Table 4.
Estimates from pooled models and restricted FE models
| Pooled models | Restricted FE models | |||||||
|---|---|---|---|---|---|---|---|---|
| Probabilities | Case numbers | Probabilities | Case numbers | |||||
| Logit | Poisson | Logit | Poisson | |||||
| Odds ratio | Multipl. effect | Odds ratio | Multipl. effect | |||||
| Est. | SE | Est. | SE | Est. | SE | Est. | SE | |
| Death | 1.012 | (0.017) | ||||||
| Inpatient care | ||||||||
| Overall | 0.994 | (0.005) | 0.986** | (0.006) | 0.985** | (0.006) | 0.981*** | (0.006) |
| Not PAH | 1.003 | (0.005) | 0.994 | (0.007) | 0.998 | (0.006) | 0.989 | (0.007) |
| PAH | 0.939*** | (0.012) | 0.936*** | (0.011) | 0.916*** | (0.013) | 0.936*** | (0.014) |
| Outpatient care | ||||||||
| Overall non-lab. | 1.169*** | (0.004) | 1.181*** | (0.004) | 1.231*** | (0.005) | 1.192*** | (0.004) |
| Overall lab. | 1.078*** | (0.004) | 1.124*** | (0.006) | 1.109*** | (0.005) | 1.147*** | (0.006) |
Probability of not PAH refers to the probability that all inpatient stays are not PAH in a year
See Table 1 for baseline probabilities and case numbers. Baseline probability of death: 1.2%
Cluster-robust standard errors are displayed for all models apart from FE logit
Controls for pooled models: cubic age interacted with gender, calendar year dummies, micro-region of residence, and health care supply variables
Controls for FE models: fixed effects, cubic age, calendar year dummies, and health care supply variables
Restricted FE models are estimated on the sample of those who did not die in the examined period
Number of observations: 7,412,000. Number of periods: 8. Number of people: 1,037,000
PAH potentially avoidable hospitalization
*** ** *