- Email: [email protected]
Introduction: Kinetic Modeling
Advances in Renal Replacement Therapy OCTOBER 1995
VOL 2, NO 4
Introduction: Kinetic Modeling When you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind: it may be the beginning of knowledge, but you have scarcely, in your thoughts, advanced to the stage of science. William Thompson, Lord Kelvin
Popular Lectures and Addresses, 1893
T
his often-quoted passage is now more than 100 years old but conveys a message for health care workers in the present. Both hemodialysis and the current application of peritoneal dialysis were developed empirically; that is, they were applied and then found to work. Neither systematic animal work nor estimates of how much dialysis would be required preceded the application of either of these treatments. In fact, the outcome of continuous ambulatory peritoneal dialysis (CAPO) surprised both the physicians and the patients. 1 When hemodialysis first became available in the 1960s, the urgency of treating the thousands of patients dying of end-stage kidney disease usurped the need to quantitate the treatment, so hemodialysis was administered according to a universal formula that dictated 6 to 8 hours of dialysis per treatment three times per week for all patients. For CAPO, which came later, the standard was four 2-liter exchanges per day. In the era before Medicare funding in the United States, some hemodialysis centers dialyzed twice weekly to accommodate more patients. 2 The difference in quality of life and health between those dialyzed twice weekly and those dialyzed thrice weekly was obvious even to the casual observer, so thrice-weekly dialysis became the standard. The duration of dialysis seemed to make a difference, yet,
within certain limits, the blood flow and dialysate flow did not seem to matter. These observations, together with the finding that the symptoms and signs of uremia could be alleviated even when the blood urea nitrogen (BUN) was not 10wered,3,4 led to the squaremeter-hour hypothesis and its derivative, the middle molecule hypothesis,5,6 which placed emphasis on gradient-limited diffusion across dialysis membranes. These working hypotheses satisfied a need for scientific logic, seemed to make sense, and were used to justify the semiquantitative approach to both hemodialysis and peritoneal dialysis. As dialysis equipment and membranes improved, the middle molecule hypothesis begged for experimental justification. The U.S. National Cooperative Dialysis Study (NCDS), sponsored by the National Institutes of Health and completed in 1980, provided evidence that urea removal was an important outcome correlate and that dialysis conditions designed to favor removal of middle molecules were of relatively little consequence? The spotlight began to focus on quantitation of urea removal during hemodialysis.s Manufacturers, who had concentrated their efforts on developing larger, thinner, and more porous membranes to remove middle molecules, redirected their attention toward removing smaller solutes. This may have been more of a philosophical redirection, because the newer membranes also removed urea and other small solutes more efficiently. Quantitation, however, proved to be a stumbling block for many. The concern about the serum concentration of hypothetical middle molecules changed to a concern about serum urea levels (BUN). A reanalysis of the NCDS by Gotch and Sargent in 1985 showed the fallacy of relying on the
Advances in Renal Replacement Therapy, Vol 2, No 4 (October), 1995: pp 283-286
283
284
Thomas A. Depner
BUN alone. 8 This came as little surprise to clinicians who, because of their experience with patients during the pre dialysis state of advancing renal failure, viewed the BUN with suspicion. In the pre dialysis state, BUN is determined by renal tubular reabsorption and protein nitrogen catabolism as well as the glomerular filtration rate (GFR), so when patients appear with obvious symptoms of uremia they have a spectrum of modestly to greatly elevated BUN. Because GFR is a measure of overall kidney function that correlates with outcome, clinicians relied more on the serum creatinine concentration as a marker for GFR, which is much less influenced by tubular reabsorption and is not affected by protein nitrogen catabolism. Serum creatinine concentrations, however, are positively related to muscle mass, and once dialysis starts, muscle mass becomes a more important predictor of outcome.9 The logical approach of monitoring serum levels of uremic toxins or their surrogates proved to be unreliable as a predictor of outcome. Neither serum urea nor serum creatinine concentrations predicted outcome as well as the fractional dialyzer clearance in patients with end-stage renal disease (ESRD). The focus on dialyzer urea clearance instead of urea levels in the dialysis patient is analogous to the focus on native kidney clearance (as reflected in the serum creatinine or creatinine clearance) instead of the BUN in the pre dialysis patient as a better indicator of outcome. After extensive evaluation of the NCDS data, attention shifted from the patient's urea concentration to fractional urea clearance by . the dialyzer. The NCDS showed that delivering a minimal amount of dialysis during each treatment was associated with fewer hospitalizations and dropouts for medical reasons.1° The amount of dialysis was quantitated as urea clearance adjusted for patient size (fractional clearance or K/V) multiplied by dialysis duration (t). Urea was the logical choice as a marker of dialyzer clearance because it is present in relatively high and easily measured concentrations in the serum, it is easily dialyzed, and serum urea concentrations also reflect protein catabolism, which the NCDS showed was another predictor of outcome.
Fractional Urea Clearance: Kt/V When clearance in milliliters per minute is divided by the patient's urea distribution volume (V) in milliliters, the result is a fractional clearance expressed per minute. When multiplied by the duration of dialysis in minutes, the result is a fractional clearance expressed per dialysis. Both K/V and Kt/V are measures of fractional clearance; the first is expressed per unit of time, and the second is expressed per dialysis. Kt/V is, in essence, a measure of the effectiveness of dialysis during a single treatment. It is difficult to measure directly, but fortunately it can easily be measured indirectly by examining urea kinetics. Direct measurement would require knowledge of the dialyzer clearance and any changes throughout the treatment, an accurate measure of total body water as the volume of urea distribution, and an accurate measure of dialysis duration. Indirect measurement requires only a measure of predialysis BUN (Co) and postdialysis BUN (C). Analysis of urea kinetics (see below) during hemodialysis shows that Kt/V is primarily a function of the log ratio of Co/C. The difference between using this approach and using the BUN as a predictor of outcome is that the parameter used for Kt/V is derived from the ratio of predialysis to postdialysis BUN independent of their absolute values. The ratio is a measure of effective dialyzer clearance, not toxin level. Target values for Kt/V were set in mid-1989 based on the NCDS data and were increased recently as a result of a National Institutes of Health (NIH) consensus derived from uncontrolled but consistent data that suggest higher values are beneficial. In this issue, Marcia Keen and Gerald Schulman discuss the various standards that have been used, limitations of the NCDS standard, and new insights that are sought from an ongoing NIH-sponsored study of hemodialysis morbidity and mortality.
Kinetic Modeling The term kinetic modeling tends to evoke an image of excitement or excessive animation in the modelers. A more proper term is modeling solute kinetics, because the animation is in the solute not in the modeler. Regardless, the term kinetic modeling seems to be irrevocably
Introduction: Kinetic Modeling
established, so we must define it. In essence, kinetic modeling is a mathematical process of fitting measured serum concentrations to a series of mathematical equations that constitute the model and that include all the forces known to influence the serum concentration. By fitting measured data to the equations, the parameters that affect the serum concentrations, such as dialyzer clearance (Kd ), urea generation rate (G), protein catabolic rate (PCR), and urea distribution volume (V), can be assigned specific values. The modeling process may refer either to development of the model, first attempted by Wolf in 1951,11 or to application of the model. Development is usually left in the hands of research specialists, including nephrologists, physiologists, and mathematicians. Application of the model can be accomplished by anyone with the proper tools. In the modern era, these tools consist of computer programs capable of iterative solutions to the mathematical equations, some of which cannot be solved explicitly by simple algebra. In this issue of ARRT devoted to dialysis kinetic modeling, several approaches to quantitation are outlined. Some require a computer, and others require only simple arithmetic. In general, the more complex programs provide more accurate analyses and more complete data such as Kct, G, PCR, and V. The simplified approaches described by John Daugirdas in this issue are especially useful for teaching and conceptualizing, for analysis of large populations in which minimal data are available, and for dialysis centers where computer facilities are not readily available. The approach based on measurement of dialysate concentrations outlined by Laurie Garred holds promise as an automated technique that will not only quantify each dialysis but also guide the dialysis provider in real time to assure the adequacy of each treatment. In the earlier days of hemodialysis, much attention was necessarily given to the equipment and less to the patient. In more recent years we have turned more of our attention to the patient and especially to the individual's response to dialysis. Kinetic modeling is an example of this individualization of dialysis. Factors such as recirculation of dialyzed blood in the access device, cardiopulmonary recirculation, fractional water volume, fluid gain, and
285
protein catabolic rates are detected or determined individually with the help of kinetic modeling so that the prescription can be adjusted to fit the patient. In this issue, Allen Kaufman et al discuss the problem of disequilibrium in the patient and how it distorts the classic single-compartment model of urea kinetics. They also discuss the recently discovered flow-related disequilibrium, including cardiopulmonary recirculation, and how to include these effects in the kinetic model.
Peritoneal Dialysis Kinetic modeling is especially valuable for hemodialysis, where the perturbations in urea concentration caused by the dialysis itself provide an opportunity to estimate certain critical parameters such as the patient's water volume and urea generation from simple measurements of serum urea concentrations. For peritoneal dialysis, Kt/V and PCR can also be measured, but the process is more direct; it requires collection of dialysate and less modeling of blood concentrations. Leonor Ponferrada and John Van Stone in this issue discuss the history, theory, and application of kinetic modeling as it applies to peritoneal dialysis. They devote special attention to the peritoneal equilibration test, methods of calculation, and to the standards or target values for Kt/V, which are derived from a consensus for lack of a controlled study such as the NCDS of hemodialyzed patients.
Protein Catabolic Rate (PCR) Although the primary goal of dialysis kinetic modeling is to quantitate and prescribe dialysis, a secondary benefit is the measurement of each patient's net protein catabolic rate from their urea generation rate. This parameter provides an estimate of the patient's dietary sufficiency as well as another outcome predictor. The positive although weaker correlation between PCR normalized to ideal body weight (PCRn) and outcome shown by the NCDS was unexpected and difficult to explain initially, because protein intake tends to increase the BUN, which correlated negatively with outcome (Fig 1). We continue to be uncertain about how to interpret PCRn with respect to
286
Thomas A. Depner
I MORBIDITY I -T'
?j
/ ,-----~
Low
Dose Dialysis
~
Reduced Protein Catabolism
"
""Other
r------,
+---
period is now in progress. This study, known as the HEMO Project, which focuses on the dialysis prescription and the patient's response to hemodialysis, will, we hope, answer some of the important questions posed by the experts in this issue of ARRT.
Disease States
Figure 1. The role of protein catabolism as an outcome determinant may be direct as shown by the dotted line or indirect as shown by the solid lines. Protein catabolism (peR) is derived from the urea generation rate obtained from analysis of dialysis urea kinetics.
outcome as shown in Figure 1, but the suspicion is that more attention should be given to patients with extreme values of PCRn, especially when PCRn is low. In this issue, Michael Flannigan, Victoria Lim, and Joetta Redlin provide an extensive discussion of PCR, of dietary protein, and of serum albumin concentrations as they relate to energy metabolism and protein nutrition, including evidence that protein and caloric nutrition in general have improved since the early days of dialysis. Most of the authors in this issue allude to the NCDS as a seminal and perhaps the only controlled study of dialysis adequacy. This NIH-sponsored study, however, was not designed specifically as a guide for clinicians to quantitate dialysis. As mentioned, its focus was on middle molecules and distinguishing membrane-limited (larger) from flow-limited (small) solute removal. Analysis of the data provides some guidelines for achievement of short-term goals, but determinants of longterm outcome in the current population using modern dialysis membranes, dialysate, and delivery systems are largely speculative. A much-needed NIH-sponsored study of multicentered hemodialyzed patients over a longer
Thomas A. Depner Guest Editor
References 1. Popovich RP, Moncrief JW, Decherd JF, et al: The definition of a novel portable/wearable equilibrium peritoneal dialysis technique. Trans Am Soc Artif Intern Organs 5:64, 1976 (abstr) 2. Barber 5, Appleton DR, Kerr DNS: Adequate dialysis. Nephron 14:209-227, 1975 3. Merrill JP, Legrain M, Hoigne R: Observations on the role of urea in uremia. AmJ Med 14:519-520, 1953 4. Johnson WI, Hagge WW, Wagoner RD, et al: Effects of urea loading in patients with far-advanced renal failure. Mayo Clin Proc 47:21-29, 1972 5. Babb AL, Popovich RP, Christopher TG, et al: The genesis of the square meter-hour hypothesis. Trans Am Soc Artif Intern Organs 17:81-91, 1971 6. Scribner BH, Farrell PC, Milutinovic I, et al: Evolution of the middle molecule hypothesis, in Villarreal H (ed): Proceedings of the Fifth International Congress of Nephrology. Basel, Karger, 1974, pp 190-199 7. Lowrie EG, Laird NM, Parker TF, et al: Effect of the hemodialysis prescription on patient morbidity: Report from the National Cooperative Dialysis Study. N Engl J Med 305:1176-1181,1981 8. Gotch FA, Sargent JA: A mechanistic analysis of the National Cooperative Dialysis Study (NCDS). Kidney Int 28:526-534, 1985 9. Lowrie EG, Lew NL: Death risk in hemodialysis patients: The predictive value of commonly measured variables and an evaluation of death rate differences between facilities. Am J Kidney Dis 15:458-482, 1990 10. Laird NM, Berkey CS, Lowrie EG: Modeling success or failure of dialysis therapy: The National Cooperative Dialysis Study. Kidney Int 23:5101-5106, 1983 (suppl13) 11. Wolf AV, Remp DG, Kiley JE, et al: Artificial kidney function: Kinetics of hemodialysis. J Clin Invest 30: 1062-1070,1951
VOL 2, NO 4
Introduction: Kinetic Modeling When you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it, when you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind: it may be the beginning of knowledge, but you have scarcely, in your thoughts, advanced to the stage of science. William Thompson, Lord Kelvin
Popular Lectures and Addresses, 1893
T
his often-quoted passage is now more than 100 years old but conveys a message for health care workers in the present. Both hemodialysis and the current application of peritoneal dialysis were developed empirically; that is, they were applied and then found to work. Neither systematic animal work nor estimates of how much dialysis would be required preceded the application of either of these treatments. In fact, the outcome of continuous ambulatory peritoneal dialysis (CAPO) surprised both the physicians and the patients. 1 When hemodialysis first became available in the 1960s, the urgency of treating the thousands of patients dying of end-stage kidney disease usurped the need to quantitate the treatment, so hemodialysis was administered according to a universal formula that dictated 6 to 8 hours of dialysis per treatment three times per week for all patients. For CAPO, which came later, the standard was four 2-liter exchanges per day. In the era before Medicare funding in the United States, some hemodialysis centers dialyzed twice weekly to accommodate more patients. 2 The difference in quality of life and health between those dialyzed twice weekly and those dialyzed thrice weekly was obvious even to the casual observer, so thrice-weekly dialysis became the standard. The duration of dialysis seemed to make a difference, yet,
within certain limits, the blood flow and dialysate flow did not seem to matter. These observations, together with the finding that the symptoms and signs of uremia could be alleviated even when the blood urea nitrogen (BUN) was not 10wered,3,4 led to the squaremeter-hour hypothesis and its derivative, the middle molecule hypothesis,5,6 which placed emphasis on gradient-limited diffusion across dialysis membranes. These working hypotheses satisfied a need for scientific logic, seemed to make sense, and were used to justify the semiquantitative approach to both hemodialysis and peritoneal dialysis. As dialysis equipment and membranes improved, the middle molecule hypothesis begged for experimental justification. The U.S. National Cooperative Dialysis Study (NCDS), sponsored by the National Institutes of Health and completed in 1980, provided evidence that urea removal was an important outcome correlate and that dialysis conditions designed to favor removal of middle molecules were of relatively little consequence? The spotlight began to focus on quantitation of urea removal during hemodialysis.s Manufacturers, who had concentrated their efforts on developing larger, thinner, and more porous membranes to remove middle molecules, redirected their attention toward removing smaller solutes. This may have been more of a philosophical redirection, because the newer membranes also removed urea and other small solutes more efficiently. Quantitation, however, proved to be a stumbling block for many. The concern about the serum concentration of hypothetical middle molecules changed to a concern about serum urea levels (BUN). A reanalysis of the NCDS by Gotch and Sargent in 1985 showed the fallacy of relying on the
Advances in Renal Replacement Therapy, Vol 2, No 4 (October), 1995: pp 283-286
283
284
Thomas A. Depner
BUN alone. 8 This came as little surprise to clinicians who, because of their experience with patients during the pre dialysis state of advancing renal failure, viewed the BUN with suspicion. In the pre dialysis state, BUN is determined by renal tubular reabsorption and protein nitrogen catabolism as well as the glomerular filtration rate (GFR), so when patients appear with obvious symptoms of uremia they have a spectrum of modestly to greatly elevated BUN. Because GFR is a measure of overall kidney function that correlates with outcome, clinicians relied more on the serum creatinine concentration as a marker for GFR, which is much less influenced by tubular reabsorption and is not affected by protein nitrogen catabolism. Serum creatinine concentrations, however, are positively related to muscle mass, and once dialysis starts, muscle mass becomes a more important predictor of outcome.9 The logical approach of monitoring serum levels of uremic toxins or their surrogates proved to be unreliable as a predictor of outcome. Neither serum urea nor serum creatinine concentrations predicted outcome as well as the fractional dialyzer clearance in patients with end-stage renal disease (ESRD). The focus on dialyzer urea clearance instead of urea levels in the dialysis patient is analogous to the focus on native kidney clearance (as reflected in the serum creatinine or creatinine clearance) instead of the BUN in the pre dialysis patient as a better indicator of outcome. After extensive evaluation of the NCDS data, attention shifted from the patient's urea concentration to fractional urea clearance by . the dialyzer. The NCDS showed that delivering a minimal amount of dialysis during each treatment was associated with fewer hospitalizations and dropouts for medical reasons.1° The amount of dialysis was quantitated as urea clearance adjusted for patient size (fractional clearance or K/V) multiplied by dialysis duration (t). Urea was the logical choice as a marker of dialyzer clearance because it is present in relatively high and easily measured concentrations in the serum, it is easily dialyzed, and serum urea concentrations also reflect protein catabolism, which the NCDS showed was another predictor of outcome.
Fractional Urea Clearance: Kt/V When clearance in milliliters per minute is divided by the patient's urea distribution volume (V) in milliliters, the result is a fractional clearance expressed per minute. When multiplied by the duration of dialysis in minutes, the result is a fractional clearance expressed per dialysis. Both K/V and Kt/V are measures of fractional clearance; the first is expressed per unit of time, and the second is expressed per dialysis. Kt/V is, in essence, a measure of the effectiveness of dialysis during a single treatment. It is difficult to measure directly, but fortunately it can easily be measured indirectly by examining urea kinetics. Direct measurement would require knowledge of the dialyzer clearance and any changes throughout the treatment, an accurate measure of total body water as the volume of urea distribution, and an accurate measure of dialysis duration. Indirect measurement requires only a measure of predialysis BUN (Co) and postdialysis BUN (C). Analysis of urea kinetics (see below) during hemodialysis shows that Kt/V is primarily a function of the log ratio of Co/C. The difference between using this approach and using the BUN as a predictor of outcome is that the parameter used for Kt/V is derived from the ratio of predialysis to postdialysis BUN independent of their absolute values. The ratio is a measure of effective dialyzer clearance, not toxin level. Target values for Kt/V were set in mid-1989 based on the NCDS data and were increased recently as a result of a National Institutes of Health (NIH) consensus derived from uncontrolled but consistent data that suggest higher values are beneficial. In this issue, Marcia Keen and Gerald Schulman discuss the various standards that have been used, limitations of the NCDS standard, and new insights that are sought from an ongoing NIH-sponsored study of hemodialysis morbidity and mortality.
Kinetic Modeling The term kinetic modeling tends to evoke an image of excitement or excessive animation in the modelers. A more proper term is modeling solute kinetics, because the animation is in the solute not in the modeler. Regardless, the term kinetic modeling seems to be irrevocably
Introduction: Kinetic Modeling
established, so we must define it. In essence, kinetic modeling is a mathematical process of fitting measured serum concentrations to a series of mathematical equations that constitute the model and that include all the forces known to influence the serum concentration. By fitting measured data to the equations, the parameters that affect the serum concentrations, such as dialyzer clearance (Kd ), urea generation rate (G), protein catabolic rate (PCR), and urea distribution volume (V), can be assigned specific values. The modeling process may refer either to development of the model, first attempted by Wolf in 1951,11 or to application of the model. Development is usually left in the hands of research specialists, including nephrologists, physiologists, and mathematicians. Application of the model can be accomplished by anyone with the proper tools. In the modern era, these tools consist of computer programs capable of iterative solutions to the mathematical equations, some of which cannot be solved explicitly by simple algebra. In this issue of ARRT devoted to dialysis kinetic modeling, several approaches to quantitation are outlined. Some require a computer, and others require only simple arithmetic. In general, the more complex programs provide more accurate analyses and more complete data such as Kct, G, PCR, and V. The simplified approaches described by John Daugirdas in this issue are especially useful for teaching and conceptualizing, for analysis of large populations in which minimal data are available, and for dialysis centers where computer facilities are not readily available. The approach based on measurement of dialysate concentrations outlined by Laurie Garred holds promise as an automated technique that will not only quantify each dialysis but also guide the dialysis provider in real time to assure the adequacy of each treatment. In the earlier days of hemodialysis, much attention was necessarily given to the equipment and less to the patient. In more recent years we have turned more of our attention to the patient and especially to the individual's response to dialysis. Kinetic modeling is an example of this individualization of dialysis. Factors such as recirculation of dialyzed blood in the access device, cardiopulmonary recirculation, fractional water volume, fluid gain, and
285
protein catabolic rates are detected or determined individually with the help of kinetic modeling so that the prescription can be adjusted to fit the patient. In this issue, Allen Kaufman et al discuss the problem of disequilibrium in the patient and how it distorts the classic single-compartment model of urea kinetics. They also discuss the recently discovered flow-related disequilibrium, including cardiopulmonary recirculation, and how to include these effects in the kinetic model.
Peritoneal Dialysis Kinetic modeling is especially valuable for hemodialysis, where the perturbations in urea concentration caused by the dialysis itself provide an opportunity to estimate certain critical parameters such as the patient's water volume and urea generation from simple measurements of serum urea concentrations. For peritoneal dialysis, Kt/V and PCR can also be measured, but the process is more direct; it requires collection of dialysate and less modeling of blood concentrations. Leonor Ponferrada and John Van Stone in this issue discuss the history, theory, and application of kinetic modeling as it applies to peritoneal dialysis. They devote special attention to the peritoneal equilibration test, methods of calculation, and to the standards or target values for Kt/V, which are derived from a consensus for lack of a controlled study such as the NCDS of hemodialyzed patients.
Protein Catabolic Rate (PCR) Although the primary goal of dialysis kinetic modeling is to quantitate and prescribe dialysis, a secondary benefit is the measurement of each patient's net protein catabolic rate from their urea generation rate. This parameter provides an estimate of the patient's dietary sufficiency as well as another outcome predictor. The positive although weaker correlation between PCR normalized to ideal body weight (PCRn) and outcome shown by the NCDS was unexpected and difficult to explain initially, because protein intake tends to increase the BUN, which correlated negatively with outcome (Fig 1). We continue to be uncertain about how to interpret PCRn with respect to
286
Thomas A. Depner
I MORBIDITY I -T'
?j
/ ,-----~
Low
Dose Dialysis
~
Reduced Protein Catabolism
"
""Other
r------,
+---
period is now in progress. This study, known as the HEMO Project, which focuses on the dialysis prescription and the patient's response to hemodialysis, will, we hope, answer some of the important questions posed by the experts in this issue of ARRT.
Disease States
Figure 1. The role of protein catabolism as an outcome determinant may be direct as shown by the dotted line or indirect as shown by the solid lines. Protein catabolism (peR) is derived from the urea generation rate obtained from analysis of dialysis urea kinetics.
outcome as shown in Figure 1, but the suspicion is that more attention should be given to patients with extreme values of PCRn, especially when PCRn is low. In this issue, Michael Flannigan, Victoria Lim, and Joetta Redlin provide an extensive discussion of PCR, of dietary protein, and of serum albumin concentrations as they relate to energy metabolism and protein nutrition, including evidence that protein and caloric nutrition in general have improved since the early days of dialysis. Most of the authors in this issue allude to the NCDS as a seminal and perhaps the only controlled study of dialysis adequacy. This NIH-sponsored study, however, was not designed specifically as a guide for clinicians to quantitate dialysis. As mentioned, its focus was on middle molecules and distinguishing membrane-limited (larger) from flow-limited (small) solute removal. Analysis of the data provides some guidelines for achievement of short-term goals, but determinants of longterm outcome in the current population using modern dialysis membranes, dialysate, and delivery systems are largely speculative. A much-needed NIH-sponsored study of multicentered hemodialyzed patients over a longer
Thomas A. Depner Guest Editor
References 1. Popovich RP, Moncrief JW, Decherd JF, et al: The definition of a novel portable/wearable equilibrium peritoneal dialysis technique. Trans Am Soc Artif Intern Organs 5:64, 1976 (abstr) 2. Barber 5, Appleton DR, Kerr DNS: Adequate dialysis. Nephron 14:209-227, 1975 3. Merrill JP, Legrain M, Hoigne R: Observations on the role of urea in uremia. AmJ Med 14:519-520, 1953 4. Johnson WI, Hagge WW, Wagoner RD, et al: Effects of urea loading in patients with far-advanced renal failure. Mayo Clin Proc 47:21-29, 1972 5. Babb AL, Popovich RP, Christopher TG, et al: The genesis of the square meter-hour hypothesis. Trans Am Soc Artif Intern Organs 17:81-91, 1971 6. Scribner BH, Farrell PC, Milutinovic I, et al: Evolution of the middle molecule hypothesis, in Villarreal H (ed): Proceedings of the Fifth International Congress of Nephrology. Basel, Karger, 1974, pp 190-199 7. Lowrie EG, Laird NM, Parker TF, et al: Effect of the hemodialysis prescription on patient morbidity: Report from the National Cooperative Dialysis Study. N Engl J Med 305:1176-1181,1981 8. Gotch FA, Sargent JA: A mechanistic analysis of the National Cooperative Dialysis Study (NCDS). Kidney Int 28:526-534, 1985 9. Lowrie EG, Lew NL: Death risk in hemodialysis patients: The predictive value of commonly measured variables and an evaluation of death rate differences between facilities. Am J Kidney Dis 15:458-482, 1990 10. Laird NM, Berkey CS, Lowrie EG: Modeling success or failure of dialysis therapy: The National Cooperative Dialysis Study. Kidney Int 23:5101-5106, 1983 (suppl13) 11. Wolf AV, Remp DG, Kiley JE, et al: Artificial kidney function: Kinetics of hemodialysis. J Clin Invest 30: 1062-1070,1951