MA & D-SNP Competitive Lens

How the numbers are built

• Unit. Plan Year + State + County + Contract ID + PBP. Benefits come from the PBP segment that serves the county. Plans are never merged on carrier name, plan name or Contract ID alone.
• Geography.
• Plans included (the market for every share, total, HHI and benchmark). Individual Medicare Advantage and SNP plans offered in the CMS Landscape for the county: HMO, HMO-POS, local PPO and regional PPO, with D-SNP, C-SNP and I-SNP selected by the segment view. Excluded: employer group (800-series) plans, PFFS, MSA, 1876 cost and PSO plans, PACE, Medicare-Medicaid plans, PDPs, and CPSC members recorded in a county the plan does not offer in the Landscape (members living outside the filed service area). The CMS reconciliation report lists these exclusions market by market.
• Enrollment vintages. Year-over-year comparisons always use the February file of CMS Monthly Enrollment by Contract/Plan/State/County (CPSC) for each plan year (post-AEP, same month each year); its analytical grain is Contract + Plan/PBP + State + County. The latest CPSC month loaded is shown separately and is never compared with a February file as if it were year-over-year growth.
• Suppressed enrollment. CMS suppresses plan-county counts of 1–10. Those cells are stored as suppressed (null), never zero. A total that includes k suppressed cells is the range [reported + k, reported + 10k]. A carrier share is shown as [A_lo ÷ (A_lo + others_hi), A_hi ÷ (A_hi + others_lo)]; changes use the matching bounds. HHI is shown with a sensitivity range (all suppressed cells at 1 vs all at 10). Charts and sorting use the range midpoint; the range is always shown in the value and the tooltip. A ⓢ marks values that include suppressed cells.
• Market benchmarks. Enrollment-weighted (default business view) — "How does our benefit position compare with the plans members are actually enrolled in?": each offering weighted by its February enrollment (suppressed cells computed at 1 and at 10; a † marks results that differ). For a plan year whose February enrollment is not yet published, it falls back to offering-weighted and says so. Offering-weighted median (the original calculation, unchanged) — "How does our offering compare with the menu of competing plans available?": every county-plan offering is one observation, so a plan offered in 30 counties contributes 30 observations. It describes the menu, not what beneficiaries experience. County-normalized — "How does our position compare when each county receives equal influence?": each county carries equal total weight, split across its offerings. Medicare-beneficiary-weighted — "How does our position compare after weighting markets by Medicare population?": each county weighted by its total Medicare beneficiaries (CMS Medicare Monthly Enrollment), split across its offerings. The same weighting is applied to carrier medians. Prevalence measures ("plans with …") use the same weights over offerings where the measure is defined.
• Direction of change. Colour comes from the metric-direction dictionary below. Where member value depends on eligibility, utilization or benefit structure (SSBCI, flex allowances, zero-dollar cost-sharing D-SNPs, plan type mix) no direction is assigned. Enrollment and share describe market position and are never coloured.
• Allowances. Converted to annual equivalents from the CMS amount and frequency (monthly ×12, quarterly ×4, semi-annual ×2, yearly ×1, every 2 years ÷2, every 3 years ÷3); "other" frequencies are not annualized and appear as not filed. Hearing aid per-ear amounts are doubled. Combined allowances are sorted into dental-only, vision-only, hearing-only, OTC-only, or multi-category "flex". SSBCI benefits apply only to eligible members and are never added to universal benefits. The raw CMS amount and frequency behind each plan value are shown in the plan comparison tooltips.
• Illustrative stay scenarios. "Illustrative 5-day inpatient member cost" and "Illustrative 30-day SNF member cost" are standardized scenarios, not any beneficiary's actual cost: one in-network, tier-1, Medicare-covered stay, first admission of the year or benefit period. Per-day copays are summed across the filed day intervals ($0 intervals add $0, days after the last interval add $0, an open-ended interval runs to the end of the stay); otherwise a per-stay copay is used. A filed inpatient deductible is added and a filed service-specific MOOP caps the result. No copay and no coinsurance filed = $0. Coinsurance designs and Medicare-defined cost sharing are not computed (they need the allowed amount or Part A parameters) and are excluded from medians. Tiered networks use tier 1, which QA confirms is the CMS-flagged lowest-cost tier. The benefit-period type is shown with each plan's structure.
• Part D. Retail 30-day tier cost sharing, using the preferred pharmacy where one exists. The 2025 Part D redesign introduced an annual out-of-pocket cap ($2,000 in 2025; $2,100 in 2026). Changes in premiums, deductibles and benefit design occurred alongside the redesign; this dataset does not attribute those changes to a single cause.
• Carriers. Grouped by the CMS parent organization reported for each Contract ID in each February file (year-aware). Documented brand-level exceptions: CDPHP is shown separately from Excellus BCBS although CMS reports Lifetime Healthcare, Inc. as the parent of both (CDPHP and The Lifetime Healthcare Companies affiliated in 2024; the brands compete separately). Fidelis Care is not split from Centene (Wellcare): CMS reports Centene as its parent and its contract was rebranded Wellcare for 2026. Highmark Wholecare counts as Highmark (CMS parent Highmark Health). When a contract changes parent between February files (for example The Cigna Group → HCSC), each year keeps its CMS-reported owner and the change is labelled an ownership transfer, never a market entry, exit or member movement. The full contract-by-year mapping is published with the QA report.
• Plan transitions. See Entries & exits. CMS Plan Crosswalk mappings take precedence; everything else is labelled as inference.
• Takeaway labels. Observed = read directly from CMS files; Derived = calculated by MedicareInsights; Interpretation = an evidence-supported reading, never presented as CMS fact; Investigate = a question worth checking, with no conclusion implied. "Explore the change" lists evidence from the same period only.
• D-SNP market-sizing proxy. D-SNP February enrollment (CMS Monthly Enrollment by Contract/Plan/State/County (CPSC)) ÷ full-benefit dual-eligible beneficiaries (CMS Medicare Monthly Enrollment, April 2026) in the counties in view. This is not an eligibility rate, conversion rate, or beneficiary enrollment rate. The numerator and denominator come from different reporting systems and vintages, and some D-SNPs also enroll partial-benefit duals.
• Missing values. Shown with their meaning: Suppressed (CMS 1–10 cell), Not filed (no value in the PBP file), Not applicable (e.g. drug deductible for a plan without Part D), Not rated (CMS: not enough data / plan too new), Awaiting CMS release (next plan year). Nothing missing is shown as 0.
• Causation. These are descriptive plan and enrollment files. The page reports what changed together; it does not claim that any benefit change caused an enrollment change.

Metric-direction dictionary

Hover ⓘ next to any measure for its definition, CMS source, formula, weighting and limitations.

Not available

• Hospital and provider network participation by year (not in these CMS files).
• Formulary coverage of specific drugs, and mail-order detail.
• SSBCI or combined-allowance dollar amounts that plans did not file as a maximum. These appear as "offered" without an amount.
• An overall benefit score. None is published, because no weighting of the dimensions above would be defensible without member-level utilization data.