Method
Sampling
A stratified purposeful sampling strategy, was used to identify first time and experienced mothers of infants who were completely immunised (for age), incompletely immunised (behind the recommended immunisation schedule), partially immunised (parents chose or advised not to have a specific immunisation) or who had no immunisations. Initially, mothers of children between 14–16 months were approached. This age allowed a range of immunisation experiences to be discussed with participants while minimising the time since the first immunisation. To obtain sufficient numbers of non-immunisers and partial immunisers this age range was broadened to include 3 to 30 months. It was initially proposed that interviewing approximately 8 mothers from each immunisation category would be sufficient to discern patterns of similarity and difference between these categories. If preferred by the mother, both parents could participate in the interview.
Participants
Possible participants were identified by Maternal and Child Health (M&CH) nurses in five metropolitan local government areas in Melbourne, Australia. Nurses were asked to approach mothers fitting the immunisation categories and to include, to the best of their knowledge, mothers of high and low education (< Year 11) and high and low income (held a Health Care Card). Parents were identified as fitting these categories from informal information available to the nurses. Parents from non-English speaking backgrounds whose English was poor were not interviewed. The Nursing Mothers Association group for the area also advertised the study.
Ethics approval was granted by the Royal Children's Hospital Ethics in Human Research Committee. Participation was voluntary, with written consent required. Participants provided informed consent to be interviewed, and for the interview to be taped. Participants were assured that they could stop the interview at any time, could choose not to answer any question if they didn't want to and that their responses would be confidential and transcripts anonymised. The interviewer did not have a dual relationship with the participants (i.e. she was neither a clinician nor provider of health services/care).
Recruiting and interviewing continued until 'saturation' occurred (i.e. no new information was obtained from the interviews) for complete, incomplete and non-immunisers or until no more new parents fitting the categories could be identified, as was the case for partial immunisers.
Over the period of data collection, 94 families were identified as possible participants. Forty-eight interviews were arranged and 45 completed. One participant withdrew consent prior to the interview (she did not believe she had anything to say). Two participants were not at home at the time scheduled for the interview and neither returned follow-up phone calls. Of those not interviewed, 17 fitted categories for which a sufficient number of interviews had been conducted (complete immunisers); 11 were not interviewed due to language difficulties; 12 could not be contacted by the nurses and 6 refused. Interviews were undertaken in 1995–96. For six interviews both mother and father participated in the interviews (three of these were non-immunising families). All interviews were conducted in the participants' homes.
Interview Structure
Semi-structured, one-on-one interviews were used to collect information from parents about their children's health, the experience of illness in the family, their understanding and interpretation of risk and how all of these related to their decision to immunise. The method of one-on-one interviews rather than focus groups, was chosen as the aim was to explore the parents' experiences and path to choosing to immunise or not, rather than a group discussion of the pros and cons of immunisation. To understand the context of parents' decisions to immunise, the interviews covered four themes: (1) how mothers keep their children healthy; (2) experience, familiarity and concerns regarding both vaccine preventable diseases and other diseases; (3) concepts and influences on risk perception and (4) the decisions, experience and outcomes regarding immunisation. The interview began with questions about health as a non-threatening introduction and to place the consequent discussions about disease and disease prevention in the framework or context of health.
Questions about diseases and the family's experience of them were included to explore the relationship between common illnesses experienced by the family and vaccine preventable diseases. What was of interest here was which diseases were familiar, which were unfamiliar, which were to be avoided if possible and which were 'just' childhood illnesses.
Interpretation of Risk Information and Omission Bias
To aid the investigation of how parents understand and interpret risk information the following two hypothetical news items about an influenza outbreak were read to the participants.
Radio News Report 1 Health authorities issued a warning today about a new strain of flu expected this winter. The flu affects the airways, making breathing difficult and causing repeated bouts of coughing. Long term effects of pneumonia and brain inflammation have been reported in some cases. This strain appears to affect adults between the ages of 20–50 years. Several deaths occurred last year from the A-strain of this virus.
Doctors have recommended that all adults should be vaccinated, especially those who are overworked, stressed and tired.
Radio News Report 2 Health authorities issued a warning today about a new strain of flu expected this winter. The flu affects the airways, making breathing difficult and causing repeated bouts of coughing. Long term effects of pneumonia and brain inflammation have been reported in some cases. Several deaths occurred last year from the A-strain of this virus. Many of those who died were children under 5 years. Doctors have recommended all young children should be vaccinated.
The description of symptoms and complications was taken from a description of the complications for pertussis. The doctors' recommendations were written so that the parents could consider themselves 'at risk' in the 1st instance, parents of young children often feeling overworked and tired, and in the 2nd scenario their child/children fitted the 'at risk' group.
Omission bias was examined in this study by asking parents to respond to the following statement:
STATEMENT 1 Some people say they won't vaccinate because they would feel worse if their child died because of the injection than if the child was not immunised and died from the disease.
Participants were asked their opinion and were then read a second statement:
STATEMENT 2 Some people say they would vaccinate because they would feel worse if their child got the disease and died or was brain damaged when they could have had an injection to prevent it.
These statements were used rather than the more complicated scenarios developed by others (e.g.) because it was believed they captured the essential element of omission bias in circumstances with which the parent could identify.
The interview concluded with discussions of the process of deciding to immunise or not and included a discussion of structural and non-structural barriers.
Interview Procedure
All interviews were conducted in the participants' homes at times convenient to them by the first author. Interviews lasted between 45 to 90 minutes. Socio-demographic information including family size and type (two or one parent family), mother's age, parental occupations and education levels was collected at the end of the interview. For those children who were immunised, immunisation status was determined from the immunisation records held by the parent. All interviews were audio-taped and fully transcribed. The interview focussed on the sole or youngest child in the family. Previous experience of disease and immunisations for older children was discussed in terms of its effect on decisions for the youngest child.
Method of Analysis
Interviews were thematically coded after all interviews had been collected. This analysis focussed on determining whether parents' descriptions of their experiences and beliefs were congruent or incongruent with theories of health behaviour, decision-making and risk perception. The coding was undertaken by the first author. No formal testing of the reliability of the coding was undertaken although discussions with colleagues about the analysis and the meanings and patterns derived from this were extensively undertaken.