This article first appeared on HealthAffairs.org.
While high drug prices and price hikes are getting lots of attention from the media, the biggest reason health care in the United States costs so much more than in other wealthy countries is that it takes more, and more highly paid, people to deliver care. And it takes more people every month. For the past decade, health care has added new jobs every month—not even taking a break during the Great Recession; from 2001 to 2016, health care jobs represented 29 percent of the 14.4 million net new jobs. While job growth may sound good, since these jobs offer attractive wages and generous benefits, the productivity of this workforce is lagging other US services industries. A new report by Nikhil Sahni and colleagues from McKinsey—“The Productivity Imperative for Healthcare Delivery in the United States”—set out to answer this paradox: How do we not just continue to increase expenses in our health care delivery system but improve productivity to bend the health care spending growth trajectory? The report analyzes value-added gross domestic product data from the Bureau of Economic Analysis and health care labor data tracked by the Bureau of Labor Statistics data in new ways.
In this post, I highlight surprising findings from that report and offer policy prescriptions for improving labor productivity. Specifically, health care jobs are being added faster than expected based upon growth in clinical demand, and most of the new health care jobs are in non-valued-added job categories. Fortunately, there are many things that can be done to improve labor productivity by improving clinical operations and reducing administrative complexity.
The Puzzle Of Health Care Hiring
Health care hiring is puzzling. Innovations in health care services and products have consistently offset increased demand for services resulting from aging. Advances have enabled the proliferation of alternative sites of care, moving more care to outpatient settings and homes every year. Additionally, changes in payment models (for example, episode-based payments, capitation) have created incentives for physicians to improve care coordination, and this has led to a large reduction in patients being readmitted to hospitals. These are just a few trends that have resulted in the reduced demand for hospital beds, and we would expect hospital hiring to slow as a result. And if the laws of economics were working, we would also expect hospitals to be unable to raise prices to pay for all of the new the jobs. Remarkably, hospital prices have grown by 250 percent over the past decade (1998–2018).
Yet, hospital bed occupancy has been remarkably flat over the past 15 years—going from 61 percent to 63 percent. Given this, it is hard to understand what all these people are doing. The US has made little progress on most clinical quality measures. Sadly, life expectancy is falling. And despite enormous investments in electronic health records, medical errors appear to be the third leading cause of death in the US. If the US health care system were adding people to dramatically improve quality, it would be a worthwhile investment, but that does not appear to be the case.
When we go a level deeper and look at the actual job titles, the story gets worse. In health care, amazingly, more than three-quarters of new jobs from 2001 to 2016 have been support roles, both clinical (half of new jobs) and nonclinical (one-quarter of new jobs). The ratio of support staff to clinicians is actually getting worse. It is now 3.6 for each clinician—exactly the opposite of what one would expect since health care providers typically have low margins, which should create disincentives to add administrative overhead. Other service industries in the US have lower ratios of support staff to value-creating people (for example, lawyers and judges in legal services). Outside of health care, the US economy is great at eliminating non-value-added jobs, so good that policy makers are debating ideas such as “universal basic income” to mitigate a future where robots and software replace many more humans. It would be easier to defend the rising premiums driven by health care job growth if the added jobs were patient-facing clinical roles, such as mental health clinicians, primary care physicians, clinicians of all types in rural communities, and clinicians serving Medicaid patients. But shortages in all of these categories persist.
Priorities For Improving Labor Productivity
Fortunately, as Sahni and colleagues note, there are many opportunities to improve labor productivity in health care delivery, especially in clinical labor, which accounts for nearly 60 percent of the workforce and is the most expensive labor (for example4, a physician salary is nearly twice that of a non-physician clinician). How do we do this? First, there is a tremendous opportunity in physician schedules: On average a physician’s schedule density is only 80 percent, with top provider systems reaching 90–95 percent. Some analysts believe that solving this problem could address almost all of the projected physician shortage. However, doing this in a vacuum, without reducing administrative tasks for clinicians, would further exasperate physician burnout, a growing concern.
The second prong of labor productivity improvement, therefore, should be focusing on reducing time-wasting, burnout-inducing administrative tasks done by clinicians. The McKinsey report notes that, shockingly, 36 percent of the work nurses do could be done by someone else with less specialized training; and internists reported spending 48 minutes per day inputting data.
Finally, the highest priority should be focusing on reducing non-value-added administrative labor. Our health care system has evolved into a byzantine set of administrative rules and processes that add time and expense with very unclear effects on outcomes. Today, hospitals employ office buildings full of people to generate the codes, bills, and clinical documentation needed to get paid. Anecdotally, a leader of one large health system shared with me that the organization employs one revenue cycle employee for every two beds. Health plans employ call centers to try to verify provider directories with little success. Massive risk-adjustment coding and data collection industries flourish to help health plans and providers get paid more from the government. All of these areas are ripe for disruption by computers, which never get tired. Even better would be rethinking the process for how payments are made and patient risk accounted for, and standardizing the rules, forms, and process steps across all health plans. Sahni and colleagues found that using known technologies, billing and insurance-related spending could be reduced by about 17 percent; another 10 percent could be captured by creating data clearinghouses, similar to those we have used in banking for decades.
Unleashing health care labor productivity is the most powerful thing that we can do to make health care more affordable. Fortunately, the US economy is the best in the world at labor productivity in other sectors, and all of actions proposed here can be done with technologies that exist today and are proven effective in other sectors. If health care labor productivity does not improve, the ramifications will be expensive and force painful budget tradeoffs. Since most health care spending is supported through taxes, without figuring out how to generate productivity, either taxes go up a lot, wages fall a lot, or both. Let’s hope that one of the benefits of big technology companies such as Amazon, Google, and Apple entering health care is that they infect the ecosystem with their 10-times better labor productivity.