Longitudinal analysis: change in forced expiratory volume in one second (FEV1) over five years and its association with change in apple intake and average apple consumption over the same period
| No. | Additional change in FEV1compared with change at baseline adjusted for: | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| (1) Age, height, age2 and height2 | (2) Model 1 + body mass index and smoking | (3) Model 2 + social class, work exercise and leisure exercise | (4)Model 3 plus total energy intake | ||||||||||||
| Additional change in FEV1 | 95% CI | Additional change in FEV1 | 95% CI | Additional change in FEV1 | 95% CI | Additional change in FEV1 | 95% CI | ||||||||
| Change in the no. of apples eaten per week4-150 | |||||||||||||||
| Decrease >2 | 249 | –21.9 | –76.6 to 32.9 | –25.5 | –79.7 to 28.7 | –23.5 | –77.5 to 30.5 | –23.2 | –77.3 to 30.9 | ||||||
| Decrease ⩽2 | 244 | –6.2 | –61.3 to 49.0 | –10.2 | –64.7 to 44.4 | –6.3 | –60.7 to 48.1 | –6.2 | –60.7 to 48.2 | ||||||
| No change | 579 | 0 | Baseline | 0 | Baseline | 0 | Baseline | 0 | Baseline | ||||||
| Increase ⩽2 | 242 | –12.4 | –67.7 to 43.0 | –25.6 | –80.6 to 29.5 | –20.8 | –75.7 to 34.1 | –20.5 | –75.5 to 34.4 | ||||||
| Increase >2 | 278 | 17.5 | –35.3 to 70.3 | 14.2 | –38.6 to 67.0 | 16.5 | –36.3 to 69.2 | 16.7 | –36.1 to 69.5 | ||||||
| Test for trend† | p = 0.136 | p = 0.164 | p = 0.175 | p = 0.174 | |||||||||||
| Average apple consumption (no. per week) | |||||||||||||||
| <1 | 401 | 0 | Baseline | 0 | Baseline | 0 | Baseline | 0 | Baseline | ||||||
| 1 | 281 | –39.6 | –95.6 to 16.4 | –44.4 | –100.0 to 11.2 | –41.3 | –96.7 to 14.2 | –41.3 | –96.7 to 14.2 | ||||||
| 2–4 | 633 | 47.0 | 1.0 to 92.9 | 25.7 | –21.1 to 72.5 | 31.3 | –15.5 to 78.1 | 31.9 | –15.0 to 78.8 | ||||||
| ⩾5 | 277 | 74.1 | 17.7 to 130.6 | 37.7 | –20.5 to 95.9 | 46.3 | –11.9 to 104.4 | 47.7 | –10.7 to 106.2 | ||||||
| Test for trend‡ | p = 0.004 | p = 0.200 | p = 0.108 | p = 0.098 | |||||||||||
↵4-150 Difference (phase II—phase I) multiplied by 5 and divided by length of follow up. †Test for trend based on ungrouped variable.‡Test for trend based on ungrouped frequency variable.