INPULSIS/Richeldi [4] | Nintedanib | Annual rate of decline in FVC (mL·year−1) | Difference in FVC change between treatment groups | 52 weeks | Linear mixed-effect model | No imputation | 2014 |
ASCEND/King [2] | Pirfenidone | Change from baseline to week 52 in FVC % predicted | Categorical analysis: FVC decline >10% or death | 52 weeks | Ranked ANCOVA | Missing values owing to death were given the worst ranks in the primary analysis (and imputed to 0 in other analyses) Missing data for other reasons were imputed by the average of the 3 closest patients | 2014 |
PANTHER-IPF/Raghu [21] and Martinez [22] | Prednisone, azathioprine and N-acetylcysteine | Change from baseline to week 60 in absolute FVC values | Difference in FVC change between treatment groups | 60 weeks | Linear mixed-effect model | No imputation | 2012 |
Shulgina [23] | Trimethoprim–sulfamethoxazole | Change from baseline to month 12 in absolute FVC values | Difference in FVC change between treatment groups | 12 months | Linear mixed-effect model | No imputation | 2013 |
MUSIC/Raghu [24] | Macitentan | Change from baseline to month 12 in absolute FVC values | Difference in FVC change between treatment groups | 12 months | Wilcoxon rank-sum test | FVC=0 if death Worst decrease if exacerbation LOCF otherwise | 2013 |
ARTEMIS/Raghu [35] | Ambrisentan | Time to disease progression defined as “respiratory hospitalisation” or categorical decrease in FVC or DLCO¶ | HR from a Cox model | Event driven | Log-rank test | No imputation | 2013 |
FIBROSAND/Crestani [26] | Octreotide | FVC decline >10% or death | Categorical analysis: FVC decline >10% or death | 48 weeks | | Imputation based on linear regression | 2012 |
ACE-IPF/Noth [27] | Warfarin | Time to: death; nonelective nonbleeding hospitalisation; or drop in FVC ≥10% | HR from a Cox model | 48 weeks | Log-rank test | No imputation | 2012 |
CAPACITY/Noble [3] | Pirfenidone | Change from baseline to week 72 in the FVC % predicted | Differences in treatment group means Categorical change in FVC | 72 weeks | Ranked ANCOVA | Missing values owing to death were given the worst ranks in the primary analysis (and imputed to 0 in other analyses) Missing data for other reasons were imputed by the average of the 3 closest patients | 2011 |
BUILD-3/King [28] | Bosentan | Time to IPF worsening or all-cause death IPF worsening defined as: drop in FVC ≥10%; drop in DLCO ≥15%; or acute exacerbation of IPF | HR from a Cox model | Event driven | Kaplan–Meier and log-rank test | No imputation | 2011 |
Imatinib-IPF/Daniels [29] | Imatinib | Time to disease progression defined as: drop in FVC ≥10%; or death | HR from a Cox model | 96 weeks | Kaplan–Meier and log-rank test | No imputation | 2010 |
Taniguchi [30] | Pirfenidone | Change from baseline to week 52 in FVC % predicted | Difference in FVC change between treatment groups | 52 weeks | ANCOVA | LOCF | 2010 |
INSPIRE/King [31] | Interferon-γ1b | Overall survival time from randomisation | HR from a Cox model | 77 weeks | Log-rank test | No imputation | 2009 |
Raghu [32] | Etanercept | Change from baseline to week 48 in FVC % predicted, DLCO % predicted and P(A–a)O2 at rest | Differences in FVC, DLCO and P(A–a)O2 at rest Changes between treatment groups | 48 weeks | ANCOVA | LOCF | 2008 |
IFIGENIA/Demedts [33] | Prednisone, azathioprine and N-acetylcysteine | Change from baseline to month 12 in absolute values of DLCO and FVC (mL·year−1) | Difference in DLCO and FVC changes between treatment groups | 12 months | ANCOVA | LOCF | 2005 |
Raghu [34] | Interferon-γ1b | PFS measured from randomisation to either disease progression (FVC decrease ≥10% or P(A–a)O2 increase at rest ≥5 mmHg) or death. | HR from a Cox model | 48 weeks after the 306th patient underwent randomisation | Cox proportional-hazards model Log-rank test | LOCF | 2004 |