Telephone coaching doesn't work for diabetes

A noble effort by the good folks at the General Practice and Primary Health Care Academic Centre at the University of Melbourne to demonstrate the effectiveness of telephone coaching by trained nurses in type 2 diabetes ended in disappointment.

As reported in the British Medical Journal, they randomized 59 primary care practices with 473 adults with A1C above 7.5%. The nurses in the  intervention practices were trained to make structured calls to their patients, engaging them in dealing with their chronic problems, managing stress, improving their lifestyle, self-monitoring their disease, and doing some other good things. The program included seven telephone coaching sessions over 10 months, a face to face coaching session at 12 months, and a final telephone session at 15 months. The nurses were paid to do the calls, overcoming an important barrier in many settings. In other words, this was a well-designed, fairly high-intensity intervention. And, the evaluation is very rigorous and well-designed with little room for random error or bias. These investigators are pros and it shows.

Unfortunately, it didn't work. After 18 months, the A1C in the intervention group was the same as the control group (P=.84). On the plus side, fewer intervention patients were overweight or obese (P=.03). However, some outcomes actually got worse: HDL-cholesterol (P=.05) and systolic blood pressure (P=.07). Given the very large number of secondary outcomes reported, and the lack of any big trends across the outcomes as a group, these apparent effects are probably random error due to multiple comparisons. All in all, there is no evidence that this intervention had important effects on the patients or the health care system.

This isn't the only evidence that high-intensity phone interventions don't work in this setting. Even if they were to generate some improvement, they are so expensive that they can only be used for the sickest patients. So for now, we are left with the proven beneficial effects of less intensive patient engagement techniques that seek to keep patients connected with the practice and activate them when they need to be, rather than educate and prepare them ahead of time. It looks like the "teachable moment" may be more valuable than all the that coaching and preparation.


Blackberry ID, Furler JS, Best JD, Chondros P, Vale M, Walker C, Dunning T, Segal L, Dunbar J, Audehm R, Liew D, Young D. Effectiveness of general practice based, practice nurse led telephone coaching on glycaemic control of type 2 diabetes: the Patient Engagement and Coaching for Health (PEACH) pragmatic cluster randomised controlled trial. BMJ. 2013 Sep 18;347:f5272. doi: 10.1136/bmj.f5272.

Missing the diagnosis of CKD

The good folks at the Oregon Rural Practice Based Research Network looked through their database of lab results and found 865 patients with laboratory evidence of chronic kidney disease (CKD). Of these, 52% had no documentation of the diagnosis in the chart. CKD in women was missed more often than in men. Importantly, the care of high blood pressure was better in those CKD patients where the provider had documented the diagnosis.

What's going on? Is this just a documentation issue?  Probably not. We've seen the same phenomenon in several other clinical settings. It's related to the difficulty of interpreting serum creatinine levels, the complexity and relative newness of the diagnostic criteria for CKD, the permanent brain cramp for all things kidney that medical school renal physiology courses induce, and the presence of the creatinine test in larger panels that are used when kidney disease is not top of mind. So, like the Oregonians, we also found that the number of known CKD cases goes up by 80-100% when the creatinine is systematically looked at (in the form of the estimated GFR value).

You can't engage patients in the care of their kidney problems and prevent end-stage renal disease if you don't even recognize that the patient has kidney disease in the first place! Happily, there are systems to help, and they are effective, easy and inexpensive. Let me know if you want to know more.

Rao MK, Morris CD, O'Malley JP, Davis MM, Mori M, Anderson S. Documentation and management of CKD in rural primary care. Clin J Am Soc Nephrol. 2013 May;8(5):739-48. doi: 10.2215/CJN.02410312. Epub 2013 Jan 31.

Does CKD increase the risk for prostate cancer?

Here's one more reason to aggressively identify, manage, and stay connected to patients with chronic kidney disease (CKD): Prostate cancer. 

A recent study from Korea analyzed a group of men with elevated prostate-specific antigen levels (PSA 4-10 ng/ml). They were all biopsied, but those with stage 3-5 CKD (eGFR < 60) were more likely to have a cancer detected.

Although I'm not sure that PSA screening is a good idea in this population (or any others, for that matter), these findings are certainly disturbing. What isn't clear is whether our strategies for slowing the progression of renal insufficiency (especially blood pressure control and angiotensin inhibition) can prevent the development of prostate cancer.

Cho SY, Park S, Lee SB, Son H, Jeong H. Differences in prostate cancer detection rates according to the level of glomerular filtration rate in patients with prostate specific antigen levels of 4.0-10.0 ng/ml. Int J Clin Pract. 2013 Jun;67(6):552-7. doi: 10.1111/j.1742-1241.2012.03014.x.

"By the way...": The pros and cons of incidental findings

Melissa Healy wrote a very interesting story for the LA Times last week on what happens when a medical test reports something unexpected.You're looking for pneumonia and the chest x-ray shows a cancer in the spine, but you don't know where it came from. A head CT to rule out subdural hematoma after a head injury shows an aneurysm...that might or might not cause trouble one day. These "incidental findings" are disturbingly common.

High tech imaging tests like CT scans and MRIs are particularly prone to this because they are not very specific in their field of view and provide a huge amount of information. If you get a CT of the abdomen to look at the liver, you're going to get all sorts of information about the gallbladder, stomach, intestines, kidneys, pancreas, etc. even if you don't want it. In my experience, lungs and brains are particularly full of findings of "uncertain significance" that are time consuming, costly and difficult to evaluate. They certainly induce a huge amount of anxiety in the patient.Whole genome studies are going to be even worse. The vast majority of genetic information in a person is not interpretable with anything like the accuracy required for clinical decision making.

The new that stimulated Healy's article was that the Presidential Committee for the Study of Bioethical Issues said professional organizations need to start writing guidelines for how to deal with these incidental results. In other words, we should stop treating them like they are not anticipated - we know we are going to get them even if we don't ask for them, so we better make a plan.

I'm all for coming up with guidance on this tricky issue, especially in cases where the consequences of over-diagnosis and unnecessary work-ups are high and the benefits are low. 

The unintended uses of diagnostic tests have a few other upsides to keep in mind, though. They can be used for public health surveillance to help develop strategies for the whole population of patients. This is done routinely with infectious disease tests, for instance, to alert us to new outbreaks.  More recently, we have been systematically scanning commonly done blood and urine chemistry tests to identify patients with undiagnosed diabetes and kidney disease. In the case of chronic kidney damage especially, even the doctors don't always recognize the condition as present. By using an automated system to monitor laboratory results, we can identify patients who need extra attention, reach out to them, connect them to the resources they need and prevent emergencies and other complications.

If you have thoughts on this issue, do leave a comment!

New data on quality of care and cardiovascular outcomes in diabetes

A recent report from The Netherlands analyzed the relationship between quality of care in diabetes and long-term cardiovascular outcomes. The authors had access to a large database of 10,058 patients with type 2 diabetes. They looked at a number of predictors including whether the patients were treated for high blood glucose, high cholesterol, high blood pressure and albuminuria (a marker of incipient kidney damage in diabetes). Over 3 years, treatment of high lipids and albuminuria were each associated with about a 25% reduction in cardiovascular events and all-cause mortality (HR = 0.77, 0.67–0.88; HR = 0.75, 0.59–0.94). Glucose lowering treatment status was associated with a similar effect, but only in patients with an elevated HbA1c level at baseline (HR = 0.72, 0.56–0.93). (Oddly, treatment of hypertension was not associated with better outcomes.)

This report is important for several reasons. First, it serves to validate that we are probably measuring the right things when we do quality improvement based on the Medicare 5-star program or other quality measurement initiatives based on the HEDIS criteria.

Second, it points out how soon we get the payoff. The effect of high quality management is apparent within 3 years, probably sooner. (Unfortunately, the report does not provide information on exactly when the effects were seen, only that they occurred over the course of the 3-year study.) You don't need to wait 20 years to see the benefits in terms of mortality, morbidity and costs.

Third, this study emphasizes the big opportunity represented by untreated albuminura. Primary care in the US is generally fair to good at tracking A1C and cholesterol in diabetes patients, but our track record with monitoring and managing albuminuria is much worse. So, a big take-home message for me is to work hard on urine monitoring.

Sidorenkov G, Voorham J, de Zeeuw D, Haaijer-Ruskamp FM, Denig P (2013) Do Treatment Quality Indicators Predict Cardiovascular Outcomes in Patients with Diabetes? PLoS ONE 8(10): e78821. doi:10.1371/journal.pone.0078821

Screening Asymptomatic Adults for Chronic Kidney Disease (CDK)

This month has seen a scuffle of sorts between the American College of Physicians (ACP), representing America's internists and with a viewpoint largely derived from the primary care providers who make up much of its membership, and the American Society of Nephrology (ASN), a group of sub-specialty kidney physicians. It started when the ACP released one of their hallmark practice guidelines. This one was about screening and management of early stage chronic kidney disease (CKD) and came to the conclusion that screening of asymptomatic adults for CKD was not warranted, largely because it has never been shown to be effective and there are clearly potential downsides. (ACP guidelines are not just evidence-based, they are downright evidence-driven.) The ASN took exception to this guideline and published a flurry of press releases calling for screening.

Full disclosure: I am a Fellow of the ACP and a big proponent of evidence-based medicine. However, I don't have a horse in this race, other than my desire to not see CKD become an inter-specialty football.) The ASN is right that lots of CKD is not being diagnosed or managed to prevent progression, but neither screening nor ignoring the problem may be the best way forward.

In my experience setting up chronic disease registries and decision support systems, 40-50% of the people in primary care with undiagnosed CKD have had an abnormal blood or urine test in the past two years. These tests weren't done for screening, and although abnormal, they didn't trigger a diagnosis by the provider. (The reasons for this are many, including the tricky nature of the serum creatinine in older adults and uncertainty about the diagnostic criteria on the part of the providers.) So, rather than go out and screen a new batch of folks and figure out how to get them plugged into care, the first thing to do is identify all the patients already in the practice with evidence of CKD and start managing them. These patients are currently under care for something, so we don't have to find them a provider. What we do need to do is build systems to identify them, bring them to the attention of their providers, and support them and the practices as they develop plans to prevent progression.

Very few practices have the tools to manage this job now. Screening a whole new population of CKD patients will just make the situation worse until we have systems in place to not only identify these patients, but engage them and their providers in the long-term process of averting, delaying, and mitigating progression to renal failure. Although Electronic Medical Records (EMRs) can help, none of the currently available EMR systems has the capacity to manage chronic care in a way that doesn't create a lot of extra work for the practice. Happily, there are systems that can do this at low cost and with good results, including documented reductions in overall utilization.

So, rather than argue about whether we should find more patients with CKD, let's work together to take better care of the patients we already have with undiagnosed and unmanaged disease.

You can read the ACP's Clinical Practice Guidelines in Annals of Internal Medicine.

Also see the ASN's Press Release here.