- Email: [email protected]
The application of the extracellular base excess to children
BIOCHEMICAI
MEDICINE
32, 67-78
(1984)
The Application of the Extracellular Base Excess to Children DANIEL J. ESSIN Pultnona~
Physiology
Laboratory. Room 3E18. Pediatric Pavilion. S&et. Los Angeles, Culjfornicr 90033
1129 Norrh
Starr
Received February 22, 1983
A number of methods for calculating an index of the metabolic component of an acid-base disturbance have been proposed. Many pediatricians who care for critically ill children with acid-base disturbances, find such an index to be useful in the management of their patients. Clinical laboratories and automated blood gas analyzers commonly report at least one of them. When severe acidosis is present, medical management may include administration of sodium bicarbonate. Early attempts at using an index to guide bicarbonate therapy occasionally led to administration of doses of bicarbonate which were too large. Consequently some clinicians use empirical formulas to modify the doses of bicarbonate calculated from the index, Others ignore the index and base their diagnosis and treatment directly on the values of pH, PCO17 and plasma bicarbonate. There is considerable disagreement among adherents to the various approaches. Little of this controversy has been aired in the literature. As a result many pediatricians trained recently have learned an empirical approach with little exposure to the details of the physiology or to subtleties of the controversy. The proliferation of automated analyzers which report a number labeled “BE” even though different formulas may be used in the calculation has contributed to the confusion. The indices represent the nonventilatory component of any disturbance in acid-base balance as the number of milliequivalents of strong acid or base which, if administered, would correct the disturbance. Each is based on a set of assumptions about the nature of the buffer systems. Since most of the work leading to the development of the indices was done in adults, one assumption is implicit: the value was calculated as if the blood sample was drawn from an adult. The purpose of this study was to explore the physiological implications of applying to children indices developed for use in adults. The hypothesis 67 0006-2944184
$3 .OO
CopyrIght $ 1984 by Academic Press. Inc A I rights of reproduction in any form reserved
6X
I)lNIEL
.I. ESSIN
that the Base Excess of the Extracellular Fluid (BE-ecf) represents the most physiologically sound index was tested by comparing it to several other indices computed on samples obtained from adults. The theoretical and experimental background which led to the formulation of this hypothesis is discussed. An adjustment for size and age was then applied to the BE-ecf. The effect was examined to test the hypothesis that adjusting this index specifically for children could produce a more accurate and more clinically useful value. METHODS The material for this study consisted of the results of blood gas analyses conducted by the Pulmonary Physiology Laboratories of the Los Angeles County-University of Southern California Medical Center. All specimens were originally collected as a part of the routine care of patients hospitalized at the Medical Center. The samples studied were selected at random; all had been drawn from 1 month to 4 years earlier and the outcome of this study had no effect on the care of any specific patient. Three separate groups of data were selected. Eight hundred and fiftythree samples were selected from adults. No information as to the age or sex distribution of the patients from whom the samples were obtained was available. Two hundred and seventy samples were selected from patients ranging from premature to 2 years of age. No information was available as to the sex distribution or weight of these patients. One hundred and sixty-three samples were selected from patients ranging from birth to 2 years of age in whom the weight was available and exceeded 2.3 kg. The sex distribution of these patients was not available. Group I. The Standard Bicarbonate (I .2), the T40 Bicarbonate (3), the Base Excess of the blood (BE-blood) (4,5), and the Base Excess of the Extracellular Fluid (BE-ecf) (6.7) and the delta Bicarbonate were calculated on the blood gas samples which were drawn from adults. For the purpose of comparison. the Plasma Bicarbonate, the Standard Bicarbonate, and the T40 Bicarbonate were expressed as differences relative to 24 meq/liter (the reference value of plasma bicarbonate) as is commonly done with the BE-blood and the BE-ecf. The delta Bicarbonate is the difference between the reference and actual values of Plasma Bicarbonate. The value of the BE-ecf was compared to each of the other indices using cross-tabulation techniques. In our laboratory, values which are less than _t3 meq/liter indicate the absence of a significant acid-base disturbance. For the purposes of the analysis the values which fell within this range were termed normal. Values outside this range were termed abnormal.
BASE
EXCESS
69
IN CHILDREN
Based on these definitions, the following computations were done: % false positives-derived from the number of samples in which the BEecf was normal but the other index was abnormal; % false negativesderived from the number of samples in which the BE-ecf was abnormal but the other index was normal; total % error-number of false positives plus number of false negatives related to total sample size. The difference between each of the other indices and the BE-ecf was calculated and will be termed an Index Difference. An Index Difference of - 1 indicates that the index under consideration yielded a value for the metabolic component which was 1 meq/liter more severe than the BE-ecf. The five basic indices thus reduce to four index differences. The magnitude of the Index Differences was compared to the absolute values of the raw data (PH. PC02, HCO?) using analysis of variance (ANOVA) techniques. The results of the ANOVA were plotted. tests for the presence of linear regression were done and Pearson’s r* was computed. In order to facilitate the computations, each of the independent variables was ranked which allowed the data to be coded into a small number of groups. Group ZZ. The index difference between delta Bicarbonate and the adult BE-ecf was calculated on the samples from the infants whose weight was unavailable. The results are presented in Table I. Group 111. An adjusted BE-ecf was calculated for the samples from infants whose weight was available by using established relationships between weight and extracellular fluid volume (8,9). The value for extracellular fluid volume was then combined with an age-adjusted plasma protein concentration and the estimated hemoglobin (hematocrit + 3) to compute an individualized buffering capacity (10). This value was then used in the standard BE-ecf equations in place of the adult value. The index differences between the adjusted and nonadjusted results were calculated. These results are presented in Table 2. TABLE INDEX
DIFFERENCE
--
-
ID _
-.. 1 0
-I -2 -3 -4 -5
AND
*
BETWEEN ADULT
I DELTA
BICARBONATE
BE-ecf
n
‘7
? 35 II8 80 ?I 2 3 770
0.74 12.96 43.70 29.63 11.48 0.74 0.74
I)ANIEI.
J. ESSIN
TABLE INDEX
DIFFEREWE
BE-ecf
2
BETWEEN ANI)
AGE-AI)JI’SI
EI)
BE-ect
ADULT
RESULTS Group I
Compared to the BE-ecf. the BE-blood yielded 10.3% false positive results, 12.6% false negative results, and a total error of 11.4%. The plasma bicarbonate was falsely positive in 4.3%, falsely negative in 8.2%, with a total error of 6.3%. The Standard Bicarbonate was falsely positive in 10.3% of samples, falsely negative in 14.2%. and the total error was 12.5%. The T40 Bicarbonate was falsely positive in 1.9%, falsely negative in 0.5%, with a total error of 1.1%. The ANOVA of the Index Differences against the pH of the samples demonstrated that each set of index differences formed a distinct group (P < 0.01) (Table 3). The groups of index differences were linearly related to the pH except in the case of the Index Difference for Standard Bicarbonate. The r2 statistic in each instance describes the proportion of the variation in the Index Difference that is linearly related to pH; the onlymoderately strongassociation(r’ = 0.86) was WiththeIndexDifference for delta-Bicarbonate. These data are presented graphically (Fig. 1). TABLE INDEX
Index
DIFFERENCES
difference
ANOVA”
delta Bicarbonate Std Bicarbonate BE-blood T40 Bicarbonate
S NS s s
“ All tests
significant
at P < 0.01 based
3 vs
pH ).2
Regression” s NS s s on the F value
unless
0.86 0.01 0.49 0. I4 indicated
otherwise.
BASE EXCESS
71
IN CHILDREN
BE-b
I ood
Std-HC03
\T40-HC03 ‘\ *\
‘\ds
I t a-l-E03
FIG. 1. The Index Differences (e.g., BE-ecf minus BE-blood) have been plotted against the pH of the specimens. This demonstrates the extent to which the other indices overor underestimate the degree of disturbance. Larger negative Index Differences represent larger overestimations.
The Index Differences were compared to PCOz in the same manner. The ANOVA revealed significant differences between the group means (Table 4 and Fig. 2). All Index Differences were linearly related to the PCOz (P < 0.01) but only in the case of the Index Differences for Standard Bicarbonate (r’ = 0.63) and BE-blood (r’ = 0.76) was even a moderate proportion of the variability in the Index Difference linearly explained by PC&.
__-..__ Index difference
TABLE 4 INDEX DIFFERENCES vs PCO, ANOVA”
.~~___ Regression“
_---
-
delta Bicarbonate S S Std Bicarbonate S S BE-blood S s T40 Bicarbonate S S -__ ” All tests significant at P < 0.01 based on the F value unless indicated otherwise,
rJ 0.22 0.63 0.76 0.25
77
‘. ‘\
-dr jj
I t a-HC03
,’ ,’
I*
3,’ ,’
---___/*
‘-T40-HC03
BE-b
I ood
FIG. 2. This figure is analogous to Fig. I except the Index Differences are plotted against the PCOz of the specimens.
The ANOVA of the Index Differences against the plasma bicarbonate also demonstrated significant differences (P < 0.01) in the group means (Table 5 and Fig. 3). In addition the Index Differences were all linearly related to the bicarbonate (P < 0.01) except for the Index Difference of delta Bicarbonate. The r* statistic demonstrates that the variation in the Index Differences for Standard Bicarbonate (? = 0.69) and T40 Bicarbonate r’ = 0.77) shows only a moderate linear association to the Plasma Bicarbonate, while the Index Differences for delta Bicarbonate and BEblood had no variability which could be linearly related to the value of the Plasma Bicarbonate. TABLE INDEX
Index difference delta Bicarbonate Std Bicarbonate BE-blood T40 Bicarbonate
DIFFERENCES
5
vs PLASMA
ANOVA” s s S s
BICARBONATE
Regression”
i
NS s s s
0.07 0.69 0.16 0.77
” All tests significant at P <: 0.01 hased on the F value unless indicated otherwise.
BASE EXCESS
IN CHILDREN
73
FIG. 3. This figure ib analogous to Fig. 1 except the Index Differences are plotted against the bicarbonate concentrations of the specimens. The point plotted for the Index Difference of the Standard Bicarbonate at an HCO, of 6 or less has a Standard Error of 2 3 meq/liter.
Group II
When the Index Differences were computed between the adult BEecf and the delta Bicarbonate in neonates and older infants, the adult BE-ecf produced estimates of the magnitude of acidosis exceeding the delta Bicarbonate by 2 or more meq/liter in 42.6% of the samples. In two samples (0.74% of the total) the overestimate was 5 meqkter (Table 1). Group III
The Index Differences were computed between the age-adjusted and adult BE-ecf values. In 4.9% of the samples the adult BE-ecf were lower by 1 meq/liter using the age-adjusted value as the point of comparison. There was exact agreement between 44.2% of the values. The adult BEecf indicated a disturbance which was 1 meq/liter more severe than that of the adjusted BE-ecf in 44.8%. The discrepancy was greater in approximately 6%. In 1 sample (0.61% of the total) the adult BE-ecf overestimate was 7 meq/liter (Table 2).
7-l
II:\NltLl.
.I. F,SSIX
DISCUSSION Since Hasselbalch (1 I) first described the relationship between plasma bicarbonate, pH, and PCO,. the plasma bicarbonate has been used air an indicator of acidosis (I?). This ignores the presence of the significant quantities of nonbicarbonate buffers which are present ( 13). The calculations required to compute an index which includes these additional buffelsystems are complex when compared to the Henderson-Hasselbalch equation. Several of the popular indices were developed, in part, because they could be represented graphically and involved few variables, not necessarily because of physiological merit. The resulting simplifications have been the subject of numerous controversies. All of the techniques attempt to remove the effect of abnormal levels of ventilation so that the magnitude of the acid-base disturbance. were the PCO:, normal (40). can be determined. Using a similar approach the pH can be adjusted to 7.4 before the metabolic component is computed. In addition, since blood has a different buffering capacity in ~‘itro than it has in \vi~*o, it is possible to calculate a variety of indices. Table 6 presents a comparison of the assumptions and mathematical corrections involved in each index. Of the five indices. the BE-ecf incorporates the most physiological set of assumptions. It removes the direct influences of pH and PCOz on the index by correcting both to a pH of 7.4 and a PCO, of 40 and thereby effectively removes the influence of any ventilatory abnormality which would tend to increase the body concentration of acid. In order to produce any index, some assumption must be made about the buffering capacity of the body. The most appropriate value for buffering capacity is one whjch can be derived theoretically and confirmed experimentally. The BE-ecf meets these criteria. The theoretical derivation makes the following assumptions: ( 1) hemoglobin. bicarbonate, and plasma proteins constitute the clinically significant buffers. I‘ABLE 6 CHARACTERISTICS
OF INDICES
Value influenced by nonbicarbonate buffers delta Bicarbonate Standard Bicarbonate Base Excess of blood T40 Bicarbonate Base Excess of ECF NA-not
applicable.
+ + +
OF
Acm-Basr;
BAI.ANCE
Standard conditiona Reference system NA
PCO, 40
PH 7.40
NA
NA
I
in in
Vitro Virro
t
t
in in
Viw b’ilo
+ *
.~ +
BASE
EXCESS
IN CHILDREN
75
(2) The intravascular and extracellular bicarbonate pools buffer as if they form a single compartment. These assumptions predict a buffering capacity of approximately 12 Slykes in adults. This has been confirmed experimentally by Brackett at al. (14). Indices of the metabolic component of acid-base balance which do not incorporate the most current physiological concepts must prove their value. If they are to be retained as diagnostic and therapeutic tools, they should be sufficiently accurate and predictable to preclude incorrect diagnostic conclusions and to avoid wide or unexpected swings in pH during therapy. For the other indices under study to be considered predictable, any deviation from the BE-ecf must be of a direction and magnitude which is proportional to the magnitude of the metabolic disturbance. Predictability was evaluated by relating the index difference (the difference between each index and the BE-ecf) to the raw data because the index used clinically to estimate the metabolic component is given different weight depending on any disturbances which are present in the raw data. If a given index maintained a constant index difference across a wide range of pH, PC02, and bicarbonate values, it could be considered highly predictable. On the other hand, large changes in the index difference would suggest that even in experienced hands, therapy based on such an index could have undesired and seemingly unpredictable effects. The BE-blood and the Standard Bicarbonate resulted in total error rates of 11.4 and 12.376, respectively, when compared to the BE-ecf. With errors of this magnitude there must be some mitigating conditions to justify their continued use. For example, were the errors directly proportional to the pH, it might be possible to learn through experience the degree to which the index should be “disbelieved.” An examination of Fig. 1 will reveal that the degree to which the BE-blood is in error is linear and consistently overestimates the magnitude of the metabolic component at low pH. The Standard Bicarbonate on the other hand is inconsistent in that it underestimates the disturbance at both low and high pH. There is no reason to retain the Standard Bicarbonate as a clinical tool in view of this behavior. The BE-blood, on the other hand. can be mentally adjusted by an experienced clinician. It is unclear why this effort should be made since the error can be eliminated by substituting the in \,ivo buffering capacity in the calculations in place of the in t’itro value. The T40 Bicarbonate would lead to erroneous conclusions about the presence or absence of metabolic acidosis in only 1.1% of the samples; however, these samples which were considered to be in error tended to underestimate the degree of acidosis at low pH. The difficulty of the
computation is comparable to that of the BE-ecf. The T40 Bicarbonate. therefore, offers no advantage over the BE-ccl’. Figure 3 demonstrates another aspect of the Standard Bicarbonate and the T40 Bicarbonate which may interfere with their therapeutic application. When the Plasma Bicarbonate is low, these two indices significantly underestimate the magnitude of the metabolic component. In this situation the Standard Bicarbonate and the T40 Bicarbonate may indicate a smaller metabolic component than the Plasma Bicarbonate leading to confusion about or distrust of the blood gas report. The delta-Bicarbonate underestimated the degree of acidosis consistently and would have led to 6.352 erroneous conclusions about the presence or absence of metabolic acidosis when compared to the BE-ecf. This tendency to underestimate acidosis is expected since there are significant buffers present in addition to bicarbonate. In the presence of severe anemia, low concentration of plasma proteins or marked increases in the extracellular fluid volume, the delta-Bicarbonate may be useful in providing a conservative guide to therapy if an appropriately corrected BE-ecf cannot be readily calculated. The acid-base physiology of children differs from that of adults. Infants and especially prematures have lower concentrations of plasma proteins. The hemoglobin concentration is age related, and in the critical care setting often highly variable. The extracellular fluid volumes vary widely. though predictably, with both pre- and postnatal age, being relatively larger in smaller babies. The renal compensatory mechanisms may be inadequate, especially in the immediate postnatal period. Consequently. most young children have a lower buffering capacity than adults. Every effort should be made to estimate the buffering capacity on an individual basis as well as to choose the most physiologically relevant formula for calculating an index of the metabolic component of an acid-base disturbance. The considerations which lead the BE-ecf to be the most appropriate index for use in adults also make it the index of choice for use in children if it is adjusted for the measured hemoglobin concentration and for the known differences in extracellular fluid volume and plasma protein concentrations. When the BE-ecf computed using adult formula was compared to a BE which was adjusted for approximate extracelhtlar fluid volume. plasma protein concentration. and hemoglobin concentration, 51% of the samples demonstrated a lesser acid-base disturbance. While most of the differences were small, occasional differences were substantial. These larger differences were in samples with the most severe acidosis (adult BE-ecfs between - 8 and - 20). Because administration of bicarbonate is much more likely in more severe acidosis, the overestimation which results from the adult BE-ecf represents a significant potential risk to the infant. Distinguishing
BASE
EXCESS
IN CHILDREN
77
between child and adult buffering capacity yields clinically useful information and should therefore be done routinely. The comparison between the delta-bicarbonate in infants and the standard (adult) method of calculating BE-ecf was done to evaluate the hypothesis that in the presence of reduced buffering capacity the delta-Bicarbonate may be a useful, conservative estimate of metabolic acidosis. In over 99% of the samples the delta-Bicarbonate was either equal to the adult BE-ecf or smaller. This suggests that if an age-adjusted BE is not available and there is uncertainty about the normality of the extracellular fluid volume, hemoglobin concentration, or the plasma protein concentration. the delta-Bicarbonate yields a conservative number which can be used safely. While the BE-ecf, when appropriately ad.justed, provides a reliable estimate of the metabolic component of an acid-base disturbance, a caveat must be observed. The foregoing discussion has been centered primarily on the utilization of the BE-ecf in the evaluation of acute changes. As Schwartz and Relman (8) have emphasized, in many clinical settings a subject’s chronic or baseline state is not normal. There may be an ongoing metabolic or ventilatory abnormality which is partially compensated by homeostatic mechanisms. Under these conditions, the BE-ecf may remain relatively constant but at a value outside the normal range. In order to assess an acute change in the subject’s status. the magnitude and direction of the change must be referred to the subject’s baseline. Thus, the effective use of the BE-ecf requires close attention 10 the context in which it is obtained as well as a thorough understanding of the physiological principles involved. SUMMARY The Base Excess of the Extracellular Fluid (BE-ecf) is the most complete model of acid-base physiology. It has gained considerable acceptance for use in adults as an index of the metabolic component of an acute acid-base disturbance. Several other commonly used indices were compared to the BE-ecf. The values of the Base Excess of Blood, Plasma Bicarbonate, Standard Bicarbonate, and T40-Bicarbonate differed significantly from the BE-ecf in 11.4, 6.3, 12.5, and 1.1% of samples, respectively. These differences are considered to be errors since the nonlinear relationship of the variables makes it difficult to clinically accomodate them. The standard (adult) form of the BE-ecf calculation overestimated the base excess by 1 meq/liter in 44.8% and by 2 meq/liter or more in 6.1% of samples from neonates and infants when compared to a form of the calculation which was individually adjusted based on the weight and hemoglobin concentration of each subject. Since it is no more difficult
to make these corrections than to ignore them. if the BE-ecf LY 10 bc used in neonates and infants the correction should be applied ACKNOWLEDGMENTS I thank John Mohler, Clarence Collier, Stanley ALen. and Miriam Wilson for their help at various stages in the preparation of this manuscript.
REFERENCES I. Roos, A.. and Thomas. L. J.. .Jr. The m vitro and in viva carbon dioxide dissociation curves of true plasma. Ane.v/lrr.sio/og.v 28, 1048 f 1967). 1. Schwartz, W. B.. and Relman. S. R. A critique of the parameters used in the evaluation of acid-base disorders. N. G/x(. J. Med. 268, 2.S (1963). 3. Armstrong, B. W., Mohler. .I. G., Jung. R. c’.. 1’1 ol The in vivo carbon dioxide titration curve. Lancer 1, 7SY ( 1966). 4. Siggard-Anderson. O., Engel. K., Jorgensen. K.. and Astrup. P. Micro method for determination of pH. carbon dioxide tension. base excess and standard bicarbonate in capillary blood. Stand. J. C/in. Ltrh. Intvst. 12, 172 (1960). 5. Astrup, P., Jorgensen, K., Siggard-Anderson. 0.. and Engel. K. Acid-base balance: New approach. Lunc,ef 1, IO35 (1960). 6. Collier, C. R., Hackney. J. D.. and Mohler. J. G. Use of extracellular base excess m diagnosis of acid-base disorders: A conceptual approach. CI~esr 61, 6s ( 1972). 7. Siggard-Anderson, 0. An acid base chart for arterial blood with normal and pathophysiological reference area\. .SU&. J. C/in. Lab. Irrwst. 27, 238 ( I971 ). 8. Cheek, D. B. Extracellular volume: Its structure and measurement and the intiuence of age and disease. J. PediutI-. 58, I ( 1961). 9. Friis-Hansen, B. Body water compartments in children: Changes during growth and related changes in body composition. Pediatrks 28, 2 ( 1961). IO. Essin, D. J. Calculation of an individualized base excess of the extracellular fluid fat children of varying size. Ahsfroc.l. C/in. Res. 26(Z). l62a (1978). des Blute\. II, Hasselbalch. K. A. Die “reduziertr” und die “regulierte” k!kiSerStOffiahi Biochern.
2. 74, 56 (1916).
I?. Van Slyke. D. D.. and Cullen. C;. E. Studies of acidosis. I. Bicarbonate concentration of blood plasma: Its significance. and its determination as a measure of acidosis. .I. Biol. Chenz. 30, 2X9 (1917). 13. Singer. R. B., and Hastings, A. B. Improved clinical method for estimation of disturbances of acid-base balance of human blood. Medicine 27, 223 ( 1948). 14. Brdckett, N. C. Jr.. Cohen. J. J.. and Schwartz. W. B. The carbon dioxide titration curve of normal man. N. EnPI. J. Med. 272, 6 (IYhSb.
MEDICINE
32, 67-78
(1984)
The Application of the Extracellular Base Excess to Children DANIEL J. ESSIN Pultnona~
Physiology
Laboratory. Room 3E18. Pediatric Pavilion. S&et. Los Angeles, Culjfornicr 90033
1129 Norrh
Starr
Received February 22, 1983
A number of methods for calculating an index of the metabolic component of an acid-base disturbance have been proposed. Many pediatricians who care for critically ill children with acid-base disturbances, find such an index to be useful in the management of their patients. Clinical laboratories and automated blood gas analyzers commonly report at least one of them. When severe acidosis is present, medical management may include administration of sodium bicarbonate. Early attempts at using an index to guide bicarbonate therapy occasionally led to administration of doses of bicarbonate which were too large. Consequently some clinicians use empirical formulas to modify the doses of bicarbonate calculated from the index, Others ignore the index and base their diagnosis and treatment directly on the values of pH, PCO17 and plasma bicarbonate. There is considerable disagreement among adherents to the various approaches. Little of this controversy has been aired in the literature. As a result many pediatricians trained recently have learned an empirical approach with little exposure to the details of the physiology or to subtleties of the controversy. The proliferation of automated analyzers which report a number labeled “BE” even though different formulas may be used in the calculation has contributed to the confusion. The indices represent the nonventilatory component of any disturbance in acid-base balance as the number of milliequivalents of strong acid or base which, if administered, would correct the disturbance. Each is based on a set of assumptions about the nature of the buffer systems. Since most of the work leading to the development of the indices was done in adults, one assumption is implicit: the value was calculated as if the blood sample was drawn from an adult. The purpose of this study was to explore the physiological implications of applying to children indices developed for use in adults. The hypothesis 67 0006-2944184
$3 .OO
CopyrIght $ 1984 by Academic Press. Inc A I rights of reproduction in any form reserved
6X
I)lNIEL
.I. ESSIN
that the Base Excess of the Extracellular Fluid (BE-ecf) represents the most physiologically sound index was tested by comparing it to several other indices computed on samples obtained from adults. The theoretical and experimental background which led to the formulation of this hypothesis is discussed. An adjustment for size and age was then applied to the BE-ecf. The effect was examined to test the hypothesis that adjusting this index specifically for children could produce a more accurate and more clinically useful value. METHODS The material for this study consisted of the results of blood gas analyses conducted by the Pulmonary Physiology Laboratories of the Los Angeles County-University of Southern California Medical Center. All specimens were originally collected as a part of the routine care of patients hospitalized at the Medical Center. The samples studied were selected at random; all had been drawn from 1 month to 4 years earlier and the outcome of this study had no effect on the care of any specific patient. Three separate groups of data were selected. Eight hundred and fiftythree samples were selected from adults. No information as to the age or sex distribution of the patients from whom the samples were obtained was available. Two hundred and seventy samples were selected from patients ranging from premature to 2 years of age. No information was available as to the sex distribution or weight of these patients. One hundred and sixty-three samples were selected from patients ranging from birth to 2 years of age in whom the weight was available and exceeded 2.3 kg. The sex distribution of these patients was not available. Group I. The Standard Bicarbonate (I .2), the T40 Bicarbonate (3), the Base Excess of the blood (BE-blood) (4,5), and the Base Excess of the Extracellular Fluid (BE-ecf) (6.7) and the delta Bicarbonate were calculated on the blood gas samples which were drawn from adults. For the purpose of comparison. the Plasma Bicarbonate, the Standard Bicarbonate, and the T40 Bicarbonate were expressed as differences relative to 24 meq/liter (the reference value of plasma bicarbonate) as is commonly done with the BE-blood and the BE-ecf. The delta Bicarbonate is the difference between the reference and actual values of Plasma Bicarbonate. The value of the BE-ecf was compared to each of the other indices using cross-tabulation techniques. In our laboratory, values which are less than _t3 meq/liter indicate the absence of a significant acid-base disturbance. For the purposes of the analysis the values which fell within this range were termed normal. Values outside this range were termed abnormal.
BASE
EXCESS
69
IN CHILDREN
Based on these definitions, the following computations were done: % false positives-derived from the number of samples in which the BEecf was normal but the other index was abnormal; % false negativesderived from the number of samples in which the BE-ecf was abnormal but the other index was normal; total % error-number of false positives plus number of false negatives related to total sample size. The difference between each of the other indices and the BE-ecf was calculated and will be termed an Index Difference. An Index Difference of - 1 indicates that the index under consideration yielded a value for the metabolic component which was 1 meq/liter more severe than the BE-ecf. The five basic indices thus reduce to four index differences. The magnitude of the Index Differences was compared to the absolute values of the raw data (PH. PC02, HCO?) using analysis of variance (ANOVA) techniques. The results of the ANOVA were plotted. tests for the presence of linear regression were done and Pearson’s r* was computed. In order to facilitate the computations, each of the independent variables was ranked which allowed the data to be coded into a small number of groups. Group ZZ. The index difference between delta Bicarbonate and the adult BE-ecf was calculated on the samples from the infants whose weight was unavailable. The results are presented in Table I. Group 111. An adjusted BE-ecf was calculated for the samples from infants whose weight was available by using established relationships between weight and extracellular fluid volume (8,9). The value for extracellular fluid volume was then combined with an age-adjusted plasma protein concentration and the estimated hemoglobin (hematocrit + 3) to compute an individualized buffering capacity (10). This value was then used in the standard BE-ecf equations in place of the adult value. The index differences between the adjusted and nonadjusted results were calculated. These results are presented in Table 2. TABLE INDEX
DIFFERENCE
--
-
ID _
-.. 1 0
-I -2 -3 -4 -5
AND
*
BETWEEN ADULT
I DELTA
BICARBONATE
BE-ecf
n
‘7
? 35 II8 80 ?I 2 3 770
0.74 12.96 43.70 29.63 11.48 0.74 0.74
I)ANIEI.
J. ESSIN
TABLE INDEX
DIFFEREWE
BE-ecf
2
BETWEEN ANI)
AGE-AI)JI’SI
EI)
BE-ect
ADULT
RESULTS Group I
Compared to the BE-ecf. the BE-blood yielded 10.3% false positive results, 12.6% false negative results, and a total error of 11.4%. The plasma bicarbonate was falsely positive in 4.3%, falsely negative in 8.2%, with a total error of 6.3%. The Standard Bicarbonate was falsely positive in 10.3% of samples, falsely negative in 14.2%. and the total error was 12.5%. The T40 Bicarbonate was falsely positive in 1.9%, falsely negative in 0.5%, with a total error of 1.1%. The ANOVA of the Index Differences against the pH of the samples demonstrated that each set of index differences formed a distinct group (P < 0.01) (Table 3). The groups of index differences were linearly related to the pH except in the case of the Index Difference for Standard Bicarbonate. The r2 statistic in each instance describes the proportion of the variation in the Index Difference that is linearly related to pH; the onlymoderately strongassociation(r’ = 0.86) was WiththeIndexDifference for delta-Bicarbonate. These data are presented graphically (Fig. 1). TABLE INDEX
Index
DIFFERENCES
difference
ANOVA”
delta Bicarbonate Std Bicarbonate BE-blood T40 Bicarbonate
S NS s s
“ All tests
significant
at P < 0.01 based
3 vs
pH ).2
Regression” s NS s s on the F value
unless
0.86 0.01 0.49 0. I4 indicated
otherwise.
BASE EXCESS
71
IN CHILDREN
BE-b
I ood
Std-HC03
\T40-HC03 ‘\ *\
‘\ds
I t a-l-E03
FIG. 1. The Index Differences (e.g., BE-ecf minus BE-blood) have been plotted against the pH of the specimens. This demonstrates the extent to which the other indices overor underestimate the degree of disturbance. Larger negative Index Differences represent larger overestimations.
The Index Differences were compared to PCOz in the same manner. The ANOVA revealed significant differences between the group means (Table 4 and Fig. 2). All Index Differences were linearly related to the PCOz (P < 0.01) but only in the case of the Index Differences for Standard Bicarbonate (r’ = 0.63) and BE-blood (r’ = 0.76) was even a moderate proportion of the variability in the Index Difference linearly explained by PC&.
__-..__ Index difference
TABLE 4 INDEX DIFFERENCES vs PCO, ANOVA”
.~~___ Regression“
_---
-
delta Bicarbonate S S Std Bicarbonate S S BE-blood S s T40 Bicarbonate S S -__ ” All tests significant at P < 0.01 based on the F value unless indicated otherwise,
rJ 0.22 0.63 0.76 0.25
77
‘. ‘\
-dr jj
I t a-HC03
,’ ,’
I*
3,’ ,’
---___/*
‘-T40-HC03
BE-b
I ood
FIG. 2. This figure is analogous to Fig. I except the Index Differences are plotted against the PCOz of the specimens.
The ANOVA of the Index Differences against the plasma bicarbonate also demonstrated significant differences (P < 0.01) in the group means (Table 5 and Fig. 3). In addition the Index Differences were all linearly related to the bicarbonate (P < 0.01) except for the Index Difference of delta Bicarbonate. The r* statistic demonstrates that the variation in the Index Differences for Standard Bicarbonate (? = 0.69) and T40 Bicarbonate r’ = 0.77) shows only a moderate linear association to the Plasma Bicarbonate, while the Index Differences for delta Bicarbonate and BEblood had no variability which could be linearly related to the value of the Plasma Bicarbonate. TABLE INDEX
Index difference delta Bicarbonate Std Bicarbonate BE-blood T40 Bicarbonate
DIFFERENCES
5
vs PLASMA
ANOVA” s s S s
BICARBONATE
Regression”
i
NS s s s
0.07 0.69 0.16 0.77
” All tests significant at P <: 0.01 hased on the F value unless indicated otherwise.
BASE EXCESS
IN CHILDREN
73
FIG. 3. This figure ib analogous to Fig. 1 except the Index Differences are plotted against the bicarbonate concentrations of the specimens. The point plotted for the Index Difference of the Standard Bicarbonate at an HCO, of 6 or less has a Standard Error of 2 3 meq/liter.
Group II
When the Index Differences were computed between the adult BEecf and the delta Bicarbonate in neonates and older infants, the adult BE-ecf produced estimates of the magnitude of acidosis exceeding the delta Bicarbonate by 2 or more meq/liter in 42.6% of the samples. In two samples (0.74% of the total) the overestimate was 5 meqkter (Table 1). Group III
The Index Differences were computed between the age-adjusted and adult BE-ecf values. In 4.9% of the samples the adult BE-ecf were lower by 1 meq/liter using the age-adjusted value as the point of comparison. There was exact agreement between 44.2% of the values. The adult BEecf indicated a disturbance which was 1 meq/liter more severe than that of the adjusted BE-ecf in 44.8%. The discrepancy was greater in approximately 6%. In 1 sample (0.61% of the total) the adult BE-ecf overestimate was 7 meq/liter (Table 2).
7-l
II:\NltLl.
.I. F,SSIX
DISCUSSION Since Hasselbalch (1 I) first described the relationship between plasma bicarbonate, pH, and PCO,. the plasma bicarbonate has been used air an indicator of acidosis (I?). This ignores the presence of the significant quantities of nonbicarbonate buffers which are present ( 13). The calculations required to compute an index which includes these additional buffelsystems are complex when compared to the Henderson-Hasselbalch equation. Several of the popular indices were developed, in part, because they could be represented graphically and involved few variables, not necessarily because of physiological merit. The resulting simplifications have been the subject of numerous controversies. All of the techniques attempt to remove the effect of abnormal levels of ventilation so that the magnitude of the acid-base disturbance. were the PCO:, normal (40). can be determined. Using a similar approach the pH can be adjusted to 7.4 before the metabolic component is computed. In addition, since blood has a different buffering capacity in ~‘itro than it has in \vi~*o, it is possible to calculate a variety of indices. Table 6 presents a comparison of the assumptions and mathematical corrections involved in each index. Of the five indices. the BE-ecf incorporates the most physiological set of assumptions. It removes the direct influences of pH and PCOz on the index by correcting both to a pH of 7.4 and a PCO, of 40 and thereby effectively removes the influence of any ventilatory abnormality which would tend to increase the body concentration of acid. In order to produce any index, some assumption must be made about the buffering capacity of the body. The most appropriate value for buffering capacity is one whjch can be derived theoretically and confirmed experimentally. The BE-ecf meets these criteria. The theoretical derivation makes the following assumptions: ( 1) hemoglobin. bicarbonate, and plasma proteins constitute the clinically significant buffers. I‘ABLE 6 CHARACTERISTICS
OF INDICES
Value influenced by nonbicarbonate buffers delta Bicarbonate Standard Bicarbonate Base Excess of blood T40 Bicarbonate Base Excess of ECF NA-not
applicable.
+ + +
OF
Acm-Basr;
BAI.ANCE
Standard conditiona Reference system NA
PCO, 40
PH 7.40
NA
NA
I
in in
Vitro Virro
t
t
in in
Viw b’ilo
+ *
.~ +
BASE
EXCESS
IN CHILDREN
75
(2) The intravascular and extracellular bicarbonate pools buffer as if they form a single compartment. These assumptions predict a buffering capacity of approximately 12 Slykes in adults. This has been confirmed experimentally by Brackett at al. (14). Indices of the metabolic component of acid-base balance which do not incorporate the most current physiological concepts must prove their value. If they are to be retained as diagnostic and therapeutic tools, they should be sufficiently accurate and predictable to preclude incorrect diagnostic conclusions and to avoid wide or unexpected swings in pH during therapy. For the other indices under study to be considered predictable, any deviation from the BE-ecf must be of a direction and magnitude which is proportional to the magnitude of the metabolic disturbance. Predictability was evaluated by relating the index difference (the difference between each index and the BE-ecf) to the raw data because the index used clinically to estimate the metabolic component is given different weight depending on any disturbances which are present in the raw data. If a given index maintained a constant index difference across a wide range of pH, PC02, and bicarbonate values, it could be considered highly predictable. On the other hand, large changes in the index difference would suggest that even in experienced hands, therapy based on such an index could have undesired and seemingly unpredictable effects. The BE-blood and the Standard Bicarbonate resulted in total error rates of 11.4 and 12.376, respectively, when compared to the BE-ecf. With errors of this magnitude there must be some mitigating conditions to justify their continued use. For example, were the errors directly proportional to the pH, it might be possible to learn through experience the degree to which the index should be “disbelieved.” An examination of Fig. 1 will reveal that the degree to which the BE-blood is in error is linear and consistently overestimates the magnitude of the metabolic component at low pH. The Standard Bicarbonate on the other hand is inconsistent in that it underestimates the disturbance at both low and high pH. There is no reason to retain the Standard Bicarbonate as a clinical tool in view of this behavior. The BE-blood, on the other hand. can be mentally adjusted by an experienced clinician. It is unclear why this effort should be made since the error can be eliminated by substituting the in \,ivo buffering capacity in the calculations in place of the in t’itro value. The T40 Bicarbonate would lead to erroneous conclusions about the presence or absence of metabolic acidosis in only 1.1% of the samples; however, these samples which were considered to be in error tended to underestimate the degree of acidosis at low pH. The difficulty of the
computation is comparable to that of the BE-ecf. The T40 Bicarbonate. therefore, offers no advantage over the BE-ccl’. Figure 3 demonstrates another aspect of the Standard Bicarbonate and the T40 Bicarbonate which may interfere with their therapeutic application. When the Plasma Bicarbonate is low, these two indices significantly underestimate the magnitude of the metabolic component. In this situation the Standard Bicarbonate and the T40 Bicarbonate may indicate a smaller metabolic component than the Plasma Bicarbonate leading to confusion about or distrust of the blood gas report. The delta-Bicarbonate underestimated the degree of acidosis consistently and would have led to 6.352 erroneous conclusions about the presence or absence of metabolic acidosis when compared to the BE-ecf. This tendency to underestimate acidosis is expected since there are significant buffers present in addition to bicarbonate. In the presence of severe anemia, low concentration of plasma proteins or marked increases in the extracellular fluid volume, the delta-Bicarbonate may be useful in providing a conservative guide to therapy if an appropriately corrected BE-ecf cannot be readily calculated. The acid-base physiology of children differs from that of adults. Infants and especially prematures have lower concentrations of plasma proteins. The hemoglobin concentration is age related, and in the critical care setting often highly variable. The extracellular fluid volumes vary widely. though predictably, with both pre- and postnatal age, being relatively larger in smaller babies. The renal compensatory mechanisms may be inadequate, especially in the immediate postnatal period. Consequently. most young children have a lower buffering capacity than adults. Every effort should be made to estimate the buffering capacity on an individual basis as well as to choose the most physiologically relevant formula for calculating an index of the metabolic component of an acid-base disturbance. The considerations which lead the BE-ecf to be the most appropriate index for use in adults also make it the index of choice for use in children if it is adjusted for the measured hemoglobin concentration and for the known differences in extracellular fluid volume and plasma protein concentrations. When the BE-ecf computed using adult formula was compared to a BE which was adjusted for approximate extracelhtlar fluid volume. plasma protein concentration. and hemoglobin concentration, 51% of the samples demonstrated a lesser acid-base disturbance. While most of the differences were small, occasional differences were substantial. These larger differences were in samples with the most severe acidosis (adult BE-ecfs between - 8 and - 20). Because administration of bicarbonate is much more likely in more severe acidosis, the overestimation which results from the adult BE-ecf represents a significant potential risk to the infant. Distinguishing
BASE
EXCESS
IN CHILDREN
77
between child and adult buffering capacity yields clinically useful information and should therefore be done routinely. The comparison between the delta-bicarbonate in infants and the standard (adult) method of calculating BE-ecf was done to evaluate the hypothesis that in the presence of reduced buffering capacity the delta-Bicarbonate may be a useful, conservative estimate of metabolic acidosis. In over 99% of the samples the delta-Bicarbonate was either equal to the adult BE-ecf or smaller. This suggests that if an age-adjusted BE is not available and there is uncertainty about the normality of the extracellular fluid volume, hemoglobin concentration, or the plasma protein concentration. the delta-Bicarbonate yields a conservative number which can be used safely. While the BE-ecf, when appropriately ad.justed, provides a reliable estimate of the metabolic component of an acid-base disturbance, a caveat must be observed. The foregoing discussion has been centered primarily on the utilization of the BE-ecf in the evaluation of acute changes. As Schwartz and Relman (8) have emphasized, in many clinical settings a subject’s chronic or baseline state is not normal. There may be an ongoing metabolic or ventilatory abnormality which is partially compensated by homeostatic mechanisms. Under these conditions, the BE-ecf may remain relatively constant but at a value outside the normal range. In order to assess an acute change in the subject’s status. the magnitude and direction of the change must be referred to the subject’s baseline. Thus, the effective use of the BE-ecf requires close attention 10 the context in which it is obtained as well as a thorough understanding of the physiological principles involved. SUMMARY The Base Excess of the Extracellular Fluid (BE-ecf) is the most complete model of acid-base physiology. It has gained considerable acceptance for use in adults as an index of the metabolic component of an acute acid-base disturbance. Several other commonly used indices were compared to the BE-ecf. The values of the Base Excess of Blood, Plasma Bicarbonate, Standard Bicarbonate, and T40-Bicarbonate differed significantly from the BE-ecf in 11.4, 6.3, 12.5, and 1.1% of samples, respectively. These differences are considered to be errors since the nonlinear relationship of the variables makes it difficult to clinically accomodate them. The standard (adult) form of the BE-ecf calculation overestimated the base excess by 1 meq/liter in 44.8% and by 2 meq/liter or more in 6.1% of samples from neonates and infants when compared to a form of the calculation which was individually adjusted based on the weight and hemoglobin concentration of each subject. Since it is no more difficult
to make these corrections than to ignore them. if the BE-ecf LY 10 bc used in neonates and infants the correction should be applied ACKNOWLEDGMENTS I thank John Mohler, Clarence Collier, Stanley ALen. and Miriam Wilson for their help at various stages in the preparation of this manuscript.
REFERENCES I. Roos, A.. and Thomas. L. J.. .Jr. The m vitro and in viva carbon dioxide dissociation curves of true plasma. Ane.v/lrr.sio/og.v 28, 1048 f 1967). 1. Schwartz, W. B.. and Relman. S. R. A critique of the parameters used in the evaluation of acid-base disorders. N. G/x(. J. Med. 268, 2.S (1963). 3. Armstrong, B. W., Mohler. .I. G., Jung. R. c’.. 1’1 ol The in vivo carbon dioxide titration curve. Lancer 1, 7SY ( 1966). 4. Siggard-Anderson. O., Engel. K., Jorgensen. K.. and Astrup. P. Micro method for determination of pH. carbon dioxide tension. base excess and standard bicarbonate in capillary blood. Stand. J. C/in. Ltrh. Intvst. 12, 172 (1960). 5. Astrup, P., Jorgensen, K., Siggard-Anderson. 0.. and Engel. K. Acid-base balance: New approach. Lunc,ef 1, IO35 (1960). 6. Collier, C. R., Hackney. J. D.. and Mohler. J. G. Use of extracellular base excess m diagnosis of acid-base disorders: A conceptual approach. CI~esr 61, 6s ( 1972). 7. Siggard-Anderson, 0. An acid base chart for arterial blood with normal and pathophysiological reference area\. .SU&. J. C/in. Lab. Irrwst. 27, 238 ( I971 ). 8. Cheek, D. B. Extracellular volume: Its structure and measurement and the intiuence of age and disease. J. PediutI-. 58, I ( 1961). 9. Friis-Hansen, B. Body water compartments in children: Changes during growth and related changes in body composition. Pediatrks 28, 2 ( 1961). IO. Essin, D. J. Calculation of an individualized base excess of the extracellular fluid fat children of varying size. Ahsfroc.l. C/in. Res. 26(Z). l62a (1978). des Blute\. II, Hasselbalch. K. A. Die “reduziertr” und die “regulierte” k!kiSerStOffiahi Biochern.
2. 74, 56 (1916).
I?. Van Slyke. D. D.. and Cullen. C;. E. Studies of acidosis. I. Bicarbonate concentration of blood plasma: Its significance. and its determination as a measure of acidosis. .I. Biol. Chenz. 30, 2X9 (1917). 13. Singer. R. B., and Hastings, A. B. Improved clinical method for estimation of disturbances of acid-base balance of human blood. Medicine 27, 223 ( 1948). 14. Brdckett, N. C. Jr.. Cohen. J. J.. and Schwartz. W. B. The carbon dioxide titration curve of normal man. N. EnPI. J. Med. 272, 6 (IYhSb.