cerebral edema in diabetic ketoacidosis, bydr. michael agus. hi, and welcome. i'll be speaking today aboutdiabetic ketoacidosis or dka. my name is michael agus. i am a pediatric intensive care doctorand a pediatric endocrinologist at boston children's hospital. i'm the director of themedicine critical care program here at children's and an assistant professor at harvard medicalschool.
now, what does a patient succumb from? whatdo they die of? in general, cerebral edema. literature review. there have been three major studies in theliterature over the past several years, fairly large studies, but retrospective, totalingabout 11,000 patients. and together, they
seem to demonstrate that there's about a 1%incidence of significant cerebral edema among patients who present with dka. among thosewho get significant cerebral edema, about 40% will die or have a severe neurologic outcome. if you do the math, that's a net populationmortality of about 0.4%. a more recent study by a colleague of mine here at children's,dr. decourcey, looked at a phis database review. phis is an administrative database we haveaccess to in the us. was able to look at a total of just over 37,000 dka patients ina single study. and to our surprise, we found a significantincreased incidence of significant cerebral edema defined as edema that prompted therapyof either mannitol or hypertonic saline by
a practicing clinician. fourfold increasein the incidence, but a 10% mortality. so a significantly reduced mortality. and onthe face, that looks like good news. we're treating more patients and they're doing better.but again, when one does the math, basically we haven't changed the ultimate mortalityat all. we're treating four times as many patients.they're one quarter as likely to have a bad outcome, and ultimately 0.4% population mortality.so despite our advances over the decades, it's not clear that we've made a major impactin overall mortality. but these numbers are low, and it's easy to have had even a 30%or 40% effect on mortality and have that really be difficult to detect with these studies.
risk factors for cerebral edema. now, what are the known risk factors for cerebraledema? well, again back to dr. glaser's paper, from 2001. now, she identified four majorrisk factors in children with significant cerebral edema. three of them are shown here.a high bun, a low paco2, and treatment with bicarbonate. now we know that a high bun isassociated with someone who's dehydrated. how about a low paco2? that's someone whohas a very low ph. and they're hyperventilating and dropping their paco2 so that they canblow off some of that additional acid. and who gets treated with bicarbonate? those whothe lowest ph. so although these were independent risk factors,the reality is that all three of them correlate,
really quite directly, with severity of illness.and while they may be independent risk factors, as clinicians, we look at patients and aremessage is the more dehydrated, the more acidotic they are, the more at risk they are. the fourthrisk factor was the one that really kind of threw us for a loop. fourth risk factor, ifyou notice, demonstrates a lower relative risk when the serum sodium rises by more than5.8 per hour. the prior three risk factors i showed you, if they were true, the relativerisk is increased. in this case, when it's true the relativerisk decreases. what that suggests is that a serum sodium rise of 5.8 per hour is a goodthing. now, we measure serum sodium every two hours. and that would mean that we wouldhave a change of 12, almost 12 millimoles
per liter, over a two hour period. and that meanspatient comes in with serum sodium 125, you check it again a couple hours later, it's137. that is an extraordinary change. and as a clinician, i would posit that this isnot, in my experience, a degree of change in serum sodium that we see very commonlyat all. so what do we with this risk factor? does it only apply when the serum sodium risesso dramatically to an extent that we almost never see? instead, we can look at it from a little bitof a different point of view. if you come over here and look at this graph i'll showyou what i mean. so if we look at the evolution of sodium and glucose concentrations overtime, in the setting of dka, we can think
about it from a point of view of being aboveand below normal. so if we consider this to be a normal range, and we start with a glucoseconcentration that is quite high, we know from our discussion that the serum sodiumwill be relatively low. and the idea in resolution of dka is thatas the glucose comes down into the normal range, the serum sodium ought to rise. thereason the serum sodium is low to begin with, you will recall, is that glucose is an osmoticagent that pulls water into the bloodstream, deluding everything that's in the bloodstream.so it's going to dilute sodium. it's going to dilute potassium and chloride as well.but sodium is the most obvious, the most prominent. and as glucose comes out of the extremelyhigh range, into the normal range, the exact
same thing ought to happen with sodium. the risk factor that we were discussing earlier,is when glucose begins to come down but sodium does not come up at the same rate. if sodiumcomes up a little bit and kind of flattens out that is considered invocation of the riskfactor that sodium has not risen fast enough. and this is considered a risk factor for developingsignificant cerebral edema. one way to follow the relationship between the hyperglycemiaand the hyponatremia, over time, is to look at the corrected serum sodium. one way to think about the sodium changesthat we talked about are to look at a value that we refer to as the corrected sodium.corrected sodium answers the question, what
would the serum sodium be if glucose was normal?we use a formula to calculate this. and we say, we take the serum sodium, and look atthe serum glucose, and we say, see how many 100s over 100 is the serum glucose. and multiplythat times 2 and add it to the serum sodium. and we get our corrected sodium. for example,serum sodium is 125 and the blood sugar is 900. well, that's 800 over 100. so 8 times2 is 16. we add 16 to the 125, we get 141. so the corrected sodium in this example is141, while the measured sodium is 125. a question that's often asked in the situationis, which is a real? which is the real serum sodium? the answer is, very definitively,that the serum sodium is 125. the sodium that the brain is seeing, it's 125. what wouldthe sodium be if the glucose was normal? 141.
but the glucose isn't normal. the glucoseis 900. and so, right now, the serum sodium is quite low. the way we use the measured,the way we use the corrected sodium to think about this risk factor is to say, the correctedsodium should be normal and stay normal. if, as we saw on the graph, the glucose comesdown and the serum sodium does not rise adequately, that should be reflected in a corrected sodiumthat is low. so if there were corrected sodium drops below 137, 138, we really need to getourselves to the bedside, do a thorough neurological exam, and determine whether or not this patienthas symptoms of significant cerebral edema. and think about treating at that point. butthis risk factor is really only present when this the sodium does not rise sufficiently.
now you may ask, why would i not just givesodium? well, there's nothing, i don't have any data to suggest that one should not treatthis scenario with sodium. but i will tell you that every study that has ever lookedat rate of sodium administration has not found it to be a predictor of significant cerebraledema. we are physicians, and when something is low, and it's a bad sign, then our inclinationis to treat it, to bring it up. the best example, the best analogy i could think of is a patientwith meningitis. let's just say that an eight-year-old hasmeningitis and has a csf white count to 10,000. now, you know that a high csf white countis a very poor prognosticator and significant risk factor. i will give you the choice. youcan either treat this patient with ceftriaxone
or you can plasmapharese the csf. which areyou going to choose? you're going to choose the ceftriaxone because you understand thatthe problem here is that there are bacteria that are eating part of the brain, and causinga massive amount of inflammation that is putting the patient at risk for a severe outcome.the white cells are only a symptom of the underlying pathology. i would posit to you that is really the exactsame thing that's going on here. when the sodium fails to rise normally it is not theproblem. it is a symptom of the problem. and it is very likely connected to inappropriateadh secretion that is triggered by an injured brain. and a brain that is becoming more andmore, either ischemic or significant swelling
with a significant pressure building up, triggeringrelease of adh, an inability to allow the kidney to clear adequate free water. thatdoesn't get any closer to a treatment approach, but it does point out that the failure ofsodium to rise has not been shown to be a component of the pathology, rather a symptomor a sign of the pathology. having said all that, when we see serum sodium fail to riseor a corrected sodium that begins to drop, the right approach, because we don't haveany better data, is to increase the concentration of sodium chloride in the fluids that we'redelivering. and we'll talk about that more in a minute. the other findings of this study were interestingin that there was a whole slew of the risk
factors that were not identified as beingsignificant. you see those on the screen. and importantly, they include initial glucose.it's very helpful for clinicians, over the years, to understand that a glucose of 1,200is no more risky than a glucose of 600. what matters is the degree of acidosis, degreeof dehydration, the factors that we talked about before. as of this study, deciding on givingan insulin bolus didn't attribute increased risk. the rate of insulin infusion didn'tseem to matter. rate of iv fluid, even the rate of sodium administration, as i mentioned. one important caveat to a retrospective study,however, is that one can only measure within the range that was done. now rate of iv fluid,for example, was not shown to be a significant
risk factor. but in animal models, we knowthat if we give a massive amount of fluid to patients with dka we will engender a lotof significant cerebral edema. and so why did that not happened here? why was that notnoticed here? and i think the answer is that patients in this study didn't give a massiveamount, didn't receive a massive amount of iv fluid. and as result, in a retrospective study whenone is practicing within reason, you may not pick up a lot of unreasonable practices. there'sa similar study done by dr. julie edge, in europe. she found similar risk factors, buttwo slightly different ones that are worth mentioning. one, is she noted a trend towardshigher risk in younger patients. and that's
something that intuitively we believed. thata younger patient with a full diaper and good urine output is not considered to be sickpatient. we see that all the time. and also, that new onset diabetics are at higher riskfor developing significant cerebral edema. again, presumably due to delayed presentationbecause the families are not skilled in noticing the symptoms of evolving dka. diagnosis of dka. now, a patient is admitted. they're in yourunit. what signs are you really looking for to diagnose cerebral edema at the time, andto consider therapy? well, dr. muir reviewed a bunch of really severe cases, and came upwith a set of diagnostic criteria. now, i
don't advocate for the strict use of thisparticular protocol and particular formula, but the lists are quite helpful and quiteinformative. on the diagnostic column, we have some symptomsthat i think all of us would jump in and treat pretty instantaneously. if patient's unresponsiveto pain, if patient has decorticate or decerebrate posturing, if a patient has a new cranialnerve palsy. and finally, if they have a neurogenic respiratory pattern, which is really somethingclose to agonal respiratory pattern. and of course, that is the third component of cushing'striad. the second component being bradycardia, anddr. meir actually quantifies this is a change of greater than 20 beats per minute from baseline.and then the first component cushing's triad,
which is actually extremely common, is hypertension,quantified by him as a diastolic blood pressure greater than 90. despite patient presentingwith hypovolemia and dehydration, they curiously have increased systemic vascular resistance,increased tone, and have relative hypertension. the other major risk factors are interesting.significantly altered mentation, mental status, and then age dependent incontinence. thisone struck me when i first read the paper, because patients will often come in in theevening and at night, and you take a 5-year-old who's been toilet trained and give them theworst night of their life, you keep them up all night, you give them a huge amount offluids, they have a ph of 7.1, and they wet the bed, is that really meaningful? and itturns out from this data that it is. and so
that's become one additional item that wefocus on in our hospital, and that nurses and house staff are really quite attuned to. the minor criteria are pretty common. vomiting,headache, lethargy, not being easily aroused. in my own estimation when i do a neurologicexam for a patient, i will stimulate the patient to whatever extent they need. but if evenafter a significant simulation, they are then able to answer questions appropriately, meaningbirthday, where they are, what they're doing here, name, that sort of thing, then i tendto allow their mental status to be characterized as acceptable. he also noticed that childrenat a young age were at increase risk, as we've seen before.
development of dka. when does one get this significant event,the cerebral edema? well in dr. glazers paper she noticed that several had it when theywalked into the ed. meaning, it can't be true that all significant cerebral edema is relatedto therapy. and it leads to the idea that there's something about the patient, somethingabout-- we've mentioned before, the size of the skull, the amount of extra axial space. and it may have more to do with the severityof the acidosis. and how long they've been ill. all these factors, we haven't been ableto identify, but what we can tell you is that some patients will get significant edema withoutany therapy.
the bulk of events, meaning development ofsignificant edema, occur within the first six hours. by hour 12, 13, you're really largelyout of the woods. but there are some stragglers. and those tend to be patients who just aren'tquite responding fully. their course ends up being a little bit longer than you mightexpect. but those are few and far between. therapies for dka now what happens if we believe we have a patientwith significant cerebral edema? we have a blown pupil, we have decorticate posturing,we have a patient who's unresponsive. well, then we need to call for our appropriate supports,local supports, hospital supports, so that if we do need to take an airway, we're ableto do that. but we also call for the appropriate
therapy. now, what is the appropriate first line therapy?there are two competing therapies out there right now, one is mannitol and one is 3% hypertonicsaline. while 3% hypertonic saline has some reasonable data that has been reported withit in traumatic brain injury, there are hardly any data reported with its use in diabeticketoacidosis. as a result and also as a result of some data i'll show you in a minute, iuse first line therapy as mannitol. we use 1 gram per kilo and we infuse it over 15 minutes. please remember that mannitol sitting on ashelf for a long period time can form some crystals, and so when one draws it from thevile, one needs to use a needle with a filter.
and it is very well tolerated. it does resultin increased urine output during the hours that follow, and so one has to keep trackof that hypovolemia that may follow my mannitol administration. second line therapy if mannitol does not workis 3% saline. there is some debate on what the actual recommended dose of 3% saline is.an equimolar dose to the mannitol does that we've use over the years is 5 ml per kilo.now, again, we infuse it over 15 minutes. however, in this particular case, if a patient'smental status improves, i would halt the infusion immediately. one does not want to run therisk of central pontine myelinolysis by acutely raising the serum sodium any further thanneeded to restore a reasonable mental status.
if the patient does not respond and they losetheir airway reflexes, intubation may be warranted and one wants to use an agent that takes intoaccount the increased intracranial pressure, for example, thiopental with or without theuse of lidocaine. after intubation, i would hyperventilate down to the end-tidal co2 thatthe patient had prior to losing their mental status. hyperventilation, while not a standardtherapy for increased intracranial pressure in cerebral edema is exactly what this patienthas been doing for many hours prior to losing their mental status, and so while one doesn'twant to go beyond what they were doing, one wants to go back to what they were doing andthen slowly raise it over time, as they would do if they were conscious.
finally, head ct is really not necessarilypart of the algorithm. the decision to treat ought to be made based on clinical symptoms.the head ct can show significant cerebral edema on virtually any patient with dka, andso the presence or absence on ct is not going to guide therapy. if patient has edema butis alert and awake, you're not going to treat, and if a patient is obtunded and non-responsiveand the ct is negative, you are going to treat. and so again, treat based on clinical symptomatology. the debate between mannitol and 3% salinereally is an active one, and in looking at some retrospective data over the last decade,it looks like in the united states the use of 3% saline has gone up quite dramatically,especially in the last few years. however,
in looking at the survival or the odds ratiofor survival when using one therapy over the other, there is a hint in this admittedlyretrospective database study that there is an increased mortality with the use of hypertonicsaline as first line therapy. that has contributed to our decision here to use mannitol exclusivelyas first line therapy. therapies for dka: update. at the end of 2013, we published a retrospectiveanalysis of mortality in dka which revealed some new data which is worth considering whenchoosing a therapy for cerebral edema in dka. the study looked at a total of 43,000 patientswith dka in an administrative database. patients were identified as having cerebral edema asa complication of dka based on the codes that
physicians and hospitals billed at the endof the admission. we were able to identify patients who had a ct scan, as well as thosemechanically ventilated or admitted to an icu. over the course of a decade, it became clearthat the use of 3% hypertonic saline has gone up dramatically over the last few years. whilestarting close to 0% in 1999, it had gone up to roughly 50% of cases by the year 2008,2009. concomitantly, the use of mannitol as a solo therapy has gone down. the use of thetwo together remains relatively infrequent. over that same time period, mortality in patientswith dka has also gone down quite significantly, from 0.47% in 2000, to a low of 0.8% in 2009.this is a logistic regression, taking account
the effective discharge year, adjusted forhospital clustering, and is statistically significant. the odds ratio of mortality inpatients treated for dka, when looked at, mannitol versus hypertonic saline, is, infact, increased slightly with the use of hypertonic saline alone to an adjusted odds ratio of2.7, which is statistically significant even after taking into account other significantfactors, such as male gender, use of mechanical ventilation, and ct scanning. this has allowed us to create some additionalthoughts that should be considered when treating someone for dka. it's noted that this is thelargest retrospective study that's ever been conducted in dka, and that, over the decadeit was studied, the rate of treatment for
cerebral edema has increased four-fold, whilethe overall mortality in dka has decreased by 80%. however, this decrease was not associatedwith the marked rise in use of 3% saline as therapy. in fact, there is a slight increasein mortality among those who receive 3% saline after conducting an adjusted analysis.
Define Diabetic Ketoacidosis,these data ought to be considered when choosinga rescue therapy for a cerebral edema in dka. i thank you very much for listening, and lookforward to interacting with you online with any questions that develop. thank you. please help us improve the content by providingus with some feedback.
Now that you know how to control your problem, you are greater willing to create the appropriate selections and adjustments in your life. Understand that these guidelines are only relevant if you are using as many of them as you possibly can, and thus you should begin now to see timely results.
No comments:
Post a Comment