We appreciate Dr Singh's interest in our recent article, ^{Reference Chen, Hu, Wei, Qin, McCracken and Copeland1} but we believe Dr Singh misunderstands our findings. First, in our two cohort studies we did not say that there was an existing dose–response association between the severity of depression and the risk of dementia. On the contrary, our data have suggested that only the most severe syndromes and cases of depression are a risk factor for developing dementia. Second, we put the two cohort studies from China and the UK in one paper because both of them used the same Geriatric Mental State–Automated Geriatric Examination for Computer Assisted Taxonomy method for the assessment and diagnosis of depression and dementia (in syndromes and cases), which would provide equivalent data between the centres for the proposed analysis. Although there are significant differences in ethnicity, social demographic background and disease patterns between the two populations, the results are consistent, suggesting our findings are robust. Third, it is known that in many Eastern countries elderly people are looked after at home, rather than in nursing homes. Traditionally, the Chinese family would take care of their frail and sick elderly. ^{Reference Chiu and Zhang2,Reference Chen, Wei, Hu, Qin, Copeland and Hemingway3} Both studies aimed to enumerate all the cases of dementia wherever they were found. Thus, we believe the composition of the Chinese cohort is similar to that of the UK cohort in terms of its predisposition to physical and depressive comorbidities. Fourth, we have mentioned in the paper that the small number of dementia cases in the Chinese cohort is one of the study limitations. Nevertheless, the UK cohort data support the Chinese findings. Therefore, we believe the findings are quite reliable. Fifth, in our sensitivity analysis excluding all participants with baseline organic syndrome at Level ⩾3, we did not say that the risk of dementia was associated with depression at Levels 1, 2 and 3 because their HRs are around 1.00 and not significant, but our data have further shown that only the most severe depression (i.e. Level ⩾4) is a risk factor for developing dementia.