Long term impact of the WHI studies on information-seeking and decision-making in menopause symptoms management: a longitudinal analysis of questions to a medicines call centre

Background While women are taking a greater role in decisions about menopause symptom management, the legacy of the Women’s Health Initiative (WHI) studies persist. Despite hormone therapy (HT) being effective in reducing all-cause mortality, many women seeking relief of menopausal symptoms exaggerate HT harms and overstate the perceived benefits or ignore the risks of alternative therapies. We aimed to explore the longitudinal impact of the widely-publicised WHI 2002 study on women’s information-seeking and describe determinants of decision-making about managing menopausal symptoms. Methods In a longitudinal analysis of both quantitative and qualitative data, we explored consumer questions about menopause-related medicines received by two Australian medicines call centres (1996–2010) before, during, and after WHI 2002. We analysed calls by age and gender of caller and patient, their relationship, postcode, enquiry type, and motivation to help-seek. We compared calls regarding HT and herbal medicines (HM) with the rest of calls, and thematically analysed question narratives across the three time-periods. Results There were 1,829 menopause-related calls received of over this time-period, with a call surge, primarily from women in their mid-fifties, in the two months after the WHI 2002 publication. Two in three calls were motivated by negative media reports as women sought support for decision-making, primarily reassurance to cease HT. While HT safety concerns persisted for eight years post-publication, the nature of information-seeking changed over time. Callers subsequently sought reassurance to use menopause treatments together with their other medicines; and pursued HT substitutes, including HM, in response to HT product discontinuation. Conclusions Women sought information or reassurance to support a decision, based on dynamic changes in internal (symptom or risk intolerance, attitude towards menopause and treatment preferences) and external factors (perceived source trust and changes in treatment availability). In assessing HT benefit versus risk, women tend to overestimate risk with HT safety concerns persisting over time. Decision-making in managing menopause symptoms is complex and dynamic. Reassurance to reach or justify decisions from a perceived trusted source can support informed decision-making. Supplementary Information The online version contains supplementary material available at 10.1186/s12905-021-01478-z.

Background Menopause is a natural and unique experience for every woman [1]; some easily pass through, while others have a difficult transition [1]. Vasomotor symptoms (VMS) are most common during the transition (prevalence 30-70%) [1] and are associated with depression [2], lower healthrelated quality of life [3,4], socio-economic factors, and increased use of healthcare services [4].
Many were uncertain about how best to manage symptoms after the publication of the Women's Health Initiative (WHI) studies in 2002 and 2004 [5,6]. These studies tested the heart-protective effect of hormone therapy (HT) in women who were older (50-79 years) than those in earlier studies (45-64 years) [7,8]. WHI 2002 was prematurely terminated after HT use by women with an intact uterus was unexpectedly associated with an increased risk of breast cancer, with no apparent beneficial cardiovascular effects [5]. Widespread negative media coverage led to a substantial and rapid decrease in HT use worldwide [9]. Many women ceased HT abruptly, with some recommencing HT due to intolerable symptoms [10]. A second WHI study of estrogen therapy (ET) in women with a hysterectomy was also prematurely terminated due to an increased risk of stroke with no increase in breast cancer risk [6] yet these conflicting findings were largely ignored by the media. Subsequent reanalysis found that both studies were confounded by age [11] with a 'window of opportunity' for cardiovascular health benefits if HT was initiated before 60 years or within ten years of menopause [11]. The 2016 Revised Global Consensus Statement on Menopausal Hormone Therapy [12] reaffirmed that HT remains the most effective treatment for vasomotor symptoms and significantly lowers the risk of osteoporosis-related fractures in postmenopausal women. HT is effective in vulvovaginal atrophy and may improve sexual function and other related symptoms such as joint and muscle pains, mood changes and sleep disturbances. However, prescriptions for HT have continued to tumble. Unfortunately, the WHI 2002 findings in older women with an intact uterus were generalised to all forms of HT, including estrogen alone in younger hysterectomised women. Prescriptions for HT declined steeply, with estimates that within 18 months of WHI 2002, half of US women using systemic HT stopped treatment. [13] Sarrel and colleagues [14] used WHI 2004 data [15] indicating a higher rate of mortality among hysterectomised women aged 50 to 59 years assigned to placebo than estrogen over a 10-year follow-up, to determine how this rate of excess mortality translated into an aggregate toll of premature death through estrogen avoidance at the population level (2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011). Their analysis concerningly suggested that between 18,601 and 91,610 excess deaths occurred among hysterectomised women aged 50-59 years following the publication of the original WHI.
Women are taking a greater role in decisions about managing symptoms [21]. The consumer's role seems to dominate the clinician's when deciding on preferred treatment [21] but professional guidance is still needed and often sought when evidence of efficacy, safety, and multiple treatment options add to confusion [19,20]. However, the legacy of the WHI studies persists. Despite hormone therapy being effective in reducing all-cause mortality [22], many women seeking relief of menopausal symptoms exaggerate HT harms, overstate the perceived benefits of alternative therapies, or ignore their risks [23][24][25].
Women's decision-making processes in managing menopause have been modelled in normal [19] and surgical menopause [20]. Existing models suggest decision-making is driven by both internal (e.g. individual characteristics, values, attitudes, beliefs and preferences) and external factors (e.g. healthcare provider context, facts and information) [19,20] but these models have limitations. Evidence was collated mainly from cross-sectional studies, with many conducted pre-WHI 2002 [19,20]. There are comparatively few recent studies focusing on women in the active process of decision-making [24,26].
Women seek information to manage their health from a range of sources including social networks, social media, and the internet [9,27]. Some women may be uncomfortable using the internet, preferring immediate feedback or an opportunity to ask questions and discuss potentially opposing ideas [28,29]. Where consultation time is limited, others might be reluctant to be perceived is complex and dynamic. Reassurance to reach or justify decisions from a perceived trusted source can support informed decision-making.
Keywords: Menopause, Estrogen replacement therapy, Information seeking behaviour, Decision making, Drug information services as ignorant or question their clinician's judgement [30]. A consumer medicines call centre (MCC) operated by pharmacists provides an alternative, anonymous source for obtaining reliable and evidence-based medicines information [31].
We sought to understand the information-seeking and decision-making behaviour of women accessing a MCC with menopause symptom management concerns, as evidence changed over time. We aimed to: (1) longitudinally explore the impact of WHI 2002 on women's information-seeking behaviours (2) compare information-seeking for HT and menopause-related HM, and (3) identify factors affecting decision-making for menopause symptoms.

Data sources
We longitudinally analysed routinely-collected health service data from two Australian MCCs operated by clinical pharmacists from Mater Health, We quantitatively and qualitatively analysed these data. The time ranges were identical for Time 1 and Time 3; but we used a deliberately narrower 'window' (10-26 July 2002, Time 2a) for the qualitative analysis ( Fig. 1) to capture questions and concerns in the period immediately following WHI 2002 and the media response. During these 13 service days, QMH pharmacists kept detailed information on callers' menopause-related questions and concerns-some with overwhelming anxiety-and any actions prior to service contact.
We conducted and reported this research in accordance with the REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) guideline [32]. We had ethical approval from the Mater Health Services Human Research Ethics Committee (HREC/13/MHS/80).

Data collection
Call data were recorded manually on a scannable collection form and transformed into an electronic dataset. Most variables were consistently recorded over time, including caller and patient demographic characteristics, question enquiry type, call motivation, medicines involved, the question, and answer. Question narratives were only electronically recorded in selected periods due to limited funding but we hand-searched forms for additional narratives across the study period.

Data classification
Medicines of interest were mapped to the Anatomical Therapeutic Chemical code: [33] G03CA (estrogens), GO3F (estrogen plus progestogen therapy, EPT), G03CX01 (tibolone), and G03BA03 (testosterone). We filtered data from female consumers by age to include calls from women who might experience premature menopause: ≥ 35 years for women calling for estrogen therapy and ≥ 48 years for EPT; calls were considered'rest of calls' (ROC) if patient age was under the cut-off. We filtered the testosterone calls by patient gender ('female' or 'blank') to enhance call retrieval and we excluded non-menopause data. We examined calls for five HM: black cohosh, wild yam, soy-based products, chaste tree, and red clover.

Variables of interest
We extracted data on variables: gender and age of the caller and patient, caller-patient relationship (self, partner, parent, child, other family, friend and client), enquiry type, call motivation, and postcode. Pharmacists classified the enquiry type and call motivation into predetermined categories [31]. We collapsed the 21 enquiry types into seven groups: (1) side effects, (2) risk/benefit, (3) pragmatics of use (e.g. administration, dose, withdrawal and efficacy, and stability/storage/ disposal), (4) interactions, (5) mechanism/profile (e.g. indication, identification, formulation), (6) treatment/ prophylaxis, and (7) logistics and miscellaneous (e.g. availability, comparison, cost, generic, subsidised medicines). The 17 call motivations were collapsed into six groups: (1) inadequate information, (2) second opinion, (3) worrying symptoms, (4) conflicting information, (5) the media and (6) other. We mapped caller postcode to the Accessibility/Remoteness Index of Australia (ARIA), which measures the remoteness of areas from service centers [34,35]. We compared proportions of calls from each ARIA with population data to generate relative call frequencies expressed as a ratio [34,35].

Quantitative analysis
We compared menopause-related calls and ROC over time and by medicines class (HT and HM). Some data (QMH) were only from Queensland so we assessed whether call characteristics differed between Queensland and other states in the Australian (ML) dataset and between Queensland callers of QMH and ML, respectively. We used a one-way analysis of variance (ANOVA) and Pearson's chi-square test to test for any differences. We conducted post-hoc analyses, including Tukey's honesty significant test and standardised residuals method [36], to further explore any differences among categories. We applied the Bonferroni correction to a p-value of 0.05 to avoid false-positive results [36]. We used Microsoft Excel 365 (Microsoft, Redmond, WA, USA) and SPSS (version 25.0, IBM Corp, Armonk, NY, USA).

Qualitative narratives
The purpose of the qualitative data analysis was to provide further insight into the results of our quantitative analyses. All question narratives were collected and analysed thematically according to the results identified in our quantitative analyses. We used a system of coding, comparison, and clustering with a basic coding frame developed from our quantitative results [37]. To enhance the rigour of this analysis two coders separately coded each question narrative. Any discrepancies or revisions to coding were resolved through consensus.

Impact of the WHI 2002 publication on call characteristics
Most menopause-related calls were made by women in their mid-50s (average 56.5-59.3 years) from areas classified as either highly accessible or accessible, and seeking help for themselves (Table 1). Call characteristics were compared among: (1) menopause-related and ROC over time; (2) Queensland and other states in the Australian dataset; and (3) Queensland callers of QMH and ML (Additional file 1). The differences in key call characteristics between menopause and ROC were consistent and stable over time. There were no differences in menopause-related calls for caller gender, relationship between caller and patient, nor ARIA when we compared calls originating from Queensland vs other states/territories; or over time (Additional file 1).
Menopause-related calls surged in the two months after WHI 2002 (July 2002, Fig. 1); two in three calls (66.7%) were motivated by media reports. Prior to WHI 2002, women were motivated to ring because of worrying symptoms and inadequate information ( Table 1). The media effect was relatively short-lived, with only Table 1 Comparison of characteristics of menopause-related calls across three time periods

Menopause narratives
We identified four primary themes (safety concerns, pragmatics of use, therapy options, and regimen comparisons) and 17 sub-themes from 749 menopause-related calls. There were 846 total questions (some narratives had more than one question): 229 from before WHI ( Sometimes the cancer fear was transferred to other symptoms or concerns. Following his wife's death from heart failure, a 60-year-old male was concerned that HT might have been the underlying cause (Table 3). Breast cancer was mentioned at other time periods; it was not a fear of the disease per se but to decide whether they could (or should) use HT or an alternative strategy.
Women asking for help to balance the risks and benefits to either use or withdraw HT, based on their individual circumstances, was a particular feature immediately after WHI (Time 2a); while before WHI and later, there was a focus on clarifying the cause of a particular symptom in relation to HT. Before and later after WHI women more commonly sought reassurance to use HT with other medicines or in the presence of other diseases.
Pragmatics of use ranked highly in Times 1 and 3, where changes in medicines access, particularly in later years, prompted a need for decision-making. Similarly, calls about therapy options and regimen comparisons at Times 1 and 3 were not evident at Time 2a. Over time, the nature of HT safety questions evolved to include whether vaginal HT produced similar effects to systemic HT (oral or transdermal) and calls asking for alternative forms of treatment. For non-hormone therapy, women asked whether HM, antidepressants, or bisphosphonates were effective in managing menopause. Women older than 60 years sought reassurance to use HT across the period of our study (1996-2010) despite guidelines after 2002 deeming HT unsuitable for this cohort [38] (Table 3).
Our data demonstrated the complex and dynamic interplay of internal and external factors that influence women who seek information for reassurance, to reach or justify a decision about managing menopause symptoms (Fig. 2). Information-seeking occurred in response to heightened uncertainty: worrying symptoms, therapy dissatisfaction, inadequate or conflicting information, perceived risk, or change to treatment availability. Any ambiguity stimulated decision reassessment. The more intense the uncertainty (e.g. negative media about HT), the more women relied on emotions to emphasise risk over benefit. Selected vignettes highlight the iterative nature of this decision-making (Additional file 2).

Discussion
WHI 2002 [5] was the 'top medical story of 2002' , creating the 'HT scare' [9]. Negative media was the main driver of information-seeking for decision-making; calls increased 400% in the first two weeks and remained high for the subsequent two months. Maladaptive risk perception is a cornerstone of cognitive models of decision making under risk and uncertainty, where worried individuals generally overestimate negative outcomes [39], and emotional reactions to perceived risk often diverge from cognitive assessments of those risks, driving behaviours such as help-seeking [40]. While the effect on helpseeking was relatively short-lived, consumer concerns about safety persisted for years after WHI 2002. Despite subsequent media reports of HT benefits [9] in specified cohorts [11], there were few calls prompted by media e Relative call frequency was calculated by dividing the proportion of calls by proportion of population in the specific ARIA index     Table 3 Selected caller quotes for each of the themes and sub-themes identified across the three specified time periods Risk concern -breast cancer  Safety concern -seeking to change to an alternative product   The subsequent HT-favourable findings of cardiovascular benefit in selected cohorts [11] did not elicit the same level of public attention as WHI 2002. Women's perception may have been influenced by other information sources (e.g. peers and clinicians) [9] or may reflect persistent concern about breast cancer [11]. Our findings support the notion that the fear of cancer overshadowed any beneficial effect of HT on cardiovascular health [42].
There were proportionally fewer menopause-related calls over time. While this might simply reflect lower HT use due to progressive product withdrawals after 2002, few studies assessing the relationship between product withdrawal and use could be retrieved [42][43][44]. In addition, distrust in medical science, given the 'pendulum swings' in HT evidence [21], potentially left women feeling they had no alternative but to bear with their symptoms without pharmacological intervention [27].
We found that women considering menopause therapy sought reassurance about their decision-making before and after WHI 2002, but their priorities differed. Afterwards, women were more likely to seek confirmation that HT or HM were safe to use with their other medicines, suggesting increased interest in HM as a substitute for HT [24,27]. The potential for interactions with over-thecounter medicines such as HM is a concern for consumers who self-medicate without consulting their clinician [45]. Women with severe menopausal symptoms might also experience other symptoms [2], or use more healthcare services [4], which may mean use of medicines other than HT. There may have also been a change in the treatment objective. Before WHI 2002 some clinicians prescribed HT for women who were symptom free; while afterwards HT use was largely limited to patient requests for symptom relief [21].
We found that changes in treatment access or availability influenced information-seeking behaviour and prompted women to seek substitute treatment. After WHI 2002, women requested information on substitute treatments when HT products were progressively removed from government subsidy and subsequently the Australian market. Calls about HM primarily originated from areas with good access to service centres where a pharmacy, health store, or naturopath are more likely to be located. Our finding is consistent with a survey of HM use by 2,020 Australian middle-aged women which found a higher prevalence of HM use for menopausal symptoms among women residing in metropolitan areas [46] perhaps reflecting increased availability and income.
To our knowledge, no other study has examined women's information-seeking and decision-making about managing menopause before and for more than seven years after WHI 2002. These data reflect the real-world issues faced by women and the significant impact of negative media reporting. The study's longitudinal nature also allowed us to examine any persisting impact of WHI 2002.
We acknowledge some study limitations. The data were collected as part of routine MCC activity so there  [19,20] are in bold dashed-line and dashed-line boxes was no opportunity to prospectively collect additional data of interest. The MCCs were in Australia, so our results might not reflect the concerns of women in other places with different cultures and ethnicities. Indications for medicines were not captured, so we were unable to explore concerns about the use of non-hormone therapy for menopause [17]. The two data sources might not be fully comparable given the differences in coverage and time frames. Our study period ended in 2010, prior to the Global Consensus Statement on Menopausal Hormone Therapy in 2013 [47]-the official reassurance made by various medical societies about the safe use of HT in specific cohorts of women. We need to know more about women's information needs for all available treatments, especially after 2013.
Building on previous models of decision-making in menopause [19,20], we identified that women information-seek for reassurance from a trusted source, to assist or justify a decision [48]. This is influenced by heightened intolerance to risk, associated with a change in competing internal factors (woman's specific symptom concerns and tolerance to risk), and external factors (the impact of negative or conflicting information, source trust and product availability (Fig. 2). Decision-making is an iterative process. Accessing HT or other therapy involves a discourse between a woman and her clinician but we observed that women sought information or reassurance to support a new decision based on dynamic changes in internal (e.g. experiencing a side effect or new symptoms) and external factors (including conflicting medical advice) about how she will ultimately manage her symptoms. Negative or conflicting source information, worrying symptoms, perceived treatment risk, and decreased treatment availability contribute to perceived risk predominance. If other factors emerge that change the dynamics of the decision-making process, the iterative cycle will recommence (Fig. 2).

Conclusions
Negative WHI 2002 media reports escalated information-seeking for decision-making about managing menopausal symptoms. While safety concerns and overestimation of risk persisted, the factors influencing the nature and intensity of information-seeking changed over time and over the woman's menopause journey, requiring reconsideration of previous decisions. Reassurance to reach or justify decisions from a perceived trusted source can support informed and shared decision-making.