No One Knows How To Diagnose CAD

And, once they diagnose it – it doesn’t seem like anyone knows what to do with it, considering all the brouhaha these days about potentially unnecessary PCI and stenting.

But, this is a prospective coronary CT angiography registry that was reviewed to determine whether any value was added with the CCTA over conventional stress testing in patients without known CAD.  They reviewed 22,551 patient records, excluded patients with known CAD, incomplete data, and patients who hadn’t undergone a recent (<3 months) cardiac stress test, and ended up with 6,198 patients.

The point the authors seem to be trying to make is that CCTA is a better test than stress testing, but that’s only part of the story.  What they note that is interesting along the way is that there is absolutely no correlation between stress testing results and CCTA results.  Patients with normal, equivocal, and abnormal stress results had, essentially, the same incidence of normal, <50%, and >50% coronary stenosis.  And, the hidden story about how CCTA is being used in their patient cohort is fascinating – a younger group with typical chest pain and normal stress tests referred to CCTA vs. an older group with less typical symptoms and abnormal stress tests referred to CCTA.

But, then, finally they compare both of their disparate tests to the “gold standard” of invasive angiography, and they find that both tests are awful at predicting >50% coronary stenosis.  Stress testing was 60.4% sensitive and 34% specific, while CCTA was 94% sensitive and 37% specific.  So, we have two tests that are wrong about the presence of disease twice as often as they’re right – and these authors are using a clinically irrelevant 50% stenosis as their “gold standard”.

Rather entertaining to observe the difficulty the cardiology literature is having reconciling all their different imaging options with clinically relevant stenoses, much less outcomes.  Good thing all these inadequate tests are cheap and harmless….

“Coronary Computed Tomography Angiography After Stress Testing”

The Tiniest Three Year Sinusitis Trial

Yet again, another article that saturated the lay press due to its publication in JAMA – this time regarding amoxicillin for acute sinusitis.

The problem is, I agree with the fundamental point the authors are making – according to the introduction, antibiotics for sinusitis account for 1 in 5 antibiotic prescriptions in the United States and they’re typically unnecessary, especially in an era where better antibiotic stewardship is needed.  However, I cannot imagine how a multicenter study of ten community clinics in St. Louis over three years only managed to enroll 166 adults into this study over the course of three years.  Their recruitment diagram states only 244 patients were assessed for eligibility – which seems like it ought to be a a couple months worth of URI presentations in an outpatient setting.

If you read the newspaper, you already know the main results – “no difference between 10 days of amoxicillin and placebo.”  But 11 of 85 intervention group patients discontinued treatment, as well of 12 of 81 placebo patients.  Due to lost data, 4 of 85 intervention patients were excluded from analysis, as well as 7 of 81 placebo patients.  Then, 32% of patients in the intervention group were non-compliant with the intervention – so, while this is valid in a real-world effectiveness sense, they’re increasingly no longer relevant to the actual efficacy of the intervention.  These are big holes in a small study.

And, bizarrely, the baseline characteristics they use to describe the two groups include more social characteristics than clinical characteristics – healthcare insurance, family income, etc.  Children living in the home, children in day care, etc., is an interesting demographic criteria, suggesting unique infectious exposure – 9% more intervention group patients had children at home, but this isn’t statistically significant because the sample sizes are so tiny.  Then, the clinical characteristics they chose only seem to partially reflect issues relevant to antibiotic efficacy – “usual health excellent or good” isn’t a very useful descriptor of whether they have impaired baseline immune function that places them at increased risk of significant bacterial superinfection.  For what it’s worth, the control group was significantly “healthier”, but also had significantly more smoking history.

Getting back to the main results – yes, the average SNOT-16 scores were equal at day 0, 3 and 10, but favored the intervention at day 7 – leading to their final conclusion that amoxicillin was of no benefit.  But, at the individual patient level, the control group patients were impaired from their usual activities almost 50% longer – 1.67 days vs. 1.15 days, and there was a 12% absolute difference in satisfaction with treatment favoring the intervention – 53% vs. 41 %.  But, due to the tiny sample size, none of these differences reached statistical significance.

In the end, it’s a fair real-world trial and addition to the literature, but it’s far too small and flawed a trial to stand on to as evidence.

Oddly, one of the authors receives royalties for the SNOT-16 scale.

“Amoxicillin for Acute Rhinosinusitis: A Randomized Controlled Trial”
http://jama.ama-assn.org/content/307/7/685

Bunnahabhain, Highland Park, Ardbeg

I had a request for a Scotch posting again – and, it has been about six months since I made a Scotch post – so, here’s what’s left on my shelf at the moment.

 – Bunnahabhain 28-year Signatory Collection.  This was a gift Scotch that is, essentially, my major celebration Scotch.  Cardinals won the World Series, bought a house, got a grant, etc.  Smooth and tastes like caramel.
 – Highland Park 18.  Has won several sort of “Scotch of the Year” type awards.  It’s not as complex as other Scotches I’ve had, but it’s sort of a sweet, smooth Scotch that will likely appeal to a wide variety of drinkers.
 – Ardbeg Alligator.  In contrast, this will not appeal to a wide variety of drinkers.  It is a peaty, burning, smoky/charcoal Scotch that has a very distinct taste acquired from being aged in charred barrels.  It’s not for everyone, and it takes a little bit of water to soften up, but the uniqueness outweighs the harshness.

I’m lucky enough to live, literally, within walking distance of the world’s largest liquor store, so typically when I’m shopping for Scotch, I have plenty to choose from.  I’ve also given the Octomore 3.1/152 as a gift, which was the fourth edition in their super-phenolic distillations, and it makes for a very uniquely peaty, nose-filling sensation that lasts forever.  Hard to find, but absolutely well worth it.  Comes in a kind of silly black matte bottle, however:

Patients: More Satisfied, More Dead?

Another article pulled out of the mainstream media – and one that highlights an issue many are familiar with: patient satisfaction.  There isn’t an ED out there whose medical director doesn’t know their patient satisfaction scores, whether Press-Ganey or their own evaluations, and many EPs compensation (or employment) is tied to their patient satisfaction.  And, we’ve argued time and time again that patient satisfaction has nothing to do with high-quality care, and that it’s insulting to degrade medical practice to customer service.

Now, this prospective cohort study of 36,428 patients from Archives demonstrates an association between patient satisfaction with their primary care physician and worse health outcomes.  They used the “Consumer Assessment of Health Plans Survey”, which included four items of interest to the authors: whether the physician listened carefully, explained things well, showed respect, and spent enough time with the patient.  There was also a fifth overall item of general health care rating for all their physician visits from the past year.
For a huge data set with a lot of granularity, the authors, unfortunately, don’t report the unadjusted mortality – which seems like it would be appropriate, when the major selling point is that mortality difference.  But, in any event, a few of the interesting adjusted associations:
 – Black race was more likely to be satisfied with their physicians. (1.17)
 – College graduates were less likely to be satisfied with their physicians. (0.78)
 – Public insurance was more likely to be satisfied with their physicians. (1.14)
 – Those in poor health were more likely to be satisfied with their physicians. (1.33)
That last item – the poor health – could potentially explain all the mortality difference.  They report unadjusted percentages for the rest of their measures, in addition to the adjusted OR, and then their main results come out: more satisfied patients are less likely to show up in the ED, more likely to be admitted, consumed slightly more healthcare dollars, and had slightly more prescription drug expenditures.  And, then, finally, the 1.26 increased hazard ratio for mortality.  Interestingly enough, when patients who have self-reported poor health and more than three chronic diseases are removed, the hazard ratio increases to 1.44.
So, satisfied patients in fairly good health, on whom more healthcare dollars are being expended, have significantly worse outcomes?  There must be more to this story than just patient satisfaction – which, unfortunately, seems to be all the lay press focuses on.
“The Cost of Satisfaction”

Automagical Problem Lists

This is a nice informatics paper that deals mostly with problem lists.  These are meticulously maintained (in theory) by inpatient and ambulatory physicians to accurately reflect a patient’s current medical issues.  Then, when they arrive in the ED, you do your quick chart biopsy from the EMR, and you can rapidly learn about your patient.  However, these lists are invariably inaccurate – studies show they’ll appropriately be updated with breast cancer 78% of the time, but as low as 4% of the time for renal insufficiency.  This is bad because, supposedly, accurate problem lists lead to higher-quality care – more CHF patients receiving ACE or ARBs if it was on their diagnosis list, etc.

These authors created a natural language processing engine, as well as a set of inference rules based on medications, lab results, and billing codes for 17 diagnoses, and implemented an alert prompt to encourage clinicians to update the problem list as necessary.  Overall, 17,043 alerts were fired during the study period, and clinicians accepted the recommendations of 41% – which could be better, but it’s really quite good for an alert.  As you might expect, the study group with the alerts generated 3 times greater additions to the patient problem lists.  These authors think this is a good thing – although, I have seen some incredible problem list bloat.

What’s interesting is that a follow-up audit of alerts to evaluate their accuracy based on clinical reading of the patient’s chart estimated the alerts were 91% accurate – which means all those ignored alerts were actually mostly correct.  So, there’s clearly still a lot of important work that needs to go into finding better ways to integrate this sort of clinical feedback into the workflow.

So, in theory, better problem lists, better outcomes.  However, updating your wife’s problem list can probably wait until after Valentine’s Day.

“Improving completeness of electronic problem lists through clinical decision support: a randomized, controlled trial.”
www.ncbi.nlm.nih.gov/pubmed/22215056

Eat Your Vegetables!

This may be a candidate for an IgNobel Prize, published as a research letter in JAMA: how to get schoolchildren to eat their vegetables!

Control group: normal lunch trays.  Intervention group: lunch trays with compartments specifically labeled with photographs of green beans and carrots.  Results: success!  Green bean choice went from 6.3% of children to 14.8% of children, and carrot choice went from 11.6% to 36.8%.  Amount of green bean and carrot consumption was stable on an individual basis, resulting in an overal net consumption of both green beans and carrots by their cohort.

Of course, this was only a single day intervention – my guess is the effect would fatigue – but, at least, for one day, children ate more vegetables.

This has far-reaching implications for Emergency Medicine.

“Photographs in Lunch Tray Compartments and Vegetable Consumption Among Children in Elementary School Cafeterias”
http://jama.ama-assn.org/content/early/2012/01/31/jama.2012.170.full

Ketamine For Acute Pain Control

So, there’s effective.  And then there’s effective, but insane.  I am aware that low-dose continuous infusions of ketamine are excellent adjunctive therapies to decrease narcotic use in trauma and orthopedic patients, but I have never seen ketamine used in bolus form to treat acute pain in the out-of-hospital setting.

But, that’s what we have.  After an initial 5mg IV bolus of morphine, patients were randomized to receive either additional morphine or ketamine boluses – 1 to 5mg of morphine every five minutes, or 10 to 20mg of ketamine every three minutes.  Pain medication was given per protocol until relief or adverse events.  And, the ketamine group was superior – pain scores dropped 5.6 points on the numerical verbal scale with ketamine and 3.2 with morphine.

However, the ketamine group also had a 39% incidence of adverse effects, compared with 14% of the morphine group.  The morphine group had mostly nausea, with one patient exhibiting a change in level of consciousness.  However, the ketamine group had multiple patients with decreased consciousness, disorientation, and emergence phenomena.  So, while the editor capsule summary states “Supplementing out-of-hospital opiods with low-dose ketamine is an effective strategy to mitigate trauma pain” he is technically correct, but the insanity of this strategy is trying to make an evidence-based decision about intracranial imaging after iatrogenically altering your patients prehospital.

What I appreciate best about this paper is how aggressive the paramedics were with treating pain – the patients receiving morphine averaged 14.4mg, with a standard deviation of 9.4mg!  I see my residents ordering 2mg at a time and it drives me nuts.

“Morphine and Ketamine Is Superior to Morphine Alone for Out-of-Hospital Trauma Analgesia: A Randomized Controlled Trial”
www.ncbi.nlm.nih.gov/pubmed/22243959

Finally, A Useful TPA Concept

Frequent readers of this site will be familiar with my distaste for TPA in stroke – not because I think it’s a therapeutically invalid option, but mostly because its use is being promoted beyond its original scope, too many stroke mimics are receiving TPA, and the published literature supporting new “innovations” in TPA have a skewed interpretation of “safe”.

This paper from Stroke is the first I’ve seen that finally tries to determine whether a patient will actually benefit from TPA in acute ischemic stroke, rather than chaining together studies in a logical fallacy to extend treatment to a larger population.  These authors have developed the “iScore” (no affiliation with Apple Computer), which was developed by logistic regression to predict outcomes in patients with ischemic stroke not treated with TPA.  The components include age, stroke severity, stroke subtype, and medical comorbidities in a scoring system that defines low (>50% good outcome), moderate (10-50%), and high-risk (<10%) groups.

These authors then apply the iScore in a retrospective fashion to their stroke database, looking both at their TPA recipients as well as propensity-matched patients in their non-TPA group.  Now, it’s not exactly prospective, randomized, controlled, but it’s an interesting trick that provides a limited comparison.  The stroke patients in the low-risk group had ~12% absolute outcomes benefit from TPA, the, the moderate group ~10% benefit, and the high-risk group ~2.6%.  There were no statistically significant benefits (or harms) from TPA in the high-risk group, but those patients were >90% disabled or dead at 30 days, regardless of therapy.

One weakness the authors point out in their study – it is sometimes clinically difficult to determine stroke subtype in the acute setting based solely off clinical presentation, particularly when baseline functional status is not perfect.  Regardless, it’s nice to see a paper that looks at better individualizing the risk/benefit equation for TPA – seems as though the 400 patients in the high-risk group did not benefit from spending $2000 on alteplase or the associated increased DRG billing associated with it.  Money isn’t free, after all….

“The iScore Predicts Effectiveness of Thrombolytic Therapy for Acute Ischemic Stroke”
http://stroke.ahajournals.org/content/early/2012/02/02/STROKEAHA.111.646265.short

Would Free Medications Help?

It’s too bad this study doesn’t actually look at what I would have hoped it would – but it’s interesting, nonetheless.  One of my hospitals is a true safety-net hospital and we see, repeatedly, repeatedly, repeatedly, the complications of neglected chronic disease.  One of our frequent laments is whether the costs of recurrent acute hospitalization wouldn’t be prevented a hundred times over if we’d simply sink some costs into preventative maintenance care, free medications, etc.

This study almost looks at that.  This is from the NEJM which compared the outcomes of patients following myocardial infarction, and they follow a group which receives completely free medication and a group that does not.  Unfortunately, the group that does not receive free medications is still receiving heavily subsidized medication support, and is only responsible for a co-pay.

Despite only needing to come up with a co-pay, there’s a significant difference in medication compliance, with an average absolute difference in full adherence with medications of ~5-6%.  With this minimal absolute difference in adherence, the full adherence group had significantly fewer future vascular events – mostly from stroke and myocardial infarction – approximately a 1% absolute decrease.  There was a non-significant decrease in total costs associated with the patients who were on the full-coverage medication plan.

Now, they don’t follow-up any medication-related adverse events, so this is the most optimistic interpretation of benefits of full-coverage, but it would seem that it is overall cheaper and more beneficial to supply medications for free.  And, it makes me wonder what the results of a similar cost/health-benefit study would show in our safety-net population.

“Full Coverage for Preventive Medications after Myocardial Infarction”

Safety-Nets & ED Length of Stay

This is a relatively intriguing public policy article in JAMA following up in a timely fashion regarding the new CMS Emergency Department quality measures.  These new measures include various time-to-X measures, including length of stay, length of time to admission from bed request, etc.  There is some concern that these quality measures may be tied to federal funding, unfairly targeting “safety-net” hospitals that are not at baseline provided with the resources to address patient flow issues.

This article is a review of the NHAMCS database, a national probability sample survey of patient visits, looking at independent predictors of increased length of stay in patients admitted and discharged from the Emergency Department.  Based on the review of this sample, they do not see a significant difference in ED length of stay – and conclude that these quality measures should not be of concern to “safety net” EDs.  However, these general time-based measures mask most of the problems encountered in “safety net” institutions.
There are some baseline differences in patient characteristics between the safety-net and non-safety-net hospitals in their sample, and they tend to work in favor of safety-net hospitals.  The safety net hospitals in this sample tended to have younger patients with lower triage acuities, which should work in favor of reduced ED overall average length of stay.  My anecdotal experience suggests that, once the quality measures track more detailed ED transit times, I believe we will see more significant deficiencies drop out in the safety-net group.
“Association of Emergency Department Length of Stay With Safety-Net Status”