Risk of Cardiovascular Disease Due to General Anesthesia and Neuraxial Anesthesia in Lower-Limb Fracture Patients: A Retrospective Population-Based Cohort Study

The purpose of this study was to analyze the relationship between elevated cardiovascular disease (CVD) risk and type of anesthesia by using the National Health Insurance Research Database (NHIRD) of Taiwan in a one-year follow-up period. We assessed whether general anesthesia (GA) or neuraxial anesthesia (NA) increased CVD occurrence in lower-limb fracture patients. Approximately 1 million patients were randomly sampled from the NHIRD registry. We identified and enrolled 3437 lower-limb fracture patients who had received anesthesia during operations conducted in the period from 2010 to 2012. Next, patients were divided into two groups, namely GA (n = 1504) and NA (n = 1933), based on the anesthetic technique received during surgery. Our results revealed that those receiving GA did not differ in their risk of CVD relative to those receiving NA, adjusted HR = 1.24 (95% CI: 0.80–1.92). Patients who received GA for more than 2 h also did not differ in their risk of CVD relative to those receiving NA for less than 2 h, adjusted HR = 1.43 (95% CI: 0.81–2.50). Moreover, in the GA group (i.e., patients aged ≥65 years and women), no significant difference for the risk of CVD events was observed. In conclusion, in our study, the difference in the risk of CVD between lower-limb fracture patients receiving NA and GA was not statistically significant. The incidence rate of CVD seemed to be more correlated with patients’ underlying characteristics such as old age, comorbidities, or admission to the intensive care unit. Due to the limited sample size in this study, a database which reviews a whole national population will be required to verify our results in the future.


Introduction
Anesthetic techniques can be classified as general anesthesia (GA), regional (caudal and neuraxial (spinal and epidural)) anesthesia, and peripheral nerve block. Choosing a suitable anesthetic technique can have benefits for patients. For some patients, physicians need to determine whether GA or neuraxial anesthesia (NA) is more suitable for a particular surgical procedure. GA has been reported to be a risk factor for dementia. In a mouse experiment, anesthetizing mice with ether vapors or sodium pentobarbital for short periods (30 s to 5 min) induced tau phosphorylation. In humans, this might increase the risk of postoperative cognitive impairment and the risk for Alzheimer's disease. [1]. Isoflurane-treated AbetaPP mice had increased beta-amyloid protein aggregates in the brain [2]. Studies on the relationship between GA and AD have yielded inconsistent results. Another example includes a primiparous female who underwent cesarean delivery with epidural anesthesia, but not with spinal anesthesia, who was suggested to have an increased risk of subsequent chronic low back pain [3].
Cardiovascular disease (CVD) is one of the most common causes of morbidity and mortality worldwide. To date, CVD is the second leading cause of death in Taiwan. According to the definition of the World Health Organization, CVD is a group of disorders of the heart and blood vessels including coronary heart disease, cerebrovascular disease, peripheral arterial disease, rheumatic heart disease, congenital heart disease, deep vein thrombosis, and pulmonary embolism. Risk factors for CVD have been divided into invariable and variable risk factors. Invariable risk factors include age, sex, and genetic status [4]. However, known variable risk factors include improper lifestyle habits, such as smoking tobacco and poor exercise and eating habits. Other risk factors associated with CVD include the presence of underlying diseases such as arterial hypertension, diabetes mellitus, dyslipidemia, cholesterol, and obesity. In addition, myocardial injury after noncardiac surgery (MINS) might be important factor in cardiovascular events during perioperative complications [5]. Hip fracture was found to be one of the risk factors for MINS during the pre-operative period [6].
Anesthesia types are unique to each individual patients with ASA (American Society of Anesthesiologists) physical status or multiple comorbidities and types of surgery, even in different hospital [7]. As such, the purpose of physicians and anesthesiologists in deciding the method of anesthesia (GA or NA) is to reduce the risk of complications, promote faster recovery, minimize surgical stress response, and improve post anesthesia outcome [8]. In previous literature reviews, the advantages and risks of GA and NA have been compared. For cesarean section, NA was linked with a shorter duration of hospitalization than GA [9]. The opioid-sparing effects of NA were associated with a better quality of recovery in patients after the surgical procedure [10]. A meta-analysis indicated that the use of NA for hip joint replacement may have better outcomes, including a lower blood transfusion requirement and a decrease in deep venous thrombosis or pulmonary embolism and serious complications [11]. In addition, the long-term effects of anesthesia remain unknown. Although NA may be associated with a relatively lower risk on several aspects and a better outcome when compared with GA, more studies are required for verification. Therefore, to determine whether GA or NA have shown benefits and better outcome for patient with lower limbs fracture surgery, we compared the association of GA and NA with CVD risk in lower-limb fracture patients at the one-year follow-up period. The purpose of this study was to investigate the risk of CVD in patients with GA and NA. The hypothesis was that GA and NA have significant differences in CVD risk.

Data Sources
The Longitudinal Health Insurance Database (LHID) is managed by the Taiwan National Health Research Institutes. One million randomly sampled beneficiaries were recruited with all outpatient and inpatient medical claims including drug medications, medical operations, procedures, and fees. The study was approved by the Ethical Review Board of Chung Shan Medical University Hospital (CS18096).

Study Groups
This study used a retrospective cohort study design. All recruited patients were aged 18 years or older, had lower-limb fractures (ICD-9-CM codes 820-827), and underwent an operation with anesthesia between 1 January 2010, and 31 December 2012. The index date was defined as the first date of lower-limb fracture. To eliminate confounding from a past disease or re-anesthesia surgery, we excluded patients with a CVD diagnosis before the index date or the requirement of anesthesia surgery again during the study period. The study population was divided into GA and NA groups by surgical codes (96005C, 96006C, 96007C, 96008C, 96017C, 96018C, 96019C, 96020C, 96021C, and 96022C). NA included spinal and epidural anesthesia.

Outcome and Covariate Measurement
The study endpoint was the diagnosis of new-onset CVD, including ischemic heart disease (ICD-9-CM codes 410, 411, 413, 414), heart failure (ICD-9-CM codes 428), and cerebrovascular disease (ICD-9-CM codes 430-437) during emergency or hospitalization. All patients were followed until one year until the occurrence of CVD, withdrawal from the social insurance system, or the end of follow-up, whichever came first. , and usage of anticoagulants (aspirin/clopidogrel/warfarin). From the LHID, we were unable to obtain personal behavioral information such as smoking, alcohol consumption, and laboratory data. In this study, we used chronic pulmonary disease and liver disease as substitutes for the indirect control of smoking and alcohol consumption interference. Other comorbidity choices, such as hypertension, hyperlipidemia, diabetes, and renal disease are all potential CVD risk factors. These covariates were defined before the index date. Furthermore, admission into the intensive care unit (ICU) was considered to control the severity of lower-limb fractures in both groups.

Statistical Analysis
To compare GA and NA, the chi-squared test or independent t test was used, as appropriate. The cumulative incidence of CVD cases was analyzed using Kaplan-Meier analysis, and the significance was calculated using the log-rank test. The Cox proportional hazard model was used to estimate the hazard ratios of CVDs. A sensitivity analysis of risk of CVD events was also conducted for different follow-up periods. Statistical analyses were performed using SPSS V.18.0 (SPSS, Chicago, Il, USA). A p-value less than 0.05 was considered statistically significant.

Characteristics of Study Subjects
A total of 6319 patients with lower-limb fractures who received anesthesia during an operation were identified from 2010 to 2012. After excluding patients with CVD diagnosis and repeated anesthesia surgery during the study period, 3437 patients were enrolled in the study. Furthermore, patients were divided into two groups. One group received GA (n = 1504) and the other group received NA (n = 1933) during the surgery for lower-limb fractures (Figure 1). The distributions of age, sex, comorbidities, ICU admission, anticoagulants, anesthesia time, and hospital levels among patients receiving GA and NA are presented in Table 1. The mean age and standard deviation of patients receiving GA and NA were 50.7 ± 20.8 and 57.6 ± 20.8 years, respectively. The proportions of male patients receiving GA and NA were 52.1% and 50.7%, respectively.  We estimated the cumulative incidence of CVD cases in patients who received GA and NA. The Kaplan-Meier curve of the cumulative probability of CVDs indicated that the GA group did not have a higher risk of CVDs within 1 year (log-rank test, p = 0.933; Figure 2). The Cox proportional hazard model revealed that the incidence of CVD did not differ in patients receiving GA than in patients receiving NA, HR = 1.24 (95% CI: 0.80-1.92). After adjustment for age, sex, comorbidities, ICU admission, and anticoagulants, patients aged 65 years or older (adjusted hazard ratio (HR): 5.77; 95% confidence interval [CI]: 3.21-10.38), male patients (adjusted HR: 1.89; 95% CI: 1.21-2.95), and patients with comorbidities, such as diabetes (adjusted HR: 2.09; 95% CI: 1.29-3.37), had a higher risk of developing a CVD. Patients who had been admitted to the ICU due to lower-limb fractures also had a higher risk of a CVD event occurring (adjusted HR: 2.59; 95% CI: 1.35-4.96; Table 2).

Risk of CVD Between General Anesthesia and Neuraxial Anesthesia Patients and Specific Subgroup Characteristics
To determine whether the duration of both GA and NA affects the risk of CVD, both anesthesia groups were divided into less than 2 h and more than 2 h groups. Compared with patients who received NA for less than 2 h, patients who received GA for more than 2 h did not differ in their risk of developing a CVD (adjusted HR: 1.43; 95% CI: 0.81-2.50; Table 3). We also conducted a subgroup analysis between GA and NA patients. In stratification of age and gender, those receiving GA did not differ in their risk of CVD relative to those receiving NA (Table 4). A sensitivity analysis was conducted, which showed that the GA and NA groups did not have statistically significant differences in different follow-up periods (Table 5).

Discussion
In this study, a total of 3437 lower-limb fracture patients who received GA (n = 1504) or NA (n = 1933) were enrolled. The risk of CVD events did not differ in general anesthesia when compared with neuraxial anesthesia in the Cox proportional hazard model. To date, no studies have reported either GA or NA as a risk factor for the development of CVD. In contrast, CVDs have been described as postoperative outcomes of any type of anesthesia [12][13][14]. Generally, the postoperative time frame can range from 24 h to 30 days [13]. In our study, the average onset of CVDs in patients who received NA and GA was 4.8 and 6.0 months, respectively (Appendix A Table A1).
Risk factors for CVDs have accumulated rapidly. Underlying diseases, such as hypertension, atrial fibrillation, kidney failure, diabetes, aging, and sex are also risk factors for CVDs that have been reported for more than 20 years [4,15]. In our study, patients aged ≥65 years, male patients, and patients with diabetes were shown to have a higher risk of developing a CVD.
In this study, we found that lower-limb fracture patients admitted to the ICU had a higher risk of developing CVD relative to those lower-limb fracture patients that underwent surgery but through direct hospital admission. In fact, several studies have indicated that VTE might be a predisposing factor of CVDs [16][17][18]. In fact, patients with particularly serious diseases staying in the ICU could also have an increased risk of CVD. A cohort study indicated that intensive care patients with acute kidney injury (AKI) had an increased risk of heart failure or myocardial infarction up to three years after hospital discharge [19].
Recently, two meta-analyses indicated an association between intra-operative hypotension (IOH) and MINS. Gu's group reviewed 14 cohort studies of IOH and MINS. They found that IOH was associated with increased risks of 30-day mortality and major adverse cardiac events (MACEs), especially myocardial injury and AKI after non-cardiac surgery [20]. Similarly, An's group reviewed 15 observational studies of IOH and MINS. They found that IOH was associated with an increased risk of postoperative AKI and myocardial injury. In more detail, the duration of IOH lasting only more than 5 min was associated with an increased risk of 30-day mortality [21]. In particular, patients suffering from pre-existing CVDs may have increased risks of perioperative cardiovascular morbidity and mortality following noncardiac surgery. Therefore, physicians and anesthesiologists need to consider cardioprotective medication, anesthetic regimen choice, blood pressure management, and transfusion regimens carefully in order to decrease such CVD risk [22]. The development of intraoperative hypotension is known to be associated with MINS, intraoperative vasopressor use, postoperative sepsis, number of ventilator days, and the types and quantities of ICU sedative and narcotic use, just to name a few contributors.
The current study has some limitations. First, we did not have access to potentially relevant personal behavioral information such as smoking habits, alcohol consumption, body mass index, and the severity of lower-limb fracture. This is significant because smoking, drinking, and BMI (obesity) are risk factors for CVD, and these factors may be diversifying the type of anesthesia. Second, the dataset did not provide information regarding the types of anesthetic technique used in patients, clinical presentations, or laboratory data. Patients' individual statuses, presence of severity of lower limbs fracture and bleeding profile laboratory data may determine the type of anesthesia method selected by anesthesiologist. Third, the study subjects were not selected from the national population. Therefore, in consequence of limiting by the sample sizes that the study results may exist a wide confidence interval and corresponding lower power.

Conclusions
In summary, our study indicates that the difference in the risk of CVD between lower-limb fracture patients receiving NA and GA was not statistically significant. Owing to the limited sample size, further experiments are required to identify whether different anesthetic procedures are associated with differential risks of CVD in the future. For the application of anesthesia in different diseases, further study is needed for verification. Therefore, physicians themselves might decide to use GA or NA under different conditions.