SAMA/SABA | P value | LAMA/LABA | P value | +ICS | |

Total respiratory hospital admissions (ICD J44–J47) | |||||

Logistic | |||||

Model 1 (n=346) | 0.41 (0.22, 0.78) | 0.007* | 0.66 (0.31, 1.39) | 0.271 | Reference |

Model 2 (n=268) | 0.47 (0.21, 1.04) | 0.064 | 0.53 (0.22, 1.27) | 0.153 | |

Model 2 A (n=268) | 0.46 (0.2, 1.03) | 0.057 | 0.53 (0.22, 1.23) | 0.154 | |

Model 2 B (n=268) | 0.43 (0.19, 0.98) | 0.043* | 0.57 (0.23, 1.39) | 0.217 | |

Model 2 C (n=268) | 0.46 (0.20, 1.04) | 0.063 | 0.52 (0.21, 1.25) | 0.144 | |

Poisson (rate) | |||||

Model 1 (n=201) | −0.57 (−1.25, 0.12) | 0.105 | −0.93 (−1.42, –0.45) | <0.001* | Reference |

Model 2 (n=157) | −0.43 (−1.27, 0.41) | 0.314 | −0.77 (−1.35, –0.19) | 0.010* | |

Model 2 A (n=157) | −0.45 (−1.33, 0.44) | 0.323 | −0.79 (−1.37, –0.21) | 0.008* | |

Model 2 B (n=157) | −0.38 (−1.24, 0.48) | 0.382 | −0.74 (−1.33, –0.16) | 0.013* | |

Model 2 C (n=157) | −0.48 (−1.32, 0.35) | 0.258 | −0. 8 (−1.4 to –0.2) | 0.009* | |

COPD with LRTI hospital admission (ICD J44.0) | |||||

Logistic | |||||

Model 1 (n=346) | 0.34 (0.17, 0.66) | 0.002* | 0.7 (0.33, 1.47) | 0.339 | Reference |

Model 2 (n=268) | 0.46 (0.2, 1.05) | 0.066 | 0.61 (0.25, 1.47) | 0.271 | |

Model 2 A (n=268) | 0.45 (0.2, 1.04) | 0.061 | 0.61 (0.25, 1.48) | 0.272 | |

Model 2 B (n=268) | 0.42 (0.18, 0.98) | 0.043* | 0.67 (0.27, 1.65) | 0.383 | |

Model 2 C (n=268) | 0.45 (0.19, 1.06) | 0.068 | 0.59 (0.24, 1.45) | 0.252 | |

Poisson (Rate) | |||||

Model 1 (n=184) | −0.61 (−1.11, –0.11) | 0.016* | −0.81 (−1.35, –0.27) | 0.003* | Reference |

Model 2 (n=143) | −0.53 (−1.15, 0.09) | 0.096 | −0.59 (−1.15, –0.02) | 0.042* | |

Model 2 A (n=143) | −0.51 (−1.14, 0.13) | 0.117 | −0.62 (−1.19, –0.05) | 0.034* | |

Model 2 B (n=143) | −0.52 (−1.14, 0.11) | 0.105 | −0.55 (−1.1, 0) | 0.050* | |

Model 2 C (n=143) | −0.53 (−1.15, 0.09) | 0.094 | −0.59 (−1.16, –0.03) | 0.039 |

Model 1—Univariate logistic or Poisson regression model showing the effect of inhaler prescriptions against the reference group of patients using +ICS.

Model 2—Model 1 additionally adjusted for age, sex and BMI.

Model 2 A/B/C—Model 2 additionally adjusted for asthma, bronchiectasis or COPD respectively.

Note that observations with high absolute residuals are dropped from the robust regression.

*p-value <0.05.

BMI, body mass index; COPD, chronic obstructive pulmonary disease; ICD, international classification of disease; ICS, inhaled corticosteroids; LABA, long-acting β-agonists; LAMA, long-acting muscarinic antagonists; LRTI, lower respiratory tract infection; SABA, short-acting bronchodilator; SAMA, short-acting muscarinic antagonists.