Category: Medicine
The Effects of Ransomware Attacks on Hospitals and Patients
As cybercriminals increasingly target health care, hospitals face the growing threat of ransomware attacks. Ransomware is a type of malicious software that prevents users from accessing electronic systems and demands a ransom to restore access. We create and link a database of hospital ransomware attacks to Medicare claims data. We quantify the effects of ransomware attacks on hospital operations and patient outcomes. Ransomware attacks decrease hospital volume by 17–24 percent during the initial attack week, with recovery occurring within 3 weeks. Among patients already admitted to the hospital when a ransomware attack begins, in-hospital mortality increases by 34–38 percent.
That is by Hannah Neprash, Claire McGlave, and Sayeh Nikpay, recently published in American Economic Journal: Economic Policy.
What happens when dating goes online?
This paper studies how online dating platforms have impacted marital outcomes, assortative matching, and sexually transmitted disease (STD) rates in the United States. We construct county-level measures of online dating usage using data from website-based platforms (2002-2013) and mobile app-based platforms (2017-2023). Leveraging county-level variation and an instrumental variable strategy, we show in the desktop era, a 1% increase in online dating sessions raises divorce rates by 0.50%, while in the mobile era, a 1% increase in online dating activity lowers marriage and divorce rates by 0.40% and 0.33%, respectively. We also document shifts in assortative matching. Desktop sites reduce sorting along education and employment dimensions, whereas mobile sites reduce sorting by employment, but increase sorting by race. Across both eras, we find no evidence that greater online dating usage increases average STD rates. Average effects are negative or statistically insignificant, but are positive for some subpopulations. We develop a search and matching model where technological changes impact search costs, market size, and market noise can explain our empirical findings.
That is from a new paper by Daniel Ershov, Jessica Fong, and Pinar Yildirim. Via the excellent Kevin Lewis.
AI Physicians At Last
In 2004 (!) I wrote:
Many people complain that medicine is too impersonal. I think it is not impersonal enough. I have nothing against my physician (a local magazine says he is one of the best in the area) but I would prefer to be diagnosed by a computer. A typical physician spends most of the day playing twenty questions. Where does it hurt? Do you have a cough? How high is the patient’s blood pressure? But an expert system can play twenty questions better than most people. An expert system can use the best knowledge in the field, it can stay current with the journals, and it never forgets.
It took longer than it should have, but we are finally here. Today, most people already use AI to help diagnose and manage medical conditions, and now:
Utah is letting artificial intelligence — not a doctor — renew certain medical prescriptions. No human involved.
It’s a pilot program for routine renewals but a welcome start. The AMA, of course, is not pleased.
In a statement, Dr. John Whyte, CEO and executive vice president at the American Medical Association, said: “While AI has limitless opportunity to transform medicine for the better, without physician input it also poses serious risks to patients and physicians alike.”
One concern is misuse or abuse, including the possibility that people struggling with addiction could try to game automated systems to obtain drugs inappropriately. Another concern is missing subtle clinical red flags or drug interactions that a doctor would catch.
It’s amazing that anyone can say these things with a straight face. As far as I know, AI has never run a pill mill, unlike human physicians. And the AI
“missing subtle clinical red flags or drug interactions that a doctor would catch.” Is this a joke?
Direct and Indirect Effects of Vaccines: Evidence from COVID-19
Sorry people, but the verdict on this one continues to come in:
We estimate direct and indirect vaccine effectiveness and assess how far the infection-reducing externality extends from the vaccinated, a key input to policy decisions. Our empirical strategy uses nearly universal microdata from a single state and relies on the six-month delay between 12- and 11-year-old COVID vaccine eligibility. Vaccination reduces cases by 80 percent, the direct effect. This protection spills over to close contacts, producing a household-level indirect effect about three-fourths as large as the direct effect. However, indirect effects do not extend to schoolmates. Our results highlight vaccine reach as important to consider when designing policy for infectious disease.
That is from American Economic Journal: Applied Economics, by Seth Freedman, Daniel W. Sacks, Kosali Simon, and Coady Wing. So many different methods and papers are pointing in the same direction…
Autism Hasn’t Increased
Autism diagnoses have increased but only because of progressively weaker standards for what counts as autism.
The autistic community is a large, growing, and heterogeneous population, and there is a need for improved methods to describe their diverse needs. Measures of adaptive functioning collected through public health surveillance may provide valuable information on functioning and support needs at a population level. We aimed to use adaptive behavior and cognitive scores abstracted from health and educational records to describe trends over time in the population prevalence of autism by adaptive level and co-occurrence of intellectual disability (ID). Using data from the Autism and Developmental Disabilities Monitoring Network, years 2000 to 2016, we estimated the prevalence of autism per 1000 8-year-old children by four levels of adaptive challenges (moderate to profound, mild, borderline, or none) and by co-occurrence of ID. The prevalence of autism with mild, borderline, or no significant adaptive challenges increased between 2000 and 2016, from 5.1 per 1000 (95% confidence interval [CI]: 4.6–5.5) to 17.6 (95% CI: 17.1–18.1) while the prevalence of autism with moderate to profound challenges decreased slightly, from 1.5 (95% CI: 1.2–1.7) to 1.2 (95% CI: 1.1–1.4). The prevalence increase was greater for autism without co-occurring ID than for autism with co-occurring ID. The increase in autism prevalence between 2000 and 2016 was confined to autism with milder phenotypes. This trend could indicate improved identification of milder forms of autism over time. It is possible that increased access to therapies that improve intellectual and adaptive functioning of children diagnosed with autism also contributed to the trends.
The data is from the US CDC.
Hat tip: Yglesias who draws the correct conclusion:
Study confirms that neither Tylenol nor vaccines is responsible for the rise in autism BECAUSE THERE IS NO RISE IN AUTISM TO EXPLAIN just a change in diagnostic standards.
Earlier Cremieux showed exactly the same thing based on data from Sweden and earlier CDC data.
Happy New Year. This is indeed good news, although oddly it will make some people angry.
The Hainan Free Trade Port
Earlier I wrote about China’s Libertarian City, Boao Hope City (officially the Boao Lecheng International Medical Tourism Pilot Zone), China’s first special economic zone for advanced healthcare. Boao Hope City is following the peer approval model I have long argued for:
Daxue: Medical institutions within the zone can import and use pharmaceuticals and medical devices already available in other countries as clinically urgent items before obtaining approval in China. This allows domestic patients to access innovative treatments without the need to travel abroad…. The medical products to be used in the pilot zone must possess a CE mark, an FDA license, or PMDA approval, which respectively indicate that they have been approved in the European Union, the US, and Japan for their safe and effective use.
Boao Hope City is part of the larger Hainan Free Trade Zone. Hainan is a large island off China’s Southern Coast, often called the Hawaii of China. The entire island is being turned into the world’s largest free trade zone. As of Dec. 18, 2025, Hainan now boasts:
- Expanded “Zero-Tariff” Coverage…“zero-tariff” eligible goods expand from about 1,900 to approximately 6,600 tariff lines, increasing coverage from 21% to 74% of total import/export items, encompassing most production equipment and raw materials. This exemption applies to import tariffs, import VAT, and consumption tax, potentially saving enterprises about 20% in tax costs on imported equipment.
- Optimized “Tariff Exemption for Value-added Processing” Policy: One of the most transformative measures, this policy sees significantly relaxed restrictions (e.g., on core business income ratios) and now allows cumulative value-added calculation across upstream and downstream enterprises. This makes it easier for businesses to meet the “over 30% value-added” threshold for tariff exemption when selling finished products into the mainland market. Companies can ship primary products or components to Hainan for substantial processing; if the value-added meets the standard, the final products can enter the mainland market tariff-free.
- “Dual 15%” Tax Incentives as a Long-term Advantage: Encouraged industries registered and substantively operating in the Hainan FTP enjoy a reduced 15% corporate income tax rate. Eligible high-end and in-demand talents benefit from an individual income tax exemption for the portion exceeding 15%, providing long-term, stable fiscal predictability.
- Enhanced Trade and Investment Liberalization/Facilitation: Measures include implementing a negative list for cross-border trade in services, relaxing foreign investment access, adopting a “commitment-based registration system” for business setup, and streamlining procedures. A visa-free policy for nationals of 59 countries is in effect, with further eased entry-exit restrictions for business personnel.
An RCT on AI and mental health
Young adults today face unprecedented mental health challenges, yet many hesitate to seek support due to barriers such as accessibility, stigma, and time constraints. Bite-sized well-being interventions offer a promising solution to preventing mental distress before it escalates to clinical levels, but have not yet been delivered through personalized, interactive, and scalable technology. We conducted the first multi-institutional, longitudinal, preregistered randomized controlled trial of a generative AI-powered mobile app (“Flourish”) designed to address this gap. Over six weeks in Fall 2024, 486 undergraduate students from three U.S. institutions were randomized to receive app access or waitlist control. Participants in the treatment condition reported significantly greater positive affect, resilience, and social well-being (i.e., increased belonging, closeness to community, and reduced loneliness) and were buffered against declines in mindfulness and flourishing. These findings suggest that, with purposeful and ethical design, generative AI can deliver proactive, population-level well-being interventions that produce measurable benefits.
That is from a new paper by Julie Y.A. Cachia, et.al. A single paper or study is hardly dispositive, even when it is an RCT. But you should beware of those, such as Jon Haidt and Jean Twenge, who are conducting an evidence-less jihad against AI for younger people.
Via the excellent Kevin Lewis.
Is involuntary hospitalization working?
From Natalia Emanuel, Valentin Bolotnyy, and Pim Welle:
The involuntary hospitalization of people experiencing a mental health crisis is a widespread practice, as common in the US as incarceration in state and federal prisons and 2.4 times as common as death from cancer. The intent of involuntary hospitalization is to prevent individuals from harming themselves or others through incapacitation, stabilization and medical treatment over a short period of time. Does involuntary hospitalization achieve its goals? We leverage quasi-random assignment of the evaluating physician and administrative data from Allegheny County, Pennsylvania to estimate the causal effects of involuntary hospitalization on harm to self (proxied by death by suicide or overdose) and harm to others (proxied by violent crime charges). For individuals whom some physicians would hospitalize but others would not, we find that hospitalization nearly doubles the probability of being charged with a violent crime and more than doubles the probability of dying by suicide or overdose in the three months after evaluation. We provide evidence of housing and earnings disruptions as potential mechanisms. Our results suggest that on the margin, the system we study is not achieving the intended effects of the policy.
Here is the abstract online at the AEA site. I am looking forward to seeing more of this work.
GDPR is worse than you had thought
We examine how data privacy regulation affects healthcare innovation and research collaboration. The European Union’s General Data Protection Regulation (GDPR) aims to enhance data security and individual privacy, but may also impose costs to data collection and sharing critical to clinical research. Focusing on the pharmaceutical sector, where timely access and the ability to share patient-level data plays an important role drug development, we use a difference-in-differences design exploiting variation in firms’ pre-GDPR reliance on EU trial sites. We find that GDPR led to a significant decline in clinical trial activity: affected firms initiated fewer trials, enrolled fewer patients, and operated at fewer trial sites. Overall collaborative clinical trials also declined, driven by a reduction in new partnerships, while collaborations with existing partners modestly increased. The decline in collaborations was driven among younger firms, with little variation by firm size. Our findings highlight a trade- off between stronger privacy protections and the efficiency of healthcare innovation, with implications for how regulation shapes the rate and composition of subsequent R&D.
That is from Jennifer Kao and Sukhun Kang, here is the online abstract for the AEA meetings.
Crime and the Welfare State
Several recent papers claim that expanding programs like Medicaid reduces crime (e.g. here). I’ve been skeptical, not because of weaknesses in any particular paper, but just because the results feel a bit too aligned with social-desirability bias and we know that the underlying research designs can be fragile. As a result, my priors haven’t moved much. The first paper using a genuine randomized controlled trial now reports no effect of Medicaid expansion on crime.
Those involved with the criminal justice system have disproportionately high rates of mental illness and substance-use disorders, prompting speculation that health insurance, by improving treatment of these conditions, could reduce crime. Using the 2008 Oregon Health Insurance Experiment, which randomly made some low-income adults eligible to apply for Medicaid, we find no statistically significant impact of Medicaid coverage on criminal charges or convictions. These null effects persist for high-risk subgroups, such as those with prior criminal cases and convictions or mental health conditions. In the full sample, our confidence intervals can rule out most quasi-experimental estimates of Medicaid’s crime-reducing impact.
Finkelstein, Miller, and Baicker (WP).
It could still be the case that very targeted interventions–say making sure that released criminals get access to mental health care–could do some good but there’s unlikely to be any general positive effect.
A similar story is found in Finland where a large RCT on a guaranteed basic income found zero effect on crime
This paper provides the first experimental evidence on the impact of providing a guaranteed basic income on criminal perpetration and victimization. We analyze a nationwide randomized controlled trial that provided 2,000 unemployed individuals in Finland with an unconditional monthly payment of 560 Euros for two years (2017-2018), while 173,222 comparable individuals remained under the existing social safety net. Using comprehensive administrative data on police reports and district court trials, we estimate precise zero effects on criminal perpetration and victimization. Point estimates are small and statistically insignificant across all crime categories. Our confidence intervals rule out reductions in perpetration of 5 percent or more for crime reports and 10 percent or more for criminal charges.
My 2011 Review of Contagion
I happened to come across my 2011 review of the Steven Soderberg movie, Contagion and was surprised at how much I was thinking about pandemics prior to COVID. In the review, I was too optimistic about the CDC but got the sequencing gains right. I continue to like the conclusion even if it is a bit too clever by half. Here’s the review (no indent):
Contagion, the Steven Soderberg film about a lethal virus that goes pandemic, succeeds well as a movie and very well as a warning. The movie is particularly good at explaining the science of contagion: how a virus can spread from hand to cup to lip, from Kowloon to Minneapolis to Calcutta, within a matter of days.
One of the few silver linings from the 9/11 and anthrax attacks is that we have invested some $50 billion in preparing for bio-terrorism. The headline project, Project Bioshield, was supposed to produce vaccines and treatments for anthrax, botulinum toxin, Ebola, and plague but that has not gone well. An unintended consequence of greater fear of bio-terrorism, however, has been a significant improvement in our ability to deal with natural attacks. In Contagion a U.S. general asks Dr. Ellis Cheever (Laurence Fishburne) of the CDC whether they could be looking at a weaponized agent. Cheever responds:
Someone doesn’t has to weaponize the bird flu. The birds are doing that.
That is exactly right. Fortunately, under the umbrella of bio-terrorism, we have invested in the public health system by building more bio-safety level 3 and 4 laboratories including the latest BSL3 at George Mason University, we have expanded the CDC and built up epidemic centers at the WHO and elsewhere and we have improved some local public health centers. Most importantly, a network of experts at the department of defense, the CDC, universities and private firms has been created. All of this has increased the speed at which we can respond to a natural or unnatural pandemic.
In 2009, as H1N1 was spreading rapidly, the Pentagon’s Defense Threat Reduction Agency asked Professor Ian Lipkin, the director of the Center for Infection and Immunity at Columbia University’s Mailman School of Public Health, to sequence the virus. Working non-stop and updating other geneticists hourly, Lipkin and his team were able to sequence the virus in 31 hours. (Professor Ian Sussman, played in the movie by Elliott Gould, is based on Lipkin.) As the movie explains, however, sequencing a virus is only the first step to developing a drug or vaccine and the latter steps are more difficult and more filled with paperwork and delay. In the case of H1N1 it took months to even get going on animal studies, in part because of the massive amount of paperwork that is required to work on animals. (Contagion also hints at the problems of bureaucracy which are notably solved in the movie by bravely ignoring the law.)
It’s common to hear today that the dangers of avian flu were exaggerated. I think that is a mistake. Keep in mind that H1N1 infected 15 to 30 percent of the U.S. population (including one of my sons). Fortunately, the death rate for H1N1 was much lower than feared. In contrast, H5N1 has killed more than half the people who have contracted it. Fortunately, the transmission rate for H5N1 was much lower than feared. In other words, we have been lucky not virtuous.
We are not wired to rationally prepare for small probability events, even when such events can be devastating on a world-wide scale. Contagion reminds us, visually and emotionally, that the most dangerous bird may be the black swan.
Innovations in Health Care
The latest issue of the journal Innovations focuses on health care and is excellent. It’s a very special issue–a double Tabarrok issue!
My paper, Operation Warp Speed: Negative and Positive Lessons for New Industrial Policy, asks what can learn from the tremendous success of OWS about an OWS for X? What are the opportunities and the dangers?
My son Maxwell Tabarrok’s paper is Peptide-DB: A Million-Peptide Database to Accelerate Science. Max’s paper combines economics and science policy. Open databases are a public good and so are underprovided. A case in point is that there is no big database for anti-microbial peptides despite the evident utility of such a database for using ML techniques to create new antibiotics. The NIH and other organizations have successfully filled this gap with databases in the past such as PubChem, the HGP, and ProteinDB. A million-peptide database is well within their reach:
The existing data infrastructure for antimicrobial peptides is tiny and scattered: a few thousand sequences with a couple of useful biological assays are scattered across dozens of data providers. No one in science today has the incentives to create this data. Pharma companies can’t make money from it and researchers can’t produce any splashy publications. This means that researchers are duplicating the expensive legwork of collating and cleaning all of this
data and are not getting optimal results, as this is simply not enough information to take full advantage of the ML approach. Scientific funding organizations, including the NIH and the NSF, can fix this problem. The scientific knowledge required to massively scale the data we have on antimicrobial peptides is well established and ready to go. It wouldn’t be too expensive or take too long to get a clean dataset of a million peptides or more, and to have detailed information on their activity against the most important resistant pathogens as well as its toxicity to human cells. This is well within the scale of the successful projects these organizations have funded in the past, including PubChem, the HGP, and ProteinDB.
Naturally, I am biased towards Tabarrok-articles but another important paper is Reorganizing the CDC for Effective Public Health Emergency Response by Gowda, Ranasinghe, and Phan. As Michael Lewis wrote in The Premonition by the time of COVID the CDC had became more akin to an academic department than a virus fighting agency:
The CDC did many things. It published learned papers on health crises, after the fact. It managed, very carefully, public perception of itself. But when the shooting started, it leapt into the nearest hole, while others took fire.
Gowda, Ranasinghe, and Phan agree.
The COVID-19 pandemic revealed significant weaknesses in the CDC’s response system. Its traditional strengths in testing, pathogen dentification, and disease investigation and tracking faltered. The legacy of Alexander Langmuir, a pioneering epidemiologist who infused the CDC with epidemiological principles in the 1950s, now seems a distant memory. Tasks as basic as collecting and providing timely COVID-19 data, along with data analysis and epidemiological modeling—both of which should have been the core capability of the CDC—became alarmingly difficult and had to be handled by nongovernmental organizations, such as the Johns Hopkins University Coronavirus Resource Center.
A closer examination of the CDC’s workforce composition reveals the root cause: a mere fraction of its employees are epidemiologists and data scientists. The agency has seen an increasing emphasis on academic exploration at the expense of on the-ground action and support for frontline health departments. (Armstrong & Griffin, 2022).
The authors propose to reinvigorate the CDC by integrating it with the more practical and active U.S. Public Health Service. This is a very good suggestion.
For one more check out Bai, Hyman and Silver as a primer on Improving Health Care. The entire issue is excellent.
Pharma supply is elastic
The crux of the problem is that the IRA imposes price caps that shorten the effective life of a patent and applies those price controls even to later-approved uses. Thirteen years after FDA approval, biologics, which are typically infused or injected, become subject to price controls. For small-molecule drugs, typically pills or tablets, the window is only nine years. The clock starts at a drug’s first approval, leaving a follow-on or alternative use, approved years later, an insufficient period to make up the cost of research.
Two weeks ago, a study I conducted with colleagues at the University of Chicago appeared in Health Affairs. It reveals how much these provisions harm cancer research. In reviewing every Food and Drug Administration-approved cancer drug between 2000 and 2024, we found a large part of innovation in cancer treatment takes place after a therapy is first approved. About 42% of the 184 cancer therapies that were initially approved during that period had follow-on approvals—involving new uses or “indications” for an existing drug—such as treating additional cancer types or being used earlier in the disease, when treatment outcomes tend to be better.
This cumulative progress through follow-on discoveries is a big driver of new cancer treatments, the largest drug class making up about 35% of the overall FDA pipeline. Cancer drugs are generally first tested in patients with late-stage disease, after which the drug is studied for use in earlier stages of that cancer and for new uses, including treating other cancers. Our study found that 60% of follow-on drugs treated earlier stages than the initial drugs. This is important because treating earlier stages is often more successful than when a cancer has spread more.
But that cumulative progress depends on incentives for sustained research well after the first FDA approval—often years of additional trials and investments. And those incentives were killed by the IRA.
Make Africa Healthy Again
In the late 1990s, South Africa’s President Thabo Mbeki decided that mainstream science had AIDS wrong. A small circle of “truth-tellers” convinced him that AIDS came from poverty and malnutrition, not a virus. He warned that anti-retroviral therapy (ART) was toxic and that pharmaceutical companies were poisoning Africans for profit.
His government stalled the rollout of ART. Health Minister Manto Tshabalala-Msimang pushed garlic, beetroot, and lemon as medicine. “Nutrition is the basis for good health,” she said, insisting that exercise and diet, not Western drugs, were the real treatment. She warned that antiretrovirals had side effects, including cancer, that the establishment was hiding. When scientists showed data, she waved it off: “No churning of figures after figures will deter me from telling the truth to the people of the country.”
The result was a public health disaster: hundreds of thousand of preventable deaths (see also here and here).
A reminder of what happens when authority trades evidence for ideology.
Big, Fat, Rich Insurance Companies
In my post, Horseshoe Theory: Trump and the Progressive Left, I said:
Trump’s political coalition isn’t policy-driven. It’s built on anger, grievance, and zero-sum thinking. With minor tweaks, there is no reason why such a coalition could not become even more leftist. Consider the grotesque canonization of Luigi Mangione, the (alleged) murderer of UnitedHealthcare CEO Brian Thompson. We already have a proposed CA ballot initiative named the Luigi Mangione Access to Health Care Act, a Luigi Mangione musical and comparisons of Mangione to Jesus. The anger is very Trumpian.
In that light, consider one of Trump’s recent postings:
THE ONLY HEALTHCARE I WILL SUPPORT OR APPROVE IS SENDING THE MONEY DIRECTLY BACK TO THE PEOPLE, WITH NOTHING GOING TO THE BIG, FAT, RICH INSURANCE COMPANIES, WHO HAVE MADE $TRILLIONS, AND RIPPED OFF AMERICA LONG ENOUGH.