The authors regret that there has been a mislabeling of the variable on book reading. The previously labelled categories “never”, “up to 1h”, and “> 1h/d” denoted in the methods section should have been labelled “up to 1h”, “1–2h/d”, and “> 2h/d”, respectively. The same change in labelling of book reading applies to Figure 1 and Table S1 in the online supplement as well as footnote 3 to Table 2. For the electronic media indicators, groups remain as indicated in the methods section and throughout the manuscript. This changes the statements regarding prevalence book reading in the abstract and results section. A total of 39% read/viewed books or were read to for up to 1 h/day, which remains a high proportion in the lowest category but is not as drastic as previously thought. This mislabeling has no effect on the association between book reading and electronic media consumption (mentioned in the results) or sleep quality (displayed in Figure 1 and Table S1). Furthermore, it has no effect on the inverse association of longer duration of book reading with worsening of CSHQ scores between ages two and three years displayed in Table 2. The authors would like to apologise for any inconvenience caused.