 |
||||
| 2007 JASN IMPACT FACTOR 7.111 | HOME AUTHOR INFO EDITORIAL BOARD SUBSCRIBE FEEDBACK ALERTS HELP | |||
| CURRENT ISSUE | ARCHIVES | JASN Express | ONLINE SUBMISSION | |
|
||||
| 2007 JASN IMPACT FACTOR 7.111 | HOME AUTHOR INFO EDITORIAL BOARD SUBSCRIBE FEEDBACK ALERTS HELP | |||
| CURRENT ISSUE | ARCHIVES | JASN Express | ONLINE SUBMISSION | |
National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland.
Correspondence to Dr. Alfred K. Cheung, Dialysis Program, 85 North Medical Drive East, Room 201, University of Utah, Salt Lake City, UT 84112. Phone: 801-581-6427; Fax: 801-581-4750; E-mail: alfred.cheung{at}hsc.utah.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
In addition to variations in BP, it is conceivable that seasonal changes might affect other bodily functions among dialysis patients, as observed for the general population. These physiologic changes would be expected to affect, in turn, the results of clinical and laboratory evaluations of the dialysis patients. This study was performed with a large cohort of chronic hemodialysis patients, to examine whether seasonal variations in 21 selected clinical and laboratory variables could be observed.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
|
Data Collection
Twenty-one variables being regularly monitored in the HEMO Study were examined for seasonal variations. The frequency and total number of measurements and the number of patients used in the analysis of each variable are presented in Table 1. Some variables, such as BP, were determined more frequently as part of routine clinical practice; they were reported to the Data Coordinating Center at regular but less frequent intervals. At least two measurements/patient per yr were required for all variables that were analyzed longitudinally. The mean number of measurements per patient during the study ranged from >20 for weight and BP to 2.03 for protein and energy intakes (calculated from Table 1). Because the frequency of measurements for different variables varied according to the HEMO Study protocol, the number of patients examined for each variable varied between 1172 and 1445.
|
Seven laboratory variables (serum creatinine, sodium, potassium, bicarbonate, and phosphorus levels, hematocrit, and leukocyte counts) were recorded monthly as part of routine care and were reported to the Data Coordinating Center every 6 mo. The median number of measurements was 4.0 to 5.0/patient and the mean follow-up period was 22.8 ± 11.3 mo (maximum, 45 mo) for these variables. For these seven variables, the first measurements in the analysis were the baseline (prerandomization) values.
The other two variables, namely dietary protein and energy intakes, were estimated from 2-d (one dialysis day and one nondialysis day), diary-assisted, dietary recalls conducted annually. The median number of measurements was 2.0/patient and the mean follow-up period was 14.4 ± 13.5 mo (maximum, 40 mo) for these variables. For these two variables, the first measurements in the analysis were the baseline (prerandomization) values.
Assays and Calculations
Blood samples were immediately centrifuged, and all serum samples were assayed for BUN and albumin levels at a central laboratory (LifeChem, Rockleigh, NJ). BUN levels were measured with an autoanalyzer (Hitachi 747-200; Boehringer Mannheim, Indianapolis, IN), using a kinetic urease method. Albumin levels were measured with a nephelometer (Array360; Beckman, Fullerton, CA), using reagents obtained from Beckman. Determination of hematocrit values, leukocyte counts, and serum creatinine, sodium, potassium, bicarbonate, and phosphorus levels was performed in the clinical laboratories associated with individual clinical centers.
The total ultrafiltration volume was estimated as the difference between the predialysis and postdialysis weights. The ultrafiltration rate was calculated as the total ultrafiltration volume divided by the duration of the dialysis session. The equilibrated normalized protein catabolic rate (enPCR) (23) and the distribution volume of urea (24) were determined by urea kinetic modeling, using predialysis and postdialysis BUN concentrations. Protein, energy, sodium, and potassium intakes were calculated from the dietary recalls by using the Nutritionist IV nutrient analysis program (First DataBank Inc., San Bruno, CA).
Climatic Data
Daily outdoor temperatures at individual clinical centers were obtained from the National Oceanic and Atmospheric Administration. The average of the highest and lowest temperatures for each day was used in subsequent analyses.
Statistical Analyses
The longitudinal variation of each of the 19 clinical and laboratory variables measured at least twice each year was investigated by analyzing the changes in the serial measurements from the initial values for each patient, using mixed-effects models to account for serially correlated data (25). First, the overall pattern of variation of a variable with calendar time was described by plotting the mean changes in the serial measurements estimated by using mixed-effects models, which also accounted for the day of the week of the measurements, the time from randomization, and the interactions with the Kt/V and flux treatment groups. Next, for evaluation of the seasonal variation of a variable, the changes in the serial measurements were related to sine and cosine functions with periods of 12 calendar months, again controlling for the day of the week, the time from randomization, and the interactions with the treatment groups. Standard trigonometric identities used in Fourier analyses were used to convert the coefficients of the sine and cosine functions, for estimation of the amplitude and phase of the seasonal variation for each outcome variable (26). The amplitude indicates the average difference between the peak and trough values, whereas the phase indicates the calendar time of the peak. This approach was feasible even for variables that were measured only twice-yearly for each patient, because the patients were enrolled in the HEMO Study at different times of the calendar year and measurements of each variable throughout the year were therefore available. Because the periodicity was set at 12 mo for the sine and cosine functions, the trough value of the seasonal component of each variable was always estimated, in this assessment of seasonal variation, to occur 6 mo after the peak value. To evaluate the possibility that some variables demonstrated systematic variations with time that deviated from this annual seasonal pattern, nonparametric curves relating each variable to calendar time were constructed by fitting cubic splines (27) to the residuals of the initial mixed-effects analyses that remained after removal of the seasonal effect.
For dietary protein and energy intakes, which were measured annually for each patient, it was not possible to evaluate seasonal variations on the basis of longitudinal changes with time, because the assessments were performed at the same time of the year for each patient. Therefore, for these two variables, the overall, seasonal, and nonseasonal variations were evaluated by using mixed-effects models, which compared the measurements for different patients who entered the study at different calendar times.
For each variable, the results are presented as the mean value at baseline, the amplitude and SEM of the seasonal effect, the t value (amplitude/SEM ratio) and P value for the seasonal effect, the amplitude/mean ratio (x100%), and the month in which the peak value of the variable was observed (counting from July 1). The amplitude of the seasonal effect is expressed as the difference between the maximal and minimal values of the sine/cosine curve for the seasonal effect. Because the sample sizes are large (2988 to 28,875 observations), the P values are mostly small. Therefore, the t values are presented to provide better indications of the statistical significance of the seasonal effect. P values of <0.01 were considered to be statistically significant. The amplitude/mean ratio provides an indication of the potential biologic or clinical significance. For indirect examination of the association between fluid accumulation and BP, similar mixed-effects analyses were used to relate the longitudinal changes in predialysis systolic or diastolic BP to the longitudinal changes in predialysis weight during the same time intervals.
In sensitivity analyses, similar results were obtained if the longitudinal changes were assessed relative to each patient’s mean value instead of the baseline value for each variable. Similar results were also obtained if more complex models, which incorporated the month of the initial measurements in the analysis, were used. Therefore, only results obtained by using changes relative to baseline values and the models described above are presented.
To examine the potential effects of outdoor temperatures on outcome variables, two separate analyses were performed. For these analyses, temperature was treated as a continuous variable. In the first analysis, the outcome measurements were related to the temperatures recorded at the designated weather station for each patient’s clinical center on the day of the outcome measurements, controlling for the day of the week, the time from randomization, and the interactions with the Kt/V and flux groups. As in the analyses described above, mixed-effects models were used to control for serially correlated data. This first analysis assessed the direct association of temperature with each variable, irrespective of seasonal effects. In the second analysis, sine and cosine terms for the seasonal variation were added to the aforementioned models for the effects of temperature. This analysis estimated the amplitudes of the seasonal effects for each variable that were not related to seasonal variations in temperature during the same time intervals.
For certain subanalyses, as described below, the 15 clinical centers in the HEMO Study were arbitrarily categorized into three geographic regions. Centers in New York City (New York), Boston (Massachusetts) (two centers), Philadelphia (Pennsylvania), Chicago (Illinois), Rochester (New York), and St. Louis (Missouri) were grouped as the Northeast/Midwest group. Centers in Durham (North Carolina), Atlanta (Georgia), Birmingham (Alabama), Dallas (Texas), Nashville (Tennessee), and Winston-Salem (North Carolina) were grouped as the South group. Sacramento (California) and Salt Lake City (Utah) were grouped as the West group. The exact locations of these cities are depicted in Figure 1. For some other subanalyses, individual clinical centers were compared with each other.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
Overview of Seasonal Variations
Three variables were selected for graphic display in Figures 2 to 4, to illustrate the seasonal component, nonseasonal component, and overall variation. Statistically significant (P < 0.01) seasonal effects were observed for 13 of the 21 variables (Table 1). Ten variables were associated with particularly large seasonal effects (t > 4.0, P < 0.001), i.e., predialysis and postdialysis BUN levels, enPCR, predialysis systolic and diastolic BP, ultrafiltration rate, serum sodium, bicarbonate, and albumin concentrations, and hematocrit values. These variables are presented in greater detail below.
|
|
|
Variations in BP, Ultrafiltration Volume, and Body Weight
The seasonal variations for predialysis systolic (t = 7.2, P < 0.001) and diastolic (t = 6.8, P < 0.001) BP were statistically strong (Figure 3), with the highest values being observed within 9-d periods in December and January, respectively. In contrast, neither postdialysis systolic BP nor postdialysis diastolic BP varied with the seasons. Of potential relevance to changes in BP, the intradialytic ultrafiltration rate (t = 4.0, P < 0.001) and total ultrafiltration volume (t = 3.9, P < 0.001) also varied with the seasons and attained their highest levels in December. Predialysis weight (t = 3.1, P = 0.002) varied with the seasons and peaked in January (Table 1). To examine the relationship between BP and volume, we related the changes in predialysis BP to the changes in predialysis weight from the initial baseline measurements for a given patient. Each 1-kg increase in predialysis body weight was associated with increases in predialysis systolic and diastolic BP of 0.67 ± 0.07 mmHg and 0.30 ± 0.04 mmHg, respectively (P < 0.0001 for both).
Variations in Other Laboratory Variables
Predialysis serum sodium (t = 6.3, P < 0.001) and bicarbonate (t = 5.0, P < 0.001) levels both attained peak values in December; the magnitude of the seasonal variation was greater for bicarbonate (2.7%) than for sodium (0.6%). Predialysis serum albumin levels (t = 8.3, P < 0.001) attained peak values in October. Hematocrit values varied with the seasons (t = 4.2, P < 0.001) and attained a peak in July (Figure 4).
Associations of Temperature with Outcome Variables
Because temperature changes with the calendar months might contribute to the seasonal variations in outcome variables, associations between outdoor temperatures and outcomes were examined. When the daily outdoor temperatures in all geographic regions were combined, the expected seasonal pattern was evident. The lowest mean temperatures were consistently observed in either December or January, which coincided with the highest predialysis BP (Figure 2).
The univariate associations between temperature and outcome variables are presented in Table 2. The outdoor temperature was inversely associated with predialysis systolic and diastolic BP, ultrafiltration rate and volume, predialysis weight, predialysis and postdialysis BUN levels, enPCR, and serum sodium and bicarbonate concentrations. The outdoor temperature was also positively associated with hematocrit values. With controlling for temperature, some outcome variables (P < 0.01 for predialysis and postdialysis BUN levels, serum sodium levels, serum albumin levels, and hematocrit values; P = 0.011 for enPCR) retained their seasonal variations. In contrast, predialysis and postdialysis BP, ultrafiltration rate and volume, predialysis weight, serum potassium levels, and serum bicarbonate levels exhibited seasonal variations without controlling for temperature (Table 1) but lost their seasonal patterns (P > 0.05) with controlling for temperature.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
Variations in BUN Levels and Protein Metabolism
A novel major finding of this study was the seasonal variation in predialysis BUN levels (Table 1). The mean fluctuation was 4% (i.e., 2.41 mg/dl). Variations in predialysis BUN levels could theoretically be attributable to variations in urea removal, protein catabolism, or protein intake. Differences in urea removal were unlikely to be the explanation, because the urea reduction ratio did not fluctuate accordingly (as calculated from the data presented in Table 1). An analysis of eKt/V values demonstrated that they did not exhibit a seasonal pattern. The consistency in eKt/V values was expected, because this variable was part of the HEMO Study intervention and was tightly controlled throughout the trial.
A second potential explanation for the variations in predialysis BUN levels involves variations in protein catabolism. Variations in enPCR, with the peak occurring at a similar time point (Table 1), are consistent with this possibility. An increase in enPCR does not distinguish, however, between an increase in dietary protein intake and excessive protein catabolism, which might occur without changes in protein intake. The seasonal variations (albeit at only marginally significant levels) in protein and energy intakes (as assessed by dietary recall), with peak values occurring at similar time points (Table 1), provide support for the former mechanism.
Seasonal variations in BUN levels and enPCR do not necessarily indicate that these variations are attributable to climatic changes. The fact that the seasonal variations persisted with controlling for contemporaneously measured outdoor temperatures argues against the dependence of BUN levels and enPCR on temperature.
The highest predialysis weights for the dialysis patients in this study were observed in December, similar to findings observed during the holidays (mid-November to mid-January) for the general population (3). Higher predialysis weights, however, do not distinguish between fluid weight and lean body mass. The highest protein and energy intakes occurred in April and the highest BUN levels were observed in March, i.e., 3 to 4 mo after the peak predialysis weights. In contrast, changes in predialysis weight were associated with changes in predialysis BP, suggesting that the highest body weights in the winter were likely attributable to extra fluid accumulation, rather than increases in fat or muscle mass. Direct assessments of lean body mass, using methods such as dual energy x-ray absorptiometry, were not performed in the HEMO Study.
Variations in BP
In the general population, both systolic and diastolic BP are highest in the winter (2,4–7). Similar findings for predialysis BP were noted in studies of 16 patients in Brazil (19), 102 patients in Uruguay (20), 53 patients in France (18), 144 patients in Japan (21), and the 1416 patients in this study (Table 1). Plausible mechanisms for the lower BP in the summer include decreased interdialytic fluid gain and thus decreased intravascular volume during that season. In contrast to the report from a single center in France, which did not note seasonal changes in interdialytic weight gain (18), our study of a larger population demonstrated that total ultrafiltration volume (presumably reflecting increased interdialytic fluid gain) and predialysis weight were lowest in the summer (Table 1). Furthermore, similar to predialysis BP, ultrafiltration volume and predialysis body weight were both correlated with the outdoor temperature (Table 2), and their seasonal variations (Table 1) disappeared with controlling for temperature. These results are consistent with the hypotheses that increased perspiration and insensible fluid loss from the body decrease interdialytic fluid gain in the summer and that BP in hemodialysis patients is at least partly volume-dependent. An alternative explanation for the lower predialysis BP in the summer is that the warm weather induces vasodilation and decreases peripheral vascular resistance. Data on precise dosages of antihypertensive medications and patient compliance are not available in the HEMO Study.
The magnitude of seasonal variations in predialysis BP in this study was smaller than that reported in an earlier study (18). The reason for this difference is not apparent. It is possible that the earlier study (18) demonstrates atypically large variations within a single center, in contrast to the HEMO Study data, which represent averaged results for 64 dialysis units.
In contrast to predialysis BP, there were no statistically significant seasonal variations in postdialysis systolic or diastolic BP (Table 1). Postdialysis BP values are presumably more responsive to the indoor temperature (because the patient has been confined in the dialysis unit for several hours) and intradialytic events, such as acute intravascular volume depletion, vascular refilling, and cardiovascular responses to changes in volume and electrolyte levels induced by ultrafiltration and dialysis.
Variations in Hematocrit
In this study, hematocrit values also varied with the seasons (Figure 4). The reasons for this seasonal variation are not apparent. Estimation of the effect of ultrafiltration volumes (reflecting interdialytic weight gain) on predialysis hematocrit values, even with the assumption that all of the shifts in body fluid were derived exclusively from the extracellular compartment, demonstrates that hemoconcentration per se seems to be insufficient to explain the observed changes in hematocrit values. Detailed data on erythropoietin dosages and blood loss are not available in the HEMO Study.
Caveats
A limitation of this study, and potentially other studies, is the fact that the data were not obtained in a manner that could definitively exclude the possibility of factors other than actual biologic seasonal variations in the variables that could affect the measurements. For example, it is possible that certain blood chemistry values might be affected by conditions during shipment of the blood samples or assays at the laboratory. The risk of the latter is particularly great for serum albumin concentrations determined with the nephelometry assay, which were measured in a single central laboratory and exhibited highly significant nonseasonal variations (P < 0.001). The hypothesis of seasonal variations in data collection, sample shipment, or laboratory assay procedures, however, would need to invoke changes in the instruments or human error on a regular seasonal basis. A further caution regarding the interpretation of these data, which is applicable to many epidemiologic studies, is that the data indicate only associations and not cause-and-effect relationships.
|
Conclusions |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
|
Acknowledgments |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  W S A Smellie What is a significant difference between sequential laboratory results?Calf muscle pain can indicate localised vasculitis J. Clin. Pathol., April 1, 2008; 61(4): 419 - 425. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S.-Y. Li, J.-Y. Chen, C.-L. Chuang, and T.-W. Chen Seasonal variations in serum sodium levels and other biochemical parameters among peritoneal dialysis patients Nephrol. Dial. Transplant., February 1, 2008; 23(2): 687 - 692. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Barany and H.-J. Muller Maintaining control over haemoglobin levels: optimizing the management of anaemia in chronic kidney disease Nephrol. Dial. Transplant., June 1, 2007; 22(suppl_4): iv10 - iv18. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Passlick-Deetjen and E. Bedenbender-Stoll Why thermosensing? A primer on thermoregulation Nephrol. Dial. Transplant., September 1, 2005; 20(9): 1784 - 1789. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
HOME
CURRENT ISSUE
ARCHIVES
JASN Express
ONLINE SUBMISSION
AUTHOR INFO
EDITORIAL BOARD SUBSCRIBE FEEDBACK ALERTS HELP |
