Health Burden in Wayne County (Detroit), Michigan: Age-Standardized DALY Methodology

Method generation v2 · CDC WONDER 2020–2022 · CDC PLACES 2024 · IHME GBD 2021 · US-2000 standard population · Sergey Soshnikov, MD PhD

This is a planning-grade decision-support estimate, not a peer-reviewed study or an official GBD/IHME county estimate. It is intended to help public-health planners, grant writers, and community stakeholders prioritize. Every input is a public dataset; every step is reproducible.

1. Overview

This analysis quantifies the disease burden of Wayne County (which contains Detroit) as Disability-Adjusted Life Years (DALYs) across seven condition groups: cancer, cardiovascular disease, chronic lower respiratory disease (COPD), stroke, diabetes, substance-use disorder, and a mental-health group. A DALY is one lost year of healthy life; it is the sum of Years of Life Lost to premature death (YLL) and Years Lived with Disability (YLD).

It uses the v2 age-standardized method, an upgrade over the crude method used for this site's smaller rural counties. The three upgrades: (a) YLL is computed age-specifically from a GBD reference life table rather than against one fixed benchmark age; (b) rates are age-standardized to the US-2000 population so counties with very different age structures can be compared fairly; (c) substance-use and mental-health mortality are directly observed by age, because Wayne is large enough that these counts are not statistically suppressed.

2. Data sources

2.1 Mortality — CDC WONDER

Deaths by cause × ten-year age group, Wayne County (FIPS 26163), 2020–2022 pooled, from the CDC WONDER Underlying Cause of Death, 2018–2024, Single Race database. The five chronic causes are the aggregate 113-cause-list rows: malignant neoplasms (C00–C97), diseases of heart (I00–I09, I11, I13, I20–I51), chronic lower respiratory (J40–J47), cerebrovascular (I60–I69), and diabetes (E10–E14). Substance use is the sum of drug-induced and alcohol-induced causes; the mental-health mortality proxy is intentional self-harm/suicide (X60–X84).

2.2 Prevalence — CDC PLACES 2024

Age-adjusted adult prevalence for each condition from the CDC PLACES county dataset (Socrata swc5-untb), which also supplies the adult (18+) population denominator. PLACES has no substance-use measure (see §4).

2.3 Disability weights & reference life expectancy

Disability weights are the IHME GBD 2021 values, reused unchanged across all counties (they are population-independent). Age-specific remaining life expectancy for YLL comes from the GBD 2019 reference life table (the same standard IHME uses), interpolated to each age group's midpoint.

2.4 Population denominators — Census PEP via CDC WONDER

Population by county × age group is drawn from the CDC WONDER Single-Race Population Estimates 2020–2022 (the Census Bureau's Population Estimates Program — the same denominators CDC uses for official age-adjusted death rates). We sum 2020, 2021 and 2022 to obtain person-years matching the 3-year pooled deaths, then annualize. Five-year population bands are collapsed to the ten-year mortality bands.

2.5 Economic parameters

Michigan GDP per capita $62,000 (BEA) for the human-capital method; income-adjusted HHS value of a statistical life $10.5M for the VSL method. Comorbidity correction factor 0.949 (Gaussian-copula net-YLD correction), reused from the Isabella pipeline.

3. Methods

3.1 DALY framework

DALY = YLL + YLD, no age-weighting and no time discounting, following the WHO Global Health Estimates convention.

3.2 Age-specific YLL

For each cause i and ten-year age group a with midpoint age ma:

YLL(i,a) = deaths(i,a) × RLE(m_a)

where RLE(m_a) is the remaining life expectancy at age ma interpolated from the GBD 2019 reference table. Pooled deaths are annualized (÷3). This replaces the crude method's single fixed 89.1-year benchmark, so a death at 40 and a death at 80 correctly lose different amounts of life.

3.3 YLD

YLD(i) = prevalence_ageadj(i) × adult_pop × DW(i) × CF

Prevalence is the PLACES age-adjusted value; DW the IHME disability weight; CF the 0.949 comorbidity factor. For the mental-health group, prevalence is PLACES depression × 0.50 (remission adjustment). PLACES provides no age-specific prevalence, so YLD carries no age structure (see §4).

3.4 Age-standardization (US-2000)

The cross-county comparison metric is the age-standardized YLL rate: the age-specific YLL rate per 100,000 in each band, weighted by the US-2000 standard population.

ASR_YLL = Σ_a [ YLL(a) / pop(a) × 100000 ] × w2000(a)

Standard weights are normalized over the ten age bands present (the <1 band is folded into 1–4 to match the population banding, and weights re-sum to 1). Wayne's age-standardized YLL rate is 14,177 per 100,000 — about 1.7× Ann Arbor's 8,251.

3.5 Substance-use & mental-health mortality

Unlike small rural counties, Wayne's substance-use and suicide deaths are numerous enough to be reported by age rather than suppressed, so both are observed directly rather than modeled from proxies. Wayne's substance-use mortality (1,107 deaths/yr) is a defining feature of its burden.

3.6 Suppression handling

CDC WONDER suppresses cell counts of 1–9. Suppressed chronic-disease cells at young ages (<45), where the true count is near zero, are imputed at 1; suppressed substance-use/suicide cells are imputed at the midpoint 5. A 0-to-9 sensitivity band brackets the choice; in a county this large the swing on the total is well under ±1%.

3.7 Economic burden

Human-capital: DALYs × $62,000. VSL: annual deaths × $10.5M (mortality only). Both are planning-grade, not audited fiscal figures.

4. Limitations & comparability

Comparability with this site's rural counties. Wayne and Washtenaw use the v2 age-standardized method; Isabella, Clare, Midland and Gratiot use the earlier v1 crude method (fixed reference life expectancy; substance-use and mental-health mortality suppressed and excluded). Rates across the two groups are not directly comparable and must not be ranked against one another. Compare within a method group only, or see the dedicated Rural vs Urban page, which discusses the seam explicitly.

5. Results summary

ConditionDeaths/yrYLLYLDDALYs
CVD (heart disease)5,531111,08733,017144,104
Cancer3,50275,34624,29699,641
Substance use1,10746,72930,03276,762
COPD73613,46020,39033,850
Mental health2249,99322,95132,944
Stroke1,01716,57615,66832,244
Diabetes64214,1248,58222,706
Total12,759287,315154,936442,251

Age-standardized YLL rate 14,177/100k · crude YLD rate 8,738/100k · crude DALY rate 24,943/100k · population 1,773,082 (2020–22 average). Rows use unrounded estimates, so YLL + YLD may differ from the DALY column by ±1.

6. Reproducibility & verification

The pipeline (Python) parses the raw CDC WONDER exports, applies the formulas above, and writes a machine-readable results file. An independent agent re-parsed the raw death files and re-implemented the engine from scratch, reproducing all 56 cause × county × metric cells and every total to the integer. Two suppression-sensitivity runs (0 and 9) bracket the central estimate. Data vintages: CDC WONDER 2020–2022, CDC PLACES 2024, CDC Single-Race Population Estimates 2020–2022, IHME GBD 2021.