Sepsis

A Balancing Act

It’s another busy day in the ED when an elderly female comes in from triage with fever, cough, and new oxygen requirement. Even before the patient comes back you are concerned for pneumonia with sepsis. The patient is tachycardic and hypotensive with a shock index greater than one. You institute early antibiotics and fluids and systematically begin to aggressively resuscitate her. The patient requires nearly four liters of normal saline before her blood pressure stabilizes. Your attending suggests that your liberal use of normal saline will induce a hyperchloremic metabolic acidosis, and perhaps you should have used lower chloride containing fluid, like lactated ringers. You perform a brief literature review on the topic of balanced resuscitation using lower chloride containing fluids.

Literature Review:
Strong Ion Difference (Kishen et al)
The main difference between normal saline and balanced fluids, such as lactated ringers, is the strong ions difference (SID), that is, the difference between cations (e.g. Na+) and anions (e.g. Cl-).  Normal saline has a SID of zero (equal parts Na+ and Cl-) where as Lactated ringers has a SID of 28, which is due to the additional cations such as Ca+, K+, and lower anion (Cl-) content.  Importantly, normal plasma SID content ranges from 38-44mmol/L, therefore balanced fluids more closely approximates physiologic SID.  As the SID becomes narrower, as is the case with significant normal saline administration, a non-gap metabolic acidosis develops. [1]

The use of normal saline in large volumes has been shown to produce a reliable drop in serum pH as demonstrated by Scheinraber et al, in a study among patients undergoing elective surgery. [2] However, the development of a hyperchloremic acidosis is of unclear clinical significance. Early animal models in dog kidneys demonstrated that compared to non-chloride fluids, chloride containing solution led to renal vasoconstriction and decline in glomerular filtration rate. Similarly a randomized, double blind crossover study in healthy humans demonstrated a significant reduction in renal blood flow and renal tissue perfusion, after the administration of two liters of normal saline compared to low chloride (98 mEq/L) Plasma-Lyte solution. [3] However, the effect of isotonic saline in acutely ill patients is still not as clear. A prospective cohort study among 175 ICU patients demonstrated that higher chloride levels (109.4 vs 115.1mEq) was an independent factor for increased mortality, although a limitation of this study was they could not distinguish the cause of hyperchloremia (iatrogenic, renal dysfunction, or endogenous hyperchloremia) [4]
Traditional and 'balanced' fluid content (crashingpatient.com)
A large retrospective cohort study of critically ill adults with vasopressor dependent sepsis showed lower in-hospital mortality in patients who received balanced (lower chloride) fluids versus isotonic saline, 19.6% versus 22.8% (RR 0.86; 95% CI,0.78-0.94). A limitation of this study was that patients receiving balanced solutions were younger, less likely to have chronic heart and renal failure, and more likely to receive steroids, colloids and invasive monitoring. [5] A 2014 retrospective study in 109,836 patients that met SIRS criteria and received crystalloid fluid resuscitation, showed that low-chloride loads were associated with lower in-hospital mortality. This mortality difference remained even after adjustment for severity of illness and total fluid volume administered. [6]

Similarly, a before and after study by Yunos et al involving 1644 ICU patients, reported the use of chloride-restricted fluids was associated with lower serum creatinine and decreased rates of renal replacement therapy (6 vs 10%) compared to controls. Like the study by Shaw et al, the difference was independent of severity of illness or total fluid volume administered. However, as mentioned by the authors, determining which component of lower-chloride fluid may have led to the observed effect is difficult, as there was simultaneous administration of lower sodium content, as well as increase in the administration of acetate, lactate, and gluconate. Importantly, this study showed no difference in mortality. [7][8]

Take home points: Administration of large volume of isotonic saline is associated with a metabolic acidosis. Animal models have demonstrated decreased renal perfusion with chloride containing fluids. Several retrospective studies indicate that chloride is an independent risk factor for mortality in acutely ill patients. More and more literature in humans seems to indicate that a ‘balanced resuscitation’ may decrease morbidity, and possibly mortality, in patients receiving large volumes of crystalloids as part of their resuscitation.  A single nonrandomized study demonstrated a correlation between low chloride fluids and decreased use of renal replacement therapy. Blinded, randomized, prospective studies are needed to further elucidate this observed effect.

Expert Commentary:

Dr. Schwarz, an Assistant Professor here at Wash U, and both an Emergency Physician and Toxicologist has provided some of his own thoughts on the topic. 

First, I’d like to thank Louis for picking a great topic and generating discussion about a very important subject.  I initially became interested in this topic a few years ago.  Originally, I was much more interested in the mechanism by which normal saline (NS) caused a non-anion gap metabolic acidosis, and that’s when I learned about the strong ion difference and a ‘balanced resuscitation.’  As a full disclosure while I found the pathophysiology really interesting, I initially didn’t think it had much clinical relevance.  However as more investigators have studied this, I’ve come to believe that my initial impressions were incorrect and changed my practice.

 The last time I reviewed the literature, I didn’t see a randomized, controlled trial comparing resuscitation with NS and lactated ringers in the ED.  However I do believe that there are studies out there that are applicable to the ED.  A retrospective study compared patients undergoing elective or emergent general surgery that received either NS or a ‘balanced fluid.’1  Unadjusted mortality and the number of patients developing major complications were higher in the group that received NS; after adjusting with propensity scoring, the mortality was no longer significantly different between the 2 groups.  However, patients that received NS were 4.8 times more likely to require dialysis. In a meta-analysis of patients with sepsis, patients that received a ‘balanced resuscitation’ had a lower mortality than patients receiving NS.2  The trend, however, was not significant.

In a promise to keep this short, I won’t review all the other literature that has been published on this topic and kept the discussion on the 2 articles that I did include short.  I’ll also concede that the literature is not perfect, and as I mentioned earlier, I’m also still waiting for that perfect ED-based study to be completed.  However the cost of NS or a ‘balanced solution’ such as lactated ringers is nearly equivalent.  I’m also not aware of significant complications from administering lactated ringers in most patients. So when the risks, costs, and benefits of implementing a ‘balanced resuscitation’ verses a standard resuscitation with NS are viewed together, I think there is enough evidence to consider changing your resuscitation strategy.

Now like many EDs, lactated ringers is not kept in our department.  It is on shortage but so is NS.  Neither of those are reasons not to use it.  So what do I do? Since I haven’t been able to convince pharmacy to keep lactated ringers in the ED yet, I do my best to guess early on which patients are going to need large-volume resuscitations.  If I think they are going to likely need more than 2-3 liters of fluid, I order additional lactated ringers from the pharmacy when I place their initial orders. In an hour after the patient has received the first few liters of NS, the lactated ringers should be there from the pharmacy.  If they need further resuscitation I can use it or return if they no longer need it.  For those that are interested to read more about this topic, I’d direct you to the upcoming May 2015 edition of Emergency Physicians Monthly. From my understanding, it’s brilliantly written! (Sorry for my shameless plug)


Jamtgaard References:
 [1] Kishen R, Honoré PM, Jacobs R, et al. Facing acid–base disorders in the third millennium – the Stewart approach revisited. International Journal of Nephrology and Renovascular Disease. 2014;7:209-217. doi:10.2147/IJNRD.S62126.
[2] Scheingraber et al. Rapid Saline infusions produces hyperchloremic acidosis in patients undergoing gynecologic surgery. Anesthesiology 1999;90;1265
[3] Chowdhury A et al. .  A randomized, controlled, double-blind crossover study on the effects of 2-L infusions of 0.9% saline and plasma-lyte® 148 on renal blood flow velocity and renal cortical tissue perfusion in healthy volunteers.  Ann Surg. 2012;256(1):18-24
 [4] Boniatti MM et al.  Is hyperchloremia associated with mortality in critically ill patients? A prospective cohort study. J Crit Care. 2011;26:175–179. doi: 10.1016/j.jcrc.2010.04.013
[5] Raghunathan K, Shaw A, Nathanson B et al. Association between the choice of IV crystalloid and in-hospital mortality among critically ill adults with sepsis*. Crit Care Med 2014; 42: 1585–91
[6] Shaw A et al.  Association between intravenous chloride load during resuscitation and in-hospital mortality among patients with SIRS. Intensive Care Medicine. 2014;40(12):1897-1905. doi:10.1007/s00134-014-3505-3.
[7] Waikar SS, Saving the Kidneys by Sparing Intravenous Chloride?.JAMA. 2012;308(15):1583-1585. doi:10.1001/jama.2012.14076.
[8] Yunos N et al. Association between a chloride-liberal vs chloride restrictive intravenous fluid administration strategy and kidney injury in critically ill adults. JAMA 2012; 308: 1566– 72.

Schwarz References
1. Shaw et al.  Major Complications, Mortality, and Resource Utilization after Open Abdominal Surgery: 0.9% saline compared to Plasma-Lyte. Ann Surg 2012;255:821-829.
2. Rochwerg et al. Fluid Resuscitation in Sepsis. A Systematic Review and Network Meta Analysis. Ann Intern Med 2014;161:347-355.


Submitted by Louis Jamtgaard +Louis Jamtgaard , PGY-3
Faculty Reviewed by Evan Schwarz @TheSchwarziee 

Assess the pipes, Carotid VTI and fluid responsiveness


Clinical Scenario:

You are working in the ED when a 75 yo F hx of CHF, DM presents with fever, cough, and hypoxia and hypotension. You are concerned for sepsis with presumed pneumonia as the source. You initiate volume resuscitation and start broad spectrum antibiotics.  Your  patient's BP initially responds to fluids, but now after your 3L your patient is still hypotensive. You perform bedside US of the inferior vena cava (IVC) with equivocal findings. You wonder, is there another way to perform rapid bedside ultrasound for volume responsiveness?  You remember a recent paper about carotid velocity time integral (VTI) , and begin to investigate

Literature review:
It seems that predicting volume responsiveness is the never-ending tale in critical care medicine, as numerous methods have been proposed over the past several years with varying degrees of success. With the expansion of ultrasound, measuring IVC collapsibility has been one of the more popular methods utilized in the emergency department. However, measuring the IVC can often be limited by body habitus, excessive intra-abdominal gas, respiratory variation, and operator experience. (1) Measuring IVC collapsibility at greater than 50% has been shown to correlate with a CVP of less than 8mmhg, and a lower CVP has been associated with volume responsiveness, but a higher CVP does not exclude volume responsiveness. (1) A recent paper by Marik et al described the novel use of Carotid VTI and passive leg raise (PLR) as a marker of volume responsiveness in hemodynamically unstable patients.  The benefit of  PLR is that it produces a hemodynamic response similar to a 200-300ml bolus, is relatively easy to perform, and is rapidly reversible.
 Courtesy Ultrasound Podcast
 By combining PLR with dynamic ultrasound, Marik et al sought to create the ideal non-invasive method of determining volume responsiveness.  They demonstrated that a 20% increase in carotid VTI had a sensitivity and specificity of 94% and 86% respectively for predicting volume responsiveness (a patient with a stroke volume increase of greater than 10% was considered volume responsiveness). 
This study was limited in that it was nonrandomized, nor blinded, and complete data was available for only 34 patient. (2)  Mike and Matt from the Ultrasound podcast provide an excellent review and explanation on how perform VTI that you can find here @ Ultrasound podcast

Take home points:
Studies have shown that only 50% of hemodynamically unstable patients are volume responders. Appropriate fluid resuscitation in sepsis is associated with improved outcomes, while excessive fluid administration is associated with increased ICU LOS and mortality. Determining fluid responsiveness is difficult but VTI combined with PLR appears to have both a high specificity and sensitivity for predicting volume responsiveness.  More studies will be needed to demonstrate validity of this method. 

Submitted by Louis Jamtgaard, PGY-3 @Lgaard
Faculty Reviewed by Deb Kane 


References

1)Nagdev A et al . Emergency department bedside ultrasonographic measurement of the caval index for noninvasive determination of low central venous pressure. Ann Emerg Med. 2010 Mar;55(3):290-5. doi: 10.1016/j.annemergmed.2009.04.021. Epub 2009 Jun 25.

2) Marik P et al. The use of bioreactance and carotid Doppler to determine volume responsiveness and blood flow redistribution following passive leg raising in hemodynamically unstable patients.
Chest. 2013 Feb 1;143(2):364-70.


Cardiac arrest, add antibiotics to the kitchen sink?

Clinical scenario:
You get a page out for a 55 yo M in cardiac arrest, EMS reports PEA on their arrival, patient has received 3 rounds of epi prior to arrival, patient achieves return of spontaneous circulation (ROSC) after 5 minutes of ACLS while in the ED. When family arrives they report the patient had been feeling unwell for several days and had a significant cough.  No obvious infiltrate was seen on initial chest xray.  The patient's BP is stable on an epi infusion. You admit the patient to the ICU. Your attending requests drawing blood cultures and starting the patient on broad spectrum antibiotics, and cites data stating antibiotics improves mortality in out of hospital cardiac arrest. You perform a brief literature review. 

Literature Review:
Out of hospital cardiac arrest (OHCA) has a very high mortality rate, where approximately only 23% make it to the hospital alive, and 7.6% survive to hospital discharge. (1) The most common etiology of out of hospital cardiac arrest is presumed to be myocardial in origin. However, several retrospective studies indicate that sepsis and bacteremia may also be a significant contributing factor to OHCA. A study by Coba et al in published in 2014 performed a prospective study to identify the incidence of bacteremia in OHCA patients. They enrolled 173 patients, where all patients had two sets of blood cultures drawn, 77 patients met exclusion criteria (trauma, pregnant, pediatric, single positive culture of skin flora). The overall incidence of bacteremia was 37% (65 patients). The most common bacterial species cultured were streptococcus and staphylococcus and Ecoli and klebsiella for gram positive and gram negative bacteria respectively. Bacteremic OHCA patient had significantly higher lactates, lower pH, and more frequent use of vasopressors. Notably the ED survival was significantly lower in the bacteremic patients (25%) compared to nonbacteremic patients (40%). However, 28 day mortality difference was insignificant in bacteremic vs nonbacteremic patients (93.8 vs 92.6%). The figure below by Coba et al lays out the proposed inter-relationships between bacteremia and sudden cardiac arrest. (1)
Proposed association between bacteremic infection and sudden cardiac arrest. From Coba et al.



Although there is very little data examining pre-existing bacteremia in OHCA, there has been a significant amount of research studying infection following ROSC in OHCA. The most commonly cited sources of post ROSC infection are lung possibly from aspiration during arrest, or gut likely from translocation of flora secondary to low flow state during arrest.  Davis et al performed a retrospective analysis on 138 patients admitted to the ICU following OHCA, and showed that 97.8% had at least one positive mark of infection within 72 hours (positive blood culture, consolidation on cxr, CRP greater than 100 or wbc greater than 11 or less than 4 x10^9 ). In this study approximately 38.4% of patients received antibiotics during the first 7 days of their ICU stay. The authors showed that mortality was significantly lower among those receiving antibiotics versus those not receiving antibiotics (56.6% versus 75.3%). However, highest mortality was within the first three days, and for patients who survived to day 3, there was no difference in mortality between those who received antibiotics already and those who had not. (2)

Take home points:
OHCA is typically presumed to be a primary myocardial event, however there is some data to suggest that sepsis is potentially a significantly under reported cause. Furthermore, there is also data to suggest that following ROSC, infection is quite common, and antibiotics may reduce early mortality. However, caution must be taken, as of yet there are no RTC's comparing prophylactic antibiotics versus placebo in OHCA.

Expert Commentary:

Dr. Holthaus one of our own critical care and sepsis guru's was nice enough to provide some of his own thoughts on this topic, and cardiac arrest in general. 

Things we'd like to see examined in future cardiac arrest RCTs:
1) Antibiotics during arrest - push dose, timing, coverage.
2) Propofol - control for this or exclude as a variable since it has been shown to cause some mitochondrial dysfunction and may be thwarting potential resuscitation benefits.
3) Epi dosing - frequency, continue 1mg push dose vs lower dose vs maximum that is less likely to cause or further exacerbate either ischemic or post-inflammatory cardiomyopathy.
4) Vasopressin-Steroid-Epi- for ED arrest. Link to VSE study in JAMA . VSE (vasopresson-steroids-epi) better than Epi alone for in-hospital arrest Vasopressin (20u, q 3-5min, max 100u, w 1 mg epi pushes)-Methypred (40mg IV x1) w better ROSC (84% vs 66%) and better CPC1/2 survival (14% vs 5%).  Major caveat is time to ACLS was very low at 2 min for both which is way faster than many we see in the ED that are frequently >10 min downtime before EMS.  Hypothesis generating, re-hinting at potential beneficial role of vasopressin and steroids for shock (like sepsis). 
5) ED ECMO for cardiac arrest or refractory/severe shock
6) Remote ischemic conditioning immediately after ROSC- 5 min thigh BP tourniquet to >20mmHg above SBP then deflate, repeat 3-4 times, reportedly induces systemic circulation of a protein that blocks CNS/cardiac opening of the "mitochondrial permeability transition pore" which is the final common pathway for ischemic reperfusion injury).  On recent ED ECMO podcast (Shiner-Bellezo) Link to podcast Remote ischemic conditioning 

Personally since everything (ACLS) isn't getting much results, if I can remember to I will do Vasopressin-Methypred-Epi dosing, I am less excited about a lot of epi (ie 3 pushes tells me if they're trending toward making it or not), I've pushed zosyn and then hung vancomycin in a code (after learning about Coba study). I have generally avoided propofol in past because of known myocardial suppression, and now with concern for mitochondrial insult, I just use fentanyl/versed. In addition I  will try thigh remote ischemic reconditioning, and continue targeted temperature management to 33-36C while hoping for ED ECMO (which I think will be the biggest game changer)  

Submitted and Edited by Louis Jamtgaard PGY-3 @Lgaard
Faculty Review by Chris Holthaus

References

 1) Coba V et al. The incidence and significance of bacteremia in out of hospital cardiac arrest.
Resuscitation. 2014 Feb;85(2):196-202. d

2)Davies K et al. Early antibiotics improve survival following out-of hospital cardiac arrest.Resuscitation 2013 May; 84 (5) : 616-9.


Sour Milk: Antibiotic Coverage For a Breast Abscess

Clinical scenario:  

Your patient is a middle-aged female who was brought in from home for altered mental status.  As EMS is moving her over to the stretcher, they say: "this lady has some kind of infection on her breast ... I saw it when I went to do her EKG".  The patient is febrile to 39.3, tachycardic in the 120’s, but maintaining a blood pressure of 150’s/80’s.  She has a large, right- sided breast abscess with some spontaneous drainage.  Clearly, this patient has severe sepsis and she needs IVF, antibiotics, and source control.



Clinical Question:  

What is the most appropriate antibiotic choice for coverage of a breast abscess?  Obviously, the patient needs an I&D, but in the meantime, what typically is growing in there?  Should anaerobic coverage be routine?




The Literature:
There are several articles that address culture results from breast abscesses in the era of community acquired MRSA.  Here are two:
One article [1]  reports the culture results of 189 drained  breast abscesses from both lactating (LA) and non-lactating (NL) women at a single center from 2003-2006. In both cases, Staph aureus was the most commonly isolated organism (67.7% from LA, 30.5% from NL, and 42.6% of all cultures overall)  The majority of these S. aureus  isolates were MSSA not MRSA (39 vs. 3.7%).  Importantly, the second most commonly isolated class of bacteria were mixed anaerobes (13.7% overall), followed by anaerobic cocci (6.3% overall).  The authors, therefore strongly suggested that anaerobic coverage be a component of all initially empiric coverage for breast abscesses.
A second article [2] similarly tracked the culture results of 46 drained breast abscesses in a community setting. Staphylococcus aureus was  again the most common aerobic organism, present in 12 cultures (32%).  In contrast to the previous article,  58% of the S. aureus  isolates  were MRSA. The remaining positive cultures yielded Coag-negative Staph (16%), diphtheroids (16%), and Pseudomonas aeruginosa (8%).    This study was severely limited for estimating the prevalence of infection with anaerobic bacteria, as only 8/46 abscesses had swabs sent for anaerobic culture. Of these 2/8 (25%) grew anaerobes.

Take-home:

In addition to arranging for I&D, cover for at least Staph aureus (MRSA if you suspect it) and Anaerobes when treating breast abscesses.  

- If the person is sick and septic like our clinical scenario, cover broadly for MRSA, anaerobes and pseudomonas as well.  Possible options include:
               Inpatient - Vancomycin & Zosyn OR Vancomycin & Unasyn.
               Outpatient - Augmentin (if nursing) OR Bactrim/Flagyl if MRSA suspected. 

References:
[1] Dabbas, N., Chand, M., Pallett, A., Royle, G. T., & Sainsbury, R. (2010). Have the Organisms that Cause Breast Abscess Changed With Time?––Implications for Appropriate Antibiotic Usage in Primary and Secondary Care. The breast journal, 16(4), 412-415.
[2] Moazzez, A., Kelso, R. L., Towfigh, S., Sohn, H., Berne, T. V., & Mason, R. J. (2007). Breast abscess bacteriologic features in the era of community-acquired methicillin-resistant Staphylococcus aureus epidemics. Archives of Surgery, 142(9), 881-884.

Contributed by Maia Dorsett, PGY-3
Faculty Reviewed by Stephen Liang