1 BACKGROUND Many Australians' health care practices include the use of complementary medicine (CM), including complementary medicine products (CMPs) like dietary supplements and herbal medicines.1-5 During pregnancy and lactation, some CMPs like iodine and folic acid supplements have been shown in clinical trials to improve outcomes for the mother and baby, and so are routinely prescribed or recommended as part of mainstream evidence-based biomedical maternity care.6 These recommendations are also endorsed by CM practitioners in Australia.7-9 As such, the definition of a CMP can vary according to context, and the prescriber or recommender. Self-prescription of CMPs, including herbal medicines and dietary supplements like folic acid, is also common.8, 10 The following operational definition of CMPs was used in this study, and was based on documents from the Pharmacy Guild of Australia,11 the Australian Government Department of Health and Ageing,12 the Australian National Health Medical Research Council,13 The Royal Australian and New Zealand College of Obstetricians and Gynaecologists [RANZCOG]6 and several Australian research papers and textbooks,7-9, 14, 15 and aligns with international definitions of CMPs16-20  (File S4):

CMPs are products like herbal medicines and vitamin and mineral supplements and probiotics. Some vitamins and minerals (e.g., iron, folate or iodine supplements) may be recommended by your doctor or other health care practitioner and have a scientific evidence base. Other CMPs like some herbal medicines may have traditional uses but may not have been scientifically researched.

Women's use of CMPs in pregnancy and lactation has been associated with women's desire to positively enhance their own health and that of their babies,9, 21, 22 including the treatment of common conditions of pregnancy (e.g., nausea and vomiting of pregnancy, preparation for labour)9, 10, 21 and lactation (e.g., blocked ducts, mastitis and concerns with breastmilk supply).9, 22, 23 Mothers' CM use has also been associated with having tertiary levels of education7, 10, 23-29 and higher income or employment levels,10, 23, 25, 28, 30 and being nonsmokers.26, 29, 30 Previous Australian qualitative research found good functional health literacy levels to be linked to CMP use in pregnancy and lactation.22, 31, 32 Previous research has also noted that women's use of CM and CMPs helps facilitate their self-determination, autonomy and control over health during pregnancy and lactation.33-38

The Health Locus of Control Form C (MHLC-C)39, 40 measures Internal, Doctors (operationalized as health care practitioners [HCPs] in this study), Other People or Chance Locus of Control beliefs,40, 41 with results indicating where a respondent believes control of her health, and in respect to pregnancy and lactation, where responsibility for the health of her unborn or breastfeeding children, lies.41 Studies focusing on Health Locus of Control (HLOC) beliefs39, 42 in pregnancy43, 44 and breastfeeding45 have found that higher Internal HLOC beliefs are associated with several different aspects of health and self-efficacy, including positive self-care behaviours in mothers with gestational diabetes46; choosing to birth in midwifery-led, low-intervention birthing units over obstetrician-led medical wards44; breastfeeding self-efficacy and success45; and positive mental health pre-47 and postnatally.45 In general populations, higher Internal HLOC beliefs in healthy adults have been associated with increased use of CM therapies and CMPs,48-51 and healthy behaviours including regular exercise.48 Previous research9, 52-56 has revealed that pregnant and breastfeeding women's use of CMPs is linked to beliefs that CMP use is health-promoting for both themselves and their babies; however, the HLOC beliefs of mothers using CMPs have not been measured before. Measuring HLOC beliefs in women who use CMPs during pregnancy and lactation could help confirm the type/s of control beliefs associated with this use, and confirm whether self-efficacy, or dependence on others, chance or HCPs influences women's CMP use during pregnancy. This would help and inform the practices of HCPs working in maternity care around the use of CMPs.

Maternal health literacy can be described as ‘the cognitive and social skills that determine the motivation and ability of women to gain access to, understand and use information in ways that promote and maintain their health and that of their children’.57 Good health literacy encourages healthy pregnancy and postpartum behaviours, and is a vital component of understanding and using the information to make health-promoting decisions, including decisions about medicines used.9, 58 Despite the potential impact that poor health literacy could have on many aspects of health and health care choices during pregnancy and lactation,58 the effects of maternal health literacy on women's reproductive health and CMP use is under-researched.9 Previous research has confirmed the high prevalence of CMP use in pregnancy and lactation7, 8, 10, 23, 24, 26, 59 and raised concerns regarding maternal health literacy and the ability to make safe decisions regarding CMP use in pregnancy and lactation.1, 60-62 Nevertheless, this previous research has not included measurements of health literacy in pregnant and breastfeeding respondents with respect to the use of CMPs.

As part of a larger, national cross-sectional study investigating factors influencing women's decision-making regarding the use of CMPs in pregnancy and lactation, this paper reports on the women's health literacy levels, HLOC beliefs and the types of CMPs used and compares the use of CMPs by the pregnant and breastfeeding cohorts.

2 METHODS 2.1 Ethical approval

Ethical approval for the study was obtained from The University of Sydney Human Research Ethics Committee, approval number 2018/1010. The survey questionnaire was completed and submitted online, and completion of the questionnaire was taken as consent to participate. The survey questionnaires were completed anonymously, and no identifying data such as name or date of birth were collected. The Participant Information Statement (PIS) informed participants of these considerations. Additionally, the PIS clearly stated that participants could withdraw their consent to participate in the study at any time before submitting their completed surveys, but that because all the data were collected anonymously, it would not be possible to extract submitted data once completed surveys had been submitted. The PIS also outlined an incentive to participate: At the end of the survey, respondents were given the option of entering their email addresses to go into the draw to win an iPad mini® and/or to receive a summary of the overall results of the study. If they chose either of these options, they were automatically redirected to a separate survey so that their email addresses were not linked to the information gathered in the study survey.

2.2 Survey design

A national, cross-sectional, online, anonymous, self-administered questionnaire was designed and set up using the Qualtrics63 platform. The questionnaire (File S1) comprised of 70 questions and took approximately 20 min to complete. The completed Checklist for Reporting Results of Internet E-Surveys (CHERRIES)64, 65 appears in File S2.

2.3 Inclusion criteria

The inclusion criteria for the study were aged 18 years or over, currently pregnant and/or breastfeeding, currently taking one or more CMPs and living in Australia. Three eligibility screening questions were used at the beginning of the survey.

2.4 Patient or public contribution

This study was designed by a multidisciplinary team of HCPs and researchers without direct public involvement. However, the survey items were informed by data from earlier qualitative research with the same population.22, 31, 32 The pilot questionnaire was designed by the research team, all of whom have experience of pregnancy and motherhood, and three of whom have the clinical experience of working with pregnant or breastfeeding women as a naturopath (L. A. J. B.), pharmacist (P. A.) and midwife (L. B.), respectively. The questionnaire was piloted by several lay-women volunteers who fulfilled the study inclusion criteria. Volunteers piloted the questionnaire on tablets, mobile telephones and laptops. Each volunteer trialled the questionnaire twice (once as a pregnant participant and once as a breastfeeding participant). They were invited to comment on its usability and relevance, and their data were not included in the final data analysis. Volunteers were asked to comment on the ease and usability of the questionnaire, as well as their understanding of the questions, which helped confirm face validity.66 Feedback was generally positive, with volunteers reporting that the survey made sense and was easy to understand, flowed well and covered topics they expected in a survey on CMP use in pregnancy and breastfeeding (content validity). The participants did not suggest any wording changes to the questions, nor did they suggest any additional questions. Furthermore, their understanding of the questions and purpose of the study was aligned with our understanding. It took between 17 and 25 min for each of the volunteers to complete the questionnaire. The first 20 completed questionnaires were also examined to ascertain how long it took for respondents to complete the survey. The Qualtrics data showed that the minimum length of time taken was 14 min and the maximum time was 30 min (average time was 22 min).

2.5 Sample size calculations

Sample sizes were based on calculated populations of pregnant and breastfeeding women. The number of registered births in Australia in one year was used as a proxy number for the population of Australian pregnant women (n = 311,104).67 Data reporting the number of Australian infants receiving any breastmilk were used as a proxy number for the number of breastfeeding mothers in Australia (n = 163,478).68 Prevalence sample size calculations with finite population corrections were performed using the online tool http://sampsize.sourceforge.net/iface/ for all outcome variables. The populations for pregnancy and breastfeeding were calculated separately. All calculations used a precision of 5% and a confidence interval of 95%. The calculated sample sizes were 384 pregnant respondents and 384 breastfeeding respondents (768 complete surveys).69

2.6 Recruitment

Recruitment occurred entirely online and primarily through posts generated from a Facebook page specific to the research project, and have been described elsewhere.70 Paid promoted posts (‘boosted’ posts) were used to advertise the study to potential participants. Purposive and snowball recruitment occurred by requesting posts to be shared on relevant Australian Facebook pages and by sharing nonboosted posts through the research team's own social network connections. All posts contained a link to the survey in the Qualtrics platform. No Internet Protocol (IP) addresses or other identifying data were retained as part of the survey data. The survey recruitment period lasted 10 weeks (6 July–17 September 2019).

2.7 Measures

The complete survey is presented in File S1. The following sections outline the specific sections relevant to this paper.

2.7.1 Demographic characteristics

Demographic questions included age, smoking status, pregnancy or breastfeeding status, number of children, gestational age of the child (pregnant participants), age of the breastfeeding child (breastfeeding participants), marital status, postcode of residence (to assess rurality), weekly household income, education levels, country of birth of the respondent and countries of birth of her parents and  the main language spoken at home.

2.7.2 CMP use

The operational definition of CMPs (see Section 1, 1) was provided to respondents at several points in the survey. Respondents were asked to indicate the dietary supplements and/or the herbal medicines that they currently consumed from lists of CMPs commonly reported as being used in pregnancy or lactation.22, 31, 32 An ‘other’ category with the option of including free-text responses was included.

2.7.3 Health literacy

Health literacy levels were measured using two validated health literacy tools: the single-item health literacy screening question,71 which measures respondents' risk of inadequate health literacy using a single, simple question, and the Newest Vital Sign (UK version),72 which measures functional health literacy levels in under 3 min using six questions about a nutrition label. Use of the two validated health literacy tests helped confirm the consistency of the results by facilitating comparisons between respondents' risk of inadequate health literacy and their functional health literacy skills.

2.7.4 Health locus of control beliefs

The validated 18-item MHLC-C39, 40 was used to test whether high Internal, Doctors (operationalized as ‘Health Care Practitioners’), Other People, or Chance Locus of Control health beliefs influenced respondents' CMP use decision-making in pregnancy and lactation. The MHLC-C was developed to be adapted for use with people living with any disease- or health-related condition.39, 41 For the purposes of this study, ‘health and well-being during pregnancy’ and ‘health and well-being as a breastfeeding mother’ were substituted for the word ‘condition’ in the MHLC-C for the pregnant and breastfeeding participants, respectively.

2.8 Data analysis

Data were analysed using IBM SPSS Statistics V24 and Excel. Data were screened and incomplete surveys were removed as per the protocol, which outlined that incomplete surveys would be removed before analysis, pending the receipt of at least 768 complete surveys to enable meaningful data analysis73 (see sample size calculations). Surveys marked ‘complete’ in Qualtrics, indicating that the respondent had progressed through all 70 survey items, were included in analyses, provided that at least 75% of the items were completed. Descriptive analyses, followed by χ2 tests, were carried out for all demographic, health literacy and CMP use data to examine differences between the pregnant and breastfeeding respondents. Missing data were not included in the statistical analyses. Statistical significance was defined as a p < .05.

The research hypotheses tested were that there would be no statistically significant differences between the two cohorts (pregnant and breastfeeding women) in the total number of dietary supplements or herbal medicines taken; that both cohorts would be similar in their functional health literacy levels and that there would be no differences between the two groups in the numbers of women at risk of inadequate health literacy; and that both cohorts would have similar HLOC scores for all four subscales. Poisson regression analysis was performed to model the count data for dietary supplements and herbal medicines, respectively, to observe whether there were significant differences in the numbers taken between the pregnant and breastfeeding respondents.

2.8.1 Health literacy levels

For the single-item health literacy screening question How confident are you filling out medical forms by yourself?,71 respondents answering ‘somewhat’ or ‘a little bit’ or ‘not at all’ were considered to be at risk of inadequate health literacy. Those answering ‘extremely’ or ‘quite a bit’ confident were not considered to be at risk of inadequate health literacy.71 For the Newest Vital Sign, respondents who scored 0–1 correct (out of six questions) were considered to have a high likelihood of limited functional health literacy skills.72, 74 Those who scored 2–3 were considered to be at risk of inadequate functional health literacy skills, and those who scored 4–6 correct were considered to have adequate functional health literacy skills.72, 74

2.8.2 Health locus of control beliefs

Means for each subscale of the MHLC-C were calculated for the two cohorts, hence providing scores on the original 1–6 subscales. To examine differences between the results for the breastfeeding and pregnancy cohorts, and calculate estimated marginal means of measure, a repeated-measures analysis of variance analysis was performed for the four HLOC subscales.

3 RESULTS 3.1 Responses collected

A total of 1418 women were enroled in the survey. Of these, 168 respondents were excluded as they did not fulfil the eligibility criteria, and a further 440 incomplete surveys were removed. A total of 810 completed surveys (57.1%) were collated for analysis.

3.2 Demographics

Of the 810 responding women, 325 (40.1%) were currently pregnant; 456 (56.3%) were currently breastfeeding; and 29 (3.6%) were currently both pregnant and breastfeeding. For all data analyses, the respondents who were currently both pregnant and breastfeeding were included in the pregnant respondents' group, resulting in two sub-samples (cohorts) being ‘currently pregnant’ (n = 354, 43.7% of the total sample) or ‘currently breastfeeding’ (n = 456, 56.3%). Almost half of the respondents (n = 363, 44.8%) reported having two or more children, and 369 (45.6%) were pregnant with (n = 161, 22%) or breastfeeding their first child (n = 208, 28.4%; Table 1).

Table 1. Pregnancy- and breastfeeding-related characteristics of the sample Number of respondents (n) Relative frequency (%) Pregnant or breastfeeding status (n = 810) Currently pregnant 325 40.1 Currently breastfeeding 456 56.3 Currently pregnant and breastfeeding 29 3.6 Total 810 100 If currently pregnant (includes those currently pregnant and breastfeeding) (n = 354) First trimester (0–12 weeks) 58 16.4 Second trimester (13–27 weeks) 140 39.5 Third trimester (28–42 weeks) 156 44.1 Total 354 100 If currently breastfeeding, age of breastfed child (n = 456) 0–2 months 104 22.8 3–5 months 106 23.2 6–8 months 67 14.7 9–11 months 43 9.4 12–15 months 47 10.3 16–18 months 19 4.2 19–23 months 30 6.6 Over 2 years old 40 8.8 Total 456 100 Number of children (n = 810) Pregnant with her first child 161 22.0 Breastfeeding her first child 208 28.4 2 Children 270 36.9 3 Or more children 93 12.7 Total 732 100 Missing data 78

Respondents ranged in age from 19 to 53 years, with the mean age being 33.8 years (SD = 4.6) and the median age being 34.0 years. Other demographic data are summarized in Table 2. There were no significant differences between the demographic characteristics of the two cohorts, except for income levels, where pregnant respondents had significantly higher income (Pearson's χ2 = 16.430, p = .021; Table 2).

Table 2. Demographic characteristics of the sample Pregnant respondents (n = 354) Relative frequency (%) Breastfeeding respondents (n = 456) Relative frequency (%) Whole sample (n = 810) Relative frequency (%) Pearson's χ2 value p Value Smoking status Currently smokes 4 1.2 5 1.3 9 1.2 35.043 0.371 Does not currently smoke 322 98.2 390 98.2 712 98.2 Prefer not to respond 2 0.6 2 0.5 4 0.6 Total 328 100 397 100 725 100 Missing data 26 59 85 Marital status Single 5 1.5 6 1.5 11 1.5 1.603 0.659 Married or in de facto relationship 322 97.3 391 97.8 713 97.5 Separated or divorced 1 0.3 2 0.5 3 0.4 Other 3 0.9 1 0.3 4 0.5 Total 331 100 400 100 731 100 Missing data 23 56 79 Highest education level Year 10 (school certificate) 4 1.2 5 1.3 9 1.2 2.718 0.91 Year 12 (high school certificate) or equivalent 9 2.7 10 2.5 19 2.6 Certificate 1–4 13 3.9 23 5.8 36 4.9 Diploma 24 7.2 31 7.8 55 7.5 Associate diploma 5 1.5 8 2 13 1.8 Bachelor's degree 126 38 156 39 282 38.5 Postgraduate studies at university 149 44.9 163 40.8 312 42.6 Other 2 0.6 4 1 6 0.8 Total 332 100 400 100 732 100 Missing data 22 56