Table 2.

Regression model forms.

Model	Form	Description
Single Pollutant (SP)	$\mathit{D}={\beta }_0+{\beta }_1\mathit{P}{\mathit{M}}_{\mathit{i}}+\sum _{j=1}^N{\beta }_j^{’}{\mathit{C}}_{\mathit{j}}$	Primary models evaluating association of PM2.5 or a component with mortality
Single Pollutant-Adjusted for total PM2.5 (SP-adj)	$\mathit{D}={\beta }_0+{\beta }_1\mathit{P}{\mathit{M}}_{\mathit{i}}+{\beta }_2\mathit{P}{\mathit{M}}_{2.5}+\sum _{j=1}^N{\beta }_j^{’}{\mathit{C}}_{\mathit{j}}$	SP model with adjustment of components for total PM2.5
Residual (R)	$\mathit{D}={\beta }_0+{\beta }_1\mathit{residua}{\mathit{l}}_{\mathit{i}}+\sum _{j=1}^N{\beta }_j^{’}{\mathit{C}}_{\mathit{j}}$ where residuali = PMi – mi × PM2.5 – bi and mi and bi are component-specific coefficients obtained from the regression: PMi = mi × PM2.5 + bi	Second adjustment model of PM2.5 components for total PM2.5 based on residuals from regressing total PM2.5 components on PM2.5
Multipollutant (MP)	$\begin{array}{c}\hfill \mathit{D}={\beta }_0+\left\{{\beta }_1\mathit{P}{\mathit{M}}_{\mathit{OA}}\right\}+{\beta }_2\mathit{P}{\mathit{M}}_{\mathit{Sea}\mathit{spray}}+{\beta }_3\mathit{P}{\mathit{M}}_{\mathit{Soot}}\hfill \\ \hfill +{\beta }_4\mathit{P}{\mathit{M}}_{\mathit{N}{\mathit{H}}_4\mathit{N}{\mathit{O}}_3}+{\beta }_5\mathit{P}{\mathit{M}}_{\mathit{S}{\mathit{O}}_4}+{\beta }_6\mathit{P}{\mathit{M}}_{\mathit{Dust}}+\sum _{j=1}^N{\beta }_j^{\prime }{\mathit{C}}_{\mathit{j}}\hfill \end{array}$	Model with 6 primary PM2.5 components including OA to evaluate their independent associations
Multipollutant OA (MP OA)	$\begin{array}{c}\hfill \mathit{D}={\beta }_0+\left\{{\beta }_1\mathit{P}{\mathit{M}}_{\mathit{SOA}}+{\beta }_2\mathit{P}{\mathit{M}}_{\mathit{POA}}\right\}+{\beta }_3\mathit{P}{\mathit{M}}_{\mathit{Sea}\mathit{spray}}\hfill \\ \hfill +{\beta }_4\mathit{P}{\mathit{M}}_{\mathit{Soot}}+{\beta }_5\mathit{P}{\mathit{M}}_{\mathit{N}{\mathit{H}}_4\mathit{N}{\mathit{O}}_3}+{\beta }_6\mathit{P}{\mathit{M}}_{\mathit{S}{\mathit{O}}_4}\hfill \\ \hfill +{\beta }_7\mathit{P}{\mathit{M}}_{\mathit{Dust}}+\sum _{j=1}^N{\beta }_j^{\prime }{\mathit{C}}_{\mathit{j}}\hfill \end{array}$	Multipollutant composition model as in MP but with refinement of OA (OA subdivided into SOA and POA)
Multipollutant SOA (MP SOA)	$\begin{array}{c}\hfill \mathit{D}={\beta }_0+\left\{\right){\beta }_1\mathit{P}{\mathit{M}}_{\mathit{SO}{\mathit{A}}_{\mathit{BVOC}}}+{\beta }_2\mathit{P}{\mathit{M}}_{\mathit{SO}{\mathit{A}}_{\mathit{AVOC}}}\hfill \\ \hfill +{\beta }_3\mathit{P}{\mathit{M}}_{\mathit{POA}}\left\}\right)+{\beta }_4\mathit{P}{\mathit{M}}_{\mathit{Sea}\mathit{spray}}+{\beta }_5\mathit{P}{\mathit{M}}_{\mathit{Soot}}\hfill \\ \hfill +{\beta }_6\mathit{P}{\mathit{M}}_{\mathit{N}{\mathit{H}}_4\mathit{N}{\mathit{O}}_3}+{\beta }_7\mathit{P}{\mathit{M}}_{\mathit{S}{\mathit{O}}_4}+{\beta }_8\mathit{P}{\mathit{M}}_{\mathit{Dust}}+\sum _{j=1}^N{\beta }_j^{\prime }{\mathit{C}}_{\mathit{j}}\hfill \end{array}$	Multipollutant composition model as in MP OA but with refinement of SOA into SOA from oxidation of VOCs that are predominantly biogenic (BVOC) and anthropogenic (AVOC) in origin.

D, the outcome, is the age-adjusted cardiovascular and respiratory disease death rate in number of deaths per 100,000 people. By default, PM_i are concentrations of component i in μg m⁻³ normalized to the IQR, also in μg m⁻³, of that pollutant. When coefficients are presented per unit mass, the IQR normalization does not occur. PM_2.5 is total PM_2.5 concentration. $\mathrm{\beta }$ values indicate regressed coefficients. Confounders (N = 28 by default) with values, C_j, are included in each model. Bold indicates a vector of county-level values. In MP models, brackets are placed around OA components for emphasis.