Response - OBSERVATIONAL STUDY DESIGNS
jaydenson
responseOBSERVATIONALSTUDYDESIGNS.docx
OE
Mar 10 6:05pm| Last reply Mar 11 2:12am
Reply from Oludapo Eludoyin
Initial Discussion Post: Methodological Appraisal of Two Epidemiologic Studies
The two assigned studies, Batty and Hamer (2020) and Hillyer et al. (2021) offer distinct, yet complementary examples of epidemiologic inquiry applied to contemporary public health problems. Although both studies address risk behaviors and health outcomes, they differ substantially in design, population, and analytic approach. A critical appraisal of their methods provides insight into the strengths and limitations of each design and the appropriateness of the authors’ conclusions.
Study 1: Batty & Hamer (2020)
Study Description, Design, and Methods
Batty and Hamer (2020) conducted a communitybased cohort study using data from the UK Biobank to examine whether vascular risk factors and the Framingham Risk Score (FRS) predicted COVID19 infection and mortality. The investigators linked baseline cardiovascular risk profiles (collected years before the pandemic) with COVID19 test results and mortality data obtained through national registries. This longitudinal design allowed the authors to assess temporal relationships between preexisting risk factors and subsequent COVID19 outcomes.
Strength and Limitation of the Design
· Strength: The prospective cohort design minimizes recall bias and establishes temporal ordering between exposure (vascular risk factors) and outcome (COVID19 infection/mortality). The large sample size and standardized baseline measurements enhance internal validity.
· Limitation: A major limitation is exposure misclassification due to the long interval between baseline risk factor assessment and the COVID19 pandemic. Cardiometabolic profiles may have changed substantially over time, potentially attenuating associations.
Population, Data Sources, and Measures of Association
· Population: Adults aged 40–69 enrolled in the UK Biobank.
· Data Sources: UK Biobank baseline assessments; national COVID19 testing and mortality registries.
· Measures of Association: Relative risks and hazard ratios comparing COVID19 outcomes across levels of vascular risk and FRS.
Appropriateness of the Design and Conclusions
The cohort design was appropriate because it enabled the authors to examine how preexisting risk factors influenced COVID19 outcomes during the pandemic. I agree with the researchers’ conclusions that vascular risk burden is associated with increased COVID19 severity; however, the long exposureoutcome lag warrants cautious interpretation. Their acknowledgment of this limitation strengthens the credibility of their findings.
Study 2: Hillyer et al. (2021)
Study Description, Design, and Methods
Hillyer et al. (2021) conducted a crosssectional study among young adult patients in a community health setting to assess ecigarette use and cooccurring lifestyle behaviors associated with cancer risk. Participants completed surveys capturing tobacco use, alcohol consumption, physical activity, and other behavioral risk factors. The study aimed to identify opportunities for clinical intervention.
Strength and Limitation of the Design
· Strength: The crosssectional design allowed for rapid assessment of behavioral patterns in a clinical population, making it wellsuited for identifying intervention opportunities in real time.
· Limitation: The design cannot establish temporality or causality. It is unclear whether ecigarette use preceded or followed other risky behaviors, limiting the ability to infer directional relationships.
Population, Data Sources, and Measures of Association
· Population: Young adult patients (ages 18–24) receiving care in a community health system.
· Data Sources: Selfreported behavioral surveys administered during clinical encounters.
· Measures of Association: Prevalence estimates and odds ratios comparing behavioral risk patterns between ecigarette users and nonusers.
Appropriateness of the Design and Conclusions
The crosssectional design was appropriate for the study’s purpose: identifying cooccurring behaviors and informing clinical screening strategies. I agree with the authors’ conclusions that ecigarette use may serve as a marker for broader lifestyle risk, though causal claims should be avoided. Their emphasis on clinical intervention opportunities aligns with the descriptive nature of the data.
Synthesis and Final Insights
Both studies demonstrate thoughtful alignment between research questions and methodological choices. Batty and Hamer (2020) appropriately leveraged a cohort design to examine temporal associations between vascular risk and COVID19 outcomes, while Hillyer et al. (2021) used a crosssectional approach to characterize behavioral patterns in a clinical population. Each design carries inherent strengths—temporal clarity in cohorts, efficiency in crosssectional surveys—and predictable limitations. Overall, the authors’ conclusions are consistent with their data and appropriately cautious given the constraints of their respective designs.
References
Batty, G. D., & Hamer, M. (2020). Vascular risk factors, Framingham risk score, and COVID19: Communitybased cohort study. Cardiovascular Research, 116(10), 1664–1665. https://doi.org/10.1093/cvr/cvaa178Links to an external site.
Hillyer, G. C., Nazareth, M., Lima, S., Schmitt, K. M., Reyes, A., Fleck, E., Schwartz, G. K., & Terry, M. B. (2021). Ecigarette use among young adult patients: The opportunity to intervene on risky lifestyle behaviors to reduce cancer risk. Journal of Community Health. Advance online publication. https://doi.org/10.1007/s10900021010277 (doi.org in Bing)
SF
Mar 10 5:29pm
Reply from Sylvie Fon
Group 1
Week 3 Discussion: Critiquing Observational Study Designs
Observational study designs are essential in epidemiology because they allow researchers to examine associations between exposures and outcomes in real-world populations when experimental designs are not feasible or would be unethical (Curley, 2024). Each design offers different advantages for describing patterns, estimating risk, and generating evidence for population health practice, but each also carries important limitations that affect interpretation.
Batty and Hamer (2020) used a community-based prospective cohort design to examine whether established cardiovascular disease risk factors and the Framingham Risk Score were associated with subsequent hospitalization for COVID-19. They drew on UK Biobank baseline data collected from 2006 to 2010 and linked these records to Public Health England COVID-19 hospital testing data from March to April 2020 (Batty & Hamer, 2020). The population was a large community sample of adults aged 40 to 69 years, with an analytic sample of 356,914 participants and 700 COVID-19 hospitalizations. Their data sources included self-reported smoking, diabetes, education, ethnicity, and physical activity, along with measured body mass index, blood pressure, and cholesterol. The epidemiological measures of association were relative risks with 95% confidence intervals (Batty & Hamer, 2020).
A major strength of this design is temporality, since exposure data were collected before the COVID-19 outcome occurred, which strengthens causal interpretation more than a cross-sectional design would (Batty & Hamer, 2020). Another strength would be the use of several objectively measured vascular risk factors. A key limitation, however, is that baseline exposures were measured up to 14 years before the pandemic, so misclassification is possible despite the authors’ retest data showing moderate to high stability for some factors (Batty & Hamer, 2020). Another limitation is that UK Biobank had a low response rate and captured hospitalized COVID-19 cases only, so generalization is limited. Overall, I agree that a cohort design was appropriate because the study question asked whether prior vascular risk predicted later COVID-19 hospitalization. I also generally agree with the authors’ conclusions, but only cautiously, because the “J-shaped” Framingham pattern appears partly driven by age, and the findings are more convincing for severe COVID-19 hospitalization than for COVID-19 risk overall, including mild or asymptomatic cases that never reached the hospital (Batty & Hamer, 2020).
On the other hand, Hillyer et al. (2021) used an observational cross-sectional design embedded within a larger health needs survey of adult primary care and oncology patients at a large urban academic medical center in New York City. Their population included 804 adults who answered all tobacco and e-cigarette items, and the sample was racially and ethnically diverse, including a substantial proportion of Hispanic participants. Data sources included emailed and in-person surveys, with participant identification partly drawn from electronic medical records. The outcome was ever use of e-cigarettes, and exposures included smoking status, age, alcohol use, secondhand smoke exposure, ethnicity, sexual orientation, and cancer history (Hillyer et al., 2021). The authors used chi-square analyses and multivariable regression, reporting relative risks and 95% confidence intervals (Hillyer et al., 2021).
A strength of this research design is efficiency. It allowed the researchers to quickly describe prevalence and correlates of e-cigarette use in a clinically relevant population, and it also identified an important intervention opportunity, especially among adults aged 25 to 40 years (Hillyer et al., 2021). However, a major limitation is the inability to establish temporality, since exposure and outcome were assessed at the same time. Another limitation is selection bias from non-probability sampling and modest response rates, along with the use of a single self-reported “ever use” question that, besides being prone to social desirability bias, could not distinguish current, frequent, or long-term use (Hillyer et al., 2021). I think the cross-sectional design was appropriate for identifying patterns and generating hypotheses, but not for determining whether e-cigarette use causes cancer risk later or whether the associated behaviors preceded vaping per se. I therefore agree with the authors’ practical conclusion that routine screening for e-cigarette use may create opportunities for counseling, but I would be more cautious than the authors about implying a stronger etiologic meaning from these associations (Hillyer et al., 2021).
Taken together, these studies show why observational design choice matters. The Batty and Hamer (2020) cohort study is stronger for temporal ordering and risk prediction, whereas the Hillyer et al. (2021) cross-sectional study is stronger for rapid clinical surveillance and identifying co-occurring risk behaviors. In both cases, the designs were appropriate for the immediate research questions, but each design also imposed clear limits on causal inference. This is exactly why epidemiologic reasoning is so important for population health practice, as the design determines not only what can be measured, but also how confidently nurses can translate evidence into intervention and policy (Curley, 2024; Friis & Sellers, 2021). Given the chance, how would you redesign one of these two studies if your goal were to strengthen causal inference while still keeping the study feasible in a real-world population health setting?
References
Batty, G. D., & Hamer, M. (2020). Vascular risk factors, Framingham risk score, and COVID-19: Community-based cohort study. Cardiovascular Research, 116(10), 1664-1665. https://doi.org/10.1093/cvr/cvaa178
Curley, A. L. C. (Ed.). (2024). Population-based nursing: Concepts and competencies for advanced practice (4th ed.). Springer.
Friis, R. H., & Sellers, T. A. (2021). Epidemiology for public health practice (6th ed.). Jones & Bartlett Learning.
Hillyer, G. C., Nazareth, M., Lima, S., Schmitt, K. M., Reyes, A., Fleck, E., ... & Terry, M. B. (2022). E-cigarette use among young adult patients: The opportunity to intervene on risky lifestyle behaviors to reduce cancer risk. Journal of Community Health, 47(1), 94-100. https://doi.org/10.1007/s10900-021-01027-7
