Chronic kidney disease epidemic: myth and reality
- First Online:
- Cite this article as:
- Mangione, F. & Dal Canton, A. Intern Emerg Med (2011) 6: 69. doi:10.1007/s11739-011-0686-4
- 168 Views
In recent years, an epidemic of chronic kidney disease (CKD) has emerged as one of the major public health problem. The prevalence of CKD is largely sustained by the inclusion of a substantial proportion of the elderly population within stage 3 CKD, according to the Kidney Disease Outcomes Quality Initiative staging system. However, some clarifications are necessary when interpreting these data. In fact, renal function “normally” declines with age, without bearing any unfavourable outcome; in addition, the Modification of Diet in Renal Disease formula used to calculate glomerular filtration rate (GFR) underestimates kidney function in the elderly and in women. Considerable interest in CKD has been generated by the evidence that predialysis CKD is associated with the increased risk of cardiovascular disease (CVD). Again, potential confounding factors must be ruled out. Age is thought to play a major role in this context. The most common causes of CKD, hypertension and diabetes mellitus, are also known to affect cardiovascular outcomes directly, thus preventing the recognition of an independent effect of kidney dysfunction on mortality by CVD. Taken together, these considerations point for a better definition of CKD. Early identification of patients at risk for accelerated decline in renal function is mandatory to plan strategies for screening and preventing CKD and its complications. At present, detection of CKD in the general population requires a multi-dimensional approach that should include the evaluation of clinical risk conditions, evaluation of albuminuria and sequential monitoring of GFR.
KeywordsChronic kidney diseaseEpidemicGFR estimationCardiovascular diseasesCardio-renal syndrome
“Medical science has made such tremendous progress that there is hardly a healthy human left.”—Aldous Huxley (1894–1963).
Chronic kidney disease: definition and staging
pathologic abnormalities: evidence of primary or secondary glomerular, tubule-interstitial or vascular diseases of the kidney; or
abnormalities in the composition of urine: presence of albuminuria (defined as urinary albumin excretion >30 mg/day or 30 mg/g of urinary creatinine in a spot urine sample) or other anomalies in the urinary sediment; or
abnormal kidney imaging studies.
Classification and staging of chronic kidney disease (CKD) according to the K/DOQI guidelines 
GFR (ml/min/1.73 m2)
Stage of CKD
With signs of kidney damage
Without signs of kidney damage
This staging system has soon gained widespread popularity, mostly because of its simplicity. Albuminuria may be detected on timed urine collection or on spot urine sample, as a ratio to urinary creatinine. Rather than rely on direct measurement of GFR, that would need renal scintigraphy or, at best, timed urine collection to calculate creatinine clearance, K/DOQI guidelines call for the use of an estimation of the glomerular filtration rate (eGFR) by the Modification of Diet in Renal Disease (MDRD) formula . MDRD formula was derived by stepwise regression to logarithmic GFR of predictors such as serum creatinine, age, race and sex, in a clinical trial that enrolled individuals with an already established CKD, most of which with reduced GFR. Of consequence, using this formula in a population that has different characteristic may bear some potential pitfalls (see below).
CKD epidemiology: the facts
Prevalence of CKD
Since its publication, the new classification system has been immediately and widely used to stage renal function in heterogeneous populations, with the great part of the individuals consisting of community-dwelling, healthy people. Surprisingly, tremendously high prevalence of CKD has been found, generating what many authors called a “CKD epidemic” [4, 5]. CKD epidemic has been claimed as a major public health issue that may substantially challenge global health-care resources . Not surprisingly, the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) has dedicated an entire volume of its Annual Data Report to pre-dialysis CKD for the first time in 2008 . In one of the largest community-based cross-sectional study, the National Health and Nutrition Examination Surveys (NHANES), overall incidence of pre-dialysis CKD (from stage 1 to stage 4) in adults (20 years or older) was 13.1%. When related to the entire US population, this would return an estimation of about 26 million adult people affected by CKD . Similar prevalence has been found in other surveys, although with slight geographic differences. In Italy, two analyses have been conducted in recent years to estimate the prevalence of CKD. The Gubbio study  investigated the frequency of CKD stages 3–5 in a limited, rural region of Central Italy. Prevalence was about 6% in the study population, whose characteristics were not well balanced upon Italian demography, however. This would give a total estimation of about 3 million individuals with eGFR <60 ml/min per 1.73 m2 in Italy. Another survey, promoted by the Italian Society of Nephrology (SIN) in cooperation with the Italian Society of General Medicine (SIMG), aimed to estimate the prevalence of low eGFR in the general population relying on data collected via computer database in use to family doctors . This analysis reported an unexpectedly high prevalence of CKD stages 3–5, which was 9.33%. Again, these results are hardly transferable to the general population because the frequency is certainly over-estimated, since serum creatinine was tested in a minority of patients, almost only in those who had known abnormality in renal function, and/or hypertension, diabetes or other high-risk condition.
Distribution of CKD by stage
In each population survey conducted so far, the largest part of CKD was attributable to stage 3. In the NHANES survey, crude prevalence of stage 3 CKD was 7.7%, accounting for 60% of the total cases of CKD . Similar data have been shown by another cross-sectional study carried out in Norway (HUNT study) . In the last NHANES survey, stage 4 CKD had a prevalence of 0.35%, 22 times lower than that of the stage 3 CKD. ESRD has the lowest prevalence, ranging from 0.07% in Italy (data from the Report of the Italian Society of Nephrology—Italian Registry of Dialysis and Transplantation, presented at the Congress of the Italian Society of Nephrology in Rimini, October 2010) to 0.11% in the US .
Distribution of CKD by age
Prevalence of CKD increases with age. This age-dependent effect is true for each stage of CKD from 1 to 5; however, the relationship is particularly strong for stage 3, whose prevalence reaches 14% in people aged 60–69 years and exceed 35% in people age 70 years or older, being lower than 5% in people between 40 and 59 years of age and lower than 2% in people younger than 40 years .
Distribution of CKD by sex
Although it is well known that the frequency of traditional risk factors for cardio-renal diseases (smoke, hypertension, diabetes, dyslipidaemia) is lower in females than in males, at least in the pre-menopausal period, the prevalence of CKD appears higher in females than in males. Pre-dialysis CKD was 33% more prevalent in females than in males in the NHANES population . Similarly, in the SIN-SIMG study, prevalence was twofold higher in women . In a Norwegian longitudinal study of 38,241 individuals 7.9% had stage 3 CKD and, among these, 70% were woman; 50% of them were 70 years or older . This circumstance is inconsistent with the observation made on individuals with ESRD on dialysis. In fact, prevalence and incidence of ESRD are about 30% higher in males than in females , while no substantial differences in rate of progression of renal disease between genders have been noted so far [12–14].
Changes in CKD epidemiology
Changes in the epidemiology of CKD over the last years have been indirectly addressed by a comparison between two consecutive NHANES surveys (1988–1994 cohort vs. 1999–2004 cohort) . Overall prevalence of CKD stages 1–4 increased over years from 10 to 13.1%; while subjects with CKD stages 2–4 grew, no statistically significant difference was detected for stage 1 between the two cohorts. While higher prevalence of diagnosed diabetes, hypertension and higher body mass index explained entirely the increase in prevalence of microalbuminuria (CKD stage 2), this was not true for stage 3, in which the increase in mean age of the population was the key factor that justified the higher proportion of patients with low GFR.
Interpreting CKD epidemic
CKD epidemic is established on a high prevalence of CKD stage 3 (moderately decreased GFR) among people aged 60 or older, a proportion of individuals that is increasingly represented in the population of the Western world. In fact, in the US 2000 census, about 16% of the population was 60 years old or over ; in the European Union 22.7% of the population was 60 years or over in 2000 . In 50 years, people older than 65 years will be over 50% of the entire population, according to projections. The need for sound evidences regarding the epidemiology of CKD is stringent, to plan adequate health policy strategies and optimize the allocation of social and economic resources. Thus, it is urgent to recognize potential bias in the definition of CKD that may have affected epidemiological data.
Does GFR decline with age?
Since CKD stage 3 is predominant in elderly, it is mandatory to rule out a direct effect of age on kidney function. It has been well known for more than 60 years that renal function does decline with age . Defining a limit between physiological and pathological decline is, however, not simple. In theory, the limit between normal and abnormal renal function would correspond to the GFR level at which kidneys lose the ability to maintain body homeostasis, or at which complications of “insufficient” renal function (such as anaemia, metabolic acidosis, secondary hyperparathyroidism and so on) begin to appear. This criterion is hardly applicable in clinical practice, because kidneys can maintain body homeostasis in a wide range of glomerular function, and different complications may appear at different levels of GFR, with also an inter-individual variability. In a recent study, our group found a mean GFR (as inulin clearance) of 84.3 ml/min per 1.73 m2 in a cohort of healthy elderly people; the lower level of “normality”, defined as the mean − 2SD, was calculated at 52 ml/min per 1.73 m2. We concluded that in healthy aged people, normal levels of renal function include GFRs below 60 ml/min per 1.73 m2 . Moreover, mild renal dysfunction of the elderly seems to be non-progressive. In fact, some authors found that elderly with stage 3 CKD were somehow protected from the development of ESRD after 10 years of follow-up. GFR declined by 1.04 ml/min per 1.73 m2 per year over 70 years of age, a rate much lower than observed in the MDRD study (−3.5 ml/min per 1.73 m2 per year) . Although high mortality rate may have prevented these patients from developing severe renal damage, this hypothesis did not entirely justify the result. In elderly, a GFR that is slightly lower than in young-adult individuals may be completely sufficient, due to the age-associated changes in muscle mass, metabolism, protein and caloric intake. Unlike patients with established kidney disease, decline of GFR is slowly progressive in elderly, arguing for physiologic, age-related decrease in renal function rather than CKD. Punctual estimation of kidney function remains mandatory in elderly, to adapt dosages of drugs with renal excretion; but labelling older individuals with mildly reduced GFR and without other signs of renal damage as affected by CKD should be considered incautious. Furthermore, outcome studies have shown that no excess of mortality is associated with mildly reduced GFR (as defined by K/DOQI guidelines) in elderly (see below).
Estimating GFR: potential pitfalls
GFR estimation is currently based on the formulas that derive creatinine generation (and therefore its clearance) from muscle mass by demographic characteristics like age, sex and race. Cockcroft and Gault , and Modification of Diet in Renal Disease  formulas are the most largely used to this scope. Both formulas are based on the assumption that creatinine generation decreases with age and is lower in females, but do not account for other factors that could affect muscle mass (and thus creatinine generation), such as meal intake, muscle exercise or CKD itself. 4-variable MDRD formula is recommended by K/DOQI guidelines, because it is considered more reliable than the simpler Cockcroft–Gault formula. As already mentioned, MDRD formula was derived in a cohort of patients with reduced GFR ; in such a population, muscle mass and creatinine generation are reduced if compared to non-CKD, age- and gender-matched individuals , suggesting that the relationship between demographics and GFR may be different between the individuals with CKD and without CKD. The use of demographic parameters as predictors of endogenous creatinine generation in elderly has been recently criticized . The use of formulas that incorporate direct measurement of muscle mass by dual energy X-ray absorptiometry (DEXA) for estimating creatinine clearance showed that traditional equations that use demographic variables largely overestimate the age-related decline in creatinine clearance and increase the likelihood of being diagnosed with CKD. As a consequence, using muscle mass measurement, rather than age as a predictor of endogenous creatinine generation, halved (from 25 to 13%) the prevalence of stage 3 CKD in the study population .
In the MDRD study, subjects older than 70 years were excluded as not eligible , and even if MDRD formula has been subsequently validated in elderly population, caution should be spent in interpreting results from an equation that has been derived in a completely different cohort of subjects. All these concerns regarding the reliability of MDRD-based estimation of GFR, in particular subset of individuals have been confirmed by a large population trial, the Nijmengen Biomedical Study . Serum creatinine was measured in 3732 community-dwelling, disease-free adults aged 18–90 years and eGFR was estimated according to MDRD. After stratifying the sample by age, and calculating a reference range by adding and subtracting two standard deviations to the mean GFR value in each layer, authors found that the lower limit for normality fell into K/DOQI stage 3 CKD for men over 35 years of age and for women at virtually every age. All these studies give evidence for a strong tendency to over-diagnosis of CKD due to bias in GFR estimation, which is particularly prominent in elderly and women.
Outcomes of CKD: cardiovascular diseases
It has already been underlined that there is a substantial disproportion between prevalence of stage 3 CKD and the one of more severe CKD, in particular ESRD. Many authors have suggested that the majority of individuals with moderately decreased GFR would die before they reach ESRD. This could easily be the case, but then the question would be: do they die because they have a reduced GFR, or rather because they are old and/or have other morbid conditions that can also bear CKD?
Age, CKD and cardiovascular disease: a complex conundrum
It is trivial to note that age is one of the most powerful predictors for all-cause mortality, and for cardiovascular mortality in particular . Irrespectively of age, gender or comorbidities, cardiovascular (CV) mortality in patients on maintenance dialysis is 10 to 30-fold higher than in general population . In 2004 evidence emerged that even mild degrees of renal dysfunction were statistically associated with the increased risk of morbidity and mortality for cardiovascular disease (CVD) . By inference, a causal relationship between CKD and CVD was determined; however, potential confounding factors like senescence and comorbidities (hypertension, diabetes and other traditional risk condition) were not convincingly ruled out. In their seminal paper, Go and colleagues  recorded rates of death, cardiovascular events and hospitalization in a cohort of over 1 million Americans, stratified according to their basal MDRD-estimated GFR. They found a direct relationship between decreasing eGFR and CV outcomes. CV morbidity and mortality were significantly increased in individuals whose eGFR was lower than 45 ml/min per 1.73 m2, while patients with an eGFR between 59 and 45 ml/min per 1.73 m2 experienced only a marginal increase in risk. Even if these outcomes were adjusted for age, individuals with moderate or severe renal dysfunction (stage 3 and 4 CKD) were more likely to be older and had higher frequency of previous CV events or comorbidities, arguing for a potential interference by these factors.
The effect of age on CV in patients with reduced GFR has been directly tested in two successive surveys. The first was conducted on 2,598,548 individuals from the US Veteran Affairs database. It was demonstrated that, in individuals with mildly decreased GFR (50–59 ml/min per 1.73 m2) of age 64 years or older, risk for mortality was unchanged when compared with individuals with GFR >60 ml/min per 1.73 m2. When older subjects were considered (85 years or more), no increase in mortality was seen in individuals with moderately reduced GFR (40–49 ml/min per 1.73 m2) in comparison to people with normal kidney function . The same results have been more recently confirmed in a British population. Both men and women 75 years of age or older with an eGFR between 45 and 59 ml/min per 1.73 m2 showed no differences in all-cause and cardiovascular mortality when compared with individuals with an eGFR >60 ml/min per 1.73 m2 . More than 40% of people with eGFR <60 ml/min was represented by individuals 75 years or older, in which no additional risk was detected due to low GFR. Taken together, these findings suggest that eGFRs of 45–59 ml/min mostly include individuals with age-related, non-progressive decline of kidney function rather than with real kidney disease. Only the latter condition, characterized by lower mean eGFRs, is associated with unfavourable cardiovascular outcomes. Not surprisingly, these evidences have been incorporated into a further classification of CKD that divides stage 3 into stage 3A (45–59 ml/min per 1.73 m2, the one associated with low CV risk) and 3B (30–44 ml/min per 1.73 m2, at a moderately increased CV risk).
Association between CKD and CVD: other potential confounding factors
Under the assumption that CKD, in its more advanced stages, is a risk factor for CVD, the strength of association still needs to be addressed. Because both CKD and CVD are highly prevalent in elderly, it is tempting to speculate that one (or more) age-related unifying factor may exist. Furthermore, the aetiology of renal diseases can affect CV outcomes by itself, as well as the two major risk factors for CKD and ESRD, diabetes mellitus and hypertension. Notably, the frequency of both conditions increases with age. Outcome studies in CKD conducted so far [25, 28, 29] included both hypertensive and diabetic patients. Statistical adjustments were performed to test the independence of the CV outcomes from comorbidities; these latter usually have been translated into dichotomous categorical variables (presence/absence of the comorbidity) for the purpose. Although this methodology is probably the unique fitting at the scope, it does not properly reflect biological reality. In fact, diabetes and hypertension are not merely categorical variables. It is well known that different degrees of glycaemic control bring to very different renal and CV outcomes in diabetic patients, independently of the presence of renal disease at baseline . Patients with CKD secondary to diabetic nephropathy usually have worse glycaemic control in their medical history than diabetic patients without renal dysfunction. In the same patients, the inveterate bad glycaemic control fully legitimizes the increased risk for CV morbidity and mortality. Atherosclerosis is the pathophysiologic link between diabetes and its complications, both CKD and CVD. The same remarks are valid for hypertension, whose prevalence was even higher than the one of diabetes in the cited studies. It is well known that even small (1–2 mmHg) differences in systolic or diastolic blood pressure can account for up to 10–15% variation in incidence of cardiovascular or cerebrovascular events . Actually, it is not possible to ascertain an independent role of CKD as a cause of CVD, excluding the effect of hypertension and diabetes, that have high prevalence in both disorders and that have been already related to them in a cause-effect connection.
Future perspectives on CKD epidemiology
Many authors have already claimed a refinement of the K/DOQI classification and staging system [32, 33]. It is growingly evident that the definition of CKD should not rely only on a single point measurement of eGFR, but rather on serial determinations, to separate individuals that have non-progressive, age-related decline in GFR, that would benefit of standard care; from subjects that have progressive kidney disease that may require strict surveillance by nephrologists.
Current perception that CKD has reached epidemic proportions largely depends on an extraordinarily high prevalence of stage 3 CKD, as a consequence of the detection of low levels of eGFR in elderly individuals within the general population. However, labelling many elderly people as being affected by kidney disease is misleading, because kidney function declines physiologically with age without bearing any excess in morbidity or mortality. This distorted perspective is emphasized by the fact that the use of the MDRD formula to calculate GFR underestimates renal function in elderly and women, inflating prevalence of CKD stage 3. Early identification of patients at risk for accelerated decline in renal function is mandatory to plan strategies for screening and preventing CKD and its complications. At present, detection of CKD in the general population requires a multi-dimensional approach that should include the evaluation of clinical risk conditions (e.g., age, diabetes, hypertension, obesity, chronic obstructive uropathy, recurrent urinary tract infections), evaluation of albuminuria and sequential monitoring of GFR.
Conflict of interest