Nephrol Dial Transplant (2014) 29: 1460–1467
doi: 10.1093/ndt/gft473
Advance Access publication 27 November 2013
NDT Perspectives
Measuring the quality of renal care: things to keep in mind when
selecting and using quality indicators
Sabine N. van der Veer1,2, Wim van Biesen2,3, Cécile Couchoud4, Charles R.V. Tomson5 and Kitty J. Jager1,6
1
Department of Medical Informatics, Academic Medical Center, Amsterdam, The Netherlands, 2European Renal Best Practice (ERBP) Methods
Support Team, University Hospital Ghent, Ghent, Belgium, 3Renal division, University Hospital Ghent, Ghent, Belgium, 4REIN Registry,
Agence de la biomédecine, Saint Denis la Plaine Cedex, France, 5The Richard Bright Kidney Unit, Southmead Hospital, Bristol,
UK and 6ERA-EDTA Registry, Academic Medical Center, Amsterdam, The Netherlands
Correspondence and offprint requests to: S.N. van der Veer; E-mail: [email protected]
and the public [1–6]. Rather than focusing on individual clinicians, these measurement initiatives evaluate systems of
healthcare delivery based on a set of quality indicators. Indicators can be categorized into different types, all with their own
advantages and disadvantages, which we describe and illustrate in the first section of this paper. There, we also discuss
how the aim of an initiative determines how the pros and cons
of those indicator types should be weighed. We argue that this
should be done with care to avoid potential unintended consequences, especially within initiatives that aim to facilitate external judgement of care quality.
In the second section we suggest using ‘causal chains’ as a
potential point of departure when composing a set of interrelated indicators. In a causal chain, each element (i.e. an aspect
of care to be translated into an indicator) is connected to the
next element by an evidence-based link. We also provide some
examples of indicator sets currently used for quality of care
measurement in nephrology.
Finally, in the last section we advocate that–apart from
(surrogate) outcome indicators—process indicators should
also be adjusted for case mix in order to enable correct interpretation of clinical quality measurements.
A B S T R AC T
This educational paper discusses a variety of indicators that
can be used to measure the quality of care in renal medicine.
Based on what aspect of care they reflect, indicators can be
grouped into four main categories: structure, process, surrogate outcome and outcome indicators. Each category has its
own advantages and disadvantages, and we give some pointers on how to balance these pros and cons while taking into
account the aim of the measurement initiative. Especially
within initiatives that link payment or reputation to indicator measurement, this balancing should be done with
utmost care to avoid potential, unintended consequences.
Furthermore, we suggest consideration of (i) a causal chain—
i.e. subsequent aspects of care connected by evidence-based
links—as a starting point for composing a performance indicator set and (ii) adequate case-mix adjustment, not only of
(surrogate) outcomes, but also of process indicators in order
to obtain fair comparisons between facilities and within facilities over time.Keywords: nephrology, quality indicators,
quality of health care
INTRODUCTION
MEASURING THE QUALITY OF CARE: BASIC
CONCEPTS AND DEFINITIONS
Initiatives worldwide are measuring the type and quality of
renal care in order to identify practice patterns, engage and
support clinicians in improving the quality of their local care
delivery system or to provide transparency to policy-makers
© The Author 2013. Published by Oxford University Press on
behalf of ERA-EDTA. All rights reserved.
Measures versus indicators
We can define measures for quantifying the things we see
in the world surrounding us. For the human body, for
1460
example, you can measure its weight in kilograms. To determine whether it concerns a healthy weight, you can relate it to
the body’s height in centimetres by calculating the body mass
index, or—for more accurate information—take into account
waist–hip ratio, skinfold thickness and body fat distribution.
All are measures of the human body.
‘Quality of care’ or ‘clinical performance’, however, cannot
be put on a scales, held against a measuring tape, or scanned
with a device to analyse its composition. Hence, there are no
direct measures of the quality of care because—unlike the
human body—it is an abstract concept that is not part of our
physical world. To quantify this abstract concept, we are thus
dependent on the measurable aspects of health care that are
only an indication of its quality. This paper is about these
measurable aspects, to which we will further refer as ‘quality
indicators’ or ‘clinical performance indicators’.
Clinical correlates and surrogate outcomes as indicators
of clinical performance
Clinical correlates are clinical or laboratory signs that can
be objectively measured and are associated with disease activity, but not necessarily causally related to patient outcomes.
The special subgroup of clinical correlates that is thought to
have a causal relationship with a subsequent health outcome is
referred to as surrogate outcomes [10]. When carefully selected
and validated, indicators based on clinical correlates and surrogate outcomes may be useful as they are more easily obtained than outcome indicators [11], while also being more
directly related to patients’ health status than process indicators. However, parameters are often called ‘surrogates’ without
Clinical performance measurement in renal
Indicator definition: selecting a numerator
and a denominator
Many performance indicators are defined as the proportion
of a numerator—specifying how many times preferred care
was provided or events occurred—relative to a denominator
reflecting the number of patients eligible or at risk. Selecting a
different numerator or denominator affects the final indicator
value, and thus, the conclusion on clinical performance. Box 1
contains an illustrative example, which shows how one may
account for the potential influence of patient characteristics
and of care delivered by healthcare professionals other than
nephrologists.
What makes a good quality indicator?
The perfect quality indicator does not exist as each has its
own advantages and disadvantages (Table 1) [11, 18, 19]. How
to balance these pros and cons depends on the aim of the
measurement initiative. There are three types of initiatives
[9, 20, 21]. First, performance-monitoring initiatives, such as
(inter)national renal registries and the Dialysis Outcomes and
Practice Patterns Study programme, which regularly review
clinical performance based on a combination of structure, process and (surrogate) outcome indicators. They aim to formulate hypotheses on the relationships between the different
aspects of care and to identify trends over time in practices
and outcomes. This requires indicators that can be measured
reliably in large populations over a long time period, accompanied by information on potential confounders. Secondly,
formative initiatives focus on internal quality control,—i.e.
without external interference—and improving care processes
[3, 4, 22]. Indicators used within such initiatives should be especially sensitive to changes in performance over time, easy to
interpret and a straightforward basis for deciding which QI initiatives should be undertaken. Lastly, summative initiatives
are characterized by external judgement of care (by governments, payers, patients, etc.) linked to direct consequences for
payment or reputation [5, 6, 23]. Selection and use of indicators for this purpose should be done with the greatest possible
care in order to avoid unjustified sanctions or rewards.
Chassin et al. [24] suggested that in this case, indicators
should reflect processes of care that (i) have a strong evidencebased link to relevant outcomes of care; (ii) can be measured
reliably; (iii) are proximate to the desired outcome and (iv)
1461
NDT PERSPECTIVES
Indicators referring to healthcare structures, processes
and outcomes
Besides being abstract, quality of care is also a multidimensional construct. One can distinguish three classic dimensions, resulting in three categories of quality indicators:
those referring to structures, to processes and to outcomes of
health care [7]. Structure indicators refer to characteristics of
the healthcare setting that affect a system’s ability to meet
healthcare needs of (a group of ) patients [8]. Often they
reflect the availability of services or resources. Examples are
the availability of a dedicated outpatient vascular access
service, the renal nurse-to-patient ratio or having a nutritional patient counselling programme in place. Process indicators refer to the care that is actually being delivered, for
instance, the timely performance of non-invasive ultrasonography of vessels in haemodialysis (HD) patients listed for
vascular access creation, or hepatitis B vaccination in seronegative patients. Lastly, outcome indicators—either observed or patient reported [9]–involve the ultimate health
status of the patient or the occurrence of (adverse) events
after having received treatment. Examples of observed
outcome indicators are cardiovascular death in patients with
end-stage renal disease (ESRD) and bacteraemia in HD patients with a tunnelled catheter, whereas the quality of life
after kidney transplantation and satisfaction with dialysis
care are typical patient-reported outcome indicators.
a robust evaluation of their relation with clinical endpoints in
large, well-designed studies that measure both surrogate and
ultimate patient outcomes. For example, based on observational data the degree of proteinuria seems a surrogate for the
decline of renal function [12, 13]. Yet, there is still debate
whether aiming for a reduction in proteinuria indeed slows
progression to ESRD [14, 15]. In addition, the validity of surrogate outcomes is specific to the context (e.g. type of patients,
targeted outcomes) in which they were evaluated, making
their interpretation complex [10]. For example, lowering blood
pressure was shown to decrease cardiovascular events in patients with diabetes [16], but this was less clear for diabetes patients who also had nephropathy [17].
Box 1. Calculation of the percentage of patients receiving HD via an arteriovenous fistula as an indicator of dialysis centres’ performance—fictitious
example
Assume we have the following vascular access data available:
All ages
All incident HD patients
536
Screened for fistula suitability at least once before start dialysis
418
Functioning AV fistula at start dialysis
171
Functioning AV fistula at day 91 of dialysis
240
We could now calculate the following values for the ‘same’ indicator by selecting different numerators and denominators:
Indicator
Numerator
Denominator
A
171
536
B
240
536
C
227
429
D
418
536
<80 years
429
364
156
227
Indicator value (%)
32
45
53
78
Indicator A
Dividing the number of patients with an AV fistula at the start of dialysis as the numerator by the total number of incident HD patients as the denominator
results in an indicator value of 32%. However, for part of the patients starting without a fistula, maturation time may have been too limited due to late referral
by general practitioners. One could argue that it is unfair to attribute these cases to the clinical performance of the dialysis centre.
Indicator B
To limit the effect of late referrals, one could use the number of patients with an ‘AV fistula at Day 91 of dialysis’ as the numerator instead. This increases the
indicator value to 45%.
NDT PERSPECTIVES
Indicator C
Patient factors may impact on a centre’s AV fistula rate. For example, in older patients it may be more difficult to establish a functioning fistula due to increased
risk of maturation failure and access-related complications. Also, elderly patients might value the expected survival gain from dialysis via a fistula less than
younger people. Instead of ascribing the effect of age to a centre’s performance, we could exclude octogenarians from the equation and focus on the 429
incident HD patients <80 years of age. This further increases the indicator value to 53%.
Indicator D
To avoid centres from using older age alone as an exclusion criterion for even considering patients for AV fistula creation, one could calculate the percentage of
all incident HD patients that were screened for fistula suitability at least once before starting dialysis. This indicator is less affected by other healthcare providers
and patient-related factors than indicators A–C. However, it only reflects a first step of fistula creation and not the required subsequent ones. Also, it is difficult
to determine whether an improved value of this indicator means that a centre changed the way of delivering vascular access care, or that they just checked more
‘fistula screening performed’ boxes.
AV, arteriovenous; HD, haemodialysis.
have minimal or no unintended consequences. Additionally,
summative initiatives should monitor if their linking performance to (financial) incentives does not result in physicians applying a ‘one-size-fits-all’ policy just to get high indicator
scores. For example, reimbursing HD via an arteriovenous (AV)
fistula at a higher rate than through a catheter—which is the case
in the UK—might drive dialysis centres to construct fistulae
while ignoring the potential harm, e.g. in older patients [25].
To illustrate how to assess the ‘goodness’ of an indicator for
specific purposes, let us look at the ‘percentage of patients with
CKD (Stage 3–5; not on dialysis) who had a fasting lipid
profile performed at least once in the last year’ [23]. It seems
suitable for a formative purpose: being a process indicator;
even small sample sizes are expected to be sufficient for detecting a change in performance [11]. Moreover, since it is hardly
influenced by patient-related factors (such as demographics,
patient preference) its interpretation is rather straightforward
as being under the influence of care providers. Also, it incorporates the preferred clinical action ( performing the fasting
1462
lipid profile), thus revealing the focus of actions needed to
improve the indicator.
When assessing its use for a summative purpose, two of the
Chassin criteria appear satisfied: availability of cholesterol and
triglyceride values in the laboratory system would be a reliable
way to determine if the profile indeed has been performed
(criterion 2); and performing the lipid profile is unlikely to be
harmful in itself (criterion 4). However, criterion 1 is only
partly fulfilled: the indicator was derived from a National
Kidney Foundation Kidney Disease Outcomes Quality Initiative guideline recommendation for which the strength of the
evidence base was graded as only ‘moderate’ [26]. Finally, criterion 3 is violated because many clinical follow-up actions
need to be executed correctly after performing the profile in
order to establish improvements that are noticeable to patients,
i.e. there is no direct link between the indicator and the
desired outcome. So, it is questionable if rewarding or punishing healthcare providers based on this indicator is fair and will
actually improve the quality of care.
S.N. van der Veer et al.
Table 1. Categories of quality indicators and their main (dis)advantages
Category
Refer to
Structures Healthcare setting characteristics
that affect the systems’ ability to
meet health care needs of patients
Processes
Examples
Advantagesa
Disadvantagesa
•
• Easy to record
• Difficult and costly to change
• Not influenced by
patient case mix
• Additional information required to
understand the indicator
•
Renal nurse-to-patient ratio
•
Having nutritional patient
counselling programme in place
Hepatitis B vaccination in HD • Sensitive to a change in • Clinical relevance is subject to changing
trends
practice
patients
Care that is actually being delivered •
•
•
Clinical
correlates
Objectively measured clinical or
laboratory signs associated with
disease activity, but not causally
related to outcomes
Availability of dedicated
outpatient vascular access
service
•
•
•
Surrogate Clinical correlates that have a causal
outcomes relationship with outcomes
Outcomes Ultimate health status
•
Prescription of ACE inhibitors • Limited delay between •
delivering care and
in CKD patients with
measuring performance •
hypertension and proteinuria
• Incorporate preferred
Performing
non-invasive ultrasonography of clinical action
vessels in HD patients listed for
vascular access creation
• More directly related to •
LDL cholesterol
ultimate health status
Blood pressure
than process indicators
Serum phosphorus
• More sensitive to
•
differences in quality of
care than outcomes
Directly influenced by patient
preference
Process indicators that regard the
performance of tests do not reflect if
necessary subsequent clinical actions
are performed.
Requires incorporation of a target
value/range, which is often subject to
discussion
Influenced by patient factors (e.g.
genetic profile, environmental factors,
compliance to treatment)
• Relevant and noticeable • Long observation period and large
to patients
sample size required to detect change
•
Occurrence of bacteraemia in
HD patients with a tunnelled
catheter
• Provide global
assessment of
performance
•
Quality of life after kidney
transplantation
•
Satisfaction with dialysis care
• Influenced by many other factors, e.g.
patient case mix, other care providers
• Potential reasons for
‘underperformance’ are not
immediately obvious
ACE, angiotensin-converting-enzyme; CKD, chronic kidney disease; HD, haemodialysis; LDL, low-density lipoprotein; PTH, parathyroid hormone.
a
Based on refs [11, 18, 19].
I N D I C AT O R S A S P A R T O F A C L I N I C A L
P E R F O R M A N C E I N D I C AT O R S E T
No single indicator covers all relevant dimensions of clinical
performance. Also, different stakeholders—clinicians, patients,
payers, policy-makers—may value certain aspects of care differently. Therefore, measurement initiatives are commonly
built around a set of indicators to create a comprehensive
picture of the quality of care provided.
Meaningful structure indicators for these sets are likely to
be found in systematic reviews of strategies to implement best
practices into daily renal care [27, 28], or in individual observational studies on the association between healthcare facility
characteristics and (surrogate) outcomes [29, 30]. For identifying process and surrogate outcome indicators Grade ‘1A’ or
‘1B’ recommendations from rigorously developed renal clinical practice guidelines may be a useful starting point [31].
These are the recommendations, which were based on evidence of high-to-moderate quality and which are expected to
apply to and be preferred by the majority of patients [32].
Finally, selection of relevant outcome indicators should be primarily guided by the purpose of the provided care. For
example, when measuring the quality of care for CKD patients
Clinical performance measurement in renal
(Stages 3–5; not on dialysis), progression to ESRD is one of the
potential outcome indicators, whereas the quality of vascular
access care may be reflected in the number of access-related infections and hospitalizations.
Evaluating causal chains in daily practice
‘Causal chains’ can be used as a starting point for composing
an indicator set [33]. Figure 1 shows an example of such a chain
for managing cardiovascular risk in adult CKD patients not on
dialysis. It consists of four elements, each belonging to a different (indicator) category; per element, we suggested a potential
indicator. The links between the elements have been evaluated
in (a meta-analysis of) randomized controlled trials (RCTs)
[34–36]. The main advantage of building an indicator set
around causal chains is that it increases the probability that a
focus on improving structure, process and surrogate outcomes
within the chain actually results in a better patient outcome.
Also, the use of a chain facilitates the evaluation of the chain in
the context of daily care in addition to previous evaluations of
the links within the often rather controlled contexts of RCTs
[37]. For example, the MASTERPLAN [34] and the SHARP
studies [35] excluded patients who were reluctant to change
drug therapy, and poorly compliant to clinic visits or prescribed
1463
NDT PERSPECTIVES
Cardiovascular mortality in
CKD patients
NDT PERSPECTIVES
F I G U R E 1 : Causal chain of reducing the cardiovascular risk in adult CKD patients not receiving dialysis, including examples of potential
performance indicators.
medication, respectively. Although often excluded in RCTs,
non-compliant patients are not uncommon in daily practice.
Therefore, it would be of interest to include this subpopulation
when monitoring performance, and evaluate if, e.g. having a
nurse practitioner is still equally effective in improving statin
prescription. However, strictly speaking, the causal chain is not
generalizable to non-compliant patients. Some initiatives
address this by having ‘patient decline’ and ‘other patient
reasons’ as criteria to exclude patients from the denominator
when calculating indicators [23], but allowing this ‘exception
coding’ is controversial. First, because it creates an opportunity
for healthcare providers to ‘game’ indicator values by excluding
all patients that did not meet the targets; especially within summative initiatives there might be an incentive to do this [38].
And secondly because it may seem to relieve physicians from
the responsibility to make an effort to convince patients to take
their medication or stop smoking [39]. So, the practice of exception coding at least warrants close monitoring to determine
whether its use was appropriate. For example, by identifying
providers with outlying exception rates or—more qualitatively—
by reviewing the medical records of samples of excluded patients.
Indicator sets in nephrology: two examples
An example of a set used for performance monitoring is
the NephroQUEST indicator data set, which resulted from the
NephroQUEST project initiated in 2007 by the European
Renal Association–European Dialysis and Transplant Association (ERA–EDTA) Registry with support from the European
Union [40]. Prior to NephroQUEST, the Registry primarily
collected data on case mix and outcomes of patients on renal
replacement therapy (RRT). In order to also enable the
1464
composition of other types of indicators, clinical and registry
experts defined additional data items (Box 2), using international renal guidelines as the starting point. To increase the
set’s sustainability over a longer period of time, the data item
definitions leave room to incorporate treatment targets at a
later stage because recommendations on which targets to
achieve are constantly evolving. As the influence of advancing
insights also applies to processes of care, the number of items
to compose process indicators is limited. The set lacks items
related to characteristics of dialysis centres for composing
structure indicators, which may be a valuable addition in the
future. For instance, frequent routine multidisciplinary conferences, or clinical decision support systems for drug prescription have been suggested to positively affect processes and
outcomes of renal care [28, 29, 41].
The extended ERA-EDTA Registry data set may be used for
evaluating parts of causal chains in daily practice. For
example, by assessing the association between achieving a
certain target and survival in European RRT patients or by investigating the relationship between potential surrogate and
final outcomes. Moreover, the data set enables exploring practice patterns among European dialysis centres, and for subsequently determining between-centre or country variation.
The Centers for Medicare and Medicaid Services (CMS) indicator set is an example of a set used for formative as well as
summative purposes. Since 2000, CMS have been measuring
clinical performance of dialysis centres in the USA. While their
first guideline-based indicator set consisted of 16—mostly
process—indicators [42], their 2010 report presented an extension of the set to a total of over 50 indicators, including also potential surrogates (e.g. haemoglobin) and outcome indicators (e.
g. HD access-related infections) [22]. Apart from formatively
S.N. van der Veer et al.
Box 2. ERA-EDTA
NephroQUEST items
registry
data
set,
including
the
additional
Structures
[None]
Processes
• Treatment with ESAs
• Dialysis duration and frequencya
• Vascular access typea
(Potential) Surrogate outcomes
• Systolic and diastolic blood pressure
• Serum albumin
• C-reactive protein
• Total cholesterol; high-density lipid cholesterol; triglycerides
• Haemoglobin; ferritin
• Calcium; phosphorus; parathormone
• Urea clearance
• Creatinine clearanceb
Outcomes
• Treatment modalityc
• Time on RRTc,d
• Survival on RRTc,d
• Source of kidney donorc
• Graft survival of kidney transplantsc,d
Patient case mix
• Age; genderd
• Height; dry weight; major amputations
• Smoking status
• Primary renal diseased
Collected only in the context of HD.
Collected only in the context of peritoneal dialysis.
c
Already collected prior to the NephroQUEST project.
d
Derived from items: date of first RRT; date/type of event or treatment; cause of death.
b
supporting QI efforts among dialysis providers, the CMS clinical performance measurement project also aims to enable patients to participate in making healthcare decisions. This
summative aspect of the initiative has been further emphasized
since the start of the CMS Quality Incentive Program [6], which
links four process indicators to reimbursement: percentage of
erythropoietin-stimulating agent (ESA)-treated patients with
haemoglobin >12 g/dL (with higher percentages getting lower
quality scores); percentage of patients with a urea reduction
ratio (URR) ≥65%; percentage of patients using a bipuncture
AV fistula during the last treatment of the month; percentage of
HD patients with an intravenous catheter in use for 90 days
before the last dialysis session. However, Fishbane et al. clearly
showed that none of these indicators fulfil all four Chassin criteria, which makes their use for summative purposes by CMS
questionable [6].
I N T E R P R E T I N G R E S U LT S O F P E R F O R M A N C E
M E A S U R E M E N T S : A D J U S T I N G F O R P AT I E N T
C H A R AC T E R I S T I C S
Besides the clinical quality of the provided care, characteristics
of the patient population can significantly affect the results of
clinical performance measurements. For example, patient
factors may explain a substantial part of differences between
CMS dialysis centres in achievement of a target URR ≥65%
Clinical performance measurement in renal
CONCLUSION
In this paper, we distinguished indicators referring to structures, processes, surrogate outcomes and outcomes of care. No
single indicator provides a comprehensive picture of clinical
performance. Therefore, indicators ought to be part of an indicator set that ideally covers elements of care that are connected
by evidence-based links. The advantages and disadvantages of
the different indicators should be balanced depending on the
aim of the measurement initiative. This is particularly important within summative initiatives, which directly link external
judgement to consequences for payment or reputation. For
this purpose, particular process indicators may be worth considering, but only if they fulfil the criteria for a suitable
‘accountability indicator’. However, we did not find any
1465
NDT PERSPECTIVES
Comorbidities at start RRTd
a
[43]. For this reason, the values of performance indicators
should be adjusted for patient factors as adequately as possible
to prevent interpretation of differences in case mix as differences in performance. The ERA-EDTA indicator data set (Box
2) contains patient characteristics frequently used for risk
adjustment in RRT populations. However, unknown, and thus
unmeasured, case-mix factors hamper complete risk correction. This is known as the ‘case-mix fallacy’ [9]. In addition,
registration of risk factors, such as comorbidities, might differ
substantially between nephrologists and dialysis centres, either
unintended or intended [44].
Risk adjustment usually seems to concern (surrogate)
outcome indicators. Although structure indicators, like having a
‘nurse practitioner trained in cardiovascular risk management’
(Figure 1), are rarely affected by patient case mix, process indicators may be. Sicker patients need more (complex) care, which
can justify a deviation from care as described in guidelines, as
well as increase the susceptibility of their treatment to errors
[9]. In addition, process indicators are more directly influenced
by patient preference than outcome indicators. For example,
older CKD patients often have more comorbidities and are
being prescribed a larger variety of medications compared with
their younger, healthier counterparts. A higher degree of multimorbidity and polypharmacy in such patients makes it more
likely that the nephrologist will decide not to prescribe statins
either to avoid adverse interactions with drug treatments recommended by different guidelines [45], or because the patient
prefers not to take ‘yet another pill’. Also, it increases the chance
that statins are omitted unintentionally from the drug regimen.
So, in Figure 1, not only the (surrogate) outcomes ‘LDL cholesterol level’ and ‘risk of cardiovascular events’, but also the
process indicator ‘percentage of patients being prescribed
statins’ requires case-mix adjustment in order to be interpreted
correctly as an aspect of the quality of care provided to CKD
patient who are not on dialysis. Unfortunately, the lack of accepted risk-adjustment methods on the one hand, and mostly
unavailable patient preference data on the other hand hinder
adequate adjustment of process indicators for patient factors.
The fact that this impedes fair comparisons between facilities
and within facilities over time is especially problematic when
using such process indicators for summative purposes.
examples within renal medicine that would meet such criteria.
When additionally taking into account the need for and difficulty of adequate case-mix adjustment of process indicators,
we feel that it is questionable for clinical performance indicators to be used for summative purposes at all. Instead, measurement initiatives should rather focus on facilitating
observational research and monitoring trends over time, and
aim to improve patient care whilst avoiding unwarranted penalization of healthcare professionals.
C O N F L I C T O F I N T E R E S T S TAT E M E N T
None declared. Furthermore, the authors declare that this
paper has not been published previously in whole or part.
NDT PERSPECTIVES
REFERENCES
1. Dialysis Outcomes and Practice Patterns Study program (2 September
2013, date last accessed)
2. Tomson C, Ford D, Ansell D. The UK Renal Registry: an overview. Br J
Hosp Med (Lond) 2008; 69: 548–549
3. van der Sande FM, Kooman JP, Ikenroth LJ et al. A system of quality management in dialysis. Semin Dial 2003; 16: 453–457
4. Stoffel MP, Barth C, Lauterbach KW et al. Evidence-based medical quality
management in dialysis—Part I: routine implementation of QiN, a
German quality management system. Clin Nephrol 2004; 62: 208–218
5. Consumer Assessment of Healthcare Providers and Systems (CAHPS).
In-center Hemodialysis survey. www.cahps.ahrq.gov/hemodialysis (19
February 2013, date last accessed).
6. Fishbane S, Miller I, Wagner JD et al. Changes to the end-stage renal disease
quality incentive program. Kidney Int 2012; 81: 1167–1171
7. Donabedian A. Evaluating the quality of medical care. 1966. Milbank Q
2005; 83: 691–729
8. Mainz J. Defining and classifying clinical indicators for quality improvement. Int J Qual Health Care 2003; 15: 523–530
9. Lilford R, Mohammed MA, Spiegelhalter D et al. Use and misuse of
process and outcome data in managing performance of acute medical
care: avoiding institutional stigma. Lancet 2004; 363: 1147–1154
10. Manns B, Owen WF, Jr, Winkelmayer WC et al. Surrogate markers in clinical studies: problems solved or created? Am J Kidney Dis 2006; 48:
159–166
11. Smith KA, Hayward RA. Performance measurement in chronic kidney
disease. J Am Soc Nephrol 2011; 22: 225–234
12. Iseki K, Iseki C, Ikemiya Y et al. Risk of developing end-stage renal disease in
a cohort of mass screening. Kidney Int 1996; 49: 800–805
13. Atkins RC, Briganti EM, Lewis JB et al. Proteinuria reduction and progression to renal failure in patients with type 2 diabetes mellitus and overt nephropathy. Am J Kidney Dis 2005; 45: 281–287
14. Cravedi P, Ruggenenti P, Remuzzi G. Proteinuria should be used as a surrogate in CKD. Nat Rev Nephrol 2012; 8: 301–306
15. Thompson A. Proteinuria as a surrogate end point—more data are
needed. Nat Rev Nephrol 2012; 8: 306–309
16. Lindholm LH, Ibsen H, Dahlof B et al. Cardiovascular morbidity and
mortality in patients with diabetes in the Losartan Intervention For Endpoint reduction in hypertension study (LIFE): a randomised trial against
atenolol. Lancet 2002; 359: 1004–1010
17. Brenner BM, Cooper ME, de Zeeuw D et al. Effects of losartan on renal
and cardiovascular outcomes in patients with type 2 diabetes and nephropathy. N Engl J Med 2001; 345: 861–869
18. Mountford J, Shojania KG. Refocusing quality measurement to best
support quality improvement: local ownership of quality measurement by
clinicians. BMJ Qual Saf 2012; 21: 519–523
19. Pringle M, Wilson T, Grol R. Measuring ‘goodness’ in individuals and
healthcare systems. BMJ 2002; 325: 704–707
1466
20. Freeman T. Using performance indicators to improve health care quality
in the public sector: a review of the literature. Health Serv Manage Res
2002; 15: 126–137
21. Solberg LI, Mosser G, McDonald S. The three faces of performance measurement: improvement, accountability, and research. Jt Comm J Qual
Improv 1997; 23: 135–147
22. Centers for Medicare and Medicaid Services Clinical Performance Measurement project. http://www.cms.gov/Medicare/End-Stage-Renal-Disease/
CPMProject (8 March 2004, date last accessed).
23. American Medical Association—Convened Physician Consortium for
Performance Improvement. Adult Kidney Disease. Performance Measurement set; Measures for Accountability. Chicago: American Medical Association, 2011
24. Chassin MR, Loeb JM, Schmaltz SP et al. Accountability measures - using
measurement to promote quality improvement. N Engl J Med 2010; 363:
683–688
25. Vachharajani TJ, Moossavi S, Jordan JR et al. Re-evaluating the fistula first
initiative in octogenarians on hemodialysis. Clin J Am Soc Nephrol 2011;
6: 1663–1667
26. K/DOQI clinical practice guidelines for management of dyslipidemias in
patients with kidney disease. Am J Kidney Dis 2003; 41: I–91
27. van der Veer SN, Jager KJ, Nache AM et al. Translating knowledge on best
practice into improving quality of RRT care: a systematic review of implementation strategies. Kidney Int 2011; 80: 1021–1034
28. Tawadrous D, Shariff SZ, Haynes RB et al. Use of clinical decision support
systems for kidney-related drug prescribing: a systematic review. Am J
Kidney Dis 2011; 58: 903–914
29. Spiegel B, Bolus R, Desai AA et al. Dialysis practices that distinguish facilities with below- versus above-expected mortality. Clin J Am Soc Nephrol
2010; 5: 2024–2033.
30. Spiegel BM, Bolus R, Desai AA et al. Dialysis practices that distinguish
top- versus bottom-performing facilities by hemoglobin outcomes. Am J
Kidney Dis 2010; 56: 86–94
31. Kotter T, Blozik E, Scherer M. Methods for the guideline-based development of quality indicators—a systematic review. Implement Sci 2012; 7: 21
32. Guyatt GH, Oxman AD, Kunz R et al. Going from evidence to recommendations. BMJ 2008; 336: 1049–1051
33. Howick J, Glasziou P, Aronson JK. Evidence-based mechanistic reasoning.
J R Soc Med 2010; 103: 433–441
34. van Zuilen AD, Bots ML, Dulger A et al. Multifactorial intervention with
nurse practitioners does not change cardiovascular outcomes in patients
with chronic kidney disease. Kidney Int 2012; 82: 710–717
35. Baigent C, Landray MJ, Reith C et al. The effects of lowering LDL cholesterol with simvastatin plus ezetimibe in patients with chronic kidney
disease (study of heart and renal protection): a randomised placebo-controlled trial. Lancet 2011; 377: 2181–2192
36. Palmer SC, Craig JC, Navaneethan SD et al. Benefits and harms of statin
therapy for persons with chronic kidney disease: a systematic review and
meta-analysis. Ann Intern Med 2012; 157: 263–275
37. Jager KJ, Stel VS, Zoccali C et al. The issue of studying the effect of interventions in renal replacement therapy—to what extent may we be deceived
by selection and competing risk? Nephrol Dial Transplant 2010; 25:
3836–3839
38. Doran T, Fullwood C, Gravelle H et al. Pay-for-performance programs in
family practices in the United Kingdom. N Engl J Med 2006; 355: 375–384
39. Rosenbaum L, Shrank WH. Taking our medicine–improving adherence in
the accountability era. N Engl J Med 2013; 369: 694–695
40. European Nephrology Quality Improvement Network - NephroQUEST.
http://www.nephro-quest.eu (6 February 2013, date last accessed)
41. Plantinga LC, Fink NE, Jaar BG et al. Frequency of sit-down patient care
rounds, attainment of clinical performance targets, hospitalization, and
mortality in hemodialysis patients. J Am Soc Nephrol 2004; 15:
3144–3153
42. Sugarman JR, Frederick PR, Frankenfield DL et al. Developing clinical
performance measures based on the Dialysis Outcomes Quality Initiative
Clinical Practice Guidelines: process, outcomes, and implications. Am J
Kidney Dis 2003; 42: 806–812
43. Tangri N, Tighiouart H, Meyer KB et al. Both patient and facility contribute to achieving the centers for medicare and medicaid services’ pay-for-
S.N. van der Veer et al.
performance target for dialysis adequacy. J Am Soc Nephrol 2011; 22:
2296–2302
44. Collier T, Steenkamp R, Tomson C et al. Patterns and effects of missing
comorbidity data for patients starting renal replacement therapy in
England, Wales and Northern Ireland. Nephrol Dial Transplant 2011; 26:
3651–3658
45. Boyd CM, Darer J, Boult C et al. Clinical practice guidelines and quality of
care for older patients with multiple comorbid diseases: implications for
pay for performance. JAMA 2005; 294: 716–724
Received for publication: 19.9.2013; Accepted in revised form: 10.10.2013
Nephrol Dial Transplant (2014) 29: 1467–1474
doi: 10.1093/ndt/gft492
Advance Access publication 22 December 2013
Integrated genomics and metabolomics in nephrology
Dorothee Atzler1,2, Edzard Schwedhelm1,2 and Tanja Zeller2,3
1
Department of Clinical Pharmacology and Toxicology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany, 2German Center
for Cardiovascular Research (DZHK), Partner Side Hamburg/Lübeck/Kiel and 3Clinic for General and Interventional Cardiology, University
Heart Center Hamburg, Hamburg, Germany
A B S T R AC T
Applying ‘omics’ approaches such as genome-wide association
studies (GWAS) and metabolome analyses, genes and metabolites have been identified to be associated with renal pathophysiology. Meta-analyses of GWAS from large epidemiologic
cohorts uncovered several novel loci linked with estimated
glomerular filtration rate and chronic kidney disease (CKD).
Sophisticated analytical technologies, including mass spectrometry and nuclear magnetic resonance spectroscopy, allow
the analyses of up to 4000 targeted and non-targeted metabolites in plasma, serum and urine. Several uraemic toxins were
found that were increased in CKD. Among them, arginine derivatives like asymmetric dimethylarginine or tryptophane
metabolites have been identified as promising candidates to
target mechanisms of kidney disease progression. This review
aims to summarize recent findings in clinical kidney diseases
research revealed by ‘omics’ approaches with a clear focus on
recent genomics and metabolomics efforts.
mechanisms. Valuable insight has been acquired from candidate gene approaches to genome-wide association studies
(GWAS); however, the underlying pathophysiological mechanisms do not only involve different genotypes but also
changes in cellular, blood and urinary metabolites. Metabolomics complements other ‘omics’ data and as a downstream
result of gene expression, changes in the metabolome are considered to reflect the activities of the cell at a functional level
(Figure 1). Building on a substantive body of work over the
past few years, we discuss herein the current knowledge of genomics and metabolomics in kidney diseases.
G E N O M E - W I D E A S S O C I AT I O N S T U D I E S
Great progress has been made in mapping genomic variations
of common disease at a genome-wide level. GWAS use
genomic variations, termed single nucleotide polymorphisms
(SNPs), to identify regions of the genome associated with the
disease status or a clinical phenotype [1, 2].
Keywords: chronic kidney disease, genomics, metabolomics,
nephrology
G WA S : T H E O R E T I C A L A S P E C T S
INTRODUCTION
Complex kidney diseases like chronic kidney disease (CKD),
diabetic nephropathy (DN) or renal hypertension, involve a
complex
bundle
of
underlying
pathophysiological
© The Author 2013. Published by Oxford University Press on
behalf of ERA-EDTA. All rights reserved.
By design, GWAS provide an unbiased survey of the effects of
common genetic variants. SNPs chosen for GWAS typically
have a minor allele frequency (MAF) of ≥0.05 and are selected
to ‘tag’ the most common haplotypes. However, the power of
detection depends directly on the sample size of the study
1467
NDT PERSPECTIVES
Correspondence and offprint requests to: Tanja Zeller; E-mail: [email protected]