strong ion difference and gap predict outcomes

Strong ion difference and gap predict outcomesafter adult burn injury

Allison E. Berndtson, MD, Tina L. Palmieri, MD, David G. Greenhalgh, MD,and Soman Sen, MD, Sacramento, California

BACKGROUND: The strong ion difference (SID) (apparent [SIDa] and effective [SIDe]) and strong ion gap (SIG) provide a comprehensivemethod of evaluating acid-base status in critically ill patients. The SID is the difference between strong cations and stronganions in plasma, while the SIG demonstrates the presence of unmeasured ions. This approach accounts for changes in apatient’s protein status, which is particularly important in those with burn injuries. We hypothesized that the SIDa, SIDe, andSIG during the first 72 hours after admission would be predictive of mortality in burn patients.

METHODS: This study is a retrospective review of adults with 20% or greater total body surface area burns admitted during a 7-year periodto a regional burn center. SIDa, SIDe, and SIG were calculated at admission and for the first 3 days. These results were thencompared with Acute Physiology and Chronic Health Evaluation II (APACHE II) and sepsis-related organ failure assessment(SOFA) scores.

RESULTS: A total of 113 patients met the criteria and had full data sets, with mean T SEM age of 45.4 T 1.4 years and total body surfacearea burn of 41.4% T 1.6%. Mortality was 27.4%. At admission, APACHE II remained most predictive of mortality (p =0.006). However, admission SIG (SIDa j SIDe) was also predictive of mortality on multivariate analysis (odds ratio, 1.11).Day 1 SIDa (Na+ + K+ + Ca2+ +Mg2+jClj) and SIDe ([1,000� 2.46� 10j11� PaCO2/10

jpH] + [[albumin]� (0.123� pHj

0.631)] + [[PO4] � (0.309) � pH j 0.469)]) were also associated with mortality (odds ratio, 1.16 and 1.13 respectively), andSIDe with length of stay and ventilator days (p G 0.05).

CONCLUSION: The SID and SIG are predictive of mortality, hospital length of stay, and ventilator days in adult burn patients. They alsoelucidate complex acid-base disorders. (J Trauma Acute Care Surg. 2013;75: 555Y561. Copyright * 2013 by LippincottWilliams & Wilkins)

LEVEL OF EVIDENCE: Prognostic study, level II.KEY WORDS: Strong ion difference; strong ion gap; burn; acid-base status.

A nalysis of acid-base status is an important part of moderncritical care, particularly as a marker of disease severity. It

also enables a focus on perfusion and resuscitation at thecellular level. A variety of methods for analyzing acid-basestatus exist, each with notable strengths and weaknesses. Themost direct analysis relies solely on pH and [HCO3j]. Whileeasy to understand, this method is limited by the use of de-pendent variables, which may be confounding. The next iter-ations, anion gap and base excess, include more variables butare subject to the same limitations. These have been extensivelystudied as potential markers of mortality in burn patients, withlimited success. Kaups et al.1 demonstrated that an admissionbase deficit greater than 6 correlated with a markedly increasedmortality rate. This was followed by Cancio et al.,2 who foundthe mean base deficit during the first 2 days after burn injuryto be useful in classifying a specific subset of patients withmidrange burns, but that it had only a negligible effect on

overall predictions. A second study by the same author3 tied theworst base deficit during the 24 hours after admission tomortality as well as to an increase in fluid resuscitation volume.Jeng et al.4 investigated both base deficit and serum lactateand found persistently elevated levels of each during the first48 hours after admission, indicating ongoing hypoperfusion,although he did not tie these to mortality. These elevated levelsdid not correlate with more traditional resuscitation measuressuch as urine output and mean arterial pressure.

More recent studies have evaluated the role of woundtissue pH and PCO2, sigmoid colon PCO2, and gastric pH andPCO2 detected via sensors, as well as burn wound perfusion asmeasured by laser Doppler imaging.5,6 Both studies hadpromising results in rats and humans and suggest a future use inguiding end points of resuscitation but do have some limitationsowing to technologic features (implanting sensors, holding en-teral feedings before gastric measurements, removing dressingsbefore laser Doppler assessments, etc.) We sought a formula-based method not subject to technical limitations but realizedthat none of the previously described methods account for ab-normal protein status, which is often pronounced in the criticallyill and particularly in burn patients. One solution is the strong iondifference (SID) method described first by Stewart7 and laterexpanded by Fencl et al.8

The SID theory of acid-base analysis is based on thepremise that only three independent variables control bloodpHVthe partial pressure of carbon dioxide (PCO2), the SID, and

AAST 2011 PLENARY PAPER

J Trauma Acute Care SurgVolume 75, Number 4 555

Submitted: September 7, 2011, Revised: June 27, 2013, Accepted: June 27, 2013.From the Department of Surgery (A.E.B., T.L.P., D.G.G., S.S.), University of

California-Davis; and Department of Burn Surgery (T.L.P., D.G.G., S.S.),Shriners Hospitals for Children Northern California, Sacramento, California.

This study was presented at the American Association for the Surgery of Trauma70th Annual Meeting, September 14Y17, 2011, in Chicago, Illinois.

Address for reprints: Allison E. Berndtson, MD, 2315 Stockton Blvd, RoomOP 512,Sacramento, CA 95817; email: [email protected].

DOI: 10.1097/TA.0b013e3182a53a03

Copyright © 2013 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

mailto:[email protected]

the total weak acid concentration ([ATOT]). A disturbance toone of these three variables is compensated via changes in theothers, and these ultimately control the concentrations of de-pendent variables such as [H+] and [HCO3j]. The SID itself isthe sum of all strong cations in a solution (Na+, K+, etc.) minusall strong anions (primarily Clj). In normal plasma, strongcations outnumber strong anions, leading to a SID of ap-proximately 40 mEq/L to 42 mEq/L.7 However, by the law ofelectroneutrality, plasma cannot be charged, so this excess ofcations is balanced by the negatively charged weak acids [PCO2]and [ATOT].

Clinically, there are two methods of estimating the trueSID in plasma. The first, the SID apparent (SIDa), is calculateddirectly from measured strong ions. The second, the SID ef-fective (SIDe) is calculated from the counterbalancing negativeweak acids. In this case, [ATOT] is the sum of [PO4] and [al-bumin], which make up the majority of weak acids in vivo.Other proteins such as globulins make an insignificant con-tribution to the weak acid component.8,9 Either of thesemethods can be used to estimate the true SID, and in an idealsituation where all variables are known, the resulting numberswill be equal.

If these two estimates of the SID are not equal, then astrong ion gap (SIG) is present. The gap (SIG = SIDaj SIDe)exists in any instance where the two estimates are differentowing to unmeasured ions. An SIG of greater than 0 indi-cates that there are unmeasured anions present that werenot captured in calculating the SIDe. They must be presentby the law of electroneutrality to counterbalance the positiveSIDa. Similarly, an SIG of less than 0 indicates unmeasuredcations. It is this ability to identify when unmeasured ions arepresent that makes the use of the SIG method promising incritical care.

The law of electroneutrality also allows the SID to affectpH, as changes to the SID are balanced by alterations in thedissociation of [H+] ions. For example, iatrogenically raisingthe [Clj] concentration via administration of normal salinecauses a decrease in SID owing to the overabundance of stronganions. This is counterbalanced electrically by a dissociationof [H+] ions, leading to a hyperchloremic metabolic acidosis.This also demonstrates the dependent role of [H+] to changesin the SID.

The correlation of SID with outcomes in critically illpatients has previously been studied in patients with majorvascular trauma.10 This retrospective trial found that the SIGand anion gap were statistically identical and outperformedother measured variablesVsuch as Injury Severity Score (ISS),pH, lactate, and standard base excessVfor the prediction ofmortality. This demonstrated the importance of unmeasuredions in acid-base balance and their utility in prediction ofoutcomes in critically injured patients. In the final multivariatemodel, SIG and ISS correlated most strongly and indepen-dently with mortality. Like trauma, burn patients frequentlyhave complex acid-base abnormalities, but the utility of theSID has not been studied in detail in this population. Thisbackground leads to our hypothesis that the SID and SIGmeasured during the first 72 hours would be predictive ofoutcomes in critically ill adult patients with severe burninjuries.

PATIENTS AND METHODS

We used our institution’s national burn repository data toretrospectively identify all adult (Q18 years) patients admittedto our regional burn center between August 31, 2003, and June30, 2010, with a 20% or greater total body surface area (TBSA)burn, for a total of 303 patients. Subjects were then excludedfor the following reasons: preexisting dialysis-dependent renalfailure (1), transfer greater than 72 hours after injury (7),comfort care within 72 hours of admission (46), dischargedalive within 72 hours of admission (2), toxic epidermal necrolysisor other skin disease as the reason for admission (23), andincomplete data (111). This left 113 patients for the analysis.

Demographic information and burn data (%TBSA, burntype, inhalation injury, and dates of burn, admission, anddischarge) were collected. Assessed outcomes included sur-vival, cause of death where applicable, hospital length of stay(LOS), total ventilator days, and the development of acute renalfailure (ARF) defined as the need for hemodialysis during thehospital admission.

Acute Physiology and Chronic Health Evaluation II(APACHE II) scores were calculated on all patients using theworst values available in the first 24 hours after admission foreach parameter.11 Sepsis-related organ failure assessment(SOFA) scores were also calculated for hospital Days 1, 2, and3 after admission using theworst values on each calendar day.12

Creatinine (Cr) was collected at admission and daily frommorning laboratory tests.

The SIDa, SIDe, and SIG were calculated both at ad-mission (using the first available set of laboratory values) andon the first three hospital days after admission (using the firstmorning laboratory values each day). This method was chosento maximize time from previous electrolyte replacement andachieve as much of a ‘‘steady state’’ as possible. Calculationswere performed via the following formulas:

& SIDa = Na+ + K+ + Ca2+ + Mg2+ j Clj

& SIDe = [1,000� 2.46� 10j11� PaCO2/10jpH] + [[albumin]�

(0.123 � pH j 0.631)] + [[PO4] � (0.309) � pH j 0.469)]& SIG = SIDa j SIDe

For the SIDa, all values were in milliequivalent per liter.Na+, K+, and Clj were provided in these units from ourelectronic medical record (EMR). Mg2+ and Ca2+ wereconverted via the following formulae:

& [Mg, mEq/L] = [Mg, mg/dL] � 0.83333& [Ca, mEq/L] = [iCa, mmol/L] � 2

For the SIDe, PaCO2 was recorded in millimeters ofmercury, albumin in grams per deciliter, and PO4 in millimolesper liter (converted from milligram per deciliter via [PO4,mmol/L] = [PO4 mg/dL] � 0.3229). Lactate was not includedin the calculation because it was not available for a majority ofour patients. Urate, which has been included in a minority ofprevious SID articles, was also excluded for this reason. Cal-culation of SID and SIG was performed via a programmedExcel spreadsheet. These results were then compared with theAPACHE II and SOFA scores for predictive value.

Student’s t tests and W2 tests were used to compare de-

mographic and outcome data as appropriate. Multivariate

J Trauma Acute Care SurgVolume 75, Number 4Berndtson et al.

556 * 2013 Lippincott Williams & Wilkins


logistic regression (for dichotomous variables) and linear re-gression (for continuous variables) were used to identify cor-relative relationships between variables. The first univariateanalysis was performed several times, each with a single in-dependent variable of SIDa, SIDe, or SIG at various timepoints. The dependent variables were hospital LOS, ventilatordays, mortality, and ARF. The next multivariate analysis wasperformed with SIDa, SIDe, or SIG as the independent variableand TBSA as the dependent variable. For admission analysesonly, age and admission Cr were added stepwise in the modelas controls for the relationship between the independent anddependent variables. All mean values are presented as themean T SEM. The University of California-Davis HumanSubject Review Committee approved the study design and datacollection protocol.

RESULTS

DemographicsA total of 113 patients were eligible and had full data sets

for analysis (Table 1). The overall mortality rate was 27.4%.Survivors and deceased patients differed significantly for age(survivor mean age, 42.0 T 1.5 years; deceased, 54.3 T 2.7years; p G 0.001); however, other admission demographicswere not significantly different. For outcomes, survivors had alonger mean hospital LOS and more ventilator-free days but asimilar number of total ventilator days. The incidence of ARFwas significantly higher in deceased patients at 48.4% versus18.3% in survivors (p = 0.001).

Admission Strong IonsAt admission, SIDa and SIDe were decreased from

normal, while the SIG was markedly elevated in both survivorsand deceased, although groups did not differ significantly(Figs. 1 and 2). On univariate analysis, SIDa, SIDe, and SIGwere not predictive of mortality (Table 2). However, on mul-tivariate analysis, after controlling for TBSA, age and admis-sion Cr, an elevated SIG predicted an increased risk ofmortality (odds ratio [OR], 1.11) (Table 3). In addition, on bothunivariate and multivariate analyses, SIDa on admission wasassociated with the hospital LOS. APACHE II score was alsopredictive of mortality (p = 0.006, not controlled for otherfactors). Base deficit had no correlation with any of our mea-sured outcomes at admission.

Hospital Day 1On hospital Day 1, the SIDs and SIG remained abnormal;

however, for the deceased group, both SIDa and SIDe weresignificantly higher than for the survivors (SIDa, 38.1 T 0.8deceased vs. 36.3 T 0.3 survivors [p G 0.05]; SIDe, 25.3 T 0.6deceased vs. 23.8 T 0.4 survivors [p G 0.05]). On univariateanalysis, both SIDa and SIDe were predictive of increasedmortality (OR, 1.16 and 1.13, respectively). In addition, withlinear regression, decreased SIDe was associated withprolonged requirement of mechanical ventilation and increasedhospital LOS (p G 0.05). Base deficit correlated only withventilator days (p = 0.04).

Hospital Day 2On hospital Day 2, the SIDs were still abnormally low in

both groups. Although both SIDa (37.9 T 0.6 vs. 36.6 T 0.3)and SIDe (27.3 T 0.7 vs. 26.7 T 0.3) were higher in the deceasedgroup compared with the survivors, these differences were notstatistically significant. On univariate analysis, SIDa, SIDe,SIG, and base deficit did not show any significant correlation tooutcomes.

Hospital Day 3On hospital Day 3, the SIDs continued to be low; how-

ever, the values did not statistically differ between deceased andsurvivors. The SIG, however, was significantly elevated in thedeceased group compared with the survivors (10.2 T 0.5 vs.8.8 T 0.2, p G 0.01) and was associated with increased mortality(OR, 1.38, p G 0.05). SIDa, SIDe, and base deficit were all

TABLE 1. Demographics

Survived Deceased

n 82 31 (27.4%)

Mean age, y 42.0 T 1.5 54.3 T 2.7*

Male, % 78.0 67.7

Mean %TBSA 40.7 T 1.8 43.5 T 3.5

Inhalation injury, % 18.3 19.4

Mean hospital LOS, d 59.2 T 3.3 29.4 T 6.3*

Mean ventilator days 36.5 T 3.1 28.2 T 6.4

ARF, % 18.3 48.4*

* = p G 0.05 between groups.

Figure 1. SIDa versus time.

Figure 2. SIDe versus time.

J Trauma Acute Care SurgVolume 75, Number 4 Berndtson et al.

* 2013 Lippincott Williams & Wilkins 557


associated with increased hospital stay and need for mechanicalventilation (p G 0.05).

DISCUSSION

As previously shown, the advantage of the SID and SIGmethod over the traditional anion gap or base excess is theability of the SID to compensate for changes in protein statusand to elucidate complex acid-base disorders. During analysisof our data, one of the most notable results was the significantlydecreased value for SIDe in burn patients. This was most ev-ident at admission and on Day 1 but persisted throughout the

time frame of the study. Breakdown of the scores showed thatthis was primarily caused by a significant acute hypoalbuminemiain almost all of our patients. By Day 1, the average albuminwas 1.92 in the survivors versus 1.95 in the deceased patients.Hypoproteinemia is usually thought of as a chronic problem,developed gradually in patients with malnutrition and thus nor-mally well compensated.13 In the acute setting, it can lead to amarked metabolic alkalosis.

That albumin levels were similar in survivors and thedeceased may have been partially artificial because our burnunit protocol is to transfuse patients to keep an albumin level of2g/dL or greater after the first hospital day.14 Despite this, itwas noted that the patient’s response to alkalosis differed be-tween groups. Nonsurvivors continued to use respiratorycompensation, with PaCO2 levels higher than 40 mm Hg, be-yond Day 3 of the study. In comparison, survivors had a moreappropriate metabolic compensation by Day 1 with increases inother anions, particularly chloride.

This limited ability to compensate metabolically in de-ceased patients may be related to their mean older age. De-ceased subjects were also much more likely to develop ARFduring the course of the study, which may have hindered theirability to adjust appropriately. Patients with complete renalfailure were excluded from our study, but to evaluate under-lying renal insufficiency, we collected baseline and daily Crvalues on all patients. Baseline Cr levels were not significantlydifferent between groups (admission Cr, 1.00 T 0.03 in sur-vivors; 1.06 T 0.06 in deceased patients; p = 0.39). However,during the first 2 days of hospitalization, the mean Cr levels ofdeceased patients rose steadily, while that of survivors declined(Table 4). There was a significant difference in Cr values be-tween groups at all times after admission.

The pervasive hypoalbuminemia seen in burn patientsalso helps to explain their unusually low SIDa and the SIDe. Aspreviously stated, a normal SID is approximately 40 mEq/L to42 mEq/L. Both our survivors and deceased patients had SIDaand SIDe measurements lower than this on each day of thestudy, with survivors initially appearing to be more abnormalthan deceased patients. As we have shown, the unusually lowSIDe is attributable to an acute loss of circulating albumin. Thisloss changes the SID steady state, with an expected compen-satory drop in SIDa via electroneutrality. Therefore, althoughour SIDa values are lower than normal for a healthy patient,they are actually higher than would be expected given our SIDefindings. This adjustment also means that deceased burn pa-tients are more abnormal than survivors, as would be expected.

Despite their abnormal protein status, burn patientsoverall remained significantly acidotic rather than alkalotic.The metabolic alkalosis attributable to hypoalbuminemia wascountered by compensatory respiratory and metabolic changes,

TABLE 3. Admission Results,Multivariate AnalysisVControlledfor TBSA, Age, and Admission Cr

Mortality LOS Ventilator Days

OR p p p

Admission

APACHE II 1.09 0.06 0.84 0.72

SIDa 1.07 0.24 0.05* 0.08

SIDe 0.96 0.40 0.10 0.07

SIG 1.11 0.05* 0.78 0.90

Base deficit 0.92 0.08 0.34 0.16

*Statistically significant.Mortality calculated via logistic regression; LOS and ventilator days calculated via

linear regression.

TABLE 2. Admission andDay 1 to 3 Results, Univariate Analysis

Mortality LOS Ventilator Days

OR p p p

Admission

APACHE II 1.12 0.006* 0.15 0.03*

SIDa 1.08 0.14 0.02* 0.08

SIDe 0.98 0.66 0.03* 0.03*

SIG 1.09 0.07 0.94 0.76

Base deficit 0.94 0.14 0.12 0.07

Day 1

SIDa 1.16 0.02* 0.11 0.46

SIDe 1.13 0.05* 0.01* 0.04*

SIG 1.03 0.66 0.30 0.12

Base deficit 1.03 0.59 0.06 0.04*

Day 2

SIDa 1.15 0.06 0.16 0.43

SIDe 1.06 0.39 0.44 0.66

SIG 1.16 0.15 0.40 0.62

Base deficit 1.03 0.64 0.18 0.20

Day 3

SIDa 1.10 0.22 G0.001* 0.002*

SIDe 0.92 0.29 0.03* 0.009*

SIG 1.38 0.004* 0.06 0.52

Base deficit 0.95 0.51 0.005* 0.001*

*Statistically significant.Mortality calculated via logistic regression; LOS and ventilator days calculated via

linear regression.

TABLE 4. Serum Cr (mg/dL) Versus Time

Admission Day 1 Day 2 Day 3

Survivors 1.00 T 0.03 1.02 T 0.03 0.99 T 0.03 0.87 T 0.02

Deceased 1.06 T 0.06 1.15 T 0.06* 1.23 T 0.10* 1.10 T 0.15*

*p G 0.05.




as previously stated, but was also complicated by a mixed acid-base picture. In the vascular trauma study by Kaplan andKellum,10 the average SIG was 2.4 T 1.8 in survivors versus10.8 T 3.2 in deceased patients. In comparison, our survivorshad a mean SIG of 14.8 T 0.5 at admission versus 16.5 T 0.9 inour deceased patients. The SIG did trend down more quickly insurvivors than deceased patients during the first 3 days (Fig. 3).When all patients were analyzed together, the minimum SIG onadmission was 5.6, with a maximum of 31.3. Throughout thecourse of the study, all calculated SIG values were positive,with no patient having a SIG of 0 or less at any time point. Thisindicates the persistent presence of a large number of unidentifiedanions in burn patients, more so than trauma patients. Theseanions remain even after clinically adequate resuscitation.

We did attempt to correlate SIDa, SIDe, and SIG valueswith fluid resuscitation during the first 24 hours of hospitali-zation and found no relationship. One hypothesis is that therewas no correlation with the amount of resuscitation received,but perhaps, there would be to the volume of intravenous fluidsa patient should have received. Suboptimal resuscitation onDay 1 would add a contraction alkalosis to the mixed acid-basestatus, and there is some evidence that this did occur in ourstudy population. After the development of an unusually lowSIDe in our patients, their SIDa decreases as well to balanceelectrical charges in the plasma. This would normally be ac-complished via an increase in anions (primarily Clj) and adecrease in cations (primarily Na+). This increase in Clj doesoccur in our patients, but the compensatory changes in Na+ arepresent only in survivors on Day 1, not in deceased patients. Onaverage, the Na+ in the nonsurviving patients actually increases,which is consistent with an inadequate fluid resuscitation andcontraction alkalosis.

There are several potential sources of the additionalunmeasured anions demonstrated in this study. Previous humanand animal researchers15Y18 have isolated a number of organicacids from critically ill subjects, including lactic acid, ketones,A-hydroxybutyrate, acetoacetate, sulfates, urate, citrate, pyruvate,gluconate, acute phase proteins, and more. Lactate has spe-cifically been included in other several SID studies but was notincluded in ours because it was rarely available retrospectively.Where it was available, none of our patients exhibited a sig-nificant lactic acidosis (maximummeasured value, 6.0 mEq/L).In addition, it has been shown that when lactate levels aremeasured, they account for less than 50% of the gap caused byunmeasured anions.15 That lactate was rarely measured on our

patients also prohibited us from comparing it with SID as apredictor of mortality. An interesting possibility is that ketonesmay constitute a significant percentage of this gap. It iswell-knownthat critically ill burn patients quickly become hypermetabolicfrom massive inflammation and severe physiologic insult. As aconsequence, ketone production in some patients may becomemarkedly increased following severe burn injury owing to pre-injurymedical conditions and poor nutrition. Alternatively, theseunmeasured anions may be specifically related to the burn injuryitself, resulting from inhaled or absorbed chemical compounds.Inhalation injury and compromise of respiratory function mayplay a role. Drug use, both illicit and prescribed, could also resultin unmeasured anions. Exogenous diuretics and corticosteroidscould cause anion loss, while blood product transfusions couldresult in cation gain. Further study would be useful in quantifyingthese effects because some results could have implications formanagement. Ultimately, the sum total of circulating unmeasuredanions will be a combination of several of these processes.

As stated, the overall mortality in our studywas 31 patientsof 113 or 27.4%. The causes of death were attributed to cardiacarrest (7), renal failure (2), sepsis (3), multiorgan failure (10),brain death (5), bowel perforation (1), and unclear causes (3).

We attempted to compare the APACHE II score, basedeficit, and the strong ion method in their ability to predictmortality, hospital LOS, and ventilator days for patients withsevere burns. While the APACHE II score was predictive ofmortality at admission, it had no correlation with other out-comes. Base deficit had some utility in predicting hospital LOSand ventilator days, but no correlation to mortality, despite itsusefulness in previous published studies.1Y4 Only the strong ionmethod was able to predict both at various time points withinour study. Given the growing ubiquity of EMRs, computers,smartphones, and handheld tablets in medicine calculating theSIDa, SIDe, and SIG is now relatively easy. A hospital’s EMRcould easily be customized to calculate and display thesevalues, as many already dowith other nonmeasured parameterssuch as glomerular filtration rate. Measurement of input pa-rameters can also be performed with the increasingly commonhandheld or in-unit point-of-care machines.4 The increasedpredictive value and decreased effort required argue for adoptingthe strong ion method as a new tool in critical care.

There are several limitations of our study, the mostprominent of which is that our data were collected retrospectivelyat a single center. This hindered our ability to analyze lactate dataand limited the number of patients available for inclusion. Inaddition, age was a confounder between our surviving and de-ceased groups, which may have influenced some of our statisticalanalyses. The logical next step would be a prospective multicenterstudy, which would be both comprehensive and definitive. If thestrong ion method is verified as predictive, which we expect, thenit would serve as an ideal benchmark for future studies investi-gating the role of proactive normalization of individual abnormalmarkers, to see if mortality or other outcomes can be improved.This would also allow a head-to-head comparison with basedeficit, lactate, or other markers of interest. A useful prospectivemarker of adequate resuscitation is the ultimate goal.

In conclusion, the SID and SIG method of acid-baseevaluation may be helpful in elucidating the complex mixeddisorders present in burn patients. This method accounts for theFigure 3. Strong ion gap versus time.




significant hypoalbuminemia seen in the burn patient popula-tion, which other formulas do not. Despite being more com-plicated to calculate, the SID method is useful in identifyingpatients with a greater metabolic derangement than wouldotherwise be suspected from anion gap or base excess dataalone. The SID does this by separating compensatory mech-anisms from mixed acid-base disorders, which may make ituseful early in the course of a patient’s illness, when correctionscan more easily be made. Further prospective study using SIDand SIG as a guide to direct resuscitation and other treatmentdecisions is warranted.

AUTHORSHIP

A.E.B. contributed in the study concept, study design, data collection,and statistical analysis and was the primary author of the abstract andthis article. T.L.P. contributed in the study concept, study design, criticalanalysis, and paper editing. D.G.G. contributed in the study concept,study design, and critical analysis. S.S. contributed in the study concept,study design, statistical analysis, and editing of the abstract and paper.

DISCLOSURE

The authors declare no conflicts of interest.

REFERENCES1. Kaups KL, Davis JW, Dominic WJ. Base deficit as an indicator or re-

suscitation needs in patients with burn injuries. J Burn Care Rehabil.1998;19(4):346Y348.

2. Cancio LC, et al. Base deficit and alveolar-arterial gradient during re-suscitation contribute independently but modestly to the prediction ofmortality after burn injury. J Burn Care Res. 2006;27(3):289Y296; dis-cussion 296Y297.

3. Cancio LC, et al. Predicting increased fluid requirements during the re-suscitation of thermally injured patients. J Trauma. 2004;56(2):404Y413;discussion 413Y414.

4. Jeng JC, et al. Serum lactate and base deficit suggest inadequate resus-citation of patients with burn injuries: application of a point-of-care lab-oratory instrument. J Burn Care Rehabil. 1997;18(5):402Y405.

5. Light TD, et al. The 2003 Carl A Moyer Award: real-time metabolicmonitors, ischemia-reperfusion, titration endpoints, and ultraprecise burnresuscitation. J Burn Care Rehabil. 2004;25(1):33Y44.

6. Jeng JC, et al. Improved markers for burn wound perfusion in the severelyburned patient: the role for tissue and gastric PCO2. J Burn Care Res.2008;29(1):49Y55.

7. Stewart PA. Modern quantitative acid-base chemistry. Can J PhysiolPharmacol. 1983;61(12):1444Y1461.

8. Fencl V, et al. Diagnosis of metabolic acid-base disturbances in critically illpatients. Am J Respir Crit Care Med. 2000;162(6):2246Y2251.

9. Wooten EW. Science review: quantitative acid-base physiology using theStewart model. Crit Care. 2004;8(6):448Y452.

10. Kaplan LJ, Kellum JA. Initial pH, base deficit, lactate, anion gap, strongion difference, and strong ion gap predict outcome from major vascularinjury. Crit Care Med. 2004;32(5):1120Y1124.

11. Knaus WA, et al. APACHE II: a severity of disease classification system.Crit Care Med. 1985;13(10):818Y829.

12. Vincent JL, et al. The SOFA (Sepsis-related Organ Failure Assessment)score to describe organ dysfunction/failure. On behalf of the WorkingGroup on Sepsis-Related Problems of the European Society of IntensiveCare Medicine. Intensive Care Med. 1996;22(7):707Y710.

13. Wilkes P. Hypoproteinemia, strong-ion difference, and acid-base status incritically ill patients. J Appl Physiol. 1998;84(5):1740Y1748.

14. Greenhalgh DG, et al. Maintenance of serum albumin levels in pediatricburn patients: a prospective, randomized trial. J Trauma. 1995;39(1):67Y73;discussion 73Y74.

15. Mecher C, et al. Unaccounted for anion in metabolic acidosis during severesepsis in humans. Crit Care Med. 1991;19(5):705Y711.

16. Kellum JA, et al. Hepatic anion flux during acute endotoxemia. J ApplPhysiol. 1995;78(6):2212Y2217.

17. Kellum JA, et al. Splanchnic buffering of metabolic acid during earlyendotoxemia. J Crit Care. 1997;12(1):7Y12.

18. Stacpoole PW, et al. Natural history and course of acquired lactic acidosisin adults. DCA-Lactic Acidosis Study Group. Am J Med. 1994;97(1):47Y54.

DISCUSSION

Dr. James Jeng (Washington, DC): Dr. Berndtson andmyother colleagues from the burn teamatUCDavis-SacramentoShriners Hospital are to be congratulated on yet another im-portant piece of research in surgical science investigation ofburn physiology.

In a nutshell, they retrospectively examined a relativelynovel indicator of acid-based physiology, the strong ion dif-ference and the difference between strong ion difference A andstrong ion difference E, as was discussed in the presentation,and correlated it to the mortality after major burn injuries. Thereis an urban legend that burn shock resuscitation is a problemthat has long been resolved and managed in a straightforwardexercise of keeping hourly urine output somewhere between30 and 50 ccs an hour.

In point of fact, though, a small but vocal group of burnresearchers have been publishing evidence for the past ten orfifteen years that all is not well and that closer inspection withserum lactate and base deficits point to the unresolved cellularshock that goes on in this population.

With this group’s current data the complexity of the acid-based abnormalities in the aftermath of major burns is furtherrevealed, especiallywith respect to the complex contribution of thedramatic hypoproteinemiaVlargely hypoalbuminemiaVthat hasbeen so elegantly described in this study.

With respect to the well-written manuscript I have thefollowing questions and comments:

Wouldn’t it be important to concurrently report the bloodgas data and serum lactate along with the strong ion differencedata?

Now, I know that you can t do the lactate becausethis study was retrospective but the base deficit should belargely available from blood gassesYbeing a straightforward, nocalculation thing.

It’s important to relate the current findings to what hasrecently been published in the last ten-fifteen years as an an-tecedent to your work just so that we have a link to all the otherpeople working in this field.

Furthermore, calculating the strong ion difference takes alittle bit more elbow grease than simply looking up the basedeficit or drawing a serum lactate.

And we really need to have a compelling reason whywe’re going to go through the extra trouble. You’ve given a lotof reasons whyVbut that argument should be looked after verycarefully.

And then I want to broaden the scope of my comments.With respect to this current work, antecedent work on lactate,base deficit, etc cetera, in future directions of your lab groupand other people in this area of interest, I’d like to underscorethe secret wish of this working group:




Can we alter mortality after burn injuries by takingclinical steps more quickly, normalizing these various markersof acid-based abnormalities and cellular shock?

We now have several good numerical talismans of re-suscitative efforts from burn-shock. It’s time we collectivelymoved ahead and tried to realize their full potential as end-points of resuscitation somewhat more specific and elegantthan simply 30Y50ccs of urine output per hour.

Thank you for the pleasure of the floor.Dr. Allison E. Berndtson (Sacramento, California):

Thank you, Dr. Jeng, for your review and questions. As yousaid, we unfortunately did not have lactate levels availableretrospectively, but in a future prospective study that would bean important component to include.

As for base deficit, we did look at that data and found nocorrelation between base deficit and mortality, length of stay, orventilator days in our population. The strong ion difference wasmore useful than base deficit at all time points.

Regarding the fact that the SID is more difficult to cal-culate, that is absolutely true. It is not something that you caneasily do in your head at the bedside like you can with an aniongap. However, it could easily be calculated by your hospital laband reported in the EMR along with other laboratory results, asis donewith other assessments such as GFR. I actually found aniPhone application that will calculate it for you as well. So itdoesn’t have to be that much work with all the computer toolsthat we have to help us calculate it.

Finally, regarding whether or not we can use eitherstrong ion difference or other endpoints to help us resuscitatepatients more quickly and more appropriately, that is abso-lutely the ultimate goal. I think that a prospective trial needsto be done using the SID as an endpoint, to see if targetingthese values specifically in addition to just urine outputwill result in faster resuscitation and an improvement in long-term outcomes.

Thank you.




strong ion difference and gap predict outcomes

Documents