Digital Social Work Benefit Enrollment Impact Analysis Case Study

Client question

A digital social work client wanted to understand whether helping Medicare Advantage members enroll in available benefit programs was associated with better healthcare outcomes and lower medical costs.

The client needed an analysis rigorous enough to support payer and partner conversations, while recognizing that observational Medicare data require careful handling of selection bias and comparability.

Data foundation

FastHSR combined Medicare Advantage encounter data, Medicare enrollment files, and Medicare Part D event data across multiple years. Because MA encounter data do not include complete payment amounts, cost outcomes were estimated using standardized pricing based on Traditional Medicare fee schedules and actual FFS payment data.

• MA encounter data: used to measure healthcare service utilization.
• Enrollment files: used to identify benefit enrollment timing, MA contract enrollment, churn, and continuous coverage.
• Part D events: used to measure medication fills and adherence.
• FFS-based pricing: used to estimate MA service costs in a standardized way.

Exposure and control cohorts

FastHSR constructed exposure cohorts based on new enrollment into benefit programs during the intervention year. Control beneficiaries met similar enrollment and coverage requirements but had not yet enrolled in the benefit category during the same period.

• Identify beneficiaries with continuous MA coverage during the baseline and follow-up period.
• Create exposure groups based on new benefit enrollment timing and benefit type.
• Construct delayed-enrollment controls to improve socioeconomic comparability.
• Exclude beneficiaries whose enrollment history did not support clean baseline and follow-up measurement.

Propensity score matching

Because benefit enrollment was not randomized, FastHSR used propensity score matching to create more comparable exposed and control groups. Matching helped balance baseline cost, demographics, clinical conditions, and other observable factors before estimating impact.

• Estimate each beneficiary's probability of entering the exposure group based on baseline characteristics.
• Match exposed beneficiaries to controls with similar propensity scores.
• Review baseline balance after matching.
• Run matched comparisons separately for each benefit category.

Difference-in-differences design

FastHSR used a difference-in-differences design to compare outcome changes before and after benefit enrollment for exposed beneficiaries against the same before-and-after changes for matched controls.

• Calculate pre/post change for the exposure group.
• Calculate pre/post change for the matched control group.
• Estimate the difference between those two changes.
• Report findings by benefit category and outcome measure.
• Extend selected outcomes into an additional follow-up year where data supported longer measurement.

Outcome measures

The evaluation focused on outcomes that matter to Medicare Advantage plans and social care organizations: spending, primary care engagement, medication adherence, and member retention.

• Cost of care: standardized MA cost estimates based on encounter utilization and FFS pricing.
• Primary care visits: E/M and Annual Wellness Visit claims with primary care provider specialties.
• Medication adherence: Part D-based adherence measures using proportion of days covered logic.
• Churn: MA contract changes measured from enrollment files across year boundaries.

Findings

The analysis found that benefit enrollment was associated with meaningful changes in several healthcare outcomes, though the magnitude and consistency differed by benefit category and follow-up period. The strongest signals appeared where benefit enrollment most directly reduced financial barriers and supported ongoing medication access.

The results also highlighted the need for careful interpretation. Churn patterns, selection into benefit programs, plan-specific implementation differences, and member-level socioeconomic factors can all influence estimates. The report therefore framed the findings as strong descriptive and quasi-experimental evidence, with additional analyses recommended for payer-specific validation.

Deliverables

• Benefit enrollment exposure and control cohort files.
• Propensity score matched analytic samples.
• Baseline balance checks after matching.
• Difference-in-differences estimates by benefit category.
• Standardized MA cost outcomes using FFS-based pricing.
• Medication adherence, primary care visit, and churn outcome summaries.
• Methodology notes for payer, partner, and investor-facing discussions.

Use cases

• Digital social work value proposition development.
• Benefit navigation ROI analysis.
• Medicare Advantage payer partnership strategy.
• Medication adherence and social needs intervention evaluation.
• Member retention and churn analysis.
• Quasi-experimental healthcare impact evaluation.

References

1. Lissenden B, Yao N "Aaron." Affordable Care Act Changes To Medicare Led To Increased Diagnoses Of Early-Stage Colorectal Cancer Among Seniors. Health Affairs. 2017;36(1):101-107. doi:10.1377/hlthaff.2016.0607
2. Jung J, Carlin C, Feldman R. Measuring resource use in Medicare Advantage using Encounter data. Health Services Research. 2022;57(1):172-181. doi:10.1111/1475-6773.13879
3. Jung J, Carlin C, Feldman R, Tran L. Implementation of resource use measures in Medicare Advantage. Health Services Research. 2022;57(4):957-962. doi:10.1111/1475-6773.13970
4. Meyers DJ, Belanger E, Joyce N, McHugh J, Rahman M, Mor V. Analysis of Drivers of Disenrollment and Plan Switching Among Medicare Advantage Beneficiaries. JAMA Internal Medicine. 2019;179(4):524-532. doi:10.1001/jamainternmed.2018.7639

Frequently asked questions

Why use delayed-enrollment controls?

Beneficiaries who enroll in similar benefits later may be more comparable to current enrollees than beneficiaries who never enroll. This helps improve socioeconomic comparability in observational data.

Why combine matching with difference-in-differences?

Matching improves baseline comparability, while difference-in-differences compares changes over time. Together, they provide a stronger observational design than a simple pre/post comparison.

Why standardize MA costs?

MA encounter data generally do not include complete payment amounts. Standardized costs estimate what observed MA services would have cost under Traditional Medicare FFS pricing structures.

For digital social work impact evaluation, benefit navigation ROI analysis, MA encounter analytics, or difference-in-differences evaluation, please email us.

See more case studies