Title: Unlocking the value of variation in chronic kidney disease prevalence
C Black1,3, SN van der Veer 2, 3
1
Institute of Applied Health Science, University of Aberdeen
2
Centre for Health Informatics, University of Manchester
3
Kidney Disease@Farr, UK Farr Institute of Health Informatics Research
Running title: Value in variation in CKD
Contact Author:
Professor Corri Black Institute of Applied Health Science, School of Medicine, Medical Science and
Nutrition, University of Aberdeen, Foresterhill, Aberdeen AB25 2ZD
+44 (0) 1224 437957
corri.black@abdn.ac.uk
Darwin, publishing his seminal work ‘On the Origin of Species’ in 1859, presented his then
revolutionary recognition of the importance of describing variation and, by studying variation, being
able to identify potentially important mechanisms of change.1
"I have hitherto sometimes spoken as if the variations [...] had been due to chance. This, of course, is
a wholly incorrect expression, but it serves to acknowledge plainly our ignorance of the cause of each
particular variation." Charles Darwin 18591
Prior to the adoption of a common definition for chronic kidney disease (CKD), the substantial
variation in prevalence reported between studies was largely attributed to differences in
methodology. The lack of a consistent definition distorted comparisons of prevalence between
countries, with important differences being lost in methodological ‘noise’.2 Since the introduction, in
2002, of the Kidney Disease Improving Global Outcomes (KDIGO) definition for CKD, there has been
a global recognition that CKD is common.3 What remains challenging, however, is to determine
precisely how common.
In 2013, the Global Burden of Disease Study ranked CKD 19th on the list of global causes of death.4
The associated risks of end stage kidney disease, cardiovascular morbidity, fractures, cognitive
impairment and anaemia mark CKD as an important global health problem.5 Estimates from the US
population, based on national health surveys, reported a prevalence of CKD of 13.1% in 1999-2004.6
The aging of the population in combination with an epidemic-like increase in diabetes mellitus,
obesity and hypertension, is anticipated to further drive substantial rises in CKD prevalence in the
coming decades.5
In this issue of JASN, Brück and colleagues report CKD prevalence across thirteen European
countries.7 They leveraged the power of nineteen studies designed to be representative of the adult
general population; ten restricted inclusion to subjects aged ≥65 years. The total study sample
consisted of over 225,000 people. The prevalence of risk factors varied substantially between
included studies. For example, the percentage of people ≥65 years with diabetes mellitus ranged
from 6.9% in Groningen (the Netherlands) to 28.2% in South-West Germany.
The authors performed analyses on individual patient-level data, which optimised comparability of
prevalence estimates between studies. It enabled application of a single definition for CKD based on
the same equation to estimate the glomerular filtration rate; the majority of studies also used the
same assays for measuring serum creatinine (Jaffe) and urinary albumin (immunoassays). They
further increased comparability by adjusting overall prevalence estimates to a standard European
age and gender distribution, and by repeating their analyses stratified for age and the main CKD risk
factors: hypertension, diabetes and obesity.
Brück's study suggests substantial geographical variation in CKD prevalence across Europe. For CKD
stage 1-5, they report overall adjusted prevalence ranging between 3.3% in Norway to over 17% in
North-East Germany. For stage 3-5, CKD varied between 1.0% in Italy and 5.9% in North-East
Germany. Stratifying the analyses for age and risk factors showed similar patterns.
Another recent publication reported the prevalence of CKD globally.8 Mills and colleagues included
33 original population-based studies, totalling over 920,000 people, of whom 81% were from highincome countries. They analysed study-level (and not patient-level) data to provide pooled estimates
of CKD prevalence for high-income and low/middle-income countries separately, while stratifying by
age and gender. Seven cohorts were included by both Mills and Brück, accounting for 39% of Brück's
total study population. Using the same definition for CKD but a different, standard global population
for age and gender standardization, Mills et al. found an overall age-adjusted CKD stage 1-5
prevalence in high-income countries that was in line with Brück's: 10.1% in men, and 12.1%, in
women; for stage 3-5, it was somewhat higher (5.4% and 8.6% for men and women, respectively).
Mills et al. convincingly show that CKD is a common public health problem worldwide, especially in
low- and middle-income countries. The importance of their findings notwithstanding, they adopted a
simplification approach by pooling of data into two large groups. The countries within each group
may share their level of gross national income, but they are likely to have different healthcare
systems and public health policies, which in turn may affect CKD prevalence. Yet, Mills'
methodological approach left any meaningful between-country differences unrevealed.
Furthermore, interpretation of between-country variation would have been hampered, firstly
because only 60% of included studies used the same equation to estimate GFR, and secondly
because analyses were not adjusted for differences in CKD risk factors.
The Brück paper reminds us that there is important knowledge to be gained by understanding
variation better. In contrast to Mills et al., Brück aimed to minimize the methodological variation
between studies, which is a salient strength of their approach. As a result, their findings contribute
to our understanding of CKD prevalence by demonstrating that the observed variation was unlikely
to be driven by methodological issues alone, and that it could neither be explained by populationbased differences in age and gender distributions, nor risk factor prevalence.
So, how do we move forward? Implicit within the variation is the potential to unlock a better
understanding of the natural history of CKD and the opportunity for natural experiments. Brück et
al. express an ambitious vision building on their work; a vision for rigorous data collection across
Europe that can be used to evaluate the impact care strategies. In essence, they call for the
infrastructure to support efficient natural experiments. While randomised controlled trials (RCT) are
the gold standard for experimental design to evaluate comparative effectiveness, there has been a
long established use of observational epidemiology to tackle situations where RCTs are impractical.
Natural experiments seek to exploit variation; whether variation in exposure between individuals, or
between populations. All natural experiments have in common that the exposure or intervention
under study has not been manipulated by the researchers.9 As such, natural experiments have an
important role to play in situations where the exposure or intervention is difficult to study
experimentally for either practical or ethical reasons; for example determining the impact of
legislation or policy, understanding the effect of environmental exposures, or studying very rare or
long-term outcomes.9
Previous natural experiments have successfully exploited variation to understand the impact of
policy, practice or exposure to risk factors.10,11 For example, in 2006, Scotland implemented a law
prohibiting smoking in enclosed public places ahead of the other three UK nations. Researchers
utilised the ‘before and after’ variation in exposure, as well as a ‘difference in differences’ approach
harnessing variation in geographical implementation.10
In nephrology, natural experiments have been used to study the impact of population-level
interventions or health policies. For example, an experiment evaluating the effectiveness of
influenza vaccination in adults on haemodialysis exploited variation between years where the
vaccine was known to be a good match for the circulating strain with years when it was not.12 The
introduction of a ‘payment for performance’ initiative lent itself to a natural experiment using timeseries analysis to study variation in markers of care before and after a national incentive scheme was
rolled out.13
Bello et al. set out clear recommendations for policy and population-wide intervention to prevent
and promote best care for people with CKD.14 Variation between countries in the type of
interventions and the timing of their implementation will be inevitable. How do we then, create an
effective and efficient system to identify best practices? Brück et al. recognise that their work,
which establishes that variation exists, is a starting point. Importantly, they also recognise the scale
of the challenge, identifying that the ‘perfect’ natural experiment is unlikely to be achievable
without significant investment into multi-country population-wide surveillance systems with
centralised collection, testing and collation of data. The UK Medical Research Council
methodological framework for natural experiments reiterates this challenge and recommends that
such quantitative endeavours should only be undertaken: where variation in exposure can be
identified; using samples large enough to detect effect; and where accurate data can be obtained
about the exposure, outcomes and potential confounders.9
So, if natural experiments are the panacea to better understand kidney disease epidemiology, risk
factors, and the impact of interventions, the reality is that we need efficient and sustainable
solutions to capture detailed international, population-based data. High quality population health
survey approaches may be part of this solution. However, a complimentary and, as yet, underutilised
resource exists in the growing array of routinely collected health information. Although now still
mainly pertaining to the clinical arena, health data collection is moving towards a more coproduced
model, where patients and the general public are increasingly encouraged and facilitated to
contribute their health data for clinical and research purposes.15 Ainsworth et al. recently described
a vision for a learning health system. Such a system brings together data from multiple sources to
describe the health of a community, and enables distributed analysis of variation through a network
of similar data clusters.16
Perhaps such an exciting future for a learning kidney health community is on the horizon but,
whether through the investment in robust specialist surveillance systems or the effective harnessing
of routine health data, key will be the collection and curation of high quality data.
Conflicts of interest: none
References
1. Darwin C. On the Origin of the Species by Means of Natural Selection,
or the Preservation of Favoured Races in the Struggle for Life. John Murray, Albemarle
Street, London 1859
2. McCullough K, Sharma P, Ali T, Khan I, Smith WC, MacLeod A, Black C Measuring the
population burden of chronic kidney disease: a systematic literature review of the estimated
prevalence of impaired kidney function. Nephrol Dial Transplant 2012;27:1812-21
3. Levey, A S., KU Eckardt, Y Tsukamoto, A Levin, J Coresh, J Rossert, D de Zeeuw, TH. Hostetter,
N Lameire, and G Eknoyan. Definition and classification of chronic kidney disease: a position
statement from Kidney Disease: Improving Global Outcomes (KDIGO). Kidney
International 2005; 67: 2089-2398
4. GBD 2013 Mortality and Causes of Death Collaborators . Global, regional, and national age–
sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a
systematic analysis for the Global Burden of Disease Study 2013. Lancet 2015; 385: 117–71
5. Jha V, Garcia-Garcia G, Iseki K, Li Z, Naicker S, Plattner B, Saran R, Wang AYM, Yang CY
Chronic kidney disease: global dimension and perspectives . Lancet 2013; 382: 260–72
6. Coresh J Selvin E, Stevens LA Manzi J, Kusek JW, eggers P, van Lente F, Levey AS Prevalence
of Chronic Kidney Disease in the United States JAMA 2007; 298: 2038-2047
7. Brück K, Stel, V, Gambaro G et al. CKD prevalence varies across European general population.
JASN
8. Mills KT, Xu Y, Zhang W, Bundy JD, Chen CS, Kelly TN, Chen J, He J A systematic analysis of
worldwide population-based data on the global burden of chronic kidney disease in 2010.
Kidney Int. 2015; 88 :950-7
9. Peter Craig, Cyrus Cooper, David Gunnell, Sally Haw, Kenny Lawson, Sally Macintyre, David
Ogilvie, Mark Petticrew, Barney Reeves, Matt Sutton, Simon Thompson. Using natural
experiments to evaluate population health interventions: new Medical Research Council
guidance. J Epidemiol Community Health 2012; 66: 1182-1186
10. Pell JP, Haw S, Cobbe S, Newby DE, Pell ACH, Fischbacher C, McConnachie A, Pringle S,
Murdoch D, Dunn F, Oldroyd K, MacIntyre P, O’Rourke B, Borland W. Smoke-free
Legislation and Hospitalizations for Acute Coronary Syndrome. N Engl J Med 2008; 359: 48291
11. Urbach DR, Govindarajan A, Saskin R, Wilton A, Baxter NN Introcution of surgical safety
checklists in Ontario, Canada NEJM 2014; 370: 1029-1038
12. McGrath LJ, Kshirsagar AV, Cole SR, Wang L, Weber DJ, Stürmer T, Brookhart AM. Evaluating
influenza vaccine effectiveness among hemodialysis patients using a natural experiment.
Arch Intern Med. 2012; 172: 548–554
13. Karunaratne K, Stevens P, Irving J, Hobbs H, Kilbride H, Kingston R, Farmer C The impact of
pay for performance on the control of blood pressure in people with chronic kidney disease
stage 3–5. Nephrol Dial Transplant 2013; 28: 2107–2116
14. Bello AK, Levin A, Manns BJ, Feehally J, Drueke T, Faruque L, Hemmelgarn B, Kernahan C,
Mann J, Klarenbach S, Remuzzi G, Tonelli M, on behalf of the Kidney Health for Life Initiative
Effective CKD Care in European Countries: Challenges and Opportunities for Health Policy.
Am J Kidney Dis 2015; 65:15-25
15. Open Humans project. https://www.openhumans.org/. Accessed November 30, 2015
16. Ainsworth J, Buchan I. Combining Health Data Uses to Ignite Health System Learning.
Methods Inf Med. 2015; 54: 479-87
References
17. Darwin C. On the Origin of the Species by Means of Natural Selection,
or the Preservation of Favoured Races in the Struggle for Life. John Murray, Albemarle
Street, London 1859
18. McCullough K, Sharma P, Ali T, Khan I, Smith WC, MacLeod A, Black C Measuring the
population burden of chronic kidney disease: a systematic literature review of the estimated
prevalence of impaired kidney function. Nephrol Dial Transplant 2012;27:1812-21
19. Levey, A S., KU Eckardt, Y Tsukamoto, A Levin, J Coresh, J Rossert, D de Zeeuw, TH. Hostetter,
N Lameire, and G Eknoyan. Definition and classification of chronic kidney disease: a position
statement from Kidney Disease: Improving Global Outcomes (KDIGO). Kidney
International 2005; 67: 2089-2398
20. GBD 2013 Mortality and Causes of Death Collaborators . Global, regional, and national age–
sex specific all-cause and cause-specific mortality for 240 causes of death, 1990–2013: a
systematic analysis for the Global Burden of Disease Study 2013. Lancet 2015; 385: 117–71
21. Jha V, Garcia-Garcia G, Iseki K, Li Z, Naicker S, Plattner B, Saran R, Wang AYM, Yang CY
Chronic kidney disease: global dimension and perspectives . Lancet 2013; 382: 260–72
22. Levey AS, Coresh J Chronic kidney disease. Lancet 2012; 379: 165–80
23. Coresh J Selvin E, Stevens LA Manzi J, Kusek JW, eggers P, van Lente F, Levey AS Prevalence
of Chronic Kidney Disease in the United States JAMA 2007; 298: 2038-2047
24. Bruck
25. Mills KT, Xu Y, Zhang W, Bundy JD, Chen CS, Kelly TN, Chen J, He J A systematic analysis of
worldwide population-based data on the global burden of chronic kidney disease in 2010.
Kidney Int. 2015; 88 :950-7
26. Peter Craig, Cyrus Cooper, David Gunnell, Sally Haw, Kenny Lawson, Sally Macintyre, David
Ogilvie, Mark Petticrew, Barney Reeves, Matt Sutton, Simon Thompson. Using natural
experiments to evaluate population health interventions: new Medical Research Council
guidance. J Epidemiol Community Health 2012; 66: 1182-1186
27. Pell JP, Haw S, Cobbe S, Newby DE, Pell ACH, Fischbacher C, McConnachie A, Pringle S,
Murdoch D, Dunn F, Oldroyd K, MacIntyre P, O’Rourke B, Borland W. Smoke-free
Legislation and Hospitalizations for Acute Coronary Syndrome. N Engl J Med 2008; 359: 48291
28. Add
29. Add
30. McGrath LJ, Kshirsagar AV, Cole SR, Wang L, Weber DJ, Stürmer T, Brookhart AM. Evaluating
influenza vaccine effectiveness among hemodialysis patients using a natural experiment.
Arch Intern Med. 2012; 172: 548–554
31. Karunaratne K, Stevens P, Irving J, Hobbs H, Kilbride H, Kingston R, Farmer C The impact of
pay for performance on the control of blood pressure in people with chronic kidney disease
stage 3–5. Nephrol Dial Transplant 2013; 28: 2107–2116
32. Bello AK, Levin A, Manns BJ, Feehally J, Drueke T, Faruque L, Hemmelgarn B, Kernahan C,
Mann J, Klarenbach S, Remuzzi G, Tonelli M, on behalf of the Kidney Health for Life Initiative
Effective CKD Care in European Countries: Challenges and Opportunities for Health Policy.
Am J Kidney Dis YEAR 65:15-25
33. Open Humans project. https://www.openhumans.org/. Accessed November 30, 2015
34. Ainsworth J, Buchan I. Combining Health Data Uses to Ignite Health System Learning.
Methods Inf Med. 2015; 54: 479-87