QUICK HIT ARTICLE: News Flash Breathing is good for you…next up Water, Wet! “How breathing can help you make better decisions…”

De Couck et al. International Journal of Psychophysiology 139 (2019) 1–9. doi: 10.1016/j.ijpsycho.2019.02.011

In this experimental study of 56 undergrad and grad students in France, utilizing specialized breathing patterns (see image below), the researchers sought to determine the effect of breathing on stressful decision making. They found that using breath patterns for 5 minutes prior to stressful written multiple-choice tests allowed the experimental group to answer an additional question correctly. Additionally, they found that the breathing group felt significantly less perceived stress via a visual analog scale difference of >4 than the sham control group.  Does this pertain to clincal scenarios where we think out loud or just to multiple choice questions? Do we really need to breath for 5 min to get an affect? Can we trust the French? Other than the last question, only time will tell…

Ok well we all know physiologically breathing is good for you but turns out it might be psychologically better for you also. This quick hit article was published in the aptly titled: Journal of pschyophysiology (no, not the physiology of psycho’s). This article looks at how breathing can impact the stress of decision making. Interestingly stress can negatively impact decision making. They cite a meta-analysis of 1829 pts that revealed stressful conditions lead participants to take a decision that was more risk taking than in non-stress conditions. In medicine we make stressful decision all the time so can breathing exercises help? This article tries to shed some light on the subject

In this study they postulate that deep slow breathing can increase vagal nerve activity, measured by heart rate variability (HRV) which they state is associated with better decision making. HRV is the physiological phenomenon of variation in the time interval between heartbeats. It is measured by the variation in the beat-to-beat interval.

There are anecdotes of tachycardia and stress in the military about soldiers running into battle without helmets. The thought here is stress leads to an aberrant decision and tachycardia (especially HR over 120) have been a marker of this (see “On Combat: The Psychology and Physiology of Deadly Conflict in War and in Peace by Grossman and Christensen). Therefore, we have to assume that HRV and vagal nerve stimulation are positively correlated with improved decision making.

Interestingly, and unlike most journals, this study reports two different studies under one article heading. I find this strange but we will look at them as two separate studies

The first study examines the effects of two breathing patterns on HRV, differing in their inspiratory to expiratory (I:E) ratios, in healthy men and women. In study 2, they examined whether one of the breathing patterns could result in better decision-making in stressful conditions. The research was conducted in a business school in France on 56 management students at the bachelors and masters level. 

Study 1 found that both types of breathing pattern, a symmetric and a skewed pattern (with a longer exhalation period), increased several parameters of HRV.

Study 2 examined 56 patients in two groups of 28 who were directed not to smoke, consume alcohol or caffeine 3 h before participation and were rewarded for participation. They define HRV-B as a method where people learn to use deep breathing methods to enhance their HRV with electronic biofeedback. Beat to beat intervals are termed NN (much like an R-R interval on an ECGl with N signifying Normal beat to beat intervals). SDNN or the standard deviation of NN intervals, reflects all the cyclic components responsible for variability in the period of recording, therefore it represents total variability. The standard deviation of the average NN intervals calculated over short periods (5 minutes in this study) they use is 50 ms and an SD of 16 ms. They make this very confusing to figure out and there might be some typos in the methods as well. HRV was derived from an ECG with a biofeedback system. They utilized TWO different breathing patterns termed symmetric and skewed. As you can imagine the symmetric pattern  was 5 minutes of inhalation on a 5 second count followed by breath holding for 2 count then 5 second count for exhalation. They skewed pattern was 5 minutes of inhalation on a 5 second count followed by breath holding for 2 count then 7 second count for exhalation.

The control group watched a 5 min “emotionally neutral” movie without music or sound. They utilized within subject experimental design (meaning these were repeated measures and paired statistical testing would be needed). A decision making scenario and questions were created as a business simulation consisting of two parts a reading part and a multiple choice decision part. They were asked to assume the role of someone in charge of a retail company and a writing test after with a strict time constraint. They state that “recently, Brugnera et al. (2018) found that verbal activity masked the vagal withdrawal through altered respiration patterns imposed by speaking. That is why, in this study, we opted for a task in which participants do not need to talk”. I find this interesting and we will get back to this later. Additionally, subjects were asked to evaluate their actual stress level using the VAS stress (for Pre-Stress levels). 

 They note in this first study that the deep breathing group had a significantly (that should be stated STATISTICALLY) higher percentage of correct choices (47% vs 32%; p=0.005) which translated to on average one more question right than the control (2.25 ± 1.35 vs 3.32 ± 1.36 correct answers). I think the more interesting part was the perceived stress by the two groups. 

 The control group had a VAS for pre- stress of 3.57 vs post stress of 8.16. The breathing group had a VAS for pre- stress of 4.40 vs post stress of 5.7. Assuming a clinically significant difference is 3 for a VAS on a 10-point scale (extrapolating from 30 on a 100 pt scale) then this would be significant. 

So put together what does this all mean? It would seem that breathing might be associated with less perceived stress but the effect here seems small with such a small and specialized sample size. Does this apply to medicine where we can sometimes think out loud? The study cited here would suggest that there is masking of the difference when one can verbalize decision. Also does this study mean we have to breath from 2-5 min to reap a benefit of this breathing pattern? What is the optimal length of time for breathing? Does it have to be isolated or can I multitask and do it while I prepare for a stressful situation? Lastly, can we really trust the French?  Certainly, my apple watch tells me that breathing will make me feel better! While all of this makes sense I’m not sure this really is conclusive evidence or always practical. I guess for my clinical practice I’ll try it out and see for myself. It certainly is interesting given the concern over bias in decision making these days.

LVAD Complications: A review

This is a nice little review by Dr. Long in Dallas entitled:

B. Long, J. Robertson, A. Koyfman, et al., Left ventricular assist devices and their complications: A review for emergency clinicians, American Journal of Emergency Medicine, https://doi.org/10.1016/j.ajem.2019.04.050


A ventricular assist device (VAD) can be placed into the right, left, or both ventricles thus the patient can have a right ventricular assist device, left ventricular assist device, or biventricular assist device. The goals of these devices include three different strategies: bridge to recovery, bridge to transplantation, or destination therapy (i.e., the patient is unlikely to recover and not a candidate for cardiac transplant). Contraindications to placement include metastatic cancer, irreversible renal/hepatic failure, and CVA with severe neurologic deficits. The LVAD has two basic designs which produce different patterns of perfusion, including the pulsatile and continuous-flow devices. More commonly these days are the non-pulsatile continuous flow devices.

Device Components

The continuous-flow LVAD has several basic parts including: the internal pump an external power source, and a control unit.The specific components of the LVAD include the inflow cannula, pumping chamber, outflow cannula, percutaneous driveline, controller, and power source. The inflow cannula, usually placed in the apex of the left ventricle (LV), provides the route for blood flow from the native LV cavity to the LVAD pumping chamber.

The pumping chamber, the component of the circuit which provides perfusion, is located in different positions depending upon the LVAD model type: the LV apex for the (HeartMate) HMIII and HVAD devices and the subdiaphragmatic space adjacent to the heart for the HMII device. The pumping chamber contains the impeller, a near-friction-less rotor with rotation speeds ranging from 2500 to 9800 rpm; these types of impeller designs can generate blood flow up to 10 L per minute. The outflow cannula provides the conduit back to the patient’s native cardiovascular system and connects the pumping chamber to the ascending aorta,

The percutaneous driveline provides a conduit for the electrical wiring, connecting the pump to the system controller. These wires not only connect the power source to the pump, but they also provide controlling and sensing functions for the LVAD. The driveline is tunneled

subcutaneously from the pump and exits the skin in the anterior abdominal area to connect to the controller. Thus, it is a frequent source of infection in the LVAD patient. The controller performs multiple functions and contains several important components. It controls LVAD functioning, including power source monitoring and regulation, overall system monitoring, data collection, and alarm system function. subcutaneously from the pump and exits the skin in the anterior abdominal area to connect to the controller. Thus, it is a frequent source of infection in the LVAD patient. 


A continuous flow LVAD will not typically produce a palpable pulse on its own, but patients may have enough native ventricular function to produce pulsatile flow and a pulse. A palpable pressure may also be due to pump thrombosis, and thus, it is important to determine if the patient has a palpable pulse at baseline. If a pulse is palpable, a standard sphygmomanometer may detect a blood pressure, which reflects a systolic blood pressure, rather than mean arterial pressure (MAP). If the pulse is not palpable, a pencil Doppler probe should be placed over the radial or brachial artery. The point at which Doppler signal returns corresponds to the MAP for continuous flow devices. If this is unobtainable, an arterial line may be required, which is the most accurate device for monitoring MAP. Invasive arterial monitoring will demonstrate minimal pulse pressure or flat arterial waveform. Caution is recommended in using pulse oximetry, as a low reading commonly reflects a lack of pulsatile flow. However, a normal value may be accurate. 

LVADs, especially those with continuous-flow, are sensitive to afterload and preload. Guidelines recommend maintaining a MAP of 70–90 mm Hg. Acute hypertensive adverse event is associated with MAP N110 mm Hg in patients with continuous flow pumps. The mechanical hum indicates device power and function. Signs of volume overload (extremity edema, ascites, elevated jugular venous pressure) can be due to subacute or chronic right ventricular failure. However, acute dyspnea, pulmonary edema, or hypotension are more commonly due to acute malfunction of the device, such as cannula obstruction or pump thrombosis. The device exit site, which is normally covered with a sterile dressing, and line should be examined with sterile gloves and mask for warmth, erythema, and discharge, which suggest infection. Finally, the patient should be asked if he/she brought the back-up battery and back-up controller. Sustained ventricular dysrhythmias may be due to underlying cardiomyopathy or decompressed left ventricle due to elevated pump speed or right ventricular failure. Patients with an LVAD will typically demonstrate normal sinus rhythm.


Chest radiograph provides important diagnostic information including position and the type of LVAD, as well as the presence or absence of an ICD or pacemaker. Deep space infection of the LVAD components requires assessment with computed tomography (CT). Sustained ventricular dysrhythmias may be due to underlying cardiomyopathy or decompressed left ventricle due to elevated pump speed or right ventricular failure. Patients with an LVAD will typically demonstrate normal sinus rhythm.

Echocardiogram can evaluate cardiac function and assess for complications such as regurgitation, right ventricular failure, and thrombus formation, though thrombi can be difficult to detect on ultrasound alone.  Key components of the assessment include valvular function, inflow/outflow abnormalities, ventricular size and function, and septal position.

Laboratory assessment includes hemoglobin/hematocrit, lactate dehydrogenase (LDH), haptoglobin, free hemoglobin, and coagulation panel. Hemoglobin and hematocrit with type and screen/cross are needed if concern of bleeding is present. Patients with LVADs are anticoagulated with a vitamin K antagonist, with a goal INR of 2–3 well as aspirin. Free hemoglobin and haptoglobin can assess for hemolysis. Elevated LDH >2.5 times the upper limit of normal suggests hemolysis, which is most commonly due to pump thrombosis in an LVAD patient.  Troponin is recommended in patients with new ECG findings, chest pain, or dyspnea. BNP is a sensitive indicator of volume overload in patients with an LVAD and may be elevated in those with new right heart failure or pump thrombosis or malfunction.

Patients should have a controller tag around their waist indicating the type of device, the institution that placed it, and a phone number. Alarms and functional parameters are shown on the external system controller. Pump speed controls flow. Pump power, flow, and speed should be noted, with assessment of alarms and battery. RPMs and pulsatility index must also be evaluated. 

LVAD specific complications:


A suction event is a common LVAD complication and is associated with low flow events, including dysrhythmia, hemorrhage, and other hypovolemic states such as diarrhea or vomiting.  Reduced LV preload results in collapse of the LV and decreased inflow into the LVAD. Low flow, speed, and power will be present on the controller. While bedside US can demonstrate decreased LV volume, this is often difficult in LVAD patients due to poor acoustic windows, and assessment of the LV diameter may assist in evaluating volume status. Treatment requires fluid resuscitation and managing the underlying etiology. With improved preload and intravascular volume, pump speed and flow will improve. 


Continuous-flow LVADs place patients at high risk of thrombosis, which may originate in the pump or the components such as the inflow or outflow cannula. Types of pump thrombi include acute catastrophic red thrombi entrapped within a fibrin mesh and white thrombi rich in platelets. Red thrombi form at the inlet and outlet areas due to blood stasis, while white thrombi typically form on the pump surface and are associated with turbulent flow. Thrombosis can result in pump dysfunction, hemolysis, emboli, stroke, and death, but patients with thrombosis present with a variety of symptoms due to these potential complications, ranging from no symptoms to cardiac arrest and death. On examination, evidence of hemolysis may be present with scleral icterus, dark urine, and fatigue. Serum LDH is typically >2.5 times normal. Urinalysis may demonstrate hematuria. Other important laboratory assessments include hemoglobin, free hemoglobin, haptoglobin, and coagulation panel. Thrombolysis may be required if patients are hemodynamically unstable. Emergent surgical pump exchange may be needed if the pump stops, the patient is unstable, or if alarms are present. 


Mechanical failure is the second most common cause of death in LVAD patients and may result from several different issues. Pump failure is the most important life-threatening complication requiring immediate care. The controller may demonstrate low flow, low voltage, and power loss. A low flow alarm should always be evaluated by first checking the power. Physicians should auscultate over the LVAD and evaluate for disconnected leads and cannula issues such as kinking or obstruction. A disconnected lead should be reconnected. However, if auscultation reveals no pump activity but all leads are in place, the clinician must assess power and power leads. If all leads are connected, the pump can be reset. If a power lead is not connected to the batteries or unit cable, the cable disconnect advisory will alarm and demonstrate a flashing symbol.  However, if the device has been off for over an hour and the patient is stable, consultation with the LVAD specialist is required, as the device should not be immediately restarted due to high risk of thromboembolic events. In the setting of hemodynamic instability, the device should be restarted immediately no matter the duration of stoppage, with continuous anticoagulation. If the clinician and/or LVAD specialist cannot restart the LVAD, pump exchange is needed, which requires discussion with the LVAD specialist and surgeon. For patients with inadequate perfusion and hemodynamic instability without an alarm activated, resuscitation with IV fluids and standard ACLS protocol is needed.

LVAD-associated complications 


Patients with LVAD are at elevated bleeding risk. Bleeding can occur from several sources: pump connections, grafts in the conduits, and most commonly, mucosal surfaces such as the gastrointestinal (GI) tract. GI bleeding affects 15–30% of patients with an LVAD. Bleeding in the immediate postoperative period is often due to hepatic congestion associated with severe heart failure and the effects of extracorporeal circulation of the bypass machine. Patients may also develop an acquired form of von Willebrand factor (vWF) disease due to the high shear stress associated with LVAD circulation resulting in cleavage and deficiency of vWF. Bleeding in elderly patients with acquired vWF is more severe. Resuscitation of patients with significant hemorrhage with LVAD includes product replacement and reversal agent administration. However, reversing anticoagulation should be weighed with the risk of thrombotic complications, and consultation with the LVAD. Lesions are typically treated with coagulation or clips. Due to the risk of sensitization and reducing the success of heart transplant, blood product transfusion should not be reflexive in patients who are hemodynamically hemodynamically stable. Leukoreduced and irradiated blood products are recommended if available. Octreotide has demonstrated efficacy in LVAD-related GI bleeding in several studies. Desmopressin can be provided, which is a synthetic analogue of vasopressin, or infusion of vWF concentrates. Discussion of platelet transfusion is needed with the LVAD specialist if the patient is thrombocytopenic and bleeding, as well as those with severe hemorrhage.


Ischemic and hemorrhagic stroke can result in poor outcomes and demonstrate a prevalence of 6.8% and 8.4%, respectively. There is an increased risk with every 5 mm Hg increase in systolic blood pressure. The ENDURANCE trial found a lower stroke rate with MAP 90 mm Hg, with patients receiving close blood pressure control demonstrating a 24.7% reduction in total neurologic events and 50% decrease in hemorrhagic stroke rate. Acute ischemic stroke more commonly affects the right cerebral hemisphere in patients with an LVAD. 


Patients with an LVAD are at high risk of sepsis, with rates of infection approaching over 42% in the first year post-implant, usually 2 weeks to 2 months. The driveline and VAD pump pocket are the most common infectious sites, with 80% of driveline infections occurring in the first 30 days of transplant. The system controller may demonstrate a high-flow alarm with distributive shock due to loss of vascular tone. LVAD-related infections include those that may occur in patients without an LVAD, but occur with greater frequency in LVAD patients such as mediastinitis, endocarditis, and bacteremia.  Non-LVAD infections include pneumonia, Clostridium difficile infection, and urinary tract infection (UTI). Within the first 3 months post implantation, the most common sources of infection typically include catheters, pneumonia, and C. difficile, while later sources of infection are more commonly related to the device.  Only half of patients will demonstrate fever, leukocytosis, or meet criteria for systemic inflammatory response syndrome. Discussion with the LVAD specialist and cardiothoracic surgery is recommended.  Deep infections typically require surgical debridement, while persistent bacteremia may require removal and implantation of a new device. 

RV Failure

RV failure is a major cause of morbidity and mortality, occurring in 15–40% of patients. Late onset right heart failure is increasingly being reported with RV dysfunction, ventricular dysrhythmias, pulmonary hypertension, tricuspid regurgitation, and device thrombosis or malfunction. This can result in reduced preload to the LV, decreasing LVAD flows and triggering a low-flow alarm. RV failure may result in elevated liver function tests, creatinine, and lactic acid. RV failure requires inotropes and/or vasopressors, pulmonary vasodilators, and LVAD specialist consultation. Patients may require careful fluid resuscitation, with 250 mL boluses. 

Ventricular Dysrhythmias

Patients may tolerate severe ventricular dysrhythmias with minimal symptoms due to the LVAD producing adequate cardiac output to meet end organ perfusion despite poor venous return. Patients often have an ICD prior to LVAD placement. Dysrhythmias may eventually result in compromised blood flow and can also contribute to RV dysfunction, suction events, thrombus formation, and poor perfusion. The controller will demonstrate low flow in patients with hypotension due to the dysrhythmia.

Aortic Regurgitation

Aortic regurgitation (AR) may develop de novo in up to 25% of patients after LVAD placement. AR more commonly occurs in patients with a closed aortic valve compared to patients in whom the valve frequently opens. AR results in decreased LVAD efficacy and may require modifications in pump speed, managed by the LVAD specialist. Patients may require aortic valve replacement.


Standard procedures for resuscitation are recommended as needed. Hypotension in LVAD patients is defined by MAP <60 mm Hg. Patients who are conscious should be assessed with history and examination, with close assessment of volume and perfusion status. ECG and bedside echocardiogram are vital components of the assessment, with analysis of LVAD components. Patients who are unresponsive and hypotensive require external chest compressions. Literature suggests no cases of dislodgement during cardiopulmonary resuscitation (CPR). If the patient has a MAP N<50 mm Hg or end tidal CO2 <20 mm Hg with a device possessing an audible hum, perfusion is likely adequate, and compressions are not necessary MAP <50 mm Hg without an audible hum in the unresponsive patient is associated with compromised perfusion and requires chest compressions at the same depth and frequency as in those without an LVAD. Defibrillation should be performed for unstable ventricular dysrhythmia. The pads should be placed distant from the pump, and if an ICD is present, the pads should not be placed directly over the ICD. In patients with adequate perfusion and respiration but who remain unconscious, evaluate for hypoglycemia, stroke, hypoxia, sedation, and coma.


Chest thoracostomy with chest tube placement in the setting of trauma with pneumothorax and/or hemothorax is recommended, but clinicians must avoid the driveline. Arterial line placement can be beneficial, and US guidance is recommended. Pericardiocentesis should be avoided due to risk of serious device complications, but it is recommended in the case of pericardial tamponade with hemodynamic compromise.





ST segment elevation in aVR is probably NOT a STEMI…BUT, damn, are they sick!

There I said it. Well Ok I’ve been saying it for a while. But say this to a room full of doctors and you might be ostracized! Well there has been mounting evidence that would argue aVR ST elevation is not an acute STEMI/ STEMI equivalent for a while now. I think this article clinches it for me. However, just because its not a STEMI equivalent doesn’t mean they are ok, nor does it mean we should forget about aVR…


In this blinded retrospective review of 99 patients with ST elevation in aVR (STE-aVR) and multi-lead ST depression, only 10% had a definite culprit lesion,  none had left main or LAD occlusions, and 40% had no disease or mild to moderate disease on PCI. However, this group had an overall in-hospital mortality of 31% compared with 6% in a matched conventional STEMI group. Patients with aVR ST elevation represent a very sick cohort of patients who need critical care and a workup for why they are having poor diffuse coronary perfusion.


This was a retrospective study with actually quite good blinding and decent methods that looked at 854 consecutive STEMI activations in a 35 month period. They identified 99 patients with ST elevation in aVR (STE-aVR) and multi-lead ST depression in the final analysis. Cardiologists reading the ECGs and catheterizations were blinded to the study outcome. The primary outcome was the number of patients presenting with STE-aVR and multilead ST depression who had an acutely occluded culprit coronary artery on PCI. Secondary outcomes were number of patients who presented with cardiac arrest, and survival-to-hospital discharge compared to the non-aVR ST elevation STEMI population. None of these 99 patients had ST elevation in 2 contiguous leads (STEMI definition). Of the original 99, 79 underwent PCI. Interestingly the 20 that did not undergo PCI were a very sick population including known severe coronary artery disease on recent coronary angiogram, neurological emergency, obvious extra-cardiac etiology for arrest, and very long down time with poor prognosis. Of the 79 patients that had PCI, 8 (10%) had evidence of a definite, acute, thrombotic, culprit coronary occlusion but none had an acutely occluded left main or left anterior descending coronary artery. 19 of these 79 had angiographically normal vessels and 40% of them had either no disease or only mild to moderate disease on PCI. However, these were a pretty sick group though: 47 developed respiratory failure, 15 developed cardiogenic shock , 48 developed acute kidney injury( 9 of them requiring hemodialysis), and 36 patients underwent in-hospital coronary revascularization (29 PCI and 7 CABG). Wow are these a sick cohort! Those with STE-aVR had a 31% in- hospital mortality, whereas those with a non-aVR STEMI had a 6.2% in-hospital mortality (p <0.00001)! The presence of a critical medical condition or severe multivessel subocclusive disease with intact distal flow was the most common etiology for this ECG finding. This emphasizes working up these patients for the underlying issue. These patients don’t likely need immediate PCI of a culprit lesion they likely need resuscitation and eventually treatment by a multi-specialty heart team for their multi-vessel disease.


Harhash, A et al. aVR ST Segment Elevation: Acute STEMI or Not? Incidence of an Acute Coronary Occlusion. The American Journal of Medicine (2019) 000:1−9. PMID: 30639554. DOI: 10.1016/j.amjmed.2018.12.021

Emergency Contraception in the ED and BMI?

We in the ED may sometimes be asked advice on emergency contraception (EC). The most popular one available is levonorgestrel.  The message from this quick hit article is pretty simple and serves as a simple reminder:

Clinicians should counsel women with BMI ≥ 26 kg/m2 on the potential limitations of oral levonorgestrelfor EC and offer more effective EC methods, including the copper intrauterine device and oral ulipristal acetate1.

In a study by Praditpann et al. it was found that after a single dose of EC, obese-BMI women are exposed to lower concentrations of levonorgestrel and similar concentrations of ulipristal, when compared to normal-BMI women2.

A large study of 6873 women in four randomized trials on EC where Participants took either 1.5 mg of levonorgestrel as either single or two separate doses, up to 120 h after unprotected intercourse showed pregnancy rates of 1.25%in women BMI <25 andpregnancy rate 2.03% in womenBMI>303.


  1. Stowers P. Use of levonorgestrel as emergency contraception in overweight women.Obes Res Clin Pract. 2019 Feb 25. pii: S1871-403X(18)30508-8. doi: 10.1016/j.orcp.2019.01.007.


  1. Praditpann et al. Pharmacokinetics of levonorgestrel and ulipristal acetate emergency contraception in women with normal and obese body mass index. Contraception 95 (2017) 464–469.


  1. Effect of BMI and body weight on pregnancy rates with LNG as emergency contraception: analysis of four WHO HRP studies. Contraception. Volume 95, Issue 1, January 2017, Pages 50-54. https://doi.org/10.1016/j.contraception.2016.08.001




Fill the Tank or Rev’ the Engine… WHOA, PUMP the BRAKES: Early fixed dose norepi for septic shock?

We all know hypotension is bad. But should we fill with fluids or rev up the heart with pressors. This begs the question does early norepi help prevent hypotension and mortality? Today’s quick hit article:

Permpikul C, et al. Early Use of Norepinephrine in Septic Shock Resuscitation (CENSER) : A Randomized Trial. Am J Respir Crit Care Med. 2019 Feb 1. doi: 10.1164/rccm.201806-1034OC. PMID 30704260.


In this prospective double blinded intention to treat analysis RCT conducted in Thailand the authors found that early fixed dose norepinephrine (0.05 mcg/kg/min) allowed for faster time to the primary outcome of “shock control” by 6 hours after diagnosis of septic shock. Shock control was defined as achievement of a MAP >65 plus urine output (>0.5 ml/kg/hr) and a lactate that cleared by 10%. They found no clinical or statistical increase in limb or organ ischemia even when norepinephrine was administered by peripheral access during the patient’s entire hospitalization. Lastly, a secondary (hypothesis generating) outcome of 28 day mortality showed a difference of 5% in favor of the fixed dose norepinephrine group which would be nice if that were real and reproducible.  


Syncope Admissions


This was a very interesting phase II, double-blind, placebo RCT with intention to treat analysis. That looked to answer the question of “shock control” in septic shock. Shock control (A COMPOSITE OUTCOME) was defined as achievement of a MAP >65 mmHg plus urine output (>0.5 ml/kg/hr) and a lactate that cleared by 10%. The interesting part comes in from their inclusion/exclusion criteria. Keep in mind that the population studied is one of the biggest predictors of generalizability in a trial. They excluded patient if they met the criteria of septic shock for MORE THAN 1 hour before randomization! I’m not really sure why they would do that. They screened 456 patients and excluded 136, 31 of which were because of shock for >1 hour.  Another weakness I found was that they did not report how they calculated the MAP and what percent had a-lines and what percent was calculated by non-invasively. This can be VERY important (See my post on MAP later). The final study randomized 155 patients to each group. The two groups looked otherwise pretty similar in their baseline characteristics. The intervention group was started on 0.5 mcg/kg/min fixed dose of epi and everyone was blinded to this intervention with a placebo in the control group. Time to getting norepi was pretty quick 93 minutes. Theoretically, you could say the norepi group could be differentiated by providers if they saw the BP go up. However, 20% of the normal saline placebo had an increase in the blood pressure.  Both groups were allowed to have open label norepi The primary outcome here was time to shock control by measured MAP, urine output, and lactate. So, that is not the most fair primary outcome since obviously time to shock control will be faster in the norepi group. 76% of the norepi group vs 48% of the placebo group met the primary outcome. An interesting feature of the study was in the COMPOSITE primary outcome less than 10% of both groups reached the lactate clearance group whereas much more of the groups (35 vs 24) reached the urine output goal. This is more reason to make me wonder how useful lactate clearance is. A secondary outcome of mortality at 28 days, although NOT statistically significant was interestingly was a 6% difference (15.5 vs 21.9) in favor of the early norepi group. THIS METRIC IS DEFINITELY SOMETHING I’D LIKE TO SEE REPEATED IN A US STUDY AS THE PRIMARY OUTCOME. Another 5% mortality is impressive. 20% is what is typical of all the other sepsis trials (ARISE, PROMISE, etc.) with sick patients showed, so that would be impressive if early norepi could decrease this subsets mortality. The last interesting point is that there was no statistical difference in limb or organ ischemia between the two groups and half of the patients had only peripheral lines. These patients were in the hospital for a median of 10 days. It would be nice to know how long the patients had the norepi running. Also, many of these patients were admitted to the wards and NOT the ICU (about 50% of both groups). These two facts in addition to all the other literature on peripheral pressor’s really makes me feel better about running norepi peripherally. However that fact that this was based in a lower resource country with admissions criteria that are different also make it not as generalizable.  So does this change practice? That waits to be seen, however, I have always felt that dumping in the magical 30ml/kg into a patient who is not hemorrhaging to be missing the point. I am biased toward giving early pressors…



Lactate for Safely Screening Sepsis?

Fernando  et. al.  HELPFUL ONLY WHEN ELEVATED: INITIAL SERUM LACTATE IN STABLE EMERGENCY DEPARTMENT PATIENTS WITH SEPSIS IS SPECIFIC, BUT NOT SENSITIVE FOR FUTURE DETERIORATION Journal of Emergency Medicine, Vol. 54, No. 6, pp. 766–773. https://doi.org/10.1016/j.jemermed.2018.01.040 


A lactate >4 in this prospective observational cohort study of almost 1000 patients with the “CMS” definition of sepsis was helpful in predicting “clinical deterioration” with an positive likelihood ratio (+LR) of 10.7 but only a negative likelihood ratio (-LR) of 0.8. Approximately 9% of patients with lactate >4 where sent home. However, a lactate of >2 only had a positive likelihood ratio (+LR) of 2 and a negative likelihood ratio(-LR) of 0.8. Lactate alone of >2 and less than 4 was not useful in predicting who had clinical deterioration or who didn’t. CMS uses a cutoff of 2 mmol/L

Ever since “sepsis” has become the only reason for a fever, the screening for mortality predictors has been the holy grail. Lactate has been promulgated as the “Lost Ark” of these biomarkers. However, just as with any single biomarker that has since been used it lacks the sensitivity and specificity we need and has rarely been tested agains physician gestalt. Remember that the THREE BIGGEST Sepsis trials all used a cutoff of lactate >4 after “fluid resuscitation”. Remember that CMS uses a lactate of >2…

Here, another article to remind us of the test characteristics of lactate arrives. In this very good prospective observational cohort study of adult ED patients satisfying the original sepsis definition the authors seek to find how good is lactate.  The authors looked at cutoff of both 2 and 4 mmol/L as the “discriminatory” zone of mortality. They do use a composite outcome of “clinical deterioration” which they define as: death, endotracheal intubation, vasoactive medication administration for a minimum of 1 h, noninvasive positive pressure ventilation (NIPPV) for a minimum of 1 h, or ICU admission for a minimum of 24 h. 985 patients met the original sepsis definition. Of these 84 patients (8.5% – shows you how good our sepsis definition is….) met the primary outcome of clinical deterioration and half of that 4% met the primary outcome while still in the ED. 

A lactate > 4.0 mmol/L  had a +LR of 10.7 (95% CI 6.3–18.3)and -LR 0.8 (95% CI, 0.7–0.9) for for predicting deterioration ( Sp 97.4% (95% CI 94.1–100%); Sn 27.4% (95% CI 17.8–36.9%). Of those patients with a lactate of 4 8.7% were discharged home and did not meet the primary outcome. 

A lacate of >2.0 mmol/L  had a +LR of 2.0 (95% CI 1.7–2.3) and -LR 0.5 (95% CI 0.4–0.7) for predicting deterioration, (Sn 67% (95% CI 55–76%); Sp of 66% (95% CI 63–69%). Of patients with a lactate < 2.0 mmol/L, 224 (56.1%) were discharged home and did not meet the primary outcome (which means half DID meet the primary outcome)

Of note those with a lactate 2.0-3.9 mmol/L 90% did NOT meet “clinical deterioration” and 10% did meet clinical deterioration 

Thus a lactate >4 was useful in predicting poor outcome but anything less was not useful either way (in reassuring no detioration or not).

This literature isn’t new but it is recent and should remind us that a normal lactate doesn’t help, nor does a lactate less than 4. Thanks CMS for more useless “meaningful use”…

To B (VM) or Not to B (VM) THAT is the question!

Source: Casey et. al. Bag-Mask Ventilation during Tracheal Intubation of Critically Ill Adults. NEJM. February 18, 2019DOI: 10.1056/NEJMoa1812405

Whether ‘tis noble in the mind to suffer the slings and arrows of aspiration or by bagging end them?

This is a great methodological (and probably landmark) trial by the impressive authors at Vanderbuilt et al. They have been publishing amazing literature recently. What a great read. The question they ask here is does BVM prevent hypoxia or is the old dogma concerned about aspiration correct? Like all research I think the answer to the question here is it depends but this certainly adds to the current literature. It also raises more questions like any good research should. It’s an interesting paper but I’m not sure its definitely says no more BVM. 

Bottom Line:

This was a randomized intention to treat, unblinded pragmatic trial of BVM vs no BVM in patients in the ICU with low risk of aspiration. While they did NOT meet the primary outcome of a 5% difference between the lowest oxygen saturation for BVM vs no ventilation, they did find an interesting 10% less patients who had oxygen saturation’s less than 80% in the BVM vs the No BVM Group. While this is a secondary (hypothesis generating) outcome it is quite striking.

This is definitely a dogma breaking study. There are definitely limitations however, I think one possible conclusion is: In patients where you are sure aspiration is not a high risk then BVM may a safe way to prevent hypoxia by giving the patient assisted ventilations from the time of induction until laryngoscopy. If you wait until saturation’s drop to 90% you will have a lower overall oxygenation and that could be associated with harm

The Details:

This was a randomized unblinded pragmatic trial of BVM for Endotracheal Intubation. Intervention was BVM from induction until the initiation of laryngoscopy vs No BVM allowed unless they had a “failed” intubation with sat <90%. However only 5 people in the No BVM group got BVM. 

The BVM group received training on best practices for BVM whereas the no BVM did not. Did this bias the results?

  1. Patient population:
  • These are ICU patients and NOT ED patients. So this cohort would naturally be more likely to have empty stomachs being in the ICU and less likely to aspirate
  • 49 pts were excluded (as listed in exclusion criteria) for being “high risk of aspiration”. That is a problem for broadly applying this study to the ED population. 
  • 50% of the patients were known to be fasting for the prior 6 hours in both groups.  (See table S3 in the supplement)
  • The trial was not blinded so was there some part of the unblinding that contributed to the difference?
  • Non-invasive pressure ventilation was not allowed between induction and laryngocospy but was allowed before. BIPAP was utilized in 16% vs 23% of BVM vs no BVM. 
  1. Primary Outcome:
  • The primary outcome was the lowest oxygen sat. So the differnce was 96 vs 93. Their study was powered to find a 5% difference and they did NOT achieve this so you can’t say the primary outcome was met. 
  • They say a post hoc  analysis adjusting for variables was 5.2 but that doesn’t appear to be prespecified and certainly wasn’t stated in the primary outcome
  1. Secondary outcome
  • The lowest sat <80 stat is interesting and almost double. However, I wonder why this no BVM group got so low. They said that a BVM was not permitted except after a failed attempt O2 <90%. So why did the non BVM group get so low? In the supplement it states only 5 patients of the no BVM trial got BVM. 
  1. Group differences
  • The BVM group had a higher (3%) first pass success rate and lower use of bougie. Was the no-ventilation group more difficult airway? It’s hard to say from the supplment table S3 (difficult airway). No overall assessment of difficulty was asked but we don’t have a great tool for that anyway. 
  • Supplemental oxygen: Why did the no-BVM have lower proportion of patients given supplemental oxygen (77 vs 100)? What does this mean? In the supplement table S3 they do say 7 patients had no preoxygenation vs 3 in the BVM group. 

The most interesting part I found was in the supplement in a table that looks something like this (I recreated their numbers):


(I made the above bar chart from their supplement pages)

Quite the striking difference. Overall will this change my practice? Maybe, I’ll probably consider bagging earlier and see if that helps.

Antibiotic StewardSHIP…Moveover Titanic…there’s a new failure in town.

Tamma, et al. Rethinking How Antibiotics Are Prescribed: Incorporating the 4 Moments of Antibiotic Decision Making Into Clinical Practice. 2018 Dec 27. doi: 10.1001/jama.2018.19509. [Epub ahead of print]


BOTTOM LINE: Providers should be asking themselves: “What is the likelihood my patient has an infection and that it requires antibiotic therapy?”

Those who know me, know my passion for medicine can occasionally turn into a status rant-icus type soliloquy on the failures of modern medicine (I feel one of those coming on me like a seizure…). One of these high horses I like to ride is on a term that has come to have as much meaning and futility as “military intelligence” or “President Trump”…that term is antibiotic stewardship.

I think its easy to know to whom I am referring. For instance, if you have ever in one breath prescribed “a zpack, prednisone, and albuterol” then I am referring to you; if you reflexively say flouroquinolone when you hear words that start with Sinu-, bronch-, or cyst- and end in -itis, then I’m referrering to you; if you have ever used the words “just to be safe” as your clinical decision aid to give antibiotics, then I’m referring to you; if you use the terms “strong” and “weak” to describe antibiotics, then I am referring to you; and lastly if anyone has ever asked you “so what infection are you actually treating?!” Then I am definitely referring to you. (By the way I think a great use of machine learning would be to develop a wristband that shocks providers who type keystrokes recognized as -itis and then Rx an antibiotic.)

Thus much like music to my ears or more appropriately propofol to my seizure I came upon the opinion paper in JAMA entitled “Rethinking How Antibiotics Are Prescribed: Incorporating the 4 Moments of Antibiotic Decision Making Into Clinical Practice”….Ahh had they only incorporated the word zen into the title then I might have been able to sleep more between these night shifts. I think this is a must read!

The article delineates a 4 step process (much shorter than 12) to thoughtfully redirect the thinking of antibiotic prescriptions. 

(Zen) Moment #1: (an antibiotic “TIMEOUT”)

This is the most important step! “Does this patient have an infection that requires antibiotics?

Yes, there are cases in severe infections where early antibiotics makes sense, but that does not translate down to an otitis media, or worse a fever! Just as all the glitters isn’t gold…all that has a temp is not an infection. Let’s go back to considering a differential (inflammation, medication, etc) and seeing if there is an alternate reason for the fever. Conversely, some causes of fever are viral and don’t need antibiotics. In fact, there is much literature currently on asymptomatic bacturia and I question giving antibiotics to all urinalyses… So ask yourself “what is the likelihood of an infection that requires antibiotic therapy.

(Zen) Moment #2:

Have I ordered appropriate cultures before starting antibiotics? What empiric antibioitcs should be started? Although not all patients need cultures we should culture appropriately. The authors state that most patients with community acquired pneumonia, abdominal, and no purple tissue cellulitis are NOT high risk for MRSA and may not need vancomycin (I am definitely guilty of not using enough empiric nafcillin/ancef)

(Zen) Moment #3:

A day or more has passed, can I stop antibiotics or narrow the therapy? Can I change from IV to oral therapy? There is much literature on physicians continuing therapy despite culture data to the opposite. 

(Zen) Moment #4:

What duration of antibiotic therapy for this patients diagnosis is needed? Remember the timelines we use are have minimal evidence behind them. Some will say that most bacterimia is cleared after 2-3 doses of IV abx. We should be using patient response to therapy rather than opinion based guidelines.

I love this article and think everyone should read it as a reminder. The more we remind ourselves of the literature the better we will do with decreasing unnecessary antibiotic use. 

Maybe the authors should have changed the name of the article to “An Antibiotic Timeout”. There more I think about this concept the more I like it. 

I better go check my EEG…

Macrolide Resistance: Is Resistance Futile?

There hasn’t been an update to the IDSA/ATS guidelines for CAP (community acquired pneumonia) since 2007 (I keep hearing they will be out soon). These guidelines recommend
 a macrolide antibiotic as first-line therapy for previously healthy patients who have no risk factors for drug-resistant S. pneumonia (SPN) infection (strong recommendation) and a combination of a macrolide and a beta-lactam for patients who require hospitalization but not in the ICU 1.  In the last few years, concerns for macrolide resistance have been discussed (I’m looking at you EM-RAP). Paradoxically no real affects on clinical outcomes have been shown. Referred to as the aptly named in vivo-in vitro paradoxthere are divergent findings of macrolide resistance and clinical outcome. Lets talk amongst ourselves, shall we?


Macrolide resistance is a thing! But the clinical studies don’t yet support the concerns of the in vitro findings of resistance.  The IDSA/ATS guidelines have not been updated since 2009 and its still OK to use macrolide monotherapy when high levels of strep pneumo resistance are unlikely. So what to do? If I’m worried about compliance I’ll still prescribe OUTPATIENT azithromycin alone (especially if I feel atypicals are the cause and at least until new IDSA guidelines come out). If I am at all “worried” I will add in a beta-lactam. For HOSPITALIZED PATIENTS OR SEVERE DISEASE NEVER USE AZITHROMYCIN ALONE.  RESISTANCE MAY BE ANNOYING BUT ITS NOT YET FUTILE


     I’m not disagreeing with macrolide resistance in general, I just don’t think it’s futile to use these even as monotherapy in CAP. Typically Azithromycin resistance is defined as an MIC >2 μg/mL) and sometimes very high resistance levels are seen. However there seems to be disparate results between MIC’s and clinical cure rates.   I think the misperception is best seen in Zhanel’s article 3on clinical cure rates. This study looked at community acquired respiratory infections (CARTI) and azithromycin resistance in streptococcus pneumoniae. Overall if you look at the study there was about a 10% difference in clinical cure rate in those treated with azithromycin for ALL respiratory tract infections. HOWEVER, in this group they included the following as CARTI: acute otitis media (sometimes air comes out of my ear when I sneeze but I don’t think its part of the respiratory tract!), CAP, acute bacterial exacerbations of chronic bronchitis and acute bacterial sinusitis. They found 1127 patients with CARTI of which they found 29% (112/388) of subjects with SPN had AZ-R. Overall, clinical cure rates in CARTI subjects treated with azithromycin were higher for in AZ-S (89.4%) versus AZ-R (78.6%; P=0.003). BUT if you look at CAP alone for AZ-S vs AZ-R the failure rates were 3/52 vs 2/27 (p=0.986) and the cure rates were 49/52 (94%) vs 25/27 (93%), respectively. These are small numbers true so lets look at some more studies.

      In this study by Yanagiharathey looked at a 3 (Three??? Yes three) day course of azithromycin in adults with mild to moderately severe CAP, and to determine whether in vitro macrolide resistance in of SPNis related to clinical failure. A good clinical response was defined as improvement of 3 of the 4 following outcomes: 1) resolution of fever 2) resolution of wbc’s, 3) improvement of CRP and 4) improvement of CXR findings. They found a good response in 13/17 (76%) and an ineffective response in 4/17 (23%). Strangely, 6 of 7 patients in whom high-level resistance was documented (MICs >256 μg/mL) showed good clinical responses. If you look at CAP alone for AZ-S vs. AZ-R the failure rates were 0/2 vs. 4/12 and the cure rates were 2/2 vs. 8/12, respectively.

     In this study, Kohno5 looked at moderate-to-severe community-acquired pneumonia.  They found that despite mostly having AZ-R strains in Japan, clinical efficacy and bacterial eradication were achieved in 10 of 11 patients. If we combine these 3 studies (not a great meta-analysis I know…) for AZ-R then you would have 43/50 (86%) good responses and 7/50 (14%) poor responses and this is in places with the highest rates of resistance… 6

     Lastly, in this study by Cillonizlooked at outcomes in AZ-S vs. AZ-R.  They found 30-day mortality in AZ-S to be 32/493 (6%) vs. AZ-R 9/133 (7%, p=0.93) and ICU admission to be 140/493 (28%) vs. 32/133 (24%, p=0.29). 

Indeed there is some thought that macrolide antibiotics have additional properties than just their bacteriostatic/bacteriocidal properties. Some studiessuggest that macrolides have anti-inflammatory properties and prevent the production of pro-inflammatory mediators and cytokines. This recent study of lowered mortality in CAP would agree with a reduction of all cause in hospital mortality and a 10% difference (not statistically significant) in 30-day mortality.  Here, Arnold et al.looked at 549 patients with CAP and bacteremia where 247 (45%) were treated with a macrolide and 302 (55%) were not.  The unadjusted 30-day mortality was 18.4% in the macrolide group, and 29.6% in the non-macrolide group (RR=0.81, CI 0.50–1.33; P = 0.41). Unadjusted in-hospital all-cause mortality was 7.3% in the macrolide group, and 18.9% in the non-macrolide group (RR= 0.54, CI 0.30–0.98; P = 0.043).

There are two major mechanisms mediating resistance to macrolides. The ermB gene encodes a methyltransferase that causes ribosomal methylation resulting in a phenotype that reduces susceptibility to macrolides (and lincosamide FYI). This mechanism results in the highest macrolide resistance. The second mechanisms is the mefA gene which codes for an antibiotic efflux pump removing the drug from the SPN. Macrolides are concentrated intra-cellularly, and this is thought to result in increased drug delivery to the site of infection, and exposure to high concentrations of drug following phagocytosis, which may overcome low level resistance. Possibly high-level resistance may be clinically relevant , however that was not exactly the case in the above Yanagihara study.

So in summary, macrolide resistance is definitely real. However, there really does not seem to be a huge affect clinically when azithromycin resistance is encountered. Furthermore, Azithromycin resistance is probably more likely in its uses other than CAP, like otitis media and sinusitis (and other infections I don’t treat with antibiotics). So how do we use this information? Well I would NEVER use macrolides alone for hospitalized patients. I still believe that for hospitalized patients macrolides and beta-lactams are the way to go. In fact, I like to add azithromycin to even my ICU patients just in case there is something magical about the macrolides. For outpatient management of mild CAP I think there is still a role for macrolide monotherapy. We all know that compliance is not great with patients and I feel in mild disease I’d rather they take 1 once a day antibiotic rather than have to add in a beta-lactam and have them not take it.


  1. Mandell LA, Wunderink RG, Anzueto A et al. Infectious Diseases Society of America/American Thoracic Society consensus guidelines on the management of community-acquired pneumonia in adults. Clin Infect Dis 2007; 44 Suppl 2, S27-72.
  2. Bishai W. The in vivo-in vitro paradox in pneumococcal respiratory tract infections. J Antimicrob Chemother 2002; 49, 433-436.
  3. Zhanel GG, Wolter KD, Calciu C et al. Clinical cure rates in subjects treated with azithromycin for community-acquired respiratory tract infections caused by azithromycin-susceptible or azithromycin-resistant Streptococcus pneumoniae: analysis of Phase 3 clinical trial data. J Antimicrob Chemother 2014; 69, 2835-2840.
  4. Yanagihara K, Izumikawa K, Higa F et al. Efficacy of Azithromycin in the Treatment of Community-acquired Pneumonia, Including Patients with Macrolide-Resistant Streptococcus pneumoniae Infection. Internal Medicine 2009; 48, 527-535.
  5. Kohno S, Tateda K, Kadota J et al. Contradiction between in vitro and clinical outcome: intravenous followed by oral azithromycin therapy demonstrated clinical efficacy in macrolide-resistant pneumococcal pneumonia. J Infect Chemother 2014; 20, 199-207.
  6. Cheng AC, Jenney AWJ. Macrolide resistance in pneumococci-is it relevant. Pneumonia (Nathan) 2016; 8, 10.
  7. Cilloniz C, Albert RK, Liapikou A et al. The Effect of Macrolide Resistance on the Presentation and Outcome of Patients Hospitalized for Streptococcus pneumoniae Pneumonia. Am J Respir Crit Care Med 2015; 191, 1265-1272.
  8. Ianaro A, Ialenti A, Maffia P et al. Anti-inflammatory activity of macrolide antibiotics. J Pharmacol Exp Ther 2000; 292, 156-163.
  9. Arnold FW, Lopardo G, Wiemken TL et al. Macrolide therapy is associated with lower mortality in community-acquired bacteraemic pneumonia. Respir Med 2018; 140, 115-121.




SUP-ICU Trial: Pantoprazole in patients at risk for Gastrointestinal Bleeding


Kart, M. Pantoprazole in patients at risk for Gastrointestinal Bleeding in the ICU. N Engl J Med 2018;379:2199-208.

Trial done in the UK. It was a multicenter, stratified, parallel group, placebo controlled blinded randomized trial. 

Included pts admitted to the ICU who were 18 yo or older and had at least one risk of actor for clinically important GIB including, shock, anticoagulants, crrt, MV >24hours, any liver disease, he of or ongoing coaulopathy

Primary outcome was death by 90 days after randomization. Secondary outcomes were GI bleeding, new onset pna, c. Diff. Predefined subgroups including SAPS II of >53 and <53

Pts were randomized to receive either 40mg IV pantoprazole daily until ICU discharge for a maximum of 90 days


  1. For the primary outcome of mortality 31.1% vs 30.4% died (RR 1.02; 0.91-1.13, p=0.76)
  2. For the outcome of clinically important GIB 2.5% of pantop group vs 4.2% of placebo had clinically important GIB. 
  3. The RR of have a SAPSII >53 was 1.13 (0.99-1.30) vs 0.92 (0.78-1.09) p=0.05.
  4. This was the subgroup with the largest separation of average RR (Despite overlapping C.I. and both crossed 1 when compared to other subgroups of shock, vent, etc.

The study was powered to detect a 5% difference which seems really large and unlikely. To power a less then 1% difference would take a vastly larger trial. The endpoint of clinically important GIB seems like that should have been the primary outcome. Since it is only hypothesis generating we cant say much about it. Taken at face value there is probably a group that would benefit more than the ones they picked. Maybe they should have enrolled ICU pts with SAPSII >53 as one of the inclusion criteria to see if there would be a bigger impact. However, this is a large trial with around 3300 pts so getting this done again will be unlikely. Probably sick ICU pts should get SUP for the prevention of GIB but not death and consider maybe a sicker population because right now the NNT for prevention of GIB is around 60.  

Last not there were a lot of these docs getting monies from drug companies…