Table 2. Effects of BMI on “Outpatient” health expenditure with respect to normo‐weight individuals.
| Indirect | Direct | Overall | ||||
|---|---|---|---|---|---|---|
| Marginal effect | % difference from the mean | Marginal effect | % difference from the mean | Marginal effect | % difference from the mean | |
| Full sample (N=2705211), average estimated expenditure = 445 euro | ||||||
| Underweight | −8.918*** | −2% | 12.66*** | 3% | 3.745* | 1% |
| Overweight | 46.19*** | 10% | −1.650** | 0% | 44.54*** | 10% |
| Obesity | 98.07*** | 22% | 12.51*** | 3% | 110.6*** | 25% |
| Severe obesity | 139.3*** | 31% | 33.77*** | 8% | 173.0*** | 39% |
| Very severe obesity | 159.3*** | 36% | 69.48*** | 16% | 228.7*** | 51% |
| 18-44 (N=934147), average estimated expenditure = 173 euro | ||||||
| Underweight | −6.019*** | −3% | −8.615*** | −5% | −14.63*** | −8% |
| Overweight | 18.08*** | 10% | 10.40*** | 6% | 28.48*** | 17% |
| Obesity | 39.05*** | 23% | 18.44*** | 11% | 57.48*** | 33% |
| Severe obesity | 63.34*** | 37% | 30.43*** | 18% | 93.76*** | 54% |
| Very severe obesity | 91.40*** | 53% | 73.45*** | 43% | 164.8*** | 95% |
| 45-64 (N=990831), average estimated expenditure = 416 euro | ||||||
| Underweight | −33.64*** | −8% | 30.61*** | 7% | −3.034 | −1% |
| Overweight | 74.71*** | 18% | 1.289 | 0% | 76.00*** | 18% |
| Obesity | 145.1*** | 35% | 14.16*** | 3% | 159.3*** | 38% |
| Severe obesity | 203.9*** | 49% | 33.27*** | 8% | 237.2*** | 57% |
| Very severe obesity | 236.3*** | 57% | 73.93*** | 18% | 310.2*** | 75% |
| 65+ (N=780233), average estimated expenditure = 796 euro | ||||||
| Underweight | −84.66*** | −11% | −0.941 | 0% | −85.60*** | −11% |
| Overweight | 65.49*** | 8% | 5.145*** | 1% | 70.64*** | 9% |
| Obesity | 115.8*** | 15% | 25.25*** | 3% | 141.1*** | 18% |
| Severely obesity | 145.9*** | 18% | 56.22*** | 7% | 202.1*** | 25% |
| Very severe obesity | 138.3*** | 17% | 78.01*** | 10% | 216.3*** | 27% |
p<0.01
p<0.05
p<0.1, each p value refers to a t test for equality of means for each category with respect to normal weight category.
Average estimated expenditure for each sample was obtained as the predicted expenditure from the health expenditure equation, calculated for the normal weight individuals, and at means of all other regressors.
Indirect marginal effects for each BMI category were computed as the sum of nonlinear combinations of parameters estimated within each pathology-specific equation with the respective pathology-specific parameter estimated within the health expenditure equation.
Direct marginal effects for each BMI category were obtained as relative parameter estimates from the health expenditure equation.
verall marginal effects for each BMI category was computed as the sum of the respective direct and indirect marginal effects.