Design
We used a mixed methods design, which incorporated a cross-sectional online survey and qualitative interviews with women with ARDs and health professionals. To enable comparisons between this UK study and a previous Australian study [2], we used similar sampling methods and inclusion criteria, and included a modified version of the Educational Needs Assessment Tool [24, 25] to assess information needs.
Online cross-sectional survey
Participants and recruitment
The survey was made available using the Bristol Online Surveys system. The patient survey was advertised through the study website, social media (Twitter & Facebook), via UK arthritis patient organisations (Lupus UK, Arthritis Care, Vasculitis UK), peer-support groups (Facebook groups for people trying to conceive/pregnancy in Lupus and vasculitis), and online networks for parents (Netmums and Mumsnet). We also used Facebook and Twitter advertising systems to promote the study. To facilitate recruitment, we offered the following incentives: a donation of 50p for each questionnaire completed to UK arthritis charities, and an option to enter a prize draw to win a £100 in gift vouchers.
Inclusion criteria
Women aged 18–49 years, who have an ARD (i.e. inflammatory arthritis or auto-immune connective tissue disease for which people would normally be under the care of a rheumatologist), and were: planning to become pregnant in the next 5 years; and/or currently pregnant; and/or had been pregnant within the last 5 years; and/or had a child (or children) under 5 years of age.
Exclusion criteria
Disease not classified as an ARD (e.g. joint hypermobility, fibromyalgia).
Measures
Information needs
A modified version of the Educational Needs Assessment Tool (ENAT) [24, 25] was used, which is a validated measure with 39 items to assess educational needs in seven domains; pain management, movement, feelings, the arthritis process, treatment from health care professionals, self-management and support from others. An international study [24] demonstrated that the ENAT is a valid tool for identification of information needs relating to rheumatic diseases, with high internal consistency. The items are scored on a five point Likert scale, providing total score ranging from 0 (lowest educational need) to 156 (highest educational need) [24]. While uploading the modified ENAT for use in the current study to the online survey software, one response category (‘fairly important’) was omitted, and consequently items were scored on a four-point scale: 1-not at all important, 2-a little important, 3-very important, 4-extremely important. To retain as much comparability as possible with previous studies, the individual item scores were transformed from a four point (1–4) to a five point (0–4) scale prior to calculation of the total score, so that its overall range would correspond with the original ENAT (0–156). The subscale total scores were Rasch transformed to provide interval rather than ordinal level data [25].
Additional items were included in the information needs section of the survey, using the four-point Likert scales to assess information needs in relation to: sex and relationships, contraception, preparation for pregnancy, how to increase chances of getting pregnant naturally, fertility treatments, options for giving birth, managing pain during childbirth, and breastfeeding. These items were developed by the research team based on the educational needs identified through previous studies [2, 10, 11], and guided by two Patient and Public Involvement representatives (both women with young children who had ARDs) who highlighted which issues were most important to them from a patient perspective. The patient representatives also requested items be included on the use of transcutaneous electrical nerve stimulation (TENS) as this may be useful for pain management, and how they could get access to their test results, which could prove difficult between appointments. The additional items from the family planning, pregnancy, and early parenting were scored separately from the original ENAT items and were not included in the total ENAT score.
Disease-related quality of life
This was assessed using the Arthritis Impact Measurement Scale Version 2 Short Form (AIMS2-SF) [26]. The AIMS2-SF is a validated 26-item measure with five factors; physical symptoms, mobility, role (work), social interaction, and affect. The AIMS2-SF has similar psychometric properties to the AIMS2, good test-retest reproducibility and sensitivity to change. Items are scored on a 5-point Likert scale from 0 to 4, and in each component scores are normalised so that they range from 0 (perfect health to 10 (worst possible health) [26]. Thus, higher scores indicate a greater impact of arthritis on each of these domains.
Lived experience and expressed support needs
Using open-response items, participants were asked what they found: i) most challenging; ii) most helpful, and; iii) what support they would have wanted while planning a family, being pregnant, or having young children. Participants were asked whether they were currently having, had previously had, or wanted the following types of support: access to a health professional to act as their main point of contact and care coordinator; physiotherapy; opportunity to talk to other people with similar experiences and to get advice (i.e. peer-support); talking therapies (e.g. counselling, Cognitive Behavioural Therapy); alternative and complementary therapies (e.g. acupuncture, aromatherapy, herbal remedies). These items were developed by the research team based on sources of information and support for women with long-term illnesses while they are building a family that were identified in the literature [2, 4, 10, 21]. The items were reviewed by two patient representatives to assess relevance, clarity and acceptability of the questions.
Clinical and demographic information
The survey included questions on type of ARD (drop down list); years since onset of ARD; current medication (drop down list); co-morbidities (open text), and; family situation (currently pregnant, planning to try to get pregnant within the next 5 years, and/or had a pregnancy in the last 5 years, had children already, and if so, how many and what their ages are). Demographic data were collected on date of birth, highest educational qualification, geographical location (postcode), marital status, ethnicity, and current employment status.
Survey data analysis
Analysis of the quantitative data was carried out using SPSS v23. Analysis was primarily descriptive, providing an overview of the information and support needs. To identify differences in information needs (ENAT) and quality of life (AIMS2-SF) of women by family status (had or did not have children) and disease group, independent t-tests were carried out, and 95% confidence intervals were calculated. Between-group differences in support received/desired, which were binary categorical variables, were calculated using Chi-square tests. To ensure that there were sufficient numbers for analysis, and due to the differences between rheumatic diseases in disease processes and pregnancy outcomes [1, 4, 5], diseases were broadly categorised as: inflammatory arthritis (rheumatoid arthritis, psoriatic arthritis, ankylosing spondylitis, idiopathic juvenile arthritis, and non-specific inflammatory arthritis), connective tissue diseases (Systemic Lupus Erythematous, systemic sclerosis, and non-specific autoimmune connective tissue disease), and vasculitis. The Holm-Bonferroni correction for multiple comparisons can be used on both parametric and non-parametric tests [27]. The Holm-Bonferroni correction included both the t-tests and Chi-square tests as these tests were conducted on the same data set. Using this method, alpha was set at 0.005. Open text data from the survey were coded thematically using an inductive approach to identify frequent, dominant, and significant themes that emerged from the data [28].
Qualitative interviews
We adopted a person-centered ethos. As women were being asked about emotional and complex issues in this study, a flexible narrative approach was used to encourage them to talk about their ‘lived experiences’ in their own words, focusing on things that were important to them, rather than being guided by a researcher-generated topics [29]. A timeline-assisted method was used, where participants were asked to create a visual representation of their histories before the interview [30]. Women were sent a ‘What to expect’ sheet was posted to participants along with stationary items, which provided guidance on some of the topics that were of interest to the research team (see Additional file 1). The timelines were used as an elicitation tool in interviews to provide cues and prompt discussion [30]. A topic guide was not used by the researchers during the interviews as the objective of the interviews was to learn about the lived experiences of women, rather than to determine the frequency of predetermined events [31]. The use of participatory approaches such as this in qualitative research can empower participants by allowing them to navigate the conversation, increase their level of comfort in discussing sensitive topics, provide positive moments and opportunities for closure, and can thereby improve the quality of data collected [30, 32].
Participants
Women who had expressed interest in being contacted for an interview through the survey were purposively sampled based on their family situation. The aim was to achieve a broadly equal representation of women who were: (i) thinking about getting pregnant; (ii) currently pregnant, or; (iii) had young children, so that the views of women who were at different stages of starting a family could be captured. Women were contacted through e-mail or telephone, based on the contact details provided in the survey. Women who expressed interest in being interviewed were sent a study information pack containing participant information sheet, consent form and stamped return envelope.
Healthcare professionals were identified through professional networks (e.g. Welsh Arthritis Research Network) and were purposively sampled for key professional groups who were involved in the care of women with ARDs in primary care (GPs), secondary care (Rheumatologists, Nephrologists, and Nurse Specialists), and maternity and children’s services (Midwives and Health Visitors – i.e. National Health Service nurses who provide advice and assistance to parents with young children). Healthcare professionals who expressed an interest in participating were e-mailed a participant information sheet and consent form.
Interview procedure
Interviews were conducted face-to-face at women’s homes, at Cardiff University, or by telephone. For pragmatic reasons, babies and young children were present during some interviews with women, but no other adults were present. Before the interviews, women were sent a resource pack, which included various items of stationary, an exemplar blank timeline template, and some examples of the themes that we were interested in covering during the interview. This encouraged participants to reflect on their experiences and to guide the discussion during the interview. The timelines provided a visual tool to enable women to map out their journey towards starting a family, noting key events and their physical and emotional responses to these. Women who had prepared timelines could use these as prompts for topics they wanted to discuss during the interviews. Women had the flexibility to use a timeline template provided by the research team, generate their own, or tell their story in their own way if they preferred.
Interviews with healthcare professionals were guided by an interview schedule (Additional file 2), which focused on the health professional’s role, challenges in providing care for women with ARDs who are starting a family, and how care could be improved. Visual timelines were drafted by the researcher at the end of the interview to map out what health professionals had talked about in terms of how healthcare services were provided along women’s journeys through pre-conception, pregnancy and early parenting, and to identify at which points extra support might be needed. The timelines were sent to the healthcare professional after the interview for participant validation, to ensure that they captured the conversation accurately, and healthcare professionals were encouraged to amend the timelines if needed.
Interviews were carried out by Denitza Williams PhD (DW) and Bethan Pell BSc (BP). DW was a post-doctoral researcher and BP a research assistant at the time the interviews were carried out, and both researchers are female. Both researchers had previous experience of carrying out qualitative interviews, and were provided with additional training and supervision in study specific procedures by the lead author (Rhiannon Phillips, PhD) and qualitative lead for this project (Aimee Grant, PhD). The interviewers had no relationship with the participants prior to the interviews and had no prior knowledge of their goals or characteristics, other than the participant information pack provided as described above. Where participants asked the interviewers about their background during interviews, the interviewers explained that they were researchers that were not from a medical background, nor were they experts in rheumatic diseases, but rather that they were interested in hearing about women’s experiences to inform further research on better meeting their information and support needs. Interviews were audio-recorded and interviewers made field notes as soon as possible after interviews. The interviewers requested a copy of the completed versions of the timelines to provide context during the analysis, although this was voluntary. Transcripts were not returned to participants for comment and participants did not comment on the findings. Health professionals were given an opportunity to review and comment on the timelines produced by the researcher to summarise their discussions. No repeat interviews were carried out.
Qualitative analysis
All interviews were audio-recorded and transcribed verbatim. The data were analysed thematically using a hybrid approach of inductive and deductive analysis, based primarily on social phenomenology [33]. The analysis focused primarily on the data-driven process of understanding how people make sense of and interpret the phenomena of their everyday world [33]. The deductive component of the analysis was far less pronounced in this study, seeking only to identify themes relating to information and healthcare needs of this and similar populations that had been highlighted in previous research [1, 2, 21]. NVivo V10 was used to facilitate analysis. DW carried out the data coding. Our protocol did not include dual coding of the data. Instead we used regular qualitative research team meetings to discuss data production, the development of the coding framework and data analysis, with each member of the qualitative research group (DW, BP, RP, AG) adding their own unique perspective to the analysis through these meetings. This approach has been identified as appropriate in qualitative research [34]. We were guided by the concept of ‘information power’ [35] rather than ‘saturation’; the research team judged the sample to provide a sufficient depth and range of knowledge to meet the study objectives.