{
  "question": "How do population size, population age composition, and age-specific underlying-heart-disease death rates contribute arithmetically to the change in U.S. heart-disease deaths between 2018 and 2024?",
  "population": "U.S. residents, all races and Hispanic origins, both sexes combined; national aggregate data only",
  "source": "CDC WONDER Underlying Cause of Death, 2018-2024, Single Race, D158; deaths and population from the same query",
  "cause_codes": ["I00-I09", "I11", "I13", "I20-I51"],
  "primary_years": [2018, 2024],
  "primary_age_groups": "WONDER ten-year groups: <1, 1-4, 5-14, 15-24, 25-34, 35-44, 45-54, 55-64, 65-74, 75-84, 85+",
  "identity": "D=N*sum(s_a*r_a); N is the summed known-age population, s_a=n_a/N, and r_a=d_a/n_a",
  "decomposition": "Replace N, the entire age-share vector s, and the entire rate vector r between the endpoints in each of the six possible orders; average each factor's signed marginal change across all orders. Report every order and the averaged contributions, in deaths and as percentages of baseline deaths. Verify exact reconciliation using rational arithmetic.",
  "primary_interpretation": "Report the signs and magnitudes of the three components and whether age composition is the larger positive demographic component. These are arithmetic counterfactual contributions, not causal effects or estimates of individual clinical risk.",
  "uncertainty": "Treat each age/year death count as an independent Poisson count, condition on the supplied populations, and use the linear coefficients of the decomposition to calculate analytic standard errors and two-sided normal 95% intervals. Label these conditional model intervals, not uncertainty about registration errors, coding or population denominators. No hypothesis tests or causal analyses.",
  "planned_sensitivity": [
    "Repeat the endpoint comparison separately for females and males",
    "Repeat with baseline 2019 and endpoint 2024",
    "Report each adjacent-year decomposition from 2018 through 2024",
    "Repeat the primary comparison with the coarser groups <65, 65-74, 75-84, and 85+",
    "Repeat excluding the open-ended 85+ group, explicitly changing the target population to ages below 85"
  ],
  "missing_and_suppression": "Exclude not-stated ages from all rate decompositions. Do not treat suppressed or unavailable counts as zero, reconstruct suppressed counts, or publish counts below the CDC disclosure threshold. Halt a claimed comparison if a required age cell has missing/suppressed deaths or nonpositive/unavailable population. Preserve source caveats and distinguish known-age totals from all-age totals.",
  "validation": "Pin raw query responses with hashes; derive rates from integer counts and denominators rather than rounded display rates; check source national totals; independently implement the symmetric decomposition as its closed-form three-factor expansion; check time-reversal symmetry and exact sum to the observed difference. Compare crude and standardized source trends only where population definitions agree.",
  "prior_knowledge": "The literature search and CDC DQS published all-population summary rates for 2018-2024 were inspected before registration. The age-specific count/population queries and the proposed decomposition have not yet been run. This is an updated replication/secondary analysis, with no claim that demographic decomposition or the broad aging mechanism is new.",
  "scope": "No individual records, no clinical advice, no inference about particular risk factors or interventions. Every planned analysis, failed check, deviation and exploratory analysis will be reported."
}