Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
Brady Harold never knew what a miracle he was.
After his car accident, he was rushed to his nearest trauma center. Unconscious, the trauma team inserted a breathing tube and resuscitated him. A CT scan of the brain revealed an epidural hematoma, life-threatening bleeding on the brain. Dr. Oliver, the local neurosurgeon, was called in and deftly removed the blood clot, preventing a catastrophic brain injury.
Brady underwent his recovery. Dr. Oliver rounded on him daily, carefully documenting his progress in his clinical notes.
“Dr. Oliver, we have to talk about your notes,” the administration would say to him. He would be called into a meeting where they highlighted how his documentation is sub-par. There were six members of the coding team, all present at the meeting. Two were MDs and four were RNs.
What was his offense? Not making Brady Harold look sicker in his notes.
“When you state his CT scan showed ‘swelling,’ we can’t risk-adjust for that. You need to use these terms approved by the coders,” the coding team would lecture him. “We need you to make the patient look as sick as possible.”
They would pester and hound Dr. Oliver, pointing out how his documentation didn’t make Brady look sick enough. They would send him angry emails, page him during surgery and call his cell phone after hours. Dr. Oliver would eventually relent and try his best to alter the documentation. It took him away from other patient care duties, but that was beside the point. After the administrators were done, they could paint the picture of Brady Harold, chronically ill patient who already had an unusually high risk of death. He never knew what a miracle he was.
So how did we get to this place? I, for one, blame baseball. Don’t get me wrong, I love the sport. It conjures up associations with warm summer days, hot dogs, and statistics. In fact, it’s the perfect statistical game. It’s zero-sum; for every offensive accomplishment, there is a corresponding defensive blight. Many kids’ understandings of statistics, probabilities, and averages comes from baseball. As the science of statistics has advanced, analysts have fine-tuned metrics that closely approximate the “true skill” of players. Batting average gave way to OPS which gave way to exit velocity and launch angle: better and better numbers to reflect the players’ skills.
If we care so much about a silly game to devote years of brainpower to its statistical analysis, surely, we could do the same with medicine. Physicians and hospitals should have their statistics published. A patient should know which doctor has the best “batting average,” right? This is where the analogy breaks down. Patient care is not a zero-sum game. There is no official scorer in medicine. There is not even consensus on the ideal outcomes.
Despite this, the government thought that incentivizing outcomes would improve patient care. The reasoning is easy to see. Instead of incentivizing more care, the government thought it would make more sense to incentivize better care. This is only necessary because there isn’t a functional free market in medicine. As Hayek eloquently stated, “Once the free working of the market is impeded beyond a certain degree, the [government] planner will be forced to extend his controls until they become all-comprehensive.”
Of course, that meant statistics must be derived. If the patient can’t determine value themselves, as is the case with a functioning market, the government must determine value. Therein lies the problem. It’s not as simple as baseball, where an objective scorekeeper can decide what is a hit or what is an error. Medicine is an infinitely complex system without defined “good” or “bad” outcomes. Instead of hits, outs, errors, and runs, what are the scorekeepers of medicine to measure?
“…as if only that which can be counted really counts.” – Jerry Z. Muller stated in The Tyranny of Metrics. Tracking statistics for physicians is not as simple as it seems. What statistics should be tracked? Comparing the survival rate between a trauma surgeon, oncologist, and pediatrician doesn’t seem appropriate. In fact, looking at the mortality rate of a pediatrician would yield very little information about the quality of said pediatrician. It’s very difficult to find reliable, objective measures of physician quality. Creating artificial metrics can have disastrous outcomes in any industry. Every centrally planned economy in history has faced this problem.
Of course, what really matters is the patient. The metrics should align with what the patients value. However, patient satisfaction scores, as measured by a number of quantifiable surveys, is highly subjective. In fact, it’s influenced by wait times, hospital décor, and cafeteria quality more than the ability of the physician. One study even showed that the patients with the highest satisfaction had the worst outcomes (along with costing the most). In some cases, notably drug-seeking patients or those wanting to self-harm, satisfying the patient’s wishes would be counter-productive to health. How does one judge value at the end of life? Some patients want to live as long as possible, while others just want to die a dignified death at home.
Early metric tracking in medicine seemed to be filled with promise. Just tracking the number of infected central lines (IVs inserted into the big veins near the heart) led to an improvement in practices and drops in the number of infections. The same principle was applied to urinary tract infections (UTI) after urinary catheters. Then hospitals realized they could game the system. A hospital-acquired UTI would count against the statistics, but not if the patient had one on arrival. All patients were suddenly tested for UTI when they enter the hospital, leading to a massive increase in testing costs in order to document UTI on arrival. Then, when a physician wanted to test for a UTI during the hospitalization, that test would be blocked by administration. You can’t find a UTI if you don’t look.
Gaming metrics reached its peak with the observed to expected complication ratio. Some hospitals have sicker patients at baseline than other hospitals. It wouldn’t be fair to penalize those hospitals with sicker patients. Thus, the metrics all must be risk-adjusted. Based on a risk-adjustment formula, hospitals would have an expected complication rate that would be compared to their observed complication rate. There’s a much bigger return on investment in making the expected complication rate as bad as possible rather than actually improving care. Just by having the coders round with the physicians, revenue on a single service was increased by 40%. This was without improving care in any way.
A whole industry has grown around this metric fixation. The US government has spent over $1.3 billion on developing quality metrics from 2008-2018. This money has gone to several private firms to devise these metrics and risk adjustment formulas. Five organizations alone were awarded nearly $900 million. On top of that, given the complexity of these metrics, consulting firms have sprung up. These firms will assist hospitals in coding and tracking metrics, improving the expected to observed ratio. This also partly explains the continued rise in administrative costs within US healthcare. More metrics require more administrators. People who say our system of multiple private insurers is what’s driving administrative growth have never dealt with Medicare.
On top of the expense in creating metrics and hospital tracking of metrics, it is handcuffing independent physician practices. Annually, the cost to physician practices in metric tracking exceeds $15 billion. Physicians spend over 12 hours every week simply entering metric data into the electronic medical record.
These expenses are necessary from the hospital standpoint, as they can make or break the bottom line. In 2019, CMS adjusted $1.9 billion in Medicare part A payments. This program is revenue-neutral, so that $1.9 billion was simply shifted from the “worst” hospitals to the “best.” Losing out on these payments could mean closing hospitals. In some communities, it leaves populations with only one choice for healthcare (or employment if you’re an HC worker). Even worse, it leaves some communities without any healthcare.
Are these value-based payments worth it? If healthcare quality improves, one could argue it is worth the cost. The data is robust: it does not help. The hospitals treating the most vulnerable patients are hurt the most. It worsens disparities. This makes sense, as Medicaid patients tend to be sicker, cost more to treat, and reimburse less. This leaves these hospitals with less money to spend on consultants to help game the numbers. It can also detract from actual attempts to improve care. As coders get better and better in making patients look as sick as possible, stagnant care will actually appear to be improving. This “improvement” in care is just a byproduct of improvement in risk adjustment coding. It has even been shown that hospitals will engage in behaviors that increase mortality in order to meet statistical benchmarks.
Shared medical decision making is the core of the patient-physician relationship. The patient and physician should arise at a treatment plan after careful discussion. Each patient will have different goals and willingness to accept treatment recommendations. This is the core of healthcare. Fostering this relationship should be the goal of government intervention. The metric-industrial complex does the opposite. It inserts metric fixation into the patient-physician relationship. Physicians are forced to care about their stats, either consciously or by aggressive administrators.
Medicine is not baseball.Published in