Cholecystectomy in Sickle Cell Anemia Patients: Perioperative Outcome of 364 Cases From the National Preoperative Transfusion Study

CHOLELITHIASIS IS VERY common in patients with sickle cell anemia (SCA), prevalence rates ranging from 30% to 70% depending on age and diagnostic criteria.1-5 Cholecystectomy for both symptomatic and asymptomatic gallstones is commonly performed in patients with this disease, and it is the most frequently performed surgical procedure in this population.6 However, debate continues concerning risks and optimal perioperative management for SCA patients undergoing cholecystectomy.

We have previously reported the results of the multiinstitutional, prospective, and randomized Preoperative Transfusion Study that investigated conservative versus aggressive transfusion in 604 surgical procedures in patients with SCA.7 In this study, cholecystectomy comprised almost 40% of the procedures performed. The study included a registry of an additional 523 patients with SCA who underwent surgery at participating centers, but who were not randomized for preoperative transfusion therapy (ie, patients who were not transfused preoperatively or who were transfused without randomization). A total of 364 patients from both the randomized population and the nonrandomized registry underwent cholecystectomy, nearly all for symptomatic cholelithiasis. In this report, we compare the patient profile, perioperative management, risk factors, and outcome among the different preoperative transfusion groups. In addition, we compare the effect of open versus laparoscopic cholecystectomy on outcome.

Patients were enrolled from 36 centers. Each institution had a principal investigator, nurse, and data coordinator, as well as an anesthesiologist and a surgeon in most cases. As previously described, a protocol was followed from patient registration through the 30-day postoperative follow-up period, and patient data were collected during this complete perioperative period. Institutions were randomly site-visited for verification of patient records.7 

Patient population.Patients were eligible for randomization to a preoperative transfusion group if they had a diagnosis of hemoglobin SS or S-beta-0 and were undergoing elective surgery (defined as surgery occurring 24 hours or longer after registration). These patients were randomized to one of two transfusion groups: group 1, Aggressive Transfusion to obtain a preoperative hemoglobin of 10 g/dL (range, 9 to 11) and hemoglobin S percent of 30 or less; or group 2, conservative transfusion, to obtain a hemoglobin of 10 g/dL (range, 9 to 11), independent of hemoglobin S percent.

Patients were eligible for study registry, but not randomization, if they had been transfused within 3 months before surgery or if they or their physicians declined randomization. These patients were registered in one of two groups: group 3, nontransfusion; or group 4, nonrandomized transfusion. Seventy percent of patients in group 3 were not transfused due to the discretion of the principal investigator; 8% due to parent/patient refusal; and 22% for unstated reasons. Thirty-two percent of patients in group 4 were excluded from randomization due to recent blood transfusion; 27% due to parent/patient refusal; 21% due to discretion of the principal investigator; the remainder for unstated reasons. Approximately 53% of the patients in groups 3 and 4 were registered late (within 3 weeks of surgery).

From August 1988 through August 1993, 398 patients with SCA undergoing cholecystectomy at a participating institution were enrolled in the study after informed consent. Thirty-four patients were subsequently removed from enrollment due to cancellation of the procedure or ineligibility due to incorrect hemoglobinopathy diagnosis.

Perioperative management and data collection.Primary surgical procedure, incidental procedures, and surgical method were determined by the attending surgeon in consultation with the attending hematologist. The standardized treatment and transfusion protocols have been described previously.7 Patients in the randomized groups were transfused so that their final preoperative hemoglobin was between 9 g/dL and 11 g/dL, but the decision to use exchange or simple transfusion was made by the local investigator. All patients received sickle-negative blood and, if there was a history of febrile transfusion reactions, leukocyte-depleted blood.8 

The patients' medical history and information regarding perioperative transfusions, surgery, anesthetic management, and recovery period were recorded. Each patient's anesthetic risk (ie, American Society of Anesthesiologists, or ASA physical status) was determined,9 and extensive perioperative laboratory information was obtained.7 Complications occurring from study entry through a 30-day follow-up period were monitored and classified as minor, serious, or life-threatening. Most intraoperative and recovery room complications were graded as serious. Six additional serious or life-threatening complications were specifically defined: alloimmunization, vaso-occlusive pain crisis, acute chest syndrome (ACS), neurologic dysfunction, renal dysfunction, and fever with infection.7 Vaso-occlusive crisis and ACS are considered to be unique to this population and were analyzed in detail as “sickle cell events.”

Statistical analysis.Serious and life-threatening complication rates were compared across the four transfusion groups using the chi-square or Fisher exact tests for proportions. Confidence intervals were two-sided and a P value of .05 or less was considered significant. Patients undergoing laparoscopic cholecystectomy were compared with those undergoing open cholecystectomy using chi-square and Fisher exact tests. For comparison of means, the Student's t-test or an analysis of variance (ANOVA) was applied where appropriate.

To determine risk factors for complications, multivariate logistic regression models were developed for the presence or absence of any serious or life-threatening complication and of a sickle cell event. The models included age and any variable with significant univariate relationship (P ≤ .05) to these outcome variables.

A multiple regression model was developed for the number of days of patient hospitalization. Independent variables included age, transfusion group, and risk factors with significant univariate relationship (P ≤ .05) to the number of days of hospitalization.

Patient profile.A total of 364 patients underwent cholecystectomy: 110 in group 1, 120 in group 2, 37 in group 3, and 97 in group 4 (Table 1). Of the total population, 21% were less than 10 years of age and the youngest was 2.6 years. Ninety-eight percent of the patients had symptomatic cholelithiasis, although complete information about chronicity of symptoms was not available for all patients. One third of all patients had a history of significant pulmonary disease, defined as previous episodes of ACS, recurrent pneumonia, emphysema, or tuberculosis. An additional 4% had a history of asthma; 4% had baseline hypoxia (ie, PaO₂ less than 60 mm Hg or oxygen saturation less than 90% in room air); 11% had an abnormal baseline chest roentgenogram. Twenty-one percent had baseline elevation of serum glutamate pyruvate transaminase. The clinical and laboratory characteristics of patients in randomized groups 1 and 2 were similar to each other and to those in the nonrandomized transfused group 4. The patients in nontransfusion group 3 were significantly older (P ≤ .001), and they tended to be female, healthier by ASA score, and smokers. For groups 3 and 4, there was no difference in all patient characteristics between those patients registered before and those registered after surgery.

Transfusion preparation.The preoperative hemoglobin and hemoglobin S levels were 10.5 g/dL) and 48%, respectively for the population at large, nearly one third of whom received an exchange transfusion in the perioperative period (Table 2). The preoperative hemoglobin values in groups 1 and 2 were similar, while their preoperative hemoglobin S percent differed, as expected by the study's transfusion protocol. Preoperative hemoglobin and S percent in group 4 were similar to that of group 1. The nontransfusion group 3 had the lowest preoperative hemoglobin and highest hemoglobin S percent. Six patients in this group received blood: two for operative blood loss, three for postoperative complications, and one for both.

Perioperative management.Preoperative, intraoperative, and postoperative management for all patient groups was generally uniform: 88% received preoperative hydration; 93% were monitored intraoperatively according to protocol; and 90% were provided with at least one intraoperative modality of body temperature conservation. General anesthesia with inhalation and intravenous agents was used in 91%, and combined epidural-general anesthesia was used in 8%. Postoperative analgesia was provided in all patients: intravenous narcotics in 92%, patient-controlled analgesia in 16%, and epidural analgesia in 10%. Postoperative oxygen therapy was provided in 96%, intravenous hydration in 100%, antibiotic therapy in 84%, and incentive spirometry in 42%. There were no differences among the four groups of patients in any of these aspects of perioperative care.

Complications.The overall complication rate was 39% percent (Table 3). A sickle cell event (ie, vaso-occlusive pain episode and/or ACS) occurred in 18% of patients and was more frequent in group 3 patients: 12% in group 1, 19% in group 2, 32% in group 3, and 18% in group 4 (P = .041). The rate of sickle cell events remained significantly higher in group 3, even when all retrospectively-registered patients were excluded from analysis. Intraoperative/recovery room events (eg, oxygen saturation <90%, temperature <35.5°C or >38.5°C, blood loss > than 10% of blood volume) occurred in 11%. Significant bleeding occurred in 22 patients (6%), 7 of whom required exploratory laparotomy following cholecystectomy for control of bleeding. Bleeding was not correlated with patient group, surgical technique (open v laparoscopic), or baseline elevation of the index liver function test. Neurological complications occurred only in group 4 patients (P = .040), many of whom were on chronic transfusion protocols for neurologic disease. Use of epidural anesthesia-analgesia in the intraoperative and/or postoperative periods had no correlation with postoperative complications.

Three deaths occurred in the study population, two in group 3 and one in group 4. A 13-year-old boy underwent open cholecystectomy without preoperative transfusion. His course was uncomplicated until 48 hours postoperatively when he developed ACS and respiratory failure, which progressed to death on the 10th postoperative day. This patient received full supportive care, although exchange transfusion was delayed due to difficulty in obtaining compatible blood. An 18-year-old boy underwent open cholecystectomy without transfusion during what appeared in retrospect to be an acute hepatic sequestration crisis. The postoperative course was complicated by hemorrhagic shock and coagulopathy. On the 10th postoperative day, the patient died after an attempted computed tomography (CT)-guided needle aspiration of an inflammatory mass around the gallbladder. A 40-year-old woman underwent open cholecystectomy following transfusions for a febrile illness thought to be acute cholecystitis, although her gallbladder showed no acute pathologic changes. She expired on the 20th postoperative day after a course complicated by ACS, pyelonephritis, Haemophilus influenzae pneumonia, coagulopathy with a large retroperitoneal hematoma, and possible transfusion reaction.

Open versus laparoscopic cholecystectomy.Fifty-eight percent of the cholecystectomies were open procedures and the remainder were laparoscopic10 (Table 4). The development of the laparoscopic technique during the course of the study was reflected by a chronologic profile of the technique's application: 14% of the procedures during the first half of the study were laparoscopic, while 70% of those during the second half were laparoscopic. Five percent of the laparoscopic procedures were converted to open procedures and were analyzed as laparoscopic. The open and laparoscopic groups were similar in makeup except that patients in the laparoscopic group were older (P = .027) and healthier by ASA physical status score (P = .011), while those in the open procedure group were more likely to have baseline elevation of the index liver function test (P = .0002). Fourteen of the patients in the laparoscopic group were less than 10 years of age, and the youngest was 3.5 years.

Patients in the open and laparoscopic groups had similar preoperative hemoglobin levels (10.6 and 10.4 g/dL, respectively) (Table 5); hemoglobin S percent was 45% and 52%, respectively (P ≤ .01). Administration of preoperative hydration, as well as the use of intraoperative monitoring and temperature control modalities occurred with similar frequency in the two groups. Combined epidural-general anesthesia was used in 13% of the open procedure patients and in only 1% of laparoscopic procedures (P ≤ .0001). Postoperative administration of supplemental oxygen therapy and intravenous fluids also occurred with similar frequency in the two groups. However, open cholecystectomy patients were more likely than laparoscopic cholecystectomy patients to receive postoperatively epidural analgesia (17% v 1% [P ≤ .0001]), parenteral antibiotics (89% v 76% [P = .002]), and incentive spirometry (50% v 31% [P = .0003]).

Open cholecystectomy patients were more likely than laparoscopic patients to have an intraoperative cholangiogram, to have common bile duct stones, and to undergo incidental procedures with cholecystectomy (Table 5). The most common of these incidental procedures were appendectomy (50 patients), liver biopsy (24 patients), and umbilical herniorrhaphy (13 patients). The two surgical groups had a similar frequency of extended procedures, ie, intended cholecystectomy, which was converted to choledochoduodenostomy, choledochojejunostomy, or hepatojejunostomy.

Total anesthesia time was shorter for open cholecystectomy than for laparoscopic (2.9 v 3.2 hours [P = .007]) (Table 5). This similarity persisted when those patients in each group who underwent incidental or extended procedures were excluded. Analysis of individual institutions that performed 10 or more laparoscopic cholecystectomies showed no chronologic decrease in anesthesia time, and the mean anesthesia time for all laparoscopic patients was 3.0 hours in the first half of the study and 3.3 hours in the second half. Hospitalization time in the open group was longer than in the laparoscopic group (9.8 v 6.4 days [P = .0001]). Total complication rate and the rate of all specific complications were similar in open and laparoscopic patients (Table 6).

Predictors of complication.To identify predictors of serious or life-threatening complications, 18 possible risk factors were abstracted from patient data (Table 7). In univariate analysis, older patients, patients with an extensive transfusion history, and those undergoing incidental procedures were more likely to have serious complications. In the multivariate logistic regression including these significant risk factors, only patient age and the presence of incidental procedures were independent predictors of any serious complication (Table 7). In the univariate analysis of 18 possible predictors of sickle cell events, absence of preoperative transfusion and a history of frequent hospitalization were significant risk factors. In the multivariate logistic regression including these two risk factors and patient age, absence of preoperative transfusion and frequent hospitalization remained as independent predictors of sickle cell events (Table 7).

Predictors of length of hospitalization.The average length of hospitalization was 8.4 days. In a univariate analysis of possible predictors of length of hospital stay, higher ASA physical status, positive transfusion history, baseline chest roentgenogram abnormality, open surgical technique, and presence of incidental procedures were factors that were significantly correlated with increased length of hospitalization. In the multivariate regression model including patient age and these risk factors, only higher ASA status, chest roentgenogram abnormality, and open surgical technique remained as significant independent predictors of increased length of hospitalization (Table 8).

Cholecystectomy is the most common surgical procedure performed in patients with sickle cell anemia.6 This report of 364 patients with SCA and symptomatic cholelithiasis undergoing cholecystectomy constitutes the largest collection of this particular surgical population, and it provides important information concerning the perioperative outcome of these patients. We demonstrate that the rate of complication is high in these patients even with high-quality perioperative care, and that the rate of specific sickle cell events in patients who did not receive preoperative transfusion is increased compared with those who did. In addition, we demonstrate that laparoscopic technique decreases hospitalization time in this population without increasing complications.

The morbidity in our patients is higher than that of previous reports describing individual institutions with sickle cell patients undergoing open11-17 and laparoscopic18 cholecystectomy for both symptomatic and asymptomatic disease. These reports contained small numbers of patients (9 to 42) managed with different transfusion regimens, and these reports did not uniformly assess patient risk factors, or include transfusion-related complications. However, Griffin et al19 did report a postoperative complication rate of 50% in a subset of their sickle cell patient population who underwent thoracotomy or laparotomy without preoperative transfusion. The mortality rate in our population is similar to that of previous reports (<5%).

The occurrence of any serious complication in the perioperative period was correlated with increased patient age and the presence of incidental surgical procedures. SCA patients beyond the second decade of life would be expected to have relatively high morbidity and mortality20 even in the absence of surgery. However, it is not clear why incidental surgical procedures should affect outcome. These surgical procedures may affect morbidity by increasing anesthesia time, although anesthesia time itself was not an independent predictor of complications. Alternatively, the particular procedures performed in our patients as incidental procedures may be associated themselves with higher risk, although these procedures are, in fact, included in surgical risk categories that are equal to or less than that of cholecystectomy.7 

The occurrence of specific sickle cell events (vaso-occlusive pain episodes and ACS) correlated with the number of previous hospitalizations and the absence of preoperative transfusion. The number of previous hospitalizations presumably reflects severity of clinical disease. The higher incidence of these events in patients who were not transfused preoperatively (group 3) compared with those who were transfused (groups 1, 2, and 4) suggests that there may be a role for preoperative transfusion in preventing sickle cell events in patients undergoing cholecystectomy.13,19,21 However, because the number of patients undergoing cholecystectomy without preoperative transfusion was small, and because these patients were not randomly assigned to nontransfusion, our findings may not be generalizable.

While there is clinical and laboratory evidence to support the beneficial effects of dilution of sickle cells on outcome,22-25 it is not certain if there is a threshold for such effects or if the effects are proportionate to the percentage of hemoglobin S. In vitro studies indicate that the interrelationship between total hemoglobin concentration and percent of hemoglobin S with blood viscosity and flow is complex and not strictly linear.21,26-28 In this study, vaso-occlusive pain episodes were highest in the nontransfusion group (mean preoperative hemoglobin S 88%) and lowest in the randomized aggressive transfusion group (hemoglobin S 34%), a finding that may suggest there is a threshold percentage of hemoglobin S at which vaso-occlusive events are more likely to occur.

The subgroup of 153 patients undergoing laparoscopic cholecystectomy offers comparisons with previous reports of this procedure in general pediatric and adult populations.29-36 The conversion rate from laparoscopic to open cholecystectomy in the general population (2% to 12%) ^29,30,32,34 is similar to that of our patients (5%), while the complication rate for the general population undergoing laparoscopic cholecystectomy (4% to 5%) is considerably lower than in our population. Surgical time reported in the general population for the laparoscopic procedure (1 to 2 hours)29-32 is shorter than that reported here (mean, 3.2 hours), although our time reflects total anesthesia time, not surgical time alone, and includes time for incidental procedures in 16% of cases. Hospitalization time for the general population is also shorter (1 to 2 days)29,31,32 than for our population (mean, 6.4 days). All of these differences would seem to be attributable largely to the underlying disease of our patients and the complex perioperative care required by them.

Surprisingly, anesthesia time was shorter in the open cholecystectomy group than in the laparoscopic group, despite a larger proportion of the cases in the open group having simultaneous incidental procedures (33% v 16%). Operative time for laparoscopic procedures has been reported to be shorter than for open procedures in the general population.29,34 The relatively long time for laparoscopic procedures in this study may reflect an institutional “learning curve” since the procedure was introduced during the course of the study, although we were unable to demonstrate this at individual participating institutions.

Perioperative outcome was similar for the open and laparoscopic cholecystectomy groups despite a higher average ASA physical status score and more frequent baseline liver dysfunction in patients in the open group. It is possible that patients in the open cholecystectomy group benefitted from more frequent use of such postoperative interventions as intravenous antibiotics, supplemental oxygen, and incentive spirometry.37 The finding that morbidity was the same in the open and laparoscopic groups in our sickle cell population (unlike what is reported for the general population29,34 ) may well reflect the importance of the underlying disease, rather than the procedure, in determining perioperative outcome in our population.

Hospitalization time was longer for the open cholecystectomy group than the laparoscopic group (mean, 9.8 days v 6.4 days), and surgical technique was found to be an independent predictor of length of hospitalization. Hospitalization time was also directly correlated with patient ASA physical status score and baseline chest roentgenogram abnormality, factors which together may reflect medical conditions that would be expected to affect length of hospital stay. That laparoscopic cholecystectomy is associated with shortened hospitalization without increased morbidity is an important consideration in the current setting of health care priorities.

In conclusion, we have found in 364 patients with SCA undergoing cholecystectomy that total perioperative morbidity is 39%. While total morbidity is not affected by preoperative transfusion, the incidence of specific sickle cell events is higher in those patients who were not transfused preoperatively than in those who were. Laparoscopic cholecystectomy was accompanied by shorter hospitalization time (6.4 days) than was open cholecystectomy (9.8 days), although perioperative outcome was similar between the two groups. Therefore, based on findings in this and our previous report, we recommend conservative preoperative transfusion and use of the laparoscopic technique, where available, for patients with SCA undergoing cholecystectomy.

We are indebted to all of the hematologists, anesthesiologists, surgeons, nurses, and data coordinators at the study centers for their participation and to Jolene Edwards for editorial assistance.

Participating Investigators: R. Nagel, Albert Einstein College of Medicine (Bronx, NY); R. Johnson, Alta Bates Hospital (Berkeley, CA); D. Sears, Baylor College of Medicine (Houston, TX); M. Wong, Bronx-Lebanon Hospital (Bronx, NY); J. Parke, Carolinas Medical Center (Charlotte, NC); G. Bray, Children's Hospital National Medical Center (Washington, DC); D. Hurst, L. Koblenz, and E. Vichinsky, Children's Hospital Oakland (Oakland, CA); P. DeAlarcon and M. Grossi, Children's Hospital of Buffalo (Buffalo, NY); K. Ohene-Frempong, Children's Hospital of Philadelphia (Philadelphia, PA); P. Waldron and M. Walters, Children's Hospital, Seattle (Seattle, WA); C. Russo, S. Murphy, and H. Smith, Children's Memorial Hospital (Chicago, IL); G. Woods, Children's Mercy Hospital (Kansas City, MO); L. Guarini, A. Hurlet, and S. Piomelli, Columbia-Presbyterian Medical Center (New York, NY); T. Kinney, and G. Phillips, Duke University Medical Center (Durham, NC); C. Daeschner, and T. Holbrook, East Carolina University (Greenville, NC); I. Buchanan, and J. Eckman, Emory University (Atlanta, GA); S. Claster, Highland Hospital (Oakland, CA); H. Hume, Hospital Ste-Justine (Montreal, Quebec, Canada); N. Olivieri, Hospital for Sick Children (Toronto, Ontario, Canada); R. Bellevue, Interfaith Medical Center (Brooklyn, NY); W. Wang, Le Bonheur Children's Medical Center (Memphis, TN); P. Milner, Medical College of Georgia (Augusta, GA); M. Abboud, and S. Jackson, Medical University of South Carolina (Charleston, SC); L. Benjamin, M. Bestak, and E. Radel, Montefiore Hospital (Bronx, NY); S. Baruchel and D. Esseltine, Montreal Children's Hospital (Montreal, Quebec, Canada) F. Blei, New York University Medical Center (New York, NY); S. Embury, and W. Mentzer, San Francisco General Hospital (San Francisco, CA); P. Gillette, S. Miller, and S. Rao, State University of New York (Brooklyn, NY); R. Grover, G. Ramirez, and D. Wethers, St Luke's/Roosevelt Hospital (New York, NY); B. Glader and C. Russo, Stanford University Children's Hospital (Stanford, CA); W. Jennings, Truman Medical Center (Kansas City, MO); D. Rucknagel and K. Kalinyak, University of Cincinnati (Cincinnatti, OH); P. Lane, University of Colorado (Denver, CO); M. Koshy, and N. Talischy, University of Illinois (Chicago, IL); C. Pegelow, University of Miami (Miami, FL); R. Iyer, and M. Steinberg, University of Mississippi (Jackson, MS); H. Cooper, and E. Orringer, University of North Carolina (Chapel Hill, NC); V. Mankad and Y. Yang, University of South Alabama (Mobile, AL); C. Johnson, and D. Powars, USC Medical Center (Los Angeles, CA).

Policy Board Members: J.A. Smith, Harlem Hospital Center, (New York, NY); K. Bailey, Mayo Clinic (Minneapolis, MN); D. Cohen, Children's Hospital of Philadelphia (Philadelphia, PA); R. Strauss, University of Iowa Hospital (Iowa City, IA); A. Schwartz, St Louis Children's Hospital (St Louis, MO); C. Reid, National Heart, Lung, and Blood Institute, National Institutes of Health (Rockville, MD); Myron Waclawiw, National Heart, Lung, and Blood Institute, National Institutes of Health (Rockville, MD); D. Bonds, Sickle Cell Disease Branch, National Institutes of Health (Rockville, MD); M. DeBaun, Washington Unviersity School of Medicine (St Louis, MO).