1 Univariate distributions

1.1 Outcome: Rate difference per 100,000 person-days

A positive rate difference indicates that wildfire is causing hospitalizations, and a negative value implies wildfire is preventing them.

rd_100k_pt_med rd_100k_pt_mean_wt rd_100k_25th rd_100k_75th
-0.059 0.179 -0.809 0.809

1.2 Intervention variables: Tree canopy and air conditioning

1.2.1 Tree canopy

tree_canopy_prop_min tree_canopy_prop_max tree_canopy_prop_mean tree_canopy_prop_med
0 0.708 0.108 0.061

1.2.2 Air conditioning

ac_prop_min ac_prop_max ac_prop_mean ac_prop_med
0 1 0.568 0.625

1.2.3 Impervious surface

(exploring as a potential intervention variable)

1.3 Other socio-demographic and geographic characteristics

1.3.1 Proportion above poverty

above_poverty_prop_min above_poverty_prop_max above_poverty_prop_mean above_poverty_prop_med
0.18 0.96 0.7 0.73

1.3.2 Proportion with insurance

insured_prop_min insured_prop_max insured_prop_mean insured_prop_med
0.58 1 0.9 0.91

1.3.3 Biome

biome_name area_km2
Temperate Conifer Forests 130,765
Deserts & Xeric Shrublands 118,685
Mediterranean Forests, Woodlands & Scrub 113,359
Temperate Grasslands, Savannas & Shrublands 46,273

1.3.4 Rural-urban classification

Definitions of rural-urban commuting codes:

  1. Metropolitan area core: primary flow within an urbanized area (UA)
  2. Metropolitan area high commuting: primary flow 30% or more to a UA
  3. Metropolitan area low commuting: primary flow 10% to 30% to a UA
  4. Micropolitan area core: primary flow within an Urban Cluster of 10,000 to 49,999 (large UC)
  5. Micropolitan high commuting: primary flow 30% or more to a large UC
  6. Micropolitan low commuting: primary flow 10% to 30% to a large UC
  7. Small town core: primary flow within an Urban Cluster of 2,500 to 9,999 (small UC)
  8. Small town high commuting: primary flow 30% or more to a small UC
  9. Small town low commuting: primary flow 10% to 30% to a small UC
  10. Rural areas: primary flow to a tract outside a UA or UC
ruca_cat pop area_km2
(0,3] 37,107,185 111,582
(3,6] 1,544,666 53,129
(6,9] 397,143 33,352
(9,10] 284,331 83,225

2 Bivariate and stratified associations

2.1 RD x Tree canopy x geographic measures

2.1.1 RD x Tree canopy

Here are scatterplots plotting the rate difference against the (square root of) proportion tree canopy.

There is a slight negative association between tree canopy and the RD, but it is rather weak.

Observations 1270
Dependent variable rd_100k_quo_pt
Type Linear regression
Est. S.E. t val. p
(Intercept) 0.396 0.095 4.159 0.000
tree_canopy_prop_sqrt -1.051 0.297 -3.545 0.000
Standard errors: MLE
overall corr_spearman corr_pearson measure
1 -0.1 -0.09 tree_canopy_prop_sqrt

2.1.2 RD x Tree canopy x biome

We stratified associations by biome, which illustrates, as above, that the tree-canopy distribution differs starkly by biome. The RD x tree canopy association appears to be negative in Mediterran Forests, Woodlands & Scrub and in Deserts & Xeric shrublands, but slightly positive in the other two.

biome_name_freq corr_spearman corr_pearson measure
Mediterranean Forests, Woodlands & Scrub -0.17 -0.14 tree_canopy_prop_sqrt
Temperate Conifer Forests -0.02 0.07 tree_canopy_prop_sqrt
Temperate Grasslands, Savannas & Shrublands 0.23 0.16 tree_canopy_prop_sqrt
Deserts & Xeric Shrublands -0.16 -0.16 tree_canopy_prop_sqrt
Observations 1270
Dependent variable rd_100k_quo_pt
Type Linear regression
Est. S.E. t val. p
(Intercept) 0.774 0.124 6.262 0.000
tree_canopy_prop_sqrt -1.936 0.417 -4.643 0.000
biome_name_freqTemperate Conifer Forests -1.386 0.684 -2.027 0.043
biome_name_freqTemperate Grasslands, Savannas & Shrublands -1.833 0.349 -5.257 0.000
biome_name_freqDeserts & Xeric Shrublands 0.319 0.527 0.605 0.545
tree_canopy_prop_sqrt:biome_name_freqTemperate Conifer Forests 2.712 1.170 2.319 0.021
tree_canopy_prop_sqrt:biome_name_freqTemperate Grasslands, Savannas & Shrublands 4.020 1.219 3.298 0.001
tree_canopy_prop_sqrt:biome_name_freqDeserts & Xeric Shrublands -5.005 3.311 -1.512 0.131
Standard errors: MLE

2.1.3 RD x Tree canopy x rural-urban

We also stratified this association by urban-rural category.

The stratified scatterplot suggests that in the most urban areas and in the least urban areas, higher tree canopy is associated with a lower rate difference. This result is in the expected direction: higher tree canopy, lower effect of wildfire. In the other two RUCA categories, the association is in the other direction: higher tree canopy is associated with a higher rate difference (i.e., suggesting a more harmful effect of wildfire on the outcome as tree canopy rises).

ruca_cat corr_spearman corr_pearson measure
(0,3] -0.16 -0.14 tree_canopy_prop_sqrt
(3,6] 0.26 0.16 tree_canopy_prop_sqrt
(6,9] 0.09 0.10 tree_canopy_prop_sqrt
(9,10] -0.18 -0.27 tree_canopy_prop_sqrt
Observations 1270
Dependent variable rd_100k_quo_pt
Type Linear regression
Est. S.E. t val. p
(Intercept) 0.599 0.113 5.293 0.000
tree_canopy_prop_sqrt -1.831 0.384 -4.767 0.000
ruca_cat(3,6] -0.895 0.367 -2.439 0.015
ruca_cat(6,9] -1.381 0.522 -2.646 0.008
ruca_cat(9,10] 0.663 0.545 1.217 0.224
tree_canopy_prop_sqrt:ruca_cat(3,6] 3.318 0.854 3.884 0.000
tree_canopy_prop_sqrt:ruca_cat(6,9] 2.763 1.402 1.970 0.049
tree_canopy_prop_sqrt:ruca_cat(9,10] -0.861 1.078 -0.799 0.425
Standard errors: MLE

2.2 RD x A/C x geographic measures

2.2.1 RD x A/C

The association with air conditioning is weakly negative.

Observations 1104 (166 missing obs. deleted)
Dependent variable rd_100k_quo_pt
Type Linear regression
Est. S.E. t val. p
(Intercept) 0.326 0.087 3.740 0.000
ac_prop -0.365 0.135 -2.703 0.007
Standard errors: MLE
overall corr_spearman corr_pearson measure
1 -0.11 -0.07 ac_prop

2.2.2 RD x A/C x biome

biome_name_freq corr_spearman corr_pearson measure
Mediterranean Forests, Woodlands & Scrub -0.08 -0.05 ac_prop
Temperate Conifer Forests 0.05 -0.04 ac_prop
Temperate Grasslands, Savannas & Shrublands -0.01 0.10 ac_prop
Deserts & Xeric Shrublands -0.18 -0.06 ac_prop
Observations 1104 (166 missing obs. deleted)
Dependent variable rd_100k_quo_pt
Type Linear regression
Est. S.E. t val. p
(Intercept) 0.376 0.094 4.005 0.000
ac_prop -0.263 0.154 -1.705 0.088
biome_name_freqTemperate Conifer Forests -0.550 0.277 -1.988 0.047
biome_name_freqTemperate Grasslands, Savannas & Shrublands -1.423 0.553 -2.575 0.010
biome_name_freqDeserts & Xeric Shrublands 0.983 0.980 1.003 0.316
ac_prop:biome_name_freqTemperate Conifer Forests -0.003 0.637 -0.005 0.996
ac_prop:biome_name_freqTemperate Grasslands, Savannas & Shrublands 0.815 0.651 1.253 0.211
ac_prop:biome_name_freqDeserts & Xeric Shrublands -0.633 1.184 -0.534 0.593
Standard errors: MLE

2.2.3 RD x A/C x rural-urban

Like with tree canopy, the RD x A/C association varies by rural-urban category. It is more negative in more populous areas, and positive–suggesting air conditioning is associated with a higher difference effect of wildfire on acute-care utilization in less populous areas.

ruca_cat corr_spearman corr_pearson measure
(0,3] -0.14 -0.10 ac_prop
(3,6] -0.18 -0.15 ac_prop
(6,9] -0.05 0.01 ac_prop
(9,10] 0.50 0.46 ac_prop
Observations 1104 (166 missing obs. deleted)
Dependent variable rd_100k_quo_pt
Type Linear regression
Est. S.E. t val. p
(Intercept) 0.392 0.092 4.247 0.000
ac_prop -0.451 0.143 -3.157 0.002
ruca_cat(3,6] 0.327 0.344 0.951 0.342
ruca_cat(6,9] -0.963 0.622 -1.550 0.122
ruca_cat(9,10] -1.992 0.470 -4.236 0.000
ac_prop:ruca_cat(3,6] -0.449 0.563 -0.797 0.425
ac_prop:ruca_cat(6,9] 0.550 0.864 0.637 0.524
ac_prop:ruca_cat(9,10] 3.296 0.731 4.507 0.000
Standard errors: MLE

2.3 RD x impervious surfaces x geographic measures

2.3.1 RD x Proportion impervious surfaces

Observations 1270
Dependent variable rd_100k_quo_pt
Type Linear regression
Est. S.E. t val. p
(Intercept) -0.289 0.103 -2.800 0.005
imperv_prop 0.837 0.209 4.014 0.000
Standard errors: MLE
overall corr_spearman corr_pearson measure
1 0.14 0.11 imperv_prop

2.3.2 RD x Proportion impervious surfaces x biome

biome_name_freq corr_spearman corr_pearson measure
Mediterranean Forests, Woodlands & Scrub 0.12 0.12 imperv_prop
Temperate Conifer Forests 0.28 0.19 imperv_prop
Temperate Grasslands, Savannas & Shrublands 0.08 -0.01 imperv_prop
Deserts & Xeric Shrublands -0.32 -0.32 imperv_prop
Observations 1270
Dependent variable rd_100k_quo_pt
Type Linear regression
Est. S.E. t val. p
(Intercept) -0.211 0.133 -1.584 0.113
imperv_prop 0.916 0.252 3.638 0.000
biome_name_freqTemperate Conifer Forests -0.338 0.282 -1.197 0.232
biome_name_freqTemperate Grasslands, Savannas & Shrublands -0.297 0.352 -0.843 0.399
biome_name_freqDeserts & Xeric Shrublands 1.814 0.437 4.146 0.000
imperv_prop:biome_name_freqTemperate Conifer Forests 1.686 1.272 1.325 0.185
imperv_prop:biome_name_freqTemperate Grasslands, Savannas & Shrublands -0.884 0.768 -1.150 0.250
imperv_prop:biome_name_freqDeserts & Xeric Shrublands -6.058 1.262 -4.802 0.000
Standard errors: MLE

2.3.3 RD x Proportion impervious surfaces x rural-urban

ruca_cat corr_spearman corr_pearson measure
(0,3] 0.17 0.15 imperv_prop
(3,6] -0.16 -0.14 imperv_prop
(6,9] -0.21 -0.15 imperv_prop
(9,10] 0.23 0.23 imperv_prop
Observations 1270
Dependent variable rd_100k_quo_pt
Type Linear regression
Est. S.E. t val. p
(Intercept) -0.496 0.132 -3.753 0.000
imperv_prop 1.208 0.252 4.791 0.000
ruca_cat(3,6] 1.196 0.330 3.626 0.000
ruca_cat(6,9] 0.718 0.625 1.149 0.251
ruca_cat(9,10] -0.241 0.393 -0.612 0.541
imperv_prop:ruca_cat(3,6] -2.894 1.130 -2.562 0.011
imperv_prop:ruca_cat(6,9] -4.128 2.412 -1.711 0.087
imperv_prop:ruca_cat(9,10] 6.408 3.309 1.936 0.053
Standard errors: MLE

2.4 Comparison table of stratified correlations

#overall
corr_tree_canopy_prop_sqrt_overall %>% 
  bind_rows(corr_ac_prop_overall,corr_imperv_prop_overall) %>% 
  kable(digits=2)
overall corr_spearman corr_pearson measure
1 -0.10 -0.09 tree_canopy_prop_sqrt
1 -0.11 -0.07 ac_prop
1 0.14 0.11 imperv_prop 
#biome
corr_tree_canopy_prop_sqrt_biome_name_freq %>% 
  bind_rows(corr_ac_prop_biome_name_freq,corr_imperv_prop_biome_name_freq) %>% 
  kable(digits=2)
biome_name_freq corr_spearman corr_pearson measure
Mediterranean Forests, Woodlands & Scrub -0.17 -0.14 tree_canopy_prop_sqrt
Temperate Conifer Forests -0.02 0.07 tree_canopy_prop_sqrt
Temperate Grasslands, Savannas & Shrublands 0.23 0.16 tree_canopy_prop_sqrt
Deserts & Xeric Shrublands -0.16 -0.16 tree_canopy_prop_sqrt
Mediterranean Forests, Woodlands & Scrub -0.08 -0.05 ac_prop
Temperate Conifer Forests 0.05 -0.04 ac_prop
Temperate Grasslands, Savannas & Shrublands -0.01 0.10 ac_prop
Deserts & Xeric Shrublands -0.18 -0.06 ac_prop
Mediterranean Forests, Woodlands & Scrub 0.12 0.12 imperv_prop
Temperate Conifer Forests 0.28 0.19 imperv_prop
Temperate Grasslands, Savannas & Shrublands 0.08 -0.01 imperv_prop
Deserts & Xeric Shrublands -0.32 -0.32 imperv_prop 
#rural-urban
corr_tree_canopy_prop_sqrt_ruca_cat %>% 
  bind_rows(corr_ac_prop_ruca_cat,corr_imperv_prop_ruca_cat) %>% 
  kable(digits=2)
ruca_cat corr_spearman corr_pearson measure
(0,3] -0.16 -0.14 tree_canopy_prop_sqrt
(3,6] 0.26 0.16 tree_canopy_prop_sqrt
(6,9] 0.09 0.10 tree_canopy_prop_sqrt
(9,10] -0.18 -0.27 tree_canopy_prop_sqrt
(0,3] -0.14 -0.10 ac_prop
(3,6] -0.18 -0.15 ac_prop
(6,9] -0.05 0.01 ac_prop
(9,10] 0.50 0.46 ac_prop
(0,3] 0.17 0.15 imperv_prop
(3,6] -0.16 -0.14 imperv_prop
(6,9] -0.21 -0.15 imperv_prop
(9,10] 0.23 0.23 imperv_prop

2.5 RD x community characteristics

2.5.1 RD x prop. above poverty

overall corr_spearman corr_pearson
1 -0.01 0.01

2.5.2 RD x prop. insured

The association between the RD and proportion with insurance is stronger than that between proportion above poverty.

overall corr_spearman corr_pearson
1 -0.06 -0.05

2.5.3 RD x biome

The RD distribution is in the positive direction in Mediterranean Forests, Woodlands & Scrub and more negative in all the others.

biome_name_freq n_zcta rd_100k_pt_med rd_100k_pt_mean_wt
Mediterranean Forests, Woodlands & Scrub 908 0.154 0.341
Temperate Conifer Forests 101 -0.282 0.043
Temperate Grasslands, Savannas & Shrublands 225 -0.498 -0.444
Deserts & Xeric Shrublands 62 -0.756 -0.534

2.5.4 RD x urban-rural category

The distribution is more negative in more rural areas, suggesting in those areas, wildfires had a preventive effect on healthcare utilization

ruca_cat n_zcta rd_100k_pt_med rd_100k_pt_mean_wt
(0,3] 1087 -0.033 0.196
(3,6] 99 -0.035 -0.164
(6,9] 51 -0.522 0.141
(9,10] 59 -0.732 -0.306

2.6 Intervention vars x community characteristics

2.6.1 Tree canopy

2.6.1.1 Tree canopy x prop. above poverty

overall corr_spearman corr_pearson
1 0.34 0.26
biome_name_freq corr_spearman corr_pearson
Mediterranean Forests, Woodlands & Scrub 0.42 0.33
Temperate Conifer Forests 0.24 0.30
Temperate Grasslands, Savannas & Shrublands 0.31 0.32
Deserts & Xeric Shrublands 0.34 0.33

2.6.1.2 Tree canopy x prop. insured

overall corr_spearman corr_pearson
1 0.38 0.29
biome_name_freq corr_spearman corr_pearson
Mediterranean Forests, Woodlands & Scrub 0.45 0.36
Temperate Conifer Forests 0.10 0.15
Temperate Grasslands, Savannas & Shrublands 0.35 0.37
Deserts & Xeric Shrublands 0.15 0.18

2.6.1.3 Tree canopy x biome

2.6.1.4 Tree canopy x rural-urban

2.6.2 A/C

2.6.2.1 A/C x prop. above poverty

overall corr_spearman corr_pearson
1 0.03 0.01
biome_name_freq corr_spearman corr_pearson
Mediterranean Forests, Woodlands & Scrub 0.17 0.17
Temperate Conifer Forests -0.01 0.01
Temperate Grasslands, Savannas & Shrublands 0.41 0.38
Deserts & Xeric Shrublands 0.22 0.16

2.6.2.2 A/C x prop. insured

overall corr_spearman corr_pearson
1 0.04 0.09
biome_name_freq corr_spearman corr_pearson
Mediterranean Forests, Woodlands & Scrub 0.11 0.15
Temperate Conifer Forests -0.02 0.00
Temperate Grasslands, Savannas & Shrublands 0.31 0.32
Deserts & Xeric Shrublands 0.04 0.14

2.6.2.3 A/C x biome

2.6.2.4 A/C x rural-urban

2.6.3 Impervious surfaces

How does imp. surfaces compare with tree canopy?