Secondary Logo

Journal Logo

Obstetric Anesthesiology: Brief Report

Neuraxial Anesthesia in Parturients with Thrombocytopenia

A Multisite Retrospective Cohort Study

Goodier, Christopher G. MD*; Lu, Jeffrey T. MD; Hebbar, Latha MD, FRCA; Segal, B. Scott MD, MHCM§; Goetzl, Laura MD, MPH

Author Information
doi: 10.1213/ANE.0000000000000882
  • Free
  • SDC
  • CE Test


Anesthesiologists are faced with a clinical dilemma when thrombocytopenic parturients request neuraxial analgesia. A platelet count of <150,000/mm3 is common, complicating 7% to 10% of pregnancies.1,2 Thrombocytopenia is considered a relative contraindication to neuraxial analgesia because of the unknown risk of hematoma.3 However, there is no general agreement on the platelet count threshold for withholding neuraxial anesthesia because the potential risk of epidural hematoma may outweigh the benefits of neuraxial anesthesia. The decision not to initiate neuraxial anesthesia results in suboptimally treated pain, a situation that is seldom acceptable in other areas of medicine. Finally, the decision to withhold neuraxial anesthesia in the setting of thrombocytopenia increases the likelihood of general anesthesia, which can result in considerable maternal morbidity. Combining the previously published case series (n = 326; Table 1), the upper limit of the 95% confidence interval (CI) of the risk of spinal-epidural hematoma is 1.7%.2,4–9 Given this uncertainty, it is not surprising to find wide practice variations.10,11

Table 1:
Total Number of Thrombocytopenic Obstetric Patients Undergoing Regional Analgesia

The goal of the current study was to increase the precision of the risk estimate for spinal-epidural hematoma after neuraxial anesthesia and analgesia in an obstetric population with thrombocytopenia (platelet count <100,000/mm3) and to provide risk estimates for complications in women with thrombocytopenia who received general, rather than neuraxial, anesthesia for cesarean delivery.


A multisite, retrospective review of obstetric medical records was performed. Records from January 1, 1997, to December 31, 2007, and from January 1, 2004, to December 31, 2007, were reviewed at the Medical University of South Carolina (MUSC) and Brigham and Women’s Hospital (BWH), respectively. IRB approval was obtained at both institutions (17297 MUSC, 2008P-001953 BWH), and both boards waived a requirement for written informed consent. At each institution, a complete blood count, which includes a platelet count, is obtained at the time of admission to the labor and delivery suite. Platelet counts are repeated as necessary in the intrapartum period based on the clinical situation.

The inclusion criterion was all patients admitted for delivery with a platelet count <100,000/mm3. There were no exclusion criteria. The MUSC Perinatal Information Network database was cross-referenced with the laboratory information system to identify all parturients admitted for delivery with a platelet count <100,000/mm3 before consideration for neuraxial anesthesia. At BWH, the Research Patient Data Registry database was searched for all patients who delivered during the specified time frame and then cross-referenced with platelet counts in the laboratory information system obtained before delivery.

Key clinical variables were extracted from the electronic and hard copy medical records using the study data collection form. Variables extracted included maternal age, gestational age, race, parity, platelet count before initiation of anesthesia, mode of anesthesia, type of cesarean delivery (elective, nonelective urgent, nonelective not urgent, and stat), anesthesia complications (e.g., postdural puncture headache, failed neuraxial, repeat tracheal intubation, hemoptysis, and aspiration), and platelet transfusion. In addition, the clinical diagnosis accounting for the low platelet count was recorded, with a special focus on the presence of preeclampsia as this disease has been linked with unstable platelet counts. The risk of general anesthesia was assessed by each anesthesiologist at the time of admission and was based on hematologic parameters and the presence of thrombocytopenia. A thorough airway examination using a combination of Mallampati score, thyromental distance, adequacy of mouth opening, and neck mobility was used to classify patients into low or high risk for anesthesia-related complications. The epidural catheter used at MUSC was the Flextip catheter (Arrow International, Reading, PA) and at BWH was the Perifix epidural catheter (Braun Medical, Bethlehem, PA).

The combined database was analyzed using SPSS statistical software (SPSS Inc., Chicago, IL). Clinical variables were compared between the 2 clinical sites and between parturients receiving neuraxial anesthesia and those who did not. Nonparametric data were compared using the Wilcoxon rank-sum test, and rates were compared using the χ2 test. The upper limit of the 95% CIs for the risk of epidural-spinal hematoma was calculated using the formula

, sometimes known as the “rule of 3,” where R is the upper limit of the 95% CI of the outcome risk (“how high could the risk be?”) and N is the number of subjects without an observed event.12 This statistical approach was chosen to calculate the upper 95% confidence limit of risk for neuraxial hematoma directly from the number of patients in the database at or below a given platelet count.

Multivariable modeling was performed to examine the odds ratio of general anesthesia for cesarean delivery in thrombocytopenic patients between sites after controlling for underlying patient differences that were significant on univariate analysis or had face validity. Candidate variables included maternal race and age, gestational age, platelet transfusion, birth weight, diagnosis of preeclampsia, emergency nature of the cesarean delivery, platelet count, and institution. A backward conditional binary logistic model was constructed, and final model parameters are reported. Graphically, using sextiles, the logit of the general anesthetic proportional rate was approximately linearly related to the platelet count, supporting the use of logistic regression.


Over the study period, there were approximately 52,000 deliveries in the 2 institutions. Two hundred eighty patients were identified that met the criterion for platelet count <100,000/mm3 (range 28,000–99,000/mm3; MUSC 190, BWH 90). Of these, 36.4% (n = 102) received epidural or combined spinal-epidural analgesia/anesthesia, and 25.4% (n = 71) patients received spinal anesthesia. Underlying diagnoses for thrombocytopenia included preeclampsia (61.4%), gestational thrombocytopenia (14.6%), idiopathic thrombocytopenic purpura (10.0%), and others (14.0%). Ninety-three percent (n = 261) of patients in our cohort had serial platelet count assessments and 24.9% (n = 65), predominantly those with preeclampsia, had a ≥10% decrease in platelet count over 24 hours. One postdural puncture headache and no cases of spinal-epidural hematoma were observed. We combined our 173 patients with the 326 total cases from previous small to moderate studies reported in the literature2,4–9 for a final sample of 499. This allowed us to revise the overall upper 95% confidence limit for the probability of spinal-epidural hematoma following neuraxial anesthesia to 0.6%.

Specific platelet count data were available for 254 of 499 patients from the previous studies2,4–9 (personal communicationa) and our current analysis. These data were used to construct a conservative estimate of the upper limit of risk by specific platelet count for patients receiving spinal analgesia (Fig. 1) or epidural analgesia (Fig. 2) based on zero observed events and the number of patients at each platelet count. It is clear from both figures that the upper estimate of risk declines with increasing platelet count because of the higher number of reported procedures at higher platelet counts (Supplemental Digital Content,

Figure 1:
Estimated risk of neuraxial hematoma with spinal anesthesia. Data are from previously published studies2 , 4–9 and the current study. The left y-axis (red) represents the upper 95% confidence limit of the risk of neuraxial hematoma using values from patients at or below a given platelet count, calculated using the “rule of 3.”12 The right y-axis (blue) represents the cumulative number of subjects with platelet counts at or below the given platelet count value.
Figure 2:
Estimated risk of neuraxial hematoma with epidural (including combined spinal-epidural) anesthesia. Data are from previously published studies2 , 4–9 and the current study. The left y-axis (red) represents the upper 95% confidence limit of the risk of neuraxial hematoma using values at or below a given platelet count, calculated using the “Rule of 3.”12 The right y-axis (blue) represents the cumulative number of subjects with platelet counts at or below the given platelet count value.
Table 2:
Logistic Regression Model for Predicting General Anesthesia in Parturients with Thrombocytopenia

Overall, cesarean delivery was performed in 190 patients: 41% (n = 77) received general anesthesia, 36% (n = 70) received spinal anesthesia, and 23% (n = 43) received epidural analgesia. One patient with neuraxial labor analgesia received general anesthesia for an urgent cesarean delivery. In patients laboring without neuraxial analgesia (n = 73), the rate of cesarean delivery was 55.8%; the rate of general anesthesia in the women requiring cesarean delivery was 50.0%. Patients receiving general anesthesia experienced 1 case of aspiration pneumonia, 3 cases of prolonged tracheal intubation (inability to extubate the trachea before leaving the operating room), and 1 case of hemoptysis for an overall aggregate morbidity rate of 6.5% (95% CI, 2.1%–14.5%). Median platelet counts were significantly higher in women receiving neuraxial anesthesia (86,000/mm3 [interquartile range 51,000–99,000/mm3]) compared with women receiving general anesthesia (60,000/mm3 [interquartile range 9,000–99,000/mm3], P < 0.001). Three patients received a platelet transfusion before placement of spinal anesthesia for cesarean delivery. There was a significant practice difference between the 2 tertiary care centers in the probability of providing general anesthesia after adjusting for risk mix, including platelet count and the urgency of the procedure (P < 0.001; Table 2).


Anesthesiologists are faced with a clinical dilemma when a thrombocytopenic parturient requires labor analgesia or surgical anesthesia. The risk of epidural hematoma after epidural analgesia in pregnant patients has been estimated at 0.2 to 3.7 per 100,00013 and is unknown after spinal anesthesia. Loo et al.14 assessed neurological complications of obstetric neuraxial anesthesia by performing a review of the English literature between 1966 and 1998. Seven cases of hematoma after epidural analgesia in obstetric patients were identified; in only 3 of these cases was a potential risk factor for hematoma identified. One patient had a low-grade ependymoma, a second patient had a prolonged partial thromboplastin time, and the third patient had cholestasis of pregnancy, although no coagulation studies were performed. None of the patients had documented thrombocytopenia.

The unknown risk of hematoma has been used to justify the decision to avoid neuraxial anesthesia in thrombocytopenic parturients. It is logical that the risk of hematoma increases with falling platelet counts. Fewer data points on patients who have safely received neuraxial analgesia at very low platelet counts increase the uncertainty around safety. Combining our patients with previously published series, we were able to narrow the risk estimate for spinal-epidural hematoma in obstetric patients who receive neuraxial procedures to 0% to 0.6%. However, a more detailed analysis suggests that there are insufficient data, even with this large data set, to confidently offer epidural analgesia to patients with platelet counts <75,000 to 80,000/mm3.

The risks of neuraxial anesthesia must be weighed against the risks of failing to provide neuraxial anesthesia, especially if needed emergently. A prospective trial would be difficult and unethical to perform. In the past, this risk benefit analysis was exceedingly difficult to perform because of the lack of accurate estimates of either risk. We provide a comprehensive report of both the likelihood of requiring general anesthesia and the associated adverse outcomes. Although complications such as aspiration pneumonia and prolonged tracheal intubation are less serious than epidural hematoma, mortality in obstetric patients receiving general anesthesia is estimated at 6.5/million.15 Furthermore, risks of general anesthesia are likely to increase over time, given the rising rates of obesity and other risk factors in the US obstetric population.16,17 Finally, our work suggests variations in clinical practice with regard to the use of general versus neuraxial anesthesia in parturients with thrombocytopenia. Large variations in care are problematic for patients, and rates of general anesthesia for cesarean delivery may be a proxy measure for quality of care and patient safety.

There are some limitations to our study. Given the retrospective design, it was impossible to determine why some thrombocytopenic patients received neuraxial anesthesia and some did not. Although we were able to control for some known differences between the 2 sites that may have influenced the observed differences in rates of general anesthesia, we did not collect data on maternal body mass index (BMI). However, given that the prevalence of obesity (BMI > 30 kg/m2) is significantly higher in South Carolina (31.6%; 95% CI, 30.4%–32.8%) than in Massachusetts (22.9%; 95% CI, 22.02%–23.8%), it is unlikely that lower BMI accounted for the increased rates of general anesthesia observed at MUSC.18 In addition, the incidence of thrombocytopenia varied between institutions, likely secondary to ascertainment methods. Despite these limitations, our data represent additional evidence for estimating the risks and benefits of neuraxial anesthesia in thrombocytopenic patients. This information may be useful to obstetric anesthesia providers and aid in minimizing practice variations.

Figure 3:
Estimate of 95% upper limit of risk of spinal-epidural hematoma by number of neuraxial blocks without complication, calculated using the “Rule of 3.”12

Overall, the results of our study suggest that even multicenter retrospective data are insufficient to guide practice regarding the safety of neuraxial analgesia in thrombocytopenic parturients. Five hundred to 1000 procedures performed without complications at each platelet count strata need to be reported to reduce the upper limit of theoretical risk to an acceptable level (Fig. 3). Given the complications we observed after general anesthesia, these data are critically needed. We call for a national registry of procedures performed in thrombocytopenic patients with a data repository managed by a national organization such as the Society for Obstetric Anesthesia and Perinatology and yearly updates of the graphs of risk estimates as data accumulate.


Name: Christopher G. Goodier, MD.

Contribution: This author helped design the study, conduct the study, analyze the data, and prepare the manuscript.

Attestation: Christopher G. Goodier approved the final manuscript.

Name: Jeffrey T. Lu, MD.

Contribution: This author helped design and conduct the study.

Attestation: Jeffrey T. Lu approved the final manuscript.

Name: Latha Hebbar, MD, FRCA.

Contribution: This author helped design the study.

Attestation: Latha Hebbar approved the final manuscript.

Name: B. Scott Segal, MD, MHCM.

Contribution: This author helped design the study.

Attestation: B. Scott Segal approved the final manuscript.

Name: Laura Goetzl, MD, MPH.

Contribution: This author helped design the study, analyze the data, and prepare the manuscript.

Attestation: Laura Goetzl approved the final manuscript.

This manuscript was handled by: Cynthia A. Wong, MD.


a Personal communication Dr. Jose Carvalho, July 22, 2014, on behalf of Tanaka et al.7 Data abstracted from report by Rasmus et al.5 and Rolbin et al.6
Cited Here


1. Sainio S, Kekomäki R, Riikonen S, Teramo K. Maternal thrombocytopenia at term: a population-based study. Acta Obstet Gynecol Scand. 2000;79:744–9
2. Shalev O, Anteby E. Epidural anesthesia can safely be performed in gestational thrombocytopenia of >50,000/ml. Blood. 1996;88:62b
3. Wlody DSuresh MS, Segal BS, Preston RL, Fernando R, Mason CL. Neurologic complications of regional anesthesia in obstetrics. In: Shnider and Levinson’s Anesthesia for Obstetrics. 20135th ed Philadelphia, PA Wolters Kluwer–Lippincott Williams & Wilkins:412–24
4. Beilin Y, Zahn J, Comerford M. Safe epidural analgesia in thirty parturients with platelet counts between 69,000 and 98,000 mm(-3). Anesth Analg. 1997;85:385–8
5. Rasmus KT, Rottman RL, Kotelko DM, Wright WC, Stone JJ, Rosenblatt RM. Unrecognized thrombocytopenia and regional anesthesia in parturients: a retrospective review. Obstet Gynecol. 1989;73:943–6
6. Rolbin SH, Abbott D, Musclow E, Papsin F, Lie LM, Freedman J. Epidural anesthesia in pregnant patients with low platelet counts. Obstet Gynecol. 1988;71:918–20
7. Tanaka M, Balki M, McLeod A, Carvalho JC. Regional anesthesia and non-preeclamptic thrombocytopenia: time to re-think the safe platelet count. Rev Bras Anestesiol. 2009;59:142–53
8. Webert KE, Mittal R, Sigouin C, Heddle NM, Kelton JG. A retrospective 11-year analysis of obstetric patients with idiopathic thrombocytopenic purpura. Blood. 2003;102:4306–11
9. Frenk V, Camann W, Shankar KB. Regional anesthesia in parturients with low platelet counts. Can J Anaesth. 2005;52:114
10. Beilin Y, Bodian CA, Haddad EM, Leibowitz AB. Practice patterns of anesthesiologists regarding situations in obstetric anesthesia where clinical management is controversial. Anesth Analg. 1996;83:735–41
11. van Veen JJ, Nokes TJ, Makris M. The risk of spinal haematoma following neuraxial anaesthesia or lumbar puncture in thrombocytopenic individuals. Br J Haematol. 2010;148:15–25
12. Hanley JA, Lippman-Hand A. If nothing goes wrong, is everything all right? Interpreting zero numerators. JAMA. 1983;249:1743–5
13. Kam PC, Thompson SA, Liew AC. Thrombocytopenia in the parturient. Anaesthesia. 2004;59:255–64
14. Loo CC, Dahlgren G, Irestedt L. Neurological complications in obstetric regional anaesthesia. Int J Obstet Anesth. 2000;9:99–124
15. Hawkins JL, Chang J, Palmer SK, Gibbs CP, Callaghan WM. Anesthesia-related maternal mortality in the United States: 1979-2002. Obstet Gynecol. 2011;117:69–74
16. Lu GC, Rouse DJ, DuBard M, Cliver S, Kimberlin D, Hauth JC. The effect of the increasing prevalence of maternal obesity on perinatal morbidity. Am J Obstet Gynecol. 2001;185:845–9
17. Soens MA, Birnbach DJ, Ranasinghe JS, van Zundert A. Obstetric anesthesia for the obese and morbidly obese patient: an ounce of prevention is worth more than a pound of treatment. Acta Anaesthesiol Scand. 2008;52:6–19
18. Centers for Disease Control. . Vital signs: state-specific prevalence of obesity among adults—United States, 2009. MMWR. 2010;59:1–5

Supplemental Digital Content

© 2015 International Anesthesia Research Society