Longitudinal analysis of the Alternative Healthy Eating Index-2010 and incident non-communicable diseases over 15 years in the 1973-1978 cohort of the Australian Longitudinal Study on Women's Health.

In studies that contain repeated measures of variables, longitudinal analysis accounting for time-varying covariates is one of the options. We aimed to explore longitudinal association between diet quality (DQ) and non-communicable diseases (NCDs). Participants from the 1973-1978 cohort of the Australian Longitudinal Study on Women's Health (ALSWH) were included, if they; responded to survey 3 (S3, 2003, aged 25-30 years) and at least one survey between survey 4 (S4, 2006) and survey 8 (S8, 2018), were free of NCDs at or before S3, and provided dietary data at S3 or S5. Outcomes were coronary heart disease (CHD), hypertension (HT), asthma, cancer (except skin cancer), diabetes mellitus (DM), depression and/or anxiety, and multimorbidity (MM). Longitudinal modelling using generalised estimation equation (GEE) approach with time-invariant (S4), time-varying (S4-S8) and lagged (S3-S7) covariates were performed. The mean (± standard deviation) of Alternative Healthy Eating Index-2010 (AHEI-2010) of participants (n = 8022) was 51·6 ± 11·0 (range: 19-91). Compared to women with the lowest DQ (AHEI-2010 quintile 1), those in quintile 5 had reduced odds of NCDs in time-invariant model (asthma: OR (95 % CI): 0·77 (0·62-0·96), time-varying model (HT: 0·71 (0·50-0·99); asthma: 0·62 (0·51-0·76); and MM: 0·75 (0·58-0·97) and lagged model (HT: 0·67 (0·49-0·91); and asthma: 0·70 (0·57-0·85). Temporal associations between diet and some NCDs were more prominent in lagged GEE analyses. Evidence of diet as NCD prevention in women aged 25-45 years is evolving, and more studies that consider different longitudinal analyses are needed.