If the heat wakes you at night and you've been told it's just the change: your health happens between visits, it belongs to you, and the coverage you already have can pay for care built that way.
By Gale Staff
Scroll
It is 3 a.m. and Dana is awake again. The heat starts under her breastbone and climbs into her face, and by the time she has kicked off the duvet her T-shirt is stuck to her back and the pillow is damp. She turns it over to the cool side. She lies there and runs through tomorrow's meetings. It is the fourth night this week.
She is 51. On a bad day she has eight hot flashes, and one of them always arrives in the ten o'clock meeting, where she can feel her neck go red in front of people who report to her. She raised it at her physical in the spring. Her doctor said it was the change, that it would pass, and asked whether she had booked her mammogram.
It will pass, eventually. For some women that takes a decade. Meanwhile Dana sleeps five hours a night, keeps a fan on her desk in October, and has started to suspect she is simply worse at her life than she used to be.
She isn't. What follows is Dana's next year, and the research behind every turn in it. It rests on three things. Her health happens between the visits, in the nights no doctor sees. It is hers: her record, her numbers, what she shares and when, and every decision. And the insurance, HSA or savings she already has can pay for care that works this way, because the visits are ordinary visits.
That is what we are building at Gale, for women from 40 to 70. Menopause is where we start.
Health happens between the visits
I
The year
One year, drawn an hour at a time.
That physical, the one where she was told it was the change, was most of the medical care Dana got all year. It is most of what anyone gets. Everything that happens to her at 3 a.m. happens in the other hours.
8,760 hours8,760
One year, one square per hour.
About one hour~1 hour
The ringed square is a year's exam time with an office-based doctor for the average American: about three visits 1 of about 18 minutes each 2.
Measured another way, a visit lasts 13 minutes 3 or 23 4. Either way, about one square.
About 21 hours~21 hours
The faint squares are the hours inside the system that can be counted: the drive, the waiting room, the emergency department, a hospital bed 567.
About two hours per office visit, counting travel and waiting 5.
With three or four chronic conditions4 · 38
Someone managing three or four long-term conditions has about 12 outpatient visits a year 8: about 4 hours of exam time, about 38 inside the system 258.
720 hours720
The shaded month is the 30 days after leaving hospital. About one hospital stay in six for people on traditional Medicare ends with a return within it 9.
Trials of seeing a doctor early after discharge found no clear effect on readmission 9.
The rest of the year8,739
The hours left in the year: about 8,739 1567. This is where the night sweats happen, and the missed meetings, and the decision to give up on a treatment.
No doctor can be in those hours, and nobody should want one there. What those hours need is the right clinician at the few moments that matter, and a way to know when those moments have come. That is the first part of this story.
II
It isn't you
How many women in a hundred take hormone therapy, then and now.
Dana's mother kept a pill case by the kitchen sink and never once mentioned a hot flash. Dana assumed her mother had been lucky. More likely she had been treated.
The treatment that works best for hot flashes went from ordinary to rare in one generation. Women didn't stop needing it.
199926.9 of 100
One hundred American women past menopause, one square each. In 1999, nearly 27 of them took hormone therapy 10.
A national health survey of 13,048 women across the years shown 10.
This counts every woman past menopause, with symptoms or without.
Black women, 20200.5
Among non-Hispanic Black women past menopause in 2020: half of one square 10.
Side by side26.9 · 4.7 · 0.5
One in four, then one in twenty, then one in two hundred 10.
The smallest grid hides the longest wait. In one large study that followed women through menopause, African American women had frequent hot flashes for a median of 10.1 years, against 7.4 overall 11.
III
Nobody asked
Four ways to measure how often a woman with symptoms goes untreated.
Dana's doctor was not careless. He was ordinary. In the surveys below, most women with frequent hot flashes had never been treated, very few doctors in training felt ready to treat them, and the women who have the least often got the least.
Never treated81%
In one survey of working women aged 40 to 65 with frequent hot flashes, 81% had never been treated 1213.
An online panel of 1,011 women, paid for by a drug company 12.
Not on the problem list23% · 6%
In one primary-care system, among 229 women with moderate or worse hot flashes, 23% had them on their problem list, and about 6% were on hormone therapy 14.
In a survey of young doctors in family medicine, internal medicine and gynecology, 6.8% felt ready to manage menopause, and 34% would not offer hormone therapy to a woman it suited 1516.
Among women 45 to 64, 10.7% of those on Medicaid had ever used hormone therapy, against 20.7% of women with private insurance 17.
Once insurance was accounted for, the gap between Black and white women was no longer significant 17.
A friend gave Dana a name: Julie, a nurse practitioner who treats menopause and nothing else. The price was on the booking page, $250, before Dana typed her card number.
It was a video visit, forty minutes, and Dana took it in her parked car. Julie asked how many hot flashes a day, how many nights, whether her periods had stopped, whether anyone in her family had had breast cancer or a blood clot. She had Dana read out her blood pressure from the cuff she had bought at the pharmacy that morning. Then she said the sentence nobody had said: this is treatable, and here are your choices.
That visit is the biggest thing in this story. For most women, being treated at all, by someone who knows menopause, is what changes the year. Everything else we build is how that care keeps working between visits.
I kept waiting for someone to tell me it was a real thing.
Dana, after her first visit with Julie
IV
Starting is not staying
Of every hundred people who start a treatment, how many stop.
They chose a low-dose estrogen patch, with progesterone to protect her uterus. Before they hung up, Julie wrote three sentences into Dana's plan. If your hot flashes aren't down by half by week six, tell me. If they come back after we taper, tell me. Any bleeding, call the office that day.
The patch went on her hip on a Sunday night. It itched at the edges for a week. By day ten the nights were a little shorter and the days were not, and Dana began to wonder, the way most women do, whether this was going to work at all. Starting is the easy part.
Hormone therapy75–88 of 100
Of 100 women who start their first hormone therapy for menopause, at least 75 have stopped within six months, and in some countries as many as 88 18.
Health records of more than 1.5 million women in the US, UK and Germany 18.
An antidepressant65–85
Prescribed for hot flashes instead: 65 to 85 of 100 stop within six months 18.
A GLP-164.8
Taken for weight, by adults without diabetes: about 65 of 100 stop within a year 19.
125,474 adults, most of them women, average age 54 19.
All three75 · 65 · 64.8
For all three, most people who start stop.
Some stopping is relief, and some is a clinician's call. The records can't show who stopped still suffering.
V
Why they stop
The reasons recorded when people stopped.
At week six Dana was having seven hot flashes a day, down from eight. That is the week a lot of women decide it isn't working and quietly peel off the patch. Nobody finds out. The next time a doctor asks, the answer is "I tried hormones once."
When researchers read why people stop, some of the reasons are ones a visit can address. One isn't.
Money47.6%
In one health system, among people who stopped a GLP-1 for weight, cost or insurance was the reason for nearly half 20.
Side effects, shortages14.6% · 11.8%
Side effects for about one in seven, and a drug shortage at the pharmacy for about one in eight 20.
Fear is a conversation. A side effect is often a dose change. Cost is different: a visit can't lower a pharmacy price. In 2021, 6% of new prescriptions were still waiting at the pharmacy two weeks later when the patient owed nothing, and 61% when she owed more than $250 22. We can show you what a visit costs before you book it. The drug's price is its own problem, and we don't pretend to solve it.
Dana didn't have to decide alone at week six. She had logged one number every evening, how many hot flashes bothered her that day, and Julie had asked to hear if that number wasn't down by half. It wasn't. The review was already on the calendar. Julie raised the dose. By week nine Dana was at three a day and sleeping through until the alarm.
VI
Who answers
What happened when medicine tried to watch people between visits.
Medicine has spent twenty years trying to reach into the other hours: weekly symptom forms, home scales, implanted sensors, follow-up texts. The results are in this chart, and they are the reason Gale's app is built around one person answering rather than around the watching itself.
How to read it0.83 → 0.99
Each row is a study result. Left of the line means fewer bad outcomes for the people watched; a bar that crosses the line means the study couldn't tell. In cancer, weekly symptom reports looked like they helped people live longer, then didn't when the trial was repeated in 52 practices 2324.
The repeat trial did find fewer emergency visits 25. A French trial was stopped early 26; a rollout across six health systems didn't cut emergency visits 27.
Heart failure0.70 → 0.88
Home monitoring with coaching calls didn't cut readmissions 28. A round-the-clock doctor-led center helped recently hospitalized patients 29 but not stable ones 30. An implanted sensor cut hospital stays 31, then missed in a broader group 32.
Watching the heartbeat0.80
A recorder implanted in older adults tripled diagnoses of an irregular heartbeat. Strokes did not fall significantly 33.
After a crisis0.56 · 0.57 · 0.52
A safety plan plus follow-up calls after an emergency visit went with about half the suicidal behavior 34; pooled studies of safety plans found much the same 35; caring texts cut attempts 36.
The first was a comparison, not a trial, and the texts missed their main goals 3436.
Someone who can act13 of 17
Now marked: 13 of the 17 results, from 11 of the 15 studies, where a named clinician or nurse got the reports and could change the treatment. Most of the studies that helped are marked. So are most that didn't 24303233.
In one lung-cancer trial, alerts sent to the patient alone did as well as alerts sent to the clinician 37.
The plainest lesson came from depression care. When symptom scores went to a clinician who used them to adjust the dose, people got well in a median of 4 weeks instead of 8, though by six months the others had caught up 38. When the scores were simply handed back to the clinician, with no set rule for acting on them, symptoms improved no more than without them 39. Watching is not care. It helps when the person reading it can change your treatment, and has promised to.
That is why Julie writes her sentences before anything is watched. The app looks for what she asked to hear and nothing else, and when it finds it, the answer is a visit with her.
VII
Why it won't beep at you all day
A watch notices something. What happens next.
In March Julie began to taper the patch, one step down every few weeks, because the goal was always the lowest dose that kept Dana sleeping. Three weeks after the last step, the heat came back. First in the school pickup line, then twice in one meeting, then at 3 a.m., the old damp pillow.
Dana's phone could have been nagging her for months. Every alert has a cost, for the person who gets it and for whoever has to answer it. One health system counted what followed a smartwatch's pulse alerts.
264 patients158 of 264
One square per patient seen after a smartwatch warned of an abnormal pulse. The shaded squares are the 158 who had a heart ultrasound 40.
3030
The lit squares are the 30 who came away with a diagnosis a doctor could act on 40.
Alerts come with worry, too. In one study, one in five wearable users with an irregular heartbeat felt anxious and always called their doctor after an alert 41.
So Dana's app stays quiet. It checks only the sentences Julie wrote. It never tells Dana a number is high or abnormal; it shows her usual range beside this week's. It sends at most two cards in 30 days, and after a Not now it waits 30 days before asking again 42.
That week in spring it sent one. It said Julie had asked to hear if the hot flashes came back after the taper, showed Dana her own log, one to three a day in the month before, six a day this week, and offered Thursday at 4:30. She tapped Book in the car at pickup.
On Thursday Julie didn't ask her to explain. The numbers were already on the screen. They went back up one step and agreed to taper more slowly in the autumn. By May, Dana had taken the fan off her desk.
Your health is yours
VIII
Where your record lives
Seven attempts to move the medical record toward the patient.
That spring visit was easy partly because Julie already had the whole picture. Getting it had not been easy. Before Dana's second visit, Julie asked for her last blood work and her mammogram report. Dana found the blood work in one portal, the mammogram in another, and the password for a third in an email from 2019. Her gynecologist, her old primary-care practice and the imaging center each held a piece of her, and none of them held the part that mattered most to Julie: what her nights had been like.
Companies have tried for twenty years to hand people their records.
Four that closed
Google Health opened in 2008 and closed in 2012. Microsoft HealthVault closed in 2019 43. In England's HealthSpace, 0.13% of the people invited opened an advanced account 44. An employer-backed record, Dossia, is the fourth bar.
Dossia's dates come from sources we could not confirm.
Three still running
Apple Health Records grew from 12 hospitals and clinics in 2018 to more than 800 by 2022, by Apple's count 45. Since April 2021, federal rules have barred providers, with narrow exceptions, from blocking your access to your own record 46, and a national network now passes records between institutions 47.
At one hospital system that launched Apple's feature, 0.27% of portal users connected it in ten months 48.
Opened at least once25% → 65%
The share of American adults who opened an online record or portal at least once in the past year: 25% in 2014, 65% in 2024 49.
One hundred American adults in 2024. Most have more than one portal 51. Seven use an app to bring them together 52.
Holding your records in one place has not, on its own, been shown to make people healthier: trials of records that patients carried found no effect 53, and an app that combined them didn't make medication lists more accurate 54. That isn't why Gale gives you yours. The app keeps what you log about yourself, your sleep, food, activity, medication, cycle and mood, and your private conversations with it, because they are about you and nobody else should decide who sees them. Nothing in it leaves your phone unless you send it.
IX
Reading your own notes
What patients find when they read what their clinicians wrote.
After the second visit Dana read Julie's note on her phone, at the kitchen counter, still in her coat. It said "8 hot flashes a week." She had said a day. She sent one line back, and the next morning the note was fixed.
She is far from alone in finding something.
22,889 readers22,889
Each square is one patient who read at least one of their own notes and answered questions about mistakes, in a survey at three American health systems 55.
4,8304,830
The lit squares are the readers who found something they thought was a mistake 55.
These are mistakes as patients saw them; nobody checked them against the chart 55.
Finding a mistake is not the same as getting it fixed. In one Swedish study, about four in ten portal users who found an error did nothing about it, and at one American health system about 0.2% of patients who asked for their chart ever filed a formal correction 56. Dana's was fixed overnight because she had someone to send it to, and that someone was the clinician who wrote it.
The note said I was not BRCA1 when I am BRCA1. My provider, who was surprised, immediately corrected it.
a patient, in a survey of note readers (JAMA Network Open, 2020)
X
What a request does
The same symptoms, and three different things to ask for.
That autumn an ad came on while Dana was folding laundry: a new pill for hot flashes, no hormones, a woman laughing on a beach. She wondered if she should be asking for it.
The easiest feature for us to build would be the one we refuse to build: an app that reads your numbers and tells you which drug to ask for. Trained actors once went to doctors with the same story, stress after a hard life event, the kind of trouble where an antidepressant usually isn't the answer. The only thing that changed was what they asked for.
Asked for nothing10
One square per visit. With no request, 10 in 100 visits ended with an antidepressant 57.
A general request for something to help: 39 in 100 57.
Asked for a brand55
A request for a brand seen in an ad: 55 in 100 57.
Together10 · 39 · 55
Same symptoms. Asking for a brand more than quintupled the prescriptions 57.
Requests cut both ways: in the same trial they also helped actors with major depression get the care they needed 57.
What you ask for shapes what you get. So the app never names a drug you aren't on and never mentions a dose. It tells you what Julie asked to hear, shows you your own numbers, and leaves the conversation to the two of you. Dana mentioned the ad at her next visit. Julie told her who the pill suits, and why Dana was not that woman yet.
The coverage you already have pays for it
XI
The appeals nobody files
What happens to a denial after it is issued.
When Julie moved her back up a step in the spring, the new patch strength needed her insurer's approval, and the insurer said no. Dana found out at the pharmacy counter, with a line behind her. Julie's office sent a one-page letter the next morning. Nine days later the answer was yes.
Most denials never get that letter.
A thousand denials1,000
Each square stands for about 4,100 of the 4.1 million times Medicare Advantage insurers said no to a request for approval in 2024 58.
The lit squares are the denials somebody appealed, usually the doctor's office: 11.5% 5859.
93 overturned93
Reversed in full or in part: four appeals in five 58.
The appeals people chose to file may not look like the ones they didn't 60.
One week in 20199 of 30
Government auditors pulled a week of denials from the 15 largest plans. Doctors reviewing them found 30 that met Medicare's own coverage rules 61, an estimated 13% of all such denials 62. By the time the auditors asked, 9 of the 30 had been reversed 61.
All 7 of the 30 that were appealed were reversed 62.
0.31%0.31%
Insurers on HealthCare.gov denied 85 million in-network claims in 2024. Patients appealed 0.31% of them 59.
A quarter of the denials were for paperwork reasons 59.
In one national survey, 85% of people with a denied claim never filed a formal appeal 63. An appeal is paperwork, and paperwork is what a woman who hasn't slept in a year has least time for. It belongs with the clinician who wrote the prescription, which is where Julie kept it.
XII
What you already have
Coverage people hold, and how little of it they know about or use.
Dana had lost money in her flexible spending account two Decembers running, because she never had a medical bill worth spending it on. That year she paid for every one of Julie's visits from her health savings account, and noticed for the first time how much of her own coverage she had never read.
Knowing what you have8% · 14%
8% of adults 35 and over were up to date on every high-priority preventive service recommended for them 64. 14% of adults with insurance through work answered basic questions about four insurance terms correctly 65.
Choosing55% · 54%
At one large employer, most workers with single coverage picked a plan that cost them more no matter what happened that year 66. Of Medicare Advantage members, 54% said they had dental coverage, though more than 95% did 67.
The average worker in such a plan could have saved about $372 66.
Money left behind57% · ~50%
In 2018 to 2020, about 57% of people who qualified for help paying Medicare premiums were getting it 68. About half of flexible-spending-account holders forfeited money in 2023, $436 on average 69.
Free, and unused23% · 0.06%
23% of people on traditional Medicare used their free yearly wellness visit 70. Of about 16 million people eligible for Medicare's diabetes prevention program, about 0.06% joined 71.
A benefit you know about is one you can use. A visit with Julie is an ordinary visit: insurance can pay for it, an HSA or FSA can, and at 65 Medicare can. We show you the price before you book so that the money you already have can go where you need it.
XIII
What it costs
Monthly prices, from what women say relief is worth to what the market charges.
Dana's year came to a $250 first visit, two $150 follow-ups, and a $75 visit in spring, the one her phone offered her. Her insurance could have paid for all of it; she chose her HSA.
Here is that next to what women say relief is worth, and what the market charges them.
What relief is worth$35–46 · $83
Women with hot flashes said they would pay $35 to $46 more a month for much better sleep or far fewer hot flashes 72. One large online clinic takes in about $83 a month per subscriber, across everything it sells 73.
The first figure comes from a drug-company survey and prices a treatment, not a service 72.
What the market charges$74–149 · $149+
Two big weight-loss programs charge $74 to $149 a month in cash, drug not included. The brand drug at its maker's cash price starts at $149 74.
A ceiling$150
$150 a month is the most a primary-care membership can charge one person in 2026 and still leave her eligible for a health savings account 75.
Gynecologists don't count as primary care under that rule 75.
We sell no membership. You pay for visits, the ordinary way, and the app comes with them. Dana's whole year, four visits, cost her about what some women pay for four months of a membership. In June she slept through the night for the eleventh night in a row, and stopped counting.
What Gale does
A clinician who treats you, a phone that only speaks when she asked it to, and a bill you already know how to pay
Between the visits: what you see
You book a video visit with a clinician trained in menopause, and you see the price before you book. You leave that first visit with a treatment and a plan, written with her. In the plan she writes down, in her own words, what she wants to hear about: tell me if the night sweats aren't down by half by week six; tell me if they come back after we taper.
Most evenings you log one number. The app watches for her sentences and nothing else. When one comes true, it tells you, in her words, with your numbers, and one tap books a visit:
GALE · NOW
Julie asked to hear if your hot flashes came back after the taper.
Your log: 1 to 3 a day in the month before your last change. 6 a day this week.
Book 20 minutes with Julie · Thu 4:30 or Fri 9:00 · $75
BookNot now
This message is from Gale's software, not from Julie.
Every card has that shape: her sentence, your numbers, a time and a price. If you tap Book, you get a visit within seven days.
Some things don't wait for a card. Bleeding after menopause, which in about 9% of women turns out to be endometrial cancer 76, gets one instruction: call the office today. Chest pain or signs of a clot while on hormone therapy get another: call 911.
Yours: what stays on your phone
The app is yours. It keeps your sleep, food, activity, medication, cycle and mood, and your private conversations with it, and nothing leaves your phone unless you send it. When you tap Book, you see exactly what Julie will receive, with a switch beside each item. Turn off the note about your marriage and she never sees it, or that it was there.
Not now means not now. Julie sees no message, no flag and no count. Nobody follows up on it.
Yours: what Julie sees
A booking, with a short brief you approved. Not a stream of data, not a dashboard of your sleep. She goes through the brief with you in the visit, where it belongs.
Already paid for: what it costs
There is no new kind of plan to buy into. Visits are ordinary visits, so the coverage and money you already have pay for them: your insurance, your HSA or FSA, cash, and at 65, Medicare. Paying yourself, a first visit is $250 and a follow-up $150, shown before you book. A visit the app offers you costs $150, $75 or $25, by income, and never more than a follow-up. If we miss the seven-day window on a cash visit, it's free. The app carries no fee.
How we'll know it works
No woman has used this yet. When she does, we will count the days women go without relief in their first three months, against usual care, and publish the result either way. We are aiming for four weeks of her life back.
For clinicians
Julie, this is the practice you trained for
You know the patient who says "I tried hormones once." You never saw her stop. She stopped at week six, alone, over something you could have fixed in ten minutes. This is a way to hear from her before she gives up, without adding one more thing to watch.
Every request that reaches you is one you wrote. You decide, at the visit, what you want to hear about and when. The app looks for that and nothing else. It will not tell her a number is abnormal, will not suggest a drug, and will not send you a reading at midnight. Primary-care physicians already get about 49 inbox messages a day 77. You get bookings, with a brief she chose to send, and she arrives with her numbers already on the screen. Go through them together: a list of questions a clinician never gets to makes visits longer and patients more anxious 78.
And the visit is paid, the way any visit is, by insurance, an HSA or cash. You aren't reviewing data for free between appointments. You are seeing a woman who needs you, at the moment she needs you, and she already trusts the message that brought her, because the words in it are yours.
How many nights last month did a hot flash wake you?
Receipts
About one hour of exam time a year computed
The National Ambulatory Medical Care Survey estimates 320.7 visits to office-based physicians per 100 people in 2019, the last year it measured those practices 1. Electronic-record timestamps from 21.0 million primary-care visits in 2017, at one vendor's clients, put the average exam at 18.0 minutes 2. 3.207 visits × 18.0 minutes = 57.7 minutes, drawn as one square. Direct observation gives 13.3 minutes a visit 3 and physicians' own reports 23.5 4, or 43 and 75 minutes a year; the drawing does not change. The survey leaves out hospital clinics, emergency departments, health centers and independent nurse practitioners, so the visit count is a floor, and it is an average across a skewed distribution, not a typical person's year 1.
About 21 hours inside the system, 8,739 outside computed
Office visits: 3.207 × 121 minutes, the patient's total time per visit including 37 minutes of travel and 84 in the clinic 5 = 6.5 hours. Emergency department: 0.427 visits per person in 2021 7 × an assumed 3 hours = 1.3 hours. Hospital: 562 inpatient days per 1,000 people in 2024 6 × 24 hours = 13.5 hours. Total 21.2 hours; 8,760 − 21 = 8,739. Calls, portal messages, pharmacy trips and forms are not counted.
About 4 hours and 38 with three or four chronic conditions computed
RAND's chartbook of the 2014 Medical Expenditure Panel Survey gives adults with three or four chronic conditions 12 outpatient visits a year, a 20% chance of an emergency visit and a 10% chance of a hospital stay 8. Exam time: 12 × 18 minutes 2 = 3.6 hours, drawn as four. Inside the system: 12 × 121 minutes 5 = 24.2 hours, plus 0.20 × 3 hours, plus 0.10 × 5.4 days × 24 hours, for 37.8 hours. The 5.4-day average stay is from hospital-association figures compiled by KFF, outside our checked evidence.
One Medicare patient in six readmitted cited
In fiscal 2023, 15.7% of traditional Medicare discharges were followed by a readmission within 30 days. Pooled trials of early physician follow-up after discharge found no significant effect on readmission 9.
26.9 in 100, then 4.7, and 0.5 cited
From the National Health and Nutrition Examination Survey, 13,048 women across the survey years: hormone therapy use among US postmenopausal women fell from 26.9% in 1999 to 4.7% in 2020, and was 0.5% among non-Hispanic Black women in 2020 10. The denominator is every postmenopausal woman, with symptoms or without, so these figures show how common the treatment became, not how many women went without it. The 10.1 years of frequent hot flashes for African American women, against 7.4 overall, is from the SWAN cohort, as its investigators reported 11.
81% never treated cited
Kapoor and colleagues, Maturitas 2026: 81.3% of 1,011 employed US women aged 40 to 65 with frequent hot flashes, surveyed through an online panel, had never been treated 12. The survey was funded by Astellas, which sells a menopause drug; we report it as its authors did 13.
23% on the problem list, about 6% on hormone therapy cited
Bevry and colleagues, Menopause 2024: a chart review linked to a survey of 229 primary-care women with moderate or worse hot flashes at one health system. About 6% used hormone therapy, about 15% a non-hormone prescription, and 23% had the symptoms on their problem list 14. The 23% counts the problem list only, not every mention in the chart.
6.8% prepared, 34% would not offer cited
Kling and colleagues, Mayo Clinic Proceedings 2019: 183 family-medicine, internal-medicine and OB-GYN residents, a 26% response rate. 6.8% felt adequately prepared to manage menopause, 34.4% would not offer hormone therapy to an appropriate candidate, and 20.3% had had no menopause lectures 1516.
10.7% against 20.7% cited
Chesnokova and colleagues, JAMA Network Open 2026, national survey data from 2013 to 2020, 1,666 women aged 45 to 64: 10.7% of Medicaid-insured women had used hormone therapy against 20.7% of privately insured women. After accounting for insurance, the Black-white difference was no longer significant 17. A survey at one point in time cannot show that insurance alone causes the gap.
75 to 88 of 100 stopped cited
Saadedine and colleagues, Climacteric: electronic records of more than 1.5 million women with natural menopause in the US, UK and Germany. Six-month discontinuation of a first treatment was 75–88% for hormone therapy and 65–85% for antidepressants, the range spanning the three countries 18. The 64.8% for GLP-1 drugs is Rodriguez and colleagues, JAMA Network Open, 125,474 adults with overweight or obesity and no type 2 diabetes, 65.4% of them women, mean age 54.4, within one year 19. Neither study shows how many stopped while still having symptoms.
47.6% stopped over cost; 31.2% on advice, 26.0% from fear cited
Gasoyan and colleagues, Obesity: 288 randomly sampled charts at one health system of people who stopped semaglutide or tirzepatide for obesity. Cost or insurance 47.6%, side effects 14.6%, shortages 11.8% 20. Bunnewell and colleagues, BJOG, pooled 69 studies of 32,213 women: 31.2% stopped hormone therapy on a clinician's advice, 26.0% for fear of risks, and 20.7% restarted 21. The unfilled-prescription figures, 6% at no cost and 61% above $250, are for new prescriptions at 14 days in 2021, before Medicare's drug-cost caps 22.
13 of 17 results marked measured
The forest plot in Section VI is the one from our previous essay, The Other Hours. The plot draws 17 results from 15 studies (PRO-TECT and TIM-HF2 each appear twice). We coded each result by who received the reports: 13 results, from 11 studies, with a named clinician or nurse who could change treatment, 2 with none (the BEAT-HF telemonitoring trial and the caring-texts trial), and 2 where no one is reported (the six-system cancer rollout and the pooled safety-planning studies). The founding cancer trial's survival figure of 0.83 came from an analysis decided after the trial ran 23; the repeat trial's 0.99 was its preregistered main result 24, and its fewer emergency visits (0.84) a secondary result 25. The French trial was stopped early at its 37th death 26. The coding is ours and no one else has checked it; no trial randomized who answered. The depression figures, 4 weeks against 8 to remission, come from a trial whose groups had converged by week 24 38.
Portal use from 25% to 65%; 59 and 7 in 100 cited
The share of all US adults who were offered an online record or portal and opened it at least once in the past year, from the national HINTS surveys as reported by ONC: 25% (2014), 27% (2017), 30% (2018), 37% (2019), 38% (2020), 57% (2022), 65% (2024) 49. The 2022 jump coincides with the survey adding a web response option 50. In 2024, 7% of US adults used an app to combine records from different portals 52; 59% of US adults had more than one online record or portal, the brief's own base being all individuals nationally, including those with none 51. Dossia's dates and reported size and HealthSpace's end date come from sources we could not verify. The null trials of patient-held records are 53 and, for the combining app, 7.5% against 7.6% of patients with no medication-list discrepancies, a wide interval 54.
One reader in five perceived a mistake cited
Bell and colleagues surveyed patients at three US health systems in 2017. Of 136,815 patients invited, 21.7% responded; 22,889 read notes and answered the error questions 55. 4,830 (21.1%) perceived a mistake; 2,043 (8.9% of readers) called it somewhat or very serious; 480 (2.1%) called it very serious. The errors were patient-reported and never checked against the chart 55. What happened next is from two studies of follow-up: about four in ten Swedish portal users who found an error did nothing, and about 0.2% of patients who requested their chart at one Michigan health system filed a formal amendment over seven years, half of them approved 56.
11.5% appealed, 80.7% overturned cited
KFF's analysis of CMS data for 2024: 52.8 million Medicare Advantage prior-authorization determinations, 4.1 million denials (7.7%), 11.5% of denials appealed, 80.7% of appeals overturned in full or in part 58. Per 1,000 denials, 115 appealed and 93 overturned (1,000 × 0.115 × 0.807 = 92.8); our computation. Appealed denials are self-selected, so this says nothing about what appealing the rest would win 60. The 85% who did not appeal is from a 2023 national survey; a 2025 survey of privately insured adults found about half of those with a denial appealed 63.
30 denials that met Medicare's rules, 9 reversed cited
The HHS inspector general sampled prior-authorization denials from June 1–7, 2019, at the 15 largest Medicare Advantage organizations and had physicians review the files 62. The report's appendix lists 30 sampled denials that met Medicare's coverage rules; 9 had been reversed by the time of the inspector general's data request 61. Weighted to all such denials, an estimated 13% met the rules 62. All 7 of the 30 that were appealed were reversed 62. The chart's 247 reviewed denials were read from the report's Appendix B (OEI-09-18-00260), outside our checked evidence.
0.31% of marketplace denials appealed cited
KFF's analysis of CMS transparency data for HealthCare.gov insurers in 2024: about 85 million of 451 million in-network claims denied, and 262,982 consumer appeals, or 0.31%. A quarter of the denials were administrative and 5% cited medical necessity 59. Insurers report these counts themselves, by claim line, and no one audits them 79.
Coverage people already have cited
8% of adults 35 and over were up to date on every high-priority preventive service they were eligible for, in 2015; the study did not check coverage 64. 14% of employer-insured adults answered questions about four insurance terms all correctly 65. At one employer with a 48-plan menu in 2010, 55% of employees with single coverage chose a plan that cost more in every outcome after adjusting for how premiums are taxed, 61% before, about $372 a year 66. 54% of Medicare Advantage enrollees reported dental coverage though more than 95% had plans that offered it 67. In 2018–2020, an estimated 56.7% of people eligible for a Medicare Savings Program were enrolled, with eligibility simulated from survey answers and before a 2023 rule meant to ease enrollment 68. About half of FSA holders forfeited money in 2023, $436 on average 69. 23% of traditional Medicare beneficiaries used the annual wellness visit in 2016 70. The Medicare diabetes prevention program enrolled 9,015 of about 16 million eligible people between 2018 and 2024: 9,015 ÷ 16,000,000 = 0.056%, rounded to 0.06% 71.
10, 39 and 55 of 100 visits cited
Kravitz and colleagues, JAMA 2005: a randomized trial with standardized patients, 298 visits to 152 physicians. For adjustment disorder, antidepressants were prescribed in 10% of visits with no request, 39% with a general request and 55% with a brand request 57. The chart draws each percentage as squares out of 100; the real groups were smaller, because the visits were split across two conditions and three kinds of request.
30 of 264 computed
Wyatt and colleagues, JAMIA: 264 patients seen at one health system after an Apple Watch abnormal-pulse alert; 59.8% had an echocardiogram and 11.4% received a clinically actionable cardiovascular diagnosis 40. 264 × 0.598 = 157.9, drawn as 158; 264 × 0.114 = 30.1, drawn as 30. The study does not say the 30 were among the 158. The one-in-five anxiety figure is from a separate study of wearable users with atrial fibrillation, as its authors reported 41. The limit of two cards in 30 days and the 30-day rest after Not now are the limits already in our code 42.
$35–46 worth of relief, and a $150 ceiling cited
Women with hot flashes said they would pay an extra $35 to $46 a month for a treatment giving much better sleep or far fewer hot flashes, in an industry-funded study 72. Hims & Hers reported $83 average monthly revenue per subscriber in 2025, across all products 73. Cash GLP-1 memberships and the maker's cash price for the brand drug are from the companies' price pages 74. The $150 cap is IRS Notice 2026-5: from 2026 a direct primary-care arrangement no longer blocks health savings account eligibility if its fees stay at or under $150 a month for one person and it provides only primary care from primary-care practitioners, among whom OB-GYNs are not listed 75. Gale's own prices, $250, $150 and the $150, $75 and $25 bands, are our plan's.
49 messages a day cited
Primary-care physicians receive an average of 49.3 inbox messages a day 77. Patient question lists that a clinician does not address make visits longer and patients more anxious 78. About 9% of women with bleeding after menopause have endometrial cancer 76.
Dana and Julie cited
Dana and Julie are composites, not real people. Dana's age, her hot-flash counts, her treatment, her taper, her denial and her bills are invented to show how the plan would work; Julie is a composite nurse practitioner. Dana's quoted line is invented too. The patient quoted on BRCA1 is real, from Bell and colleagues, JAMA Network Open, 2020. No real woman has used any part of the product described here: Gale runs on synthetic data until it can legally hold real health records.
How this was made measured
AI agents working for us did the research, checked the headline findings against their sources, and drafted this essay; we edited it and we stand behind it. We run Gale, and we have a commercial interest in you believing it. Findings with a W id that were not separately checked are given as their authors reported them; findings from our previous essay, The Other Hours, are cited only where its checks confirmed or corrected them. We asked an outside reviewer, an AI from a different company, to score the plan for its chance of a dramatic improvement in women's care; it scored it 7 out of 10, and said that score belongs to the care itself, not to the app alone. What we don't know yet: we have no way yet for women on Medicaid to pay for Gale, and the study that would show whether it works has still to be run. Two defaults in our software changed before any real woman used it: a clinician could see how many items a woman held back, and a switch that sends her record to the AI model we use started on 8081. On September 22, 2026 the count was removed, and the switch now stays off until she says yes.
References
NCHS, National Ambulatory Medical Care Survey 2019 National Summary Tables (Tables 1, 3); NCHS Data Brief 408 for 2018. In 2019 — still the most recent year NCHS measured office-based physician practice, since every NAMCS release from 2021 onward covers only the health center component — Americans made an estimated…https://www.cdc.gov/nchs/data/ahcd/namcs_summary/2019-namcs-web-tables-508.pdf[E-099 · national probability survey of office-based physicians (398 ]
Neprash HT et al., Medical Care 59(1):62-66, 2021. EHR activity timestamps from 21.0 million primary-care visits in 2017, drawn from one vendor's national client base (athenahealth, whose employee co-authored the study), put the average exam at…https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:33301282&format=json&…[E-101 · observational analysis of national EHR (athenahealth) timest]
Gottschalk A, Flocke SA, Annals of Family Medicine 3(6):488-93, 2005; Tai-Seale M et al., Health Services Research 42(5):1871-94, 2007. Direct observation of family physicians found 13.3 minutes of face time plus visit-specific work per visit versus 18.7 minutes physician-reported in NAMCS 2003, i.e., national self-reported visit…https://pmc.ncbi.nlm.nih.gov/articles/PMC1466945/[E-102 · time-motion direct observation (11 physicians, 611 visits, 1]
NCHS, NAMCS 2019 National Summary Tables (Tables 26–27). Physician-reported time with the physician at office visits averaged 23.5 minutes (median 19.4) in 2019, with 39.2% of visits at 15 minutes or less; the field was missing for 32% of visits and…https://www.cdc.gov/nchs/data/ahcd/namcs_summary/2019-namcs-web-tables-508.pdf[E-100 · national survey, physician/record report with imputation]
Ray KN, Chari AV, Engberg J, Bertolet M, Mehrotra A, American Journal of Managed Care 21(8):567-74, 2015. Each ambulatory visit costs the patient 121 minutes on average (37 minutes travel, 84 in clinic), so physician face time is on the order of one-fifth of clinic time and about 15% of the total time…https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&id=26295356&rettype=ab…[E-103 · secondary analysis of American Time Use Survey 2003–2010]
KFF analysis of AHA Annual Survey (Key Facts About Hospitals; State Health Facts); MedPAC Data Book July 2024 Charts 6-12 and 6-15; Tipping MD et al., J Hosp Med 2010. KFF's compilation of AHA Annual Survey data shows US community hospitals delivered 562 inpatient days per 1,000 population as of the most recent year shown (2024), down from 681.9 in 2000 — not…https://www.kff.org/other-health/state-indicator/inpatient-days-by-ownership/[E-105 · administrative data (AHA Annual Survey; Medicare MedPAR clai]
NCHS, National Hospital Ambulatory Medical Care Survey 2021 Emergency Department Summary Tables (Tables 1, 3, 23); NCHS Data Brief 478. There were 42.7 ED visits per 100 persons in 2021 (49.6 for 65+, 65.5 for 75+); by expected payment source, Medicaid/CHIP accounted for 37.3% of ED visits and private insurance for 32.3% (the…https://www.cdc.gov/nchs/data/nhamcs/web_tables/2021-nhamcs-ed-web-tables-508.pdf[E-106 · national probability survey of ED visits (16,207 sampled vis]
Buttorff C, Ruder T, Bauman M, Multiple Chronic Conditions in the United States, RAND TL-221, 2017 (MEPS 2014). The most recent national chartbook to break health-care use down by condition count — RAND's Multiple Chronic Conditions in the United States, drawn from the 2014 Medical Expenditure Panel Survey…https://www.rand.org/content/dam/rand/pubs/tools/TL200/TL221/RAND_TL221.pdf[E-104 · chartbook of MEPS 2014 household survey (values read from ch]
Jencks SF, Williams MV, Coleman EA; NEJM 2009;360:1418 (2003-04 claims). In 2003–04 Medicare fee-for-service claims, 19.6% of 11.9 million discharges were followed by a rehospitalisation within 30 days, and — on the figure NEJM corrected in 2011 — 50.2% of the medical…https://www.nejm.org/doi/full/10.1056/NEJMsa0803563[E-127 · administrative data]
Yang & Toriola, JAMA Health Forum 2024 (NHANES, n=13,048). Menopausal hormone therapy use among US postmenopausal women fell from 26.9% in 1999 to 4.7% in 2020, and was 0.5% among non-Hispanic Black women in 2020.https://pubmed.ncbi.nlm.nih.gov/39331372/[W-037 · serial cross-sectional, nationally representative]
Avis et al., JAMA Intern Med 2015. African American women had the longest median duration of frequent vasomotor symptoms in SWAN.https://pubmed.ncbi.nlm.nih.gov/25686030/[W-087 · prospective multiethnic cohort, n=1449 with frequent VMS]
Kapoor et al., Maturitas 2026 (Astellas-funded). 81.3% of US women aged 40-65 with frequent hot flashes had never received treatment.https://pubmed.ncbi.nlm.nih.gov/42617245/[W-041 · online panel survey, n=1,011, industry-funded]
Kapoor et al., Maturitas 2026 (industry-funded). Among employed US women with frequent hot flashes, most had never been treated and few reported employer support.https://pubmed.ncbi.nlm.nih.gov/42617245/[W-089 · online panel survey, n=1011]
Bevry et al., Menopause 2024. Among 229 primary-care women with moderate-or-worse vasomotor symptoms, about 6% used hormone therapy, about 15% a non-hormone prescription, and only 23% had the symptoms on their problem list.https://pubmed.ncbi.nlm.nih.gov/39352126/[W-038 · retrospective chart review linked to a survey, single health]
Kling et al., Mayo Clin Proc 2019. Only 6.8% of US family medicine, internal medicine and OB-GYN residents felt adequately prepared to manage menopause, and 20.3% had no menopause lectures.https://pubmed.ncbi.nlm.nih.gov/30711122/[W-039 · cross-sectional survey, n=183, 26% response rate]
Kling et al., Mayo Clin Proc 2019. Only a small minority of US residents in family medicine, internal medicine and OB/GYN felt adequately prepared to manage menopause, and a third would not offer hormone therapy to an appropriate…https://pubmed.ncbi.nlm.nih.gov/30711122/[W-088 · cross-sectional survey, 183 residents, 26% response]
Chesnokova et al., JAMA Netw Open 2026. Medicaid-insured women aged 45-64 were half as likely as privately insured women to have used menopausal hormone therapy, and the Black-White difference was attenuated to non-significance after…https://pubmed.ncbi.nlm.nih.gov/42467433/[W-085 · cross-sectional, NHANES 2013-2020, n=1666]
Saadedine M et al., Climacteric. In >1.5 million women with natural menopause in the US, UK and Germany, most stopped their first symptom treatment within 6 months.https://pubmed.ncbi.nlm.nih.gov/40735862/[W-015 · EHR cohort]
Rodriguez PJ et al., JAMA Netw Open. Among US adults with overweight/obesity starting a GLP-1 RA (65.4% women, mean age 54.4), 64.8% without type 2 diabetes stopped within one year, and GI adverse events raised the hazard of stopping…https://pubmed.ncbi.nlm.nih.gov/39888616/[W-013 · retrospective EHR cohort, n=125,474]
Gasoyan H et al., Obesity. In charts of 288 randomly sampled patients who stopped semaglutide or tirzepatide for obesity, cost or insurance was the most common reason, far ahead of side effects.https://pubmed.ncbi.nlm.nih.gov/41039650/[W-014 · chart review, single health system]
Bunnewell S et al., BJOG. Across 69 studies, HRT discontinuation averaged 51.3%, led by clinician recommendation, fear of risks and preference for a natural approach; symptoms returned often and about one in five restarted.https://pubmed.ncbi.nlm.nih.gov/40999909/[W-016 · systematic review, 32,213 women]
IQVIA Institute, The Use of Medicines in the U.S. 2022, Exhibit 35 (LAAD sample claims projected nationally). In 2021, one in ten new-to-product prescriptions prepared at US pharmacies — 81 million — was still undispensed 14 days after the pharmacy prepared it, and abandonment rose steeply with the…https://static1.squarespace.com/static/54d50ceee4b05797b34869cf/t/6262dbae038fd5690b1543be…[E-128 · administrative data (pharmacy claims sample, projected)]
Basch E et al., JAMA 2017;318:197-198. The same trial's long-term follow-up found a 5-month median survival advantage, but the survival analysis was decided post hoc and the trial enrolled 2007-2011 at one academic center.https://pmc.ncbi.nlm.nih.gov/articles/PMC5817466/[E-044 · RCT secondary analysis, post hoc endpoint]
Basch E et al., Nature Medicine 2025;31:1225-1232. PRO-TECT's pre-registered primary endpoint, overall survival, showed no difference between ePRO monitoring and usual care.https://pmc.ncbi.nlm.nih.gov/articles/PMC12184200/[E-047 · cluster RCT, 52 practices, primary endpoint]
Basch E et al., Nature Medicine 2025;31:1225-1232. In the same trial, these benefits appeared in secondary outcomes rather than the primary one: ePRO monitoring prolonged time to first emergency visit (HR 0.84, 95% CI 0.71–0.98; nominal P=0.03,…https://pubmed.ncbi.nlm.nih.gov/39920394/[E-048 · cluster RCT, secondary outcomes]
Denis F et al., JNCI 2017;109(9):djx029 and JAMA 2019;321:306-307. In France, weekly web symptom reports e-mailed straight to the oncologist after lung-cancer treatment detected relapse earlier and prolonged survival, but the trial was stopped at the 37th death…https://academic.oup.com/jnci/article/109/9/djx029/3573360[E-045 · multicenter phase III RCT, stopped early at interim]
Hassett MJ et al., J Clin Oncol 2025;43(16_suppl):11001 (eSyM/SIMPRO, ASCO 2025 abstract). The largest pragmatic deployment, an EHR-integrated ePRO program across six health systems, did not reduce emergency or inpatient events; only half of exposed patients ever reported symptoms and…https://www.semanticscholar.org/paper/f391b137c1903149d5d6029b41077ac7308a151e[E-050 · stepped-wedge cluster RCT, conference abstract (text obtaine]
Ong MK et al. (BEAT-HF), JAMA Internal Medicine, 2016. Combining health-coaching calls with daily home telemonitoring after a heart-failure hospitalisation did not reduce 180-day all-cause readmission compared with usual care.https://pubmed.ncbi.nlm.nih.gov/26857383/[E-323 · RCT]
Koehler et al., TIM-HF, Circulation, 2011. A 24/7 physician-led telemedical centre with high adherence still produced no mortality benefit when applied to stable ambulatory heart-failure patients, showing that a response apparatus is…https://pubmed.ncbi.nlm.nih.gov/21444883/[E-143 · RCT]
Abraham et al., Lancet, 2011; Brugts et al., MONITOR-HF, Lancet, 2023. CHAMPION (Lancet 2011, n=550): an implanted pulmonary-artery pressure sensor whose readings clinicians acted on cut heart-failure hospitalisation at 6 months, HR 0.70 (95% CI 0.60-0.84), and HR…https://pubmed.ncbi.nlm.nih.gov/37220768/[E-144 · RCT]
Svendsen JH et al. (LOOP Study), The Lancet, 2021. Continuous rhythm monitoring with an implantable loop recorder in older adults with stroke risk factors tripled atrial-fibrillation diagnoses and anticoagulation but did not significantly reduce…https://pubmed.ncbi.nlm.nih.gov/34469766/[E-324 · RCT]
Stanley B et al., JAMA Psychiatry, 2018. In a non-randomised comparison across nine VA emergency departments (n = 1,640; 88% men), safety planning plus at least two structured follow-up calls was associated with roughly half the odds of…https://pmc.ncbi.nlm.nih.gov/articles/?term=jamapsychiatry.2018.1776[E-021 · cohort comparison (non-randomised, 9 VA EDs, propensity adju]
Nuij C et al., British Journal of Psychiatry, 2021; Doupnik SK et al., JAMA Psychiatry, 2020. Pooled across trials, safety-planning-type interventions reduce suicidal behaviour but have no detectable effect on suicidal ideation; single-encounter acute-care interventions also reduce…https://pubmed.ncbi.nlm.nih.gov/35048835/[E-022 · meta-analyses (6 studies / 3,536; 14 studies / 4,270)]
Comtois KA et al., JAMA Psychiatry, 2019. Eleven caring text messages over a year to at-risk Soldiers and Marines did not change the primary outcomes (current ideation, risk incidents) but were associated with fewer suicide attempts and…https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:30758491&format=json&…[E-024 · RCT (3 military installations; 70.2% followed to 12 months)]
Billingy NE et al., JAMA Netw Open 2024;7(8):e2428975 (SYMPRO-Lung). In Dutch lung-cancer care, weekly ePRO monitoring improved quality of life for a year, and sending alerts to the patient alone (self-management advice) worked as well as sending them to the…https://pubmed.ncbi.nlm.nih.gov/39186274/[E-052 · stepped-wedge cluster RCT]
Husain N et al., 2025; Kendrick T et al., Cochrane Database of Systematic Reviews, 2016; Rognstad 2023, Barkham 2023, de Jong 2021. Measurement-based care replicated on speed, not magnitude: the 2025 multicentre assessor-blinded RCT found median time to remission 4 weeks vs 8 with no difference in remission or response rates…[E-368 · RCT; Cochrane meta-analysis]
Kendrick T et al., Cochrane Database of Systematic Reviews, 2016. Routine outcome monitoring with feedback (OQ-45/ORS) showed no evidence of symptom benefit across pooled trials, and no change in number of sessions.https://pmc.ncbi.nlm.nih.gov/articles/PMC6472430/[E-017 · Cochrane systematic review / meta-analysis of 17 RCTs]
Wyatt et al., JAMIA. Among 264 patients seen after an Apple Watch abnormal-pulse alert, 59.8% had an echocardiogram and only 11.4% received a clinically actionable cardiovascular diagnosis.https://pubmed.ncbi.nlm.nih.gov/32979046/[W-080 · retrospective chart review]
J Am Heart Assoc. Patients with AF who used wearables showed more symptom preoccupation and more AF-specific health-care use than matched non-users, and 20% felt anxious and always contacted their doctor after an…https://pubmed.ncbi.nlm.nih.gov/39011944/[W-081 · retrospective propensity-matched, n=172]
proactive/policy.ts, engine.ts. The proactive engine is opt-in, capped at 2 unprompted check-ins per 30 days with a 10-day minimum gap and a 30-day cooldown after 'not now', triggers only on an activity drop, and is fail-closed…[W-100 · shipped code]
Google Official Blog (Aaron Brown), 'An update on Google Health and Google PowerMeter', 2011. Google retired Google Health on January 1, 2012, kept data downloadable for one more year, and said "any data that remains in Google Health after that point will be permanently deleted"; the…https://googleblog.blogspot.com/2011/06/update-on-google-health-and-google.html[E-183 · other (primary corporate statement)]
Ko H, Turner T, Jones C, Hill C. Quality & Safety in Health Care, 2010; Greenhalgh T et al. BMJ, 2010. Patient-held records have historically failed on adoption and shown no clear clinical benefit: 14 controlled studies of paper patient-held records found none, and the NHS's national personal…https://pubmed.ncbi.nlm.nih.gov/20511601/[E-070 · systematic review of comparative studies; mixed-method natio]
Apple Newsroom press releases 2018-2022. Apple Health Records grew from 12 institutions at its January 2018 beta to more than 800 institutions across 12,000 locations by July 2022, because it pulls records through FHIR APIs that federal…https://www.apple.com/newsroom/2022/07/how-apple-is-empowering-people-with-their-health-in…[E-194 · other (company statements)]
ONC/HHS, Federal Register 85 FR 70064 (interim final rule), 2020. ONC's COVID-era interim final rule (85 FR 70064, Nov. 4, 2020) extended the Cures Act Final Rule's timetable rather than creating the regime: information blocking became applicable April 5, 2021;…https://www.federalregister.gov/documents/full_text/html/2020/11/04/2020-24376.html[E-113 · federal regulation (primary text)]
The Sequoia Project (TEFCA RCE), 2026 ASTP Annual Meeting deck and RCE press releases, 2026. The Sequoia Project — the TEFCA Recognized Coordinating Entity, paid by ASTP/ONC under a five-year contract awarded in August 2023 — reports that documents moved across TEFCA rails rose from…https://www.healthit.gov/wp-content/uploads/2025/09/2026AnnualMeeting_TEFCA-from-A-to-Z.pd…[E-114 · administrative data (self-reported by the RCE; counting meth]
Rolnick et al., University of Pennsylvania Health System, JMIR 2022. At one of Apple Health Records' launch health systems, 0.27% of patient-portal users connected the feature in its first ten months, and adopters skewed male, white and privately insured.https://pmc.ncbi.nlm.nih.gov/articles/PMC8826150/[E-193 · cross-sectional survey linked to EHR data (single center)]
ASTP/ONC Data Brief 77 (HINTS 4-7), 2025. The share of US adults who were offered and accessed their online medical record rose from 38% in 2020 to 57% in 2022 and 65% in 2024, with app-based access rising from 38% to 57%; the largest…https://www.healthit.gov/data/data-briefs/individuals-access-and-use-patient-portals-and-s…[E-116 · repeated cross-sectional national survey (HINTS); HINTS 7 re]
ONC Data Briefs 69 (Oct 2023) and 77 (July 2025), HINTS 4-7. The share of US adults offered and having accessed an online medical record or patient portal at least once in the prior year more than doubled from 25% in 2014 to 65% in 2024, and among those…https://www.healthit.gov/data/data-briefs/individuals-access-and-use-patient-portals-and-s…[E-191 · government report on nationally representative cross-section]
ASTP Data Brief 77 (July 2025) Appendix Table 1; HINTS 6 and HINTS 7 instruments. The reported rise in using an app to combine portals, from 2% (2022) to 7% (2024) of adults, crosses a change in who was asked; among people with more than one portal the change was 5% to 7%, and…https://healthit.gov/wp-content/uploads/2025/07/2024-HINTS-Patient-Access-DB77_508.pdf[E-600 · cross-sectional survey (HINTS 7, n=7,278, response rate 27.3]
Williams JG et al. Qual Health Care 2001; Cornbleet MA et al. Palliat Med 2002; Lester H et al. BJGP 2003; Warner JP et al. Br J Psychiatry 2000; Bjerkeli Grovdal L et al. Scand J Prim Health Care 2006. Randomised trials of paper patient-held records in cancer, palliative care, schizophrenia, long-term mental illness and child health found no effect on quality of life, service use, admissions or…https://www.ebi.ac.uk/europepmc/webservices/rest/search?query=EXT_ID:11533423%20AND%20SRC:…[E-507 · RCTs and cluster RCTs]
Gionfriddo MR et al., JMIR Formative Research 2022. An EHR-integrated app combining patient-reported, EHR and pharmacy data did not improve medication-list accuracy against a pharmacist's best-possible medication history.https://pmc.ncbi.nlm.nih.gov/articles/PMC8941436/[E-510 · cluster RCT]
Bell SK, Delbanco T, Elmore JG et al., JAMA Network Open, 2020. In a 2017 survey of 22,889 note readers at three US health systems (21.7% response rate), one in five reported perceiving a mistake in their notes, and 42% of those rated it somewhat or very…https://pmc.ncbi.nlm.nih.gov/articles/PMC7284300/[E-001 · cross-sectional survey (3 US health systems, 79 practices; 1]
Bärkås A et al., JMIR, 2023; Hanauer DA et al., JAMIA, 2014. Finding an error rarely leads to fixing it: most portal users who find one take no action, and the formal HIPAA amendment right is almost never used and is denied about half the time.https://pmc.ncbi.nlm.nih.gov/articles/PMC10656659/[E-004 · national cross-sectional portal survey; 7-year administrativ]
Kravitz et al., JAMA 2005. Patients' drug requests sharply raised antidepressant prescribing, including for adjustment disorder where antidepressants are usually not indicated, while also raising minimally acceptable care…https://pubmed.ncbi.nlm.nih.gov/15855433/[W-029 · RCT with standardized patients, 298 visits, 152 physicians]
KFF analysis of CMS Part C reporting data, 2026-01-28 (series 2019-2024). In 2024 — the most recent year of CMS Part C reporting data — Medicare Advantage insurers made 52.8 million prior-authorization determinations and issued 4.1 million adverse determinations (7.7%),…https://www.kff.org/medicare/medicare-advantage-insurers-made-nearly-53-million-prior-auth…[E-085 · administrative data]
KFF analyses of CMS transparency data (2023 data; 2024 data published 2026-03-24). HealthCare.gov marketplace insurers denied 19% of in-network claims in both 2023 (86 million of 436 million) and 2024 (85 million of 451 million) — but most reported denials are administrative or…https://www.kff.org/patient-consumer-protections/claims-denials-and-appeals-in-aca-marketp…[E-086 · administrative data]
KFF analysis of CMS Part C reporting data, January 2026. Of the 4.1 million prior-authorization requests Medicare Advantage insurers denied in 2024 — 7.7% of 52.8 million determinations — just 11.5% were appealed to the insurer, and 80.7% of those…https://www.kff.org/medicare/issue-brief/nearly-50-million-prior-authorization-requests-we…[E-463 · administrative data]
HHS OIG, Some Medicare Advantage Organization Denials of Prior Authorization Requests Raise Concerns About Beneficiary Access to Medically Necessary Care, OEI-09-18-00260 (2022), Appendix B. The inspector general's report lists, in its Appendix B, 30 sampled prior-authorization denials that met Medicare coverage rules; 9 of them had been reversed by the time of OIG's data request and…https://oig.hhs.gov/documents/evaluation/3150/OEI-09-18-00260-Complete%20Report.pdf[E-621 · government report (audit of a stratified random sample)]
HHS Office of Inspector General, OEI-09-18-00260, 2022-04-27. A 2022 HHS Inspector General audit drew a stratified random sample of 250 prior-authorization denials from a single week — June 1-7, 2019 — at the 15 largest Medicare Advantage organizations,…https://oig.hhs.gov/documents/evaluation/3150/OEI-09-18-00260-Complete%20Report.pdf[E-087 · government audit with stratified random sample and physician]
KFF Survey of Consumer Experiences with Health Insurance, 2023. In KFF's 2023 Survey of Consumer Experiences with Health Insurance (3,605 insured adults, Feb 21 to Mar 14, 2023), 18% of insured adults said their insurer had denied a claim for care they thought…https://www.kff.org/affordable-care-act/consumer-survey-highlights-problems-with-denied-he…[E-595 · cross-sectional survey]
Borsky A, Zhan C, Miller T, Ngo-Metzger Q, Bierman AS, Meyers D; Health Affairs 37(6):925-928; 2018. As of 2015 — still the most recent national measurement of this composite — only 8% of US adults aged 35 and older were up to date on *every* high-priority preventive service they were eligible…https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&id=29863918&retmode=xm…[E-239 · cross-sectional survey (MEPS)]
Loewenstein G, Friedman JY, McGill B, et al.; Journal of Health Economics 32(5):850-862; 2013 (full text verified via HBS open copy). Only 14% of employer-insured US adults aged 25–64 answered basic questions about all four of deductible, copay, coinsurance and out-of-pocket maximum correctly — accuracy ran from 78% on…https://www.hbs.edu/faculty/Publication%20Files/misunderstanding_health_insurance_2d7d6a99…[E-240 · cross-sectional survey (KnowledgePanel probability sample, e]
Bhargava S, Loewenstein G, Sydnor J; NBER w21160 (2015) / QJE 132(3):1319-1372 (2017). In 2010 one Fortune 100 US employer offered an unusual menu of 48 plans, in which 35 of the 36 lower-deductible plans cost more than an otherwise identical higher-deductible plan in every outcome.…https://www.nber.org/system/files/working_papers/w21160/w21160.pdf[E-241 · administrative plan-choice data (48-plan menu, 35 dominated)]
Cai CL, Iyengar S, Woolhandler S, Himmelstein DU; JAMA Network Open 8(1):e2454699; 2025. Pooled national survey data from 2017 to 2021 (MEPS and MCBS, 76,557 beneficiaries not also on Medicaid) found that only 54.2% of these Medicare Advantage enrollees reported having dental coverage…https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&id=39808428&retmode=xm…[E-242 · cross-sectional, pooled MEPS + MCBS 2017-2021, nationally we]
Kotb S, Su A, Sinaiko AD; JAMA Network Open 8(10):e2535408; 2025 (and Koma JW et al., Health Affairs 44(7):878-886, 2025). In 2018–2020 an estimated 20.9% of community-dwelling Medicare beneficiaries (95% CI, 19.8–22.0) met Medicare Savings Program income and asset criteria, and only 56.7% of them (95% CI, 54.5–59.0)…https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&id=41042507&retmode=xm…[E-243 · cross-sectional national survey (MCBS) with simulated eligib]
Employee Benefit Research Institute, FSA Database Issue Briefs (May 2024; May 2025), Jake Spiegel. Roughly half of health FSA accountholders forfeited funds to their employer in 2023, with an average forfeiture of $436 (2022: roughly half, $441); the average 2023 contribution was $1,413 and 88%…https://www.ebri.org/health/hsa-database/content/updates-from-ebri-s-flexible-spending-acc…[E-246 · administrative account data (recordkeeper panel, 3.2 million]
Lind KE, Hildreth KL, Perraillon MC; Medical Care 57(12):984-989; 2019 (and Misra A, Lloyd JT; Prev Med 2019). Claims-based use of Medicare's free annual wellness visit rose from 8.1% of fee-for-service beneficiaries in 2011 to 23.0% in 2016 (24% in 2017 in a separate claims sample), with use 10-12 points…https://pubmed.ncbi.nlm.nih.gov/31584462/[E-247 · MCBS-linked Medicare claims, nationally weighted]
CMS Innovation Center, MDPP Expanded Model Final Evaluation, Findings at a Glance; 2025. Between April 2018 and March 2024 the Medicare Diabetes Prevention Program enrolled 9,015 beneficiaries although an estimated 16 million people aged 65+ are eligible; only 414 suppliers existed as…https://www.cms.gov/priorities/innovation/data-and-reports/2025/mdpp-finalevalrpt-fg[E-248 · government program evaluation (independent contractor)]
Shiozawa et al., Expert Rev Pharmacoecon Outcomes Res 2023 (Astellas-funded). US women with vasomotor symptoms would pay an extra $35-46 a month for substantially better sleep or large reductions in hot flashes; this measures willingness to pay for a treatment, not for a…https://pubmed.ncbi.nlm.nih.gov/37650213/[W-042 · discrete choice experiment, n=467]
Ro, WeightWatchers and NovoCare pricing pages. Cash GLP-1 programmes charge a membership fee that excludes the drug (Ro $149/month month-to-month or from $74/month prepaid; WeightWatchers Med+ $74-84/month), while Novo's self-pay Wegovy costs…https://www.novocare.com/obesity/products/wegovy/savings-offer.html[W-045 · company price lists]
IRS Notice 2026-5 (OBBBA s.71308). Section 71308 of the OBBBA (not 71306) amended section 223(c)(1) of the Code to provide that months beginning after Dec 31, 2025, a direct primary care service arrangement (DPCSA) is not treated…https://www.irs.gov/pub/irs-drop/n-26-05.pdf[W-049 · law/reg]
Clarke et al., JAMA Intern Med. In women with postmenopausal bleeding, the pooled risk of endometrial cancer is 9%, and 91% of endometrial cancers present with such bleeding.https://pubmed.ncbi.nlm.nih.gov/30083701/[W-082 · systematic review and meta-analysis, 129 studies]
Nath B et al. 2021 (GAP-2-29); Tai-Seale M et al. 2019 (GAP-2-28); Singh H et al. 2009 and 2010 (GAP-2-32/33) — v2 ledger rows carried into v3 COUNTER-EVIDENCE section 12. An added between-visit signal lands on a clinician inbox already receiving 49.3 messages a day, where above-average system-generated message volume went with 40% higher probability of burnout, and…[E-375 · cross-sectional surveys and administrative EHR alert data]
Brown, Butow et al., Br J Cancer, 2001; Ramlakhan et al. scoping review PMC10349246, 2023 (as summarised in PATIENT_VIEW_EVIDENCE.md). Giving patients pre-visit question lists backfires when the clinician does not engage with the list: consultations lengthen and anxiety rises, so patient-side preparation without a provider loop…PATIENT_VIEW_EVIDENCE.md:18, 27-28[E-231 · RCT]
KFF analysis of CMS transparency data, 2024 data (published 2026). In 2024, insurers on HealthCare.gov denied about 19% of roughly 451 million in-network claims, with rates ranging from 3% to 36% by insurer. Consumers formally appealed fewer than 1% of denied…https://www.kff.org/affordable-care-act/claims-denials-and-appeals-in-aca-marketplace-plan…[E-267 · administrative data]
BriefIncludeView.swift; claudia-advocate-api contracts/brief.ts. Before anything reaches the clinician, the patient sees each brief item with its own toggle. Until September 22, 2026 the clinician also saw a count of what she left out; the brief now carries only what she sends.[W-098 · shipped code, updated]
collections.ts:162-172; chat/pipeline.ts:1086; events.py; library.py; documents.py. Until September 22, 2026 the model context switch was on by default and web article saves and document uploads reached the clinician's digest on their own. Both are fixed: the switch is off until she says yes, and nothing reaches the clinician unless she sends it.[W-108 · shipped code, updated]