Don't jump to conclusions about that JAMA surgical readmissions paper

On February 3, JAMA published a paper online about readmission rates after surgery. The focus of most tweets was on the most common cause for readmission—surgical site infections (SSIs)—in 19.5% of readmitted patients.

At first glance, this suggests that infection rates after surgery were 19.5%, but that is not so. The paper said that 19.5% of the readmissions were caused by infections.

Of 498,875 total operations reviewed, only 30,270 (6.1%) were readmitted for any reason, and only 5576 (1%) of all patients were readmitted for SSIs.

According to the full text of the paper, the authors had two main points:

Readmissions: Sometimes it's the patients

My Twitter friend Dan Diamond (@ddiamond) posted a picture of a slide that said a hospitalized patient was taught to inject insulin using an orange to practice on. When he was readmitted to the hospital with a very high blood sugar, it turned out that instead of injecting himself at home, the patient was injecting his insulin dose into an orange, and then eating it.

We've all heard stories about patients who took suppositories by mouth instead of the way they were intended.

Since doctors get blamed for just about everything, some would say that patients who take suppositories by mouth or eat an orange filled with insulin do so because they were not properly taught by their doctors (or nurses).

I have blogged before about the problem of who is at fault if patients do not follow up. Although I feel that much of the time it's the patient who decides not to return for follow-up, it seems prevailing sentiment and possibly even the courts say it's the physician who should be held responsible.

But how do you explain this? A study in Heart, a BMJ journal, found that of 208 hypertensive patients referred to a clinic for suboptimal blood pressure control, 52 (25%) were either completely or partially non-adherent [aka non-compliant] with their antihypertensive medications as determined by urine mass spectrometry.

The authors concluded that urine testing for medications or their metabolites would help doctors avoid ordering unnecessary investigations for patients whose blood pressures were not well-controlled.

The reasons for patient non-adherence were not mentioned. Could all 52 patients not have been told about the importance of taking their medications? I doubt it.

You might think the 15% who were partially non-adherent may have forgotten to take the drugs occasionally, but it turns out that most of those in this group took adequate doses of most of other their prescribed medications. This suggests that they selectively omitted some doses of one or more drugs.

The only explanation I can fathom for the 10% who had no traces of any BP meds in their urine is that they just said "to hell with it" and didn't take their meds at all.

I know someone with type 2 diabetes who doesn't watch her weight or what she eats and doesn't check her blood sugars. She says, "You've got to die of something. I'd rather live my life the way I want to."

Is it that doctors and nurses aren't educating the patients or are the patients at fault?

The answer to this question has important implications because of the newly established financial penalties for hospitals with high readmission rates.

Older methods that may improve adherence are tracking prescription refills and having pharmacists or nurses specifically assigned to explain medications to patients in detail.

Here's something that might help.

A recent meta-analysis showed that adherence to HIV/AIDS antiretroviral therapy was modestly improved when patients were sent reminders to take their medications by text message. Those who were more adherent had lower viral loads and better CD4 counts.

Of course, such an intervention assumes that patients have mobile phones or pagers capable of receiving texts, will check for messages, and will act upon the advice. Compared to patients with HIV/AIDS, those with hypertension might tend to be much older and possibly not as technologically savvy.

So what is the solution? I don't know, but sometimes the problem is the patients.

Duh? Postoperative complications lead to readmissions

God knows I’ve written more than my share of papers that the Nobel Prize Committee has rightfully chosen to ignore. I understand that academicians need to publish in order to keep their jobs. Writing a paper is hard work and I don’t really want to demean it.

A paper reports that complications of surgery are linked to increased rates of readmission, and this seems rather obvious to me. It got a lot of media attention, and comment is needed.

The study, published in the Journal of the American College of Surgeons, looked at the records of over 1400 patients who had general surgical operations and found that 163 (11.3) were readmitted within 30 days of discharge.

The authors make some good points such as readmissions were not related to age, race, sex, or certain co-morbidities such as diabetes, smoking status, COPD, ascites, hypertension, steroid use, unintentional preoperative weight loss, history of bleeding disorders or renal disease. Readmissions were significantly more apt to occur if patients had preoperative dyspnea, open wounds or disseminated cancer.

But the main findings that readmissions were due to complications and the more complications a patient had, the more likely he was to have been readmitted, are not exactly earth-shattering.

The press release and articles accompanying the paper’s publication were a little over the top.

Here are some quotes from a few of the many articles about this research:

From the American College of Surgeons website: Findings published in the Journal of the American College of Surgeons lead researchers to devise a patient safety plan to decrease complications for the benefit of patients and hospitals.

From a site called Redorbit: [The surgeon-author] reported that the results of this investigation provided a framework for his research team to develop a simple complication-prevention plan that minimizes the risk of surgical patients developing complications.  

No such plan is mentioned in the paper.

From Infection Control Today: This patient safety approach includes engaging the postoperative care team to start transition-of-care planning early—especially for high risk patients—to encourage early discharge from the hospital. This too is not specifically stated in the paper nor is it supported by the data. In fact, early discharge might result in more readmissions. Who knows?

Incidentally, the press release, in most cases printed verbatim by the medical news websites, was wrong about the data forming the basis of the study. It says, “Researchers conducting this retrospective study analyzed patient records from hospitals that were enrolled in the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP), which is the leading nationally validated, risk-adjusted, outcomes-based program to measure and improve the quality of care in private sector hospitals. Data from Emory University Hospital was merged with ACS NSQIP data to identify unplanned readmissions.” Not so. The paper included patients who were operated on only at Emory University Hospital.

Slight digression: Even if readmission rates were not increased by complications, I would be in favor of a plan to reduce them.

One of the reasons I am skeptical about a lot of things is that when you look into the details, you often find that the “spin” produced by stories about a research paper is not always matched by its content.

I’m not the only one. A French study on the PLoS One website in September found that the spin generated by press releases and abstracts is highly likely to influence news reports about the research and overstate its beneficial effects.

I've written before about the need to read the whole paper and not just the abstract. You must be cautious about what is written in the press release too.

How to lower hospital readmission rates: Let 'em die

As new Medicare rules kick in, some 2200 hospitals nationwide are facing financial penalties for high 30-day readmission rates for myocardial infarction, congestive heart failure and pneumonia. Medicare payments will be lowered by as much as 1%.

Investigators at the Skeptical Scalpel Institute for Evidence-Based Outcomes and Advanced Research (SSIEBOAR, catchy acronym, don’t you think?) have come up with a plan that is certain to lower readmission rates across the board. Some have said the idea should be patented but the institute is not-for-profit and thus is willing to share.

The solution is quite simple—let the patients die. Yes, death reduces readmission rates for all diseases, not just MI, CHF and pneumonia.

Oh, there may be some resistance and relatives of the patients may complain, but at least Medicare will be satisfied and after all, isn’t that why we became doctors?

Another outcome measure, hospital length of stay is also positively impacted by death. For example, if the average length of stay for a patient with a heart attack is 4 days, a patient who dies on hospital day #2 would lower the hospital’s average. Death also results in fewer resources being utilized, which saves the hospital money for those patients whose reimbursement is based on the DRG.

I confess. I’m not serious, and the idea is not original.

There are many issues. In most cases, as length of stay is ratcheted down, readmission rates will rise. One way to reduce readmissions is to keep patients in the hospital longer. And what about the things the hospitals and doctors can’t control? A recent study found that only 63% of Medicaid patients with diabetes, hypertension and hypercholesterolemia actually took their medications regularly.

So what is the solution?

Assessing quality of care in hospitals is a difficult task. People like me have complained about focusing on processes such as the Surgical Care Improvement Project because adherence to process measures does not always correlate with good outcomes. [See previous blog here.] However, processes are much easier to track than outcomes.

The problem with outcome measures is that experts can’t agree on which ones to measure. Even something as seemingly straightforward as death can actually be complex. A 2010 paper in the British Medical Journal on this subject was reviewed in a blog, which pointed out the difficulties with death as a benchmark. This holds true even when death is adjusted for risk.

Readmission rates are also controlled by physicians, not hospitals. Even concurrent review of readmissions by hospital utilization staffs has not been effective in reducing these numbers.

There is another factor. Here’s an anecdote that might help you understand the problem. An elderly woman was admitted for congestive heart failure. After a few days of intense medical care, she was discharged. She was readmitted for CHF three days later. When interviewed during her history and physical exam, she admitted that she 1) did not take any of her prescribed medications at home, 2) continued to smoke cigarettes and 3) did not follow her cardiac diet.

Is it really fair to penalize hospitals for readmissions, many of which cannot be prevented?