BMI and its association with death and the initiation of renal replacement therapy (RRT) in a cohort of patients with chronic kidney disease (CKD)

Obesity was one of the major themes of World Kidney Day in 2017 [1]. Over the past two decades, a multitude of observational studies have illustrated that obese patients on renal replacement therapy (RRT) have better survival than those with normal BMI. This concept of “reverse epidemiology” has also been observed in patients with heart failure, chronic obstructive pulmonary disease, liver cirrhosis and metastatic cancer, as well as in the geriatric population [2‐7]. Recent studies have confirmed this “obesity paradox” in contemporary cohorts across different ethnicities and races, as well as different geographic regions on the world [2‐7].

In a multicenter prospective observational cohort study of 2833 patients in Korea on maintenance hemodialysis, individuals with higher BMI had better survival over an average follow-up of 24.2 months [8]. Some groups have also reported that obesity in non-dialysis dependent CKD was associated with lower risk for all-cause mortality [9‐13].

We recently showed that patients with chronic kidney disease (CKD) in three metropolitan public renal practices in an Australian setting had strikingly higher rates of obesity than the general population [14]. We described the cross-sectional associations of obesity with clinical characteristics [14], but we did not report the outcomes of RRT and death without RRT in the context of BMI. Here our specific objectives are to examine the association between BMI and death prior to RRT and starting RRT in this same cohort, followed for up to 5 years.

A larger proportion of the non-dialysis CKD population in this Australian study is overweight or obese. Fifty one point two (51.2%) percent of subjects were obese, compared with 28.4% of a contemporaneous National Health Survey (NHS) sample of the general Australian population [18]. Even patients without diabetes in our CKD study population were notably more often overweight or obese than the NHS sample population [14, 18]. Although obesity is widely considered a health risk [19], when followed for 5 years, overweight or mildly obese patients in this cohort of patients with CKD, particularly those males who were above the group median age of 68 years at enrolment, had a significantly lower risk for death without RRT than males with a normal BMI. There were no significant associations between higher BMI and starting RRT in either males or females.

The phenomenon of lower mortality in older males with CKD who were overweight or obese is consistent with recent studies in which dialysis patients of higher BMI had lower rates of death than those of “normal” BMI, termed the “obesity paradox” [8, 20]. The explanation for this phenomenon remains elusive. Does it represent exacerbated mortality risk in “normal” BMI patients and “protective” factors in overweight and obese subjects, or contributions from both [21, 22]? The phenomenon persists with adjustment for age, so that lower BMIs in older patients are not the major explanation.

There were no clear differences in “cause” of death by BMI category, with about 60% in each BMI category dying with cardiovascular conditions, comparable percentages dying with cancer (about 16%) and marginally fewer people of normal BMI (17.0%) dying with ESKF than those of higher BMIs (23%). The association persisted after adjustment for all the common co-morbidities that were independently associated with mortality risk.

Some have speculated that overweight or obese CKD patients may have better and more frequent care than others, and have adopted a healthier lifestyle. However, the standard settings and policies in these major public specialty CKD practices make it unlikely that normal weight CKD patients are relatively neglected in these domains of health care. Similarly, better nutritional status in those CKD patients with higher BMI may conceivably be a protective factor for mortality. However, no data of nutritional status were collected in this cohort. Others propose that CKD patients of higher BMI may better tolerate the side effects of some interventions [23, 24]. We note, too, that the patients’ BMIs in our report were ascertained more than 3.5 years prior to death on average, and might have changed substantially since first measured. Weight loss is common with progression of kidney diseases and BMIs close to death might have been substantially lower than documented on recruitment into the CKD.QLD cohort [23]. It has also been argued that, because CKD patients are more likely to develop fluid retention, BMI might not be an ideal measure for examining the risk in CKD [23], and using both BMI and waist circumference (WC), which reflects visceral adiposity, would better represent the associations between muscle mass, adipose tissue and various causes of death in CKD [21, 25‐29]. Some of these factors seem unlikely to be major contributors to this phenomenon. There are undoubtedly also unrevealed confounders affecting this association. Given that this “reverse epidemiology” applies in many settings, groups of patients and conditions, CKD-specific explanations can have only limited applicability. An intuitive explanation might be that, in a population where most adults are overweight or obese, the presence of lower or “normal BMIs” can flag significant health disadvantage or frank illness for a substantial proportion of people. Our exclusion of patients with BMIs < 18.5 in this analysis would not entirely mitigate against this influence.

Major strengths of this study include the large cohort of well-studied CKD patients, followed up for 5 years, the availability of data on all-cause mortality without RRT and on starting RRT, and ability to adjust for some important comorbidities including cancer, diabetes, PVD, chronic lung disease, CAD and all other CVD, which are potentially related to the risk of mortality. Their continued follow up will expose more findings over time, as numbers of death and RRT starts increase. However, it should be emphasised that these CKD patients do not represent the general community-based CKD population, as this cohort is selected for severity by being already referred to renal specialty practices. In addition some other confounders that may affect the associations were not included in our analyses.

In summary, more than 80% of these CKD patients were overweight or obese. Higher BMI seemed to be a significant “protective” factor against death without RRT in males, but there was not a significant relationship in females. Higher BMI was not a risk factor for predicting RRT in either males or female CKD patients.