Delivery of Predialysis Care in an Academic Referral Nephrology Practice

Abstract

The purpose of this study was to obtain a detailed picture of the delivery of all aspects of pre-end stage renal disease (ESRD) care in an academic nephrology practice. The study consisted of a cross-sectional review of the charts of 111 patients with chronic kidney disease (CKD) (Creatinine > 1.5 mg/dL for males, Creatinine > 1.3 mg/dL for females) followed either in a private practice or a public hospital clinic. Charts were reviewed for evidence of a number of quality of care items including the degree of blood pressure control, the use of angiotensin II blockade, treatment of anemia, bone disease, and cardiovascular risk factors, as well as evidence of dialysis orientation and vascular access placement. Compared with previous published studies, this study shows encouraging trends towards better outcomes in at least two areas with evidence of improvement in hemoglobin levels, use of angiotensin II blockade, and degree of blood pressure control. However it also reveals many areas where care could be improved. This was particularly true in the areas of access placement, bone and mineral metabolism, and cardiovascular disease. It will be interesting to see whether the recent introduction of CKD guidelines by the National Kidney Foundation (NKF) and the associated educational initiatives for primary care physicians and nephrologists that are planned will further serve to improve the management of CKD patients in the years to come.

Keywords:

• chronic kidney disease
• care delivery
• hypertension
• anemia
• metabolic bone disease

Introduction

Recent years have seen the development of significant advances in the care of the predialysis patient with chronic kidney disease (CKD). Primary among these is the recognition that vigorous treatment of blood pressure, particularly with agents that block the actions of angiotensin II can slow, and perhaps even prevent, the progression of CKD.[1-5] Equally important is the recognition that early identification and treatment of the comorbidities of CKD such as anemia,[6] dyslipidemia,[7] and bone disease[8] may be important not only in progression, but also in limiting the significantly increased risk of cardiovascular morbidity and mortality that characterizes this group of patients. Several studies suggest, however, that many patients with chronic kidney disease are not receiving such treatments and thus valuable opportunities to prevent progression of chronic kidney disease, and its associated morbidities, are being lost.[9-13] Potential reasons for this situation are multiple; the lack of symptoms in early CKD, lack of access to care for socioeconomic reasons, lack of familiarity of primary care physicians with treatment recommendations, and delayed subspecialist referral. Few studies to date have analyzed the delivery of predialysis care to the CKD patient by the nephrologist to see if there are also opportunities for improvement of care in that area also.[10] The purposes of this study were to obtain an in-depth cross-sectional picture of the care being delivered in the practice of an academic nephrology group, to assess how effectively treatment guidelines for CKD were being met, and also to compare the results with the previous study of Kausz et al.[10] to see if there was any evidence of improvement.

Methods

Between the months of September 2001 and April 2002, chart reviews of all patients being followed in an academic nephrology practice with evidence of impaired renal function (Creatinine > 1.5 mg/dL for males, Creatinine > 1.3 mg/dL for females) were reviewed to see what tests had been ordered within the previous 6 months and to see how many patients were achieving certain levels of care. The patients were followed at two sites: a hospital-based outpatient clinic and a university practice plan private clinic. Patients were managed by five attending nephrologists. Renal fellows were also involved in the management of patients in the outpatient hospital clinic under the supervision of three of the five nephrologists.

We looked for documentation of results of the following tests: In evaluating renal function we looked for serum blood urea nitrogen and creatinine, isotopic estimation of glomerular filtration rate (GFR), creatinine clearance, urinary urea nitrogen, serum albumen, quantitation of urinary protein, and microalbumin. For anemia assessment we looked for hemoglobin/hematocrit, serum iron, total iron binding capacity, transferrin saturation (Tsat), and serum ferritin. For bone disease assessment we looked for serum calcium, phosphorous, parathyroid hormone (PTH), and measurement of bone density by DEXA scan. While the majority of these determinations were performed at our local hospital laboratory, many determinations were made at a variety of different laboratories. We used the actual values for each test obtained and did not attempt to standardize these measurements in any way.

In diabetics we looked for a recent estimation of hemoglobin A1c. For assessment of cardiovascular disease potential we looked for systolic and diastolic blood pressure, history of coronary artery or peripheral vascular disease, and results of echocardiography as well as measurement of serum cholesterol, lipid profile, homocysteine, and C-reactive protein levels. We also looked for results of stress testing and coronary angiography, peripheral Doppler exams, and angiography.

In the area of treatment we assessed how many patients met the following treatment goals.

• BLOOD PRESSURE:[14]

  • BP < 130/85 in all patients < 60 years old

  • BP < 125/75 if proteinuria > 1 gm/day and patient < 60 years old.

  • BP < 145/75, if patient > 60 years old unless contraindicated (e.g., cerebrovascular disease).

• ANEMIA:[15]

  • Hemoglobin (Hgb)> 11 g/dL

  • Erythropoietin therapy if Hgb < 11 g/dL

  • Iron therapy if Tsat < 20%

• BONE DISEASE:[16]

  • Serum Phosphorous < 4.5 mg/dL

  • Serum Calcium between 8.5 and 10.5 mg/dL

  • Phosphate binding therapy for serum phosphorous > 4.5 mg/dL

  • Vitamin D analog therapy if PTH > 100

• DYSLIPIDEMIA:

  • Serum cholesterol < 200 mg/dL

  • Lipid lowering agent if Cholesterol > 200 mg/dL

• PREPARATION FOR DIALYSIS

  • Documentation of discussion re ESRD modalities

  • Placement of dialysis access for patients with estimated GFR < 20.

The chart review was performed by two renal fellows (KM, VV).

Results

In all, 111 charts were reviewed and the demographic data on the patients are outlined in Table 1. The majority were Caucasian, but included one-third African-American, and one-third diabetic. There was a slight preponderance of males. Most were covered by commercial insurance, Medicare, or Medicaid but 14% had no insurance. One-fifth were current smokers.

Table 1. Patient demographics

		Number of Patients for whom data available
Age (yrs)	58 ± 1.6	111
Race		101
Caucasian	53%
African American	34%
Hispanic	2%
Asian	2%
Not recorded	9%
Gender		111
female	45%
male	55%
Primary Insurance		111
Commercial	48%
Medicare	29%
Medicaid	9%
None	14%
Smoking	21%	93
Diabetes	32%	111
Causes of CKD		104
Glomerulonephritis	27.9%
Hypertension	23.4%
Diabetes	18.9%
Renovascular	6.3%
Other	14.4%
None	9.0%
BUN (mg/dL)	37.7 ± 1.8	110
Creatinine (mg/dL)	2.5 ± 0.1	111
Creatinine clearance (mL/min)	44.1 ± 2.5	77
MDRD GFR	32.3 ± 1.4	101

Assessment of Renal Function

The majority of patients had a serum creatinine of less than 3 mg/dL (see Figure 1A). It was possible to calculate an estimated GFR in 101 patients by the simplified Modification of Diet in Renal Disease (MDRD) equation and this showed a mean GFR of 32.3 ± 1.3 mL/min. According to the proposed K/DOQI guidelines for classification of chronic kidney disease, essentially half the patients had Stage 3 disease with most of the remainder having Stages 4 and 5 (see Figure 1B). Only a handful of patients had estimated GFRs above 60 mL/min. The distribution of patients between stages were similar for males and females.

Figure 1 A. Distribution of patient by serum creatinine. B. Distribution of patient by stage of chronic kidney diseae (CKD).

Approximately 70% of patients had 24 hour urine collections of urine for estimation of creatinine clearance and protein excretion. No chart had a documented isotopic GFR. The mean creatinine clearance was 44.1 ± 2.4 mL/min numerically higher than the estimated GFR measurements. Of the 84 patients with proteinuria, 20% had nephrotic range proteinuria, 32% had proteinuria in the 1–3 gm/day range, 40% proteinuria < 1 gm/day, while 8% had normal protein excretion. These data were collected before the publication of the recent NKF clinical guidelines advocating the use of GFR by the modified MDRD equation and the use of spot urines for estimation of protein excretion.

Hypertension

Nearly all patient charts had documented blood pressure measurements and the mean values were 136 mm Hg systolic and 79 mm Hg diastolic (Table 2). We estimated how many of the patients achieved the following targets as recommended by the NIH.[14]

Table

Patients with proteinuria < 1 gm	< 135/85
Patients with proteinuria > 1 gm	< 125/75
Patients over 60	< 145/75

Table 2. Management of hypertension

		Number of Patients	Percentage of patients	Mean Value
Systolic		109	97%	136
Diastolic		109	97%	79
Controlled		45	41%	117/72
Uncontrolled		64	59%
	Sys + Dias	41	38%	148/84
	Sys	18	17%	148/91
	Dias	5	5%	128/90
Classes of Antihypertensives
ACE inhibitor/AII receptor blocker			62%
Diuretic			49%
Calcium Channel Blocker			53%
Beta-Blocker			33%
Vasodilator			21%
Other			18%

By these criteria 41% were felt to have adequately controlled blood pressure on the basis of the most recent measurement. Of the 59% who were inadequately controlled, in most cases it was the systolic pressure that did not meet targets. Patients were prescribed a mean of 2.3 ± 0.1 antihypertensive medications (see Table 2).

Agents that block Angiotensin II (AII) were the most frequently used followed by calcium channel blockers and diuretics (see Table 2). Thirty-eight percent of patients were not receiving either ACE inhibitors or AII-receptor blockers, which are felt to have unique renoprotective properties. Of the 43 patients not receiving these agents, 10 patients had not tolerated them while 22 had proteinuria of less than 1 gm. This left 19 patients whom we would consider candidates for AII blockade who were not receiving either type of agent. Interestingly six of these had blood pressure that already met target goals.

Anemia

Determinations of hemoglobin were available in 109 patients but iron parameters were estimated in less than half of patients. Patients were more likely to have estimations of iron parameters if they were either on Erythropoietin (EPO) or had a Hgb of less than 11.0 g/dL (77% as against 28% of patients with Hgb > 11.0 g/dL and no EPO). Sixty-two percent of iron saturation measurements showed evidence of iron deficiency (iron saturation < 20%).

The mean Hgb was 12.5 ± 0.2 g/dL and the mean hematocrit 37.1 ± 0.5%. Twenty percent of patients had a Hgb < 11 g/dL and 7% had a Hgb < 10 g/dL. In all, 30 patients (28%) were either on EPO or had indications for the medication (Hgb < 11.0 g/dL). Indications for EPO increased as the stage of kidney disease advanced and were present in close to half of patients with Stages 4 and 5, but only 14% of those in Stage 3 (Figure 2). Patients were slightly more likely to receive the drug (76% vs. 50%) if they had more advanced renal insufficiency. Seventy-seven percent of patients with indications for EPO had measurements of iron parameters but iron deficiency was undertreated with only six of 28 iron deficient patients (21%) on either oral (n = 4) or intravenous (n = 2) iron replacement.

Figure 2 Percentage of patients with indications for erythropoietin therapy and who are and are not receiving the medication by stage of CKD.

Bone and Mineral Metabolism

Nearly all patients had determinations of serum calcium and hypocalcemia was infrequent (8%) and mostly corrected when adjusted for the serum albumen (2%) (Figure 3A). Serum phosphorous was measured in only 70% of patients and was elevated in 16% of these. Only three of these 12 patients were on phosphate binders, though an additional nine patients with normal phosphorus were also receiving binders. All binders prescribed were calcium containing. No patient had a calcium-phosphorous product of > 55. Figure 3B shows the breakdown of phosphate binder therapy by stage of CKD. For this analysis, only patients with a measurable MDRD GFR and serum phosphorous were included (n=72). As can be seen, the need for binder therapy was confined mostly to patients with Stage 4 and 5 disease. Of note, however, in Stage 5 disease, about one third of patients were either not receiving binders or had an inadequately controlled serum phosphorus.

Figure 3 A. Number of patients who had determinations of calcium, phosphorus, and intact PTH levels. B. Percentage of patients with measurements of serum phosphorous who had indications for phosphorous binders (phosphorus > 4.5 mg/dL) by stage of CKD, whether they were receiving binder therapy and whether phosphorus was controlled.

Determinations of serum parathyroid hormone (PTH) were infrequent with documented values present in only 9% of charts. Five of these 10 patients had elevated PTH values and three of these patients were receiving Vitamin D analogs. Only two patients had documented studies of bone mineral density. One was a 66-year-old woman and the second a 42-year-old man with a previous history of steroid therapy.

Cardiovascular Disease

Only 13% of patients had a documented history of coronary artery disease (angina or myocardial infarction). Seventy percent had measurements of serum cholesterol, but very few (< 5%) had determinations of either plasma homocysteine or C-reactive protein levels. Thirty-seven patients were receiving therapy with statins. Of those with serum cholesterol measurements, the value exceeded 200 mg/dL in 48%. Results of echocardiograms were available in 27%; stress tests in 4%, and 7% of patients had had coronary angiograms (see Figure 4).

Figure 4 Percentage of patients who had various screens for cardiovascular disease.

Preparation for Dialysis

Nineteen patients had a GFR by the modified MDRD equation of less than 20 mL/min. In only six cases was there documentation in the medical record that placement of dialysis access had been discussed with the patient. Only one patient had a functioning access (a PD catheter) in place and one other was awaiting fistula placement. A third patient had refused access placement. Access placement plans were unclear in the other three. These findings recently caused us to perform a follow-up study on these 19 patients to determine how many progressed to ESRD over the subsequent two years and how many had access placed in a timely fashion. Five patients were lost to follow-up within 3 months after completion of the chart review. Three of these were elderly patients (79, 83, and 89). Two additional patients were aged 52 and 64 and had creatinines of 5.9 and 5.8 when last seen. The chart of a sixth patient (aged 67) could not be located. Of the remaining 13 patients, four still have functioning kidneys as of 9/30/03. The other nine progressed to ESRD. Four were started on peritoneal dialysis and five on hemodialysis. Only two of the patients starting hemodialysis had functioning access (fistulas) at the time of dialysis initiation, which occurred in the outpatient setting for all except one patient. None of the patients with functioning kidneys (latest serum creatinines range from 3.2 to 5.1 mg/dL) have had dialysis access placed.

Opportunities to Improve Care

Figure 5 shows the percentage of patients who were identified during this project as having clinical abnormalities or persistent risk factors that could be potentially improved and consequently represent an opportunity for improvement of care. The most marked deficiencies were seen in the area of bone and mineral metabolism and also in preparation for dialysis modality. Additional areas with potential for improvement included meeting target levels for blood pressure control and cholesterol level. However, given that several other parameters were not even assessed in a significant number of patients, this probably represents an underestimate of potential beneficial interventions.

Figure 5 Opportunities for improved car in this group of CKD patients.

Discussion

This study consisted of a cross-sectional review of prevalent practice in an academic nephrology practice in early 2002. A total of 111 patients with significant elevation of serum creatinine and with Stage 3–5 chronic kidney disease were included. The study revealed a number of interesting findings and several potential areas for clinical improvement particularly in the areas of blood pressure control, preparation for dialysis, and management of bone and mineral metabolism.

The patients constituted a fairly broad spectrum of disease and socioeconomic groups. They were drawn from both a practice plan office setting and a public hospital subspecialty clinic. While the majority were covered by insurance, a significant percentage (14%) were not. African-Americans were overrepresented as compared to their prevalence in the local population and this probably reflected both the urban public hospital setting, as well as their propensity to chronic kidney disease.[17] The commonest cause of chronic kidney disease was glomerulonephritis, followed by hypertension and diabetic nephropathy. The majority of cases of glomerulonephritis had been confirmed by biopsy so we do not feel this is an overestimate. It contrasts with studies of end stage renal disease, where glomerulonephritis is usually surpassed by both diabetes and hypertension.[18] Two factors may have contributed to this. The first is a referral bias, in that patients with idiopathic glomerulonephritis may be referred earlier to a tertiary referral practice like ours than cases of diabetic/hypertensive nephropathy, because the latter diagnoses are usually made “clinically” whereas the former usually requires a renal biopsy. Secondly, at least in the case of diabetes, renal failure may progress more rapidly than in most idiopathic glomerulonephritides, which would lead to a disproportionate retention of the latter patients in a pre-ESRD cohort.

This cohort review was completed just before publication of the K/DOQI guidelines for the classification, diagnosis, and management of chronic kidney disease.[18] It is, therefore, not surprising that estimates of kidney function and injury consisted exclusively of serum creatinine and 24-hour urine collections for creatinine clearance and protein excretion. It was possible, however, to retrospectively calculate an estimated GFR by the simple MDRD equation and, thus, classify the patients by stage of CKD. The largest group fell in the Stage 3 category with few if any being Stage 2, suggesting that the gender specific cutoff for serum creatinine chosen corresponded fairly closely to Stage 3 and worse. Estimated GFR was on average about 10 mL/min lower than corresponding measures of creatinine clearance, when these were available.

Two previous studies have looked at delivery of care to patients with CKD and serve as baselines with which to compare our data. Kausz et al.[10] reviewed 602 patients followed between 1994 and 1998 at three Boston nephrology clinics. Their patients had slightly more advanced CKD than did those in our study, by virtue of the use of higher serum creatinine cutoff so that initial serum creatinine was 3.1 mg/dL and GFR 22.2 mL/min. They also reported on anemia management in this cohort in more detail in a second publication.[19] The majority of these patients were covered by HMOs. Israni et al.[13] reviewed a cohort of 56 patients with CKD (Creatinine > 1.7 mg/dL) followed at an academic primary care clinic, also in Boston, between 1997 and 1999. Mean serum creatinine in that study was 2.3 mg/dL with a mean creatinine clearance of 39 mL/min, values very close to the mean values in our patients. Both studies also reported considerable opportunities to improve the care received.

Nearly all patients in our cohort had recent measurements of blood pressure and many were receiving antihypertensive therapy. However, less than half had achieved recommended target blood pressure levels. On the other hand less than 40% of patients had BP > 140/90 at the last measurement, which is somewhat better than the 52% reported by Israni.[13] It is true that we did not distinguish or record the length of follow-up of patients to see if the degree of blood pressure control tended to improve over time. Only 60% of patients were receiving ACE inhibitor/angiotensin II receptor blocker (ARB) therapy, but these percentages are higher than the 49% and 41% found by Kausz[10] and Israni,[13] respectively. The percentage of patients prescribed ACEI/ARBs was considerably higher if proteinuria was greater than 1 gm/day (78% vs. 44% for proteinuria of < 1 gm). There is evidence that patients with heavier degrees of proteinuria derive greater benefit from angiotensin II blockade, while its benefit in nondiabetic kidney disease with proteinuria of less than 1 gm was less well-established at the time of this study. Whether or not a patient was on an ACEI/ARB did not appear to be related to their gender, race, serum creatinine, or insurance status (data not shown). Of the 43 patients not receiving these agents, 10 patients had not tolerated them while 22 had proteinuria of less than 1gm. This left 19 patients whom we would consider candidates for AII blockade who were not receiving either type of agent. Interestingly six of these had blood pressure that already met target goals. The rate of intolerance of angiotensin II blockade found in this study (10%) is considerably higher than that seen in randomized controlled trials and may imply a lesser tolerance of side-effects such as cough and mild hyperkalemia in the “real” world of clinical practice. Similar rates of intolerance to AII blockade have been found in other studies.[12&13]

The incidence of severe anemia (Hgb < 11 g/dL) in this study (20%) was lower than that found by Israni (38%) and Kazmi[19] (62%) in their studies.[10], [13] The reasons for this are not clear. Seventeen percent of our patients were receiving erythropoietin (EPO) therapy for anemia, higher than the 7% seen in the primary care population of Israni[13] but less than the 30% in the nephrology practices of Kazmi[19]. The latter differences may be accounted for by the more advanced stage of CKD of patients in that study (see previous information). We identified 11% of our patients who had a Hgb < 11.0 g/dL who were not on EPO and, therefore, might have benefited from therapy with this agent. When patients receiving EPO were compared with those with a Hgb < 11 not on EPO, there were no obvious differences in terms of practice site (clinic vs. private), insurance status, mean Hgb, systolic, or diastolic blood pressure. Because we looked only at the most recent hemoglobin we could not compare the hemoglobin level at which EPO therapy was instituted in those who were receiving the drug. Of note, in the group with Hgb < 11 but not receiving EPO, only two patients had hemoglobin levels below 10 g/dL. Initially, Medicare required that a Hgb level of < 10 g/dL be documented before therapy with EPO could be instituted (Medicare Intermediary Manual § 3112.4), while the policies of private insurers varied but usually required preapproval. Medicare has since revised its policy to allow coverage for EPO when the hematocrit (Hct) is < 36.0. This may explain why the majority of these patients were not started on EPO. Most of the 31 patients either on EPO or with a Hgb < 11 had had estimations of iron status, which is also a requirement before starting EPO; however, only 21% had adequate iron stores. Fourteen of the remaining 19 were being prescribed oral iron supplements, while two had received intravenous iron. Treatment with iron was more common in those receiving EPO (74% vs. 33%). These findings are similar to those previously reported by Kazmi et al., [19] who also found low rates of testing for iron status and use of iron supplementation in patients with CKD. This contrasts with the heavy utilization of iron in patients on ESRD where the facility of giving intravenous iron and its documented benefit in enhancing the response to EPO are well recognized. Only 7% of patients in this study had a Hgb < 10 g/dL and 75% of them were receiving EPO, somewhat better numbers than those seen by Kazmi et al. (39% and 55% respectively).[19] Data from the (US Renal Data System) USRDS does suggest that there has been a steady trend to greater use of EPO and consequently in mean hematocrit in patients starting dialysis in the last few years.[20]

The need for phosphate binder therapy was largely confined to patients with Stage 4 and 5 CKD, and again, there was evidence of room for improvement in that just 12 of 21 patients with indications for phosphate binder therapy were actually receiving such agents. Of interest, all patients were prescribed calcium-based binders. The choice to use calcium-based binders appear to have been a medical decision on the part of the nephrologists since at the time only two patients prescribed binders lacked prescription coverage and none of the local health insurers had any specific restriction on prescribing the newer sevelamer binder to pre-ESRD patients. Indeed many of the patients receiving calcium binders in Stage 2 and 3 may have been on these for other indications, e.g., protection against osteoporosis. The low number of estimations of PTH was surprising. The PTH estimations were done only marginally more often in those with Stages 4 and 5 CKD (12%) than in those with Stage 3 (8%). Previous studies[9&10], [13] have not looked at the question of calcium/phosphorous/PTH treatment in any detail, though Kausz and coworkers[10] also found a low incidence of PTH measurements in their CKD patients (15%). This suggests that PTH estimations were not regarded as an important component of patient care in this group of patients by their nephrologists. One possible explanation is that the nephrologists were too preoccupied with other aspects of care (e.g., blood pressure and anemia) to pay attention to PTH or were not convinced of the clinical benefits of intervention. The recent publication of guidelines on bone and mineral metabolism by the NKF may well have a significant effect on such practices.[21]

Cardiovascular disease is probably the major cause of morbidity and mortality in patients with CKD and there was clinical evidence of symptomatic coronary artery disease in 13% of our patients. Only a minority had had any significant cardiac evaluations with just 27% having had an echocardiogram and less than 10% either a stress test or coronary angiogram. At this time there are, however, no well established guidelines for when and how often such tests should be performed in patients with CKD if they are asymptomatic and most such tests are performed after some sentinel clinical event (chest pain, MI of CHF) or as part of a pretransplant workup. Further prospective studies will be required to address these issues as well as the potential role of noninvasive testing, such as computed tomographic studies of cardiac calcification. Seventy percent of patients had had estimations of serum lipids, though most of these consisted of isolated measurements of serum cholesterol and triglycerides. There was a relatively high utilization of statins (37%), but the target cholesterol value of less than 200 mg/dL had been achieved in only half the patients.

Another disappointing finding was the poor documentation of access planning in the office chart. This does not, of course, mean that discussions did not take place, but that if they did they were not well documented. Many patients, when the prospect of access is first discussed are reluctant to agree to such surgery, which in the initial denial phase they may regard as “unnecessary.” They may also vacillate between the options of hemodialysis and peritoneal dialysis, thus further postponing access placement. Whatever the reason, the apparent poor level of access placement in this study was of concern, and again suggests that access placement, even in the hands of nephrologists, may be unduly delayed. Our subsequent follow-up study over the subsequent 2 years is somewhat reassuring in that seven of nine patients with an initial GFR < 20 who progressed to requiring dialysis had access placed in a timely fashion and only two required central catheters at the initiation of dialysis. In only one of the nine was dialysis instituted in the inpatient setting.

There are a number of limitations to this study. It is a retrospective chart review and cross-sectional in nature so that it included patients regardless of how often that they had been seen or how long they had been followed up. Thus the nephrologist may not have had the opportunity to respond to certain laboratory abnormalities or the effects of an instituted therapy may not have been realized as yet. Furthermore we confined ourselves only to the documentation that was immediately available in the nephrology clinic chart, It is possible that certain tests may have been performed elsewhere, e.g., in the primary care physician's or other subspecialist's (cardiologist, endocrinologist) office. These factors would, if anything, have had the effect of making delivery of care appear worse than it actually was. While most charts contained multiple determinations of serum creatinine and a diagnosis of CKD made by the attending nephrologist, in a few cases with little or no follow-up, the determination of CKD may have been based on only one serum creatinine and not have met the strict definition of the NKF which requires that the abnormality be present for at least 90 days.[18] Thus we may have inadvertently included some cases of acute renal failure. We have used primarily two previous studies of CKD for comparison but both studied patients at an earlier time period (1994–1998 and 1997–1999). While many of our findings were similar, there were subtle differences, particularly in the use of ACE inhibitors and the treatment of anemia. Whether these were due to an overall trend towards improved outcomes with time or to regional differences in medical care cannot be determined.

In summary, we have analyzed in detail the delivery of care to a cohort of CKD patients being managed in an academic nephrology practice. This study serves as a useful update in comparison with previous studies on this issue, particularly that by Kausz et al.,[10] performed in an earlier time period. Our study shows encouraging trends towards better outcomes in at least two areas with evidence of improvement in hemoglobin levels, use of angiotensin II blockade, and degree of blood pressure control. However it also reveals many areas where care could be improved. This was particularly true in the areas of access placement, bone and mineral metabolism, and cardiovascular disease. It will be interesting to see whether the recent introduction of CKD guidelines by the NKF and the associated educational initiatives for primary care physicians and nephrologists that are planned will further serve to improve the management of CKD patients in the years to come.