Incidence and Risk Factors for Pruritus in Patients with Nondialysis CKD

Visual Abstract

Abstract

Background

Pruritus is a common symptom experienced by patients with nondialysis CKD, but risk factors for incident pruritus in this patient population have not been evaluated.

Methods

We identified 1951 participants with CKD in the Chronic Renal Insufficiency Cohort Study without pruritus at the baseline assessment. Pruritus was assessed by the Kidney Disease Quality of Life-36 (KDQOL-36) instrument, and moderate-to-severe pruritus was defined as a response of 3 or higher on a Likert scale of 1–5. We used time-updated multivariable joint models to evaluate the association of patient clinical characteristics, eGFR, and laboratory parameters with incident pruritus.

Results

Over a median follow-up of 6 years, 660 (34%) participants developed incident moderate-to-severe pruritus, with a higher incidence rate observed among participants with more advanced CKD. In multivariable models, the hazard ratio (95% confidence interval [CI]) for pruritus associated with a 10 ml/min per 1.73 m² lower eGFR was 1.16 (95% CI, 1.10 to 1.23). Older age (≥65 years), higher body mass index, diabetes, current smoking, opioid use, depressive symptoms, and serum parathyroid hormone were also associated with a higher risk of incident pruritus, whereas low serum calcium (<9 mg/dl) was associated with a lower risk (all P<0.05). Serum phosphate was not associated with incident pruritus in the primary analysis.

Conclusions

A substantial proportion of patients with nondialysis CKD develop moderate-to-severe pruritus. Although lower eGFR is associated with the risk of pruritus, other comorbidities, particularly depressive symptoms, were potential risk factors.

Podcast

This article contains a podcast at https://dts.podtrac.com/redirect.mp3/www.asn-online.org/media/podcast/CJASN/2023_02_08_CJN09480822.mp3

Introduction

Although often thought of as a symptom of advanced CKD approaching dialysis, the prevalence of pruritus in patients with eGFR between 30 and 60 ml/min per 1.73 m² may approach or exceed 40%.^1,2 Reduced kidney function and the resultant accumulation of uremic toxins is generally considered the dominant risk factor for pruritus. Indeed, in a cross-sectional analysis, the CKD Outcomes and Practice Patterns Study (CKDopps) demonstrated a higher likelihood of experiencing pruritus among patients with more advanced CKD¹; however, this finding was not replicated in a separate study of the association between CKD stage and prevalent pruritus.³ Importantly, the association between kidney function and other risk factors for incident pruritus has never been evaluated in a prospective cohort of patients with nondialysis CKD.

Apart from changes in GFR, patients with CKD may also be at risk for pruritus due to dysregulation of calcium, phosphate, and parathyroid hormone (PTH) homeostasis or increased activation of inflammatory pathways.^4–6 Existing evidence to support these theories is either inconsistent, as is the case with markers of mineral metabolism,^7–11 or, as for inflammatory pathways, has yet to be explored in the CKD patient population. Finally, there is a growing appreciation that social and psychological determinants contribute to symptom burden in chronic disease, both the experience of symptoms and a patient's ability to cope with them.¹² For example, the prevalence of depressive symptoms among patients with CKD is high,¹³ and, in cross-sectional analyses, depressive symptoms are associated with pruritus¹ and overall symptom burden.¹⁴ Furthermore, elucidating potential biopsychosocial mechanisms causing pruritus in CKD is crucial to improving treatment strategies.

Thus, the primary purpose of the current investigation was to rigorously investigate clinical factors associated with incident moderate-to-severe pruritus among patients with nondialysis CKD. To address this goal, we used the rich clinical and biochemical data of the Chronic Renal Insufficiency Cohort (CRIC) Study, an ongoing longitudinal prospective cohort study of patients with CKD. The prospective nature of the CRIC Study allowed us to identify patients with CKD free of pruritus at the initial assessment and follow them longitudinally for the development of pruritus and evaluation of associated risk factors.

Methods

Study Design

The CRIC Study is a prospective, multicenter observational study of adults with CKD in the United States. A total of 3939 participants not on dialysis, aged 21–74 years, and with an eGFR between 20 and 70 ml/min per 1.73 m² were enrolled across seven clinical centers in the United States between 2003 and 2008. The CRIC Study inclusion and exclusion criteria have been previously published.¹⁵ Participants completed annual study visits at which numerous health measures were assessed and blood and urine specimens were collected. A total of 1951 CRIC Study participants were included in the present analysis. It is performed after excluding participants if they were missing an assessment of pruritus at the baseline visit (n=25), reported any degree of pruritus at baseline (n=1783), or were missing baseline assessment of any covariates included in our main model (model 2; n=180).

All participants provided written consent for the CRIC Study. The institutional review boards of the participating centers approved the study, and the research was conducted in accordance with the Declaration of Helsinki.

Study Data

eGFR was calculated using the CRIC GFR estimating equation, which includes serum creatinine and cystatin C levels in addition to age, sex, and race.¹⁶ The CRIC equation was selected because of its internal validity with reference to measured GFR. We considered risk factors for pruritus across multiple clinical domains, chosen on the basis of previous data from cross-sectional studies conducted primarily in the dialysis-dependent patient population, as noted below. Demographic risk factors were age, sex, race, and ethnicity. Clinical and behavioral factors included body mass index (BMI),¹⁷ history of diabetes,¹⁷ hypertension,¹⁸ coronary artery disease,¹⁷ peripheral vascular disease, congestive heart failure,⁹ lung disease (asthma or chronic obstructive pulmonary disease),^1,9 use of prescription opioid medications,¹⁹ depressive symptoms,¹ and current smoking.¹⁸ Depressive symptoms were assessed with the Beck's Depression Inventory (BDI) and classified as none, moderate, or severe (BDI scores 0–10, 11–13, and ≥14, respectively) on the basis of previous findings.^20,21 Markers of mineral metabolism were albumin-corrected serum calcium, phosphate, and intact parathyroid hormone (iPTH). Anemia was measured by hemoglobin level, and serum albumin, C-reactive protein (CRP), and interleukin 6 (IL-6) were included as markers of inflammation.

Outcomes and Censoring Events

Pruritus in CRIC was captured annually by a single item on the Kidney Disease Quality of Life-36 (KDQOL-36) instrument,²² which has been used by previous studies of symptoms in CKD to assess individual symptoms.^1,2,23 Participants reported to what extent they were bothered by “itchy skin” during the past 4 weeks on a Likert scale with five response options: (1) “Not at all bothered,” (2) “Somewhat bothered,” (3) “Moderately bothered,” (4) “Very much bothered,” and (5) “Extremely bothered.” Consistent with previous cross-sectional analyses, the outcome of moderate-to-severe pruritus was defined as the first occurrence of a response of 3–5 on this scale.¹

Participants were censored at the time of progression to dialysis-dependent CKD, receipt of a kidney transplant, death, loss to follow-up, or the end of the follow-up period, whichever occurred first. Ascertainment of outcomes in CRIC has been described previously.¹⁵ The end of follow-up for each participant was defined as the final study visit because pruritus was not assessed past this time point, and the date of administrative censoring is December 2020.

Statistical Analyses

Baseline characteristics were summarized by CKD stage using frequency and percentage for categorical variables and mean±SD for normal continuous variables or median and interquartile range (IQR) for continuous variables with a non-normal distribution. The incidence rate for pruritus was stratified by baseline characteristics, and the log-rank test for equality of survivor functions was used for unadjusted comparisons across groups.

Associations between risk factors and incident pruritus were examined using joint models. Using the stjm command in Stata, a linear mixed effects model was used to estimate the trajectory of eGFR (modeled as a continuous variable) on an individual level, which was then included simultaneously as an exposure variable in a Weibull proportional hazards survival model, thus capturing the relationship between the longitudinal variable and the survival outcome.^24,25 The Weibull model was chosen among parametric survival models on the basis of the lowest Akaike Information Criterion.²⁶ Primary models included only the current value of eGFR from the linear mixed effect model. Secondary models included both the current eGFR value and the derivative (i.e., slope) of eGFR to test whether the rate of change in eGFR was associated with the risk of pruritus. Nonlinearities of the association of eGFR with incident pruritus were tested by adding higher-order polynomials to the models.

We used a set of nested models to examine the relationship between risk factors and incident pruritus. Model 1 includes time-updated eGFR and baseline demographic features and clinical site. Model 2, our primary model, added time-updated clinical and behavioral covariates and baseline BDI score to Model 1. Model 3A included all covariates from Model 2, plus baseline markers of mineral metabolism. Model 3B included all covariates from Model 2, plus baseline hemoglobin and markers of inflammation. We included only the baseline values of depressive symptoms and laboratory parameters in our models because these variables may be a sequela of decreasing eGFR, and therefore, including them as time-updated covariates would bias our estimates.²⁷ All laboratory parameters were included as categorical variables in models as they did not seem to meet the assumption of linearity.

In an exploratory analysis, we used single-measure lasso logistic regression to quantify the relative prognostic value of all covariates in predicting an individual's risk for developing pruritus. To capture change in eGFR for this analysis, we restricted our sample size to 1668 participants with at least two measures of eGFR. We created variables to represent different metrics of eGFR: baseline eGFR, mean annual change in eGFR (computed for each participant using linear regression of eGFR on time), and absolute overall change in eGFR (computed as the change in eGFR from baseline to the final measure taken at the time of first pruritus reported or censoring). We then used lasso logistic regression with moderate-to-severe pruritus versus mild or no pruritus as the outcome for variable selection comparing three different methods for selection of the regularization parameter (λ): 10-fold cross-validation, minimum Bayesian information criterion, and adaptive lasso.

In a secondary analysis, proteinuria was added to all models as a log-transformed linear variable. In a sensitivity analysis, to test the robustness of our results, we repeated the primary analyses using new race-free CKD-EPI eGFR equations.²⁸ We also computed hazard ratios (HRs) for incident pruritus associated with a 10 ml/min per 1.73 m² lower eGFR for incremental durations of follow-up time to understand the potential for heterogeneity in period-specific HRs to affect overall HRs.²⁹

Missing values for covariates are shown in Supplemental Table 1. For missing covariates after the baseline visit, the last observed value was carried forward, except for eGFR, which was estimated by the joint model. Models 3A and 3B included 1876 and 1893 participants, respectively, because of a small percentage of participants with missing laboratory parameters at the baseline visit. All analyses were performed using Stata SE 17.0 (StataCorp, College Station, TX).

Results

At the baseline evaluation, CRIC Study participants without pruritus had a mean (±SD) age of 58±11 years, 44% identified as female, and 41% as Black (Table 1). Participants with more advanced CKD at baseline were more likely to be older, female, Black or Hispanic, current smokers, with higher BMI and higher prevalence of most comorbidities. The characteristics of CRIC Study participants reporting any degree of pruritus at the baseline visit (46%, n=1783), and thus, not included in the current study, are shown in Supplemental Table 2. On average, participants reporting pruritus at the baseline visit had lower eGFR, higher proteinuria, higher prevalence of comorbidities, and were more likely to report depressive symptoms.

Table 1.

Characteristics of 1951 Chronic Renal Insufficiency Cohort Study participants without pruritus at the baseline visit, stratified by baseline eGFR

Characteristics	Overall	Estimated GFR (ml/min per 1.73 m2)
		≥60	45–59	30–44	≤29
N (%)	1951	402 (21)	594 (30)	632 (32)	323 (17)
Demographic factors
Mean age, yr (SD)	58 (±11)	53 (±11)	58 (±11)	59 (±11)	58 (±11)
Female, N (%)	853 (44)	163 (41)	239 (40)	295 (47)	156 (48)
Race/ethnicity, N (%)
Non-Hispanic White	854 (44)	227 (56)	265 (45)	259 (41)	103 (32)
Non-Hispanic Black	805 (41)	134 (33)	238 (40)	276 (44)	157 (49)
Hispanic	217 (11)	18 (4)	67 (11)	77 (12)	55 (17)
Other	75 (4)	23 (6)	24 (4)	20 (3)	8 (2)
Kidney measures
Mean eGFR (SD), ml/min per 1.73 m2	47 (±17)	71 (±10)	52 (±4)	38 (±4)	24 (±4)
Median urine protein (IQR), mg/gCr	132 (54–622)	63 (39–135)	95 (50–313)	220 (66–876)	701 (148–2567)
Clinical factors
Mean body mass index (SD), kg/m2	32 (±7)	30 (±6)	32 (±7)	33 (±8)	32 (±8)
Diabetes, N (%)	886 (45)	114 (28)	253 (43)	333 (53)	186 (58)
Hypertension, N (%)	1673 (86)	276 (69)	523 (88)	575 (91)	299 (93)
Coronary artery disease, N (%)	378 (19)	44 (11)	110 (19)	136 (22)	88 (27)
Peripheral vascular disease, N (%)	96 (5)	7 (2)	18 (3)	41 (6)	30 (9)
Congestive heart failure, N (%)	132 (7)	14 (3)	26 (4)	46 (7)	46 (14)
Lung disease, N (%)	249 (13)	55 (14)	73 (12)	79 (12)	42 (13)
Current smoking, N (%)	237 (12)	36 (9)	63 (11)	84 (13)	54 (17)
Opioid use, N (%)	178 (9)	27 (7)	52 (9)	69 (11)	30 (9)
Depressive symptoms, N (%)
None (BDI 0–10)	1582 (81)	345 (86)	496 (84)	506 (80)	235 (73)
Moderate (BDI 11–13)	133 (7)	20 (5)	32 (5)	48 (8)	33 (10)
Severe (BDI ≥14)	236 (12)	37 (9)	66 (11)	78 (12)	55 (17)
Mineral metabolism markers
Mean serum calcium (SD), mg/dl	9.2 (±0.5)	9.2 (±0.4)	9.2 (±0.4)	9.3 (±0.5)	9.2 (±0.5)
Mean serum phosphate (SD), mg/dl	3.7 (±0.6)	3.4 (±0.5)	3.5 (±0.6)	3.8 (±0.6)	4.1 (±0.7)
Median intact PTH (IQR), pg/ml	51 (33–83)	34 (26–46)	44 (31–66)	61 (39–93)	108 (65–177)
Hemoglobin
Male: Mean hemoglobin (SD), g/dl	13.2 (±1.8)	14.1 (±1.4)	13.4 (±1.7)	12.9 (±1.6)	12.0 (±1.7)
Female: Mean hemoglobin (SD), g/dl	12.1 (±1.6)	12.9 (±1.4)	12.4 (±1.4)	11.8 (±1.5)	11.2 (±1.4)
Inflammatory markers
Median hsCRP (IQR), mg/L	2.3 (0.9–5.6)	1.3 (0.7–3.3)	2.4 (1.0–5.1)	2.6 (1.1–6.4)	2.8 (1.1–7.3)
Mean serum albumin (SD), g/dl	4.0 (±0.5)	4.1 (±0.4)	4.0 (±0.5)	3.9 (±0.4)	3.8 (±0.5)
Median IL-6 (IQR), pg/ml	1.7 (1.0–2.8)	1.0 (0.7–1.7)	1.6 (1.1–2.6)	2.0 (1.3–3.1)	2.5 (1.6–4.0)

Groups compared using ANOVA, Kruskal-Wallis, or Pearson's chi-squared test as appropriate. All P for trend <0.05, with the exception of lung disease and opioid use. Missing: 1.5% serum calcium, 1.6% serum phosphate, 1.9% intact PTH, 0.4% hemoglobin, 0.1% hsCRP, 1.5% serum albumin, 1.5% IL-6. IQR, interquartile range; BDI, Beck's Depression Inventory; PTH, parathyroid hormone; hsCRP, high-sensitivity C-reactive protein; IL-6, interleukin-6.

The median time window for this analysis was 1.1 (IQR 0.9–1.1) years, with a median of 5 (IQR 3–10) observation windows per participant. During follow-up, 660 (34%) participants developed incident moderate-to-severe pruritus (incidence rate: 5.0 per 100 person-years), with higher unadjusted incidence of pruritus among participants with more advanced CKD at baseline (Figure 1). Statistically significant (P<0.05) differences in unadjusted incidence rates were observed across categories of all considered risk factors except age, coronary artery disease, and lung disease (Supplemental Table 3). Participants with severe depressive symptoms, peripheral vascular disease, and prescription opioid use had the highest rates of incident pruritus. Of the 660 participants who developed moderate-to-severe pruritus, 107 (16%) did not have a subsequent assessment of pruritus due to censoring (i.e., progressed to dialysis-dependent CKD, receipt of kidney transplant, loss to follow-up, or end of observation period). Of the participants with a follow-up pruritus assessment, 82 (12%) did not report pruritus again, whereas 181 (27%) reported some degree of pruritus at every subsequent visit. The remaining 290 (44%) participants reported fluctuating pruritus status at follow-up assessments. A total of 159 of the 660 participants who developed moderate-to-severe pruritus eventually progressed to dialysis-dependent CKD, compared with 344 participants progressing to dialysis-dependent CKD among 1291 participants who did not develop pruritus.

Figure 1.

Proportion of 1951 Chronic Renal Insufficiency Cohort (CRIC) Study participants developing moderate-to-severe pruritus, stratified by baseline estimated GFR. Unadjusted Kaplan-Meier failure curve. Groups compared using the log-rank test. The time by which 25% of the patients developed incident moderate-to-severe pruritus was 4.1 (95% confidence interval [95% CI], 3.1 to 5.1) years for eGFR ≤29 ml/min per 1.73 m², 4.1 (95% CI, 3.2 to 5.0) years for eGFR 30–44 ml/min per 1.73 m², 5.1 (95% CI, 4.8 to 7.0) years for eGFR 45–59 ml/min per 1.73 m², and 10.0 (95% CI, 6.9 to 13.1) for eGFR ≥60 ml/min per 1.73 m². The interpretation of these times is as follows for the group with eGFR 30–44 ml/min per 1.73 m²: Among all individuals with eGFR in this range and no pruritus at baseline, one in four will develop moderate-to-severe pruritus by 4.1 years.

The estimated mean change in eGFR among participants in this study was −1.40 (95% confidence interval [95% CI], −1.50 to −1.30) ml/min per 1.73 m² per year, and 513 (32%) participants with eGFR ≥30 ml/min per 1.73 m² at the baseline visit progressed to CKD stage G4 during follow-up. Trajectories of eGFR over the follow-up duration appeared similar between participants who did and did not develop moderate-to-severe pruritus (Figure 2), yet eGFR was significantly associated with the risk of incident pruritus in all multivariable joint models. A 10 ml/min per 1.73 m² lower eGFR was associated with a 16% higher risk of incident moderate-to-severe pruritus in our primary model (Table 2, Model 2; 95% CI, 1.10 to 1.23). The slope of eGFR (e.g., annual rate of eGFR change), however, was not significantly associated with the risk of incident pruritus (results not shown). Older age (≥65), higher BMI, diabetes, current smoking, use of prescription opioid medications, and moderate and severe depressive symptoms were also significantly associated with a higher risk of incident pruritus (Table 2, Model 2). The magnitude of these associations remained relatively stable in models that further adjusted for markers of mineral metabolism, hemoglobin, and inflammatory markers (Table 2, Models 3A and 3B).

Figure 2.

Plot of longitudinal trajectory of eGFR in 1000 randomly selected CRIC Study participants included in the current study, stratified by outcome status. Initial eGFR values and eGFR trajectories are relatively similar between CRIC Study participants who do and do not develop moderate-to-severe pruritus. The time scale is adjusted relative to each participant's outcome event. The left-hand panel depicts eGFR trajectories for participants who did not develop pruritus (and were ultimately censored), whereas the right-hand panel depicts trajectories for participants who developed moderate-to-severe pruritus. Solid lines are lowess smoothers.

Table 2.

Multivariable associations of time-updated eGFR, time-updated clinical factors, baseline demographic factors, and baseline laboratory markers of mineral metabolism, anemia, and inflammation with risk of incident moderate-to-severe pruritus among Chronic Renal Insufficiency Cohort (CRIC) Study participants

Risk Factors	N Events	Unadjusted IR, per 100 Person-Years (95% CI)	Model 1 HR (95% CI)	Model 2: Main Model HR (95% CI)	Model 3A: Additional Analysis HR (95% CI)	Model 3B: Additional Analysis HR (95% CI)
eGFR (time-updated) per 10 ml/min per 1.73 m2 lower	—	—	1.20 (1.13 to 1.26)	1.16 (1.10 to 1.23)	1.16 (1.08 to 1.23)	1.14 (1.07 to 1.21)
Demographic factors (baseline only)
Age, yr
≤44	82	4.2 (3.4 to 5.3)	0.86 (0.67 to 1.10)	0.94 (0.73 to 1.21)	0.96 (0.74 to 1.25)	0.93 (0.72 to 1.21)
45–64	380	4.8 (4.4 to 5.4)	1.0 Ref	1.0 Ref	1.0 Ref	1.0 Ref
≥65	198	5.6 (4.9 to 6.5)	1.20 (1.00 to 1.43)	1.31 (1.09 to 1.57)	1.33 (1.10 to 1.60)	1.29 (1.08 to 1.56)
Sex
Male	341	4.5 (4.1 to 5.0)	1.0 Ref	1.0 Ref	1.0 Ref	1.0 Ref
Female	319	5.6 (5.0 to 6.2)	1.25 (1.06 to 1.46)	1.15 (0.98 to 1.36)	1.14 (0.96 to 1.35)	1.15 (0.97 to 1.36)
Race/ethnicity
Non-Hispanic White	282	4.2 (3.7 to 4.7)	1.0 Ref	1.0 Ref	1.0 Ref	1.0 Ref
Non-Hispanic Black	284	5.8 (5.1 to 6.5)	1.31 (1.11 to 1.56)	1.13 (0.94 to 1.35)	1.03 (0.85 to 1.25)	1.09 (0.91 to 1.32)
Hispanic	69	6.2 (4.9 to 7.8)	1.12 (0.80 to 1.55)	0.92 (0.66 to 1.28)	0.86 (0.61 to 1.20)	0.89 (0.63 to 1.25)
Other	25	4.9 (3.3 to 7.2)	1.09 (0.71 to 1.66)	1.11 (0.73 to 1.71)	1.07 (0.69 to 1.65)	1.10 (0.71 to 1.69)
Clinical factors (time-updated)
BMI, per 5 kg/m2 higher	—	—	—	1.10 (1.04 to 1.16)	1.09 (1.04 to 1.15)	1.08 (1.02 to 1.15)
Diabetes
No	339	4.1 (3.7 to 4.6)	—	1.0 Ref	1.0 Ref	1.0 Ref
Yes	321	6.3 (5.7 to 7.1)	—	1.24 (1.05 to 1.46)	1.24 (1.04 to 1.47)	1.18 (0.99 to 1.40)
Hypertension
No	90	4.0 (3.3 to 4.9)	—	1.0 Ref	1.0 Ref	1.0 Ref
Yes	570	5.2 (4.8 to 5.6)	—	0.89 (0.66 to 1.21)	0.89 (0.65 to 1.21)	0.93 (0.68 to 1.27)
Coronary artery disease
No	528	4.8 (4.4 to 5.2)	—	1.0 Ref	1.0 Ref	1.0 Ref
Yes	132	5.9 (5.0 to 7.0)	—	1.03 (0.85 to 1.25)	1.01 (0.83 to 1.24)	1.01 (0.83 to 1.23)
Peripheral vascular disease
No	621	4.8 (4.5 to 5.2)	—	1.0 Ref	1.0 Ref	1.0 Ref
Yes	39	8.5 (6.2 to 11.7)	—	1.22 (0.89 to 1.66)	1.10 (0.79 to 1.53)	1.16 (0.85 to 1.60)
Congestive heart failure
No	615	4.8 (4.5 to 5.2)	—	1.0 Ref	1.0 Ref	1.0 Ref
Yes	45	7.7 (5.7 to 10.3)	—	1.13 (0.84 to 1.52)	1.17 (0.87 to 1.58)	1.10 (0.82 to 1.47)
Lung disease
No	568	4.9 (4.5 to 5.3)	—	1.0 Ref	1.0 Ref	1.0 Ref
Yes	92	5.7 (4.7 to 7.0)	—	1.10 (0.90 to 1.35)	1.09 (0.89 to 1.34)	1.08 (0.88 to 1.33)
Opioid use
No	577	4.7 (4.3 to 5.1)	—	1.0 Ref	1.0 Ref	1.0 Ref
Yes	83	8.0 (6.5 to 9.9)	—	1.27 (1.01 to 1.59)	1.30 (1.03 to 1.64)	1.27 (1.01 to 1.60)
Current smoking
No	560	4.7 (4.3 to 5.1)	—	1.0 Ref	1.0 Ref	1.0 Ref
Yes	100	7.9 (6.5 to 9.6)	—	1.60 (1.27 to 2.01)	1.59 (1.26 to 2.01)	1.55 (1.23 to 1.96)
Depressive symptomsa
None (BDI 0–10)	500	4.4 (4.0 to 4.8)	—	1.0 Ref	1.0 Ref	1.0 Ref
Moderate (BDI 11–13)	49	6.9 (5.2 to 9.1)	—	1.41 (1.04 to 1.92)	1.50 (1.10 to 2.04)	1.46 (1.08 to 1.99)
Severe (BDI ≥14)	111	9.6 (8.0 to 11.6)	—	2.01 (1.61 to 2.51)	2.03 (1.61 to 2.55)	1.98 (1.57 to 2.48)
Markers of mineral metabolism (baseline only)
Serum calcium,a mg/dl
<9	167	4.1 (3.5 to 4.8)	—	—	0.80 (0.66 to 0.97)	—
9–9.4	320	5.3 (4.8 to 6.0)	—	—	1.0 Ref	—
≥9.5	164	5.4 (4.6 to 6.3)	—	—	0.94 (0.78 to 1.15)	—
Serum phosphate,a mg/dl
<3	66	3.9 (3.1 to 5.0)	—	—	0.94 (0.71 to 1.24)	—
3–3.9	394	4.8 (4.3 to 5.3)	—	—	1.0 Ref	—
≥4.0	186	6.0 (5.2 to 6.9)	—	—	1.10 (0.92 to 1.33)	—
Intact PTH,a pg/ml
0–64	410	4.4 (4.0 to 4.8)	—	—	1.0 Ref	—
≥65	233	6.2 (5.5 to 7.1)	—	—	1.30 (1.08 to 1.55)	—
Markers of anemia and inflammation (baseline only)
Hemoglobina
M: <12 g/dl; F: <11 g/dl	153	6.9 (5.9 to 8.1)	—	—	—	1.19 (0.97 to 1.46)
M: 12–14 g/dl; F: 11–13 g/dl	503	4.6 (4.2 to 5.0)	—	—	—	1.0 Ref
hsCRP,a mg/L
0–3.0	359	4.5 (4 to 4.9)	—	—	—	1.0 Ref
>3	300	5.7 (5.1 to 6.4)	—	—	—	1.03 (0.87 to 1.23)
Serum albumin,a g/dl
<3.5	75	7.3 (5.8 to 9.2)	—	—	—	1.31 (1.00 to 1.73)
3.5–3.9	238	5.8 (5.1 to 6.5)	—	—	—	1.20 (1.01 to 1.43)
≥4.0	338	4.3 (3.8 to 4.8)	—	—	—	1.0 Ref
IL-6,a pg/ml
0–3.0	492	4.5 (4.1 to 4.9)	—	—	—	1.0 Ref
>3	158	7.2 (6.2 to 8.5)	—	—	—	1.17 (0.95 to 1.43)

Bold font denotes a P value <0.05. Results shown above are from joint modeling of linear mixed effects models and proportional hazards survival models. Models were also adjusted for clinical center site. Model 2, the main model, included 1951 participants with complete covariate data. Model 3A included 1876 CRIC participants and model 3B included 1893 CRIC participants because of a small percentage of missing values for either calcium, phosphate, PTH or hemoglobin, hsCRP, albumin, or IL-6, respectively, at the baseline visit. Unadjusted incidence rates were calculated using the value of the variable at the baseline visit; however, some of these variables (e.g., clinical factors) were included as time-updated variables in joint models. IR, incidence rate; 95% CI, 95% confidence interval; HR, hazard ratio; BMI, body mass index; BDI, Beck's Depression Inventory; PTH, parathyroid hormone; M, male; F, female; hsCRP, high-sensitivity C-reactive protein; IL-6, interleukin-6.

^aTo reduce biases in our estimates that can result from controlling for sequelae of changing eGFR, only baseline values of these variables were included in models. All other variables included as time-updated variables in models.

A serum calcium level <9 mg/dl was associated with a lower risk of incident pruritus (HR 0.80; 95% CI, 0.66 to 0.97), and a PTH level ≥65 pg/ml was associated with a higher risk of moderate-to-severe pruritus (HR, 1.30; 95% CI, 1.08 to 1.55). Of the markers of anemia and inflammation tested, only a serum albumin of 3.5–3.9 g/dl was significantly associated with a higher risk of incident pruritus (HR, 1.20; 95% CI, 1.01 to 1.43).

In exploratory analyses, the absolute value of the change in eGFR from the first observed measurement to the last observed measurement was the strongest predictor of pruritus (Table 3). A 10% increase or decrease in weight from baseline, sex, BMI at baseline, and depressive symptoms were also consistently chosen by all three lasso model variations. The mean annual rate of change in eGFR was not identified as a predictor of pruritus status by any lasso model, and the only laboratory parameter selected by two of the three lasso models was serum phosphate level at baseline.

Table 3.

Use of lasso logistic regression to select variables best at predicting the outcome of moderate-to-severe pruritus among 1668 Chronic Renal Insufficiency Cohort Study participants

Selected Predictors (Listed in Descending Order of Coefficient Magnitude)	Method of Lasso Lambda (λ) Selectiona
	Cross-Validation	Minimize BIC	Adaptive
Change in eGFR, absolute value	✓	✓	✓
10% decrease in weight from baseline	✓	✓	✓
Sex	✓	✓	✓
10% increase in weight from baseline	✓	✓	✓
Body mass index at baseline	✓	✓	✓
BDI score ≥11	✓	✓	✓
Baseline eGFR	✓		✓
Serum phosphate at baseline	✓		✓
New lung disease	✓		✓
New congestive heart failure	✓		✓
New peripheral vascular disease	✓		✓
New prescription opioid use	✓		✓
New hypertension	✓
Prescription opioid use at baseline	✓
Diabetes at baseline	✓
Peripheral vascular disease at baseline	✓
CRP at baseline	✓
PTH at baseline	✓
New diabetes	✓
Urine protein	✓
Race/ethnicity	✓
New coronary artery disease	✓

Variables are listed in order of the absolute values of their coefficients, with largest values first. All variables were standardized to facilitate comparisons of relative size of regression coefficients. Variables selected repeatedly by lasso regression models using different lambda tuning parameters demonstrate a more robust association with the outcome. In addition to comorbidities at baseline and baseline laboratory parameters, variables were created to represent a new diagnosis of diabetes, hypertension, coronary artery disease, peripheral vascular disease, congestive heart failure, or new use of opioid medications during follow-up. A 10% increase or decrease in weight refers to the percent change in weight as compared with the weight at the baseline visit. The absolute change in eGFR was calculated by subtracting the eGFR at the time of the outcome or censoring event from the baseline value. Because we wanted to create variables to capture change in eGFR during follow-up, this analysis was restricted to the 1668 participants with at least two measurements of eGFR. BDI, Beck's Depression Inventory; CRP, C-reactive protein; PTH, parathyroid hormone.

^aCross-validation chooses lambda on the basis of the model that minimizes the 10-fold cross-validation function. Minimize BIC refers to the lambda that results in the model with the smallest value for the Bayesian information criterion (BIC). Adaptive lasso also uses cross-validation but uses a different penalization criterion in which coefficients receive different data-driven weights.

When proteinuria was added to our primary model, along with eGFR, there was no significant association between proteinuria and the risk of pruritus (Supplemental Table 4).

When the analyses were repeated using CKD-EPI race-free eGFR equations, the results remained overall unchanged (Supplemental Table 5). Finally, hazard ratios for the primary exposure variable, eGFR, calculated using incremental durations of follow-up time remained relatively constant (Supplemental Table 6).

Discussion

In this prospective cohort study of CRIC participants, we found that the incidence of moderate-to-severe pruritus was high across all stages of CKD and that lower eGFR was associated with higher risk of pruritus. We identified the absolute change in eGFR as the most significant risk factor for incident pruritus and found that the rate of change in eGFR was not associated with risk for pruritus. Our findings also suggest additional clinical factors, including higher BMI, current smoking, prescription opioid use, and depressive symptoms, that are associated with a higher risk of pruritus. Although lower serum calcium and high intact PTH were associated with pruritus risk, serum phosphate and markers of anemia and inflammation were not.

Ours is the first study of incident pruritus in nondialysis CKD. We demonstrate that one third of patients with CKD will develop pruritus. Notably, at least one in four participants who started with an eGFR ≥60 ml/min per 1.73 m² eventually developed pruritus in this study. Other studies in nondialysis CKD report a prevalence of pruritus ranging from 19% to 29%,^1,3,7 in comparison with an overall prevalence of pruritus among dialysis patients of around 55%.³⁰ Given that pruritus is associated with a higher risk for reduced quality of life and mortality among patients with kidney disease,^9,17,31,32 increased awareness and consideration of treatment of this symptom by CKD providers is warranted.

To the best of our knowledge, this study is the first to rigorously establish lower eGFR as a risk factor for incident pruritus using a prospective cohort of patients with CKD. The findings from both our primary and exploratory analyses suggest that it is the absolute change in eGFR, regardless of the starting value of eGFR or the rate of change in eGFR that is independently associated with a higher risk of developing pruritus. The association between eGFR and incident pruritus supports the hypothesis that the accumulation of uremic toxins may play a role in the pathogenesis of pruritus, a hypothesis that is further supported by observations in kidney transplant recipients of a marked improvement in pruritus after transplant.^23,33 Future studies utilizing metabolomics or proteomics in the CKD population could help to identify the specific uremic toxins associated with pruritus.

In the current study, depressive symptoms were a strong predictor of pruritus status. This is similar to a large study of Japanese hemodialysis patients in which depressive symptoms were significantly associated with the risk of developing pruritus³⁴ and a recent finding that depressive symptoms are associated with increased symptom burden in US hemodialysis patients.¹⁴ Physician-diagnosed depression was also a risk factor for prevalent pruritus in participants of CKDopps with nondialysis CKD.¹ It is possible that the association between depressive symptoms and pruritus may be mediated by poor sleep, as key regulatory mechanisms of the skin are altered during sleep.³⁵ Or perhaps, more generally, depressive symptoms impair a patient's ability to cope with symptoms of chronic disease, thus resulting in a more intense symptomatic experience. The results from our exploratory analysis suggest that a 10% weight change in either direction is a strong predictor of pruritus, relative to other clinical factors tested, and a higher BMI was associated with a higher risk of pruritus in our primary analysis. We believe our study is the first to demonstrate an association between weight and pruritus in nondialysis CKD; these results are similar to those in patients with psoriasis and may be a result of increased weight leading to higher ambient body temperature and increased itch sensation.^36,37 Our findings of higher risk for pruritus among older participants are supported by existing literature demonstrating a greater symptom burden in this population.³⁸ The association of diabetes with pruritus in our study is similar to that found in patients on hemodialysis and may be mediated by peripheral neuropathy.¹⁷ Similarly, smoking has also been linked to higher prevalence of pruritus among hemodialysis patients^9,18 and may be mediated by higher levels of pruritus-inducing cytokines, such as interleukin-31.^39,40 Finally, we demonstrate that use of prescription opioid medications is associated with a higher risk for developing pruritus. Opioid medications, such as morphine, which act as agonists on µ-opioid receptors, inhibit pain but can induce or enhance pruritus.^41,42 On the other hand, peripheral activation of κ-opioid receptors by medications such as difelikefalin has been shown to significantly reduce pruritus in patients on hemodialysis,⁴³ as has the use of µ-opioid receptor antagonists such as naloxone,⁴⁴ supporting the notion that an imbalance of opioid receptor activation seems to play a role in the pathogenesis of pruritus.

Markers of mineral metabolism have been inconsistently associated with pruritus in nondialysis CKD^1,3,7,8 and hemodialysis^9,11,45 populations. Contrary to common dogma, we found no evidence for an association between phosphate levels and the risk of pruritus in our main models, and ours is the first study to examine this risk factor prospectively in nondialysis CKD. Increased levels of extracellular free calcium ions have been observed in the deep epidermis of patients with kidney disease–associated pruritus.⁴⁶ This observation may somewhat account for the protective role of lower serum calcium levels that we observed. Our results also demonstrated a higher risk for pruritus associated with elevated PTH, independent of serum calcium. There is historic evidence for improvement of pruritus after parathyroidectomy,⁴⁷ but the causal pathway between PTH and pruritus is not well established. Finally, we considered the hypothesis that uremic pruritus is because of a chronic proinflammatory state.⁴⁸ Although we found an association between lower serum albumin and the risk of pruritus, we did not find an association between CRP or IL-6 and incident pruritus.

The strengths of our study include that CRIC is a large, multicenter, prospective study of US adult patients with CKD, capturing repeated assessments of pruritus over time. The prospective nature of the CRIC Study allowed for a longitudinal assessment of incident pruritus and associated risk factors, thus our results are more robust than cross-sectional analyses. Besides, given the diversity of the CRIC Study patient population, our results are generalizable to most adult patients with CKD in the United States. Several limitations should, however, be noted. This study used a widely available measure, the KDQOL-36, to capture patient-reported pruritus. This is consistent with other studies in the field,^1,2,23 although we recognize that additional work is needed to validate this approach. In addition, the single question on the KDQOL-36 was not able to capture information regarding the location, timing, or frequency of pruritus. Similarly, we were not able to account for known liver disease or possible co-occurring dermatologic conditions among participants, e.g., eczema or psoriasis that could cause pruritus.

In conclusion, our study characterized the high incidence of pruritus in nondialysis CKD and identified several clinical factors, in addition to lower eGFR that were associated with higher risk of moderate-to-severe pruritus. In particular, depressive symptoms and weight warrant further investigation as potentially modifiable risk factors that could be targeted for a multifaceted approach to the treatment of pruritus in CKD. We suggest that providers caring for patients with CKD should anticipate a high likelihood that their patients may develop pruritus and, thus, incorporate frequent screening for pruritus and associated risk factors into routine clinical care.

Supplementary Material

cjasn-18-193-s001.pdf ^{(246.9KB, pdf)}

Disclosures

S. Kalim reports serving on the speakers bureau for Fresenius Kabi. T. Shafi reports consultancy agreements with Cara Therapeutics and Siemens; research funding from Baxter (clinical trial site investigator), Bayer (clinical trial site investigator), CVS (clinical trial site investigator), Goldfinch (clinical trial site investigator), Natera (clinical trial site investigator), and Vertex (clinical trial site investigator); grants from the NIH during the conduct of the study; honoraria from Cara Therapeutics and Siemens; and advisory or leadership roles for American Journal of Kidney Diseases, American Journal of Medicine, CJASN, and Kidney360. All remaining authors have nothing to disclose.

Funding

K.E. Wulczyn is supported by the National Institutes of Health (NIH) grants T32DK007540-34 and F32DK131793. S. Kalim is supported by NIH grant R01DK124453. E.P. Rhee and T. Shafi are supported by NIH grant R01-NR017399. L. Myint reports funding from NIH grant R01-NR017399.

Author Contributions

S. Kalim, E.P. Rhee, T. Shafi, and K.E. Wulczyn conceptualized the study; K.E. Wulczyn was responsible for data curation and the formal analysis; L. Myint, T. Shafi, and K.E. Wulczyn were responsible for the methodology; S. Kalim and T. Shafi were responsible for supervision; K.E. Wulczyn was responsible for writing the original draft; and the remaining authors contributed to reviewing and editing the manuscript.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/CJN/B557.

Supplemental Table 1. Number of observations included in present analysis stratified by visit year, and percentage of missing values for variables among CRIC Study participants included in the current analysis.

Supplemental Table 2. Baseline characteristics of the CRIC cohort stratified by prevalent pruritus at the baseline visit.

Supplemental Table 3. Incidence rates of moderate-to-severe pruritus stratified by baseline characteristics among CRIC Study participants.

Supplemental Table 4. Multivariable associations of demographic and clinical factors, including eGFR and urine protein, and risk of incident moderate-to-severe pruritus among 1951 CRIC Study participants.

Supplemental Table 5. Multivariable associations of demographic and clinical factors, using new race-free eGFR equations, and risk of incident moderate-to-severe pruritus among 1951 CRIC Study participants.

Supplemental Table 6. Hazard ratios for the risk of incident pruritus associated with a 10 ml/min/1.73 m² lower eGFR calculated for incremental durations of follow-up time.