Fearless chief Ted has organized the first of our Tuesday Night Journal Clubs (with apologies to Sheryl Crow). The goal of these meetings is to informally discuss several relevant and provocative papers on a certain topic of interest in EM. Particularly, we wanted to give attendees a foundation in the literature, and some ammunition when dealing with other services. We picked a dinner theme, with an appetizer, main course, and dessert.

Our appetizer came from the classic JAMA Rational Clinical Exam series, “Does This Patient Have Appendicitis?” (Wagner et al. JAMA Nov 20 1996, Vol 276, No. 19, PMID 8918857). Some interesting and quotable tidbits from this comprehensive review:

• RLQ pain had a positive likelihood ratio of about 8, and negative LR of between 0 and 0.28
• Rigidity wasn’t very sensitive, but had a positive LR of 3.8
• Migration of pain had a LR of 3.2
• Psoas sign, rebound, guarding, nausea, vomiting, fever and anorexia had relatively unimpressive positive and negative LR’s. Having ‘no similar pain previously’ gave a negative LR of 0.32.

The main course and desert follows below!

Our main course featured two papers on dealing with contrast CT in diagnosing appendicitis:

“Advantages of Focused Helical CT Scanning with Rectal Contrast Only vs. Triple Contrast in the Diagnosis of Clinically Uncertain Acute Appendicitis.” Mittal, Goliath et al. in Arch Surg 2004 (139) p495-500 (PMID 15136349).

The authors found the sensitivity, specificity, PPV and NPV were comparable in the triple-contrast vs. rectal-contrast groups. The rectal contrast group tolerated the study better (1.5 L in, by the way – compared to 900 oral and 900 rectal in the triple groups), cost less ($305 vs. $620) and had no missed diagnoses with a better negative appy rate. They also noted a lower perforation rate compared to triple-contrast, but that was likely due to a failure of randomization – the triple-contrast group waited about 40 hours before presenting, compared to about 30 hours for the rectal group. ED-to-OR times were comparable in both groups, despite the time saved avoiding PO contrast.

Our group had a lot of confusion about the methodology. This is mostly because the authors note that 684 appendectomies were performed during the study period, but only 91 were enrolled in the study. Why? Because “only patients with uncertain diagnosis of acute appendicitis” were entered. We took this to mean cases in which the ED docs consulted surgery before the CT was done. Of course, we have no idea how many CT’s were done without consulting surgery, and how these were performed. And we were surprised by their determination that only 91 out of 684 of their appendicitis diagnoses were “uncertain” and thus needed a CT – the rest went straight to the OR! But this is a paper out of Southfield, Michigan, and the data was collected in 2000-2002, perhaps a period before CT gained the prominence it has in New York these past few years. (It’s worth noting in the discussion section, one of the authors floats the idea of a moratorium on ED docs ordering CT scans before the surgeon sees the patient – in my limited experience with surgeons, they’d prefer a moratorium on consults pre-CT).

The second paper of the main course was, “A systematic review of whether Oral Contrast is Necessary for the CT diagnosis of Appendicitis in Adults,” by Anderson et al, Am J Surg 190 (2005) 474-478 (PMID 16105539).

This was a less-than-rigorous review – using only MeSH terms on pre-2004 papers – of patients 16 or over, in which CT scans were used to diagnose appendicitis. The papers they picked had to provide enough data to calculate sensitivity, specificity, and NPV and PPV. Studies had to also be consistent in how contrast was given, too, and could not be just on women or children. That narrowed things down from 189 hits to 23 English-language papers.

Their analysis of those 23 papers showed that CT techniques that didn’t involve oral contrast were equivalent, if not better (!) than those with oral contrast (so IV +/- rectal was better than triple contrast, or oral + IV, or rectal alone). How is that even possible? IV contrast enhances the wall of the inflamed appendix, and oral/rectal contrast works by opacifying the cecum, allowing for visualization of appendiceal obstruction. But we kinda believed oral contrast wouldn’t hurt interpretation. Yet they found that the sensitivity and PPV of CT, sans PO contrast, was better than with it (specificity and NPV were about the same). They can’t explain it, and suggest maybe publication bias was at fault – maybe studies diminishing the sensitivity of noncontrast CT weren’t published, or residents were reading those films.Also, the introduction of helical CT’s during the study period may be confounding results.

This 2005 study was also notable because it references a 2000 statement from the Group Health Cooperative that states on 20% of patients undergoing appy have had a preoperative CT scan – that was a few years after Rao’s landmark studies and several years before we achieved the near-100% rate we have now.

For ‘desert’, we wanted to end on a light but interesting note, so we chose a paper from the EM literature (as opposed to the surgery papers from the main course). When it comes to appendicitis, the EM literature has the benefit of enrolling all comers (or at least, more comers) and so may be more pertinent to our practice.

The paper discussed was, “The use of the clinical scoring system by Alvarado in the Decision to Perform CT for acute appendicitis in the ED,” by Robert McKay et al in AJEM 2007 (25) 489-493 (PMID 17543650). These authors applied the previously-devised Alvarado clinical scoring system to decide which ED patients need a CT. The scoring is described below — each criteria can be remembered by the irrelevant mnemonic, MANTRELS:

Migration: — 1 point
Anorexia-acetone:– 1 point
Nausea-vomiting — 1 point

Tenderness in RLQ — 2 points
Rebound pain — 1 point
Elev. Temp over 37.8C — 1 point
Leukocytosis — 2 points
Shift to the left — 1 point

For a maximum score of ten. The authors retrospectively looked at 150 charts of patients aged 7 and older, finding Alvarado scores 3 or lower were 96.2% sensitive for not finding appendicitis with a specificity of 67%. Patients with Alvarado scores 7 or higher had an incidence of acute appendicitis of 77.7% (28/36).

For those equivocal cases of scores between 4-6, the sensitivity of scores in predicting acute appendicitis was 35.6%, and the specificity 94%. The sensitivity and specificity of CT scans in patients with equivocal Alvarado scores remained
high, at 90.4% and 95%, respectively.

From this, the authors concluded that, “In the equivocal clinical presentation of appendicitis as defined by Alvarado scores of 4 to 6, adjunctive CT is recommended to confirm the diagnosis in the ED setting. If clinical presentation suggests acute appendicitis by an Alvarado score of 7 or higher, surgical consultation is recommended. Computed tomography is not indicated in patients with Alvarado scores of 3 or lower to diagnose acute appendicitis.”

Our problems with this paper stem from our problems with the Alvarado system: any patient with a fever, white count, left shift, and nausea has four points and is therefore in the “equivocal” zone for appendicitis — even if their chief complaint was diarrhea or Crohn’s flare. Still, this is a fairly easy-to-learn tool to help justify CT scans and calls to surgery, if you’re on the fence or the patient has an ambiguous presentation.

So, there you have it — our first Tuesday Night Journal Club, and some useful facts and guidance for navigating the diagnosis and disposition of appendicitis patients.

The first Journal Club of the year got off to a thunderous start (there was lightning and flooding, too, and an explosion a little bit later). Those who braved the elements participated in a discussion about thrombolysis before angioplasty for STEMI, centered on two papers with two different conclusions about its benefit.

The primary paper we covered was from the GRACIA-2 non-inferiority RCT, called Primary angioplasty vs. early routine post-fibrinolysis angioplasty for acute myocardial infarction with ST-segment elevation (from the European Heart Journal 2007 Vol. 28, p949-960).

Our secondary paper was from the ASSENT-4 RCT, called Primary verses tenecteplase-facilitated percutaneous coronary intervention in patients with ST-segment elevation acute MI (from the Lancet 2006: Vol 367, p569-578).

The GRACIA authors had their hearts in the right place — they were thinking about the STEMI patients that experience significant delays before PCI, and wondered if thrombolytics beforehand were safe or helped. Thus, the timings for lysis and PCI is different than in the ASSENT trial — GRACIA patients got PCI within 3-12 hours of lytics (average: 5.9 hours) whereas ASSENT patients waited just 135 minutes on average (interval quartiles were 90 and 210 minutes).

In 2002-2003, the GRACIA authors enrolled (presumably) all patients over 18 from fifteen hospitals in Spain and Portugal (10 of which had interventional facilities). They were included if they had symptom onset within 12 hours, lasting longer than 30 minutes, with ST-segment elevation of at least 0.1 mV in at least two limb leads, or at least 0.2 mV in two or more contiguous precordial leads, or left bundle-branch block or paced rhythm. Patients were exlcuded if they were in shock, had a terminal illness, had evidence of cardiac rupture, liver dysfunction, creatinine over 2.5, or contraindication to lytics, aspirin, plavix, heparin (or more arcane reasons described in the text).

Patients who met criteria were randomized to receive PCI of the culprit artery within three hours of randomization, or fibrinolysis plus delayed PCI. All patients received some amount of aspirin (200-500 mg) in the ED. The fibrinolysis-first group, 104 patients in all, received a 5 second bolus of tenecteplase over a period of 5 s based on bodyweight (between 30-50 mg) and an enoxaparin bolus of 30 mg followed by a subcutaneous dose of 1 mg/kg. Then, between 3 and 12 hours later, PCI was performed on 87 of these patients. The culprit vessel was dilated if its TIMI flow was less than 3 or it was more than 50% stenosed. They’d stent more lesions if a lot of myocardium was threatened, and patients who remained in pain or maintained ST elevations 90 minutes after lytics went straight to PCI.

The 108 primary-PCI patients received unfractionated heparin (to achieve an activated clotting time of 350–450 s during the invasive procedure) and abciximab, given as bolus of 0.25 mg/kg, followed by a drip rate of 0.125 mg/kg/min. One patient died before catheterization. Of the PCI grouped patients, one died, 107 got angiography, and of those patients, 95 received Guidant stents. Of the 104 fibrinolysis patients, 84% continued to PCI and received stents. “Beta blockers, ACE inhibitors, statins, and post-interventional antithrombotic therapy were administered to all patients, as outlined in the international guidelines.” No patients got more heparin or lovenox if recanalization was successful.

Results made sense in terms of reperfusion — 67% of the early fibrinolysis group had TIMI 3 flow in the culprit artery by the time angioplasty was performed, compared to 14% in the early PCI group. Post-procedure numbers were similar, at around 85%, and all but one patient maintained TIMI-3 flow at 6-week followup. Infarct size was similar for both groups, and both groups had similar good LV function at 6-week followup.

In terms of bleeding events, “there was no significant difference in major bleeding rate among the groups; two patients (2%) in the post-fibrinolysis angioplasty group experienced major bleeding, with one episode of intracranial haemorrhage, and three patients (3%) in the primary angioplasty group.”

At six months, “the incidence of death, reinfarction, stroke (no cases of stroke were observed during follow-up), or revascularization was 10% in the early routine post-fibrinolysis angioplasty group compared with 12% in the primary angioplasty group, (relative risk, 0.80; 95% confidence interval, .37–1.74).”

In table 4, you see that the primary PCI group had 6 deaths (5.6%) and the early fibrinolysis group had three deaths (2.9%). Based on this we calculated a number-needed-to-treat of 37 (taking a 5.6% in the PCI-alone group, minus 2.9% in the lytics+PCI group = 2.7% absolute risk reduction) . That’s not a bad NNT, considering what’s at stake and the availability of lytics. But can we put faith in these results?

Our biggest problem with the GRACIA-2 data is in Table 1 — it seems the PCI-only patients were sicker. They were much more likely to have three-vessel disease (15 vs. 3 for the lytics+PCI group), degree of stenosis was greater, and number of non-stenotic vessels was less. The authors try to write this off as an effect of the lytics, but we were skeptical: I’d trust tenecteplase to relieve a vessel occluded by thrombus, but not to widen every stenotic coronary. Rather, we thought the degree of sick hearts in the PCI arm was a failure of randomization, either by fluke or by nonblinded decisionmaking from the caregivers. We found ourselves wishing patients had undergone CT-angio to better ensure equality.

Also, when you compare this data to the ASSENT trial, it makes less sense. The ASSENT trial took patients from 24 countries in an attempt to see if patients with large MIs did better with “facilitated PCI” than with delayed PCI. They wanted 4000 patients, but had to stop early because of higher in-hospital mortality in the group receiving PCI and tenecteplase (6% vs. 3% in PCI alone). The facilitated group had more strokes, more reinfarction, and more need for revascularization. This was puzzling, as the infarct patency was improved with fibrinolysis before angioplasty (similar to GRACIA data).

The ASSENT authors speculate the bad outcomes with lytics + PCI are due to suboptimal antithrombotic cotherapy, such as withholding heparin — but the authors were concerned about intracranial bleeds (which turned out to be a nonissue). They also point to treatment delays in the lytics+PCI group, compared to PCI alone.

So, with the disturbing ASSENT-4 data and the inapplicability of GRACIA to our patient population (for instance, the Iberian diet, while delicious, is quite different than ours) we cannot apply these findings to our practice. As much as we’d like to give lytics before PCI, the data at this time doesn’t support it.

Of course, ACEP provides guidelines on fibrinolytic therapy in acute MI, and this summer the AHA came out with new recommendations for unstable angina / NSTEMI. And, as always, EMcrit.org has useful references, including the 2004 STEMI treatment guidelines and a section on PCI vs. lytics (scroll down).

For the final journal club of the year, Ravi picked an article from Goldstein (and others)  entitled, “A Randomized Controlled Trial of Multi-Sliced Coronary Computed Tomography for Evaluation of Acute Chest Pain” (available from the Journal of the American College of Cardiology, 2007 Vol 49, No. 8, pp863-871).

It’s not the definitive, practice-changing study we’re waiting for, but it draws upon previous work (such as JAMA 2005; 293:2471-8) estabilishing multi-slice CT coronary angiography as effective in diagnosing occlusion, and attempts to compare the safety, efficacy and efficiency of MSCT against established protocols.

Specifically, the authors looked at safety (in terms of major adverse cardiac events — MACE — at 6 months), diagnostic efficacy (difined as correctly diagnosing, or excluding, CAD as the cause of chest pain, without repeat testing or MACE within 6 months), and efficiency (a combination of time-to-diagnosis and cost during first ED visit).  We must note that the trial received funding from Minestrelli Advanced Cardiac Research Imaging (sometimes spelled Ministrelli).

They enrolled 203 “low risk” patients with acute chest pain (older than 25 years, symptoms of angina or equivalent within 12 hours, and low risk of infarction according to criteria from Goldman and others). Patients were excluded if they had known CAD, EKG changes, elevated enzymes, or cardiomyopathy with low EF, arrhythmia, history of contrast reactions or beta-blocker allergy, morbid obesity, renal insufficiency, or recent contrast CT.

If the patients’ 0-hr and 4-hr EKG and serum biomarkers were normal, they were randomized into the MSCT or the standard-of-care group. Enrollment was 24 hrs a day, but MSCT and SPECT was only available 7a-6p, so the cost of waiting was also figured into their results.

The 99 patients in the MSCT arm got beta-blockade (if they weren’t already on it) of atenonol 50-100mg +/- metoprolol 5-30mg to get their HR below 65 (though greater HR was not a contraindication to the study). Everyone also got a sublingual nitrogen a minute before the CT. The coronary lesions seen on CT were clasified on a 5 point scale, with 0 = no stenosis and 5 = total occlusion.

These patients in the MSCT arm would go home if their coronary scores were 1 or lower (or if their calcium score was under 100 Agatstons), or they’d get invasive angiography (catheterization) if their scores were 4 or higher (>70% occlusion). In-between, or non-diagnostic scans would get a nuclear stress test.

The 98 patients in the Standard of Care (SOC) arm received serial EKG’s and biomarker studies at 4h and 8hr after presentation. Troponins greater than 1.5 ng/mL, CK-MB greater than 5 ng/mL, or myoglobins over 98 ng/mL were considered abnormal. Patients received same-day SPECT (single photon emission CT) myocardial perfusion stress-tests or nuclear stress tests.

These SOC patients would go home if their EKG’s, biomarkers, and stress tests were normal. They’d get cathed if they had any abnormalities in EKG, enzymes, or stress test. (The final decision to cath, it should be noted, rested in the PMD’s hands).

It’s helpful to keep some concepts in mind: this study really just compared ways of diagnosing coronary artery disease, specifically,  CT vs. stress testing. There was no straight-up comparison to gold standard of PCI. And the enzymes/EKG portion was just a warmup to rule-out MI and stratify out the high-risk group, who would go straight to PCI anyway.

In terms of safety, well, there were no MACE within 6 months of the 88 MSCT patients (out of 99 total) that were discharged home from the ED (this is 67 patients with no or minimal coronary occlusion, plus 21 patients with intermediate or non-diagnostic MSCT who had negative stress tests). Similarly, there was no 6-month adverse events in the 95 (of the 98) SOC patients who went home.

In terms of diagnostic efficacy, MSCT alone was not adequate for diagnosis is 24 of 99 cases. Twelve of 99 underwent catheterization — eight of these were true positives, three were false positives, and one was a true negative. As the authors state, “96 of 99 patients were judged to have a clinically correct diagnosis.” Compare this to the SOC patients — only 7 of which (out of 98) underwent catheterization (one was a true positive, two were false positives, and four were false negatives). In total, “96 of the 98 SOC patients were judged to have a clinically correct diagnosis.” This stands as a pretty remarkable validation of standard-of-care risk stratification, based on the Goldman criteria.

Finally, in terms of efficiency, the time-to-diagnosis was found to be faster with MSCT (3.4 h vs. 15 hours) even though nuclear stress and MSCT were only available during daylight hours. Because of this reduced length-of-stay, the cost of care for patients in the MSCT arm was less ($1586 vs $1872 for SOC).

We noted that, while no negative outcomes in 6 months of followup is nice, it really suggests a larger trial is needed. Especially because we’re already starting with low-risk patients. Another way of knowing you need a bigger trial is calculating the positive and negative likelihood ratios of CAD with findings on MSCT. The positive LR was 4 (meaning, positive findings on the MSCT essentially quadrupled the odds the patient had significant occlusion), but the negative LR was 0… which just can’t be true. A larger sample is needed.

In the same issue, this journal also published editorial comments from de Fuyter and van Pelt. The editorial acknowledges that the real power of coronary CT might be in calcium scoring and total plaque burden calculation, but for now, we’re left wondering the value of the new tools we do have, especially in comparison to standbys like stress testing and PCI. The authors helpfully put some numbers on a concept we brought up in discussion: how much radiation are we talking about? (because, let’s not forget, some unfortunate patients had to undergo MSCT, nuclear stress, then PCI):

The editorialists also noted, wisely, that, ”A comprehensive evaluation of thoracic structures for noncardiac disease such as pneumonia, pleural calcification, and hiatal hernia might add to the usefulness of MSCT as a diagnostic tool when compared with other modalities such as nuclear and dobutamine stress testing.”

The ACEP clinical policy group just weighed in on NSTE ACS – but sadly does not mention the role of coronary CT in evaluating chest pain. Fortunately, EMCrit.org has a spectacular page on ACS — including likelihood ratios of physical exam findings, many important risk-stratification schemes, pathways for cases of renal insufficiency and cocaine (among others)… but this site, too, currently lacks guidance on the role of coronary CT. But because CT can visualize vessel walls and other sources of chest discomfort, we might someday look to CT as superior to PCI.

Thanks to all who stopped by our Innovations inEM Education booth at SAEM these past few days. Your input and encouragement were appreciated! The abstract is now online and an MS-Word version of our informational pamphlet is available for download.  

 For those of you who wanted to discuss things further, my email is Nicholas.Genes /at/ mssm.edu.

Other EM journal club websites with analyses of articles:

• Washington University (archives)
• University of Michigan (archives)
• Emory (archives)
• East Virginia Medical School

If you know of more, please share! The idea of putting EM Journal Clubs online, for prompt bedside access, goes back a few years…

At our last Journal Club, Tom presented a 2005 paper from JEM on new CT scanners in the evaluation of SAH. The paper was called Subarachnoid Hemorrhage Diagnosis By Computed Tomography and Lumbar Puncture: Are Fifth Generation CT Scanners Better at Identifying SAH? by Boesiger and Shiber, and it appears in Journal of Emergency Medicine (2005: Vol. 29, No. 1 pp23-27).

The article is motivated by the fact that 1% of headache patients in the ED have SAH. Most are traumatic, but those that aren’t are usually from Circle-of-Willis aneurysm ruptures, which often kill or disable otherwise healthy people. EM physicians hate that sort of unsettling risk, and the situation is further complicated by the 20-50% of SAHers who present with a sentinel bleed. So there’s a real opportunity to help some potentially moribund patients —  but if you ask most interns, they’ll say they’re shoving too many needles into the backs of people who probably just needed some exedrin.

Maybe we can change our practice, based on recent upgrades in CT scanner technology. These authors were the first to look at the new scanners with an eye toward sensitivity in SAH diagnosis. More below…

We were all curious about this ‘fifth generation’ — it turns out if you google “fifth generation CT” you get a wikipedia entry and … this very paper! The wikipedia article is lacking in citation but states:

Although numbered sequentially, the 3rd and 4th generation designs developed at approximately the same time. The concept of electron beam CT, which some authors have called 5th generation, followed later. Some authors have described up to 7 generations of CT design. However, it is only generations one to four that are widely, and consistently, recognised.

The authors used GE Lightspeed 2.X scanners, which don’t use electron beams, as far as I can tell by wading through old press releases (the LightSpeed is different from the GE eSpeed, which does use EBT). Perhaps it’s just easier to say they’re using a multidetector CT, instead of parotting this confusing ‘fifth-generation’ terminology.

The authors retrospectively looked at patient reconrds from 2002 at a rural Level-1 Trauma Center ED (a prospective study would’ve been a tough sell to the IRB). If anyone in their ED with a headache went on to have a CT and LP to evaluate for SAH, they were included in this study (except for those under age 17, and those with recent trauma or neurosurgery). The records needed to mention something about ruling out SAH to be included. Headache did not need to be the chief complaint, but needed to be a reported symptom.

They didn’t exclude anyone with anemia, which might’ve caused some false negatives (if blood is not brighter than brain, SAH is harder to see). And they didn’t record the patient’s report of symptom onset, which would’ve been interesting.

The CTs had 5-mm cuts through the cerebrum and posterior fossa, and were read by attending radiologists. Why just 5mm slices, when they could cut thinner? Well, as it turns out, 5 mm was sensitive enough, and maybe cutting thinner would’ve caused more false positives.

Their gold standard, LP, was positive if tube 1 had at least 400 RBCs, and the RBCs did not clear by tenfold in Tube 4. LPs could also be considered positive if xanthochromia was visually observed. SAH was further evaluated by CT angiography, or telephone followup.

Of the 569 LP’s they performed in 2002, 177 were listed as part of an evaluation for SAH. Of those 177, 159 had negative CTs and negative LPs. Eleven CTs were negative but had positive LP. Six CT scans in 2002 were positive for SAH — one LP was done in this group and this was also positive. And, notably, there was one positive CT that had a negative LP (it was read as questionable for intraventricular blood, and the patient had no further complications, so this was the one false positive they noted.)

Looking through their ED database, they saw 31 SAH in 2002 — 25 were traumatic (81%, but they say it’s 77%), and all traumatic and nontraumatic SAHs were seen on CT.

None of the 170 patients who had a negative CT scan for SAH was found to have one by LP or telephone followup. Of the 11 patients with negative CT but positive LP, none had further headache complications on phone followup, and none had xanthochromia, so these were called ‘LP false positives’.

From this data, they calculated a sensitivity of CT for SAH of 100%, which 95% confidence interval ranging from 61.0-100%. And the specificity of CT for SAH was 99.4% (95%CI was 96.8-99.9%).

The authors don’t calculate likelihood ratios, which we use to augment our pretest probability assignments for various diagnoses. If they did calculate LR, well, a positive CT scan would have a huge LR — even if you took the low end of the confidence interval, LR would be 19 — overwhelmingly likely to make your diagnosis.

However, if you use the low end of their confidence intervals, a negative CT scan only has a negative LR of 0.4, which is not good enough to budge your pretest probability into a diagnosis of SAH.

Really, this size of this study was small, and it was underpowered to provide meaningful values for the sensitivity and specificity. A larger trial, involving rural and nonrural centers, might produce a confidence interval narrow enough to give a meaningful likelihood ratio.  

An article in Postgrad Medical Journal (2005;8:470-473) cites the familiar statistic, that CT is 98% sensitive in diagnosing SAH within 12 hours, 95% sensitive within 24 hours, and dips to 75% sensitivity if the SAH is between 24-72 hours old. (That data seems to come from a paper by Adams, in Neurology, 1983: Vol. 33, pp981-988). LP, on the other hand, doesn’t hit 100% sensitivity until after 12 hours – so the combo of CT/LP has been a powerful tool for emergency physicians. It will stay that way, for now — it’s far too soon to consider abandoning LP.

More practice resources, from the 2002 ACEP guidelines on the approach to headache:

EMCrit.org cites a variety of sources in the EM and neurology literature to provide some useful, evidence-based recommendations in evaluating SAH and the results of your LP.

Lynn was telling me about a case she saw this year – if I recall, it was a young man who developed uncontrollable hiccuping after an inguinal hernia repair. I don’t know what happened to the guy, but I saw my first hiccuping patient shortly thereafter. As I surf the web, I keep coming across remedies for this unusual but vexing complaint:

• Methylcellulose (to expand the stomach) 
• NG Tube (to decompress the stomach)
• An implantable vagal nerve stimulator
• Pressing on the ears while drinking through a straw
• Digital rectal massage (also reported in Annals of EM)
• Chlorpromazine 25-50mg IV (the only FDA-approved agent for intractable hiccups)

Our anesthesia colleagues have looked at this issue, as hiccuping in the OR is a pressing concern. Unfortunately, a systematic review (Kranke, Eur J Anaesthesiol 2003 Mar;20(3):239-44) turned up lots of anecdotes, but only one (inconclusive) RCT:

A large variety of interventions have been proposed for the treatment of hiccup during anaesthesia and sedation. However, perioperative treatment is still based on empirical findings and no treatment is ‘evidence-based’. Thus, no valid recommendations for the treatment of hiccup can be derived. Uncontrolled observations are inadequate to establish treatment efficacy.

More drug suggestions below, along with some background on hiccups…

 EMedicine has an exhaustive list of drug therapies:

Chlorpromazine is the most studied and appears to be the drug of choice. Increments of 25-50 mg IV/IM are effective in 80% of cases. To avoid or minimize hypotension from the agent, it is advisable to preload the patient with 500-1000 mL of IV fluid.

Another major tranquilizer, haloperidol, is effective in doses of 2-5 mg. Metoclopramide has been used successfully in a dose of 10 mg every 8 hours.

Several anticonvulsant agents have been used to treat intractable hiccups. Phenytoin, valproic acid, and carbamazepine have been effective when used in typical anticonvulsant doses. More recently, gabapentin has been shown to be effective where CNS lesions are present.

Of the anesthetic agents, ketamine has been the most successful at a dose of 0.4 mg/kg (one fifth of the usual anesthetic dose). Baclofen, a centrally acting muscle relaxant, administered at 10 mg 4 times a day orally, particularly is useful in patients for whom other agents are contraindicated (eg, those with renal impairment). IV lidocaine in a loading dose of 1 mg/kg, followed by an infusion of 2 mg/min, has cured patients after other agents were unsuccessful.

Other agents reported to be beneficial include muscle relaxants, sedatives, analgesics (eg, orphenadrine, amitriptyline, chloral hydrate, morphine), stimulants (eg, ephedrine, methylphenidate, amphetamine, nikethamide) and a miscellaneous group including edrophonium, dexamethasone, amantadine, and nifedipine. Benzodiazepines have been shown to exacerbate or precipitate hiccups and should be avoided. 

 Also called hiccoughs, and medically termed “singultus”, the pathophysiology of hiccups is still poorly understood.

A hiccup is a sudden, involuntary spasmodic contraction of the diaphragm and external intercostal muscles, that results in inspiration which abruptly ends with closure of the glottis.¹ Although the physiological role remains unknown, hiccups are usually short lived, uncomplicated, and affect healthy subjects on occasion. Rarely, hiccups can become persistent, intractable and even refractory to a variety of therapeutic modalities.

The hiccup reflex arc is comprised of afferent pathwaysvagal, phrenic, and sympathetic (T6-T12) branches. The efferent pathways are composed of the phrenic nerve to the diaphragm and nerves to the glottis and the external intercostal muscles. The central connection is the spinal cord between segments C3 and C5, possibly controlled by supraspinal pathways.

Here’s an evolutionary theory of hiccups:

Hiccoughs are characterized by glottal closure during inspiration and by early development in relation to lung ventilation. They are inhibited when the concentration of inhaled CO2 is increased and they can be abolished by the drug baclofen (an agonist of the GABA-B receptor). These properties are shared by ventilatory motor patterns of lower vertebrates, leading to the hypothesis that hiccough is the expression of archaic motor patterns and particularly the motor pattern of gill ventilation in bimodal breathers such as most frogs. A circuit that can generate hiccoughs may persist in mammals because it has permitted the development of pattern generators for other useful functions of the pharynx and chest wall muscles, such as suckling or eupneic breathing.

Jack tackled a controversial topic at this month’s Journal Club — what’s the evidence for giving antibiotics within four hours for CAP patients? It’s a good question, because how well we perform at this task is a big part of how our hospitals are measured. Ineed, pneumonia antibiotic timing is one of JCAHO’s Core Measures and there are only more such metrics down the road — so we’d like to think that our funding depends on rock-solid science and proven benefits.

Well…

As Jack noted, the 4-hr policy is based primarily on four papers,

1) Kahn et al, JAMA 1990 Oct 17 264(15) 1969-73 with comments 1995-6

2) McGarvey et al, Quality Review Bulletin April 13(4) 124-30

3) Meehan, Houck et al.  JAMA 1997 Dec 17 278(23) 2080-4 

4) Houck et al, Arch Intern Med. 2004; 164(6):637-644.

It’s this last paper we’re going to discuss — a retrospective study derived from a national sample of medicare patients with pneumonia.  

Full disclosure — the lead author, Peter Houck, works for the Oklahmona Foundation for Medical Quality. Sure, they’re not-for-profit, but there’s an inherent conflict-of-interest. It’s kind of like how JCAHO reviews hospital performance, but also has a consulting arm that avises hospitals on how to pass their review, for a fee. The line between “the patient’s best interest” and “the regulatory agency’s best interest” becomes blurry, when someone from a quality-assurance agency is researching measurable hospital policies.

Since the ‘therapy’ being tested here is timing to administration of antibiotics, this study can’t ethically be done prospectively. Their retrospective sample is worth noting, however: They took their numbers from a database of Medicare (over 65) with a primary or secondary ICD-9 code of pneumonia sampled from July 1- Dec 31, 1998, or Sept 1, 1998-March 31, 1999 (this database had 346,105 patients, though it seems possible they sampled the same patients twice). The authors randomly selected up to 850 cases from each state (more like the Senate than the House of Representatives — Wyoming’s handling of pneumonia could be weighed equally with California’s), giving a study set of 39,242 patients .

Exclusion criteria included death or discharge on day of admission, age under 65, lack of radiographic evidence of pneumonia, history of immunocompromised status, recent hospitalization, or lack of ABx timing documentation. This eliminated over half their set, leaving 18,209 patients from 3732 hospitals (avg = 3/hospital, but one hospital accounted for 122 patients in the sample).

Their outcomes were in-house mortality, mortality within 30 days of admission, LOS in house, and readmission within 30 days of discharge. Jack recommends we start assessing the data with Table 4, which reports 30-day mortality, divided up by who got ABx when. So patients that got antibiotics within one hour, compared to those that got ABx after one hour, showed no significant difference in mortality. Ditto with two hours. But those patients that got ABx within 3 hours had a mortality of 11.7% (95%CI 11.0-12.6) compared with 12.3% (95%CI 11.5-13.0) for patients getting antibiotics after 3 hours, giving an adjusted odds ratio of 0.88, which was significantly different. Three hours! So why isn’t that the rule? Because, ah, four hour cutoffs are “commonly used in quality improvement activities.” OK…

 The benefit of ABx before X hours, compared to after X hours, persists for values of X from 3 to 9 (if you don’t get antibiotics before 10 hours, it makes no significant difference in mortality if you get them after 10 hours.

Since the authors decided to focus on the 4 hour cutoff, they generated a lot of data around that timepoint. Their patient characteristics (Table 1) shows sufficient randomization in many categories, though it looks like women, minorities, and those over 85 were more likely to get ABx after 4 hours, rather than before. In Table 2, we see that city hospitals were less likely to get ABx administered before four hours, compared to nonmetropolitan hospitals. The same slowness was seen true in for-profit hospitals, as well as teaching hospitals. Make of that what you will.

Table 5 shows the breakdown of all outcomes, depending on whether ABx were given before or after 4 hrs. Length of stay, in-house mortality, and 30 day mortality are all significantly lower in the “within 4 hr” group (30-day readmission is unaffected). Looking closer at 30-day mortality, we see that 12.7% of patients die with antibiotics after 4hr, compared with 11.6% who die with antibiotics before (confidence intervals do not intersect). That 1.1% difference means the NNT is 91  (if we give 91 patients ABx before 4 hrs, we’ll save one more life). Of the 600,000 medicare patients admitted for pneumonia each year, that translates to 6600 lives saved.

Also striking about Table 5– there’s a huge mortality difference between PSI (pneumonia severity index) class IV and V, compared to class II and III… but all patients who got ABx within 4 hours fared better than those who did not, in terms of in-hospital mortality, 30-day mortality, and shorter LOS.

So, early administration is obviously a good thing, right? Well, no – it turns out that patients who got antibiotics before two hours actually did worse (see Table 3)  in terms of higher in-hospital mortality (7.4%, CI 6.6-8.3)  compared to the 2-4hr group (6.3%, CI 5.6-7.0), and 30 day mortality (12.5% for the under 2 hr group, compared to 10.9% for the 2-4 hr group).  The authors are perplexed by this — it doesn’t fit with their model that pneumonia’s progressive injury to the lung can be interrupted by antibiotics. They recommend further study (because recommending to withhold antibiotics until two hours had elapsed might be controversial).

A few more points about all the ‘benefits’ of early antibiotics  – JCAHO’s elevation of the 4 hr rule as a proxy for quality has encouraged hospitals to agressively, but superficially, treat coughing patients that walks into the ED. It’s hard to study whether this rule is contributing to antibiotic resistance, or whether rushing to give PO biaxin is really helping anyone, or whether the community benefits when we let pneumonia patients jump the triage queue at the expense of possibly sicker patients.

For further reading, emcrit.org has good resources on CAP, the blood culture controversy, and citations for optimal coverage for all of pneumonia’s different flavors.

Also, Jack previewed the new guidelines from the Infections Disease Society of America / American Thoracic Society, reprinted from Clinical Infectious Disease 2007: 44:S27-72:  (I love how the first author is, in fact, based in Canada.)

29. For patients admitted through the ED, the first antibiotic dose should be administered while still in the ED. (Moderate recommendation; level III evidence.)

Surprisingly mild!! The authors elaborate:

Blockbuster stuff! Let’s see if JCAHO modifies their recommendations in light of this authoritative, evidence-based recommendation.

When I was preparing an M+M last fall, I came across this notable study called Management of Acute Undifferentiated Agitation in the ED: A Randomized Double-Blind Trial of Droperidol, Ziprasidone, and Midazolam . It’s by Martel et al (including Michelle Biros, who’s editor of Academic Emergency Medicine), and appeared in Academic Emergency Medicine in December 2005 (Vol 12, No 12, pp1167 – coincidentally, right after Dr. Richardson’s EMPATH study).

It’s a good study that we might otherwise overlook, because it came out on the eve of the ACEP guidelines for agitation management and thus, wasn’t included in that extensive lit review.

We all know how important it is to manage agitation — for our safety, and also from a diagnostic perspective to figure out what’s really wrong with the patient. If you’ll recall from my talk, interpreting the agitation management studies isn’t easy — there are different scoring scales, different modes of delivery. And, of course, different practice environments. Some studies exclude alcohol agitation, others populations are dominated by alcoholics. Some studies seem applicable to us, until you find out it’s a study of consenting psych ward patients.

So these authors tried to keep it straightforward to better aid practice — they looked at a convenience sample of adults presenting to a large urban ED with undifferentiated agitation — could be drugs, psych, anything. A “public consent” was used — local facilities and shelters were notified of the study, and therefore individuals could be enrolled in the study, exempt from informed consent (proxies were used when available).

After being selected by an ED doc as having “undifferentiated agitation, the patients (assuming they were not pregnant, not incarcerated, and not allergic) were randomized to receive either 5 mg of midazolam (versed) IM, 5 mg of inapsine(droperidol) IM, or 20 mg of ziprasidone (geodon) IM.

College and medical students then scored the patients on the Altered Mental Status Scale (AMS – a modified version of BARS) at the time of med administration and frequently for two hours. Pulse ox and end-tidal CO2 were also monitored, which I believe is a first for agitation sedation studies.

They enrolled about 50 patients in each arm of the study. Though most characteristics were randomized, the midazolam arm ended up with more head-injured patients. It’s important to note that 90% of their patients were eventually shown to have alcohol on board.

What did they find? Midazolam and droperidol are fastest — adequate sedation (an AMS score less than or equal to 0) was achieved at 15 minutes after the IM dose, compared to 30 minutes for those receiving ziprasidone.

Specifically, 33 of the 48 midazolam patients were sedated at 15 minutes, and 30 of the 50 droperidol patients were sedated. Only 18 of the 46 zirprasidone (geodon) patients were sedated at 15 minutes, though by 30 minutes all groups were statistically similar, and by 45 minutes, the midazolam group was starting to get agitated again (14 of 48 were agitated, compared to droperidol’s 9 of 50, and ziprasidone’s 9 of 46.

Time to discharge / dispo was not affected by drug choice — it was always around 6-7 hours. The droperidol patients needed the least amount of “rescue” sedation — only five patients, for a total of six doses. For geodon, nine patients needed 11 doses, and significantly, 24 midazolam patients — half of them — needed more sedation (30 doses altogether).

A few patients (six altogether) got akathesia. Interestingly, four of those patients were geodon recipients — I thought that was a selling point of these new antipsychotics.

The respiratory issue was of particular concern to the authors, and yet they didn’t really report anything specific about end-tidal CO2 results. About half of all patients had some respiratory depression, no significant differences between arms. Twenty of the 48 midazolam (versed) patients needed O2, but that was not significantly worse than the other groups. No one got intubated. No one had dysrhythmias.

So, I liked this study because they took pains to make it applicable to practice — they didn’t try to separate out the causes of agitation. Nor did they lose patients by trying to obtain individual informed consent. Unfortunately, the vast majority of their study group had alcohol on board (although that may better simulate the EHC experience). As one critic noted, if the undifferentiated agitation was due to cocaine or alcohol withdrawal, you give benzos and don’t want to waste time with other drugs. Furethermore,

The effectiveness of sedation in this study was defined as an AMS score of less than 0; however, we think that the most important measures of effective sedation are the time to establishment of an intravenous line, as well as determination of a fingerstick glucose level and a rectal temperature, thus allowing for administration of medications and identifying rapidly reversible causes of AUA.

It’s a good point, but if you interpret this literally you end up physically restraining everyone until you get a diagnosis, and that puts the staff and patient at risk. And how often do we really need stat labs and tox on our agitation patients? As always, it depends on the center — a place like EHC gets a  lot of drunks, but a patient at MSH might be more likely to have drug interactions causing agitation. Use your judgment.

Another commenter noted that the FDA black-box warning on droperidol mandates EKGs and 2-3 hours of tele monitoring. However, Martel – and the ACEP guidelines — note that most everyone has had a good and safe experience with droperidol.

This study by itself isn’t enough to change practice, but taken together with the ACEP guidelines it makes a pretty strong case for the safety and efficacy of benzos in general and midazolam in particular. Just be ready to give more when it wears off. I’ll excerpt some from page 6 of the ACEP recommendations, which make much of a Class II RCT by Nobay (see the guidelines for the reference) comparing ativan, versed and haldol :

The authors compared IM midazolam (5 mg) to IM lorazepam (2 mg) or IM haloperidol (5 mg). Midazolam had a significantly shorter time to sedation than did lorazepam or haloperidol. The mean time to sedation was 18.3 minutes for midazolam, 28.3 minutes for haloperidol, and 32.2 minutes for lorazepam. The time to arousal (81.9 minutes) in patients given midazolam was also significantly shorter than that of the other therapies.

In addition, several class III studies found midazolam (2.5 to 3 mg IM) to be efficacious in reducing agitation. It produced rapid sedation, within 6 to 8 minutes, in a small series of acutely agitated patients. Midazolam was significantly better than haloperidol in controlling motor agitation in a small study of schizophrenic patients. A large series reported by the TREC Collaborative group found that midazolam (15 mg) was superior to haloperidol (5 mg) plus promethazine (50 mg) in producing rapid sedation at 20- and 40- minute endpoints.

Again, take care to note the different agitation scoring methods, different routes of admin, and the different patient populations. It’s explained well on the ACEP review. And I should note, in summary, ACEP’s final recommendations — all class B — are for benzos (midazolam or lorazepam) or typical antipsycholtics (haldol or droperidol) for effective monotherapy of an undifferentiated agitation. Atypical antipsychotics should be reserved for agitated patients with known psych history — the case against them in undifferentiated agitation has been eloquently stated. EMcrit.org has some other studies cited as well, plus some practice notes for sedating violent patients.

This week in journal club, Matt reviewed a nice little trial submitted by a group of Texans to the Canadian Journal of Emergency Medicine. They studied IV dexamethasone in preventing benign headache recurrence (Can J Emerg Med 2006;8(6):393-400, PDF) – something I had never tried, but apparently has been bouncing around the neurology and EM literature for 20 years.

It turns out that migraines may not be simply a vascular disorder, but rather an inflammatory disease. And, as Matt pointed out, it’s very difficult to diagnose migraines; it might be simpler for us ED folk to say headaches exist on a continuum between tension and migraine, and maybe ED patients with primary headache would benefit from a steroid.

The authors had two aims: to see if patients discharged from the ED with benign headache had less recurrence if they received dex, and to see if dexamethasone could reduce the need for future doctor’s visits and allow a return to normal daily activity. To answer these questions, they designed a randomized, double-blind, placebo-controlled trial of adults presenting to the ED with benign headache (as diagnosed by the ED doc). Exclusions were reasonable: pregnancy, fever, stiff neck, focal neuro deficit, or cause to worry about immunosupression (active PUD, DM I, systemic fungal infection, or already taking steroids).

Of note, the dexamethasone was given IV, and the IV could not be placed just for the purposes of enrolling into the study. So those presenting with headaches had to be sufficiently worrisome so as to warrant IV placement, either for labs or fluids or other meds. Also of note - this was a convenience sample, and it took them ten months to enroll just 57 patients – it’s unclear whether they just didn’t get that many headaches requiring IV, or if the authors were seeking a higher level of convenience than I’m used to seeing.

The patients got 10 mg IV dex before discharge, rated their pain, and were followed up between 48 and 72 hours by phone. They thought they’d need 94 patients to achieve a clinically significant decrease in recurrence, but they achieved that early and so, unfortunately, stopped the study before the secondary goal of assessing quality-of-life was completely answered.

What did they find? Just 9.7% of patients who received dexamethasone experienced headache recurrence, whereas 58.3% of placebo patients had recurrence (a number consistent with previous studies of headache recurrence). Pretty awesome, right? We calculated the NNT (number needed to treat) for dex at around two, which is incredible. But quizically, when the information was collected in a different way, 12.9% of dexamethasone patients reported either headache severe enough to require a physician visit, or precluding a return to normal life (compared to 33.3% of placebo patients). The vast majority of dexamethasone patients (87.1%) needed no further treatment, or only OTC treatment – compared to 66.7% of placebo patients. Why don’t these “secondary goal” numbers match up with the primary goal data? It’s unclear, and tarnished the otherwise compelling results.

Who were these patients, anyway? Well, both placebo and dexamethasone patients were predominantly female, in their early thirties, and just under half reported a history of migraines). Both groups had about a 13% LP rate. For a randomized group, the dexamethasone patients seemed luckier – compared to controls, they got more of every class of pain medication (more NSAIDs, more opioids, more antiemetics, and more tylenol, caffeine, etc). Their initial pain rating (VAS) and pain at discharge were comparable to the placebos.

Side effects were negligible in this tiny study – with the placebo group reporting nausea, cramps, and auditory hallicuinations, and the dex group mostly complaining of tingling. The feared reaction of disseminated strongyloides did not materialize from this one-time dex dose.

There are a number of little things in this study that might seem like drawbacks, but probably aren’t – such as the small size of the study, and the use of a young, generally healthy population. And the lack of distinction between migraine and other headaches, and the lack of standardizing treatment – all of these things better reflect practice, and I think make the results more impressive. But it’s not clear why the investigators limited themselves to one dose of IV dex when IM would be just as easy to randomize and might have allowed more enrollment.

Still, many of us were impressed by this well-executed study of a simple intervention to a common complaint. It shows that RCTs don’t need to be multimillion dollar projects, and can still generate findings that can change our practice. I think of this study like flomax for nephrolithiasis  –  it’s using a common medication for a common complaint in a new way. The positive results let us feel like we’re actually doing something novel and effective for these otherwise hurting and often hard-to-satisfy patients. So, along with encouraging some caffeine consumption, many of us at journal club were looking forward to administering dexamethasone to our future benign headache patients.

More resources: ACEP policy on headaches (focus on diagnosis, neuroimaging and LP), and EMCrit.org has a great page on headache differential and workup.

In the growing backlog of articles and journal club presentations I’d like to write up, I came across this publication which was presented a few months back — Does End-tidal Carbon Dioxide Monitoring Detect Respiratory Events Prior to Current Sedation Monitoring Practices? from Burton, Harrah, Germann and Dillon in Academic Emergency Medicine 2006; 13:500-504. My notes on the presentation have long since disappeared (I believe it was given by… Tim?) but my interest in the topic was rekindled after a recent M+M.

 The authors note that, in the OR, end-tidal monitoring is standard practice, but no EM organizations have advocated its routine use in ED procedural sedation. Thus, the goal of this study was to see if ETCO2 monitoring could detect acute respiratory events before current methods.  They used a prospective observational case series with a convenience sample of ED patients undergoing PSA at a 50,000-pt/year ED among consenting patients when investigators were present. Controls got SpO2, RR, HR and BP monitoring, and an investigator not involved with pt care “assured” the quality of waveform data for SpO2 and ETCO2, as well as proper positioning of monitoring equipment, and was the only person present who saw the ETCO2 waveform and numerical CO2 display. Measurements were recorded every 30 seconds, as well as during med administration, observed acute respiratory events, interventions, and return to baseline.

What was a respiratory event? The authors define it as a desat to 92% or less, an increase above 2LNC in administered O2, hypoventilation, or use of bag-valve mask or ventilatory assistance, use of physical or verbal stimulation, or administering reversal meds. Changes in ETCO2 of more than 10 mmHg from baseline, or an intrasedation ETCO2 of less than 30 mmHg or more than 50 mmHg, were considered investigational acute respiratory events.

They planned to enroll 250 patients but, in dramatic fashion, had to stop the study early, after just 59 patients (60 sedations). Abnormal ETCO2 findings (as described above) were found in 36 of the 60 encounters — 32 low and five high (yeah, one guy had both low and high).

Of course, 44% of these “abnormal” ETCO2 readings were not associated with any adverse respiratory event or intervention. Some were just single, abnormal blips in recordings. But the investigators did log 20 acute respiratory events in their 60 encounters (33%) and 17 of them (85%) were associated with abnormal ETCO2 recordings. Stay with me here — of these 17 encounters, 14 (70%) had the abnormal ETCO2 recording before changes in SpO2 or RR.

So, what to make of this? Well, just browse through the tables and see what kind of events we’re talking about. First, 18 of the 20 acute respiratory events occured on propofol, which we can’t use in the ED. Ketamine and Etomidate account for the others, while midazolam was event-free (but they only used it three times). Only four of their twenty acute resp events required bag-valve mask ventilation, and two of those had no warning (ie, the ETCO2 didn’t help predict the event).  Most of these abnormalities were low ETCO2, it only went high once (and you’d think low RR and CO2 retention would be a more common finding — as seen in previous studies). No one was hyperventilating, either.

In addition to not providing adequate warning in some of their worst acute respiratory events, ETCO2 provided warning 3-4 minutes ahead of some relatively minor events (requiring just some verbal stimulation). Is that really valuable? And don’t forget the fact that 44% of ETCO2 abnormalities were not accompanied by clinically observed respiratory events.

So, while ETCO2 monitoring is certainly becoming more widespread during sedation, I don’t think its advocates can really point to this study as demonstranting its relevance. I think you could argue that this study by Miner et al from Academic Emergency Medicine 2002 Apr;9(4):275-80 made the case for ETCO2 better. But in discussing that Miner study, and others,  the 2004 ACEP policy guidelines for procedural sedation have this to say about capnometry:

 If you’re curious about monitoring sedated patients who aren’t undergoing procedural sedation (beause, say, they were agitated) I’ve only found one study that addressed this — it’s from Martel, Miner and Biros and was also printed in Academic Emergency Medicine 2005 Dec;12(12):1167-72. Unfortunately it wasn’t included in the recent ACEP agitation guidelines.

Emcrit.org is uncharacteristicly quiet on the issue of ETCO2 in procedural sedation, concluding it’s safer and summarizing the above paper in one line. But if you’re looking for a hands on guide to the different sedation agents we can give, their dosing and drawbacks, it’s a great source.