Original editorial illustration, not a study figure or study data. Source discussed: Kling et al., 2026.
An app can collect symptoms, organise questions and offer information before a consultation. But does an interactive digital format achieve more than a well-prepared pamphlet? In a Mayo Clinic pilot study, the emmii app did not demonstrate an advantage over an educational pamphlet in the frequency of new estrogen prescription orders. That matters—but it does not establish that the two approaches are equivalent or that apps have no value.
Kling and colleagues published the study on 16 September 2026 in Frontiers in Reproductive Health. Its interpretation requires separating two comparisons, two sets of denominators and several meanings of “starting treatment”. These distinctions make the findings more useful than a simple verdict for or against digital health.
What was compared?
The investigators invited women aged 45–55 who had an upcoming primary-care appointment at Mayo Clinic. Eligibility required at least 5 points on the Menopause Rating Scale, a symptom questionnaire. Menopause stage was not directly assessed. The sample should therefore not be described as a representative group of all postmenopausal women.
Of 14 726 women invited, 653 participated and were randomly allocated: 319 received access to emmii and 334 received a digital educational pamphlet. Both groups received information before their visit. The pamphlet was an active comparator covering symptoms, treatment options and lifestyle strategies. It was not an empty document or absence of support. The app offered additional interactive and personalised features.
This distinction defines the question answered by the trial: whether one way of delivering education outperformed another in the chosen setting. It was not a randomised comparison of education against no education. A neutral result against a meaningful alternative can be valuable, while still leaving room for individual preferences and other outcomes that were not established here.
What counted as treatment initiation?
The outcome was a new estrogen prescription order in the electronic health record within 30 days after the appointment, with no qualifying order during the preceding year. The definition included both systemic and vaginal estrogen. Describing it simply as initiation of systemic menopausal hormone therapy would therefore be inaccurate.
A prescription order is not the same as dispensing a medicine, taking it consistently, experiencing symptom relief or achieving a favourable overall treatment outcome. The study does not settle all those later steps. More prescriptions are not automatically better care, just as fewer prescriptions are not automatically evidence of neglect. The relevant clinical objective is an appropriate, informed decision—not maximising a prescribing percentage.
The randomised result:9.09% versus 9.58%
New orders were recorded for 29 of 319 women allocated to the app, or 9.09%, and 32 of 334 allocated to the pamphlet, or 9.58%. The app-minus-pamphlet difference was −0.49 percentage points. Its 95% confidence interval ranged from −4.95 to 3.97 percentage points; p=0.83.
The defensible conclusion is that the app did not demonstrate superiority for this outcome. It is not that equivalence was established. The confidence interval includes differences in either direction. A non-significant test does not eliminate uncertainty, and two similar percentages do not turn a study into an equivalence trial.
Nor does this result establish what every digital education programme would achieve. The trial evaluated a particular app, a particular pamphlet and a particular appointment pathway. Different content, patterns of use or outcomes would require their own evaluation. The findings should not be extended to a product simply because it also runs on a phone.
Why the comparison with nonparticipants is different
The authors also compared all trial participants with a reference population receiving care during the same period. New orders occurred in 61 of 653 participants, or 9.34%, compared with 623 of 14 032 reference patients, or 4.44%. These figures could easily become a headline saying education doubled treatment initiation. That would overstate the evidence.
The reference population was not a third randomly assigned arm. Participants volunteered and had to meet a symptom threshold. MRS scores were not available for the reference group, and prior hormone-therapy use also differed. The authors explicitly acknowledge that baseline differences and selection could explain the association rather than an effect of education.
Randomisation protects the comparison in which it was actually used. It does not automatically confer causal meaning on every additional comparison appearing in the same paper. A useful reading habit is to ask exactly which groups were formed by random allocation and which were assembled in another way.
Survey findings use another denominator
Both surveys were completed by 383 women: 171 in the app arm and 212 in the pamphlet arm. These are the respondents whose experiences could be analysed, not the denominator for the estrogen-order comparison. Switching those numbers would mix different outcomes and different levels of follow-up.
Many app users reported improved understanding and found the tool useful before their appointment. But the control group was not asked the same additional emmii questions. These responses cannot demonstrate superiority in objectively measured knowledge. The app questionnaire was also longer, and completion differed between groups, raising the possibility of response bias.
Perceived usefulness is still worth examining. It simply needs to be described as what was measured: a user report, rather than an established improvement in health. Ease of use, understanding, prescribing and symptom relief are related questions, not interchangeable outcomes.
What can patients take into an appointment?
The practical message is neither “you must use an app” nor “avoid apps”. An understandable resource can help someone prepare a list of symptoms and questions for a consultation. This is an editorial suggestion for organising the conversation, not an additional intervention proven by this trial.
When choosing a resource, useful questions include who produced the information, which sources it uses, what it does not assess and how it supports contact with a clinician. Data-handling arrangements also deserve attention. This discussion list is not a certification of any product. The study does not justify starting, stopping or adjusting treatment independently.
For healthcare teams, the result encourages a more specific evaluation question than whether a service looks modern. What extra function is being offered, who can use it, and which outcome would justify choosing it over an accessible alternative? Those are questions for further assessment, not benefits demonstrated in the present paper.
Limits and transparency
Follow-up lasted 30 days. Participants came from one care network, had limited racial and ethnic diversity, and were largely well educated and insured. Only a small proportion of those invited enrolled. Generalisation to different services, underserved groups or longer-term treatment outcomes needs additional evidence.
The authors declared no financial support for the work or publication and acknowledged the Mayo Clinic Kern Center. Relevant relationships were disclosed: coauthor Arjan Karens is founder and CEO of BettrHealth, which codeveloped emmii. Other authors disclosed advisory, educational or research relationships with companies. The methods state that statistical analysis was conducted and verified independently of BettrHealth. Reporting these declarations is not an independent audit of interests.
My disclosure: I am a co-creator and co-owner of the iYoni application. This study evaluated emmii, not iYoni. Its findings are not evidence of effectiveness for my application.
The central conclusion is deliberately narrow: this pilot did not show that an interactive app improved the frequency of new estrogen orders compared with a pamphlet. Choosing an educational format does not replace assessing its content, the quality of the consultation or the appropriateness of an individual treatment decision.
Source: Kling JM et al. A mobile app compared with an educational pamphlet on menopause hormone therapy initiation: a randomized pilot study. Frontiers in Reproductive Health. 2026;8:1952673. DOI10.3389/frph.2026.1952673. Educational information, not individual medical advice.