The Ethics of Upcoding: Does Illegality Matter?
In the United States, upcoding is punishable under the federal False Claims Act.
The World Health Organization (WHO) is insistent: any newborn who weighs less than 2500 g gets stamped with the “low birth weight” label.
That's not quite a scarlet letter, but it does warn that there may be some trouble around the bend. Babies with low birth weight can have a rough go of it. In some developing countries, for instance, low birth weight is responsible for more than half of infant mortality.1 Stateside, babies with low birth weight are more prone to neonatal infections, reduced brain volume, and growth inhibition.2 Even years later, when childhood is squarely in the rearview, the scars still show: babies with very low birth weight (under 1500 g) are more than 10% less likely to graduate from high school and 40% less likely go to college.3 It turns out that that delivery room scale doubles as a crystal ball.
Because birth weight can give us a peek into the future, physicians use it to help figure out which newborns may benefit from a little extra attention. To keep things simple — and to avoid the hassle of wrestling with continuous variables — we typically use a taxonomy that the WHO drew up: low birth weight is less than 2500 g; very low birth weight is under 1500 g; and extremely low birth weight is anything below 1000 g, with an occasional subdivision at 2000 g by a third-party reimbursement scheme. The system makes sense, both for practitioners and the accountants perched dutifully over our shoulders, as long as each successively lighter category of infant is at a greater risk for poor outcomes.
The data back this up, at least at first glance. In 2017, a German study looked at babies who were just a little bit lighter — 25 g or less — than a threshold weight and concluded that they tended to have longer hospital stays and receive more procedures than those who were just above the threshold.4 Smaller infants also happened to have better mortality rates. All of that comfortably squares with intuition. We presume that babies with lower birth weight are a bit more fragile than their heavier counterparts, so hospitals might choose to direct more resources to those infants and exercise more cautious around sending them home. However, because the difference in weight — and, it follows, baseline health — between the 2 groups was tiny, it also made sense that the extra care that the lighter kids received would propel them to comparatively stronger outcomes. So far, so good.
When researchers ran the numbers again, however, this time controlling for variables that they guessed might also be associated with neonatal health markers, like gestational age and multiplicity, the differences around the thresholds almost entirely vanished. Babies on either side of the dividing lines all of a sudden had statistically indistinguishable inpatient durations and mortality rates. It seemed that once the study accounted for other data points indicating the newborn might be unhealthy, the thresholds no longer mattered.
This isn't actually all that much of a mystery. The study's authors figured that the hospital staff was, every now and again, underreporting newborn birth weights. Recording a weight as below rather than above a threshold is basically the perfect crime. The hospital can net upward of $20,000 USD in extra reimbursement,5 and because newborns lose about 10% of their birth weight in their first few days, the chances of getting caught are, to borrow a term, extremely low. This bit of anthropometric chicanery eliminates the threshold differences because instead of being relatively similar, the newborns immediately above or below the divider are now very different. The staff's intervention dictates that the ones below the line are assumed to be sicker and because of that are expected to hang around the hospital longer and eat up more resources. Once those other variables are taken into account, it becomes clear that it's the baby's baseline health — and not the threshold — that causes the difference in outcomes. More importantly, by underreporting birth weight, hospital staff can funnel high-risk kids into greater acuity of care. The misinformation pushes the potential problem child into a new diagnosis group, which can in turn justify more aggressive intervention and attentive monitoring. This is the same added care, of course, that allows babies with (reported) weights just below the threshold to do better than kids just above it. What a difference a gram makes.
This pattern of selective fibbing, otherwise known as upcoding, is everywhere in medicine. It's universally illegal (in the United States it is punishable under the federal False Claims Act), but that doesn't stop physicians and nurses the world over from labeling simple infections as septicemia or pretending that routine office visits involve comprehensive examinations. At its worst, upcoding is self-serving fraud that greases the illegitimate transfer of wealth from credulous taxpayers to avaricious physicians. Upcoding the condition of a vulnerable infant into a better-funded diagnosis group lives on the opposite end of that spectrum. It elegantly shunts scarce medical resources toward infants in need, and those infants achieve better outcomes as a result. It does exactly what diagnostic thresholds aspire to — focus care to where it's needed most — but better.
Conditioning fiscal reimbursement and clinical management guidelines on birth weight guarantees that delivery room staff will be presented with a sort of sliding doors moment every time a newborn comes in just above a threshold. One path easily accommodates whatever treatment might be needed; the other, less so. Even though the decision ultimately boils down to whether to tell a little white lie, this isn't truly an ethical dilemma — at least, not in the classic medical sense. Upcoding the weight of a borderline neonate is perfectly consistent with the ideas of beneficence and nonmaleficence, and it's hard to imagine a parent asserting autonomy by insisting that the child's weight be downcoded. The only obvious ethical quandary is whether birthweight upcoding unjustly diverts resources from other babies, but even then, it's worth noting that newborns with low birth weight less than 25 g above the various thresholds account for only about 0.3% of total births in a given year. Not exactly overwhelming numbers.
And, anyway, birthweight upcoding appears to be well targeted. The German study concluded that nearly all of the interthreshold difference in mortality rates could be explained by selective upcoding in favor of more fragile infants. The pursuit of just outcomes elevates the spirit of the law over its letter. Reimbursement thresholds exist precisely to encourage the just allocation of resources — to make sure that every patient can get whatever treatment is required. Selective upcoding of birthweight is a step toward that reality. Ultimately, upcoding is a way for neonatologists, experts who encounter borderline babies every day, to evaluate a set of data considerably more comprehensive than just birthweight, and then make a confident prediction as to the baby's prospects. It turns out that that's a lot more valuable than some dusty old crystal ball.
- Stevens-Simon C, Orleans M. Low-birthweight prevention programs: the enigma of failure. Birth. 1999;26(3):184-191.
- Kowlessar NM, Jiang HJ, Steiner C. Hospital stays for newborns, 2011: statistical brief #163. In: Healthcare Cost and Utilization Project (HCUP) Statistical Briefs. Rockville, MD: Agency for Healthcare Research and Quality; 2006.
- Hack M, Flannery DJ, Schluchter M, Cartar L, Borawski E, Klein N. Outcomes in young adulthood for very-low-birth-weight infants. N Engl J Med. 2002;346:149-157.
- Reif S, Wichert S, Wuppermann A. Is it good to be too light? Birth weight thresholds in hospital reimbursement systems. J Health Econ. 2018;59:1-25.
- Jürges H, Köberlein J. What explains DRG upcoding in neonatology? The roles of financial incentives and infant health. J Health Econ. 2015;43:13-26.