CKD Prognosis: Beyond the Traditional Outcomes

In this issue of the American Journal of Kidney Diseases, the article by James and colleagues titled “CKD and Risk of Hospitalization and Death With Pneumonia”1 provides an excellent opportunity to examine our preconceptions about the outcomes we consider in evaluating the prognosis of chronic kidney disease (CKD). The article provides a powerful demonstration of the strong associations between decreased kidney function (estimated glomerular filtration rate [GFR]) and risk of hospitalization with pneumonia and subsequent mortality. The data support pneumonia as an important CKD outcome since it is common, shows a strong association with CKD, is preventable by vaccination, and is treatable. James and colleagues cite previous reports indicating that fewer than 50% of patients with CKD receive influenza vaccination and less than 10% receive pneumococcal vaccination before initiation of renal replacement therapy, making it clear that the level of attention given to pneumonia prevention is not high enough. Even among solid organ transplant recipients enrolled in health maintenance organizations, influenza vaccination coverage was only 52% from 1995 to 2005.2

The number of CKD patients with elevated risk of pneumonia is large and spans a wide range of kidney function, suggesting that the implications for increased vigilance to prevent pneumonia and decrease mortality after its occurrence apply well beyond the nephrology community. Some have suggested referral to nephrologists of all patients with estimated GFR less than 30 mL/min/1.73 m2 (0.5 mL/s/1.73 m2, stage 4) and patients with albuminuria, progressive disease, or uncontrolled risk factors (approximately 19% of CKD stage 3).3 The article by James and colleagues suggests an increased risk among a broader group of CKD patients. Unfortunately, all too often CKD patients receive less rather than more preventative care than lower-risk populations. The standard trivalent influenza vaccine results in seroprotection among more than 80% of dialysis patients,4 suggesting the challenge is to improve coverage. Pneumococcal vaccine provides an additional challenge since its efficacy among high-risk groups is controversial.5 CKD patients would benefit from research to determine efficacy in CKD and guide informed action.

The authors have done an impressive job of showing both absolute and relative risk. The study population of 252,516 participants with 4,253 first hospitalizations with pneumonia is large enough to allow detailed stratification by both age and estimated GFR. The main modeling strategy focuses on relative risks and appropriately points out that the relative risks decrease with age. However, the authors also point out that the absolute risk of hospitalization with pneumonia increases dramatically with age. They conclude both that “[t]he increased relative risk of pneumonia at lower eGFRs was more pronounced in younger patients” and that “[n]onetheless, the absolute increases in risk with decreased eGFR were greatest in older patients, and thus interventions to prevent and treat pneumonia would be expected to have their greatest impact at a population level in this segment of the population.” While we already have recommendations to vaccinate all older adults against pneumococcus and influenza, these data are an impetus to make sure the recommendations are fully implemented in CKD. The counterbalancing trends of higher absolute risk but lower relative risk in older adults are seen for most age-related conditions. They complicate interpretation, but are worthy of attention each time, since large relative risks tend to be more impressive and easier to communicate, but moderate relative risks in high-risk subgroups often lead to the highest attributable risk and potential for prevention.6

This study of pneumonia risk is observational and hence needs to be interpreted with care. It has a number of strengths, including the very large population size and high sensitivity and specificity of pneumonia identification using administrative data. Use of a single laboratory with information on creatinine calibration is a rare situation with studies of this size, where often the use of multiple laboratories and uncertain calibration over time limits the interpretation of the absolute level of GFR estimates. The authors also conduct a number of useful sensitivity analyses, assuring us that the finding is not dependent on whether baseline or time-varying measures of estimated GFR are used, or whether emergency room visits for pneumonia which do not lead to admission are included or excluded. The authors do not lump all participants with an estimated GFR greater than 60 mL/min/1.73 m2 (1.0 mL/s/1.73 m2) into a single reference group. Instead, they find that pneumonia, like many other conditions, has an increased risk among participants with a very low serum creatinine and consequently high estimated GFR. This paper uses greater than 105 mL/min/1.73 m2 (1.75 mL/s/1.73 m2) to explore the “U shape” of risk associated with high estimated GFR and finds substantially elevated risk in all age groups. Separating this group from the reference group (60 to 104 mL/min/1.73 m2 [1.0 to 1.73 mL/s/1.73 m2]) is important to avoid inappropriately raising the baseline risk of this reference group, which then obscures elevations in risk at moderately decreased GFR, below 60 mL/min/1.73 m2 (1.0 mL/s/1.73 m2). A continuous analysis would shed even more light on this issue and the possibility that among the oldest group the risk of pneumonia is higher in the upper part of the reference group. This might change some of the inferences about the inflection point where pneumonia risk begins rising with lower eGFR among patients older than 75 years. The authors appropriately point out the limitation that despite their adjustment for comorbid conditions, residual confounding may remain but is unlikely to explain the very strong associations observed. To what extent CKD marks or causes an elevated pneumonia risk cannot be determined with observational data. However, for some interventions, such as pneumonia vaccinations, a marker of increased risk is sufficient to make intervention more cost-effective and warranted. Better data on programs seeking to achieve higher compliance with recommendations, and on the cost-effectiveness of different programs, including those seeking to decrease pneumonia case fatality in CKD, would be helpful.

As pointed out by the authors, the outcome which receives much of our attention in thinking about the prognosis of CKD has been cardiovascular disease (CVD) risk. This is appropriate since CVD is the leading cause of mortality and a frequent contributor to morbidity. The other natural outcomes of CKD to examine are the kidney outcomes, with kidney failure treated by dialysis and transplantation (end-stage renal disease) often being the most studied. However, the range of outcomes associated with CKD is much broader. Kidney failure leading to death without treatment is important, though difficult to study. In a recent report from Norway, Hallan et al7 found that after 10.3 years of follow up of 65,589 patients, 58 were treated by dialysis or transplantation and 66 others died of advanced CKD (ie, a documented CKD diagnosis and a documented stable estimated GFR less than 15 mL/min/1.73 m2 [0.25 mL/s/1.73 m2] or an indication for renal replacement therapy before death) without dialysis or transplantation. Thus, even with national insurance, the number of untreated kidney failure cases may outnumber the number of treated cases. Hallan et al also document the risk of kidney failure by both estimated GFR and albuminuria. Together, the 2 factors account for an over 4,000-fold risk gradient, after adjustment for demographics and other risk factors. This article provides a useful example of how estimated GFR, albuminuria, and other risk factors can be combined in risk prediction.

In summary, systematic study of CKD prognosis should include a broad range of outcomes, with likely implications beyond the traditional outcomes and patients seen by nephrologists. Table 1 summarizes some of the key outcomes. Development of the CKD stages focused on concurrent complications, since data on these was more readily available and treatment of severe CKD often focuses on managing concurrent complications.8 As we focus on earlier stages of CKD, the importance of understanding and minimizing the risk of future complications increases. The table highlights the importance of considering both estimated GFR and albuminuria, as well as the other strong risk factors, for each outcome. For some outcomes, such as kidney failure, it may be that albuminuria and estimated GFR may capture most of the risk information, while for other outcomes other risk factors may be more important (ie, history of CVD and heart failure for CVD risk). The first challenge will be to document risk rigorously for each outcome, taking into account absolute risk as well as interactions with age and other factors. Next, is the need to develop, test, and implement strategies for optimal treatment of concurrent complications and reduction of future risk tailored to the wide spectrum of CKD.

Table 1. Prognosis of CKD: Range of Outcomes and Risk Factor Associations

	Outcome	Risk Factors for Different Outcomes
		eGFR	Albuminuria	Other Major Risk Factors
	Concurrent complications⁎	++	?	Vary for different complications, although most increase with older age
	Prognosis (10-year risk)
	CVD or mortality	++	++	History of CVD (absent, subclinical, clinical) Major CVD risk factors (blood pressure, cholesterol, diabetes, age, sex, race)
	Kidney failure	+++	++	Type of kidney disease†, blood pressure, male sex, race, age (older age decreases risk of starting renal replacement therapy)
	Rate of decline in GFR	+	+++	Type of kidney disease†, blood pressure, race
	Acute kidney injury, drug toxicity	+++	+++?	Older age, medication use
	Hospitalizations for pneumonia	+++	?	Older age, smoking, COPD, no vaccination

Note: The number of + signs indicates the strength of association. Abbreviations: CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; CVD, cardiovascular disease; COPD, chronic obstructive pulmonary disease.

⁎ Hypertension, anemia, malnutrition, bone disease, neuropathy, decreased quality of life, electrolyte and acid base abnormalities. † Examples are diabetic, glomerular, vascular, tubulointerstitial, and cystic.