Anesth Analg 2003;97:383-389
© 2003 International Anesthesia Research Society
PEDIATRIC ANESTHESIA
Adverse Events After Protamine Administration Following Cardiopulmonary Bypass in Infants and Children
Harry A. Seifert, MD MSCE*,
,
David R. Jobes, MD,
Thomas Ten Have, PhD MPH*,
Stephen E. Kimmel, MD MSCE, FACC*,
Lisa M. Montenegro, MD,
James M. Steven, MD SM, FAAP,
Susan C. Nicolson, MD, and
Brian L. Strom, MD MPH, FACP, FACE
*Center for Clinical Epidemiology and Biostatistics, Department of Biostatistics and Epidemiology, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania; and
Division of Cardiothoracic Anesthesiology, Department of Anesthesiology & Critical Care, The Children’s Hospital of Philadelphia, University of Pennsylvania, Philadelphia, Pennsylvania
Address correspondence to Harry A. Seifert, MD, Global Clinical Safety and Pharmacovigilance, GlaxoSmithKline, MC-4340, 1250 S. Collegeville Rd., Collegeville, PA 19426. Address e-mail to harryaseifert{at}comcast.net Reprints will not be available from the author.
 |
Abstract
|
|---|
We performed this study to determine the incidence of and risk factors for adverse events (AEs) in infants and children after the IV administration of protamine after cardiopulmonary bypass. In a retrospective cohort study, all relevant anesthesia records from a 3-yr period were examined to identify AEs after protamine. The AEs were then grouped into three categories by applying increasingly strict criteria. Among 1249 anesthesia records, there were no documented episodes of isolated or hypotension-associated right-sided cardiac failure or acute pulmonary dysfunction. The incidence of systemic hypotension after protamine was between 1.76% (95% confidence interval [CI], 1.11%–2.65%) and 2.88% (95% CI, 2.03%–3.97%), depending on the strictness of case definition. To identify risk factors, we performed a nested case-control study in which unmatched controls were randomly selected from the parent cohort at a 4:1 ratio to cases. Cases of hypotension after protamine were more likely during operations on girls (odds ratio [OR], 6.47; 95% CI, 1.66–32.8), after larger doses of protamine (OR, 1.88; 95% CI, 1.03–3.63), or after smaller doses of heparin (OR, 0.49; 95% CI, 0.17–0.67).
IMPLICATIONS: Systemic hypotension after protamine administration occurred in 1.76%–2.88% of pediatric patients having cardiac surgery. Female sex, larger protamine dose, and smaller heparin dose were each associated with increased risk. The development of protamine alternatives or prophylactic therapies may be useful for reducing the frequency of these events.
|
Introduction
|
|---|
Protamine is administered during cardiac surgery to neutralize the anticoagulant effects of heparin. Systemic reactions to protamine range in severity from mild to fatal. The most common reaction in adults is a transient decrease in systemic mean arterial blood pressure (MAP), which is usually associated with the rate of administration (1–4). The incidence and severity of reactions vary with clinical setting, patient population, and case definition. For adult cardiac surgery patients, the incidence of clinically significant protamine reactions varies from 0.1% (5) to 24% (6); most estimates are from 3.8% to 6% (7). The incidence of protamine reactions in children has not been established.
The diverse clinical presentations of protamine reactions have implicated a variety of physiologic mediators, and attempts to identify a single immune mechanism have yielded inconsistent results. A small prospective trial of preoperative histamine (H1 and H2) antagonists prevented the decreases in systemic blood pressure observed in 33% of untreated adult patients (8). In another study, one third of adults with histories of severe protamine reactions demonstrated neither immunoglobulin G nor immunoglobulin E antibodies to protamine (9). A prospective immunologic evaluation has been unsuccessful in predicting clinically significant reactions (10).
Risk factors for serious reactions to protamine in adults include prior reactions to the drug, allergies to other medications (11), allergies to fish (12), and previous exposure to IV protamine (6) or protamine-containing insulin preparations (11,13,14). The risk factors for protamine reactions may differ in children, because of their potentially less frequent or shorter duration of exposures to cross-reacting antigens and a lesser sensitivity of their pulmonary intravascular macrophages to heparin-protamine complexes (15).
We retrospectively examined a cohort to identify the frequency of adverse events (AEs) after intraoperative protamine administration in infants and children undergoing surgery that required cardiopulmonary bypass (CPB). We then performed a case-control study, nested within this cohort (i.e., both cases and controls were drawn from the parent cohort), to identify factors that were independently associated with AEs after protamine.
|
Methods
|
|---|
This study was approved by the Committee for the Protection of Human Subjects. The cohort consisted of electronic anesthesia records from all operations requiring CPB and protamine administration that were performed on children younger than 16 yr at The Children’s Hospital of Philadelphia from 1995 through 1997. Records were excluded from the cohort if there was a major surgical complication (defined as any complication requiring surgical correction) after protamine administration. These were excluded because they can cause cardiopulmonary dysfunction independent of protamine. The dates for inclusion in the cohort were selected to coincide with the installation of an automated anesthesia record system (CompuRecord®; Phillips Medical Systems, Andover, MA), which improved the accuracy of recorded variables and the detection of critical events (16) and facilitated retrieval of data. The CompuRecord system allowed manual entry of demographic information at the time of surgery, and it automatically collected and stored data from various monitoring devices every 15 s.
AEs were defined by three criteria. First, one of the following must have been documented: 1) a decrease of systemic MAP of at least 25% of baseline (defined as the average MAP during the 5 min preceding protamine administration) or of at least 10% and requiring inotropes, vasopressors, or reinstitution of CPB; 2) an increase in right atrial or central venous pressure of at least 10% from baseline associated with a decrease in MAP of at least 25% or requiring any intervention, as described above; 3) noncardiogenic pulmonary edema, evidenced by any decrease in oxygen saturation requiring an increase of inspired oxygen concentration, ventilatory rate, tidal volume, or ventilating pressures; or 4) bronchospasm, evidenced as wheezing, a peak airway pressure increase of at least 2 cm H2O, or bronchodilator therapy. Second, the events had to have occurred within 30 min after the administration of protamine (5,13). Finally, the events had to have lasted for at least 5 min. These criteria are based on those used for studies of protamine reactions in adults (5–11,13) and on the normal and abnormal hemodynamic effects of protamine in adults (2,3,17,18).
Because of the difficulty in ascribing causality to an event that temporally follows drug administration in a clinical setting in which there are many possible causes of that event, we grouped the AEs of interest into three categories by applying increasingly strict criteria. These classifications, which were based on those used in a similar study in adults (13), eliminated from successive categories those cases that were less likely to have been caused by protamine. The most inclusive category ("all cases") included all cases defined by the above criteria. Then cases identified by events that began >10 min after protamine administration were excluded, leaving those that are designated as "early cases." Finally, cases that could have been caused by factors other than protamine—such as increased doses of drugs known to cause hypotension—were excluded, leaving "early cases without concomitant events."
Descriptive statistics were used to characterize the cohort. The unit of analysis was the anesthesia record, each of which uniquely corresponded to one operation on one patient. The incidence of AEs after protamine administration per operation, with 95% confidence intervals (CIs), was calculated by dividing the numbers of events identified by each of the three case definitions by the number of operations in which a patient was at risk of experiencing an event. Demographic characteristics of subjects with and without AEs were compared by using Fisher’s exact test for binary variables or Wilcoxon’s ranked sum test for continuous variables.
A case-control study was nested within the cohort. Cases were identified as previously described. Because the number of cases was fixed, the statistical power of the study was enhanced by randomly selecting four times as many unmatched controls from the parent cohort (19). Data on risk factors and potential confounders in both cases and controls were obtained from the anesthesia record database and perfusionists’ hand-written records. The original printed copies of all anesthesia records from operations with AEs and a 30% random sample of controls were reviewed to ensure that hand-written comments did not contradict the electronically recorded data or document a potential etiology of an AE that met our case definitions. None of the sampled records contained any such comments.
The potential risk factors are listed in Tables 1 and 2. Some risk factors from other studies, including vasectomy (20) and the rate of protamine administration (2), were not considered in our study. This was because they were not germane for this cohort, in which no patients had undergone vasectomy and all patients received protamine as a bolus injection rather than as an infusion. The following factors were evaluated as potential confounders, because they could have been independently related to both protamine administration and the AEs of interest: year of operation, type of operation, ASA physical status (PS), elective versus emergency operation, surgeon, anesthesiologist, history of fish allergy, use of volatile anesthetics, duration of CPB, use of circulatory arrest, use of inotropes or vasopressors after CPB but before protamine administration, administration of IV calcium after CPB but before protamine, and histamine H1 or H2 receptor blockade before surgery. Because operation type was encoded as a free-text field, it was converted into a categorical variable. This was accomplished by convening a panel of experienced pediatric cardiac anesthesiologists to devise a classification scheme based on underlying pathophysiology. A history of fish allergy was not evaluated as a potential risk factor because of the expected rarity of this factor and the anticipation that the size of the cohort would limit the number of putative risk factors evaluated. The comment field of every case and control was reviewed to determine whether any notes of surgical manipulation accompanied hemodynamic perturbations after CPB.
View this table:
[in this window]
[in a new window]
|
Table 1. Unstratified, Unadjusted Univariate Evaluation of Risk Factors That Were Encoded as Binary Variables by Using the Primary Case Definition (i.e., Early Cases Without Concomitant Events)
|
|
View this table:
[in this window]
[in a new window]
|
Table 2. Unstratified Evaluation of Risk Factors That Were Encoded as Continuous Variables by Using the Primary Case Definition (i.e., Early Cases Without Concomitant Events)
|
|
The odds ratio (OR) was used to estimate the relative risk of AEs for operations with and without each binary risk factor. Fisher’s exact statistic and two-sided 95% CIs were determined for each OR that described the relationship between each binary risk factor and the AE. The Mann-Whitney test was used to compare, between cases and controls, the distribution of each risk factor that was evaluated as a continuous variable.
Stratified multivariate logistic regression was performed to determine the OR for each risk factor while adjusting for other risk factors and potential confounding factors. Relevant continuous variables were transformed into categorical variables by division into quartiles to allow their inclusion in exact logistic regression modeling. Potential confounding factors were included in the final models if their inclusion changed the unadjusted OR for any risk factor by >15% (21). To assess the specificity of the primary case definition and to better evaluate whether nominal risk factors were predictive of protamine reactions rather than hemodynamic perturbations unrelated to protamine administration, analyses were repeated and results compared as cases were defined by increasingly strict criteria.
Statistical analyses were performed with Stata 6.0 (Stata Corp., College Park, TX), StatXact 4 (Cytel Software Corp., Cambridge, MA), and LogXact (Cytel Software Corp.). Statistical significance was defined as a two-sided P value <0.05 or an OR whose exact 95% CI excluded unity.
|
Results
|
|---|
Inclusion criteria were met by 1249 anesthesia records, representing operations on 1074 patients. The number of operations (records) per patient ranged from 1 to 5; 87% of patients had only one record in the dataset. Fifty-seven percent (718 of 1249) of the operations were performed on boys. The median patient age at the time of operation was 208 days (range, 0–5799 days), and the median weight was 6.4 kg (range, <1 to 120 kg). Fifty-three percent of operations involved ASA PS 3 patients. There were very few PS 1 (3 of 1248; 0.24%) or PS 5 (17 of 1248; 1.36%) patients. One record lacked a PS classification. Emergencies accounted for 6.2% (77 of 1249) of operations.
Eight anesthesiologists and four surgeons participated in operations in the cohort. Two anesthesiologists were responsible for 70% of the records. One surgeon was responsible for 63% of the operations. The actual numbers of cases identified by each criterion for each case definition and the incidence of severe AEs after protamine administration are summarized in Table 3.
Exposures of cases and controls, by using the primary case definition (i.e., early cases without concomitant events) to identify the potential risk factors, are summarized in Tables 1 and 2. In unadjusted, univariate analyses with our primary case definition, female sex, smaller heparin dose, larger protamine dose, and having not received modified ultrafiltration (MUF) were associated with statistically significant increases in the risk of AEs. The variable "surgeon" was found to be a significant confounder for all of these risk factors except heparin dose. In the multivariate analysis (Table 4), female sex, larger protamine dose, and smaller heparin dose remained significantly associated with the risk of AEs by the primary case definition. "Surgeon" significantly confounded the associations of sex, protamine dose, and MUF with severe events. With stratification by surgeon, the protective effect of MUF was no longer statistically significant. However, removing MUF from the model did not substantially affect the results for the other risk factors. No instances of surgical manipulation associated with hemodynamic perturbations were documented in either the cases or controls.
View this table:
[in this window]
[in a new window]
|
Table 4. Univariate and Adjusted Multivariate Analyses of Risk Factors Having Significant Associations with Severe Events by Using the Primary Case Definition (Early Cases Without Concomitant Events)
|
|
The adjusted, multivariate risk factors significantly associated with AEs are shown in Table 4. The ORs for female sex and larger protamine dose each increased with progressively specific case definitions and were statistically significant, supporting the hypothesis that each of these factors is associated with an increased risk of protamine AEs (Table 5). The OR for heparin dose decreased by approximately 25% from the most inclusive case definition to the definition of intermediate sensitivity and specificity, but it did not change between this and the most specific definition, so the nature of the association of this factor with protamine-associated AEs is less clear. The association between sex and the primary case definition was significantly confounded by heparin dose. The OR for MUF was not statistically distinguishable from unity.
View this table:
[in this window]
[in a new window]
|
Table 5. Associations Between Risk Factors and Severe Adverse Events by Using Progressively More Specific Case Definitions: Multivariate Odds Ratios Have Been Adjusted as Described in the Text
|
|
|
Discussion
|
|---|
This study demonstrates that hemodynamic events, manifested as systemic arterial hypotension sustained for at least 5 minutes, occur after protamine administration in as many as 2.88% (95% CI, 2.03%–3.97%) of cardiac operations performed on patients younger than 16 years. There are many potential reasons for hemodynamic instability after CPB, so it is likely that not all of the AEs identified by this definition were caused by protamine. By applying more stringent criteria to limit the time after drug administration and exclude known, potential causes of hemodynamic compromise, the measured risk decreased to 1.76% (95% CI, 1.11%–2.65%). However, even the most conservative incidence, as determined by our primary case definition, probably includes some events not caused by protamine.
The challenge of ascribing causality on an individual patient basis produces the need for an epidemiologic study such as ours. However, we believe that it is unlikely that unrecorded factors account for a significant proportion of the cases of arterial hypotension after protamine, for at least three reasons. First, at our institution, we can recall no instances in which mechanical manipulation of the heart by the surgeon produced hemodynamic changes that persisted for five minutes or longer, and this would be minimum duration required to fulfill our case definitions. Second, review of the hard-copy records from all cases and a 30% random sample of controls revealed no electronic or hand-written comments or notes of other events that produced the hypotension. Third, our analysis of risk factors by specificity of case definition showed that the ORs associated with female sex and protamine dose progressively increased and the OR associated with heparin dose decreased with increasingly specific case definitions. If the case definitions were not increasingly specific for protamine reactions, these trends would have been unlikely to have occurred; i.e., if a significant proportion of what we identified as AEs were actually other events unassociated with protamine, it would be likely that the risk factors associated with these other events would be different from the risk factors for protamine reactions, making the monotonically progressive increase or decrease (with application of increasingly specific case definitions) of the OR for a given risk factor unlikely. Furthermore, excluding from the primary case definition operations in which there was concomitant administration of vasodilators or cardiac depressants may have also excluded true protamine reactions. Thus, we believe that is unlikely that our primary definition significantly overestimates the actual incidence of protamine-associated reactions.
This is the first published study that quantifies the incidence of AEs after protamine in infants and children. Among adults, studies have identified the incidence of AEs attributed to protamine as 0.1%–33% (5,8,11). In general, studies that relied on physician attribution identified a decreased risk, and those that applied objective criteria identified a larger risk. Some studies that identified fewer incidences of AEs also focused on pulmonary hypertension or anaphylactic reactions (11). A study that was similar to ours, but that relied on hemodynamic variables that were manually recorded during coronary artery bypass surgery on adults, identified a 12.9% incidence of AEs after protamine administration. After events with other potential causes were excluded, the incidence was 2.6% (7), which is consistent with the incidence identified in this study.
Our study also demonstrated that female sex, a larger protamine dose, and a smaller heparin dose are each independent risk factors. The other potential risk factors were not associated with increased probabilities of AEs. The risk factors identified as significant in our study differ from those identified in a similar study of adults (11). However, the relative rarity in the pediatric population of at least two of the risk factors for reactions in adults—neutral protamine Hagedorn insulin use and fish allergy—decreases the potential clinical importance of those factors. The reason for the more than sixfold increase in risk for girls is obscure but does not appear to have been related to sex differences in age at operation, weight, type of operation, or other measured, potentially confounding factors. Drug-associated reactions in general are more likely to be reported for females (22), although the relative contribution of increased underlying risk versus increased reporting is undefined. The biological reasons for this sex difference remain obscure.
The design of this study eliminates the possibility of reporting bias. The modest increase in risk associated with larger protamine doses suggests a threshold or dose-dependent mechanism for at least a proportion of the events. The protective effect of larger heparin doses may imply that larger amounts of residual heparin bind more of the administered protamine and that heparin-protamine complexes are less likely than unbound protamine to cause hypotension. Protamine and heparin dosages were not significant factors in a similar study of adults (13), implying that the methods of selecting heparin and protamine doses, and the pharmacologic mechanisms of AEs after protamine, may differ between children and adults.
The analysis by specificity of definition of AEs supports the hypotheses that female sex, larger protamine dose, and, possibly, smaller heparin dose are each associated with protamine reactions. The use of MUF appeared to be a significant protective factor in the unstratified, univariate analysis. In this cohort, any specific effect of MUF on AEs could not be distinguished from differences associated with different surgeons. In addition, MUF did not demonstrate a steadily decreasing OR with increasingly specific case definitions, and this would be expected if it were truly associated with the reduced risk of AEs after protamine administration.
One of the greatest limitations of our study is that it is not possible to identify what proportion of severe events was caused by unrecorded factors. Our primary case definition yielded an incidence of 1.76%, which could be considered as an upper limit of the actual incidence of AEs caused by protamine. Another limitation is that our cohort was drawn from a single institution and may not be generalizable to other institutions. In addition, our statistical power to assess the risks associated with factors that were found to be significant by other studies (i.e., prior exposure to protamine-containing insulin preparations, prior protamine reactions, and history of nonprotamine drug allergies) is limited by the rarity of these exposures in our cohort.
The results of this study also must be interpreted with the cautions applied to all observational studies. Identifying an association by random chance (type I error) is unlikely, because we focused on a limited number of clinically plausible risk factors. Not identifying true associations between risk factors and severe AEs (type II errors) is unlikely for all factors except history of protamine allergy or history of other drug allergies. Finally, patient outcomes have not been linked to the AEs identified in this study. The influence of these events on survival or neurological outcome remains unknown for this patient population. However, in one study in adults, AEs after protamine administration were associated with an increased risk of mortality (23).
Our study also had several strengths. Severe reactions were identified on the basis of predefined criteria that were independent of clinician attribution (7). The hemodynamic variables on which the case definitions were based were recorded by an automated system, which recorded data that are more precise and free of bias than standard, hand-written records (15). All source data were available, thus eliminating one potential source of selection bias.
In conclusion, this retrospective cohort study demonstrated that the incidence of AEs after protamine administration during operations on children younger than 16 years is similar to the incidence for adults. Female sex, larger protamine doses, and smaller heparin doses were independent risk factors for severe events in this pediatric population. Because there are no proven techniques to prevent severe reactions to protamine in children, future research is warranted to develop prophylactic strategies or alternative therapies in this population.
|
Acknowledgments
|
|---|
This study was supported by a grant from Ibex Corp.
|
References
|
|---|
- Conahan T, Andrews R, MacVaugh H. Cardiovascular effects of protamine sulfate in man. Anesth Analg 1981; 60: 33–6.[Abstract/Free Full Text]
- Horrow JC. Protamine: a review of its toxicity. Anesth Analg 1985; 64: 348–61.[Free Full Text]
- Ovrum E, Lindberg H, Holen EA, et al. Systemic and pulmonary circulatory effects of protamine following cardiopulmonary bypass in man. Scand J Thorac Cardiovasc Surg 1991; 25: 19–24.[ISI][Medline]
- Schapira M, Christman BW. Neutralization of heparin by protamine. Circulation 1990; 82: 1877–9.[Free Full Text]
- Levy JH, Schwieger IM, Zaidan JR, et al. Evaluation of patients at risk for protamine reactions. J Thorac Cardiovasc Surg 1989; 98: 200–4.[Abstract]
- Weiler JM, Gellhaus MA, Carter JG, et al. A prospective study of the risk of an immediate adverse reaction to protamine sulfate during cardiopulmonary bypass surgery. J Allergy Clin Immunol 1990; 85: 713–9.[ISI][Medline]
- Kimmel S, Sekeres M, Berlin J, et al. Adverse events after protamine administration in patients undergoing cardiopulmonary bypass: risks and predictors of under-reporting. J Clin Epidemiol 1998; 51: 1–10.[ISI][Medline]
- Kambam J, Meszaros R, Merrill W, et al. Prophylactic administration of histamine1 and histamine2 receptor blockers in the prevention of protamine-related haemodynamic effects. Can J Anaesth 1990; 37: 420–2.
- Weiss M, Nyhan D, Peng Z. Association of protamine IgE and IgG antibodies with life-threatening reactions to intravenous protamine. N Engl J Med 1989; 320: 886–92.[Abstract]
- Levy JH, Zaidan JR, Faraj B. Prospective evaluation of risk of protamine reactions in patients with NPH insulin-dependent diabetes. Anesth Analg 1986; 65: 739–42.[Abstract/Free Full Text]
- Gottschlich GM, Gravlee GP, Georgitis JW. Adverse reactions to protamine sulfate during cardiac surgery in diabetic and non-diabetic patients. Ann Allergy 1988; 61: 277–81.[ISI][Medline]
- Knape J, Schuler J, Haan PD. An anaphylactic reaction to protamine in a patient allergic to fish. Anesthesiology 1981; 55: 324–5.[ISI][Medline]
- Kimmel S, Sekeres M, Berlin J, et al. Risk factors for clinically important adverse events after protamine administration following cardiopulmonary bypass. J Am Coll Cardiol 1998; 32: 1916–22.[Abstract/Free Full Text]
- Gupta SK, Veith FJ, Ascer E, et al. Anaphylactoid reactions to protamine: an often lethal complication in insulin-dependent diabetic patients undergoing vascular surgery. J Vasc Surg 1989; 9: 342–50.[ISI][Medline]
- Horiguchi T, Enzan K, Kawamura K, Suzuki M. Pulmonary responses to heparin-protamine complexes: the effects of age and species. J Appl Physiol 1996; 80: 56–61.[Abstract/Free Full Text]
- Sanborn KV, Castro J, Kuroda M, Thys DM. Detection of intraoperative incidents by electronic scanning of computerized anesthesia records. Anesthesiology 1996; 85: 977–87.[ISI][Medline]
- Kirklin JK, Chenoweth DE, Naftel DC, et al. Effects of protamine administration after cardiopulmonary bypass on complement, blood elements, and the hemodynamic state. Ann Thorac Surg 1986; 41: 193–9.[Abstract]
- Shapira N, Schaff HV, Piehler JM, et al. Cardiovascular effects of protamine sulfate in man. Thorac Cardiovasc Surg 1982; 84: 505–14.
- Schlesselman JJ. Case-control studies: design, conduct, analysis. New York: Oxford University Press, 1982.
- Adourian U, Shampaine E, Hirshman C, et al. High-titer protamine-specific IgG antibody associated with anaphylaxis: report of a case and quantitative analysis of antibody in vasectomized men. Anesthesiology 1993; 78: 368–72.[ISI][Medline]
- Mickey RM, Greenland S. The impact of confounder selection criteria on effect estimation. Am J Epidemiol 1989; 129: 125–37.[Abstract/Free Full Text]
- van der Klauw AM, Wilson JHP, Stricker BHC. Drug-associated anaphylaxis: 20 years of reporting in the Netherlands (1974–1994) and review of the literature. Clin Exp Allergy 1996; 26: 1355–63.[ISI][Medline]
- Kimmel SE, Sekeres M, Berlin JA, Ellison N. Mortality and adverse events following protamine administration in patients undergoing cardiopulmonary bypass. Anesth Analg 2002; 94: 1402–8.[Abstract/Free Full Text]
Accepted for publication April 3, 2003.
Top
Abstract
Introduction
