August 03, 2014

The Robot Will See You Now

“Disease management” is all the rage. The reason? In the US and other developed countries, most diseases are chronic rather than acute and chronic diseases can generally be treated but not cured; ergo, they need to be “managed.” Interesting that it’s the disease that’s supposed to be managed rather than the patient. In any event, despite the popularity of the concept, disease management programs haven’t proven terribly effective: one review of 35 Medicare-funded projects in 22 states involving 300,000 patients showed that most did not improve the quality of care or reduce the cost of treatment.  Nonetheless, with the continuing growth in the number of people with chronic diseases—68% of Medicare patients have at least 2 chronic diseases and 14% have six or more—disease management remains popular as a common-sense approach that seems as though it ought to work. And a study in this week’s Annals of Internal Medicine suggests that maybe it does.

The article is a “systematic review and meta-analysis,” or an attempt to get at the truth by combining findings from many different studies. As an aside, the authors initially identified nearly 3000 articles about disease management, but ended up analyzing only 18 since the overwhelming majority were not of sufficiently high quality to be worth including. These 18 clinical trials all used “nurse-managed protocols” to guide the outpatient treatment of such diseases as diabetes, high blood pressure, and elevated cholesterol. The results? Use of these protocols by nurses led to small but statistically significant falls in blood sugar levels, blood pressure, and LDL (the “bad” cholesterol). The article concludes that nurses “are in an ideal position to collaborate with other team members in the delivery of more accessible and effective chronic disease care.” In other words, nurses should take over the function of managing chronic disease.

I have no reservations about handing over large chunks of primary care medicine to nurses—in fact, I think that nurses, particularly nurse practitioners, are better suited to primary care than many physicians and should do more than just follow flow charts. If all that's needed is to stick to clearcut guidelines, then a smart machine would be better than a nurse. But I do have concerns about the mindless application of algorithms to patient care, at least for older patients with multiple chronic illnesses. In younger, less complicated patients, preferably those with only a single disease that requires managing, simply following the optimized treatment strategy is likely to be a good idea. But in older patients with “multimorbidity,” as having multiple chronic conditions is increasingly called, practicing algorithmic medicine leads to disaster.

Suppose I’m seeing an 85-year-old woman with high blood pressure and diabetes and Parkinson’s disease. Call her Janet Dover. Optimal medical management of the high blood pressure means use of a diuretic. Tight control of diabetes means keeping the average blood sugar down, even if that means occasional dips to dangerously low levels. And control of Parkinson’s disease involves using a drug such as Sinemet. But there’s a problem with giving this patient a diuretic and a hypoglycemic agent and Sinemet all together, even if each individually would be a good idea. Both diuretics and Sinemet tend to make blood pressure fall when a person stands up and low blood sugar tends to make a person unsteady on her feet. So give Mrs. Dover all 3 medicines and the next thing you know, she will stand up, be dizzy and off balance—and then she will fall and break her hip.

Good geriatric care is all about figuring out how to treat someone like Janet Dover. It’s not easy and it depends both on her particular combination of diseases and on her willingness to make certain tradeoffs. But it’s not something that can be done by following a protocol, whoever is in charge of the protocol.

No comments: