Systemic Lupus Erythematosus
Short-Term Perioperative All-Cause Mortality and Cardiovascular Events in Women With Systemic Lupus Erythematosus
Address correspondence to Ali Yazdanyar, DO, PhD, Section of Hospital Medicine, Department of Medicine, The Reading Hospital and Medical Center, Sixth Avenue and Spruce Street, West Reading, PA 19611. E-mail: firstname.lastname@example.org.
Persons with systemic lupus erythematosus (SLE) are at an increased risk of cardiovascular disease (CVD) events, but this excess CVD burden in the perioperative setting is yet to be determined. We aimed to determine the risk of perioperative short-term all-cause mortality and CVD events among women with SLE compared to those without SLE.
We conducted a cross-sectional analysis of pooled hospital discharge data of the Nationwide Inpatient Sample from 1998–2002. We abstracted diseases and procedures using International Classification of Diseases, Ninth Revision, Clinical Modification codes. The principal procedure was categorized into either a low, intermediate, or high risk level. Survey logistic regression adjusting for potential confounders provided estimates for stratum-specific odds of adverse events in women with SLE relative to those without SLE for each procedure risk level.
All-cause mortality was significantly greater among women with SLE having a low- (odds ratio [OR] 1.54, 95% confidence interval [95% CI] 1.00–2.37) or a high-risk principal procedure (OR 2.52, 95% CI 1.34–4.75) relative to women without SLE, but did not differ significantly among persons with intermediate-risk procedures. Women with SLE with a low-risk procedure were also more likely to experience a composite CVD event relative to women without SLE (OR 1.40, 95% CI 1.04–1.87).
Women with SLE are at an increased risk for short-term perioperative adverse events. These results highlight a need for greater scrutiny during perioperative evaluation and management of women with SLE.
Cardiovascular (CV) events are a leading cause of morbidity and mortality in the perioperative setting [1, 2]. Disease states associated with increased CV disease (CVD) burden, such as diabetes mellitus and chronic kidney disease, have also been shown to pose an increased risk of adverse events in the perioperative setting. Although patients with systemic lupus erythematosus (SLE; lupus) have significant morbidity and premature mortality from CVD, the risk of perioperative adverse events is unknown.
Lupus is a systemic autoimmune inflammatory disease affecting predominantly women. The major cause of premature morbidity and mortality in women with lupus has been reported to be secondary to CVD [3-14]. Traditional risk factors for CVD, hypercoagulability, and chronic inflammation are thought to contribute to accelerated atherosclerosis in SLE [15-17]. We hypothesized that patients with SLE would have a greater risk of perioperative CVD events and all-cause mortality compared to patients without SLE. We aimed to determine the risk of perioperative CVD events and all-cause mortality during elective hospitalizations among patients with SLE compared to patients without SLE who underwent a noncardiac procedure.
Significance & Innovations
- The perioperative setting represents a time of heightened risk for morbidity and mortality, especially in individuals at an increased risk of cardiovascular disease.
- Women with systemic lupus erythematosus (SLE) are at an increased risk of premature morbidity and mortality attributed to atherosclerotic disease.
- The results of our study, which revealed an increased risk of adverse perioperative events in women with SLE, indicates a need for greater scrutiny in perioperative clinical care, in addition to further investigation as to the noncardiovascular causes of increased perioperative mortality observed in women with SLE.
SUBJECTS AND METHODS
The Nationwide Inpatient Sample (NIS) of the Healthcare Cost and Utilization Project (HCUP) is a survey design database that includes hospital discharge data from nonfederal (excludes veterans hospitals and other federal facilities), nonrehabilitation, acute care, short-term hospitals. It contains data abstracted from each inpatient hospitalization, including patient age, sex, diagnoses, and procedures (if any) performed during the hospitalization. Annually, the NIS includes data from approximately 6.8 million hospital discharges sampled from more than 984 hospitals across more than 22 states. The number of hospitals and states included in NIS data increased from 984 hospitals across 22 states in 1998 to 995 hospitals across 35 states in 2002. The data and supplemental files, including the NIS annual discharge data and the NIS Trends Supplemental Files, which were used to pool multiple years of NIS data, were obtained from the HCUP of the Agency for Healthcare Research and Quality (AHRQ) [18-21].
The study population included women ages ≥18 years with a hospitalization designated as elective occurring between 1998 and 2002 and that included a noncardiac principal procedure. We focused on hospitalizations designated as elective since, in contrast to emergent or urgent surgeries, it is within this population that preoperative clearance is most relevant. Additionally, given the lack of an indicator for present on admission status for listed International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnoses in the NIS, the choice of elective hospitalization provides the greatest likelihood that the procedure preceded composite CVD study end points. Finally, we focused on noncardiac procedures in order to eliminate the possibility of confounding by indication, which would exist for patients undergoing cardiac procedures.
Principal and secondary procedures
Principal procedures and secondary procedures were abstracted using ICD-9-CM codes (see Supplementary Appendix A, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.21915/abstract). The principal procedure was clustered into clinically meaningful categories (see Supplementary Appendix B, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.21915/abstract) using the Clinical Classifications Software, a software tool developed by the AHRQ, prior to further categorization into low, intermediate, or high risk for cardiac events based on guidelines published by the American College of Cardiology/American Heart Association [18, 22]. For example, endoscopic procedures, ophthalmologic procedures, and procedures on the breast are categorized as low risk, whereas orthopedic and intraabdominal procedures are categorized as intermediate risk. High-risk surgery includes peripheral vascular surgery, with the exception of carotid endarterectomy and noncardiac procedures within the cardiothoracic cavity. Secondary procedures were categorized into major and minor therapeutic procedures using the AHRQ Procedure Classification software tool prior to creating tally count variables . The tally count secondary minor or major therapeutic procedure variables were included as potential confounders in the fully adjusted statistical models.
Patient-level characteristics included as covariates were age and comorbid diagnoses recognized in previous studies as predictors of adverse perioperative events, including hypertension, diabetes mellitus, congestive heart failure (CHF) without pulmonary edema, valvular heart diseases, coronary artery disease, and chronic kidney disease. Comorbid diseases were abstracted using ICD-9-CM codes (see Supplementary Appendix A, available in the online version of this article at http://onlinelibrary.wiley.com/doi/10.1002/acr.21915/abstract).
An adverse perioperative event was defined as the occurrence of either in-hospital all-cause mortality or a composite CVD event. A composite CVD event included an acute myocardial infarction, a non–ST-segment elevation myocardial infarction, CHF with pulmonary edema, and/or an acute cerebrovascular accident.
Associations between categorical variables were determined using the design-adjusted Rao-Scott chi-square test . Survey logistic regression was used to determine the odds of the composite CVD outcome and all-cause mortality among hospitalized individuals with SLE compared to those without SLE. Survey logistic regression analysis was performed separately for each principal procedure risk level using subpopulation analyses. A significance level of 0.05 with a 2-sided test was used for all hypotheses. Statistical analyses were conducted using SAS and Stata, version 11.1, and accounted for the survey design features of the NIS data in order to provide population-based estimates.
The study included 5,267,589 elective hospitalizations with a noncardiac principal procedure among women with and without SLE between 1998 and 2002, of which 3,640,994 were categorized as low-, 1,513,597 as intermediate-, and 112,998 as high-risk procedures. Table 1 shows the characteristics of the study population by procedure risk and the presence or absence of SLE.
Table 1. Characteristics of hospitalized women by procedure risk level and lupus statusa
|Age, mean ± SE years||49.3 ± 0.6||45.1 ± 0.5||< 0.001||50.3 ± 0.5||54.7 ± 0.2||< 0.001||48.9 ± 1.3||67.1 ± 0.3||< 0.001|
|Hypertension||29.0 (1.1)||16.7 (0.5)||< 0.001||34.0 (1.4)||28.1 (0.4)||< 0.001||32.8 (3.4)||45.8 (0.7)||< 0.001|
|CAD||10.6 (0.7)||8.1 (0.4)||< 0.001||5.5 (0.7)||4.7 (0.1)||0.26||13.5 (2.3)||28.0 (0.7)||< 0.001|
|CHF||7.7 (0.6)||5.4 (0.2)||< 0.001||3.2 (0.5)||2.0 (0.1)||0.02||9.4 (1.9)||14.6 (0.4)||0.01|
|Valvular heart disease||0.8 (0.2)||0.6 (0.1)||0.25||0.6 (0.2)||0.3 (0.0)||0.13||3.4 (1.4)||6.0 (0.4)||0.07|
|DM||12.1 (0.7)||8.8 (0.3)||< 0.001||8.1 (0.7)||10.4 (0.1)||0.002||14.8 (2.9)||28.2 (0.7)||< 0.001|
|CKD||3.6 (0.5)||0.5 (0.03)||< 0.001||3.9 (0.8)||0.3 (0.03)||< 0.001||26.4 (3.9)||3.8 (0.4)||< 0.001|
Among women who underwent a low-risk principal procedure, women with SLE were older than the women in the comparison group, whereas among women in the intermediate- and high-risk principal procedure strata, women with SLE were younger (Table 1). Although women with SLE who underwent a high-risk procedure were more likely to have chronic kidney disease, there was a greater prevalence of coronary artery disease, diabetes mellitus, hypertension, valvular heart disease, and CHF among women without SLE. Coronary artery disease was more common among women with SLE in the low-risk principal procedure stratum but less common in the high-risk principal procedure stratum than women in the comparison group. Chronic kidney disease was more common in women with SLE than in the comparison group in each principal procedure risk stratum. Diabetes mellitus was more common among women with SLE in the low-risk principal procedure stratum but less common in the intermediate- and high-risk principal procedure strata than women in the comparison group. Hypertension and CHF were more common in women with SLE than in the comparison group in the low- and intermediate-risk principal procedure strata.
Perioperative all-cause mortality
Among women who underwent a low-, intermediate-, and high-risk principal procedure, there were 38,020 (1.04%), 5,444 (0.36%), and 3,333 deaths (2.95%), respectively. All-cause in-hospital mortality was more common among women with SLE compared to women without SLE who underwent a low-risk principal procedure (224 [1.69%] versus 37,796 [1.04%]; P = 0.02) or a high-risk principal procedure (54 [5.28%] versus 3,279 [2.93%]; P = 0.05), whereas the difference among persons with an intermediate-risk principal procedure was not statistically significant (33 [0.52%] versus 5,412 [0.36%]; P = 0.37).
Table 2 shows the odds of all-cause mortality among persons with a principal procedure. The odds of all-cause mortality were greater in women with SLE who underwent either a low- or high-risk principal procedure relative to the women in the comparison group, which persisted after adjusting for potential confounders. For women who underwent an intermediate-risk procedure, the odds of all-cause mortality did not differ significantly in women with SLE compared to women without SLE.
Table 2. Odds of adverse perioperative events by principal procedure riska
|Low-risk procedures|| || || || |
|Model A||1.00||1.64 (1.09–2.44)||1.00||1.50 (1.14–1.98)|
|Model B||1.00||1.91 (1.26–2.91)||1.00||1.73 (1.30–2.31)|
|Model C||1.00||1.56 (1.01–2.39)||1.00||1.42 (1.06–1.90)|
|Model D||1.00||1.54 (1.00–2.37)||1.00||1.40 (1.04–1.87)|
|Intermediate-risk procedures|| || || || |
|Model A||1.00||1.46 (0.63–3.35)||1.00||0.81 (0.31–2.10)|
|Model B||1.00||2.27 (0.99–5.21)||1.00||1.38 (0.52–3.63)|
|Model C||1.00||1.78 (0.78–4.09)||1.00||1.05 (0.39–2.83)|
|Model D||1.00||1.90 (0.83–4.36)||1.00||1.06 (0.39–2.84)|
|High-risk procedures|| || || || |
|Model A||1.00||1.85 (0.99–3.45)||–||–|
|Model B||1.00||2.98 (1.56–5.69)||–||–|
|Model C||1.00||2.65 (1.40–4.99)||–||–|
|Model D||1.00||2.52 (1.34–4.75)||–||–|
Composite CV outcome
A total of 57,285 (1.57%), 5,831 (0.39%), and 2,385 composite CVD events (2.11%) occurred among women who underwent a low-, intermediate-, and high-risk principal procedure, respectively. Women with SLE who underwent a low-risk principal procedure were significantly more likely to have a composite CVD outcome as compared to women without SLE (311 [2.34%] versus 56,974 [1.57%]; P = 0.003), whereas the difference in composite CVD events among women who underwent an intermediate-risk principal procedure was not significant (20 [0.31%] versus 5,812 [0.39%]; P = 0.66). There were no composite CVD events among women with SLE who underwent a high-risk principal procedure. After adjusting for potential confounders, women with SLE who underwent a low-risk principal procedure were at an increased risk of a composite CVD event relative to women without SLE (Table 2). A similar pattern of association was not observed among women who underwent an intermediate-risk procedure, and the presence of SLE perfectly predicted the absence of a composite CVD event among women who underwent a high-risk principal procedure.
The perioperative setting represents a time of heightened risk for adverse events. Given the literature on the association between SLE and CVD, we sought to determine whether women with SLE are at an increased risk for adverse perioperative events using hospital discharge data. Our findings revealed that women with SLE who underwent either a low- or high-risk principal procedure had greater than 1.5-fold odds of all-cause in-hospital mortality. In addition, women with SLE who underwent a low-risk procedure had a 40% greater relative risk of short-term CVD events as compared to women without lupus.
Over several previous decades, studies on the association between SLE and premature CVD morbidity and mortality have been accumulating. Deaths in SLE follow a bimodal pattern, with deaths later in the disease course attributed to atherosclerotic events with coronary events and cerebrovascular diseases accounting for up to 20% and 15% of deaths, respectively [3, 5, 9-11, 14, 24]. In terms of CVD morbidity, the prevalence of clinically significant ischemic heart disease has been estimated to approach 17% [4, 7, 8, 12, 25]. In an analysis of the California Hospital Discharge Database, young women with SLE were reported to be at a more than 8-fold greater risk for myocardial infarction and strokes and were 11 times more likely to have heart failure relative to women without SLE . In another report by Manzi et al comparing the rate of CVD events in women with SLE to women in the Framingham Offspring Study, women with SLE had a greater than 2.5- to 52-fold higher rate of CVD events . It is likely that the increased risk of CVD morbidity and mortality in persons with SLE is due in part to a higher prevalence of traditional CVD risk factors, prothrombotic state, and chronic inflammation [15-17].
Several risk indices have been developed to aid clinicians in the preoperative risk stratification of persons awaiting a noncardiac procedure [27-30]. The components of such risk indices include previously recognized CVD, such as coronary heart disease, CHF, and valvular heart disease, in addition to various CVD risk factors, including age, diabetes mellitus, hypertension, and chronic kidney disease. Given that the likelihood of an adverse perioperative coronary event or cardiac-specific mortality is greatest among persons with known or suspected coronary or atherosclerotic disease, it is of clinical relevance whether the perioperative period could represent a period of increased vulnerability to adverse events for persons with SLE. If so, there would be a need for greater scrutiny during perioperative risk assessment and management. More specifically, persons with SLE would then require a perioperative CVD risk stratification and management similar to other disease states such as diabetes mellitus.
Among women undergoing a low-risk procedure, those with SLE had an increased risk of adverse in-hospital perioperative events, including all-cause mortality and composite CVD events, relative to women without SLE. Because these procedures are considered low risk with respect to CVD complications, patients in this group likely included not only those with very low CVD risk, but also those with unsuspected or subclinical CVD and those with recognized CVD but who were judged to be appropriate candidates for the procedures. Therefore, the comparisons in this subgroup were likely least affected by selection factors. In turn, the relatively unselected nature of this subgroup may have helped uncover the perioperative risks associated with SLE. The risk associated with SLE was present in analyses that adjusted for known CVD and its major risk factors, but we could not adjust for subclinical CVD, which might have been present in a higher proportion of women with SLE and was unmasked by the procedure. A preoperative CVD risk assessment may have detected subclinical disease before the event.
In contrast, we did not find a significantly increased risk of perioperative CVD events or in-hospital mortality in those having intermediate-risk procedures, although the risk of mortality tended to be higher in women with SLE. In contrast to low-risk procedures, procedures in the intermediate-risk category are major surgeries recognized to entail an increased risk of complications, including possible CVD complications. Some of these procedures, including many orthopedic procedures, may be discretionary, and patients at an increased risk of CVD complications may elect not to undergo these procedures. Evidence for selection based on risk profile is seen in the comparison of patients undergoing intermediate-risk procedures and those undergoing low-risk procedures. The overall frequency of composite CVD events in the intermediate-risk procedure group was only 35% of that in the low-risk group, and in-hospital mortality was 25% of that in the low-risk procedure group. Moreover, both women with and without SLE in the intermediate-risk procedure group had a much lower prevalence of CVD and CHF than women in the low-risk procedure group. Selection based on known CVD risk and the discretionary nature of the procedures may have served to mitigate any difference in risk of adverse outcomes associated with SLE. Direct evidence of the extent to which selection was responsible for the results would require additional comparisons with women who had indications for these procedures but who did not have them.
Relatively few patients with SLE had high-risk procedures, which limited our ability to estimate differences in the risk of CVD events. However, the risk of in-hospital mortality was significantly higher among women with SLE in this subgroup. The absolute risk of death was also high in this subgroup, reflecting the high-risk nature of these procedures. The increased risk of death in the absence of an increased risk of composite CVD outcomes suggests that noncardiac causes of death predominated in the patients with SLE. However, part of the difference may also be due to sudden cardiac deaths that, although CV in origin, were not accompanied by or identified with a separate premorbid CVD event. Unfortunately, cause of death information was not available.
Although we studied a large and nationally representative sample and examined outcomes by strata of procedural risk, our study also has several limitations. First, the cross-sectional design has inherent limitations. Second, although we adjusted for the presence of known CVD and its major risk factors, the data did not permit adjustment for the severity of these conditions, which may have differed between the women with SLE and those without SLE. There may also be residual confounding by other comorbid conditions or by procedure-related factors such as duration of the procedure or type of anesthesia, although these factors would not be predicted to differ between patients with and without SLE. Third, diagnoses and procedures were abstracted using ICD-9-CM codes; therefore, one must consider the possibility of coding errors or misclassification. Fourth, women with SLE may not have been referred for elective procedures due to its recognized association with CVD risk, but this seems unlikely because the timeframe for the data predated the majority of published literature regarding SLE and CV risk. Fifth, we could not assess the impact of medications on perioperative adverse events due to lack of medication information in the NIS data set. Sixth, we studied only women, since there were too few men with SLE for meaningful analysis by procedure risk group. Finally, our study was only able to determine the risk of short-term in-hospital adverse events, and due to the lack of long-term followup data, does not address the risk of long-term postoperative adverse CVD or mortality events (i.e., 30 days or 60 days).
In conclusion, in a population of hospitalized women who underwent a noncardiac procedure, we found a significant association between SLE and perioperative composite CVD events and all-cause mortality, particularly among those undergoing low-risk procedures. Our study serves as the initial step in attempting to quantify the significance of the atherosclerotic disease in SLE during the perioperative period. Prospective studies are needed to further elucidate the relationship between SLE and perioperative risk, including cause-specific mortality and long-term postoperative CVD and non-CVD morbidity and mortality, and potential benefits of different perioperative risk assessment strategies.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Yazdanyar had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Yazdanyar, Wasko, Scalzi, Kraemer, Ward.
Acquisition of data. Yazdanyar.
Analysis and interpretation of data. Yazdanyar, Wasko, Scalzi, Ward.