General Anesthesia for Cesarean Delivery for Thrombocytopenia in Hypertensive Disorders of Pregnancy: Findings From the Obstetric Airway Management Registry

Abstract

BACKGROUND: 

In resource-limited environments, spinal anesthesia (SA) is preferred for cesarean delivery. In women at risk of spinal epidural hematoma, particularly those with hypertensive disorders of pregnancy, thrombocytopenia should be excluded before neuraxial blockade. In the context of emergency surgery for fetal distress, this investigation may be hampered by laboratory services being unavailable or off-site.

METHODS: 

The Obstetric Airway Management Registry (ObAMR) is currently active across all anesthesia training institutions affiliated with the University of Cape Town. This multicenter observational study aimed to estimate the proportion of patients receiving general anesthesia (GA) for either confirmed or suspected thrombocytopenia, which was not excluded due to unavailability of laboratory results. To establish the number of GA uses that may have been avoided if platelet counts were available, we retrospectively searched for subsequent platelet counts in patients for whom thrombocytopenia was suspected. An algorithm was proposed, including a simple decision aid for estimating risk versus benefit of SA versus GA, to be followed in the setting of hypertensive disorders of pregnancy and thrombocytopenia.

RESULTS: 

Thrombocytopenia was the indication for GA in 100 of 591 patients (16.9%) captured in the registry. In total, 48 of 591 (8.1%) had confirmed thrombocytopenia, and 52 of 591 (8.8%) had suspected thrombocytopenia. Of these patients, 91 of 100 had a hypertensive disorder of pregnancy. In the confirmed thrombocytopenia group, the indication for GA was a platelet count <75 × 10⁹/L. In the suspected thrombocytopenia group, 46 of 52 (88.5%) platelet counts could be retrospectively traced. The median (interquartile range) platelet count was 178 × 10⁹/L (93 – 233 × 10⁹/L), and platelets exceeded 75 × 10⁹/L in 41 of 46 patients (89.1%). In the 5 of 46 patients with retrospectively confirmed thrombocytopenia, 2 had hemolysis, elevated liver enzymes, and low platelets (HELLP) syndrome, 2 had antepartum hemorrhage with preeclampsia, and 1 had isolated thrombocytopenia with preeclampsia.

CONCLUSIONS: 

In 17% of patients, the indication for GA was thrombocytopenia. Of these, 52 of 100, or nearly 9% of the total of 591, received GA because a platelet count was unavailable at the time of surgery. The importance of early laboratory assessment, when available, should be emphasized. Overall, 41 of 591 (6.9%) had a platelet count >75 × 10⁹/L and would not have needed GA if their platelet count had been known. After following the constructed algorithm and applying the decision aid to assess risk and benefit, there may be circumstances in which the clinician justifiably opts for SA when a platelet count is indicated but unavailable.

KEY POINTS

• Question: What proportion of patients received general anesthesia either for confirmed or suspected thrombocytopenia, the latter of whom had not been excluded due to unavailability of a platelet count?
• Finding: In 100 patients (17% of 591), the indication for general anesthesia was thrombocytopenia; 52 received general anesthesia due to the unavailability of a platelet count, and 41 of 46 of these patients were subsequently shown to have had a platelet count permitting spinal anesthesia.
• Meaning: Using a constructed algorithm including a decision aid to assess risk–benefit of SA versus GA in obstetrics, there may be circumstances in which the clinician opts for SA when a platelet count is indicated but unavailable.

African mothers who give birth via cesarean delivery (CS) are at least 50× more likely to die than mothers in high-income countries.¹ The 2 most significant independent risk factors for maternal mortality in Africa are hemorrhage and complications of anesthesia.¹ General anesthesia (GA) is associated with increased risk of a difficult airway (including upper airway edema) or failed tracheal intubation, pulmonary aspiration, awareness, and greater blood loss than regional anesthesia (RA).² In patients with a hypertensive disorder of pregnancy (defined as gestational hypertension, preeclampsia, or eclampsia), there is more rapid desaturation during apnea after induction of anesthesia and during tracheal intubation.³ There is also a higher risk of neurological complications due to hypertension during tracheal intubation and more neonatal respiratory depression.⁴ Therefore, in the absence of epidural anesthesia for labor, the method of choice for CS is spinal anesthesia (SA), particularly in resource-limited settings.

A rare but potentially catastrophic complication of SA is the development of spinal epidural hematoma.⁵ Therefore, it is appropriate to assess the maternal platelet count before performing SA in high-risk individuals who may have thrombocytopenia, such as women with preeclampsia with severe features. In the context of emergency surgery for fetal distress and in limited-resource environments such as much of South Africa, this investigation may be hampered by laboratory services being unavailable or off-site.

Data from a multicenter Obstetric Airway Management Registry (ObAMR) established by the Department of Anesthesia and Perioperative Medicine of the University of Cape Town (UCT) have shown that approximately 20% of obstetric GA is performed either on the basis of confirmed maternal thrombocytopenia, or in patients for whom coagulopathy was suspected but a platelet count was unavailable despite being clinically indicated.⁶ The primary aim of this prospective descriptive observational study was to establish the proportion of patients for whom GA was performed on the basis of known thrombocytopenia or because it was suspected but not excluded. As a secondary aim, when possible, perioperative platelet counts were subsequently retrospectively obtained from the laboratory records of the latter group to establish the number and proportion of GA use that might have been avoided if a platelet count had been known at the time of surgery. We further propose an algorithm, including a simple decision aid for estimating risk versus benefit of SA versus GA, which is recommended for every patient with gestational hypertension, preeclampsia, or eclampsia, incorporating the presence or absence of a platelet count.

METHODS

In 2018, the ObAMR was initiated at all sites providing obstetric anesthesia under supervision of the Department of Anesthesia and Perioperative Medicine of UCT, including 1 tertiary and 2 secondary level institutions. Approval for the registry was granted by the human research ethics committee (HREC) of the UCT faculty of health sciences (R025/2018). Written consent was waived by the HREC, and verbal consent was required from each patient to be included in the registry. Verbal consent was ensured by the data capturers indicated in the electronic data capture form. The electronic data capture form on a secure Research Electronic Data Capture (REDCap) server (https://www.project-redcap.org/) is completed during or immediately after each case by specialist and trainee anesthetists in the obstetric theaters using an electronic link sent to their respective smartphones, or access was gained by scanning a Quick Response code, which is present in all obstetric theaters. Inclusion criteria are any elective or emergency obstetric surgery with GA performed from 20 weeks of gestation until 7 days postpartum. After verbal consent is attained, all data are anonymized.

This was a prospective multicenter descriptive observational study using data from the ongoing ObAMR. The reporting in this article adheres to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. Ethics approval for the conduct of this study was granted by the UCT HREC (807/2019). In addition, a substudy was approved in which the platelet count was retrospectively obtained from laboratory records in those patients for whom the investigation and/or result was not available before surgery. We analyzed data collected from November 29, 2019, to May 31, 2021. In our institution, GA is performed for CS if the platelet count is <75 × 10⁹/L within 12 hours of the procedure or if the investigation is clinically indicated and no platelet count is available at the time of surgery. Patients with further indications for GA related to coagulation status (abnormal international normalized ratio [INR] or thromboelastography tracing or the administration of heparin or other anticoagulants within the time frame recommended by the American Society of Regional Anesthesia guidelines) were documented in the registry but excluded from this analysis.

We traced platelet counts retrospectively via the National Health Laboratory Service (NHLS) website using patient folder numbers obtained from the operating theater records. As the ObAMR is anonymized, we matched registry entries with theater records using a combination of the following: surgery under GA, date of surgery, location of surgery, patient age, gestational age, gravidity, and status with respect to hypertensive disorders of pregnancy. If an entry could not be reliably matched based on these data or blood was not drawn for a platelet count on the day of, before, or after the CS, it was excluded.

Statistical Analysis

We extracted data from REDCap that had been collected during the study period to an Excel spreadsheet (Microsoft). The primary outcome was the proportion of the total number of obstetric operations with GA performed for suspected or confirmed thrombocytopenia. Patient characteristics were reported as mean (standard deviation [SD]) for continuous, normally distributed variables, median (interquartile range [IQR]) for data not normally distributed, and number (percentage) for categorical variables.

The proportion of patients receiving GA for suspected or confirmed thrombocytopenia in this population is unknown. For the purposes of sample size calculation, we hypothesized that approximately 5% received GA. For a 5% margin of error (95% confidence interval [CI], 0–10), this would require a sample size of 80 cases (G*Power. Version 3.1.9.6. Germany: Universität Kiel; 2019). However, as there was uncertainty as to the true incidence, we included 100 records in the study.

RESULTS

Data from 591 obstetric operations with GA had been entered into the ObAMR during an 18-month period. Demographic and obstetric data are shown in Table 1. Of parturients who received GA, 33.8% (n = 200) had developed a hypertensive disorder of pregnancy: 27% (n = 159) had preeclampsia, 3.9% (n = 23) had eclampsia, and 3% (n = 18) had gestational hypertension.

Table 1. - Demographic and Obstetric Data

Characteristics	n	Mean (SD) or median (IQR)
Age (y)	591	29.6 (6.6)
Weight (kg)	581	81.0 (22.3)
Gestational age (wk)	543	36.1 (4.1)
Gravidity	586	2 (1–5)
Parity	582	1 (0–2)

Abbreviations: IQR, interquartile range; SD, standard deviation.

Indications for GA are shown in Table 2. For 100 of 591 patients (16.9%; 95% CI, 14.0–20.2), the indications were confirmed or suspected thrombocytopenia (Figure 1). Thrombocytopenia was confirmed in 48 of 591 (8.1%; 95% CI, 6.2–10.7) and suspected/not excluded in 52 of 591 patients (8.8%; 95% CI, 6.5–11.1). Two of the 591 patients had received anticoagulant therapy. Most obstetric operations with GA were performed by anesthesiology residents in training with 1 to 5 years of anesthesia experience (<1 year [45, 7.6%]; 1–5 years [325, 55.0%]; >5 years [221, 37.4%]).

Table 2. - Indications for GA (n = 590)

Indication	n (proportion %)
Failed spinal anesthesia	127 (21.5)
Severe fetal distress/bradycardia	102 (17.3)
Thrombocytopenia (confirmed or suspected, but not excluded)	100 (16.9)
Postpartum hemorrhage	72 (12.2)
Abnormal placentation	69 (11.7)
Decreased level of consciousness	15 (2.5)
Retained products of conception	26 (4.4)
Prolonged surgery	25 (4.2)
Maternal cardiac disease	19 (3.2)
Other	142 (24.1)

Abbreviation: GA, general anesthesia.

Figure 1.: Flow diagram of patients who received GA for thrombocytopenia. GA indicates general anesthesia; HELLP, hemolysis, elevated liver enzymes, and low platelets; hpt, hypertension.

Overall, 91 of 100 patients had a hypertensive disorder of pregnancy. Of these, 78 (76.5%) had preeclampsia, 8 (7.8%) had eclampsia, and 5 (4.9%) had gestational hypertension. Of the 100 patients who received GA, there were 94 entries in response to questions on the case report form about hemolysis, elevated liver enzymes, low platelet count (HELLP) syndrome; 41 of 94 patients (43.6%) had suspected or confirmed HELLP syndrome, for 34 of 94 (36.2%), the diagnosis was uncertain, and for 19 (20.2%), HELLP syndrome was not present.

Of the 9 patients recorded as not having a hypertensive disorder of pregnancy, 4 had confirmed thrombocytopenia, and in 5, it was suspected but not excluded. Seven of these patients had additional indications for GA, including hemorrhage (3), pulmonary edema (1), von Willebrand disease (1), and isolated thrombocytopenia of unknown cause (2). On review of medical records, 2 patients with possible HELLP syndrome were incorrectly categorized as not having a hypertensive disorder.

In the group of patients with suspected but unconfirmed thrombocytopenia, platelet counts could be traced retrospectively in 46 of 52. These were taken either on the date of surgery (35 of 46 cases) or within 1 day before or after surgery (11 of 46). Of these, 41 of 46 (89.1%) had platelet counts >75 × 10⁹/L and received GA. The median (IQR) platelet count for these patients was 178 (93–233) × 10⁹/L. Therefore, of all patients receiving GA, 41 of 591 (6.9%; 95% CI, 5.0–9.3) would not have needed GA if their platelet count had been known.

There were 5 patients with counts <75 × 10⁹/L. Two had HELLP syndrome, 2 had antepartum hemorrhage and preeclampsia, and 1 had preeclampsia with isolated thrombocytopenia, as confirmed in the clinical notes. No patients in the sample of 591 had clinical signs of spinal epidural hematoma during the study period.

DISCUSSION

In this study, data extracted from the ObAMR showed that the indication for GA in our sample of the obstetrics population was thrombocytopenia in 100 of 591 patients (16.9%). Decisions were based on laboratory data in only approximately 50%. Patients who received GA with possible thrombocytopenia but no laboratory confirmation comprised 52 (8.8%) of all GAs performed during an 18-month period. We subsequently confirmed that platelet counts were <75 × 10⁹/L in only 5 of 46 (10.9%) of these patients at the time of surgery. Thus, 41 of 46 patients were exposed to the risk of GA because no recent platelet count was available at the time of surgery. A simple algorithm for guidance in decision-making with respect to the performance of SA versus GA in the setting of hypertensive disorders of pregnancy and thrombocytopenia is proposed in Figure 2. The included decision aid allows for contextualization of the risk–benefit of SA versus GA.

Figure 2.: Proposed decision algorithm for SA versus GA in the setting of hypertensive disorders of pregnancy and thrombocytopenia. GA indicates general anesthesia; SA, spinal anesthesia; HELLP, hemolysis, elevated liver enzymes, and low platelets.

The safe lower limit of the platelet count with respect to the risk of spinal epidural hematoma in preeclampsia or eclampsia is controversial. A recent systematic review and Delphi process performed by a Society of Obstetric Anesthesia and Perinatology (SOAP) task force concluded that the risk is low in patients with thrombocytopenia with a platelet count >70 × 10⁹/L associated with hypertensive disorders other than HELLP syndrome, immune thrombocytopenic purpura, thrombotic thrombocytopenic purpura, and unknown etiology.⁷ SOAP also stipulates that in some circumstances, RA may be undertaken at a count of 50–70 × 10⁹/L, but not <50 × 10⁹/L.⁷

The most recent National Institute for Health and Care Excellence guidelines do not recommend a specific platelet count below which RA is contraindicated, but they suggest that the decision should be individualized.⁸ The timing of the platelet count before surgery is controversial. In the case of HELLP syndrome, SOAP recommends a platelet count that is taken a maximum of 6 hours before the procedure due to potentially rapid changes.⁷ When formal laboratory results are unavailable, point-of-care testing, such as the use of thromboelastography and thromboelastometry before RA, has been investigated in the context of pregnancy. Two articles using thromboelastography in preeclampsia suggested hypocoagulability at platelet counts <54 × 10⁹/L (95% CI, 40–75 × 10⁹/L) and 75 × 10⁹/L, respectively.^9,10

The method of assessment of the risk of spinal epidural hematoma is unclear in the setting of a high prevalence of hypertensive disorders of pregnancy and urgent CS. A history of bleeding may serve as a guide, but this has not been validated in pregnancy or in patients with thrombocytopenia.¹¹ Nonetheless, a bleeding history and examination for clinical signs, such as bleeding intravenous sites and mucocutaneous bleeding, is still advocated.⁷

In a retrospective analysis over a 10-year period involving 1,260,000 spinal anesthetics and 450,000 epidural anesthetics, the incidence of spinal epidural hematoma was reported as 1:200,000 in the obstetric population. In this study, epidural hematoma occurred in parturients with HELLP syndrome rather than in patients with preeclampsia without severe features other than hypertension.¹² In a recent investigation aiming to establish the risk in the setting of thrombocytopenia in obstetrics, a Multicenter Perioperative Outcomes Group data base identified 573 parturients with a platelet count <100,000 × 10⁹/L who received neuraxial anesthesia for CS. This number was expanded to 1524 when data from a systematic review were added. Although no cases of spinal epidural hematoma requiring surgical intervention were identified, the upper limits of the 95% CI of the risk were progressively higher in patients with lower platelet counts due to limited observations (3% for a platelet count of 50–69 × 10⁹/L, and 11% for 0–49 × 10⁹/L).¹³

A recent article describes the process of risk–benefit decision analysis in obstetric anesthesia.¹⁴ In the context of thrombocytopenia, SOAP has also provided a decision aid for when to proceed with a neuraxial procedure.⁷ Regarding the performance of SA or GA in patients with a hypertensive disorder of pregnancy without a platelet count, the decision depends upon many factors, including the experience of the anesthesia practitioner (62.4% of procedures with GA in this study were performed by junior anesthesia practitioners with <5 years of experience), maternal comorbidities, including an anticipated difficult airway and/or morbid obesity, and the likelihood of thrombocytopenia.

In a recent study of a series of 1015 South African parturients, which examined the incidence and predictors of thrombocytopenia before CS, 117 of 1015 had either preeclampsia (90) or eclampsia (27). Only 7 of the 117 had platelet counts <75 × 10⁹/L. Of these, 5 had eclampsia and 2 had preeclampsia. Six patients had HELLP syndrome, only 1 had preeclampsia with no severe features other than hypertension.¹⁵ Similarly, in the suspected thrombocytopenia group in our study, of the 5 of 46 patients with platelet counts <75 × 10⁹/L, only 1 had a hypertensive disorder of pregnancy that was not complicated by either antepartum hemorrhage or HELLP syndrome. The above findings suggest, together with the SOAP task force recommendations, that a platelet count <75 × 10⁹/L is uncommon in preeclampsia in the absence of complications such as HELLP syndrome or eclampsia.

Although every attempt should be made to obtain laboratory confirmation of the coagulation status soon after admission (to be repeated if necessary), all of the above considerations suggest that there are certain parturients with a hypertensive disorder of pregnancy for whom a decision to perform surgery with SA is justifiable in the absence of a platelet count.

The present study has several limitations. The decision to perform surgery using GA cannot be fully validated because full details of individual clinical cases were not available, and only the main indication for GA was recorded in the ObAMR. The overall rate of capture into the registry was dependent on the attending anesthesia provider and was approximately 80% in a previous validation study,⁶ therefore, at least 20% of procedures using GA may not have been recorded. The coronavirus disease 2019 pandemic affected the risk assessment for GA in terms of aerosolization and the use of anticoagulants in hypoxemic patients, which, in many instances, contraindicated SA. Therefore, severe acute respiratory syndrome coronavirus 2–positive obstetric patients who presented for GA were not included in the study.

In conclusion, confirmed or suspected thrombocytopenia was the indication for GA in 17% of patients in this review of our obstetrics registry. Approximately 50% of these patients received GA due to the unavailability of laboratory confirmation of platelet counts. Retrospective tracing of platelet counts showed that 6.9% of the sample would not have received GA if their platelet count had been known. The importance of early laboratory assessment should be emphasized. We propose an algorithm incorporating a simple decision aid considering the risk–benefit of SA versus GA, which is recommended for each patient. After such an analysis, there may be circumstances in which the clinician justifiably opts for SA when a platelet count is indicated but unavailable.

ACKNOWLEDGMENTS

The authors would like to thank their colleagues in the Department of Anesthesia and Perioperative Medicine of the University of Cape Town for collecting ObAMR data and our patients for consenting to participate and expand medical knowledge. They would also like to acknowledge the ongoing support toward clinical research made possible by the Western Cape Department of Health, Groote Schuur, Mowbray Maternity, and New Somerset Hospitals. Anthony Reed, MBChB, DA(SA), MMed (UCT), FRCA, from the Department of Anaesthesia and Perioperative Medicine, Faculty of Health Sciences, University of Cape Town, South Africa, coordinated the ObAMR at New Somerset Hospital. Bruce Biccard, MBChB, DA(SA), FCA, PhD, Department of Anaesthesia and Perioperative Medicine, Faculty of Health Sciences, University of Cape Town, South Africa, assisted with statistical analysis and proofread the manuscript.

DISCLOSURES

Name: Lisa M. Seymour, MBChB, DA(SA).

Contribution: This author provided substantial contributions toward the conception or design of the work, data analysis and interpretation, drafting and revising the work critically for important intellectual content, and final approval of the version to be published.

Name: Nicole L. Fernandes, FCA(SA).

Contribution: This author provided substantial contributions toward the conception or design of the work; data analysis and interpretation; drafting and revising the work critically for important intellectual content; and final approval of the version to be published.

Name: Robert A. Dyer, FCA(SA), PhD.

Contribution: This author provided substantial contributions toward the conception or design of the work, data analysis and interpretation, drafting and revising the work critically for important intellectual content, and final approval of the version to be published.

Name: Maretha I. Smit, FCA(SA).

Contribution: This author provided contributions toward formation of the Obstetric Airway Management Registry (ObAMR), data analysis and interpretation for the work, revising the work critically for important intellectual content, and final approval of the version to be published.

Name: Dominique van Dyk, FCA(SA).

Contribution: This author coordinated the ObAMR at Mowbray Maternity Hospital and Groote Schuur Hospital, and provided substantial contributions toward data interpretation, formation of the figures in the work, drafting and revising the work critically for important intellectual content, and final approval of the version to be published.

Name: Ross Hofmeyr, FCA(SA), FAWM.

Contribution: This author provided contributions toward formation of the ObAMR, conception or design of the work, amendments to the data-capturing forms, revising the work critically for important intellectual content, and final approval of the version to be published.

This manuscript was handled by: Angela Enright, MB, FRCPC