Study design and setting
Approval was obtained from the Ministry of Health Ethics Committee. General practices in two NZ regions (Auckland and Midland) were enrolled during 2005 and 2006. Approximately 50% of all NZ children aged 0 to 4 years reside in the study region.
A random sample of practices was recruited, with stratification by region and oversampling of Maori governance practices (these are independent Maori health providers that target services primarily towards Maori and have a Maori management and governance structure).18 There were 11 such Maori governance practices in Auckland and 50 in Midland.
A total of 213 of the 517 practices in the study region were selected. In Auckland this included all Maori governance practices plus a random 29% of the non-Maori governance practices. In Midland it included a random sample of 62% of the Maori governance practices and 61% of non-Maori governance practices. From this sample it was possible to recruit 124 practices: 72 from Auckland and 52 from Midland.
Data collection and measurements
Practice immunisation coverage and timeliness were measured by electronic audit of the immunisation records for all children aged 6 weeks to 23 months. During this study the NZ immunisation schedule included a ‘6 week, 3 month, 5 month’ primary series of diphtheria, tetanus, pertussis (DTaP); polio; Haemophilus influenzae type b (Hib); and hepatitis B vaccines and; at age 15 months, measles, mumps, rubella vaccine plus a booster dose of DTaP and Hib vaccines.2
‘Practice structure and organisation’ was defined as those aspects of the practice that were independent of the characteristics of a specific GP or practice nurse, but were liable to influence the care they delivered when working within that practice. These characteristics were divided into those relevant to all aspects of primary healthcare delivery by the practice and those that were specific to immunisation delivery. For example, characteristics relevant to all aspects of primary care delivery include the age at which children being seen at the practice were registered with the practice and whether or not the practice had staff shortages (that is, unfilled positions for doctors, nurses, or other practice staff). In NZ, when this study was conducted, patients could be seen as casual patients at any practice. Registration with the practice indicated that the practice was the provider of this child's preventive health care and that the practice would maintain a record of the health care received by the child. Examples of characteristics specific to immunisation delivery include whether the practice had specific immunisation clinics and whether GPs at the practice sometimes gave immunisations.
The practice structure and organisation was described from a face-to-face interview with the practice manager, lead GP, or senior practice nurse. Practice population demographics (ethnicity and socioeconomic deprivation) were obtained from the Ministry of Health. Socioeconomic deprivation was measured using the NZDep2001 Index of Deprivation, a small area-based measure that combines nine variables from the 2001 Census that reflect aspects of household material and social deprivation, and is used to group NZ households into socioeconomic quintiles or deciles.19
Immunisation coverage was measured based on the child's age in relation to the immunisation schedule. For example, a child aged 10 weeks was fully immunised if they had received all of the immunisations scheduled to be given at age 6 weeks, and a child aged 6 months was fully immunised if they had received all of the immunisations scheduled to be given at age 6 weeks, 3 months, and 5 months.
Immunisation coverage was defined using the third dose assumption. If the third in a series of vaccine doses was recorded as given, then it was assumed that the previous doses had also been given.20 This assumption results in a small overestimate of coverage that is less than the underestimate that occurs if only recorded doses are counted.20
The Ministry of Health's National Immunisation Register definition of timeliness was used.21 An immunisation was defined as delayed if not received within 4 weeks of the first due date for the 6 week immunisations, and within 6 weeks for 3 month, 5 month, and 15 month immunisations.21
Data analysis
Immunisation coverage and timeliness were described for the children aged 6 weeks to 23 months registered at each practice. The proportion of children fully immunised and the proportion with no delayed immunisation were transformed to facilitate analysis using the arcsin of the square root of the variable. This transformation made the variance constant across the distribution of coverage and number of children at each practice. Examination was carried out for interaction with either region or practice governance for any of the variables associated with coverage. Variables for which such interaction was present were not included in the statistical models.
A general linear model was created with arcsin of the square root of the proportion immunised as the outcome variable. For the regression analyses a base model was created that included region, practice governance, socioeconomic deprivation, and the age, and age at registration, of the children as explanatory variables.
Additional explanatory variables that described practice structure and organisation were then added to this model. Practice structure and organisational variables were retained in the final models of coverage and timeliness if they increased the model's explanatory power or remained significantly associated with practice coverage or timeliness. Analyses were performed using SAS (version 9.1; SAS Institute, Cary, NC, US).
