2.1. Variables used

The key outcome variables used were: i) facing a barrier to seeing a GP; and ii) having had a hospital admission for a range of non-injury related conditions. ‘Facing a barrier to seeing a GP’ was defined as a positive response to the question “In the last 12 months, has there been any time when [child] needed to see a GP or family doctor about his/her health, but didn’t get to see any doctor at all?”. The same question was asked at 24 and 54 months.

Having been admitted to hospital in the previous year was reported at 54 months. The relevant questions for hospital admissions excluded injuries, and related to the following specific health conditions: non-food allergies; hay-fever; ear infections; asthma; whooping cough (pertussis); other respiratory disorders including chest infections, bronchiolitis, bronchitis, pneumonia; cough lasting more than four weeks; wheezing in the chest; gastroenteritis (three or more watery or looser-than-normal bowel movements or diarrhoea within a 24 h period); eczema or dermatitis; throat infection or tonsillitis; skin infections. ASH-related conditions were defined as a hospital admission in the previous 12 months of one or more of the following: ear infection, asthma/wheeze, whooping cough, gastroenteritis, eczema/dermatitis, skin infections or throat infection/tonsillitis. These were chosen as being the categories that were most closely related to the conditions used to define childhood ASH (Health QualitySafety Commission).

The sex of the child was reported by the mother at age 9 months. Maternal age was self-reported when the child was age 54 months. At that time, maternal age ranged from 20 to 50 years and was analysed in five-year age bands. At that same timepoint, the child’s ethnicity was reported by the mother on behalf of the child.

In this study, we report ethnicity as Māori, Pacific People, and NZ European/Other. People who reported their ethnicity as “New Zealander” were included in the ‘Other’ group, unless they also identified as Māori or Pacific, as the majority of people who report their ethnicity as New Zealander are New Zealand Europeans (Cormack & Robson, 2010). For analysis, the total Māori population was compared with non-Māori population and the total Pacific population compared with non-Pacific population. The New Zealand European group was defined as people who identified as New Zealand European, but did not identify as Māori or Pacific ethnicity. This group does not include “Other” ethnicities.

Relevant primary health care measures used were i) having a regular GP or practice; ii) whether the mother reported that the child’s visit to the GP usually incurred fees; and iii) number of GP visits in the previous 12 months. Health was measured using maternal-reported child health at 24- and 54-months, in five categories (excellent, very good, good, fair, poor).

Various measures of social determinants of health were used. These included an area-based measure of material deprivation (Salmond & Crampton, 2012), based on place of domicile using measures from the 2006 (at age 24 months) and 2013 (at age 54 months) censuses, known as NZDep2006 and NZDep2013; maternal employment at age 24 months; self-reported standard of living at age 24 months in five categories (low, fairly low, medium, fairly high, high); self-reported income and sufficiency of income at age 24 months in four categories (not enough, just enough, enough, more than enough); and overcrowding, defined as more than one person per room, measured at 9 months.

We included maternal experience of racism in the health service, reported when the child was 24 months old as an additional potential confounder or mediator of the relationship between barriers to primary care and later hospitalisations. This measure was defined as a mother answering that they had been treated unfairly in/by the health service because of their ethnicity.