Table 1

Hospitalisation, mortality and discontinuation parameters

Mortality and hospitalisation in postadmission states
DefinitionProbabilityBeta distribution, α, βSource
COPD-related death during admission2017 National UK COPD audit9
 Men0.046654, 13 552
 Women0.035570, 15 514
COPD-related 90-day death post admission2017 National UK COPD audit9
 Men0.0731043, 13 163
 Women0.0661058, 15 027
90-day COPD-related readmission2017 National UK COPD audit9
 Men0.1712423, 11 784
 Women0.1622609, 13 476
All-cause COPD mortality post admission2017 National UK COPD audit9
 90-day men0.046650, 13 556
 90-day women0.041660, 15 425
Hospitalisation rate and risk of non-COPD-related mortality in the stable health state for the post hospital population
DefinitionRateSample sizeSource
Base case  Garcia-Aymerich et al 12
 COPD admissions per year1.6340
 Non-COPD-related mortality0.071340
Lowest rates  Bucknall et al 14
 COPD admissions per year0.47464
 Non-COPD-related mortality0.194464
Highest rates  Budweiser et al 13 and Heinemann et al 15
 Base case risks multiplied by 2  
Treatment and exacerbation rates for stable population
Severity stageAnnual exacerbation rateProportion of exacerbations hospitalisedSource
 
%Beta distribution, α, β
GOLD 32.35610.22267, 22 062Rothnie et al 10
GOLD 42.91413.4862, 6429Rothnie et al 10
Base case rate ratios for admission to domiciliary NIV relative to usual care
Rate Ratio (95% CI)*Source
All studies estimates
Posthospital population
 Pooled mean0.494 (0.382 to 0.638)Pooled result of 11 RCTs8
 Best-case NIV0.333 (0.187 to 0.596)Li et al 30
 Worst-case NIV1.372 (1.067 to 1.763)Struik et al 31
Stable population 
 Pooled mean0.606 (0.482 to 0.760)Pooled result of 10 RCTs8
 Best-case NIV0.346 (0.276 to 0.433)Luyang et al 32
 Worst-case NIV1.166 (0.601 to 2.264)Kamiński et al 33
Western setting estimates†  
Posthospital population 
 Pooled mean0.728 (0.212 to 2.504)Pooled result of 2 RCTs8
 Best-case NIV0.389 (0.321 to 0.470)Murphy et al 1
 Worst-case NIV1.372 (1.067 to 1.763)Struik et al 31
Stable population 
 Pooled mean0.718 (0.617 to 0.836)Pooled result of 4 RCTs8
 Best-case NIV0.643 (0.432 to 0.956)Clini et al 34
 Worst-case NIV1.166 (0.601 to 2.264)Kaminski et al 33
Discontinuation
Health stateDiscontinuation RateNormal distribution, 95% CISource
All patients15%10% to 20%Assumption and Dretzke et al 8

  • *Western settings included studies from Italy, Poland, the Netherlands, Germany and the UK. Non-Western studies contributing to meta-analyses are set in China.

  • †Natural log rate ratios and standard errors used in the probabilistic sensitivity analysis are shown in online supplemental appendix 6.

  • GOLD, Global Iniative for Chronic Obstructive Lung Disease; NIV, non-invasive ventilation; RCT, randomised controlled trial.