Introduction

Anesthesia-related adverse events in obstetric populations in developed countries are rare (~1/1,000 anesthetic interventions).1 Nevertheless, anesthesia-related morbidity has been recognized as an appropriate and useful indicator for obstetric care quality and could serve as a patient safety indicator for childbirth.2,3 Studies on the adverse events of obstetric anesthesia and analgesia are limited to specific complications of neuraxial anesthesia and have been presented as case reports and case series analyses.4,5,6,7,8,9,10,11,12,13,14,15 To ensure the safety of anesthetic practice, it is important to evaluate a multitude of complications and determine their risk in larger cohorts.

The Society for Obstetric Anesthesia and Perinatology created the Serious Complication Repository (SCORE) project to establish the incidence of serious adverse events related to obstetric anesthesia and to identify associated risk factors. High neuraxial block, respiratory arrest in labour and delivery, and unrecognized spinal catheterization were the most frequent complications encountered. A serious adverse event was reported in approximately 1:3,000 (95% confidence interval [CI], 1:2,443 to 1:3,782) obstetric anesthetics.16 A study from New York found that despite their low incidence, anesthesia-related adverse events in obstetrics remain a cause of concern, particularly in parturients undergoing Cesarean delivery, living in rural areas, or having pre-existing medical conditions.17

Analysis of a cohort of Canadian parturients hospitalized for delivery from 2002 to 2015 identified that anesthesia-related adverse events were likely responsible for precipitating cardiac arrest in 12.6% of pregnant people (adjusted OR, 16.53; 95% CI, 10.65 to 25.66).18 Data from the UK Obstetric Surveillance System showed that about 28% of pregnant people who received basic life support had cardiac arrest as a direct complication of obstetric anesthesia.19 Nevertheless, the survival rate following this cause was 100% in both studies. A study by Joseph et al.20 estimated the rate of maternal morbidity in Canada due to anesthesia-related adverse events to be 13.5 (95% CI, 11.6 to 15.6) per 100,000 deliveries; however, the details, types, and the estimation of risk factors for their occurrence were not studied. Anesthesia was claimed responsible for 1% of direct maternal deaths in the UK.21 Anesthesia-related adverse events accounted for 0.4% of all pregnancy-related deaths in 2014–2017.22

There is no dedicated obstetric anesthesia registry; therefore, Canadian data on such adverse events are unknown. This poses difficulties when discussing the risks of anesthesia during informed consent, as well as while dealing with medicolegal issues in obstetric patients. Furthermore, knowing such risks will help us better prepare to prevent and manage them more effectively. The objective of this study was to determine the frequency, types, temporal trend, and associated risk factors of anesthesia-related adverse events during hospitalization for delivery in Canada.

Methods

This retrospective population-based cohort study was completed after approval by the University of Calgary Ethics Review Board and Canadian Institute for Health Information (CIHI). The study used hospitalization data from the CIHI Discharge Abstract Database relating to obstetric deliveries in Canada for 13 fiscal years from 1 April 2004 to 31 March 2017. Data from all jurisdictions in Canada (except the province of Quebec) were included in the study. The records include approximately 75% of all obstetric deliveries during the study period and are representative of births in Canada.23

Canadian Institute for Health Information data, prepared under strict confidentiality guidelines, contain information extracted from the medical records of parturients admitted for delivery. The study population consisted of all pregnant people with gestational age of 20 weeks or more hospitalized for childbirth, excluding early pregnancies with abortive outcome ectopic and molar pregnancies. Diagnoses were coded according to the enhanced Canadian version of the tenth revision of the International Statistical Classification of Diseases (ICD-10-CA) and Related Health Problems for diagnoses.24 Procedures were coded according to the Canadian Classification of Health Interventions.25Anesthesia-related adverse events and medical and obstetrical conditions were defined based on diagnostic codes.24 Specific anesthesia-related events investigated as defined per CIHI included aspiration pneumonitis; other pulmonary, cardiac, and central nervous system complications of anesthesia; toxic reaction to local anesthesia; spinal and epidural anesthesia-induced headache (postdural puncture headache [PDPH]); failed or difficult intubation; other complications of anesthesia; unspecified complications of anesthesia; malignant hyperthermia due to anesthesia; mechanical complications of epidural and subdural infusion catheter; and intraspinal abscess (Electronic Supplementary Material, eAppendix). Data on patient demographics, medical conditions, obstetrical complications during labour and delivery, and potential etiologies of anesthesia-related adverse events were collected. The primary outcome was the population-based incidence of anesthesia-related adverse events in pregnant people hospitalized for childbirth in Canada. The secondary outcomes included the association of anesthesia (general or neuraxial) complications with medical and obstetric conditions.

Statistical analysis

Hospitalized parturients receiving an anesthetic for delivery were identified by the administration of at least one of a general, spinal, epidural, combined general and epidural, combined spinal and epidural, and/or nerve block anesthetic and/or monitored/unmonitored care. For analysis purposes, spinal, epidural, spinal/epidural combination, and nerve block anesthesia were grouped into a broad heading of neuraxial anesthesia. Population demographics are presented as mean (standard deviation), median [interquartile range], or frequency (percentage). Anesthesia-related adverse events and annual administrations of anesthesia modalities for the population are presented using incidence and intervention rates per 100,000 total anesthesia interventions with their 95% CIs. Multiple types of anesthesia may have been administered during a single hospitalization. In such circumstances, pairing of complications with type of anesthesia given was not possible. For the presentation of anesthesia event rates specific to general, neuraxial, and general plus neuraxial anesthesia, included hospitalizations were restricted to those that received only general anesthesia, neuraxial anesthesia, or combined general and neuraxial anesthesia. Such restriction permitted the presentation of anesthesia-related adverse event incidence rates specific to these modalities of anesthesia. Serious adverse events were defined as failed or difficult intubation, aspiration pneumonitis, pulmonary, cardiac and central nervous system complications due to anesthesia, toxic reaction to local anesthesia, malignant hyperthermia, and intraspinal abscess. Trends in annual adverse event rates, anesthesia interventions, and delivery method were assessed using univariate linear regression and a conservative approach towards the calculation of standard errors using a robust sandwich estimator. Data are presented as the average annual increase/decrease in interventions and adverse event rates between 2004 and 2013 with their 95% CIs.

Associations between hospitalizations with an anesthesia-related adverse event and patient characteristics, delivery method, and modality of anesthesia were assessed using multivariate logistic regression with adjustment for demographics (parturient age, gestational age, parity, and admission status), delivery type (Cesarean only or vaginal only), anesthesia modality (general anesthesia only, neuraxial anesthesia only, or general plus neuraxial combination only). Associations between anesthesia-related adverse events and obstetric and pre-existing medical conditions were investigated using multivariate logistic regression after adjustment for equivalent covariates as listed prior. The specific obstetric and pre-existing medical conditions used in the regression analysis to assess associations with anesthesia-related events were selected upon discussion and agreement between two study authors with specialty training and core focus in obstetrical anesthesia. Hospitalizations included in the analysis were restricted to those that received only one modality of anesthesia intervention (general anesthesia only, neuraxial anesthesia only, or general plus neuraxial combination only) and were associated with a single type of delivery (Cesarean only or vaginal only). Data for regression analyses are presented as frequency (percentage) and unadjusted and adjusted OR (95% CI). Statistical analysis was completed using SPSS 25.0 (IBM Corp., Armonk, NY, USA) and R Studio version 1.2.5019 with R statistical software version 3.6.1 (The R Project for Statistical Computing, Boston, MA, USA).

Results

A total of 3,194,875 anesthesia interventions occurred in 2,601,034 hospitalizations for childbirth during 2004–2017. We identified 8,361 events during 8,130 hospitalizations complicated by anesthesia-related adverse events over the 13-year period, i.e., one per 382 anesthetic interventions (0.3%; average annual incidence, 262 events per 100,000 interventions; 95% CI, 256 to 267) and one per 311 hospitalizations (0.3%). The annual incidence of anesthesia-related adverse events reduced significantly from a high of 327 (95% CI, 303 to 352) to 222 (95% CI, 204 to 240) per 100,000 interventions (average annual decrease of 7 events per 100,000 interventions; 95% CI, -10 to -4; P < 0.001) (Fig. 1). Two deaths were recorded in our cohort, potentially due to amniotic and pulmonary embolism (25 per 100,000 hospitalizations; 95% CI, 5 to 79).

Fig. 1.
figure 1

Annual incidence of anesthesia-related adverse events per 100,000 anesthetic interventions from fiscal years 2004/2005 to 2016/2017. Shaded areas represent 95% confidence intervals around line of best fit.

Full size image

Neuraxial (n = 2,847,749) and general (n = 91,800) anesthesia accounted for 89% and 3% of all anesthesia interventions, respectively (Table 1). Annual total anesthesia interventions showed an increasing trend with time varying from 209,956 per year to 266,139 per year (average annual increase of 3,248 interventions; 95% CI, 1,307 to 5,189; P = 0.004) (Fig. 2a). The annual rate of general anesthesia administration showed a significant decreasing trend with time (average annual decrease of 146 administrations; 95% CI, -188 to -104; P < 0.001) (Fig. 2b), while those of neuraxial (average annual decrease of 160 administrations; 95% CI, -438 to 119; P = 0.23) and general plus neuraxial (average annual increase of 6 administrations; 95% CI, -11 to 23; P = 0.45) administrations showed no significant change (Fig. 2b). Vaginal (n = 1,591,633) and Cesarean (n = 1,021,870) births accounted for 61% and 39% of deliveries, respectively. Average annual Cesarean deliveries decreased by 217 (95% CI, -377 to -57; P = 0.01) while average annual vaginal deliveries increased by 209 (95% CI, 48 to 369; P = 0.02) between 2004 and 2017.

Table 1 Demographic characteristics, type of delivery, and method of anesthesia interventions in hospitalizations requiring at least one anesthetic administration
Full size table
Fig. 2
figure 2

Annual number of total anesthesia interventions (A) and type of anesthesia per 100,000 total anesthesia interventions (B) stratified by general anesthesia, neuraxial anesthesia, and general/neuraxial anesthesia between the fiscal years 2004/2005 and 2016/2017. Shaded areas represent 95% confidence intervals around line of best fit.

Full size image

Anesthesia modality-specific complication rates

General anesthesia

Three hundred and forty anesthesia-related adverse events were associated with 62,954 hospitalizations that received only general anesthesia for an adverse event incidence rate of 488 (95% CI, 438 to 542) per 100,000 general anesthesia administrations (Table 2). Annual adverse event incidence ranged from 374 (95% CI, 244 to 552) to 607 (95% CI, 402 to 881) per 100,000 interventions, showing no significant trend over time (P = 0.65). A total of 135 serious adverse events (40%) were reported with the annual incidence varying from 118 (95% CI, 56 to 222) to 364 (95% CI, 213 to 585) per 100,000 interventions (P = 0.13) (Fig. 3a). Of the 135 serious adverse events, 88 (65%) were classified as a failed or difficult intubation.

Table 2 Frequency and event rate of anesthesia-related adverse events in hospitalizations restricted to the interventions of either general, neuraxial or general/neuraxial anesthesia
Full size table
Fig. 3
figure 3

Annual incidence of anesthesia-related serious adverse events (A) and spinal and epidural induced headache (B) per 100,000 interventions stratified by general anesthesia, neuraxial anesthesia, and general/neuraxial anesthesia between the fiscal years 2004/2005 and 2016/2017.

Full size image

Neuraxial anesthesia

Five thousand nine hundred and ninety-seven adverse events were associated with 2,212,090 hospitalizations (2,669,453 administrations) that received only neuraxial anesthesia giving an event incidence rate of 225 (95% CI, 219 to 230) per 100,000 neuraxial interventions (Table 2). Annual event incidence rates showed a significant decreasing trend ranging from 286 (95% CI, 262 to 312) to 185 (95% CI, 166 to 204) per 100,000 interventions (P < 0.001). A total of 385 serious adverse events (6%) were reported with the annual incidence rates varying from 10 (95% CI, 6 to 15) to 23 (95% CI, 16 to 30) per 100,000 interventions (P = 0.37) (Fig. 3a). Postdural puncture headache accounted for 86% (95% CI, 85 to 87) of anesthesia-related adverse events. Annual rates of PDPH showed a significant decreasing trend from 236 (95% CI, 215 to 259) to 159 (95% CI, 142 to 177) per 100,000 interventions (P < 0.001) (Fig. 3b).

General plus neuraxial anesthesia

Two hundred and five anesthesia-related adverse events were associated with 12,965 hospitalizations in parturients receiving a general plus neuraxial combination (13,887 interventions) (Table 2), giving an overall incident rate of 1,476 (95% CI, 1,284 to 1,689) per 100,000 general plus neuraxial anesthesia interventions (Table 2). Annual adverse event incidence rates displayed fluctuations from 2,923 (95% CI, 1,921 to 4,277) to 797 (95% CI, 435 to 1,351) per 100,000 interventions presenting no significant trend (P = 0.35). A total of 62 (30%) serious adverse events were reported with an annual rate varying from 187 (95% CI, 37 to 599) to 654 (95% CI, 308 to 1,233) per 100,000 interventions (P = 0.14) (Fig. 3a). Postdural puncture headache represented 50% (95% CI, 43 to 57) of anesthesia-related events. The overall annual PDPH rate was 735 (95% CI, 602 to 888) ranging from 2,192 (95% CI, 1,346 to 3,391) to 342 (95% CI, 95 to 913) presenting no significant trend over time (P = 0.31) (Fig. 3b).

Frequency of anesthesia-related adverse events regardless of the type of anesthesia is summarized in Table 3.

Table 3 Frequency of anesthesia-related adverse events in hospitalizations that received an anesthesia intervention
Full size table

Risk factors

The associations of demographic, obstetrical data, delivery type, and anesthetic modality with an anesthesia-related adverse event are presented in Table 4. Gestational age between 29 and 36 weeks, one or more comorbidities, Cesarean delivery, and use of general or general plus neuraxial anesthesia were significantly associated with an adverse event. Medical and obstetric conditions associated with an anesthesia-related adverse event included eclampsia, preeclampsia, cardiomyopathy, obesity, and asthma (Table 5).

Table 4 Associations between anesthesia-related adverse events and demographic characteristics, type of delivery and anesthesia intervention
Full size table
Table 5 Associations between anesthesia-related adverse events and maternal and obstetric conditions
Full size table

Discussion

Our analysis shows that, over a period of 13 years, one in 382 anesthetic interventions and one in 311 hospitalizations for delivery were complicated by anesthesia-related adverse events, with a decreasing trend over time. Postdural puncture headache and failed or difficult intubation were the most common adverse events related to neuraxial and general anesthesia, with an event rate of 1:462 and 1:15,895, respectively. No deaths were directly attributed to anesthesia-related adverse events.

Despite a changing population, characteristics including increasing rate of obesity among parturients, advanced age,26,27 and a higher incidence of complications during labour and delivery (such as postpartum hemorrhage and thromboembolism),28 it is reassuring to observe that the overall rate of anesthesia-related adverse events is declining over time. This trend is likely due to various changes resulting in improved patient safety such as universal use of dilute local anesthetic solutions during labour and the increased use of neuraxial anesthesia during labour and for Cesarean delivery (Fig. 4).29

The incidence of anesthesia-related adverse events in our study (0.3%) was slightly lower than reported in other studies using administrative databases (0.3% to 0.5%). 3,16,17 Nevertheless, it was more than seven-fold compared with reports directly collected from participating institutions in the SCORE project.16 This might be related to discrepancies in the duration of the data collection, smaller sample size, and heavy predominance of academic centres in their study.

These complication rates are two-fold and seven-fold higher per 100,000 interventions with general and general plus neuraxial anesthesia, respectively compared with neuraxial anesthesia alone. Serious adverse events constituted 9% of all adverse events. Specifically, 40% and 30% of adverse events in parturients administered only general or only general/neuraxial anesthesia, respectively, were considered serious. The incidence of failed or difficult intubation in our cohort was 1:792 general anesthetics and 1:604 in combined general and neuraxial anesthesia. This incidence remained unchanged over the study period, although there was a reduction in the rate of general anesthesia administration for Cesarean deliveries (Fig. 2). Our finding is in agreement with the incidence of failed intubation reported from the UK and US ranging from 1:300 to 1:533 of general anesthetics for 2000–2014.16,21,22,30,31 A recent retrospective audit from the UK found no cases of failed intubation in 3,430 general anesthetics for Cesarean deliveries, which is attributed to the higher than average rate of obstetric general anesthetics administered at their hospital resulting in greater experience of the technique.32 In a large, multicentre, contemporary study from the US examining over 14,000 general anesthetics for Cesarean delivery from 2000 to 2018, the reported risk of difficult intubation was 1:55 and the reported risk of failed intubation 1:1,250.33 Because the risk of failed intubation remains stable and modern airway equipment is so effective, it is recommended to have equipment for managing the difficult airway readily available in the labour suite and obstetric operating rooms.34

Postdural puncture headache remained the most common anesthesia-related adverse event and complicated 1 in 521 neuraxial anesthetics. Nevertheless, the rate of PDPH significantly decreased over the study period. The incidence was higher in the SCORE project at 1:144 neuraxial anesthetics16 and in the US cohort from 2010–2016 at 1:323 deliveries.35

The rate of subdural hematoma was 1.5 per 100,000 deliveries in the study by Moore et al.35 In another study from the US (1998–2010), the incidence of spinal hematoma was estimated at 0.6 per 100,000 (95% CI, 0.3 to 1.0) epidural catheterizations in the obstetric population, and the incidence of epidural abscess was zero.36 The incidence of serious complications such as spinal hematoma, abscess, purulent meningitis after neuraxial anesthesia in a Swedish cohort was estimated to be 1:20,000 to 1:30,000.37 In our study, we identified six cases of intraspinal, epidural, extradural, and subdural abscesses corresponding to a rate of 0.19 per 100,000 interventions (95% CI, 0.08 to 0.39). Surprisingly, 115 cases of spinal headaches were identified in the subgroup of patients that received general anesthesia. This finding is most likely due to a coding error that resulted in diagnosing any postpartum headache as a postdural puncture headache.

The rate of cardiac, pulmonary, and central nervous system adverse events related to anesthesia was low. The observed pulmonary adverse events could be attributed to exacerbation of pre-existing respiratory issues, opioid use, fluid overload, or failed or difficult intubation.38 The low incidence of aspiration in our cohort is consistent with the report on maternal mortality in Michigan, where no deaths were associated with aspiration.39 No cases of aspiration related to general anesthesia during Cesarean delivery were reported in the SCORE project.16 Due to the nature of ICD-10-CA codes, we were unable to identify the incidence of specific serious cardiorespiratory conditions resulting from administration of anesthesia such as cardiac arrest, major cardiac event, and pulmonary embolism. The Cardiac Arrest in Pregnancy Study (CAPS) from the UK estimated that nearly a quarter of cases of cardiac arrest in their cohort were precipitated by neuraxial anesthesia-related complications, although their survival rate was reported as 100%.19

We identified several risk factors such as cardiomyopathy, obesity, eclampsia, and asthma associated with anesthesia-related adverse events (Table 5). Pregnancy is poorly tolerated in parturients with cardiomyopathy, and the risk of mortality is approximately 7% if the patient is classified as a New York Heart Association functional class III or IV.40,41 Results of a registry of the European Society of Cardiology indicate that pregnant people with structural and ischemic heart disease, including cardiomyopathy, had a higher rate of Cesarean delivery, almost half of those being emergency Cesarean deliveries.42 Obesity and preeclampsia/eclampsia contributed to about 12% of anesthesia-related adverse events. Obese parturients are at increased risk of develo** gestational hypertension, diabetes, and thromboembolic events and have a higher incidence of operative or Cesarean delivery. They are also at a higher risk for difficult airway management, difficult neuraxial techniques, and anesthesia-related mortality compared with non-obese parturients.43,44 In developed countries, eclampsia ranges from one in 2,000 to one in 3,448 pregnancies and is responsible for considerable morbidity and mortality.45,46,47 Severe complications in these patients include antepartum hemorrhage due to placental abruption, cerebrovascular events, organ failure, and disseminated intravascular coagulation, which can further complicate anesthesia management.48

Based on the review of existing literature, this is the first Canadian population-based cohort study on the epidemiology of anesthesia-related adverse events in pregnant people hospitalized for delivery. Compared with other similar studies, we utilized the largest database for hospitalizations for delivery and captured all important complications related to anesthesia. The limitations of this study are similar to any analysis using an administrative data set. The hospitalization records do not directly identify an underlying cause of anesthesia-related adverse events. The inability to determine sequential relationships between delivery methods and anesthesia-related complications makes it impossible to recommend the safest mode of delivery in parturients with known risk factors. There is also a possibility of unrecorded interventions in the database, such as combined administration of general and neuraxial anesthesia, and misdiagnosing postpartum headaches as PDPH. These two factors in combination could account for diagnosis of PDPH in patients receiving general anesthesia. Finally, we were not able to examine anesthesia-related adverse events during hospitalization for delivery in Quebec because data from this province are not available through the CIHI Discharge Abstract Database for administrative reasons. The quality of information in the hospitalization data has been investigated previously.23,49 The number of data errors in terms of illogical or out-of-range values was found to be low, and the estimated prevalence of parturients’ comorbid conditions was similar to that reported in the literature. The hospitalization data have been successfully used in previous studies of maternal morbidity and infant health outcomes,50,51 which provides reassurance regarding the validity of our findings.

An interesting observation was made about feasibility of calculating the frequency of anesthesia-related adverse events and frequency of these events for individual practitioners in a given hospital using an administrative database in combination with hospital-specific intervention rates and staffing patterns. Future research is needed to validate the model. Although it is not possible to create evidence-based practice advisories from the study findings, the results can be used to guide discussions involving informed consent and to make practical recommendations. This information could also serve as an audit to evaluate and assess the inherent risks associated with providing anesthetic care during pregnancy through the puerperium. Obtaining information on many more adverse events is a worthwhile patient safety goal. Next steps in this process should include the creation of a national obstetric anesthesia database and/or a serious complication registry.