Mental Health Awareness Month. 27 prior authorizations, 60+ hours of invisible labor, and the healthcare AI systems scaling this inequity without examining it.
Access is Power | Mental Health + Healthcare AI | 16 min read
This is the sixth installment of the Access is Power series. Mental Health Awareness Month. Thread series this week: Lyfgenia: A Public Record — Lyfgenia: A Public Record Ep 3 | Counting the Invisible Hours. This week’s thesis: there is a kind of labor that does not show up on any org chart, any performance review, or any AI training dataset. The administrative burden of complex treatment in the US healthcare system is itself a health equity crisis — invisible in clinical data, compounding the harm of the illness itself, and being automated by AI systems that have never been asked to measure it.
There is a kind of labor that does not show up on any org chart, any performance review, or any AI training dataset.
I have been quantifying mine.
LYFGENIA: A PUBLIC RECORD — EPISODE 3: COUNTING THE INVISIBLE HOURS
Since beginning Lyfgenia gene therapy for sickle cell disease, here is what I have tracked:
27 prior authorization requests filed, some multiple times after initial denial.
Over 60 hours spent on insurance-related calls and paperwork.
6 specialist departments coordinated, none of which share records natively.
11 separate forms on which I have translated my own medical history into bureaucratic language.
All of this conducted while managing the physical effects of a treatment that temporarily depletes your immune system.
None of this is in the clinical literature when you look up Lyfgenia. None of it is tracked in patient outcome data. None of it is what doctors describe when they explain the treatment process.
But it IS the treatment process — for patients without dedicated administrative support systems. For patients without the expertise to navigate insurance bureaucracy. For patients without the time and resources to absorb this invisible labor on top of an illness that is already extracting everything from them.
Let me break these numbers down so they are not abstract.
The 27 prior authorizations. The prior authorization process requires a treating physician to document, in language specifically designed to satisfy insurance reviewers rather than describe clinical reality, why a treatment that a specialist has already determined is medically necessary is medically necessary. My first authorization was denied. The second required specific language changes. The third covered a different component of the treatment and started the process again. Each denial requires a patient advocate, a physician, and an administrative team to coordinate a response within a specific window. If you do not have a patient advocate — a role that most patients do not have access to and most hospitals do not provide — you navigate this yourself. On the phone. While sick.
The 60-plus hours. I tracked these because I wanted to know. They include time on hold, time in phone queues that disconnect after extended waits and require calling again, time on calls where the representative did not have access to the information needed to answer my question and transferred me to someone who also did not have it.
The 6 departments. Hematology, oncology, cardiology, stem cell collection, infusion, and primary care. These departments use systems that are technically interoperable but practically require manual coordination by the patient. My job, at multiple points in this process, has been to relay information from one specialist department to another because the shared record system does not surface the specific data each team needs. I have repeated my full medical history 11 times in the last two months.
I want to be precise about what this represents in terms of equity.
I have significant advantages. I am educated. I speak the bureaucratic language of healthcare and insurance fluently. I have professional expertise in navigating complex systems. I have support. I have flexibility in my schedule to spend hours on hold.
The patients who most commonly require Lyfgenia — Black Americans with sickle cell disease, disproportionately from communities with limited financial and administrative resources — frequently have none of these advantages. The invisible labor I am documenting at a relatively high resource level is the same labor that, for patients without my advantages, becomes an insurmountable barrier. Not a medical barrier. An administrative one. Built into the system that is supposed to help them.
THE MENTAL HEALTH DIMENSION
Mental Health Awareness Month asks us to name the hidden costs.
The administrative burden of complex treatment in the US healthcare system is itself a health equity issue. It falls disproportionately on patients with the fewest resources to carry it. It is invisible in clinical data. It compounds the harm of the illness itself.
And it is the kind of friction that AI-mediated healthcare systems are automating and scaling — without examining whether they are automating equity or automating exclusion.
I am counting it. Publicly. In real time.
Because if we do not count it, it stays invisible. And invisible labor gets assigned to the people least able to refuse it.
THE DATA BEHIND THE BURDEN
The data confirms this is not anecdotal.
A 2026 KFF Health Tracking Poll found that 39 percent of insured adults with chronic conditions identify prior authorization as the single biggest burden in accessing care — more than twice the share who cite any other obstacle. Seven in ten insured adults describe the process as burdensome. Nearly half (47 percent) of all insured adults have had a service denied or delayed in the past two years, rising to 57 percent among those with chronic conditions. [Source: KFF Health Tracking Poll, February 2026]
The PAN Foundation’s 2026 Center for Patient Research survey found that among patients requiring prior authorization, 27 percent waited over a week for an insurance decision, and 32 percent waited over a week to receive prescribed care. One in three patients (34 percent) were initially told their request was approved — only to be denied later. [Source: PAN Foundation Center for Patient Research, March 2026]
The American Medical Association surveys consistently show that physicians submit an average of nearly 40 prior authorization requests per week, with 86 percent of physicians reporting that prior authorization burdens have increased over the past five years. An alarming 92 percent of physicians report that prior authorization can have a negative impact on patient clinical outcomes. [Source: AMA Prior Authorization Physician Survey, 2024]
For sickle cell patients specifically, the economic burden is staggering. A 2025 analysis found that commercially insured adults with SCD have healthcare costs averaging $31,445 annually compared to $2,844 for matched controls — an 11-fold difference driven primarily by hospitalizations and outpatient visits. [Source: Economic Burden of Sickle Cell Disease in the United States, PMC, April 2025]
The prior authorizations I filed are not just paperwork. They are a tax on survival. And AI systems are being built to automate this process without examining whether the historical patterns they learn from — patterns shaped by decades of underfunding and systemic neglect — should be replicated or corrected.
WHAT AI MISSES ABOUT INVISIBLE LABOR
AI systems are being deployed at scale across healthcare administration: coverage decisions, prior authorization review, care management flagging, patient risk stratification.
The organizations deploying these systems typically measure their AI tools by clinical outcome metrics — accuracy of diagnosis support, reduction in clinical error rates, improvement in treatment adherence.
They do not measure administrative burden generation.
When an AI prior authorization system denies a claim incorrectly — which happens at measurable rates — the burden of the correction falls on the patient and the physician. The AI system does not track the hours the patient spends on the appeal. It does not measure the cognitive load of managing a denial while managing the disease the treatment is supposed to address. It does not account for the patients who give up — who do not appeal, who do not call back, who accept the denial because they do not have the resources to fight it.
Those patients disappear from the data. The AI system sees a denied claim that was not appealed and learns that denying similar claims produces no negative signal. The system optimizes for a metric that excludes the harm it generates.
This is not a hypothetical. This is how AI-mediated prior authorization works right now, in hospitals and insurance systems across the country. The invisible labor is invisible to the system that creates it.
REST AS RESISTANCE
Rest as Resistance is the philosophy I have built my retreats around. This month, I understand it differently than I ever have.
The retreats I run in Japan are built on the principle that rest is not earned through productivity — it is a prerequisite for sustainable work. Small groups. Women executives. A garden in Kyoto. The deliberate construction of space that is not performing productivity for anyone else’s benefit.
This month, rest is not aspirational for me. It is medical. And the system makes it extraordinarily difficult to access.
Survival requires continuous labor. Rest is the thing I am fighting for the right to have. The administrative burden of this treatment is not just time. It is the theft of rest from a person who medically requires it. The system that is supposed to heal me is also the system that prevents me from resting while it heals me.
That is the mental load no model measures.
THE AI GOVERNANCE CONNECTION
AI is mediating more of this system every day. Coverage decisions, prior authorization approvals, care management flagging. When those AI tools are built without examining the administrative burden they create or perpetuate, they do not make the system more equitable. They make the existing inequity faster.
Efficiently producing inequitable outcomes is not an improvement.
AI trained on historical denial patterns does not correct historical bias in coverage decisions. It codifies it. The patterns these systems learn from — patterns shaped by decades of underfunding and systemic neglect — are being replicated at scale without anyone asking whether they should be replicated or corrected.
The governance document this series is building is not abstract policy analysis. It is a record of what happens when the system’s invisible labor falls on one patient who decided to make it visible.
DISCUSSION QUESTIONS:
Does your organization measure the administrative burden its AI systems generate for patients, customers, or users — not just the efficiency gains for the organization? If not, who is absorbing the cost of that unmeasured labor?
When an AI prior authorization system denies a claim that is later overturned on appeal, does your organization track the patient hours spent on that appeal? If not, your AI is optimizing against a metric that excludes the harm it generates.
The invisible labor of healthcare navigation falls disproportionately on the patients with the fewest resources. What would it mean for your organization to treat administrative burden as a measurable health equity outcome — and to hold your AI systems accountable for it?
NEXT WEEK
This is Week 6 of the Access is Power series — the April–May 2026 arc. Read the full series at dr-dede.com/blog.
Next week: What Japan Taught Me About Designing for Everyone — the most personal essay of the series.
Dr. Dédé Tetsubayashi is a Black, queer, first-generation Togolese immigrant and transracial adoptee living with sickle cell disease. She is a TEDx speaker and global advisor on AI governance, disability innovation, and inclusive technology strategy. She has been the person coded out. She is now the person organizations call to fix it. She would prefer to be called before the crisis. → Work with D. Dédé
📅 Ready to move from awareness to accountability? → Book a strategy session
📧 Stay connected for ongoing thought leadership: → Join the newsletter
🎬 Watch the TEDx talk that started it all: → Watch on YouTube
🌿 Rest as Resistance retreats: → Learn more
About Dr. Dédé Tetsubayashi
Dr. Dédé is a global advisor on AI governance, disability innovation, and inclusive technology strategy. She helps organizations navigate the intersection of AI regulation, accessibility, and responsible innovation.
Work With Dr. Dédé