Reducing Hospitalization
Hospitalization is expensive, often unpleasant, and is associated with iatrogenic harm, yet we have evolved a healthcare system that is hospital-centric. One notion embodied in the phrase “value-based purchasing” is that many hospitalizations are preventable through wide use of guideline directed care. This notion is based on results from trials in small populations. Vasquez and associates conducted a pragmatic, cluster- randomized controlled trial of the effect of electronic health record prompts and local care coordinators on hospitalization rates in 11,182 patients who had diabetes, hypertension, and chronic kidney disease. 1 These patients represented the majority of my practice, and the risk factors for coronary artery and progression of kidney disease in the study patients are representative. The headline results? “…use of an EHR-based algorithm and practice facilitators embedded in primary care clinics did not translate into reduce hospitalizations at 1 year.” Strengths of the study include the large number of patients and the high rate of hospitalizations—1139/5508 (20.7%) in the intervention group and 1160/5492 (21.1%) in the control group. 30-day readmission rates were 24.3% and 22.6% respectively; dialysis was administered in 0.7% and 0.6% respectively, and there were 129 deaths in the intervention group (2.3%) and 148 (2.7%) in the control group. A concern is the difference in rates of utilizing recommended medications was not very large between the intervention and control groups, partly because of already high uptake in the control group. Does this mean recommended therapies don’t work? There are numerous studies showing a relative risk reduction of a composite end-point of 30-50% with the recommended medications. But the absolute risk of death in this high-risk group was 2.5%, which is also typical for population studies. Even if we include kidney death and round up, the absolute risk is only 4%, so at best we would expect an absolute risk reduction of 2%. Said another way, you would have to treat 50 patients to prevent one death, and realistically the NNT is likely closer to 100. And the treatments are also costly and have side-effects. An issue not often addressed is that delaying one outcome, say need for dialysis, does not cure the patient, it buys time for something else to kill them. Now avoiding the need for dialysis is a highly valued outcome for the patient and may reduce the size of the budget for dialysis, but will increase costs elsewhere. I have found many times when trying to improve the quality of delivered dialysis, if we pushed too hard, something else would become a problem as a result. My mental model for this problem is squeezing on a balloon. If you slowly apply even pressure on all sides simultaneously, you can avoid popping the balloon, but it is really hard to do. Would the results be different if all the recommendations were adhered to? In this study the goals requiring patient action ranged from about 56% (glycated hemoglobin, unchanged by intervention) to 69% (blood pressure, improved from 49%.) This, too, is consistent with both my experience and the literature. The natural number for success seems to be about 50%, which, with consistent effort can be increased to about 70%. Getting to 80% is uncommon and requires massive, expensive, effort. (Think COVID 1 Vazques MA, Oliver G, Amarasingham R, et al. Pragmatic Trial of Hospitalization in Chronic Kidney Disease. NEJM 2024;390:1192-1206. doi:10.1056/NEJMoa2311708. vaccination rates when they were given to people for “free.”) I don’t think there is any reason to think better “compliance” is a reasonable goal. Another issue, present in the study and in this discussion is time frame of interest. For the patient, the time frame is “the rest of my life.” For the clinician and researcher, the time frame might be as long as five years. For the insurance company, the time frame is rarely longer than the current fiscal year. Given that patient “churn” can be upwards of 30% annually, this short horizon makes some sense, but it raises questions about population health. Which population and for how long? Who is going to pay for the investments needed to really change the way care is provided? How much of what passes for value-based purchasing is really just arranging the deck chairs on the Titanic? What makes sense in trying to codify optimal care for individual patients and to achieve it in as many of them as possible likely will never lead to saving money this year or maybe not even in this decade. This may be just another example of Goodhart’s law: “When a measure becomes a target, it ceases to be a good measure.” 2 What he actually wrote in a paper on monetary policy in 1975 was less pithy. “Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes.” The study by Vasquez shows the limits of using a simple measure, hospitalization rate, to capture a complex clinical reality. We should beware of anyone claiming to know the secret of how to save money over the short term and focus instead on getting in tune with the patient’s perspective—how do we get the best possible outcome over the rest of my life? 12 April 2024 2 Accessed 8 April 2024 at https://en.wikipedia.org/wiki/Goodhart%27s_law. |
Further Reading
High-Cost High-Need Patients A small number of patients use a disproportionate share of medical resources—the high-need, high-cost patient. If you can identify such patients prospectively, care management resources might be deployed cost-effectively, as most studies of care management methods yield disappointing results. Can this approach work? Linking Guidelines and Process Improvement I fear working on the difficult issues of improving health care delivery and linking guidelines and process improvement won’t stop us from ramming the iceberg created by current economic realities of high inflation, low unemployment, wage growth, lack of qualified applicants for positions requiring credentials, and pent-up patient demand that was deferred by the pandemic. Paying for What We Don't Want Do you believe the proverb "you get what you pay for"? What if you pay for what you don't want? Population Health Population health is a phrase that disguises some hard realities as illustrated by two recent reports. Swimming Upstream Our current cultural norms make following traditional medical advice, like eating less and exercising more, difficult for most people to do. Improving health may have more to do with modifying these forces, which is beyond the competence of health care providers and organizations. Vaccine Hesitancy Vaccine hesitancy is a case study with implications for what we can expect from current efforts to create "good healthcare". |