DPH 810 Module 7 Staged Study Proposal Example

Food Swamps and Diabetes Incidence: A Staged Proposal for a Community Cohort Study

Student Name

Doctor of Public Health Program, Aspen University

DPH 810: Advanced Epidemiology in Public Health

Instructor Name

Month Day, Year

Food Swamps and Diabetes Incidence: A Staged Proposal for a Community Cohort Study

This proposal brings together the earlier modules into a staged plan for a doctoral study testing if food swamp exposure raises type 2 diabetes incidence in adults of a composite county. It sets out aims, significance, design, measures, analysis, safeguards against bias, timeline, ethics and dissemination.

Specific Aims

Aim 1: estimate how the neighborhood food swamp ratio relates to new type 2 diabetes over five years in adults who start the study without diabetes. Aim 2: examine whether the association differs by neighborhood income and urban or rural setting. Aim 3: describe food shopping behaviors and perceptions by food swamp quartile through a community survey to aid interpretation.

Background and Significance

Neighborhood food environments have been linked to obesity and diabetes. County-level analyses suggest food swamps predict obesity more strongly than food deserts (Cooksey-Stowers et al., 2017), and cohort studies have found better neighborhood resources associated with lower diabetes incidence (Auchincloss et al., 2009). US cohort evidence using food swamp measures in mixed urban and rural counties remains limited, and local zoning decisions depend on such evidence.

Design Overview

Stage one is a retrospective cohort of about 80,000 diabetes-free adults drawn from two health systems' records, followed for five years, with geocoded addresses linked to food outlet data. Stage two is a stratified community survey of 1,600 adults. Stage three is a natural experiment around two supermarket openings. Each stage produces a stand-alone deliverable, so progress is visible even if later stages are delayed.

Aims, Data and Analysis

The table links aims to data and analysis.

Measures

Exposure is the ratio of fast-food and convenience outlets to supermarkets and produce markets within distance buffers. The outcome is incident type 2 diabetes defined by diagnosis codes, A1C and medication records. Covariates include age, sex, race and ethnicity, insurance, tract deprivation and visit frequency. All measures will be defined in a data dictionary shared with the committee.

Analysis Plan

Incidence rates will be calculated by quartile. Cox proportional hazards models will estimate hazard ratios with time-varying exposure, adjusting for confounders identified in a causal diagram. Robust standard errors will account for clustering by tract. Results will be reported as estimates with confidence intervals rather than significance alone (Greenland et al., 2016).

Safeguards Against Bias

Planned safeguards include sensitivity analyses among long-term residents, exclusion of early diagnoses, adjustment for visit frequency, validation audits of the outlet database, a negative control outcome and quantitative bias analysis. Neighborhood research has long emphasized separating area effects from individual characteristics (Diez Roux, 2001).

Timeline

Months one to four: approvals and data agreements. Months five to ten: data linkage and cleaning. Months eleven to sixteen: cohort analysis. Months nine to eighteen: survey fielding and analysis. Months seventeen to twenty-two: natural experiment analysis and integration. Months twenty-three to twenty-four: writing and dissemination.

Data Management and Ethics

De-identified data will be stored on a secure university server with access limited to the study team. The institutional review board will review the cohort and survey components. Geographic results will be reported at levels that protect privacy.

Dissemination

Findings will be shared with the county planning commission, health department, health systems and community groups through briefs and meetings, and submitted to a peer-reviewed journal. Messaging will focus on retail environments rather than residents, to avoid stigmatizing neighborhoods.

Expected Contributions

The study will provide US cohort evidence on food swamps and diabetes in a mixed urban and rural county, test methods to address self-selection and detection bias and inform local zoning and healthy retail policies.

Power and Sample

About 80,000 adults followed at an expected 9 cases per 1,000 person-years will give the cohort roughly 3,600 cases over five years, providing ample power for the primary comparison and reasonable power for interactions by income and rurality.

Stakeholder Engagement

A steering group including the health department, planning staff, health system researchers and community representatives will meet quarterly to review progress and help interpret findings. Engagement increases the likelihood that results inform policy.

Budget Overview

The main costs are data linkage and management, the community survey, geospatial analysis software and dissemination, estimated at $96,000 over two years, to be sought from a foundation dissertation grant and the health department's data modernization funds.

Limitations Anticipated

Limitations include reliance on records from insured patients, possible outcome misclassification, residual confounding and limited diet data. The proposal addresses each where possible and will discuss remaining limits openly.

Quality Assurance

Data quality checks will include range checks on laboratory values, verification of geocodes for a random sample of addresses and duplicate record detection. A second analyst will independently reproduce key analyses before results are finalized.

Plan for Null Results

If no association is found, the study will still inform policy by showing that food swamp zoning alone may not reduce diabetes, and by reporting the confidence interval to show which effect sizes remain plausible.

Integration of Findings

The final analysis will integrate cohort, survey and natural experiment results, noting where they agree and differ. Agreement across methods with different biases would strengthen the conclusion; disagreement would guide interpretation and future research.

Roles and Responsibilities

The student will lead design, analysis and writing; the chair will supervise methods; the spatial methods member will guide exposure measurement; and the practice member will ensure policy relevance and data access. Health system analysts will extract data under agreements.

Risks to the Timeline

Data agreements, geocoding problems and survey response could delay the project. Starting agreements early, piloting linkage and planning extra survey follow-up protect the schedule.

Conclusion

This staged proposal combines a records-based cohort, a community survey and a natural experiment to test whether food swamps raise diabetes incidence. With clear aims, appropriate measures, a rigorous analysis plan, safeguards against bias, a realistic timeline and attention to ethics and dissemination, it is ready to develop into the doctoral project.

References

Auchincloss, A. H., Diez Roux, A. V., Mujahid, M. S., Shen, M., Bertoni, A. G., & Carnethon, M. R. (2009). Neighborhood resources for physical activity and healthy foods and incidence of type 2 diabetes mellitus: The Multi-Ethnic Study of Atherosclerosis. Archives of Internal Medicine, 169(18), 1698-1704. https://doi.org/10.1001/archinternmed.2009.302

Cooksey-Stowers, K., Schwartz, M. B., & Brownell, K. D. (2017). Food swamps predict obesity rates better than food deserts in the United States. International Journal of Environmental Research and Public Health, 14(11), Article 1366. https://doi.org/10.3390/ijerph14111366

Diez Roux, A. V. (2001). Investigating neighborhood and area effects on health. American Journal of Public Health, 91(11), 1783-1789. https://doi.org/10.2105/AJPH.91.11.1783

Greenland, S., Senn, S. J., Rothman, K. J., Carlin, J. B., Poole, C., Goodman, S. N., & Altman, D. G. (2016). Statistical tests, P values, confidence intervals, and power: A guide to misinterpretations. European Journal of Epidemiology, 31(4), 337-350. https://doi.org/10.1007/s10654-016-0149-3