Table 2.
Logistic, linear (for BMI), and negative binomial (for number of disordered eating behaviours) regression models
| Complete exposure, imputed confounders and outcome (sample A) Odds ratio (95%CI) | Complete cases (sample B) Odds ratio (95%CI) | ||
|---|---|---|---|
| Binge eating | n=6361 | n=1940 | |
| Crude model | 2·23 (1·36–3·69), p=0·0018 | 1·79 (0·96–3·30), p=0·062 | |
| Adjusted model 1* | 2·01 (1·20–3.37), p=0·0082 | 1·57 (0·84–2·95), p=0·16 | |
| Adjusted model 2† | 1·99 (1·18–3.33), p=0.0097 | 1·57 (0·83–2·94), p=0·16 | |
| Adjusted model 3‡ | 2·04 (1·21–3·43), p=0·0081 | 1·60 (0·85–3·01), p=0·14 | |
| Adjusted model 4§ | 1·49 (0·88–2·54), p=0·14 | 1·27 (0·66–2·45), p=0·48 | |
| Purging | n=6361 | n=1926 | |
| Crude model | 1·88 (1·24–2·84), p=0·0031 | 1·99 (1·18–3·35), p=0·0089 | |
| Adjusted model 1* | 1·74 (1·13–2·68), p=0·013 | 1·85 (1·08–3·15), p=0·025 | |
| Adjusted model 2† | 1·73 (1·12–2·67), p=0·013 | 1·85 (1·08–3·15), p=0·025 | |
| Adjusted model 3‡ | 1·76 (1·14–2·72), p=0·011 | 1·85 (1·08–3·16), p=0·025 | |
| Adjusted model 4§ | 1·51 (0·96–2·36), p=0·071 | 1·85 (1·05–3·23), p=0·032 | |
| Excessive exercise | n=6361 | n=1929 | |
| Crude model | 0·84 (0·40–1·78), p=0·65 | 0·53 (0·19–1·47), p=0·22 | |
| Adjusted model 1* | 0·79 (0·37–1·71), p=0·55 | 0·50 (0·18–1·39), p=0·18 | |
| Adjusted model 2† | 0·79 (0·37–1·71), p=0·55 | 0·49 (0·18–1·38), p=0·18 | |
| Adjusted model 3‡ | 0·80 (0·37–1·73), p=0·57 | 0·49 (0·18–1·39), p=0·18 | |
| Adjusted model 4§ | 0·71 (0·32–1·57), p=0·39 | 0·46 (0·16–1·32), p=0·15 | |
| Fasting | n=6361 | n=1943 | |
| Crude model | 2·25 (1·64–3·09), p<0·0001 | 2·33 (1·58–3·44), p<0·0001 | |
| Adjusted model 1* | 2·09 (1·49–2·94), p<0·0001 | 2·26 (1·51–3·40), p<0·0001 | |
| Adjusted model 2† | 2·05 (1·46–2·89), p<0·0001 | 2·25 (1·50–3·38), p<0·0001 | |
| Adjusted model 3‡ | 2·10 (1·49–2·96), p<0·0001 | 2·25 (1·50–3·39), p<0·0001 | |
| Adjusted model 4§ | 1·65 (1·15–2·38), p<0·0001 | 1·94 (1·26–2·97), p=0·0025 | |
| Any disordered eating behaviour | n=6361 | n=1898 | |
| Crude model | 1·92 (1·46–2·52), p<0·0001 | 2·06 (1·41–2·90), p<0·0001 | |
| Adjusted model 1* | 1·82 (1·35–2·44), p<0·0001 | 1·97 (1·37–2·83), p<0·0001 | |
| Adjusted model 2† | 1·80 (1·34–2·41), p<0·0001 | 1·97 (1·37–2·83), p<0·0001 | |
| Adjusted model 3‡ | 1·83 (1·36–2·46), p<0·0001 | 1·98 (1·37–2·87), p<0·0001 | |
| Adjusted model 4§ | 1·50 (1·10–2·03), p=0·010 | 1·70 (1·16–2·49), p=0·0070 | |
| Number of disordered-eating behaviours¶ | n=6361 | n=1898 | |
| Crude model | 0·58 (0·32–0·84), p<0·0001 | 0·54 (0·19–0·88), p=0·0021 | |
| Adjusted model 1* | 0·49 (0·23–0·75), p<0·0001 | 0·46 (0·13–0·79), p=0·0058 | |
| Adjusted model 2† | 0·48 (0·22–0·74), p<0·0001 | 0·46 (0·14–0·79), p=0·0054 | |
| Adjusted model 3‡ | 0·49 (0·24–0·75), p<0·0001 | 0·48 (0·16–0·80), p=0·0036 | |
| Adjusted model 4§ | 0·32 (0·06–0·57), p=0·017 | 0·38 (0·05–0·71), p=0·024 | |
| BMI¶e | n=6361 | n=2957 | |
| Crude model | 0·026 (–0·06–0.11), p=0·54 | 0·016 (–0·09–0·12), p=0·77 | |
| Adjusted model 1* | 0·006 (–0·07–0·09), p=0·90 | –0·011 (–0·12–0·10), p=0·83 | |
| Adjusted model 2† | –0·001 (–0·08–0·09), p=0·99 | –0·018 (–0·13–0·09), p=0·74 | |
| Adjusted model 3‡ | –0·003 (–0·06–0.05), p=0·92 | 0·019 (–0·05–0·09), p=0·58 | |
| Adjusted model 4§ | 0·001 (–0·06–0·06), p=0·99 | 0·025 (–0·04–0·09), p=0·47 | |
Models tested the associations between suspected or definite (vs none) psychotic experiences at age 13 years and disordered eating behaviours and body-mass index (BMI) at age 18 years. Data presented are odds ratios (95% CI) from complete case analyses (sample B) and analyses done on a sample of children for whom we had with complete exposure data and imputed confounders and outcomes (sample A, main analyses).
Adjusted model 1: adjusted for maternal: age, marital status, education, social class, depression; and child's sex.
Adjusted model 2: adjusted model 1 plus autistic traits at age 7 years.
Adjusted model 3: adjusted model 2 plus BMI at age 13 years.
Adjusted model 4: adjusted model 3 plus depressive symptoms at age 13 years.
Data given for this variable are coefficients (95% CI).