Sampling and recruitment
Study One analysed data collected from the “Having a Baby in Queensland” Pilot Survey, conducted by the Queensland Centre for Mothers and Babies (QCMB) in November 2009. The QCMB partnered with the Queensland Registry of Births, Deaths and Marriages (BDM) to recruit participants, based on formal notifications by care providers of all births in Queensland in a two-week period. Women were eligible for participation in the survey if they had a single, live born baby during this period and a complete address recorded (n = 2306).
Eligible women were sent a survey invitation package when their infants were approximately three months old. The invitation package contained a paper version of the survey, a reply-paid envelope and a fridge magnet as a gift. Two weeks after the first package was sent, a follow-up package was posted, regardless of whether women had responded. Confidential sampling via the Registry of Births, Deaths and Marriages meant we were unable to identify, and send tailored reminders to, non-responders specifically. Women were able to complete the survey on paper (and return using the reply-paid envelope), online or over the telephone (by free call with trained interviewers or in any language with an interpreter). In total, 2240 survey packages were delivered (66 packages were returned undelivered) and 693 women responded to the survey, yielding a response rate of 31%. Of the 693 survey respondents, 627 provided sufficient information about their pre-pregnancy weight and height to allow for the calculation of their pre-pregnancy BMI.
The survey and collection methods were approved by the Behavioural and Social Sciences Ethical Review Committee at The University of Queensland (Ethics Clearance #2009001531).
All major institutions throughout Australia offering a medical degree and/or an undergraduate midwifery degree were invited to recruit participants. Administrators at these institutions were asked to forward an email from the researcher to eligible students, inviting them to participate in an online survey. Participants were offered entry into a draw to win a $100 shopping voucher in exchange for participation.
Of 18 medical schools contacted, six consented to inviting their students to participate. Approximately 2200 medical students were invited to participate, and 215 completed the survey, yielding a response rate of about 10%. Of nine midwifery departments, three consented to invite their students, and surveys were administered in person to 22 midwifery students from one institution. Students from the other two institutions completed the survey online. Of 55 midwifery students invited to participate overall, 33 responded, yielding a response rate of about 60%.
The study and collection procedure were approved by a subcommittee of Behavioural and Social Sciences Ethical Review Committee at The University of Queensland (approval # 10-PSYCH-4-114-JM).
Participants were 627 women who gave birth in Queensland in 2009.
Participants were 248 pre-service maternity care providers in Australia, comprising 215 medical students from six universities, and 33 midwifery students from three universities. Medical students were in the final two years of their medical degree, and had completed, or were currently completing, a General Practice rotationb. All midwifery students were in their final (third) year of an undergraduate midwifery degree.
The “Having a Baby in Queensland” survey employed a cross-sectional design, and included retrospective self-reported measures of the quality of maternity care provided, information and support provided, health outcomes, and socio-demographic variables. Measures relevant to the current study are detailed below.
Pre-pregnancy body size
Pre-pregnancy BMI was calculated by dividing participants’ self-reported pre-pregnancy weight in kilograms by the square of their height in metres (weight (kgs)/height (m)2).
Participants’ reported highest level of education was coded from 1 (No school) to 12 (Higher University Degree); a higher score indicated a higher education level. Country of birth was also assessed (“In what country were you born?”). Women’s age at birth was calculated from participants’ reported date of birth, and the date of their baby’s birth. Participants reported their infant’s gestational age at birth (“How many weeks pregnant were you when your baby was born?”). The approximate age of each participant’s baby at survey completion was calculated from the reported date of their baby’s birth and date of survey receipt.
Perceived quality of treatment
Participants responded to a set of statements assessing separately their perceived quality of treatment during pregnancy, during labour and birth, and after birth. Participants were asked, “Thinking about your care during your pregnancy/during your labour and birth/after having your baby, how much do you agree or disagree with the following statements?”, and responded to the same set of statements in reference to each time period. The statements, each of which utilised a Likert response format (from 1 = “strongly disagree” to 5 = “strongly agree”) formed two scales reflecting participants’ “perceived positive treatment” and “perceived negative treatment”.
Perceived positive treatment was derived by averaging participant responses to the following four statements: “My carers treated me with respect”, “My carers treated me with kindness and understanding”, “My carers respected my privacy”, and “My carers genuinely cared about my wellbeing”, with higher scores indicating higher perceived positive treatment. The scale demonstrated high internal consistency for each time period (αs > .89).
Perceived negative treatment was derived by averaging participant responses to the following four statements: “One or more of my carers did not treat me with respect”, “One or more of my carers did not treat me with kindness and understanding”, “One or more of my carers were not open and honest”, and “One or more of my carers did not genuinely care about my wellbeing”, with higher scores indicating more perceived negative treatment. The scale demonstrated high internal consistency for each time period (αs > .90)
A novel survey was developed to assess care providers’ attitudes towards, and perceptions of, normal-weight, overweight and obese pregnant women. Participants first read a hypothetical patient case about a pregnant woman who was either normal-weight, overweight or obese. They were then asked questions about perceptions of, and attitudes towards, caring for the patient described. The survey was administered online to the majority of participants, and randomly generated either the normal-weight, overweight or obese patient case for each participant. The second half of the survey included demographic questions and a measure of weight stigmatising attitudes.
Development of patient case presentation
Participants were presented with a hypothetical case detailing a pregnant woman presenting for an initial appointment in pregnancy, with either a general practitioner (for medical students) or a midwife (for midwifery students). The BMI of the hypothetical patient was manipulated in case presentations such that participants were randomly allocated to read about a pregnant patient with either a normal-weight BMI, overweight BMI, or obese BMI. Descriptions of the patient’s appearance in terms of weight (e.g., “Debbie appears to be overweight”) as well as measures of height, weight and BMI were provided to ensure effective manipulation of the independent variable. A range of information was included to enhance ecological validity and to indicate that the patients were equal on all other indices of health, and were not at risk for other complications. This information was kept consistent across all conditions.
Perceptions of patient self-management and health
Perceptions of patient self-management and health was measured using a 3-item scale, adapted from Hebl and Xu , with a Likert response format from 1 (“Highly Unlikely”) to 7 (“Highly Likely”). Items included “Overall, the patient is healthy”, “Overall, the patient takes care of herself”, and “The patient is self-disciplined”. Scores for the scale were calculated by averaging participants’ scores across items. A higher score was associated with more positive perception of the patient’s self-management and health. The scale demonstrated high internal consistency with the current sample (α= .88).
Attitudes towards caring for patient
Attitudes towards caring for the patient were measured via a 6-item scale adapted from Hebl and Xu , which used the same response format and scoring methods as above. Scale items were “This sort of patient would make me like my job”, “I would have a lot of patience with this patient”, “This patient would annoy me”, “I would have a significant personal desire to help and support this patient”, “Overall, I would feel positive towards this patient”, and “Seeing this patient would feel like a waste of my time”. Scores for the scale were calculated by averaging participants’ scores across items, with the two items endorsing a negative attitude reverse-scored. A higher score was associated with a more positive attitude towards caring for the patient. The scale had high internal consistency in this sample (α= .86).
Gender and age were self-reported.
Participants were asked to report their own height and weight. A continuous measure of participant BMI was calculated, as in Study One.
Weight stigmatising attitudes
Participants’ weight stigma attitudes were assessed using Lewis et al.’s  Anti-Fat Attitudes Test (AFAT). The AFAT provides a measure of an individual’s general level of weight stigmatising attitudes and beliefs, incorporating the extent to which individuals attribute negative characteristics and stereotypes to overweight and obese people. The scale contains items indicative of anti-fat attitudes (e.g., “I’d lose respect for a friend who started getting fat”; “Most fat people are lazy”), with response options from 1 (“definitely disagree”) to 5 (“definitely agree”). Total scale scores were calculated by averaging responses across all items. Six items endorsing positive or neutral attitudes (e.g. “If I were single, I would date a fat person”) were reverse-scored. Overall, higher scores indicated higher weight stigmatising attitudes. The scale had high internal consistency (α= .91).
T-tests and chi-square tests were conducted to determine differences between women who did not report their weight and/or height (excluded from the sample) and women in the final sample. Six hierarchical linear regression analyses were conducted to examine relationships between pre-pregnancy BMI and the dependent variables (perceived positive and negative quality of treatment from care providers in pregnancy, labour and birth, and after birth). Past research has found that pre-pregnancy BMI is negatively related to education , so all analyses routinely controlled for education to ensure that any observed effects were attributable to BMI. In each regression analysis, maternal education level was entered in Block 1, followed by pre-pregnancy BMI in Block 2. Significance was set to p < 0.05 for all analyses. Notably, only a very small number of women (<1%) had a BMI in the underweight range (BMI < 18.5), which precluded any separate analysis of this group.
Surveys administered in person
Surveys were administered in person to 22 midwifery students. Surveys were handed out in such an order that participants were randomly assigned to one of the three BMI conditions. Participants were first given the patient case and all measures except the AFAT. Participants were given the AFAT only after completion and collection of the first part of the survey, to prevent potential response bias for earlier survey questions due to exposure to the AFAT.
Surveys administered online
Participants accessed the online survey via a link in the invitation email. One of three patient cases (varying on patient BMI) was then randomly presented to participants, followed by the survey measures. Participants were unable to return to previous survey pages, to prevent bias associated with awareness of the experimental manipulation when they were presented with the AFAT.
Chi-square tests and one-way Analyses of Variance were conducted to assess differences between experimental groups on demographic characteristics and AFAT scores. The effect of patient BMI on pre-service care providers’ perceptions of, and attitudes towards caring for, the patient, as well as potential moderating effects of anti-fat attitudes and care providers’ BMI, were investigated with two hierarchical moderated regression analyses. In the first regression, perceptions of patient self-management and health was the criterion variable, and in the second, attitudes towards caring for the patient was the criterion variable. Using unweighted effect coding, two patient BMI contrasts were created to allow for the comparison of perceptions and attitudes between the different patient BMI conditions. The first variable contrasted the normal-weight patient BMI condition to those in the overweight and obese conditions. The second variable contrasted the overweight patient BMI condition to the obese patient BMI condition. Mean-centred AFAT scores and participant BMI were entered at Step 1 of the regression analyses, followed by the two patient BMI contrasts entered at Step 2. To explore the moderating effects of anti-fat attitudes, two interaction terms for each of the patient BMI condition contrasts and anti-fat attitudes (AFAT) scores were entered at Step 3. Additionally, to explore any moderating effects of participant BMI, two interaction terms for each of the patient BMI condition contrasts and participant BMI were entered at Step 3. Power analyses indicated that 159 participants were required to have an 80% chance of detecting medium effect sizes , thus the analyses were collapsed across medical and midwifery pre-service care providers to maximise power.