Cesarean delivery (CD) is associated with significantly increased risks of anesthesia-related adverse events (ARAEs) and nonanesthetic perioperative morbidity compared with vaginal delivery. Temporal trends in these adverse outcomes remain unknown despite efforts to improve maternal safety. This study examines temporal trends in ARAEs and nonanesthetic perioperative complications in CDs in New York hospitals.
Data are from the State Inpatient Database for New York, 2003–2012. ARAEs, including minor and major ARAEs, and nonanesthetic perioperative complications were identified through International Classification of Diseases, Ninth Revision, Clinical Modification codes. Statistical significance in time trends was assessed using the Cochran–Armitage test and multivariable logistic regression.
Of the 785,854 CDs studied, 5,715 (730 per 100,000; 95% CI, 710 to 750) had at least one ARAE and 7,040 had at least one perioperative complication (890 per 100,000; 95% CI, 870 to 920). The overall annual rate of ARAEs decreased from 890 per 100,000 in 2003 to 660 in 2012 (25% decrease; 95% CI, 16 to 34; P < 0.0001). The rate of minor ARAEs decreased 23% (95% CI, 13 to 32) and of major ARAEs decreased 43% (95% CI, 23 to 63). No decrease was observed in the rate of ARAEs for CDs performed under general anesthesia. The rate of nonanesthetic complications increased 47% (95% CI, 31 to 63; P < 0.0001).
Anesthesia-related outcomes in cesarean deliveries appear to have improved significantly across hospitals in New York in the past decade. Perioperative nonanesthetic complications remain a serious healthcare issue.
In a review of approximately 800,000 cesarean deliveries in New York state from 2003 to 2012, the overall rate of anesthesia-related adverse events was 730 per 100,000, and there was a decrease in major events by 43% and minor events by 23% during this time. During the same time period, the incidence of nonanesthesia complications increased 47%.
Supplemental Digital Content is available in the text.
Cesarean delivery is associated with more anesthesia-related adverse events and perioperative morbidity than vaginal delivery, but temporal trends in the former have not been recently examined
In a review of approximately 800,000 cesarean deliveries in New York state from 2003 to 2012, the overall rate of anesthesia-related adverse events was 730 per 100,000, and there was a decrease in major events by 43% and minor events by 23% during this time
During the same time period, the incidence of nonanesthetic complications increased 47%
DURING the past decade, extensive efforts across clinical specialties have been deployed to improve perioperative patient safety and outcomes.1,2 In the area of obstetric anesthesia, evidence-based practice guidelines and reviews of mortality and morbidity, including closed malpractice claims, have been used to foster continual improvement in maternal safety and perioperative outcomes.3–6 However, the effectiveness of these efforts in improving maternal outcomes has not been evaluated. Cesarean delivery is the most commonly performed inpatient surgical procedure in the United States, totaling 1.3 million in 2010, with a regular increase in cesarean delivery rate through 2009.7,8 Compared with vaginal delivery, cesarean delivery is associated with a significantly increased risk of anesthesia-related adverse events (ARAEs), nonanesthetic perioperative morbidity, and mortality.9–11
Obstetric ARAEs are usually preventable because they are often caused by substandard care.5,6,12,13 The risk of ARAEs during cesarean delivery is especially high when the procedure is unplanned, when it is performed under general anesthesia, and in women with preexisting comorbidities.9,14,15 To minimize the risk of ARAEs, the American Society of Anesthesiologists recommends neuraxial anesthetic techniques for cesarean delivery, whenever possible.4 However, the rising prevalence of maternal comorbidities in the context of the increasing rate of cesarean deliveries, particularly in unplanned cases, poses specific challenges to anesthesiologists caring for such patients.16–19 The goal of this study was to examine the temporal trends in ARAEs, nonanesthetic perioperative complications, and overall mortality in cesarean deliveries in New York state between 2003 and 2012.
Materials and Methods
The study protocol was reviewed by the institutional review board of Columbia University Medical Center. It was granted exemption under 45 Code of Federal Regulation 46 (not human subjects research). The study adhered to the STROBE statement.20
The study sample consisted of all women undergoing cesarean delivery in the state of New York between January 1, 2003 and December 31, 2012. Hospital discharge record data for these women collected in the deidentified New York State Inpatient Database (SID) were analyzed. SIDs are part of the Healthcare Cost and Utilization Project sponsored by the Agency for Healthcare Research and Quality. SID databases have captured all inpatient discharge records from nonfederal acute care community hospitals in participating states since 1988. Nonfederal community hospitals account for 85% of U.S. hospitals. For each record, the SID includes patients’ demographic, economic and outcome characteristics, one hospital identifier, and up to 15 procedural and 25 diagnostic codes defined in the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM).
Records with neonatal and/or maternal diagnoses and procedures identified with the NEOMAT code provided by the SID database were first selected. Records indicating labor and delivery were identified with a combination of ICD-9-CM diagnosis and procedure codes, as described by Kuklina et al.21 Finally, the subset of cesarean deliveries was selected. Codes of the ICD-9-CM used in this study are available in Supplemental Digital Content 1, http://links.lww.com/ALN/B196.
Planned cesarean delivery was defined as a cesarean delivery record with no ICD-9-CM code indicating labor and no ICD-9-CM code indicating one of the following maternal or fetal condition requiring rapid delivery: abnormal fetal heart rate tracing, placental abruption, uterine rupture, eclampsia, or umbilical cord prolapse. The proportion of cesarean deliveries among delivery-related discharges was calculated for each year of the study.
Patient and Hospital Variables
The following variables were recorded directly from the SID data: maternal age and type of anesthesia. Anesthesia type is reported in the SID data as a categorical variable with values corresponding to general, regional, other, local, none, and missing. Each record contains a maximum of one value for anesthesia type. For the purpose of the study, the variable was categorized as general anesthesia, nongeneral anesthesia, and missing. Charlson and Bateman comorbidity indexes were calculated with algorithms based on ICD-9-CM codes and weights as previously described.22–24 The Bateman index for use in obstetric patients was chosen because it includes, in contrast to the Charlson index, pregnancy-related risk factors for adverse maternal outcomes such as severe preeclampsia and eclampsia or multiple gestation. The proportions of general anesthetics, women older than 40 yr, Charlson comorbidity index greater than 0, and Bateman comorbidity index greater than 1 were calculated for each year of the study.
Whenever the hospital identifier was available in the SID data for each year of the 10-yr study period, the following three characteristics were recorded: annual number of cesarean deliveries, residency training accreditation by the Accreditation Council for Graduate Medical Education, and location (urban vs. rural) defined by the Core-Based Statistical Area in the American Hospital Association file. Hospitals were further classified for each year into three categories based on the annual number of cesarean deliveries performed. Hospitals were defined as low- and high-volume hospitals if the annual number of cesarean deliveries was lower than the 25th and greater than the 75th percentiles of the considered year, respectively.
ARAEs and Temporal Trends
Anesthesia-related adverse events were identified with a combination of ICD-9-CM codes as previously described by Cheesman et al.9 Based on ICD-9-CM codes, adverse events were categorized into the following seven groups: pulmonary complications, cardiac complications, central nervous system complications, adverse events related to anesthetic drugs, other and unspecified systemic adverse events, systemic adverse events without specification of the location, and adverse events related to neuraxial anesthesia and local anesthetics. In addition, adverse events were further categorized into major and minor adverse events. Major adverse events were defined as adverse events with an associated risk of death or cardiac arrest greater than 1%. Cardiac arrest was identified with ICD-9-CM codes, and maternal death was identified directly from SID-recorded data.
A multivariable logistic regression model was fitted to assess the association between ARAEs and adverse maternal outcome. The model used death or cardiac arrest as the composite outcome of interest and ARAEs, patient characteristics, and delivery characteristics as predictor variables.
The overall crude rate of ARAEs was calculated as the ratio of the number of records indicating at least one adverse event to the total number of records. The temporal trends in this rate were analyzed according to the type of cesarean section (planned or unplanned), hospital characteristics (annual cesarean volume, rural or urban location, and presence of a residency program), and type of anesthesia (general or nongeneral). The crude rates of ARAEs for the seven types of adverse events previously defined, for major adverse events and for minor adverse events, were also calculated.
Nonanesthetic Perioperative Complications and Mortality
Nonanesthetic perioperative complications included the following diagnoses identified with the corresponding ICD-9-CM codes and adopted from the study by Callaghan et al.18 : acute myocardial infarction or ischemia, acute heart failure, acute respiratory failure, blood-clot pulmonary embolism or deep venous thrombosis, disseminated intravascular coagulation, acute renal failure, sepsis or septic shock, and stroke. The crude rate of nonanesthetic perioperative complications and in-hospital mortality was calculated as described in the section “ARAEs and Temporal Trends.” The temporal trends in the rates of nonanesthetic perioperative complications were also analyzed according to the type of cesarean delivery (planned or unplanned).
The statistical analysis was performed with R version 3.0.2 (R Foundation for Statistical Computing, Austria). Results are expressed as number (% or per 100,000).
To assess the association between ARAEs and the composite outcome of death or cardiac arrest, a logistic regression model was developed. First, comparisons of discharges with and without the composite outcome used chi-square tests or Fisher exact tests for discrete variables. Odds ratios were obtained with univariate logistic regression. Second, variables with a P value less than 0.2 in the univariate analysis or deemed clinically significant, along with the variable indicating the year of delivery, were forced as predictors into a logistic regression model. Discrimination of the model was assessed with the c-index and calibration with the Hosmer–Lemeshow test.
For each year of the study, proportions and crude rates were calculated. The 95% confidence interval (CI) was calculated with a binomial distribution. Missing data were observed for the following variables used to calculate proportions and rates: type of anesthesia (n = 48,030) and in-hospital death (n = 2). Discharges with missing data were excluded from the calculation of proportions or rates. The percent change in proportions (or rates) during the 10-yr study period was calculated as the difference between the proportion (or rate) in 2012 and the proportion (or rate) in 2003 divided by the proportion (or rate) in 2003. The 95% CI for the percent change was calculated.
For proportions, the Cochran–Armitage test for trends was used to assess the statistical significance of changes during the 10-yr study period. For rates, the statistical significance of changes over time was assessed with a multivariate logistic regression including the following three independent variables treated as continuous variables: year of delivery, Charlson comorbidity index, and Bateman comorbidity index. The strength of the linear relation between rates and proportions was measured with the Pearson correlation coefficient (r).
The selection of the study sample is presented in figure 1. During the study period, 785,854 discharges indicating cesarean delivery were recorded. Characteristics of the sample studied and the rate of adverse maternal outcomes are presented in table 1.
Cesarean Delivery Rate and Maternal Comorbidities
The cesarean delivery rate increased from 28.7% in 2003 to 34.7% in 2009 (19% increase; 95% CI, 18 to 20; P < 0.0001) and remained stable from 2010 to 2012. The increase was statistically significant for both planned (21%; 95% CI, 19 to 22; P < 0.0001) and unplanned (17%; 95% CI, 15 to 18; P < 0.0001) cesarean deliveries. There were significant increases in the proportion of women older than 40 yr (13%; 95% CI, 7 to 18; P < 0.0001), Charlson comorbidity index greater than 0 (84%; 95% CI, 75 to 92; P < 0.0001), and Bateman comorbidity index greater than 1 (9%; 95% CI, 8 to 11; P < 0.0001); these increases were observed in both planned and unplanned cesarean deliveries (data not shown).
ARAEs and Temporal Trends
A total of 11,093 adverse events were recorded among 5,715 discharges (730 per 100,000; 95% CI, 710 to 750). In discharges recording at least 1 adverse event, the median number of adverse events reported was 2 (interquartile range, 1 to 3). Distribution of the types of adverse events according to the seven groups defined with ICD-9-CM is presented in table 2. Four types of adverse events were major with the most common being pulmonary complications (2.9%), followed by cardiac complications (1.9%), central nervous system complications (0.5%), and complications related to anesthetic drugs (0.4%). Major adverse events accounted for 5.7% of all ARAEs. In the logistic regression model, the adjusted odds ratio of death or cardiac arrest associated with ARAEs was 17.9 (95% CI, 12.4 to 25.9) (table 3).
A statistically significant decrease in the overall rate of ARAEs was observed, from 890 per 100,000 (95% CI, 820 to 960) in 2003 to 660 per 100,000 (95% CI, 610 to 720) in 2012 (25% decrease; 95% CI, 16 to 34; P < 0.0001) (fig. 2A). The decrease was observed for the seven types of adverse events defined with ICD-9-CM except for systemic adverse events without specification of the location (no change) and adverse events related to anesthetic drugs (increase) (fig. 2B). The rate of major adverse events decreased from 111 per 100,000 in 2003 (95% CI, 88 to 139) to 63 per 100,000 in 2012 (95% CI, 47 to 83) (43% decrease; 95% CI, 23 to 63; P < 0.0001). The rate of minor adverse events decreased from 781 per 100,000 in 2003 (95% CI, 716 to 850) to 603 per 100,000 in 2012 (95% CI, 551 to 660) (23% decrease; 95% CI, 13 to 32; P < 0.0001).
The decrease in the overall rate of ARAEs was observed in both planned cesarean deliveries (34%; 95% CI, 23 to 45; P < 0.0001) and unplanned cesarean deliveries (16%; 95% CI, 2 to 29; P = 0.0003). It was observed across hospitals regardless of annual cesarean volume, teaching status, and rural or urban location (fig. 3).
Anesthesia data were available for 737,824 discharges (94%). The proportion of cesarean deliveries under general anesthesia decreased from 7.5% in 2003 to 6.0% in 2012 (21% decrease; 95% CI, 18 to 24; P < 0.0001). This decrease was observed in unplanned cesarean deliveries (from 9.1% in 2003 to 6.3% in 2012; 31% decrease; 95% CI, 28 to 34; P < 0.0001) but not in planned cesarean deliveries (5.7% in 2003 and 5.6% in 2012; 1% change; 95% CI, −8 to 5; P = 0.95). There was no decrease in the overall rate of ARAEs in cesarean deliveries performed under general anesthesia (1,070 per 100,000 in 2003 and 1,100 per 100,000 in 2012; 3% change; 95% CI, −36 to 42; P = 0.68), regardless of the severity of adverse events (data not shown). Conversely, a decrease was observed in the rate of adverse events in cesarean deliveries performed under nongeneral anesthesia, from 870 per 100,000 in 2003 to 650 per 100,000 in 2012 (25% decrease; 95% CI, 16 to 35; P < 0.0001).
Nonanesthetic Perioperative Complications and Mortality
The rate of nonanesthetic perioperative complications increased from 770 per 100,000 (95% CI, 700 to 830) in 2003 to 1,130 per 100,000 (95% CI, 1,050 to 1,200) in 2012 (47% increase; 95% CI, 31 to 63; P < 0.0001) (fig. 4). This increase was statistically significant for both planned cesarean deliveries (36% increase; 95% CI, 14 to 57; P < 0.0001) and unplanned cesarean deliveries (57% increase; 95% CI; 34 to 79; P < 0.0001). The increase was observed for all the components of the composite indicator except for myocardial ischemia or infarction (no change; P = 0.18), acute respiratory failure (no change; P = 0.10), stroke (43% decrease; 95% CI, 19 to 67; P = 0.0004), and deep venous thrombosis or pulmonary embolism (30% decrease; 95% CI, 13 to 47; P < 0.0001). A significant correlation was observed between the proportion of women with a Charlson comorbidity index greater than 0 and the perioperative complication rate (r = 0.98; 95% CI, 0.92 to 0.99).
The maternal mortality rate decreased from 34 per 100,000 (95% CI, 21 to 50) in 2003 to 16 per 100,000 (95% CI, 9 to 28) in 2012 (51% decrease; 95% CI, 18 to 84; P < 0.0001).
The major findings of this study from 2003 to 2012 in the state of New York are a marked decrease in major and minor ARAEs and in-hospital mortality after cesarean delivery and a continuous increase in nonanesthetic perioperative complications.
Recent studies examining temporal trends in anesthesia safety in obstetrics in the United States have focused on mortality.15,18 They report a decrease in the rate of anesthesia-related deaths. A similar decrease in major adverse events has been reported in Western Europe countries.6 The current study assesses both major and minor ARAEs, extends these findings to a more recent time period, and establishes that these events are strongly associated with cardiac arrest and death. Importantly, a global improvement was observed regardless of the hospital annual volume of cesarean deliveries, hospital teaching status, or hospital location.
Although unlikely to lead to harm or disability, minor adverse events may lead to mothers’ dissatisfaction; most women expect an uneventful birth.25 Claims for maternal minor injuries (e.g., headache, back pain, pain during surgery, or emotional distress) made up 28% of obstetric claims in the last Closed Claims Analysis of Liability Associated with Obstetric Anesthesia.5 In the current study, the number of minor adverse events was much higher (94.3%) than the number of major events (5.7%). Although there was a reduction in the overall rate of ARAEs (25%), the decrease was more pronounced for major adverse events (43%) than for minor adverse events (23%), suggesting further possible improvement for minor adverse events. For instance, the decrease in adverse events related to neuraxial anesthesia that made up 54% of the adverse events was only 16%. Ninety-six percent of adverse events related to neuraxial anesthesia were postdural puncture headaches. This highlights the importance of preventing, appropriately diagnosing, and managing this condition, which has possible long-term consequences.26,27
The 2007 American Society of Anesthesiologists Practice Guidelines for Obstetric Anesthesia stating “neuraxial techniques are preferred to (general anesthesia) for most cesarean deliveries” contrasted with the 1999 statement that “the decision to use a particular anesthetic technique should be individualized.”3,4 This evidence-based, guideline-supported change in anesthesia practice may explain the decrease in the proportion of general anesthetics for unplanned (but not planned) cesarean deliveries from 9.1 to 6.3% observed during the study period. Although we observed an overall decrease in ARAE, we did not observe a decrease in ARAEs, minor or major, for cesarean deliveries performed under general anesthesia. Unpreparedness and insufficient skills in rare and unplanned situations were suggested as probable causes of the heightened risk of cesarean delivery under general anesthesia in the reports of the Confidential Enquiry into Maternal Deaths in the United Kingdom.6,19 Two interventions, namely protocols and team training, have been suggested to improve skills and preparedness for rare and unplanned situations and subsequently resulting in improved outcomes. To date, however, these interventions have not been widely implemented or deployed.
In a 2010 Sentinel Event Alert, the Joint Commission recommended that protocol-driven patient care be tailored to each institution providing obstetric care.28 In 2013, the American College of Obstetricians and Gynecologists (ACOG) and the Society for Maternal Fetal Medicine introduced the concept of Patient Safety Bundles for managing obstetric emergencies.29 Our results suggest that algorithms for management of general anesthesia during labor and delivery, including airway management, should be part of these bundles.30 However, protocols or safety bundles are not yet consistently used in labor and delivery units across the United States. For instance, an audit performed in 2012 identified that 23% of academic obstetric anesthesia units in the United States did not have a protocol for management of major obstetric hemorrhages.31 In 2005, the Agency for Healthcare Research and Quality recommended multidisciplinary team training to improve patient safety.32 Numerous studies suggest that team training in obstetric emergencies, including simulation training, results in improvement in technical and communication skills, team performance, and clinical outcomes.33
The cesarean delivery rate in New York state appears to have reached a plateau in 2009. This trend is consistent with national data.8,34 The prevalence of advanced maternal age and preexisting maternal chronic comorbidities, however, continue to increase, as indicated in the proportion of women older than 40 yr with a Charlson comorbidity index greater than 0, or a Bateman comorbidity index greater than 1. This increase in maternal age and burden of maternal disease has been suggested as an explanation for the observed increase in severe maternal morbidity and mortality in the United States in the past 2 to 3 decades.16,18,35,36 We also observed an increase in nonanesthetic-related perioperative morbidity. The strong relation between the proportion of discharges with Charlson comorbidity index greater than 0 and perioperative morbidity suggests that this increase may be explained by the increased prevalence of chronic comorbidities during pregnancy over the years of the study.6,18 The increase in chronic comorbidities and associated increase in perioperative morbidity highlights the need for better screening, counseling, and management of these high-risk women. It also supports the recent guidelines on levels of maternal care issued by the ACOG and the Society for Fetal-Maternal Medicine, which provide uniform designations for facilities so that appropriate patients are transferred to facilities that have the resources to provide the required level of care.37
The decrease in the proportion of thromboembolic disease and strokes in our analysis merits some discussion as these are the only two nonanesthetic perioperative complications that have decreased during the study period. This decrease also highlights that improvement is possible and that efforts should focus now on other types of morbid events or conditions such as the care of mothers with cardiac disease. The trend for venous thromboembolic disease mimics a decrease in maternal death due to thromboembolic disease observed in the United Kingdom.6 It may reflect better prophylaxis of venous thromboembolic disease as outlined by the Joint Commission in 2010 and promulgated by the ACOG in a 2011 Practice Bulletin.28,38 The decrease in stroke may be explained by better management of preeclampsia and eclampsia during pregnancy as outlined by the 2002 ACOG Practice Bulletin.39 It may also be explained by decreased use of general anesthetics that have been associated with an increased risk of stroke in preeclamptic women undergoing cesarean delivery.40
Consistent with previous studies examining postoperative mortality in nonobstetrical populations, we observed a marked decrease in the overall mortality rate in women undergoing cesarean delivery.41,42 This finding may appear to be inconsistent with the increase in overall maternal mortality reported in the United States over the past 2 decades.35 Our analysis focused only on cesarean deliveries and in-hospital outcomes, whereas national maternal mortality data encompass vaginal and cesarean deliveries, along with direct and indirect deaths within 42 days of delivery (maternal mortality ratio) or death from any cause up to 1-yr postdelivery (pregnancy-associated death).35,36
We acknowledge several limitations of our analysis. The findings are from a single state, New York, and may not be generalizable across the United States. A major drawback in our analysis of administrative data is that we were not able to assess the temporal relation between surgery and perioperative complications and to distinguish between preventable and nonpreventable adverse events. We recognize that the definition used for planned and unplanned cesarean delivery was derived from ICD-9-CM codes rather than a clinical diagnosis. Therefore, we were not able to differentiate urgent, emergent, or perimortem cesarean deliveries among unplanned cesarean deliveries.
In conclusion, a significant improvement in obstetric anesthesia-related outcomes was observed in New York state hospitals between 2003 and 2012. However, the lack of decrease for ARAEs during general anesthesia and the increase in non–anesthesia-related perioperative morbidity are concerning and may pose serious concerns for patient safety.
Support was provided solely from institutional and/or departmental sources.
The authors declare no competing interests.