Back Home Next



‘Optimal Haemodialysis’

 Is ‘adequate’ still good enough?


Haemodialysis ‘adequacy’


1.    The current ‘measure’ of haemodialysis adequacy is a mathematical ‘concept’ called Kt/V* – more of this later …


2.    A ‘poor-man’s’ Kt/V called the Urea Reduction Ratio (URR)* is also used as it provides a good approximation to the Kt/V


We talk of a dialysis treatment being ‘adequate’ if this measure, the Kt/V*, can reach 1.3 per treatment … regardless of dialysis duration (time) or frequency.


For those who use the URR, adequacy is achieved if the reduction in urea for any given treatment is >65%.


Kt/V and URR will be described later in this section


Most dialysis units/services measure Kt/V or URR once every 2-3 months on a random mid-week dialysis. They ‘assume’ that the result obtained is representative of ‘all’ dialysis treatments in that measurement period.

That seems a naïve assumption at best … !


The inadequacies of ‘adequacy’

‘Adequate’ is defined as … *

‘sufficient’  …  but sufficient for what?

to remove all symptoms of disease?

to normalise all biochemistry


to simply enable ‘activity of daily living’?

to simply extend life/stay alive?

‘satisfactory’ … but this implies ‘barely so’

‘enough’ … but ‘enough’ proportionate to what?

The Australian Concise Oxford Dictionary

Adequate adequacy


Haemodialysis can only be adequate when:

all symptoms and signs of uraemia are eradicated

the patient is fully rehabilitated

dialysis does not interrupt activities of daily living


… thus, true ‘haemodialysis adequacy’ is, in practice, rarely if ever reached



But …


We talk of our current measures of haemodialysis ‘adequacy’ as if we think that what we do now is ‘optimal’, that these measures somehow, magically, confer all these benefits ... and confer them in such a way that the treatment is deemed ‘good’ dialysis’.


But ...


Current adequate dialysis is NOT good dialysis


Current ‘adequacy’ measures are restricted to the measurement of the clearance (removal) by dialysis of one, small and very unrepresentative solute (waste), urea


… and good dialysis should be and is far more than that!




The ideal ‘marker’ of adequacy


A marker of ‘adequacy’ to determine ‘how much dialysis is enough dialysis’ has long been sought.


As ‘clearance’ is (essentially) synonymous with ‘renal function’, a measure of ‘dialysis clearance’ seemed to be a good way to assess ‘dialysis function’

This, of course, begs the question that clearance is the ‘be-all’ and ‘end-all’ marker of ‘goodness’


Vanholder has defined that an ‘ideal marker of adequacy’ should be:-

Retained in renal failure

Eliminated completely by dialysis

Proven to have dose-related toxicity

Furthermore …

Its generation/removal should be accurately representative of other toxins

It should be easily and reliably measured


Uraemic toxins and possible markers


Some potential candidate toxins considered as the marker have include:


Small Soluble molecules





Protein-bound molecules


Indoles and phenols



‘Middle’ Molecules


β2 microglobulin



And the winner was … urea !


Urea – the ‘practical’ marker...

Urea reflects

the efficiency of removal of other small molecular weight uraemic toxins…

but ...

importantly, urea does not always reflect the clearance of larger solutes

It also reflects the dietary protein intake

Urea losses through dialysis and in the faeces also usefully allow the calculation of a nutritional constant – the Protein Catabolic Rate (PCR)

But ...

Urea has, in my view sadly, become the ‘surrogate marker’ of dialysis adequacy, despite many critics and criticisms



Should adequacy be more than this?


Beyond the simplistic concept of efficient small solute clearance, I believe that optimal haemodialysis should include the:

Resolution of all overt disease

Regression of covert (hidden) CKD pathology

Avoidance/abolition of dialysis-related side-effects

Reduction in the ultrafiltration rate to less than the equilibration rate

Allowance for a normal diet and fluid intake

Rehabilitation of lifestyle and employment capacity

Reversal of all adverse morbidity/mortality

… and, for the provider, even cost efficiency


To exclude all of these markers of ‘good dialysis’

from an assessment of dialysis adequacy is




yet that is exactly what we do!




Kt/V and the URR … (ugh!) … and the origins of adequacy


It all began with the NCDS study in 1981.


This study was designed to test the relative toxicity and importance of small (and middle) molecules.

In brief, it showed that the clearance of small solutes (eg urea) correlated ‘best’ with outcome.

It ‘conceived and birthed’ the concept of ‘Kt/V’ as a mathematical measure of dialysis adequacy.




K           = clearance of urea (ml/min)

t             = time (minutes)

V            = volume of body water (litres)


Kt/V depends upon

dialyser surface area

dialyser membrane characteristics

the blood flow rate (Qb) and dialysate flow rate (Qd)

the convective component of ultrafiltration


… in brief, Kt/V became the mathematicians dream!


The ‘purists’ approach ...


There are several formulae to calculate Kt/V (groan!)


Kt/V = - log(R - 0.03) + [(4 - 3.5 R) x (UF/W)]


Or ...


Kt/V = [- log((post BUN/pre BUN) - 0.008 x HD hrs)] + ((4 - 3.5 x post BUN/pre BUN) x (wt loss/post wt))

Where ...

UF          = UF volume (in litres)

W           = post dialysis weight (in kgm)

R            = ratio of post-dialysis to pre-dialysis BUN


The ‘simpletons’ approach …


For the ‘mathematically challenged’, Kt/V can be thought of as:

K =    Membrane characteristics & flow rates

t =     Time

V =    Body water as proportion of body weight


These three factors – in the prevailing view – are the key determinants of dialysis adequacy but, to me, this reliance on a mathematical formula based solely on the dialysis process fails my key determinants of patient outcome …


well-being of the individual


relief of symptoms


return to work and family-related pursuits


individual self-determination.


None of these are represented by the Kt/V formula.



Interpreting Kt/V


In addition, Kt/V is diminished by a multitude of confounding factors. These confirm Kt/V as a far too simplistic expression of ‘adequacy’.


These include:

The ‘in-parallel’ nature of haemodialysis

The inescapable facts of cardio-pulmonary and access recirculation

The diminished reliability of Kt/V as a ‘true’ measure of urea clearance without making allowances for ‘equilibration time’.

The inaccuracy of dialyser specifications and dialysis machine estimates of blood (Qb) and dialysate (Qd) rates

The inevitability of hollow fibre clotting and consequent ‘non-contribution’

The ‘cribbing’ of dialysis time - far commoner than we admit …

Errors in post dialysis urea measurement – the complex issues of single pool v double pool kinetics (not an issue for detailed discussion here)


In addition … early interpretations of Kt/V were misleading:

Initially, Sargeant and Gotch divided the NCDS dialysis population into two groups, those with a Kt/V of < 0.8 and those > 0.8. They concluded that:

A Kt/V < 0.8 was associated with high morbidity/mortality

A Kt/V > 0.8 was associated with a lower morbidity/mortality

Early on, a Kt/V > 0.8 became the 1st ‘Holy Grail’


The ‘80’s & ‘90’s were controversial decades with a gradual acceptance that higher Kt/V’s were needed for ‘adequate’ dialysis.


In 1998, Held first clearly showed a decline in mortality with increasing dialysis ‘dose’ and re-drew the Sargeant and Gotch graph.


He showed that morbidity and mortality clearly lessened as the Kt/V rose and that this advantage appeared to continue ‘ad infinitum’


Meanwhile, along came ‘easy adequacy’ - the URR






Daugirdas proposed a simpler measure of adequacy – the urea reduction ratio (URR)

URR = pre-dialysis urea - post-dialysis urea

pre-dialysis urea


The percent reduction urea (PRU) ... which simply expresses the URR as a percentage


The URR ...

Is easily calculated

Is epidemiologically useful

But ...

It contains all Kt/V’s inaccuracies, plus some


URR variations are primarily due to:

urea removal with ultrafiltration (which is not considered in Kt/V)

timing inaccuracies of post dialysis blood samples

single v double pool (urea equilibration time)

intra-dialytic urea generation


Kt/V derived from the URR …


Several equations have been developed to convert Kt/V to URR:

Kt/V = (0.026 x URR) - 0.460

Kt/V = (0.024 x URR) - 0.276

Kt/V - URR = (0.04 x URR) - 1.2

But … a URR-derived Kt/V is significantly inaccurate and imports and compounds all the variabilities of both measurements


In addition, there are other ‘chemistries’ to consider …

Chemistries other than urea clearly have an influence on adequacy …

The protein catabolic rate (PCR) and nutritional adequacy is an important contributor to overall dialysis adequacy but is, in part, confounded by intra-dialytic urea generation

The serum albumin has long been regarded as a major determinant of outcome … if not the most important of all  (Lowrie)

Acute phase reactants – CRP, ferritin and ? others have gained believers as important factors in determining outcome

Then there are …

B2 microglobulin

Oxidation end products


etc …

If you have been confused by all this … welcome to my nightmare


At the end of the day, a mathematical representation of dialysis adequacy is, in my view, far too simplistic and fails to include many of the things that matter most.


Adequacy IS clearly more than Kt/V.


To think that haemodialysis adequacy can ‘adequately’ modeled simply by measuring (badly) the clearance of a one small solute (urea) as the lone determinant is, at best, naïve and, at worst …


… but then, I had better not say!

So … what DOES seem to matter … ?


Is it flow rate (Qb), membrane area (m2), membrane permeability (flux) … or is it time?


In the 90’s, emphasis (especially in the US) was placed on:

Higher and higher flow rates (Qb)

Bigger and bigger (m2) dialysers

Leakier and leakier (flux) dialysers


But ...

Shorter and shorter (t) time


In the ‘90’s, major survival differences surfaced - US v ‘the rest’.


Even after taking into consideration differences between patient populations, US patient outcomes did not appear as good as those of patients in Japan and Europe.


This stimulated an international study to look at dialysis outcomes and practice profiles (DOPPS), a study followed by even more extensive studies involving an even more countries and dialysis services (DOPPS II, DOPPS III and DOPPS IV) to attempt to identify those practices which were associated with poor outcomes and those with better outcomes.



All that seems clear so far is that ...

1.    Mortality correlates inversely with delivered dialysis dose

2.    Dialysis dose = a factor of clearance and time



The HEMO study


In 1994, a 7 yr multi-centre US study, the HEMO study, was initiated.


Sadly and in retrospect, I think it asked the wrong questions.

It aimed to identify whether

membrane type – high flux (leaky) or low flux (less leaky)

the amount of dialysis (as measured by delivered Kt/V)

or both

… affected outcome


HEMO compared 4 groups:

Low flux + low Kt/V            }

High flux + low Kt/V           }         no direct comparison of

Low flux + high Kt/V          }                  dose with time

High flux + high Kt/V         }


The initial results were released in late 2002



Outcomes of the HEMO trial

High flux did not improve average patient survival

Higher dose ‘tended to’ significance only

eKt/V(sp) 1.25 ‘as good as’ 1.65 (but it only used urea kinetics)

Unfortunately, there are lots and lots of confounders

And ... importantly, it did not ask the KEY question – is TIME ON DIALYSIS, (ie: dialysis membrane contact time) what really matters?



Time and the dialysis syndrome


Longer dialysis duration permits:

Slower ultrafiltration rates and less intra-dialytic hypotension

Proportionately greater middle molecular v small solute clearance

Better (reaching ideal) volume and BP control

More easily achieved ‘dry weight’

Slower/lesser intra-dialytic biochemical ‘assault’

Intra-dialytic v post-dialytic compartmental equilibration

Reduced cardiovascular ‘turmoil’ and morbidity


Frequency - is this the key?


Daily short hour dialysis

Increased frequency without increased time

Better clearances

... and some volume advantage


The following graph schematically represents the rate of removal of, for example, urea over a 4hr dialysis treatment.

Waste removal in 1st 2 hrs

Waste removal in 2nd 2 hrs


As can be seen, the rate of removal is fastest at the start and slowest at the end of dialysis. By having 6 x 2hr treatments/week, waste removal is maximised when compared to 3 x 4hr treatments/week. This is the ‘rationale’ behind short hour daily dialysis.

Long nightly dialysis removes even more waste than a short daily regime, but does so at a far slower and gentler rate, thus maximising waste and fluid removal but at the same time, minimising the speed of change and thus minimising the symptoms associated with dialysis


Nightly long hour dialysis

= Increased frequency with increased time

Provides maximum solute and fluid management

Uses minimum ultrafiltration rates

Attains maximum Kt/V (1.7-1.8)

The Haemodialysis Product (HDP)


In January 2002, Scribner & Oreopoulos published a new concept – the haemodialysis Product (HDP) – in Dialysis and Transplantation


They proposed adequacy be determined solely by the HDP – based only on time (t) and frequency (f) and expressed as


The ‘Haemodialysis Product’ = Hours/session x (Sessions/week)2


HDP = t x (f)2


They believed ‘f’ more important than‘t’, though recognised that both are important, inter-related and are the primary determinants of good dialysis


They also ‘debunked’ Kt/V and URR and recommended haemodialysis adequacy be simply expressed simply as t x (f)2









Totally inadequate. Severe Malnutrition




Inadequate. A high % = malnourished




Borderline. Some malnutrition. Poor BP control




The only 3 day/wk schedule ‘proven’ to be sufficient




No data available yet




No data available. Should give easier BP control




Good well-being. BPcontrol good if Na+ limited




Best so far. Phosphate normalized. BP control easy


Modified from Scribner and Oreopoulos: Dial & Transpl. Jan 2002: 13-15


HDP = an attractive concept but

As the HDP embraces ONLY time and frequency (consistent with my own beliefs) … why am I still uneasy that HDP is the right answer?

Optimum dialysis is and must be more …

Optimum HD can only be modeled by including:

Better clearance data – but more than just urea

Inclusion of morbidity data – diabetes, age, CV disease, Ca/PO4 product etc

Inclusion of functional data – activity capacity, rehabilitation

… and, the unmentionable – expenditure impact!



Towards optimum haemodialysis


Abandon ‘adequate’ and aim for ‘optimum’


Recognise the flaws of Kt/V and URR


Accept the lack of a true measure of optimum HD


Focus on time and frequency rather than simply on process


Consider using ‘smart’ computing modelling software on accumulated databases (eg: multi-variate analysis or, possibly better yet, artificial intelligence through ‘machine learning’ programs)


Incorporate functional (FSI) and outcome measure (OMI) data





Have I answered the question – “What is ‘optimal’ haemodialysis?” …


Sadly, I don’t think I have …


Moreover, I don’t think I yet can …


… but, this discussion may have rattled a few old skeletons, have provided some food for thought and have suggested a more inclusory process to develop for the future.

… if it has done so, it has been a useful exercise



 Back Home Next


Authored by Prof John Agar. Copyright © 2012
Nocturnal Haemodialysis Program, Barwon Health.
All rights reserved. Revised: July 1st 2012