Comorbidites in patients with chronic kidney disease – Masao Iwagami

27 Sep 2016

In August 2016, our monthly Journal Club featured a Nephrology article entitled “Comorbidity as a driver of adverse outcomes in people with chronic kidney disease” (M Tonelli et al., Kidney Int., 2015 [88], 859-66).Iwagami_Masao-1

Chronic kidney disease (CKD), defined as kidney damage (albuminuria or proteinuria) or as decreased kidney function (estimated glomerular filtration rate [eGFR] <60 mL/min/1.73m2), is a common condition in the community. It affects around 10% of the general population. It is almost impossible to dissociate CKD from other health conditions because there are a number of CKD risk factors (e.g. diabetes and hypertension) and CKD is associated with several outcomes (e.g. heart failure, myocardial infarction, and stroke). Therefore, it may sound natural that a patient with CKD has multiple morbidities. Yet very few studies have formally assessed the burden of multiple morbidities in patients with CKD. Also, as the authors of the study suggested, kidney researchers have tended to focus on ‘concordant’ comorbidities and have paid less attention to ‘discordant’ and ‘mental-health/pain’ comorbidities. The concepts of ‘concordant’ and ‘discordant’ comorbidities were originally proposed by Piette et al. for the field of diabetes (Diabetes Care, 2006 [29], 725-31). These concepts can be applied to the kidney field. For example, ‘concordant’ CKD comorbidities include hypertension, diabetes, heart failure, myocardial infarction, and stroke, while ‘discordant’ CKD comorbidities include cancer, COPD, and inflammatory bowel disease.

Using a Canadian provincial database from Alberta, the authors examined the association between multiple morbidities (29 comorbidities, classified into ‘concordant’, ‘discordant’, and ‘mental-health/pain’ subgroups) and several outcomes (mortality, number of hospitalizations, hospital length of stay, and acute myocardial infarction [AMI]) in patients with CKD. Their current work builds on their two previous groundwork articles. Their first paper established a CKD cohort using serum creatinine and urinalysis records (BMC Nephrol., 2009 [10], 30). Their second paper defined 30 chronic conditions in the database (BMC Med Inform Decis Mak., 2015 [15], 31). Their groundwork might be far from perfect. For example, 73% of the patients in the CKD cohort received only one serum creatinine measurement before cohort entry, so patients with acute kidney injury (AKI) might have been inadvertently included. Their “validated” algorithms to identify 30 chronic conditions included those examined outside of Canada (e.g. in the US), which might not be applicable in Alberta. Having said that, their groundwork was greatly appreciated as a start to the current study.

Using their CKD cohort, the authors counted the number of comorbidities for each patient. They examined the most frequent pattern or combination of comorbidities, calculated the population attributable risk (PAR) for each of the 29 morbidities (although we could raise a question about whether or not the assumptions behind the PAR calculations are appropriate), and identified a list of comorbidities for which the PAR of death or hospitalization was greater than the median. Finally, they examined the associations between the number of comorbidities and outcomes by differentiating the ‘concordant’, ‘discordant’, and ‘mental-health/pain’ categories. As a result, the authors demonstrated that a substantial proportion (25%) of people with CKD had three or more comorbiditiesand a smaller but still substantial proportion (7%) had five or more comorbidities. The authors further determined that both concordant and discordant comorbidity were exceedingly common among patients with CKD—and so were mental health conditions and chronic pain. There was a strong, graded, and independent relation between all three of these forms of comorbidity and the risk of death and hospitalization. The authors’ key message is: Given that most previous studies have focused on the importance of concordant comorbidity, this suggests an urgent need to refocus the research agenda for people with CKD on the management of discordant comorbidities as well as mental health and chronic pain.

We could criticize that this phenomenon might not be specific to the CKD population. In any study population (e.g. patients with diabetes, or even patients without CKD), cancer, COPD, dementia, depression, and pain are expected to be associated with patient prognoses. However, it is also true that kidney researchers have been inclined to focus on ‘concordant’ comorbidities, so that this paper reminds us of the importance of ‘discordant’ and ‘mental-health/pain’ comorbidities in our clinical practice and research. This paper also gave us a good opportunity to discuss how to take multiple morbidities into account in our research. We sometimes adjust for consultation frequency as a proxy for multiple morbidities, but this may not be always appropriate. Of course, the answer would be dependent on our research question, or exposure and outcome of the study. Still, the appropriate understanding and handling of multi-morbidity is becoming ever more important in this era of an aging society.

Looking forward to the next monthly Journal Club.

Masao Iwagami