Welcome to another addition of the Articles of the Month. And yes, I know this is no longer a monthly publication, but it is always published in A month and you get what you pay for. The podcast version can be found over on Broome Docs. Enjoy.
Let’s put PESIT to bed
Oqab Z, Ganshorn H, Sheldon R. Prevalence of pulmonary embolism in patients presenting with syncope. A systematic review and meta-analysis. The American journal of emergency medicine. 2017; PMID: 28947223
For rational physicians, one of the major criticisms of the PESIT trial was that it was so inconsistent with previous large syncope databases. No prior study had revealed large numbers of syncope patients being diagnosed with PE. This is a systematic review and meta-analysis looking at the rate of PE among syncope patients. They identified 12 studies (9 focused on the emergency department with 6608 total patients and 3 focused on hospitalized patients with 975 total patients). The pooled prevalence of PE among ED patients was 0.8% (95% CI 0.5-1.3%). The prevalence among hospitalized patients was 1.0% (95% CI 0.5-1.9%). This data doesn’t tell us, but my guess is that most of the patients with PE would have had signs, symptoms, or risk factors indicating the diagnosis. Either way, these numbers are below the test threshold, and much more in keeping with reality than the PESIT numbers. However, the data here is not perfect, and suffers from incomplete work up bias.
Bottom line: We all already knew this. Work up PE in patients who have signs, symptoms, and risk factors for PE. Isolated syncope is not an indication for a PE workup.
Finding incidental PEs won’t help your patients
Beck KS, et al. Incidental Pulmonary Embolism After Coronary Artery Bypass Surgery: Long-Term Clinical Follow-Up. American journal of roentgenology. 2018; 210(1):52-57. PMID: 29064757
This study provides great evidence that pulmonary emboli are not all the same. They are often incidental and self-resolving, so we need to be very careful who we decide to test. As part of the routine follow-up for coronary bypass surgery, many surgeons are ordering coronary CT angiograms to assess the patency of the grafts. These contrast CTs of the chest frequently find asymptomatic PEs. This is a retrospective chart review that looks at 353 patients who had CABG done at a single hospital and then had a CCTA in the following 2 weeks (median 5 days post-op). 22 patients (6.2%) had PEs founds on these scans. None of these patients had any symptoms of PE or DVT, all were hemodynamically stable, and none had prior history of thromboembolic disease. Because of the post-op risks of anticoagulants, only 2 of the 22 patients were started on an anticoagulant. 21 patients had another CCTA a median of 149 days later, and 20 showed complete resolution of the PE. One patient had signs of residual PE, but it had decreased from and obstruction index of 12.5% to 2.5%, and another follow up CT 3 years later showed complete resolution. There were no deaths or recurrent PEs during the follow-up period. Obviously, postoperative CABG patients are not the same population that we see in the emergency department, but this study is a good reminder that PEs are frequently incidental, and will resolve without any treatment. If these patients had been started on anticoagulation, the harms would have certainly outweighed the (non-existent) benefits. This is why we have to be very careful about who we work up for PE. This is why we are not doing our patients any favours when we retrospectively justify our PE workup because we happened to get a positive CT scan despite the patient being low risk and PERC negative. Exposing patients with incidentally found PEs to anticoagulation exposes them to harm, with no clear benefit.
Bottom line: PEs can be incidental and self-resolving.
Routine pelvic exams also won’t help your patients
Linden JA, Grimmnitz B, Hagopian L. Is the Pelvic Examination Still Crucial in Patients Presenting to the Emergency Department With Vaginal Bleeding or Abdominal Pain When an Intrauterine Pregnancy Is Identified on Ultrasonography? A Randomized Controlled Trial. Annals of emergency medicine. 2017; 70(6):825-834. PMID: 28935285
This study asks a great question, but unfortunately has too many problems to provide a definitive question. They were looking to see whether a pelvic exam was truly needed for women in the first trimester with abdominal pain or vaginal bleeding who already had an ultrasound demonstrating an intrauterine pregnancy. It was a prospective, single-centre, equivalence RCT that enrolled a convenience sample of women up to 16 weeks pregnant. At this hospital, ultrasounds are ordered at triage before the patient is seen by a physician, so randomization occurred after the ultrasound but before the exam. Some patients had already had a pelvic, and so were excluded. They also excluded patients who were hemodynamically unstable, with heavy bleeding, with an IUD in place, suspicion for heterotopic pregnancy, and in who the physician though there was an alternative diagnosis (such as PID) that required a pelvic exam. The outcome was a 30 day compositie of further treatment, unscheduled visits, transfusion, infection, and subsequent identification of another sources of their symptoms. (I am not sure all parts of this composite are of equal value). The big reason their study falls apart: they screened 1280 patients, but excluded 875 (primarily because they didn’t speak English) and ultimately only included 202 in the final analysis. Their power calculation said they needed a sample size of 720. There was no difference between the groups (19.6% had one of the composite outcomes in the no pelvic exam group versus 22.0% in the pelvic exam group.) They say they cannot conclude equivalence because of large confidence intervals, but that really isn’t fair. You need a proper sample size to make appropriate comparisons. The primary diagnostic reason to perform a pelvic exam in the emergency department is to identify an alternative source of the symptoms. In the pelvic exam group, they identified an alternative diagnosis in 2 of 100 patients, although they don’t provide us with any details about those patients. These results aren’t surprising, and fit with previous research (see the next article). Routine performance of any invasive exam doesn’t make sense. Tests need to be considered in the context of pretest probability, harms, benefit of intervention, and severity of condition. Clinician judgement is essential.
Isoardi K. Review article: the use of pelvic examination within the emergency department in the assessment of early pregnancy bleeding. Emergency medicine Australasia. 2009; 21(6):440-8. PMID: 20002713 [free full text]
This is a review article that addresses the same question. They identified 20 prospective observational trials, 7 retrospective chart reviews, and a handful of other articles looking at the utility of the pelvic exam in first trimester bleeding. Unsurprisingly, pelvic exam is not great for ectopic pregnancy, with adnexal masses and tenderness only present 14% and 61% of the time respectively. Ultrasound is far more likely to identify an adnexal mass than a bimanual exam. Inter-rater reliability of the pelvic exam is poor. An open os is only present in 24% of miscarriages, but more importantly there are false positives (12 of 214 open os exams were normal pregnancies in one study). Overall, the pelvic exam leads to a change of management in only 1.3% of patients. These studies are imperfect, and often exclude hemodynamically unstable patients. The results are pretty clear though: there is no reason to perform a routine pelvic exam. Some patients will certainly require one, though, and clinical judgement is essential. If there is heavy bleeding or significant pain, a pelvic exam can be therapeutic by removing clot or products of conception from the cervical os. If ultrasound isn’t available, I would be more likely to perform a pelvic. Similarly, if there is clinical suspicion of an alternative diagnosis, such as trauma, infection, or a foreign body, a pelvic exam in essential.
Overall bottom line: Invasive exams shouldn’t be done routinely. Like the rectal exam in trauma, there is a role for the pelvic exam in first trimester bleeding, but certainly not on every patient. (This almost certainly applies to non-pregnant patients too. See this discussion.)
A (not very convincing) rebuttal of overdiagnosis and overtreatment
Although I disagree with many of the individual points raised, there is one important message here: medical care is incredibly complex and all too often our conversations about it (and those of policy makers) are oversimplified. Beyond that, this piece is littered with logical fallacies and inconsistencies. The author tries to spin her previous moniker of “Ms. Appropriate” into evidence that she is an expert on the topic. She implies that the overuse of cardiac stents is driven by just a handful of outlying “crooks”, not normal doctors, which simply isn’t supported by the numbers. She goes on to talk a lot about how stents might actually be helpful outside of acute coronary syndrome, but without providing any evidence. In fact, the only studies she cites concluded the exact opposite (and the entirety of the literature is pretty clear on this topic). I also have a problem with her discussion of bias. She says (correctly) that people arguing for a less is more strategy have a bias, but because the bias is less obvious that the bias of money in industry, it is more dangerous. Everyone is biased, but trying to claim that the bias of physicians practicing less is more (trying to find the best possible care for their patients) is somehow the same or worse than the financial bias of physicians who profit from ordering more tests and doing more procedures is just crazy. Perhaps the most bizzare part of this essay is when she admits that when she was being worked up as a patient, she was actually fine with uncertainty about the final diagnosis, but the physician was uncomfortable with uncertainty and convinced her to have more testing that she did not want to have. Somehow that anecdote about more testing making the physician, but not the patient, more comfortable is supposed to convince us that doing more in medicine is a good idea. On the whole, I think it’s is clear that we are doing far too much in medicine; that overtreatment and overdiagnosis are far bigger problems than under-treatment and under-diagnosis. However, the author is spot on when she says, “the most accurate conclusion is that sometimes less is more, sometimes more is more, and often we just don’t know”. I would just add: most of the time it is pretty obvious that less is more. This article has spawned a lot of controversy. It is worth reading to understand why.
Bottom line: This essay is worth a read as reminder that policy talk often becomes oversimplified, but in no way counters the mass of evidence that overdiagnosis and overtreatment are huge problems in modern medicine.
Verapamil for infants?
Lapage MJ, Bradley DJ, Dick M. Verapamil in infants: an exaggerated fear? Pediatric cardiology. 2013; 34(7):1532-4. PMID: 23800976
There is an accepted wisdom, not just in emergency medicine, but in pediatrics as well, that intravenous calcium channel blockers should be avoided in infants, usually translated to children under 2 years of age. This paper reviews some of the background for that contraindication. The data is shaky at best, relying mostly on case reports, many of which involved children given calcium channel blockers after already being given beta-blockers, or children given unusually high doses of calcium channel blockers. The initial recommendation was to avoid using CCBs in neonates (less than 6 weeks of age). This recommendation seems to have turned into infants without any real evidence. Although the use of calcium channel blockers in infants will be rare, and I will always do so in conjunction with my specialist colleagues, it is good to know that this contraindication may be more dogma than science.
Bottom line: Some interesting medical history, but still lots of reason to be careful in infants with arrhythmias.
Spoiler alert: EpiPen expiration dates
Cantrell FL, Cantrell P, Wen A, Gerona R. Epinephrine Concentrations in EpiPens After the Expiration Date. Annals of internal medicine. 2017; 166(12):918-919. PMID: 28492859
Expiration dates on medications have always raised questions for me. Manufacturers obviously have a financial interest in setting shorter expiration dates, so that more medication will be bought. On the other hand, we definitely don’t want patients to be using unsafe medications. I had previously read a study that looked at epinephrine that was stocked on an army base since the 1950s, and found that it remained in its original concentration without obvious contamination. This study collected 31 EpiPens and 9 EpiPen juniors that were between 1 and 50 months past their expiration date from patients at a community clinic. Nineteen (65%) of the EpiPens and 5 (56%) of the EpiPen juniors contained at least 90% of their original dose. All of the devices had more than 80% of the stated dose. (I do wonder about manufacturing, and what these percentages would have been on the day the drugs left the factory.) These numbers aren’t as good as I had seen in the past. Clearly we shouldn’t be advising people to cut corners by holding on to old EpiPens, but it is very reasonable to use an old autoinjector if there isn’t another option.
Bottom line: Expired medications are not perfect, but will frequently contain a large percentage of the original dose. I would advise patients to use an expired EpiPen if that was all they had available.
Worst paper of the month: Door to furosemide time
Matsue Y, Damman K, Voors AA. Time-to-Furosemide Treatment and Mortality in Patients Hospitalized With Acute Heart Failure. Journal of the American College of Cardiology. 2017; 69(25):3042-3051. PMID: 28641794
I didn’t really want to include it, but I have heard this paper discussed a few times already, and it is just so bad that people need to know about it (so that we can all ignore it). This is a prospective observational trial of 1291 adult patients who were ultimately admitted to hospital through with emergency department with acute heart failure. (Problem #1: we don’t know which patients will be admitted, or what their diagnosis will be when we are first seeing them, so it is impossible to apply these results to our patients.) They excluded patients without a BNP measured, further limiting application of the results. In these patients, the looked at the time from arrival to first dose of IV furosemide. Using a cutoff of 60 minutes, they say 37% of patients were treated early and 63% of patients were treated late. In hospital mortality was lower in the group treated early (2.3% vs 6%, p=0.002), which is why you might incorrectly hear that you should be treating CHF patients with furosemide early. This trial cannot support that claim at all. It is an observational trial, which means that it provides us with associations only. The two groups were quite different from each other. The early group looked sicker, but what that really means is that they were easier to identify. Complex patients with multiple comorbidities and subtle presentations will not be as easy to identify as acute CHF, may have more than one active diagnosis, and may have more concerning underlying pathology as the basis of their CHF. (CHF is not a single disease. Cardiogenic shock has a very different prognosis that simple volume overload in mild LV failure.) I would expect physicians to take time assessing these sicker patients, and to weigh the risks and benefits of IV furosemide, but I would also expect this group to have a higher mortality. The association here should not surprise us at all, nor should it change our management. After all, furosemide has never been shown to have a mortality benefit in any context, so how could giving it a little bit earlier in the emergency department end up saving lives? Just like with antibiotics for sepsis, we should never delay appropriate treatment once we know it is indicated (and we never purposefully would), but we also should never apply time to treatment criteria retrospectively, because it would require treating many people who don’t require treatment in order to catch all those who turn out to have the disease.
Bottom line: We should not be thinking about door to furosemide time.
To IV or not? #SGEMHOP
Hawkins T, Greenslade JH, Suna J. Peripheral Intravenous Cannula Insertion and Use in the Emergency Department: An Intervention Study. Academic emergency medicine. 2018; 25(1):26-32. PMID: 29044739 [article]
This article is discussed on this week’s SGEM HOP. It is a before and after study looking at IV use in adult patients in the emergency department. They performed a 10 week multifaceted educational intervention aimed at reducing IV access and looked at the 12 weeks before and after that period. The educational intervention centered around the simple question: do you think that this patient has an 80% chance of needing this IV? There were 4,173 patients included in the trial, and they saw a decrease in IV insertion in the post-intervention period (41% before and 32% after, ARR 9.8%; 95% CI 6.9-12.7%). They also saw an increase in the number of IV being used in the department (67% before and 79% after). There are a couple issues with the study. Any before/after study design is subject to the Hawthorn effect, although they performed an analysis that indicates that the Hawthorne effect doesn’t explain their results. The biggest problem might be external validity. In the Australian system, a large number of IVs are placed by medical trainees. In Canada, almost every IV is placed by a nurse. That might explain why their rate of multiple IV attempts (31%) and time to obtain an IV (15 minutes) seem higher than I am used to. I am all for reducing unnecessary interventions, but this study doesn’t tell us anything about possible harms. How many patients needed to be poked multiple times, when it was determined later that an IV really was necessary? How many patients crashed and required immediate vascular access that wasn’t available? It is hard to interpret the benefit of reducing IV insertion rates without any information about harms.
Bottom line: Thinking is good in medicine. Simply asking the question, is this patient going to need an IV, before placing one makes a lot of sense.
Time to start wearing costumes to work
White RE, Prager EO, Schaefer C, Kross E, Duckworth AL, Carlson SM. The “Batman Effect”: Improving Perseverance in Young Children. Child development. 2017; 88(5):1563-1571. PMID: 27982409
After reading this fun little study you might decide to show up for your next shift wearing a batman costume. Children were asked to do a boring repetitive task, but were allowed to take a break and play video games whenever they wanted. They were told it was an important activity and asked to work as long as they could. They then tested three types of self reflection. For internal reflection children asked themselves “am I working hard?” For third person reflection, children asked “is [child’s name] working hard?” And then for an exemplar reflection, the children were given props to dress up like a familiar character (like Batman), and asked “is Batman working hard?” The children taking the outsider’s perspective (Batman) spent more time working than those with the internal perspective, with the third person reflection somewhere in between. So maybe picturing yourself as a hero will allow you to persevere longer at your job? There are a number of limitations to the study. Most importantly when reading the psychological literature, one has to be very aware of the major problems that have been reported with p hacking and lack of replicability of studies.
Bottom line: This was a fun study that you can use as support if you want to show up to work in a Batman costume. Also, this might be a way to trick your kids into getting their chores done.
Cheesy Joke of the Month
My therapist told me that a great way to let go of your anger is to write letters to people you hate and burn them.
I did that, and I feel much better, but I am wondering: do I keep the letters?