Table 2.
Explanatory models for the relationship between sociodemographic factors and mortality in hospitalised children with COVID-19 in Brazil
| Models | OR | 95% CI | P value |
| Model 1 (n=5856) | |||
| Comorbidity | 4.81 | 3.82 to 6.06 | <0.001 |
| Model 2 (n=4400) | |||
| Comorbidity | 4.71 | 3.63 to 6.11 | <0.001 |
| Model 3 (n=4400) | |||
| Comorbidity | 2.89 | 1.02 to 8.22 | 0.046 |
| White | 1.00 | – | – |
| Pardo | 2.13 | 1.29 to 3.54 | 0.003 |
| Black | 1.44 | 0.54 to 3.85 | 0.460 |
| Asian | 4.68 | 1.07 to 20.49 | 0.040 |
| Indigenous | 9.32 | 3.01 to 28.84 | <0.001 |
| South | 1.00 | – | – |
| North | 1.76 | 1.08 to 2.86 | 0.023 |
| Low GeoSES | 1.00 | – | – |
| Middle GeoSES | 0.51 | 0.31 to 0.81 | 0.005 |
| High GeoSES | 0.34 | 0.17 to 0.69 | 0.002 |
| Male | 1.00 | – | – |
| Female | 0.99 | 0.79 to 1.25 | 0.959 |
| Model 4 (n=2907) | |||
| Comorbidity | 2.85 | 0.87 to 9.34 | 0.083 |
| White | 1.00 | – | – |
| Pardo | 2.17 | 1.14 to 4.11 | 0.018 |
| Black | 2.13 | 0.63 to 7.17 | 0.220 |
| Asian | 5.79 | 0.92 to 36.30 | 0.061 |
| Indigenous | 15.78 | 4.18 to 59.54 | <0.001 |
| South | 1.00 | – | – |
| North | 1.76 | 0.95 to 3.25 | 0.070 |
| Low GeoSES | 1.00 | – | – |
| Middle GeoSES | 0.58 | 0.30 to 1.12 | 0.105 |
| High GeoSES | 0.27 | 0.13 to 0.59 | 0.001 |
| Male | 1.00 | – | – |
| Female | 1.22 | 0.90 to 1.66 | 0.195 |
| Model 5 (n=1493) | |||
| Comorbidity | 5.33 | 1.55 to 18.27 | 0.008 |
| White | 1.00 | – | – |
| Pardo | 2.22 | 0.94 to 5.20 | 0.067 |
| Black | 0.88 | 0.16 to 4.78 | 0.880 |
| Asian | 3.04 | 0.24 to 39.15 | 0.393 |
| Indigenous | 2.18 | 0.13 to 35.38 | 0.584 |
| South | 1.00 | – | – |
| North | 1.70 | 0.74 to 3.89 | 0.206 |
| Low GeoSES | 1.00 | – | – |
| Middle GeoSES | 0.49 | 0.23 to 1.06 | 0.071 |
| High GeoSES | 0.35 | 0.12 to 0.97 | 0.043 |
| Male | 1.00 | – | – |
| Female | 0.68 | 0.46 to 1.00 | 0.055 |
Results of multilevel mixed-effects generalised linear models, with municipality and hospital as random effects. Model 1 considers exclusively the association of comorbidities (at least one, asthma excluded) with mortality, adjusted for sex. Model 2 includes adjustment for sociodemographic factors (ethnicity, region, age and GeoSES). Model 3 expands model 2, including the assessment of possible interactions between comorbidities and sociodemographic factors. Models 4 and 5 are equivalent to model 3, filtering for age group (up to 10 years old or older than 10 years old, respectively). The reference category for ‘comorbidity’ was patients without previous conditions.
SES, socioeconomic status.