Dataset Information

Evaluation of Selection Bias in an Internet-based Study of Pregnancy Planners.

ABSTRACT: Selection bias is a potential concern in all epidemiologic studies, but it is usually difficult to assess. Recently, concerns have been raised that internet-based prospective cohort studies may be particularly prone to selection bias. Although use of the internet is efficient and facilitates recruitment of subjects that are otherwise difficult to enroll, any compromise in internal validity would be of great concern. Few studies have evaluated selection bias in internet-based prospective cohort studies. Using data from the Danish Medical Birth Registry from 2008 to 2012, we compared six well-known perinatal associations (e.g., smoking and birth weight) in an internet-based preconception cohort (Snart Gravid n = 4,801) with the total population of singleton live births in the registry (n = 239,791). We used log-binomial models to estimate risk ratios (RRs) and 95% confidence intervals (CIs) for each association. We found that most results in both populations were very similar. For example, maternal obesity was associated with an increased risk of delivering a macrosomic infant in Snart Gravid (RR = 1.5; 95% CI: 1.2, 1.7) and the total population (RR = 1.5; 95% CI: 1.45, 1.53), and maternal smoking of >10 cigarettes per day was associated with a higher risk of low birth weight (RR = 2.7; 95% CI: 1.2, 5.9 vs. RR = 2.9; 95% CI: 2.6, 3.1) in Snart Gravid and the total population, respectively. We cannot be certain that our results would apply to other associations or different populations. Nevertheless, our results suggest that recruitment of reproductive aged women via the internet may be no more prone to selection bias than traditional methods of recruitment.

SUBMITTER: Hatch EE

PROVIDER: S-EPMC4819248 | biostudies-literature | 2016 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Evaluation of Selection Bias in an Internet-based Study of Pregnancy Planners.

Hatch Elizabeth E EE Hahn Kristen A KA Wise Lauren A LA Mikkelsen Ellen M EM Kumar Ramya R Fox Matthew P MP Brooks Daniel R DR Riis Anders H AH Sorensen Henrik Toft HT Rothman Kenneth J KJ

Epidemiology (Cambridge, Mass.) 20160101 1

Selection bias is a potential concern in all epidemiologic studies, but it is usually difficult to assess. Recently, concerns have been raised that internet-based prospective cohort studies may be particularly prone to selection bias. Although use of the internet is efficient and facilitates recruitment of subjects that are otherwise difficult to enroll, any compromise in internal validity would be of great concern. Few studies have evaluated selection bias in internet-based prospective cohort s ...[more]

PMID: 26484423

Similar Datasets

Project description:Study questionIs a history of periodontitis among women associated with reduced fecundability?Summary answerA history of periodontitis, as assessed by three different self-reported measures, may be associated with reduced fecundability.What is known alreadyPeriodontitis is a chronic inflammatory condition affecting the hard and soft tissues surrounding the teeth. Few studies have evaluated the association between periodontitis and time to pregnancy, and findings are mixed. It is hypothesized that periodontitis may adversely affect time to pregnancy.Study design, size, durationWe conducted a prospective cohort study of 2764 female pregnancy planners residing in North America (March 2015-June 2020).Participants/materials, setting, methodsEligible participants had been attempting pregnancy for six or fewer menstrual cycles at enrollment and were not using fertility treatment. Women answered questions about their oral health. Pregnancy was ascertained via bi-monthly follow-up questionnaires. We used proportional probabilities regression models to estimate fecundability ratios (FRs) and 95% confidence intervals (CIs) for three different measures indicative of a history of periodontitis: ever diagnosed with periodontitis (N = 265), ever received treatment for periodontitis (N = 299), and ever had an adult tooth become loose on its own (N = 83). We adjusted for potential confounders and precision variables. Women at risk of misclassification of periodontitis diagnosis due to pregnancy-related gingivitis were reclassified in a sensitivity analysis.Main results and the role of chanceAll three indices of periodontitis may be associated with reduced fecundability. FRs were 0.89 (95% CI 0.75-1.06) comparing women with and without a previous periodontitis diagnosis, 0.79 (95% CI 0.67-0.94) comparing women with and without previous periodontitis treatment, and 0.71 (95% CI 0.44-1.16) comparing women with and without a tooth that became loose. After reclassification of pregnancy-related gingivitis in the sensitivity analysis, the FR for periodontitis diagnosis was 0.83 (95% CI 0.68-1.00). Weaker FRs were observed among parous women as compared with nulliparous women for periodontitis diagnosis and tooth becoming loose, but not for periodontitis treatment.Limitations, reasons for cautionThough we used validated self-report measures of periodontitis, clinical confirmation is the gold standard. These questions may be functioning as markers of different levels of periodontitis severity, but we were unable to measure disease severity in this population. Finally, we cannot eliminate the possibility of unmeasured confounding.Wider implications of the findingsThis is the first preconception prospective cohort study to evaluate the association between self-reported periodontitis and fecundability. Our results indicate that periodontitis may be associated with lower fecundability.Study funding/competing interest(s)This work was partially funded by R01HD086742/Eunice Kennedy Shriver National Institute of Child Health and Human Development and R21HD072326/Eunice Kennedy Shriver National Institute of Child Health and Human Development. PRESTO has received in-kind donations from Swiss Precision Diagnostics, Sandstone Diagnostics, FertilityFriend.com, and Kindara.com for primary data collection. L.A.W. is a fibroid consultant for AbbVie, Inc. J.C.B., S.W., J.Y., K.J.R., E.E.H., and B.H. have no conflicts of interest to disclose.Trial registration numberN/A.

Project description:BackgroundSelection bias and unmeasured confounding are fundamental problems in epidemiology that threaten study internal and external validity. These phenomena are particularly dangerous in internet-based public health surveillance, where traditional mitigation and adjustment methods are inapplicable, unavailable, or out of date. Recent theoretical advances in causal modeling can mitigate these threats, but these innovations have not been widely deployed in the epidemiological community.ObjectiveThe purpose of our paper is to demonstrate the practical utility of causal modeling to both detect unmeasured confounding and selection bias and guide model selection to minimize bias. We implemented this approach in an applied epidemiological study of the COVID-19 cumulative infection rate in the New York City (NYC) spring 2020 epidemic.MethodsWe collected primary data from Qualtrics surveys of Amazon Mechanical Turk (MTurk) crowd workers residing in New Jersey and New York State across 2 sampling periods: April 11-14 and May 8-11, 2020. The surveys queried the subjects on household health status and demographic characteristics. We constructed a set of possible causal models of household infection and survey selection mechanisms and ranked them by compatibility with the collected survey data. The most compatible causal model was then used to estimate the cumulative infection rate in each survey period.ResultsThere were 527 and 513 responses collected for the 2 periods, respectively. Response demographics were highly skewed toward a younger age in both survey periods. Despite the extremely strong relationship between age and COVID-19 symptoms, we recovered minimally biased estimates of the cumulative infection rate using only primary data and the most compatible causal model, with a relative bias of +3.8% and -1.9% from the reported cumulative infection rate for the first and second survey periods, respectively.ConclusionsWe successfully recovered accurate estimates of the cumulative infection rate from an internet-based crowdsourced sample despite considerable selection bias and unmeasured confounding in the primary data. This implementation demonstrates how simple applications of structural causal modeling can be effectively used to determine falsifiable model conditions, detect selection bias and confounding factors, and minimize estimate bias through model selection in a novel epidemiological context. As the disease and social dynamics of COVID-19 continue to evolve, public health surveillance protocols must continue to adapt; the emergence of Omicron variants and shift to at-home testing as recent challenges. Rigorous and transparent methods to develop, deploy, and diagnosis adapted surveillance protocols will be critical to their success.

Dataset Information

Evaluation of Selection Bias in an Internet-based Study of Pregnancy Planners.

Publications

Evaluation of Selection Bias in an Internet-based Study of Pregnancy Planners.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets