Sensitivity analysis
| % FEV1 | % FVC | FEV1/FVC (%) | ||||||||||
| Studies | Participants | Effect estimate | I2 | Studies | Participants | Effect estimate | I2 | Studies | Participants | Effect estimate | I2 | |
| Statistical analysis method | ||||||||||||
| Random effect | 41 | 49 705 | −7.15 (−8.27, −6.03) | 99% | 35 | 49 152 | −9.21 (−11.15, −7.26) | 100% | 45 | 50 711 | −0.27 (−1.63, 1.08) | 100% |
| Fixed effect | −3.18 (−3.20, −3.16) | −4.82 (−4.85, −4.80) | 0.01 (−0.00, 0.02) | |||||||||
| Publication year | ||||||||||||
| Before 2000 | 4 | 244 | −1.59 (−3.19, 0.01) | 0% | 2 | 109 | −2.81 (−6.84, 1.21) | 0% | 2 | 109 | −1.94 (−4.92, 1.05) | 0% |
| 2000–2009 | 14 | 12 142 | −8.14 (−10.96, −5.31) | 93% | 11 | 11 789 | −11.96 (−18.66, −5.26) | 95% | 15 | 12 759 | 1.16 (0.27, 2.05) | 86% |
| 2010–2019 | 23 | 37 319 | −8.00 (−9.48, −6.51) | 99% | 22 | 37 254 | −8.67 (−11.04, −6.29) | 100% | 28 | 37 843 | −0.53 (−2.39, 1.33) | 100% |
| Type-2 diabetes group size | ||||||||||||
| <50 patients | 21 | 1207 | −10.29 (−13.88, −6.70) | 86% | 19 | 1155 | −9.45 (−13.67, −5.22) | 89% | 20 | 1168 | −1.69 (−4.86, 1.48) | 99% |
| ≥50 patients | 20 | 48 498 | −5.68 (−7.01, −4.34) | 99% | 16 | 47 997 | −9.16 (−11.80, −6.51) | 100% | 25 | 49 543 | 0.69 (−0.93, 2.30) | 100% |
| Study quality | ||||||||||||
| Only good quality studies | 32 | 48 201 | −7.95 (−9.30, −6.60) | 99% | 31 | 48 276 | −9.53 (−11.59, −7.47) | 100% | 38 | 49 359 | −0.52 (−2.13, 1.10) | 100% |
| Including predatory journals and grey literature | 57 | 51 845 | −1.27 (−1.83, −0.71) | 100% | 48 | 50 973 | −1.73 (−2.39, −1.08) | 100% | 69 | 53 875 | 0.13 (−0.02, −0.28) | 96% |
| Excluding the highest-weight study | ||||||||||||
| 40 | 40 921 | −7.37 (−8.53, −6.20) | 96% | 34 | 40 368 | −9.39 (−11.42, −7.36) | 98% | 44 | 26 029 | −0.21 (−1.66, 1.23) | 100% | |
| % FEF25–75%1 | % PEF | % DLCO | ||||||||||
| Studies | Participants | Effect estimate | I2 | Studies | Participants | Effect estimate | I2 | Studies | Participants | Effect estimate | I2 | |
| Statistical analysis method | ||||||||||||
| Random effect | 13 | 1034 | −9.89 (−14.42, −5.36) | 76% | 15 | 1291 | −9.79 (−13.42, −6.15) | 92% | 12 | 851 | −7.13 (−10.62, −3.64) | 93% |
| Fixed effect | −9.02 (−10.68, −7.36) | −8.73 (−9.62, −7.85) | −3.79 (−4.78, −2.80) | |||||||||
| Publication year | ||||||||||||
| Before 2000 | 1 | 21 | NA | NA | 0 | 0 | NA | NA | 0 | 0 | NA | NA |
| 2000–2009 | 2 | 82 | −2.80 (−13.85, 8.25) | 42% | 2 | 59 | −15.15 (−36.78, 6.47) | 90% | 6 | 347 | −8.26 (−14.58, −1.94) | 72% |
| 2010–2019 | 10 | 931 | −11.85 (−16.68, −7.03) | 78% | 13 | 1232 | −8.93 (−12.50, −5.36) | 91% | 6 | 504 | −5.99 (−9.80, −2.18) | 76% |
| Type 2 diabetes group size | ||||||||||||
| <50 patients | 9 | 614 | −9.01 (−15.84, −2.19) | 72% | 7 | 436 | −9.47 (−17.73, −1.22) | 93% | 8 | 368 | −7.29 (−11.66, −2.93) | 66% |
| ≥50 patients | 4 | 420 | −11.37 (−18.96, −3.78) | 86% | 8 | 855 | −9.84 (−13.56, −6.12) | 88% | 4 | 486 | −7.01 (−13.43, −0.58) | 84% |
| Study quality | ||||||||||||
| Only good quality studies | 12 | 1013 | −10.60 (−15.03, −6.17) | 75% | 15 | 1291 | −9.79 (−13.42, −6.15) | 92% | 10 | 749 | −7.08 (−10.92, −3.25) | 83% |
| Including predatory journals and grey literature | 18 | 1722 | −0.57 (−0.81, −0.32) | 82% | 21 | 1961 | −0.73 (−1.01, −0.46) | 88% | 13 | 971 | −0.51 (−0.76, −0.26) | 69% |
| Excluding the highest-weight study | ||||||||||||
| 12 | 914 | −10.07 (−16.01, −4.13) | 78% | 14 | 1196 | −10.10 (−14.20, −6.01) | 89% | 11 | 756 | −7.91 (−11.40, −4.43) | 63% | |
FEV1: forced expiratory volume in 1 s; FVC: forced vital capacity; FEF25–75%: forced expiratory flow between 25% and 75% of total lung capacity; DLCO: diffusion capacity of the lung for carbon monoxide; PEF: peak expiratory flow; NA: not applicable.