR-Egger OR: 0.59, 95%CI: 0.38-0.90, p=0.01) FUT3 and FUT5 signals colocalized with IPF signals, with posterior probabilities of a shared genetic signal of 99.9% and 97.7%. Further transcriptomic investigations supported the protective effects of FUT3 for IPF.An efficient MR scan of 834 circulating proteins provided evidence that genetically increased circulating FUT3 level is associated with reduced risk of IPF.

Preference-based instrumental variables in health research rely on important and underreported assumptions: a systematic review

OBJECTIVE

Preference-based instrumental variables (PP IV) designs can identify causal effects when patients receive treatment due to variation in providers' treatment preference. We offer a systematic review and methodological assessment of PP IV applications in health research.

STUDY DESIGN AND SETTING

We included studies that applied PP IV for evaluation of any treatment in any population in health research (PROSPERO: CRD42020165014). We searched within four databases (Medline, Web of Science, ScienceDirect, SpringerLink) and four journals (including full-text and title and abstract sources) between January 1, 1998, and March 5, 2020. We extracted data on areas of applications and methodology, including assumptions using Swanson and Hernan's (2013) guideline.

RESULTS

We included 185 of 1087 identified studies. The use of PP IV has increased, being predominantly used for treatment effects in cancer, cardiovascular disease, and mental health. The most common PP IV was treatment variation at the facility-level, followed by physician- and regional-level. Only 12 percent of applications report the four main assumptions for PP IV. Selection on treatment may be a potential issue in 46 percent of studies.

CONCLUSION

The assumptions of PP IV are not sufficiently reported in existing work. PP IV-studies should use reporting guidelines.

Mendelian Randomization in Cardiovascular Research: Establishing Causality When There Are Unmeasured Confounders

Mendelian randomization is an epidemiological approach to making causal inferences using observational data. It makes use of the natural randomization that occurs in the generation of an individual's genetic makeup in a way that is analogous to the study design of a randomized controlled trial and uses instrumental variable analysis where the genetic variant(s) are the instrument (analogous to random allocation to treatment group in an randomized controlled trial). As with any instrumental variable, there are 3 assumptions that must be made about the genetic instrument: (1) it is associated (not necessarily causally) with the exposure (relevance condition); (2) it is associated with the outcome only through the exposure (exclusion restriction condition); and (3) it does not share a common cause with the outcome (ie, no confounders of the genetic instrument and outcome, independence condition). Using the example of type II diabetes and coronary artery disease, we demonstrate how the method may be used to investigate causality and discuss potential benefits and pitfalls. We conclude that although Mendelian randomization studies can usually not establish causality on their own, they may usefully contribute to the evidence base and increase our certainty about the effectiveness (or otherwise) of interventions to reduce cardiovascular disease.

Formulating causal questions and principled statistical answers

Although review papers on causal inference methods are now available, there is a lack of introductory overviews on what they can render and on the guiding criteria for choosing one particular method. This tutorial gives an overview in situations where an exposure of interest is set at a chosen baseline ("point exposure") and the target outcome arises at a later time point. We first phrase relevant causal questions and make a case for being specific about the possible exposure levels involved and the populations for which the question is relevant. Using the potential outcomes framework, we describe principled definitions of causal effects and of estimation approaches classified according to whether they invoke the no unmeasured confounding assumption (including outcome regression and propensity score-based methods) or an instrumental variable with added assumptions. We mainly focus on continuous outcomes and causal average treatment effects. We discuss interpretation, challenges, and potential pitfalls and illustrate application using a "simulation learner," that mimics the effect of various breastfeeding interventions on a child's later development. This involves a typical simulation component with generated exposure, covariate, and outcome data inspired by a randomized intervention study. The simulation learner further generates various (linked) exposure types with a set of possible values per observation unit, from which observed as well as potential outcome data are generated. It thus provides true values of several causal effects. R code for data generation and analysis is available on www.ofcaus.org, where SAS and Stata code for analysis is also provided.

Does Transfer to Intensive Care Units Reduce Mortality? A Comparison of an Instrumental Variables Design to Risk Adjustment

BACKGROUND

Instrumental variable (IV) analysis can estimate treatment effects in the presence of residual or unmeasured confounding. In settings wherein measures of baseline risk severity are unavailable, IV designs are, therefore, particularly appealing, but, where established measures of risk severity are available, it is unclear whether IV methods are preferable.

OBJECTIVE

We compared regression with an IV design to estimate the effect of intensive care unit (ICU) transfer on mortality in a study with well-established measures of risk severity.

RESEARCH DESIGN

We use ICU bed availability at the time of assessment for ICU transfer as an instrument. Bed availability increases the chance of ICU admission, contains little information about patient characteristics, and it is unlikely that bed availability has any direct effect on in-hospital mortality.

SUBJECTS

We used a cohort study of deteriorating ward patients assessed for critical care unit admission, in 49 UK National Health Service hospitals between November 1, 2010, and December 31, 2011.

MEASURES

Detailed demographic, physiological, and comorbidity data were collected for all patients.

RESULTS

The risk adjustment methods reported that, after controlling for all measured covariates including measures of risk severity, ICU transfer was associated with higher 28-day mortality, with a risk difference of 7.2% (95% confidence interval=5.3%-9.1%). The IV estimate of ICU transfer was -5.4% (95% confidence interval=-47.1% to 36.3%) and applies to the subsample of patients whose transfer was "encouraged" by bed availability.

CONCLUSIONS

IV estimates indicate that ICU care is beneficial but are imprecisely estimated. Risk-adjusted estimates are more precise but, even with a rich set of covariates, report that ICU care is harmful.

Assessing causal treatment effect estimation when using large observational datasets

Background

Recently, there has been a heightened interest in developing and evaluating different methods for analysing observational data. This has been driven by the increased availability of large data resources such as Electronic Health Record (EHR) data alongside known limitations and changing characteristics of randomised controlled trials (RCTs). A wide range of methods are available for analysing observational data. However, various, sometimes strict, and often unverifiable assumptions must be made in order for the resulting effect estimates to have a causal interpretation. In this paper we will compare some common approaches to estimating treatment effects from observational data in order to highlight the importance of considering, and justifying, the relevant assumptions prior to conducting an observational analysis.

Methods

A simulation study was conducted based upon a small cohort of patients with chronic obstructive pulmonary disease. Two-stage least squares instrumental variables, propensity score, and linear regression models were compared under a range of different scenarios including different strengths of instrumental variable and unmeasured confounding. The effects of violating the assumptions of the instrumental variables analysis were also assessed. Sample sizes of up to 200,000 patients were considered.

Results

Two-stage least squares instrumental variable methods can yield unbiased treatment effect estimates in the presence of unmeasured confounding provided the sample size is sufficiently large. Adjusting for measured covariates in the analysis reduces the variability in the two-stage least squares estimates. In the simulation study, propensity score methods produced very similar results to linear regression for all scenarios. A weak instrument or strong unmeasured confounding led to an increase in uncertainty in the two-stage least squares instrumental variable effect estimates. A violation of the instrumental variable assumptions led to bias in the two-stage least squares effect estimates. Indeed, these were sometimes even more biased than those from a naïve linear regression model.

Conclusions

Instrumental variable methods can perform better than naïve regression and propensity scores. However, the assumptions need to be carefully considered and justified prior to conducting an analysis or performance may be worse than if the problem of unmeasured confounding had been ignored altogether.

Methodological Challenges When Studying Distance to Care as an Exposure in Health Research

Distance to care is a common exposure and proposed instrumental variable in health research, but it is vulnerable to violations of fundamental identifiability conditions for causal inference. We used data collected from the Botswana Birth Outcomes Surveillance study between 2014 and 2016 to outline 4 challenges and potential biases when using distance to care as an exposure and as a proposed instrument: selection bias, unmeasured confounding, lack of sufficiently well-defined interventions, and measurement error. We describe how these issues can arise, and we propose sensitivity analyses for estimating the degree of bias.

Causal Effect of Sleep Duration on Body Weight in Adolescents: A Population-based Study Using a Natural Experiment

Supplemental Digital Content is available in the text.

A large number of observational epidemiologic studies have reported consistent associations between short sleep duration and increased body weight, particularly in children and adolescents. Causal evidence on the effect of sleep duration on body weight is still limited, however.

Epidemiology, genetic epidemiology and Mendelian randomisation: more need than ever to attend to detail

In the current era, with increasing availability of results from genetic association studies, finding genetic instruments for inferring causality in observational epidemiology has become apparently simple. Mendelian randomisation (MR) analyses are hence growing in popularity and, in particular, methods that can incorporate multiple instruments are being rapidly developed for these applications. Such analyses have enormous potential, but they all rely on strong, different, and inherently untestable assumptions. These have to be clearly stated and carefully justified for every application in order to avoid conclusions that cannot be replicated. In this article, we review the instrumental variable assumptions and discuss the popular linear additive structural model. We advocate the use of tests for the null hypothesis of ‘no causal effect’ and calculation of the bounds for a causal effect, whenever possible, as these do not rely on parametric modelling assumptions. We clarify the difference between a randomised trial and an MR study and we comment on the importance of validating instruments, especially when considering them for joint use in an analysis. We urge researchers to stand by their convictions, if satisfied that the relevant assumptions hold, and to interpret their results causally since that is the only reason for performing an MR analysis in the first place.

Estimating cluster-level local average treatment effects in cluster randomised trials with non-adherence

Non-adherence to assigned treatment is a common issue in cluster randomised trials. In these settings, the efficacy estimand may also be of interest. Many methodological contributions in recent years have advocated using instrumental variables to identify and estimate the local average treatment effect. However, the clustered nature of randomisation in cluster randomised trials adds to the complexity of such analyses. In this paper, we show that the local average treatment effect can be estimated via two-stage least squares regression using cluster-level summaries of the outcome and treatment received under certain assumptions. We propose the use of baseline variables to adjust the cluster-level summaries before performing two-stage least squares in order to improve efficiency. Implementation needs to account for the reduced sample size, as well as the possible heteroscedasticity, to obtain valid inferences. Simulations are used to assess the performance of two-stage least squares of cluster-level summaries under cluster-level or individual-level non-adherence, with and without weighting and robust standard errors. The impact of adjusting for baseline covariates and of appropriate degrees of freedom correction for inference is also explored. The methods are then illustrated by re-analysing a cluster randomised trial carried out in a specific UK primary care setting. Two-stage least squares estimation using cluster-level summaries provides estimates with small to negligible bias and coverage close to nominal level, provided the appropriate small sample degrees of freedom correction and robust standard errors are used for inference.

Reporting of Baseline Characteristics to Accompany Analysis by Instrumental Variables

A property and criticism of instrumental variable (IV) estimators is that they estimate the exposure effect in the compliers, individuals whose exposure is influenced by the instrument. It is conceivable that the exposure effect is different in individuals whose exposure is not influenced by the instrument. For that reason, it is useful to know who the compliers are. We present methods for reporting averages and other statistics and comparing them between the compliers and noncompliers, which are applicable to scenarios in which the instrument and exposure are fixed in time and dichotomous. The methods are illustrated in a comparison of outcomes between teaching and nonteaching hospitals in patients undergoing brain or spinal surgery.

Evaluating the potential role of pleiotropy in Mendelian randomization studies

Pleiotropy, the phenomenon of a single genetic variant influencing multiple traits, is likely widespread in the human genome. If pleiotropy arises because the single nucleotide polymorphism (SNP) influences one trait, which in turn influences another ('vertical pleiotropy'), then Mendelian randomization (MR) can be used to estimate the causal influence between the traits. Of prime focus among the many limitations to MR is the unprovable assumption that apparent pleiotropic associations are mediated by the exposure (i.e. reflect vertical pleiotropy), and do not arise due to SNPs influencing the two traits through independent pathways ('horizontal pleiotropy'). The burgeoning treasure trove of genetic associations yielded through genome wide association studies makes for a tantalizing prospect of phenome-wide causal inference. Recent years have seen substantial attention devoted to the problem of horizontal pleiotropy, and in this review we outline how newly developed methods can be used together to improve the reliability of MR.

Understanding the Assumptions Underlying Instrumental Variable Analyses: a Brief Review of Falsification Strategies and Related Tools

PURPOSE OF REVIEW

Instrumental variable (IV) methods continue to be applied to questions ranging from genetic to social epidemiology. In the epidemiologic literature, discussion of whether the assumptions underlying IV analyses hold is often limited to only certain assumptions and even then, arguments are mostly made using subject matter knowledge. To complement subject matter knowledge, there exist a variety of falsification strategies and other tools for weighing the plausibility of the assumptions underlying IV analyses.

RECENT FINDINGS

There are many tools that can refute the IV assumptions or help estimate the magnitude or direction of possible bias if the conditions do not hold perfectly. Many of these tools, including both recently developed strategies and strategies described decades ago, are underused or only used in specific applications of IV methods in epidemiology.

SUMMARY

Although estimating causal effects with IV analyses relies on unverifiable assumptions, the assumptions can sometimes be refuted. We suggest that the epidemiologists using IV analyses employ all the falsification strategies that apply to their research question in order to avoid settings that demonstrably violate a core condition for valid inference.

Nature as a Trialist?: Deconstructing the Analogy Between Mendelian Randomization and Randomized Trials

Mendelian randomization (MR) studies are often described as naturally occurring randomized trials in which genetic factors are randomly assigned by nature. Conceptualizing MR studies as randomized trials has profound implications for their design, conduct, reporting, and interpretation. For example, analytic practices that are discouraged in randomized trials should also be discouraged in MR studies. Here, we deconstruct the oft-made analogy between MR and randomized trials. We describe four key threats to the analogy between MR studies and randomized trials: (1) exchangeability is not guaranteed; (2) time zero (and therefore the time for setting eligibility criteria) is unclear; (3) the treatment assignment is often measured with error; and (4) adherence is poorly defined. By precisely defining the causal effects being estimated, we underscore that MR estimates are often vaguely analogous to per-protocol effects in randomized trials, and that current MR methods for estimating analogues of per-protocol effects could be biased in practice. We conclude that the analogy between randomized trials and MR studies provides further perspective on both the strengths and the limitations of MR studies as currently implemented, as well as future directions for MR methodology development and application. In particular, the analogy highlights potential future directions for some MR studies to produce more interpretable and informative numerical estimates.

Causal null hypotheses of sustained treatment strategies: What can be tested with an instrumental variable?

Sometimes instrumental variable methods are used to test whether a causal effect is null rather than to estimate the magnitude of a causal effect. However, when instrumental variable methods are applied to time-varying exposures, as in many Mendelian randomization studies, it is unclear what causal null hypothesis is tested. Here, we consider different versions of causal null hypotheses for time-varying exposures, show that the instrumental variable conditions alone are insufficient to test some of them, and describe additional assumptions that can be made to test a wider range of causal null hypotheses, including both sharp and average causal null hypotheses. Implications for interpretation and reporting of instrumental variable results are discussed.

Instrumental variable methods for a binary outcome were used to informatively address noncompliance in a randomized trial in surgery

Objectives

Randomization can be used as an instrumental variable (IV) to account for unmeasured confounding when seeking to assess the impact of noncompliance with treatment allocation in a randomized trial. We present and compare different methods to calculate the treatment effect on a binary outcome as a rate ratio in a randomized surgical trial.

Study Design and Setting

The effectiveness of peeling versus not peeling the internal limiting membrane of the retina as part of the surgery for a full thickness macular hole. We compared the IV-based estimates (nonparametric causal bound and two-stage residual inclusion approach [2SRI]) with standard treatment effect measures (intention to treat, per protocol and treatment received [TR]). Compliance was defined in two ways (initial and up to the time point of interest). Poisson regression was used for the model-based approaches with robust standard errors to calculate the risk ratio (RR) with 95% confidence intervals.

Results

Results were similar for 1-month macular hole status across methods. For 3- and 6-month macular hole status, nonparametric causal bounds provided a narrower range of uncertainty than other methods, though still had substantial imprecision. For 3-month macular hole status, the TR estimate was substantially different from the other point estimates.

Conclusion

Nonparametric causal bound approaches are a useful addition to an IV estimation approach, which tend to have large levels of uncertainty. Methods which allow RRs to be calculated when addressing noncompliance in randomized trials exist and may be superior to standard estimates. Further research is needed to explore the properties of different IV methods in a broad range of randomized controlled trial scenarios.

Instrumental Variable Analyses in Pharmacoepidemiology: What Target Trials Do We Emulate?

Purpose of review

When leveraging observational data to estimate treatment effects, it is useful to explicitly specify the “target trial” the investigators aspire to emulate. One concern is whether a proposed analysis plan can address the realities of the differences between the available non-randomized observational study and the target trial. When large or unknown sources of unmeasured confounding are suspected, investigators might consider turning to instrumental variable (IV) methods. Of course, the interpretation and appropriateness of IV analyses need to be considered carefully. The purpose of this review is to summarize recent methodologic advancements in how epidemiologists weigh the validity of an IV analysis and to place these methodologic advancements in the context of the feasible target trial’s protocol components.

Recent findings

There have been increased development and application of tools for sensitivity analyses, falsification strategies, and the identification of previously overlooked problems with IV analyses as applied in pharmacoepidemiology. Many of these recent insights can be seen as articulating restrictions on or tradeoffs between the types of target trials that can be validly emulated when using a classical IV analysis.

Summary

Putting classical IV methods in the context of target trials underscores the importance of recent methodologic developments and, more generally, when and how an IV analysis would be appropriate. We see that some tradeoffs in defining the target trials are unavoidable, that some tradeoffs may be offset or explored via sensitivity analyses, and that this serves as a framework for scientific discourse regarding IV and non-IV results emulating potentially different trials with different tradeoffs.