1 Nephrology and Dialysis Unit, Provincial Hospital, Camposampiero, Padova, 2 Nephrology and Dialysis Division, Regional Hospital, Treviso, 3 Laboratory, Provincial Hospital, Camposampiero, Padova and 4 Laboratory, Regional Hospital, Treviso, Italy
Correspondence and offprint requests to: Giovambattista Virga, Nephrologia e Dialisi, Ospedale P. Cosma, Via P. Cosma, 35012 Camposampiero, Padova, Italy.
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Methods. Twenty-four patients (14 CAPD, 10 APD) at two centres were studied by means of a triple dialysate and urine collection for a period of 1 week. Variability in the findings for a given patient was expressed by the coefficient of variation (CV%) calculated for peritoneal (p), renal, and total (tot) adequacy parameters. The target Kt/V and CCr values were recalculated on the basis of variability.
Results. Kt/V was less variable (CV 4.0 and 4.4% for peritoneal Kt/V (pKt/V) and total Kt/V (totKt/V) respectively) than CCr (4.7 and 6.0% for peritoneal creatinine clearance (pCCr) and total creatinine clearance (totCCr) respectively) and proved to be a more reliable indicator of adequacy in terms of the CV. Both variability parameters became worse if renal clearance was added to peritoneal clearance. CV in APD proved to be no different from CAPD for all the parameters considered. In our centres dialysis adequacy target correction for variability provided safe values for weekly Kt/V (pKt/V=1.782.10 and totKt/V=1.822.15 target 1.72.0) and CCr/1.73 (pCCr=53.764.4 l and totCCr=55.166.1 l; target 5060 l).
Conclusions. Evaluating the adequacy of PD by means of a single measurement should take into account the weekly variability as demonstrated by a triple dialysate and urine collection. Standard adequacy targets can be corrected to allow for variability. Thus one can obtain safe values for prescription decisions based on a single collection result.
Keywords: adequacy target; coefficient of variation; creatinine clearance; Kt/V; peritoneal dialysis
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Initially, a theoretical approach set the target level for weekly Kt/V and CCr at 2.0 and 60 l, respectively [6], whereas the clinical approach suggested 1.7 and 50 l respectively [7]. Recently, the CANUSA study on CAPD suggested that higher clearances are associated with a better survival and lower morbidity, so weekly total Kt/V from 2.1 to 2.3 and CCr from 70 to 80 l were each associated with an improvement in the expected 2-year survival from 78 to 81% [5]. The latest and most thorough PD guidelines, published as the Dialysis Outcome Quality Initiative by the National Kidney Foundation [8], have set the weekly targets at 2.0 for Kt/V and 60 l for CCr in CAPD with an increase of 5% for CCPD and of 10% for NIPD based on the opinion of the major experts that intermittent or variably efficient treatments need an elevation of the targets [8].
It is considered optimal practice in PD to collect biological fluids for the calculation of adequacy parameters every 4 months [8], but this is usually done no more than twice a year. A prescription decision based on a single measurement can thus influence clinical outcome for many months.
In general, the evaluation of a measurement is based on the concept of accuracy, which can be split into bias and precision. Adequacy studies based on direct measurements are generally considered as reference or true values (there being no external reference available), so it is impossible to assess bias and the accuracy coincides with the precision (the time-to-time variability of the test). As a result, only the scatter or variability in repeated measurements can be assessed. Data have been published on the variability in triple dialysis adequacy measurements over 1 week of PD and demonstrated a variability judged to be clinically significant [9].
The aim of our study was to assess the day-to-day variability of common dialysis adequacy parameters and to evaluate its impact on the adequacy indexes in PD.
![]() |
Subjects and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Patients on APD were treated for 810 h/night (tidal 5075%) and 02 daily dwells while those on CAPD had four exchanges every 24 h. The daily volume prescription was 21.1±3.3 l in APD (range 14.525.8 l) and 8.5±0.7 l in CAPD (range 6.810.4 l). Four patients on APD were anuric.
During the week of the study, the dialysis prescription remained unchanged. All patients were metabolically stable, had been free from acute disease or peritonitis for at least 3 months, and had performed a standard peritoneal equilibration test [10] no more than 6 months before the study. All patients enrolled in the study gave their informed consent.
Study procedure
Dialysate and urine were collected for 24 h three times in one week by all patients. Samples of dialysate and urine were obtained after they had been weighed and mixed. Dialysate volume was assumed to equate to its weight (1 kg=1 l), disregarding any difference between dialysate and distilled water. Dialysate was analysed for urea, Cr, and glucose, and urine was analysed for urea and Cr, using standard laboratory methods.
For the Camposampiero patients the dialysate Cr was corrected for glucose with an over-estimation correction factor of 0.0001806 mg/dl of Cr every mg/dl of glucose, while for the Treviso patients an enzymatic assay was used with no need to adjust for glucose interference (Bayer, Tarrytown, NY, USA). At the time of each biological fluid collection, patients were weighed and a blood sample was obtained for Cr and urea assay in steady-state conditions (at 8 a.m., fasting, for CAPD patients; between 2 and 5 p.m., not fasting, for APD patients).
Eight parameters were studied: peritoneal (pCCr) and total creatinine clearance (totCCr), peritoneal, renal, and total Kt/V (pKt/V, rKt/V, totKt/V), glomerular filtration rate (GFR), diuresis, and ultrafiltration (UF). GFR, pCCr and totCCr were all normalized for a 1.73 m2 body surface area (BSA). BSA was calculated using the du Bois formula [11] and body water (V) was calculated using the Watson formula [12]. The clearances were expressed as weekly values calculated by multiplying the 24-h value by 7. UF (calculated as the difference between 24-h drainage and load volume) and diuresis were expressed as daily values (ml/day) and were not normalized for BSA or V. GFR (l/week) was estimated as an average of urea and Cr renal clearances [13]. totCCr was calculated adding pCCr to GFR to approximate the amount of renal CCr due to glomerular filtration excluding tubular excretion [14].
Analytical (inter-assay) variability was assessed using a triple assay of urea and Cr over a period of 1 week in urine, blood, and dialysate samples obtained from a different group of 24 uraemic and PD patients (Camposampiero 19, Treviso 5) in order to investigate separately the influence of laboratory variability.
Statistical analysis
Continuous variables were expressed as means (M)±standard deviation (SD) or as medians (interquartile range) if data were not normally distributed.
Variability between a given patient's three assessments was expressed by the coefficient of variation (CV%=SD/Mx100) calculated for each study parameter. CV was also calculated for BSA and V. Individual CVs for each parameter were summarized using the median value (interquartile range) and its 95% confidence interval [15] and expressed study population variability between the three collections. A comparison between APD CVs and CAPD CVs was drawn using the non-parametric MannWhitney U-test.
It is assumed that the variability observed randomly affects every single adequacy measurement, but we considered only the positive portion as clinically dangerous because a single false high measurement can lead to a prescription being considered adequate when in fact it is not. Consequently, the standard target values assumed from the literature for Kt/V and CCr were recalculated to give `safe' target values, including the upper limit of the 95% confidence interval of the median CV value as a correction factor (CF) as follows: safe target=standard target+(standard targetxCF).
The null hypothesis was rejected for all tests with two-tailed alpha values lower than 0.05. JMP 3.02 (SAS, Cary, NC, USA) software and Instat 2.03 (Graphpad, San Diego, CA, USA) software on Macintosh (Apple, Cupertino, CA, USA) hardware were used for the statistical analysis.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
|
|
|
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
A controversial aspect concerns the reliability of a single collection in representing the outcome of a 4-month period of dialysis (if current guidelines are followed [8]) or of a 6-month period if the test is done twice a year, as is commonly the case.
The major problem with these data concerns the day-to-day variability of renal and peritoneal parameters. This variability, which can also be expressed as error or uncertainty, has biological and analytical causes. In fact, it can stem from physiological oscillations and BW and V measurement inaccuracies, or from hypothetical variations in renal function or peritoneal permeability, but some degree of analytical variability in the laboratory dosages might also be taken into account. Moreover, other causes could be involved, such as a discontinuous compliance with therapy prescription or collection instructions, or methodological differences in the weighing, mixing, and sampling of dialysate and urine.
Individual variability in three peritoneal and renal clearances by the direct quantification method is expressed by the SD of this clearance and the best clearance estimation by the M of real values. To make this uncertainty or error comparable, the SD/M ratio (CV) is commonly used, expressed as a percentage [16]. In our study population, CVs were not normally distributed, so we have expressed data as median values and interquartile ranges without making any other assumptions on distributions.
The day-to-day variability observed should be borne in mind in prescribing the dialysis dose on the strength of data based on a single measurement and an adequacy target is considered as a minimum value. The aim of this study on variability in PD is to prompt a better interpretation of adequacy results and guide prescription decisions.
In our study, CCr is affected by a higher variability than Kt/V (Table 3) that appears to be more reliable in terms of the CV. Renal parameters (rKt/V and GFR) demonstrate a higher CV and adding their values to peritoneal clearances makes the variability worse, probably due to a physiological variability in the severely limited renal function of uraemic patients and to errors in urine collection (Table 2
).
Laboratory dosage variability played a significant part in the global variability of adequacy indexes in PD (Table 5). Laboratory dosage variability was 12%, which is far from negligible for PD adequacy parameters that show a CV with a range of 4 to 6% (Table 3
).
Variability between APD and CAPD failed to reveal any statistically significant difference in our study, but CCr in APD demonstrated a high CV, which can be considered as one of the causes of the well-known inconsistency between Kt/V and CCr in PD [17]. This latter aspect of adequacy evaluation in APD is probably more evident for intermittent techniques (NIPDNTPD) than for continuous treatments (CCPDCTPD).
Our population shows a disagreement between pKt/V and pCCr/1.73 (Table 1) that is due mainly to the presence of 10 (42% of the sample) APD patients. In PD, the causes of discrepancy are primarily mathematical, e.g. the different normalization, to V and BSA, of the two indexes and the non-linear relationship between V and BSA [18]. Two other causes of inconsistency are physiological, i.e. the presence of a more than negligible GFR and peritoneal permeability. The former often leads to total CCr/1.73 values appearing adequate without an optimal Kt/V value [19] because if we add peritoneal CL to GFR, the CCr/Kt/V ratio increases because the renal CCr is higher than the renal Kt/V. The latter cause of inconsistency is due to a low peritoneal permeability that affects the transport of Cr more than that of urea, and this condition can suggest an adequate Kt/V with a low CCr/1.73 [17].
In APD, both short dwell times [20] and low peritoneal permeability [21] lower the CCr/Kt/V ratio. In anuric CTPD patients, a discrepancy between Kt/V and CCr/1.73 has only been demonstrated in patients with a peritoneal permeability lower than the mean [22]. In short, GFR and high peritoneal permeability tend to increase CCr/Kt/V while short dwell times and low peritoneal permeability make it lower.
The only paper published, to our knowledge, on day-to-day variability in the adequacy of PD is by Rodby et al. [9]. The greater variation in renal parameters is confirmed by their data, too, so both Kt/V and CCr variability become worse if renal clearance is combined with peritoneal clearance findings (Table 3). This result is also consistent with the significant relationship between intra-method variability in total Cr excretion and residual renal function reported for multiple (35) collections in children on PD [23]. The CVs calculated by Rodby et al. were about 75% higher than in our study. Unfortunately, UF, GFR, and totCCr with GFR CVs were not reported. To compare our CV for UF we elaborated data published by Fisher et al. [24] on a triple collection to study compliance in PD: the calculated median UF CV of 37.7% (14.853.9) was higher than ours, but to the same extent (75%) as the other parameters considered by Rodby et al.
We corrected our target values by the median CV, using the upper limit of the 95% confidence intervals in order to consider the variability of our PD patient population. Using this approach, we have used a safe value of variability with a 95% precision. For example, if we consider as adequate a minimum pKt/V=1.7 or CCr/1.73=50 l we must accept (considering the variability emerging from this study) a single measurement value of about 1.78 and 53.7 l (Table 6). If we set targets at 2.0 for totKt/V and 60 l for totCCr in CAPD, following DOQI guidelines [8], only a dialysis prescription able to obtain 2.15 and 66.1 l respectively from a single measurement can ensure that the real adequacy values are not below the targets (Table 6
). The higher values thus calculated represent not new targets, but `safe' values that guarantee the minimum clearances targeted.
In conclusion, our study suggests that assessments of the adequacy of PD should take the variability phenomenon into account. Kt/V seems to be more reliable in terms of variability than CCr, considering only peritoneal, renal, and total clearances. To ensure that the adequacy target is safely reached in PD it would be wise to use an appropriate correction for each value if, as is usually the case, dialysis quantification is evaluated on the basis of a single measurement.
![]() |
Acknowledgments |
---|
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|