Patient satisfaction ≠ quality, round ∞

One thing that’s different about medicine as a business compared to most other businesses is the role of customer satisfaction, or, in the case of medicine, patient satisfaction. In most businesses, the overall goal is to keep the customer satisfied, so that he or she is more likely to think well of your product or service, purchase more of it, and provide good word-of-mouth to friends and family. Of course, price, quality, functionality, options, and many other factors go into producing customer satisfaction, which is why inexpensive crap can sometimes be very popular. Here’s where the difference with medicine comes in. In medicine there are definite standards and evidence to tell us as physicians, advance practice nurses, physician’s assistance, nurses, and any other provider what is and is not quality care. The problem is that those standards often conflict with what it takes to maintain high patient satisfaction scores.

Unfortunately, the more like a business medicine has become, the more patient satisfaction has come to the fore as a metric by which doctors and health care facilities are judged. These days, we live and die by our Press-Ganey scores, the result of the most commonly used patient satisfaction survey used by health care facilities. Indeed, Press-Ganey itself sells its services as “driving performance excellence” in health care. The inherent assumption behind patient satisfaction surveys like Press-Ganey is that if patients are satisfied then the doctors and facilities producing high satisfaction scores must be doing a good job. But it’s subtler than that. The underlying assumption is actually that patient satisfaction equals quality, and the further assumption is that Press-Ganey scores reflect patient satisfaction. Never mind that Press-Ganey scores include questions about a whole host of things that have little or nothing to do with the quality of care being delivered, such as parking and food. That’s why, as I’ve said before many times, patient satisfaction does not equal quality. Sometimes it can, but often it does not, and sometimes it even equals the opposite of quality.

So it was with interest that I saw a story on NPR, Patients Give Doctors High Marks For Prescribing Antibiotics For Common Sniffles:

When they’re sick, Americans seem to know what they want: antibiotics. And if they don’t get them, their doctors’ reputations may suffer.

A study published Monday finds that patients rated themselves happiest with their doctor’s visit when they got an antibiotic after seeking care for a respiratory tract infection, such as a common cold, whether they needed the drug or not.

I’m not a primary care doctor; so I seldom have to deal with this particular request from a patient, antibiotics for what is almost certainly a viral illness, but I know a fair number of internists and primary care doctors who almost certainly groaned in recognition when they saw that headline.

Here’s the study, which was published in JAMA Internal Medicine, Association Between Antibiotic Prescribing for Respiratory Tract Infections and Patient Satisfaction in Direct-to-Consumer Telemedicine. Interestingly, the study came from the Cleveland Clinic, an institution that I’ve been lambasting regularly over the last five years or so for its tight embrace of the quackery of traditional Chinese medicine and functional medicine, an embrace that contributed to the director of its Wellness Institute publishing an antivaccine rant and later going full quack. It’s good to see the useful side of the Cleveland Clinic, which used to be a fierce rival up up the street from University Hospitals of Cleveland, where I did my surgery residency, and one that I used to think of as scientifically rigorous and thus a worthy rival.

The authors point out at the beginning of their brief study:

Outpatient respiratory tract infections (RTIs) are mostly viral in nature and rarely warrant treatment with antibiotics, yet physicians frequently prescribe antibiotics for such infections.1 This decision to prescribe antibiotics for RTIs may be owing to physician assumptions that patient satisfaction will be lower if antibiotics are not prescribed.2 However, evidence supporting these assumptions is mixed.3-6

Direct-to-consumer telemedicine is an ideal setting in which to evaluate the association between antibiotic prescribing for RTIs and satisfaction ratings among patients. Respiratory tract infections are the most common reason that individuals seek medical care in this setting and every encounter concludes with a prompt for patients to rate their satisfaction. We assessed the association between antibiotic prescribing for RTIs and patient satisfaction ratings in the Online Care Group direct-to-consumer telemedicine platform.

So here’s what they did. They studied patient encounters completed between January 1, 2013, and August 31, 2016 through American Well’s Online Care Group, a national provider of telemedicine services to consumers, which ended up being a total of 8,437 appointments for evaluation of upper respiratory symptoms. Patients with respiratory tract infections (RTIs) were defined as those with International Classification of Diseases, Ninth Revision (ICD-9), or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10, which I discussed yesterday in the context of discussing the forthcoming ICD-11), codes for sinusitis, pharyngitis, bronchitis, or other RTI. The authors then categorized prescription outcome as no prescription, prescription of an antibiotic, or prescription of a nonantibiotic medication. Patients rated satisfaction with their physician on scales of 0 to 5 stars (where 5 is most satisfied and 0 is not satisfied at all), dichotomized as 5 stars vs fewer than 5 stars. Individual physicians’ adjusted mean rates of antibiotic prescribing were assessed for correlation with their adjusted mean satisfaction scores, all adjusted for patient, physician, and encounter characteristics. Amixed-effects logistic regression model was then used to evaluate whether satisfaction varied by prescription outcome, accounting for clustering by physician.

Here are the results:

Among 8437 encounters for RTIs with 85 physicians, 5580 (66.1%) resulted in prescription of an antibiotic, 1309 (15.5%) resulted in prescription of a nonantibiotic medication, and 1548 (18.3%) resulted in no prescription (Table). Most encounters (87%) garnered a 5 star satisfaction rating.

A total of 1123 of 1548 patients who received no prescription (72.5%) rated their satisfaction as 5 stars, compared with 5075 of 5580 (90.9%) of those who received a prescription for an antibiotic and 1126 of 1309 (86.0%) of those who received a prescription for a nonantibiotic medication. Compared with receiving no prescription, receipt of a prescription for an antibiotic was strongly associated with rating care 5 stars (adjusted odds ratio, 3.23; 95% CI, 2.67-3.91), as was receiving a prescription for a nonantibiotic medication (adjusted odds ratio, 2.21; 95% CI, 1.80-2.71.) Physicians’ mean adjusted rates of antibiotic prescribing ranged from 19% to 90% (interquartile range, 56%-77%) and adjusted satisfaction ratings correlated with adjusted antibiotic prescribing rates (Pearson correlation, 0.41; P < .001) (Figure).

So, basically, for a study like this, there was a pretty decent correlation between the physicians writing a prescription for antibiotics for a viral upper respiratory infection and the odds that they would receive a five-star rating for the encounter. True, satisfaction levels were still pretty high, even for physicians who didn’t write antibiotics prescriptions, but writing those prescriptions pushed them significantly higher. The authors noted that the absolute difference in patient satisfaction rates wasn’t that high but pointed out that physicians are usually judged by the quartile they land in, not by absolute scores and that patients do often distinguish between physicians with patient satisfaction scores of, say, 4.7 stars versus 4.9 stars.

Indeed the authors note:

For individual physicians, frequent prescription of antibiotics was associated with better satisfaction ratings. Few physicians achieved even the 50th percentile of satisfaction while maintaining low rates of antibiotic prescribing. To reach the top quartile, a physician had to prescribe antibiotics at least half the time; almost all physicians above the 90th percentile had a rate of antibiotic prescribing greater than 75%.

The authors also pointed out that, based on their results, no “other patient or physician factor was as strongly associated with patient satisfaction as receipt of a prescription for an antibiotic.” By way of comparison, actually nothing the patients received was associated with patient satisfaction besides an antibiotic prescription was associated with a five star rating other than a non-antibiotic prescription (adjusted odds ratio 2.21) or using a coupon for a free or reduced-cost visit (adjusted odds ratio 1.58). For doctors, giving these patients a prescription for antibiotics far and away was the best means to increase the odds of receiving a five star rating for any particular patient encounter for an RTI.

Another striking finding of this study is just how wide the variation in prescribing practices is, with the mean adjusted rates of antibiotic prescribing ranging from 19% to 90% for RTIs. This variation is far beyond what can be explained based on evidence and suggests that there are widely differing interpretations of the role of antibiotics in treating RTIs. Either that, or some physicians are better at placating patients without writing for an antibiotic than others.

There are limitations in this study, as with all studies. For one thing, it’s observational. For another, it’s a specific type of practice, meaning that its results might not be generalizable to typical primary care practices. Also, no judgment was made on the appropriateness of the antibiotic prescriptions. Rates were adjusted for diagnosis, though, and the authors note that the “high rate of antibiotic prescriptions we observed is likely inappropriate for the diagnoses.”

I like to point out that there is one particular specialty in medicine that emphasizes patient satisfaction above all. Indeed, at the Cleveland Clinic itself (sorry, authors, I couldn’t resist, but this is the best example), Dr. Mark Hyman’s functional medicine clinic is touted for its popularity and for how satisfied the patients are with the care there, even though functional medicine is rank quackery that combines the worst of both worlds, namely the massive overtesting and overtreatment to which conventional medicine can be prone, with quackery like acupuncture, traditional Chinese medicine, naturopathy, and even homeopathy. Indeed, “integrative medicine” often produces very high patient satisfaction, even though it is an illegitimate specialty that exists to “integrate” quackery with conventional medicine.

It’s not just integrative and alternative medicine, though. There are lots of examples to echo this example of how treating RTIs with antibiotics (and risking causing antibiotic-associated complications for no benefit) is associated with high patient satisfaction. For instance, surgery and invasive procedures are, understandably, often associated with lower satisfaction because of the pain and discomfort, but overly aggressive preventive medicine is associated with high scores. As I discussed, overaggressive screening can lead to overdiagnosis and overtreatment whose harm outweighs the benefit in terms of lives saved by early detection and intervention, but patients (and doctors) love it, because it feels as though they’re “doing something.” There is even evidence that in some situations, increased patient satisfaction is associated with worse, not better, outcomes. One study that I discussed a few years back found that patients in the study who demonstrated the highest level of satisfaction were more likely to have an inpatient admission (adjusted odds ratio 1.12) than those with the lowest levels of satisfaction and also accounted for 8.8% more health care expenditures, including greater prescription drug expenditures. Worst of all, they demonstrated a higher mortality, with an odds ratio of 1.26, which means they had a 26% higher chance of dying. Also, higher patient satisfaction appears to be correlated with higher resource utilization. In addition, denying patient requests (such as for referral, pain medication, and laboratory tests) have been shown to be associated with poorer patient satisfaction scores.

In fairness, though, it must be noted that the relationship between patient satisfaction is complex. For some surgical measures, for instance, high patient satisfaction is correlated with measures of surgical quality. In other words, I’m not saying that patient satisfaction is never associated with higher quality medicine. I’m merely pointing out that the assumption that it’s always (or even frequently) associated with quality is clearly incorrect. Worse, in the case of common conditions like RTIs, the emphasis on patient satisfaction über alles can actually result in harm, as increasingly hospitals and health systems incentivize physician behavior based on patient satisfaction scores. Medicine is about making the patient better, and, unfortunately, that sometimes means not doing what the patient wants.

Source of cartoon used for this post: