Role of Creatinine Clearance as a Screening Test in Persons With Spinal Cord Injury

Methods 

The study was approved by our hospital’s institutional review board. We selected 70 male patients with SCI who had had annual inpatient evaluations for 5 separate years. The annual evaluation is a comprehensive history and physical examination designed to optimize the patient’s health maintenance and well-being. Many of our patients are admitted to the hospital for these evaluations because of the extent of the testing, the time required to travel to and from the clinic, and the socioeconomic limitations of this cohort. Evaluations are offered annually to patients. Because of patient preferences, scheduling conflicts, and other factors, the interval between evaluations frequently exceeds 1 year. Using a locally maintained patient registry, we obtained a list of all patients who had completed annual evaluations in the preceding 2 years. Beginning with the patient who most recently completed an evaluation, we consecutively selected all patients who met the inclusion and exclusion criteria. Only those patients admitted to the hospital in each of 5 consecutive annual evaluations were selected for this study to ensure that subjects’ urine would be collected and maintained by trained nursing staff, as opposed to collecting urine in an outpatient setting. All patients were at least 1 year beyond their spinal injury at the time of the first CCr measurement. The few women who are among our clinic’s more than 650 patients did not meet the criteria for the review.

The nursing staff collected urine for a 24-hour period from each patient, using his current mode of bladder management (eg, indwelling catheters, condom catheters, intermittent catheterization, spontaneous voiding). The urine was collected in an appropriate laboratory container and stored in a refrigerator until the collection was completed. Urine volumes and urinary creatinine concentrations were determined by our hospital’s laboratory personnel. Serum creatinine was measured during the 24-hour urine collection period.

Demographic data were collected about each subject. Test results pertinent to renal function were culled from the medical records from each of 5 consecutive annual evaluations (time 1 through time 5). These included renal ultrasound findings, serum creatinine, and 24-hour urine collection data (total volume, total urine creatinine, calculated CCr). When there was an abnormally low CCr value or significant change from the previous year’s measurement, we reviewed the medical record to see if there were any subsequent changes made in medical management (eg, medications, bladder management) as a result of the CCr value.

All statistical analyses were performed using SPSSa and Stata.b A P value less than .05 was considered significant. We used 3 methods to characterize the variability of annual CCr and serum creatinine measurements. The first, called repeatability, was described by Bland and Altman.10 It uses the within-subject standard deviation (SD) to estimate the maximum expected difference between 95% of paired observations on the same subject. We also used the within-subject SD and the quantiles of a normal distribution to estimate the percentage of patients that would be expected to change from a sample mean CCr to levels indicative of varying degrees of renal failure the following year, in the absence of any real change in renal function. Classification of degree of renal failure were taken from the National Kidney Foundation’s (NKF)11 classification of chronic kidney disease, without adjusting GFR for height. Finally, we computed for both tests the discriminant ratio,8, 12 which is an estimate of the subject-to-subject SD of true values divided by the within-subject SD of repeated measurements. The higher the discriminant ratio, the greater the ability of the test to discriminate between subjects, or between different health states for 1 person. With the use of these methods, it was assumed that there was no underlying change in the renal function of our subjects during our study period, which was consistent with our data; it was also assumed that SDs do not depend on mean values, which we confirmed with appropriate plots.13

Results 

The demographics of the study cohort are summarized in table 1. The mean intervals between consecutive measurements were 1.41, 1.42, 1.38, and 1.37 years, with a mean interval ± SD between the first and fifth CCr determinations of 5.57±2.13 years. The most common bladder management at time 5 was an indwelling catheter (urethral or suprapubic), which was used by 41 (59%) patients.

Table 1. Patient Demographic Data

	Variables	Patients (N=70)	%
	Level of injury
	Cervical	48	68.6
	Thoracic	18	25.7
	Lumbar	4	5.7
	Sacral	0	0
	ASIA Impairment Scale
	A	32	45.7
	B	16	22.9
	C	11	15.7
	D	11	15.7
	Bladder management at time 5
	Indwelling catheter	41	58.6
	Condom catheter	13	18.6
	Spontaneous voiding	9	12.9
	Intermittent catheterization	5	7.1
	Ileal conduit	2	2.9

Abbreviation: ASIA, American Spinal Injury Association.

Mean Serum Creatinine, CCr, and Urine Creatinine 

Mean serum creatinine, CCr, and urine creatinine at the 5 time points are shown in table 2 and normalized values are shown in figure 1. As expected, serum creatinine values between subjects were highly variable. There was a slight decrease in mean serum creatinine between time 1 and time 5 (.68±.21mg/dL vs .65±.24mg/dL; paired 2-tailed t test, P=.079,). There was no significant change in CCr between time 1 and time 5 (105.4±38.8mL/min vs 101.0±39.6mL/min, P=.42). The mean urine creatinine decreased from 1013±396mg/24h at time 1 to 883±399mg/24h at time 5 (P=.008), indicating a mean 12.8% drop in urine creatinine over a mean 5.57 years, 3.2% per time period, or 2.3% per year. Forty-two patients (60%) had at least 1 CCr measurement below the normative range (85–125mL/min). Two patients had a measured CCr less than 20mL/min, which is a value typically considered as renal failure (ie, requiring dialysis). Neither of these patients had any clinical signs of renal insufficiency, and both had a normal renal ultrasound and stable serum creatinine at the time of the abnormal CCr value.

Table 2. Measurements from Consecutive Evaluations on 70 Subjects

	Time Points	Serum Creatinine (mg/dL)	CCr (mL/min)	Urine Creatinine (mg/24h)
	Time1	0.68±0.21	105.4±38.8	1013±396
	Time2	0.67±0.19	105.9±35.3	996±377
	Time3	0.66±0.22	105.6±33.1	974±398
	Time4	0.67±0.25	107.0±39.9	946±380
	Time5	0.65±0.24	101.0±39.6	883±399

NOTE. Values are mean ± SD.

View Large Image Download to PowerPoint

Fig 1. Mean values ± standard error of the mean for the 70 subjects at each time point are normalized to the value at time 1 (serum creatinine, .68mg/dL; CCr, 105.4mL/min; urine creatinine, 1013mg/24h).

Discussion 

We found no evidence that annual CCr calculated from a 24-hour urine collection affects the management of patients with SCI. The CCr test results were highly variable, and the test’s variability diminished its clinical utility. The variability is in part due to the multiple sources of measurement error, many of which involve determination of urinary creatinine excretion. These include incomplete collection of all urine produced during 24 hours, inaccurate measurement of urine volume, and variability of the laboratory test for urinary creatinine concentration. Furthermore, although there were many instances of abnormal CCr values, no changes in medical management were made based on these values, because in all instances there were no other clinical data to corroborate the findings. When patient management of renal function was changed, decisions were based on clinical data other than CCr, or on nonmedical factors such as social support and ease of care. Serum creatinine values varied considerably between subjects, reflecting the differences in intact muscle mass, body habitus, chronicity of injury, and GFRs between subjects. No attempt was made to correlate these values with level of injury or other variables because this has been reported previously6 and was found not to be useful, other than to conclude that generally, decreased muscle mass in SCI patients results in lower serum creatinine measurements.

In our SCI unit and others in the United States, the 24-hour CCr has been applied as a screening test for renal insufficiency in SCI patients. One definition of screening is “the identification of unrecognized disease by the application of tests or examinations to apparently well persons to sort out those who probably have a disease from those who probably do not.”14(p1776) People who screen positively require a diagnostic workup to determine whether or not they actually have the disease.

One rationale for using the 24-hour CCr as a screening test is that the normative range of serum creatinine in able-bodied subjects is not applicable to persons with SCI.5, 6 Serum creatinine is in part a reflection of creatinine production, and in patients with SCI, muscle atrophy results in lower creatinine production, with resultant lower serum creatinine measurements. Thus, it is possible that a patient with SCI and a serum creatinine in the normative range may have a severely reduced GFR. A more sensitive measure of renal insufficiency, the 24-hour measured CCr, could theoretically detect loss of kidney function in the SCI population. In this test, differences in steady-state creatinine production resulting from differences in muscle mass should not affect creatinine clearance.

There are, however, several pitfalls in using CCr as the primary screening test for renal disease. First, CCr is dependent on several factors, including diet, medication, body mass, age, functional reserve of nephrons, and method of urine collection.5 There are many problems inherent in the collection of a 24-hour urine, such as careful timing of the collection and storing the urine. Even with hospital admission and supervision of the collection by a trained nursing staff in our study, there was tremendous year-to-year variability in the CCr values for each patient. Inadequate collections may also be in part due to patient noncompliance. Rosano and Brown15 state that the inconsistency is such that consecutive measurements of serum creatinine are a more sensitive index of glomerular pathology than are consecutive clearance measurements, when analytic and biologic variability are considered. We do not believe that all the CCr values we found were a true reflection of the clinical condition of the patient; indeed, there were more than 20 CCr values that were in the supraphysiologic range (>150mL/min).

Second, CCr is primarily used as a measure of existing renal insufficiency, as a measure of severity of renal failure. In patients with SCI, the CCr is advocated as a screening test for renal disease, without an existing diagnosis of renal insufficiency. A screening test is one that can be applied easily and inexpensively, and can detect disease before it becomes clinically apparent, with the knowledge that early detection and treatment results in preservation of function or prolongation of life.14 The 24-hour CCr is not easily administered, is fairly labor intensive (and thus, expensive), and, by our accounts, has not detected any renal disease or problem that was not manifested on other tests that are of higher yield (eg, renal ultrasound). Thus the test is applied inappropriately as a screening measure in the SCI population.

The variability in longitudinal CCr measurements has not previously been described in the SCI population because most studies have only correlated measurements with different techniques taken at a single point in time. MacDiarmid et al7 recently reported good correlation (r=.71) of the 24-hour CCr measurement with 99mTc diethylenetriaminepentaacetic acid (DTPA) clearance in patients with SCI, and concluded that the CCr was the most practical test for measuring renal function in this population. They advocated use of the test for routine surveillance. This group, and others advocating the use of CCr in SCI, did not report on whether this information resulted in improved detection of renal insufficiency or had any clinical impact on their patient populations over time.6, 7 Additionally, their method of CCr measurement differed from ours in that they performed 2 consecutive 24-hour urine collections and used the average of the 2 for comparison with 99mTc DTPA clearance. The mean CCr measurements calculated from the 2 urine collections performed during the same hospitalization differed by 16.4%. To some degree, their variability was minimized by the use of a single measurement of serum creatinine for both CCr calculations, so the 16.4% reflects the day-to-day variability of urine creatinine measurement, not CCr measurement. Our data demonstrate considerably greater variation between consecutive CCr measurements when performed over a longer interval (1.4y vs 24h apart) and without mandatory bladder catheterization for removal of residual urine. Tan et al8 showed a somewhat lower within-subject SD for CCr (17.6mL/min vs 25.9mL/min in this study) for 2 measurements taken 4 weeks apart in 40 subjects with normative range CCr8; however, in that study, the GFR calculated using serum creatinine (Cockcroft-Gault formula) showed lower variability than CCr.

As shown by the repeatability measures, 5% of patients would show a difference of at least 71.8mL/min for successive CCr measurements, despite presumably stable renal function. For a patient with a CCr equal to the median for this sample (103.2mL/min), there is a 30.3% chance of having the test show at least mild renal failure (<90mL/min) and a 5% chance of showing at least moderate renal failure (<60mL/min). The discriminant ratio for CCr is less than 1, indicating that there is greater within-subject variability than between-subject variability. This degree of variability on consecutive measurements results in a low specificity for this test because of frequent false-positives, potentially leading to additional unnecessary testing. The repeatability measure for serum creatinine indicates that a change of at least .34mg/dL would be seen for 5% of patients on successive tests. A greater degree of change than this in serum creatinine should prompt a work-up for potentially treatable causes (including short-term reversible ones like dehydration), even though for many patients the absolute value of serum creatinine would still be well within the normative range (<1.3mg/dL). While single measurements of CCr may correlate more closely with more precise measurements of GFR (eg, 99mTc DTPA clearance), changes in serum creatinine could be as sensitive or more sensitive than CCr changes.

One issue in this study is why so many abnormal CCr values did not result in additional diagnostic testing, particularly for the very severely depressed values. Typically, identifying an acute decrease in renal function prompts a workup for potential causes, or at least a more sensitive measure of GFR (eg, radionuclide scan). However, in all cases, there were stable renal ultrasounds, serum creatinines or both, and no clinical evidence of renal insufficiency. Therefore, there was little to indicate the need for further diagnostic testing, particularly if laboratory values (eg, extremely low urinary creatinine or urine volume) were indicative of inadequate urine collection.

There are some who believe that CCr as a measure of GFR should be abandoned.16, 17, 18 Our demonstration of the extreme variability of the test supports this belief. Another issue to be entertained is whether there is a need for additional screening for renal disease in SCI patients, other than with a serum creatinine, renal ultrasound, and KUB. In this sample of 70 patients with 390 patient-years of follow-up, renal function presumably remained stable over approximately 5 years, based on nonsignificant changes in both mean CCr and serum creatinine over time. For all comparisons of repeatability, variability, and reliability, including mean differences and correlations between consecutive measurements, serum creatinine was superior to CCr. A new method of determining CCr in SCI patients that involves the measurement of serum cystatin C19, 20 may be a more appropriate and sensitive determinant of renal function in this population, and should be investigated.

Conclusions 

Current laboratory testing of renal function in patients with SCI shows considerable variability over consecutive measurements. We conclude that the CCr test has little value as a screening measure for renal disease in the SCI patient.

Suppliers