1. Results of meta‐analysis individual patient data.
| Outcomes | Trial | Number (nocturnal‐NIPPV/control) | Tx Effect* | 95% CI |
Homogeneity of Tx Effect, P Value |
| PaO2 (3 months) | Casanova 2000; Clini 2002; Gay 1996; Meecham Jones 1995; Sin 2007; Strumpf 1991 | 79/83 | 1.30 | ‐0.71 to 3.30 | 0.4787 |
| PaO2 (12 months) | Clini 2002; McEvoy 2009 | 62/56 | ‐1.77 | ‐8.60 to 5.07 | 0.2412 |
| PaCO2 (3 months) | Casanova 2000; Clini 2002; Gay 1996; Meecham Jones 1995; Sin 2007; Strumpf 1991 | 79/83 | ‐2.50 | ‐5.28 to 0.29 | 0.2607 |
| PaCO2 (12 months) | Clini 2002; McEvoy 2009 | 62/56 | ‐0.96 | ‐3.55 to 1.64 | 0.8290 |
| 6MWD (3 months) | Gay 1996; Meecham Jones 1995; Sin 2007 | 21/19 | 27.7 | ‐11.0 to 66.3 | 0.5662 |
| 6MWD (12 months) | Clini 2002 | 25/21 | 3.2 | ‐49.7 to 56.1 | ‐ |
| SGRQ Total (12 months) | Clini 2002; McEvoy 2009 | 50/53 | 0.90 | ‐19.21 to 21.01 | 0.0288 |
| FEV1 (3 months) | Casanova 2000; Gay 1996; Meecham Jones 1995; Sin 2007; Strumpf 1991 | 42/41 | ‐0.01 | ‐0.09 to 0.07 | 0.2413 |
| FEV1 (12 months) | Clini 2002; McEvoy 2009 | 63/62 | ‐0.01 | ‐0.07 to 0.04 | 0.7445 |
| FVC (3 months) | Casanova 2000; Clini 2002; Gay 1996; Meecham Jones 1995; Sin 2007; Strumpf 1991 | 42/40 | 0.00 | ‐0.13 to 0.14 | 0.9570 |
| FVC (12 months) | Clini 2002; McEvoy 2009 | 63/62 | 0.04 | ‐0.12 to 0.20 | 0.4510 |
| PImax (3 months) | Casanova 2000; Gay 1996; Strumpf 1991 | 24/24 | 4.87 | ‐1.48 to 11.21 | 0.5538 |
| PImax (12 months) | Clini 2002 | 29/23 | ‐2.31 | ‐9.50, 4.89 | ‐ |
| PEmax (3 months) | Casanova 2000; Gay 1996; Strumpf 1991 | 24/24 | 22.09 | ‐23.53 to 67.70 | 0.0002 |
| Sleep efficiency (3 months) | Gay 1996; Meecham Jones 1995; Strumpf 1991 | 13/11 | ‐9.11 | ‐38.09 to 19.86 | 0.0022 |
6MWD: 6‐minute walking distance; CI: confidence interval; FEV1: forced expiratory volume in 1 second; FVC: forced vital capacity; NIPPV: non‐invasive positive pressure ventilation; PaCO2: arterial carbon dioxide tension; PaO2:arterial oxygen tension; PEmax: maximal expiratory pressure; PImax: maximal inspiratory pressure; SGRQ‐Total: total score on the St. George's Respiratory Questionnaire; Tx effect: treatment effect.
* An overall treatment effect (mean difference and associated 95% CI) was obtained from the difference in scores under each study condition (NIPPV minus controls). Individual patient data were analysed using a linear mixed model to compare the treatment effects. Treatment and time of follow‐up (3 and 12 months) were analysed with interaction terms as fixed factors. To consider the homogeneity among trials, a random factor was defined in the statistical models. Statistical significance (P value < 0.05) in the test of homogeneity suggested that the observed difference in the treatment effects was in part attributable to the study effect.