Importance of creatinine clearance for drug dosing in nursing home residents

Abstract

Serum creatinine is often not an adequate measure of renal function, especially in advanced age or in physically debilitated patients. Estimated creatinine clearance is necessary to decide on usage of drugs such as Metformin. This study included 64 nursing home residents with diabetes treated with Metformin. Creatinine clearance (CrCl) was calculated by the Cockcroft–Gault equation modified for ideal body weight. CrCl more than or equal to 60/mL min was used as a cut-off for appropriate use of Metformin. In our sample, 20.3% had renal failure when measured by serum creatinine while 56.3% had renal failure when measured by CrCl. Age >65 years and women were more likely to be classified as normal for serum creatinine but have abnormal creatinine clearance. Use of estimated creatinine clearance should be advocated instead of serum creatinine when prescribing Metformin, especially for those of older age and among women.

• Creatinine
• diabetes mellitus
• elderly
• metformin
• renal insufficiency

Introduction

Creatinine clearance (CrCl) is a more accurate measurement of renal function than serum creatinine.1 One major factor that affects serum creatinine levels is muscle mass. As people age, their muscle mass decreases;2,3 simultaneously many older people also undergo a decline in renal function.3,4 Since serum creatinine falls as muscle mass decreases, this renal function decline may appear as normal serum creatinine but with abnormal CrCl. Many nursing home residents are older and/or physically debilitated, with reduced muscle mass. They are particularly vulnerable to this phenomenon of renal function discrepancy between serum creatinine and CrCl measurements.

As muscle mass plays an important role in determining serum creatinine, nursing home residents with reduced muscle mass require a more accurate measure of their renal function (e.g., CrCl to determine drug dosing). This is particularly relevant for drugs whose clearance or potential side effects are affected by renal function. For example, Metformin, used to treat type 2 diabetes,5,6 has been associated with lactic acidosis in patients with reduced CrCl.7,8 In the USA, 17% of patients diagnosed with lactic acidosis due to Metformin were over 80 years of age.9 Therefore, nursing homes residents are potentially at risk for such concerns.

In a pharmacokinetic evaluation of Metformin using CrCl to assess renal function, patients with normal renal function had a significantly lower mean half-life as compared to patients with renal dysfunction.10 It is recommended to avoid Metformin in renal impairment, defined as serum creatinine levels of 1.5 mg/dL in men and 1.4 mg/dL in women or with abnormal CrCl.11 Varying levels of CrCl ranging from 30 mL/min to 70 mL/min,12–17 have all been suggested as limits below which Metformin should not be prescribed. However, clinical studies have not established a specific limit of CrCl below which Metformin would be inappropriate for use.

The risks and benefits of the use of Metformin need to be weighed in deciding whether patients with renal impairment should be treated with Metformin. However, the studies above were generally performed in community-based practices. To our knowledge no study exclusively focused on a nursing home population. We evaluated nursing home residents with diabetes to assess if appropriate considerations with regard to renal function were made before prescribing Metformin. We studied whether there are any discrepancies between serum creatinine and CrCl, the variables which would predict abnormal creatinine clearance, and analyzed the usage of Metformin in a nursing home setting.

Methods

This is a retrospective cross-sectional study which included all 64 residents of a suburban area nursing home with a history of diabetes mellitus type 2 being treated with Metformin during August 2009. The study received Institutional Review Board (IRB) approval and was performed in accordance with the ethical standards of the 2000 Declaration of Helsinki and 2008 Declaration of Istanbul. The study received ethics approval to not provide informed consent due to the retrospective nature of the study. Data regarding age, sex, race, height and creatinine were obtained from nursing home charts. None of the patients were on medications that are nephrotoxic or that inhibit secretion of creatinine. Ideal body weight was calculated with height of the patient and sex as per the ideal body weight (IBW) calculator developed by Robinson et al.18 For men of all frame sizes, IBW = 51.65 kg + 1.85 kg/inch for height greater than 5 feet. For women of all frame sizes, IBW = 48.67 kg + 1.65 kg/inch for height greater than 5 feet. Creatinine clearance was calculated using the Cockcroft–Gault equation.19

For adult men, the calculation is as follows:

For adult women, the obtained CrCl value was multiplied by 0.85 to reflect smaller muscle mass. Based on a review of prior recommendations, we used a cut-off value of 60 mL/min for CrCl to determine abnormal renal function for both men and women, when Metformin should be evaluated for risks and benefits, before choosing to use it as a treatment for type 2 diabetes mellitus.9,15,16

Statistical analyses

Descriptive statistics of mean and standard deviation were used to describe the continuous variables and percentage and frequency were used to describe the categorical variables. Analysis of variance (ANOVA) was used to compare the normal and abnormal creatinine clearance groups with the outcome variable of Metformin dose. As serum creatinine had a skewed distribution, the Spearman correlation was used for correlation analyses of Metformin dose, serum creatinine, and creatinine clearance. Due to small subgroups, the Fisher’s exact test was used to compare normal and abnormal classifications for serum creatinine and creatinine clearance in the whole sample and also when stratifying for age groups, sex groups, and race/ethnicity groups. Logistic regression was used for the separate outcome variables of classification discrepancy and abnormal creatinine clearance. All p values were two-sided and considered significant at p < 0.05. SPSS Statistics Version 19.0 (Chicago, IL) was used for all analyses.

Results

Table 1 shows the characteristics of the sample of those with diabetes that were prescribed metformin. Age was almost equally distributed among those below and above 65 years with slightly more individuals below 65 years. Sex was almost equally distributed among men and women with slightly more men. With regard to race/ethnicity, Whites and African Americans comprised slightly more than three-quarters of the sample. Almost one-fifth of the serum creatinine levels were abnormal consistent with renal failure, while slightly more than half of the creatinine clearances were abnormal consistent with renal failure. The mean Metformin dose was approximately 1400 mg/day.

Table 1. Demographic characteristics of the sample of those with diabetes prescribed Metformin from a nursing home setting.

Variables	Percentage (frequency)	Mean (standard deviation)
Age		65.7 (13.2)
Younger (≤65 years)	57.8% (37)
Older (>65 years)	42.2% (27)
Sex
Men	54.7% (35)
Women	45.3% (29)
Race/ethnicity
White	40.6% (26)
African American	37.5% (24)
Asian	18.8% (12)
Hispanic	3.1% (2)
Serum creatinine		1.3 (1.3)
Normal ≤1.3 mg/dL	79.7% (51)
Abnormal >1.3 mg/dL	20.3% (13)
Creatinine clearance		61.7 (33.9)
Normal (>60) mL/min	43.8% (28)
Abnormal (<60) mL/min	56.3% (36)
Metformin dose (mg/day)		1,403.9 (505.9)

There were no significant differences (p = 0.62) for Metformin dose for the normal (Metformin = 1367.9 mg/day, SD = 515.5) and abnormal (Metformin = 1431.9 mg/day, SD = 503.9) creatinine clearance groups by ANOVA. Spearman correlation analyses did not show any significant correlation of Metformin dose used with either creatinine clearance or serum creatinine.

Table 2 shows comparisons of classifications for normal and abnormal serum creatinine and creatinine clearance. Among the whole sample, the percentages of abnormal creatinine clearance significantly differed for those classified as normal and abnormal for serum creatinine. This included that 47.1% classified as normal for serum creatinine was classified as abnormal for creatinine clearance. Moreover, for both older age (>65 years) and women, 85% and 81%, respectively, were classified as normal for serum creatinine but had abnormal creatinine clearance, whereas only 22.6% and 23.3% younger age residents and men respectively had normal serum creatinine levels with abnormal creatinine clearances.

Table 2. Comparisons of serum creatinine and creatinine clearance.

Variables	Serum creatinine Normal <1.3 mg/dL % (Frequency)	Serum creatinine Abnormal >1.3 mg/dL % (Frequency)
Whole sample (n = 64)
Creatinine clearance
Normal (>60 mL/min)	52.9% (27)	7.7% (1)
Abnormal (≤60 mL/min)	47.1% (24)	92.3% (12)**
Age
Younger (≤65 years) (n = 37)
Creatinine clearance
Normal (≥60 mL/min)	77.4% (24)	16.7% (1)
Abnormal (≤60 mL/min)	22.6% (7)	83.3% (5)**
Older (>65 years) (n = 27)
Creatinine clearance
Normal (>60 mL/min)	15.0% (3)	0.0% (0)
Abnormal (≤60 mL/min)	85.0% (17)	100.0% (7)
Sex
Men (n = 35)
Creatinine clearance
Normal (>60 mL/min)	76.7% (23)	20.0% (1)
Abnormal (≤60 mL/min)	23.3% (7)	80.0% (4)*
Women (n = 29)
Creatinine clearance
Normal (>60 mL/min)	19.0% (4)	0.0% (0)
Abnormal (≤60 mL/min)	81.0% (17)	100.0% (8)
Race/ethnicity
White (n = 26)
Creatinine clearance
Normal (>60 mL/min)	59.1% (13)	25.0% (1)
Abnormal (≤60 mL/min)	40.9% (9)	75.0% (3)
African American (n = 24)
Creatinine clearance
Normal (>60 mL/min)	58.8% (10)	0.0% (0)
Abnormal (≤60 mL/min)	41.2% (7)	100.0% (7)*
Asian (n = 12)
Creatinine clearance
Normal (>60 mL/min)	27.3% (3)	0.0% (0)
Abnormal (≤60 mL/min)	72.7% (8)	100.0% (1)

Notes: Comparisons performed with the Fisher’s exact test. Analyses were not performed for the race/ethnicity category of Hispanic due to too few participants (n = 2).

*p < 0.05; **p < 0.01.

Table 3 shows analyses for predictors of classification discrepancy for serum creatinine and creatinine clearance. In both univariate and multivariate analyses, older age, women, and normal serum creatinine had significant high odds ratios for predicting this classification discrepancy. Race/ethnicity did not significantly predict classification discrepancy.

Table 3. Predictors of normal/abnormal classification discrepancy for serum creatinine and creatinine clearance.

Variables	Univariate discrepancy (n = 64) OR (95% CI)	Multivariate discrepancy (n = 64) OR (95% CI)
Age
Younger (≤65 years)	1.00	1.00
Older (>65 years)	6.16 (2.04, 18.62)**	11.76 (2.52, 54.96)**
Sex
Men	1.00	1.00
Women	4.78 (1.62, 14.09)**	7.52 (1.74, 32.62)**
Race/ethnicity
White, Asian, Hispanic	1.00	1.00
African American	0.50 (0.17, 1.48)	0.44 (0.13, 2.43)
Serum creatinine
Abnormal >1.3 mg/dL	1.00	1.00
Normal ≤1.3 mg/dL	10.67 (1.29, 88.23)*	50.12 (3.85, 652.93)**

Notes: Discrepancy is indicated by either (1) serum creatinine as abnormal or creatinine clearance as normal (n = 1), or (2) serum creatinine as normal and creatinine clearance as not normal (n = 24). OR = odds ratio, CI = confidence interval.

*p < 0.05; **p < 0.01.

Table 4 shows analyses for predictors of abnormal creatinine clearance. In both univariate and multivariate analyses, older age and women had significant high odds ratios for predicting abnormal creatinine clearance. Older age had slightly larger odds ratios than women and abnormal serum creatinine for these significant predictors in both the univariate and multivariate analyses. Race/ethnicity did not significantly predict abnormal creatinine clearance.

Table 4. Predictors of abnormal creatinine clearance.

Variables	Univariate abnormal creatinine clearance (n = 64) OR (95% CI)	Multivariate abnormal creatinine clearance (n = 64) OR (95% CI)
Age
Younger (≤65 years)	1.00	1.00
Older (>65 years)	16.67 (4.18, 66.49)***	27.78 (4.17, 185.03)**
Sex
Men	1.00	1.00
Women	13.64 (3.81, 48.76)***	20.38 (3.39, 122.65)**
Race/ethnicity
White, Asian, Hispanic	1.00	1.00
African American	1.15 (0.41, 3.19)	0.95 (0.19, 4.85)
Serum creatinine
Normal ≤1.3 mg/dL	1.00	1.00
Abnormal >1.3 mg/dL	13.50 (1.63, 111.66)*	20.54 (1.45, 290.98)*

Notes: OR = odds ratio, CI = confidence interval.

*p < 0.05; **p < 0.01; ***p < 0.001.

Discussion

Our study demonstrates that more than half of elderly diabetic nursing home residents were prescribed Metformin despite having an estimated CrCl <60 mL/minute clearance. This still represents an underestimation of renal impairment in this generally debilitated population, particularly in the elderly and among female residents. The Cockcroft–Gault equation using ideal body weight as per the Robinson equation is essentially using height as a measure of ideal body weight. In physically debilitated patients this may be an overestimation of their muscle mass and therefore their renal function.

In our nursing home setting, among those with diabetes on Metformin, 20.3% had renal failure when measured by serum creatinine while more than double that number at 56.3% had renal failure when measured by CrCl. We suggest that this discrepancy is due to the presence of decreased muscle mass which is considered as part of the calculation for creatinine clearance.

In our study we found that nursing home residents younger than 65 years had a significant relationship between creatinine and normal creatinine clearance. However, for both older age (>65 years) and women, 85% and 81%, respectively, were classified as normal for serum creatinine but had abnormal creatinine clearance. This is potentially troublesome when prescribing for older age and women Metformin or any other medication which can be affected by renal clearance.

It is not surprising that in our study in both univariate and multivariate analyses, older age and women had significant high odds ratios for predicting abnormal creatinine clearance and classification discrepancy when comparing serum creatinine to creatinine clearance. This is because with older age muscle mass goes down and usually women have less muscle mass as compared to men in all age groups.

One may have expected a better relationship of normal serum creatinine with normal creatinine clearance for African Americans since they tend to have a higher amount of muscle mass as compared to Caucasians.4 However, since the Robinson equation18 takes only height into consideration, creatinine clearance in African Americans may have been underestimated. Of note, although nursing home populations generally have high female to male ratios, the specific diabetic population on Metformin that we studied had an almost 1:1 female to male ratio. Despite the similar percentages, we still found a significant discrepancy between serum creatinine and creatinine clearance in our sample.

There are number of study limitations. First, we only included those from a single nursing home. Second, lactic acid levels could not be obtained due to the retrospective design of the study, where lactic acid was not previously measured. Future research with a prospective design along with measurement of lactic acid would be useful. Third, since only one value of creatinine was measured, AKI cannot be definitively ruled out in some patients. However, as this study was performed in outpatients who were not on nephrotoxic drugs and whose creatinine levels were sent as routine labs, acute kidney injury is unlikely. Fourth, although our small sample size is potentially a limitation, we nevertheless report high levels of measurement discrepancy.

In conclusion, serum creatinine (SCr) and creatinine clearance (CrCl) have measurement discrepancy for abnormal renal function among nursing home residents. As CrCl is a more accurate measurement of renal function than SCr, we advocate the use of estimated CrCl instead of SCr when prescribing Metformin, with extra caution in debilitated patients. Particularly in a nursing home population, an estimated CrCl <60 mL/min warrants consideration in deciding the use and dosing of Metformin since CrCl estimated by Cockcroft–Gault likely represent underestimations in many nursing home residents. More frequent monitoring for worsening renal function and metabolic acidosis may be indicated. Clinical judgment is paramount in weighing risks and benefits of using Metformin in nursing home residents with renal impairment, especially for those of older age and among women.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.