The obesity paradox revisited – influence on the results of percutaneous coronary interventions

Coronary artery disease is a leading cause of death, with hundreds of thousands of percutaneous coronary interventions (PCI) being performed each year in the United States1. PCI is associated with many risks, including perforation, tamponade, haemodynamic collapse, failure to recanalise the vessel, and death2. Thus, it is imperative to understand patient risk factors for such complications, especially mortality. One potential risk factor that requires further investigation is the patient body habitus. According to the Centers for Disease Control and Prevention (CDC), over 70% of adults in the United States are overweight, with over 40% being considered obese3. It is well established that obesity is a strong risk factor for the development of cardiovascular diseases, such as hypertension, as well as metabolic syndrome45. Thus, it is commonly presumed that, since obesity is such a strong risk factor for the development of cardiovascular disease, it must also be a risk factor for poor outcomes once a patient has developed such disease. However, some studies have shown that in many cases obese patients may have better outcomes in the context of cardiovascular disease than their non-obese counterparts – termed the “obesity paradox”6. In particular, multiple studies have reported the obesity paradox in the context of PCI7891011. Despite such evidence, the obesity paradox remains controversial, with criticisms of the paradox including failing to distinguish between metabolically healthy and unhealthy obesity as well as attributing the paradox as a whole to statistical bias in observational studies1213. Considering the number of overweight and obese Americans, the frequency with which PCI is performed, and the persistent controversy surrounding the obesity paradox, it is clear that further investigation and evidence are needed. This study, the largest retrospective cohort investigation of the relationship between obesity and mortality following PCI, seeks to further elucidate the obesity paradox and the risk or protection conferred by increased body habitus to patients undergoing PCI.

Methods

Data source

The study population was drawn from the National Inpatient Sample (NIS), a publicly available dataset provided by the Agency for Healthcare Research and Quality. The NIS dataset is one of the largest publicly available, nationally representative inpatient datasets, as it approximates a sample of about 20% of US community hospitals, and about 98% of the total US population, using discharge weights14. The NIS dataset is publicly available and includes no identifiable information, precluding the study from requiring institutional review board approval.

Study population

Patient data were drawn from the NIS, for the years 2016 to 2020, and International Classification of Diseases, Tenth Revision (ICD-10) codes were used to search the database and further stratify the study populations. As outlined in the previous study by Nathan et al, the following ICD-10 procedure codes were used to identify patients who had undergone PCI: 02703(4-7)Z, 02703(D-G)Z, 02703TZ, 02713(4-7)Z, 02713(D-G)Z, 02713TZ, 02723(4-7)Z, 02723(D-G)Z, 02723TZ, 02733(4-7)Z, 02733(D-G)Z, 02733TZ, 02H(0-3)3DZ, 02H(0-3)3YZ, 027(0-3)3ZZ, 02C(0-3)3Z7, 02C(0-3)3ZZ, 02F(0-3)3ZZ15. ICD-10 diagnosis codes were then used to classify patients as cachectic (R64), overweight (E66.3), obese (E66.9, E66.8, E66.0), or morbidly obese (E66.01, E66.2).

Study outcome and statistical analysis

The primary study outcome was patient mortality following PCI during hospital admission. In addition to analysing mortality, patient characteristics and demographics were also analysed and incorporated into multivariate analysis of mortality. These patient characteristics include smoking history (F17.20, Z72.0, Z87.891), diabetes (E08-E13), hypertension (I10, I11.0, I11.9, I120, I129, I13.0, I13.10, I13.11, I13.2, I15.0, I15.1, I15.2, I15.9, I16.0, I16.1, I16.9), chronic obstructive pulmonary disease (COPD; J41.0, J41.1, J41.8, J42, J43.0, J43.1, J43.2, J43.8, J43.9, J44.0, J44.1, J44.9, J47.0, J47.1, J47.9, J684), chronic kidney disease (CKD; I13.11, I13.2, N289, Q613, N181, N182, N183, N1830, N1831, N1832, N184, N185, N186, N189, R880, N19), ST-segment elevation myocardial infarction (STEMI; I21.01, I21.02, I21.09, I21.11, I21.19, I21.21, I21.29, I21.3, I21.9, I21.A1, I21.A9, I22.0, I22.1, I22.5, I22.9), non-STEMI (I21.4, I22.2), and previous myocardial infarction (I25.2). The patient demographics considered included sex and race. Patient demographic information and mortality are reported as means with 95% confidence intervals. Binary clinical outcomes were ascertained using logistic regression, and multivariate analysis was performed to adjust for confounders. All analysis was performed using population discharge weights. The reported p-values are two-sided, with p<0.05 considered statistically significant. Data were analysed using Stata 17 (StataCorp).

Results

A weighted total of 10,069,454 patients were identified who had undergone PCI. Of these patients, 8,110,634 were of normal weight, 60,690 were cachectic, 55,765 were overweight, 1,134,765 were obese, and 716,085 were morbidly obese. The average age for the whole study population was 66.99 years. The average age decreased as the patient’s body mass index (BMI) increased (cachectic: 74.43 years, normal weight: 70.98 years, overweight: 68.20 years, obese: 66.43 years, morbidly obese: 64.20 years). The study population was primarily comprised of males (63.67%) (Table 1). When compared to patients in the normal-weight category, cachectic patients had the highest mortality at 9.78% (odds ratio [OR] 3.88, 95% confidence interval [CI]: 3.65-4.12; p<0.001). Mortality was lowest in overweight patients at 1.28% (OR 0.46, 95% CI: 0.39-0.55; p<0.001), followed by obese patients at 1.61% (OR 0.58, 95% CI: 0.56-0.61; p<0.001). In the morbidly obese category, this protective effect was much less, with mortality being measured at 2.05% (OR 0.75, 95% CI: 0.42-0.78; p<0.001; vs 2.72% in the normal-weight category) (Figure 1). After multivariate analysis adjusting for baseline characteristics and comorbidities, mortality remained high in cachectic patients (OR 3.65, 95% CI: 3.42-3.90; p<0.001) and remained low in overweight (OR 0.51, 95% CI: 0.43-0.61; p<0.001) and obese (OR 0.68, 95% CI: 0.66-0.71; p<0.001) patients, but the protective value of weight almost disappeared in the morbidly obese category (OR 0.96, 95% CI: 0.96-1.00; p=0.04) when compared with normal-weight patients (Figure 2).

Table 1. Patient demographic information, overall and by BMI class.

Figure 1. Percentage of mortality by weight categories in patients undergoing percutaneous coronary intervention.

Figure 2. Odds ratios of mortality in patients undergoing percutaneous coronary intervention, according to body mass index.

Discussion

Our results demonstrate that the obesity paradox holds partially true in the context of patients undergoing PCI. Univariate analysis found that mortality after PCI for cachectic patients was 8.5% higher than that for normal or overweight patients. However, the overweight category had the lowest mortality, followed by the obese category, with almost complete loss of protection in morbidly obese patients. These results are consistent with multiple other studies investigating the obesity paradox in patients undergoing PCI. For example, a meta-analysis conducted by Liu et al considering over 200 studies found that overweight patients (OR 0.66) and obese patients (OR 0.60) were at lower risk of mortality following PCI for STEMI than normal-weight patients8. The magnitude of the risk reduction in their study is consistent with that found in this current investigation. The obesity paradox was also supported by the retrospective database investigation conducted by Li et al, which found a U-shaped relationship between BMI and mortality following coronary artery bypass grafting and an L-shaped curve in the case of PCI9. Thus, while the magnitude of protection conveyed by obesity and morbid obesity may be less consistent among studies, it is clear that being underweight or of normal weight conveys a higher risk of mortality following PCI than being overweight, as is further demonstrated in our study. Even studies with longer follow-up, such as that conducted by Ueshima et al with a follow-up time of 3 years, found that overweight patients had lower rates of major adverse cardiac events following drug-eluting stent placement than non-overweight patients11. Considering the results of these studies and the results of this current study outlined previously, it is clear there is substantial evidence supporting the validity of the obesity paradox in the context of patients undergoing PCI. However, in our study, overweight patients had the best outcome of the weight categories, suggesting that excessive weight may have a negative effect, and morbid obesity may lose any protective effect seen in the obesity paradox. While the evidence for the existence of the obesity paradox is strong, the cause of the paradox remains uncertain. One proposed mechanism is that overweight and obese patients are often younger at presentation and may be managed with more aggressive medical therapy. For example, a study conducted by Bundhun et al, which lends support to the obesity paradox, found that overweight patients were more likely to have intensive use of medications than non-obese patients16. More intense medication use post-PCI in obese patients following discharge was also observed by Tan et al in their meta-analysis17. Beyond more intense medical management of obese patients, age alone may contribute to the observed paradox. A Swedish study investigating outcomes of out-of-hospital cardiac arrest found that obese patients were significantly younger at the time of arrest than non-obese patients18. While their study investigating out-of-hospital cardiac arrest did not support the obesity paradox, our results do demonstrate that obese and overweight patients are significantly younger at the time of PCI than normal-weight patients. As reported in the results section, the average age decreased as patient BMI increased (cachectic: 74.43 years, normal weight: 70.98 years, overweight: 68.20 years, obese: 66.43 years, morbidly obese: 64.20 years). While decades of research have demonstrated that obesity plays a strong role in the development of cardiovascular disease, the notion that overweight and obese patients develop such disease at a younger age may be an advantage when it comes to surviving cardiac procedures, contributing to the observed obesity paradox. Another factor potentially involved in the aetiology of the obesity paradox is the difference between metabolically healthy and unhealthy obesity and the distinction between visceral versus subcutaneous adiposity. It is well established that central adiposity carries a higher metabolic risk than peripheral adiposity19. One study investigating obesity in the context of Takotsubo syndrome found that subcutaneous adiposity was protective, leading to better in-hospital outcomes and overall clinical course. The author continued to discuss the observed benefits contributing to lower autonomic sympathetic nervous system activity in patients with subcutaneous obesity, leading to lower stress placed on the damaged heart20. With this in mind, it is possible the observed obesity paradox may be due, in part, to benefits experienced by patients with subcutaneous obesity, versus those with central and visceral adiposity. However, the NIS database does not provide information regarding fat distribution, thereby limiting our results. Critics of the obesity paradox often claim it is merely correlation and not causation, citing statistical bias as the underlying cause12. Additionally, more specific criticism of studies supporting the obesity paradox includes failing to control for smoking status as a confounding variable, as not only does smoking increase cardiac risk, but it is often associated with lower body mass due to its effects on metabolic rate11. While a valid concern, our study has controlled for not only smoking status, but also COPD, CKD, hypertension, diabetes, race, sex, and age; when such a control is considered alongside the sheer size and representativeness of the NIS database, our results are strongly supported. While our study did consider smoking as a possible confounding variable, it still had multiple limitations. First, the NIS database did not allow for the specification of whether mortality was due to a cardiac or non-cardiac cause. This is an important implication and should be further studied in future investigations. Additionally, while previously discussed studies considered medication differences between overweight and non-overweight patients, the NIS database does not provide such information. Finally, the NIS database does not include any information related to long-term survival or quality of life following discharge, which should be further investigated. The cause of the obesity paradox is not known. Metabolic reserve appears to be the most important reason for the obesity paradox. Patients in critical conditions have high metabolic demands that could protect obese patients from nutritional-related mortality20. The fact that cachexia has the highest mortality is consistent with this hypothesis. There are some data indicating that obese patients may present at an earlier stage in the course of their illness, and therefore, they are less sick21. Furthermore, the definition of normal weight versus overweight may need adjustment as many overweight patients may have better fitness than patients defined in the normal-weight definition of a BMI between 18.5 and 25. If a slightly higher BMI were categorised as normal weight, we could show a lower mortality in the normal-weight category. Furthermore, at the lower end of the normal-weight category, many patients could be closer to the cachexia category, despite being called normal weight. This could result in high-risk cachectic patients being included in the normal-weight category, falsely elevating the mortality rate in this population2223.

Limitations

Our study is a retrospective analysis and not a randomised trial, limiting our results. We used ICD-10 coding for weight categories and did not have true BMI. However, ICD-10 coding followed recorded BMI and therefore should be accurate. We analysed the inpatient population; therefore, our results cannot evaluate any effect of obesity in a stable outpatient population. We had no access to medication data, preventing us from adjusting for medication used. Furthermore, our mortality data only included total mortality without the ability to assess cardiac mortality.

Conclusion

The “obesity paradox” held only partially, with the lowest mortality in the overweight category followed by the obese category, with an almost complete loss of protection in morbidly obese patients; mortality was highest in cachectic patients. The partial confirmation of the obesity paradox, with loss of protection with increasing weight, suggests that we should continue to advise against obesity and morbid obesity in our population. The cause of the most protective effect of overweight status in PCI patients is unknown and warrants further investigation.

Impact on daily practice

Based on our report, obesity has a positive effect on mortality and therefore should not be an obstacle for performing percutaneous coronary intervention (PCI) in obese or morbidly obese patients. However, extra caution must be taken when PCI is planned in patients with cachexia. These patients should be informed about this higher risk.

Conflict of interest statement

The authors have no conflicts of interest to declare.