What changes when women’s experiences become part of the baseline?
Why is women’s pain misunderstood? Part of the answer lies in gaps in research and care: pain studies have historically relied heavily on male animals, and evidence shows sex bias in some pain-treatment decisions. Understanding women’s pain requires research that examines relevant differences and clinical records that capture symptoms and their impact. [1] [4]
When I studied economics, we began with a reassuring little phrase: ceteris paribus. All other things being equal.
You isolate a variable. Hold the rest constant. Build your model.
Later, working in analytics in the luxury industry, I learned how much effort goes into that word “equal”. Campaigns change behaviour. Seasonality shifts. Last year’s comparison stops making sense. Before trusting the result, you check the baseline.
Then I spent years living with pain before my endometriosis was diagnosed. And I started asking the same question about medicine.
Who was the baseline built around?
In pain research, the answer has too often been male. That history matters when we try to understand women’s symptoms, decide whose pain gets treated, and build the next generation of medical AI.
What is the male baseline in pain research?
A historical review of animal studies published in Pain found that at least 79% used only male animals. Just 8% used females only, and 4% were explicitly designed to test sex differences. The figures cover a ten-year period reviewed in the mid-2000s; they describe the foundations of the field, rather than its entire present-day output. [1]
I find that unsettling. We were trying to understand a human experience while studying a narrow slice of biology.
Research has moved forward. The US National Institutes of Health now expects sex to be considered in the design, analysis, and reporting of relevant studies. [2] But putting females into a study does not automatically tell us whether a finding applies equally to them. That depends on what researchers measure, analyse, and report.
In analytics, an average can hide the very difference you need to see. The same question belongs in medicine: what disappeared when we combined the data?
What does research show about bias in women’s pain treatment?
A recent study of two large US health-record databases found higher female prevalence in 72.9% of the pain categories assessed in All of Us and 82.6% in Epic Cosmos. These percentages describe how widespread the pattern was across categories, including different locations and types of pain. [3]
Then there is what happens when a patient asks for help.
A 2024 study of emergency-department records found that female patients were less likely to be prescribed pain-relief medication, even after accounting for reported pain scores and other factors. Their pain scores were also less likely to be recorded by nurses. [4]
That is a difficult combination: gaps in the evidence about pain, followed by gaps in how pain is documented and treated.
Biological differences and gender bias are separate questions, and both deserve attention. Studying how bodies differ does not explain away unequal treatment. Taking a patient seriously is part of collecting useful information in the first place.
Why track endometriosis symptoms between appointments?
According to the World Health Organization, endometriosis affects an estimated one in ten women of reproductive age worldwide. Access to diagnosis and treatment is often delayed. [5]
My body does not organise its symptoms around my appointment calendar.
There is the pain I have today. The fatigue that followed it yesterday. The plans I cancelled last week. The pattern I can feel happening again, but struggle to reconstruct when somebody asks me to summarise the last few months.
A pain score can be useful. So can knowing whether I slept, worked, walked, or spent the afternoon lying down. Timing and daily impact give that number context.
When those details remain scattered across memory, messages, and occasional consultations, we risk losing the pattern.
What happens when AI learns from those gaps?
This is where my background in data makes me particularly cautious.
A model learns from what we give it. If a record is missing symptoms, understates their impact, or reflects unequal treatment, those limitations can travel into the model. More sophisticated analysis cannot guarantee that the missing experience becomes visible.
We need to ask practical questions. Whose data are represented? What counts as a successful outcome? Does the tool perform as well for different patient groups? How are those differences checked?
For me, those questions belong at the beginning of product development.
Building evidence that reflects daily life
This is part of why I’m building ENDOless. I want the experiences that disappear between appointments to become easier to record, describe, and study.
Useful progress means:
- Studying relevant differences: including females in research and analysing results by sex where appropriate.
- Following symptoms over time: capturing their timing, recurrence, and effect on everyday life.
- Checking the tools: evaluating clinical and AI performance across the people expected to use them.
- Making research participation trustworthy: using clear consent and appropriate safeguards for health data.
The economic stakes are substantial too. The McKinsey Health Institute estimates that closing the broader women’s health gap could add at least $1 trillion annually to the global economy by 2040. [6] Behind that projection are people who could spend more of their lives well enough to work, participate, and make plans.
I return to the same habit I learned in analytics: before improving a model, examine what it assumes.
When women’s experiences are missing, questioning the baseline is a necessary part of improving the evidence.
Women’s pain deserves research built to see it.
Frequently asked questions
What does “the gender pain gap” actually mean?
It’s the gap in how women’s and men’s pain is understood and treated. Part of it starts in research: whose bodies were studied, and whose experiences were missing? Part of it shows up in care, when pain is recorded or treated differently. Biology and bias are different questions, and we need to take both seriously. [1] [4]
Are women really less likely to receive pain relief?
In some settings, yes. A 2024 study of emergency-department records found that women were less likely to be prescribed pain relief, even after researchers accounted for their reported pain scores and other factors. That doesn’t describe every doctor or every appointment. But it is a pattern worth taking seriously. [4]
How common is endometriosis?
More common than many people realise: the WHO estimates it affects around one in ten women of reproductive age worldwide. Yet getting a diagnosis and access to treatment can still take time. Being common hasn’t made it easy to get answers. [5]
Learn more about ENDOless and the story behind the project.
Sources
- Studying sex and gender differences in pain and analgesia: a consensus report — historical animal-research figures.
- NIH: Sex as a Biological Variable — research policy.
- Sex differences in pain in two large and diverse US databases — All of Us and Epic Cosmos study.
- Sex bias in pain management decisions — PNAS, 2024.
- WHO: Endometriosis — prevalence and access to care.
- Closing the women’s health gap — McKinsey Health Institute, 2024.
