In this large population-based study of hospitalized individuals, we sought to determine the association between the absence or presence of either or both diagnoses of RA and DM with perioperative CV events and all-cause mortality. In our analyses, we found that hospitalized RA patients who underwent a non-cardiac principal procedure were not at an increased risk of in-hospital short-term perioperative adverse CV events or mortality compared with those without RA or DM. These findings were consistent across all three levels of procedural risk and after controlling for comorbid diseases and additional secondary hospital-based procedures. To our knowledge, this is the first study to investigate the association between RA and perioperative mortality and CV risk.
Recent studies comparing the risk of CV disease in RA with DM have found an elevated risk of CV events in RA comparable in magnitude to DM.16,17
In the perioperative setting, we compared the prevalence of CV comorbid disease and the risk of perioperative CV events among hospitalized persons with either RA, DM, neither, or both conditions. In comparison to persons without RA or DM, persons with RA did indeed have a higher prevalence of CV comorbid diseases including coronary artery disease, hypertension, and congestive heart failure. Regarding the risk of adverse perioperative events, our results were consistent with previous literature in that persons with DM remained at an increased risk of adverse perioperative events.29,30
However, despite the higher prevalence of CV comorbid disease, our study did not find persons with only RA to have an elevated risk of CV events. Given the high prevalence of CV comorbid disease in persons with RA in our study population, we addressed the possibility of model over-adjustment by including a multivariable model adjusting for only age and sex. However, the results of this intermediate multivariable model did not reveal that over-adjustment accounted for the lack of association between RA and adverse perioperative events.
The results of our supplementary analyses were in large part consistent with our primary analyses. In order to account for the possibility of procedure case-mix within each procedure risk category, we conducted a separate procedure-specific analysis to determine the odds of adverse perioperative events among elective hospitalizations with a principal procedure of total knee replacement. While the result of this supplementary analysis was consistent among persons with RA alone, it did suggest that the subgroup of persons with both RA and DM who may be at heightened risk for adverse perioperative events. Due to the multi-factorial and possible non-atherosclerotic etiology of decompensated heart failure we conducted additional analyses excluding acute heart failure as a component of the composite CVD endpoint results of which did alter study conclusion.
Given the large body of evidence suggesting an increased burden of CV disease in RA, we consider some limitations and potential explanations for our somewhat surprising findings. First, our study in limited due to factors related to its observational and cross-sectional design. Secondly, some possible explanations for our results include a differential pre-hospitalization preoperative risk assessment and modification. Specifically, patients with active or aggressive RA would probably be less likely to undergo an elective non-cardiac surgical procedure. Also, RA patients may not have been referred for an elective surgical procedures because the recognition of the overall increase in risk of CV events leading to a selection process that excluded RA patients from elective surgical procedures. Given that much of the epidemiologic research highlighting the association of RA with CV disease has been carried out in the past decade, and post-dates the time of this dataset (1998–2002) this possibility seems unlikely. Likewise, patients with DM who at greater perioperative risk may have been identified prior to an elective surgery because of pre-operative scrutiny, therefore underestimating the risk in these patients relative to an unselected group of patients with DM. Third, a number of RA-specific factors which were not available in the NIS dataset, such as the disease duration, magnitude of systemic inflammation, use of disease modifying drugs, as well as functional status may have had an impact on the CV disease profile.4, 31–33
Fourth, one must consider the possibility of coding errors or misclassification when using ICD-9 codes for abstraction of diseases and procedures.34–36
For instance, the misclassification of a non-inflammatory arthritis as RA may have led to an attenuation of the association between disease and perioperative CV events. Fifth, while a mix of procedures within each procedure risk group, especially if these differed by diagnosis, may have introduced residual confounding, the results of our supplementary analysis make this possibility less likely. Lastly, there was a lack of an anticipated graded increase in the rate of adverse perioperative events across low, intermediate, and high risk procedures which may have been due to above mentioned possibilities of differential pre-operative screening and limitations inherent to use of ICD-9 codes and other limitations such as the lack of additional information on procedure-related factors, including indication and duration.
The strengths of our study include the large population-based sample of patients, the comparison with both unaffected controls and patients with DM, and examination of risks within surgical risk strata. It serves as the initial step in attempting to quantify the significance of the increased atherosclerotic burden of patients with RA in the perioperative period serving well those who encounter this task routinely in clinical practice. Several risk stratification indices have been developed to aid clinicians in preoperative risk stratification with their components including various recognized CV disease states, such as coronary heart disease, congestive heart failure and valvular heart disease.28
In addition, these indices include characteristics such as age, DM, chronic kidney disease, and hypertension. Accumulating evidence for the CV disease burden in RA prompted our investigation of the risk associated with RA in the perioperative setting.18
Our findings did not reveal a significant association RA and mortality or perioperative CV events. The results of our cross-sectional observational study we present here do not support the need for a change in the currently practiced preoperative screening for CV disease in persons with RA being evaluated before elective non-cardiac procedures. However, our result reflect short-term in-hospital adverse outcomes and do not exclude the possibility of an increased risk of adverse post-operative events after longer follow-up time after surgery. therefore, prospective studies are warranted to further elucidate the relationship between RA and perioperative CV risk.