They are the 40,000 kids living with chronic poverty-related illnesses, who see the inside of hospital more often than a kindergarten or kohanga reo.The obvious place to test the validity of this is the Child Health monitor (now renamed Child Poverty Monitor) which I regard as a reliable statistical source. One with which the writer will be familiar.
There are 40,765 average annual hospital admissions for conditions with a social gradient in children aged 0–14 years - clearly where the "40,000" comes from.
Firstly, admissions are much higher amongst very young children who would yet to have even enrolled at a "kindergarten or kohanga reo".
Secondly, multiple admissions may apply to one individual. A child who is chronically sick may be admitted 2 or more times in the space of a year. Therefore the likely number of individual children admitted is considerably lower than 40,000.
Thirdly, and most obviously, an admission (or discharge) statistic alone tells us nothing about the length of stay.
Despite extensive searching the following is the best information I can find. It applies to all patients however, not just children.
On balance the claim looks like a significant statistical exaggeration and, with respect to the majority of admissions applying to children who wouldn't be of "kindergarten or kohanga reo" age, highly misleading.