Diuretics and Bioimpedance‐Measured Fluid Spaces in Hypertensive Patients

Abstract

The authors examined the relationship between thiazide‐type diuretics and fluid spaces in a cohort of hypertensive patients in a retrospective study of 60 stable hypertensive patients without renal abnormalities who underwent whole‐body bioimpedance analysis. Overhydration was greater in the diuretic group, but only to a nonsignificant degree (5.9 vs 2.9%; P=.21). The total body water did not differ in the two groups (41.8 L vs 40.5 L; P=.64). Extracellular fluid volume (ECV) (19.7 L vs 18.5 L; P=.35) and intracellular fluid volume (ICV) spaces (20.8 L vs 21.3 L; P=.75) were also not significantly different in the two groups. The ratio of ICV:ECV, however, appeared different: 1.05 vs 1.15 (P=.017) and the effect was maintained in the linear regression–adjusted model (β coefficient: −0.143; P=.001). The diuretic‐related distortion of ICV:ECV ratio indicates potential fluid redistribution in hypertensive patients, with ICV participating in the process.

Diuretics are a cornerstone of most antihypertension guidelines, including the Joint National Committee on the management of hypertension in adults1 and the European Society of Hypertension and the European Society of Cardiology.1 While both recommendations clearly endorse the use of diuretics, their current recommendations seem somewhat less vocal than previous editions of the same. These newer guidelines take a much more cautionary approach both in treatment goals and in the unequivocal endorsement of one agent over another for the primary treatment of essential hypertension. For example, while the 2013 version of the European Societies' recommendation discusses controversies of thiazide diuretic use in detail in section 5.2.1.2, the 2003 version simply makes the recommendations of grade 2B diuretics only.2 Potential side effects of diuretic therapy include new‐onset diabetes and hyperuricemia,3 polypharmacy in the elderly,4 hypokalemia,5 and even adverse cardiac outcome.6 Clearly, some but not all of the untoward side effects associated with diuretics, such as dehydration and the increased risk of falls in the elderly,7 may be related to the changes of extracellular fluid volume (ECV) induced by diuretics.8, 9 However, classic methods for measuring total‐body water space,10, 11 estimation of the extracellular and intracellular water spaces (with sodium bromide and potassium isotopes, respectively),12 inulin‐based volume measurements,13 and the gold standard measurement of fat tissue content14, 15, 16 are not conducive to daily practice. Bioimpedance is an emerging technology to measure the body's fat content and both ECV and intracellular fluid volume (ICV) spaces17 and may have the potential to optimize diuretic therapy.18, 19 In this current paper, we sought to evaluate the influence of diuretics on the various fluid compartments in a treated hypertensive cohort with predominantly normal kidney function.

Methods

This is a retrospective cohort study of patients visiting a specialty hypertension clinic at the Semmelweis University in Budapest, Hungary, between 2010 and 2013. The study had been approved by the university's ethics committee (institutional review board equivalent). There are a large number of patients at this hypertension clinic who regularly follow up with regard to blood pressure (BP) control. A total of 60 consecutive patients in the hypertension clinic were referred for undergoing bioimpedance fluid space measurements in order to elucidate an underlying fluid excess contributing to difficult‐to‐control hypertension. All patients had been screened for underlying metabolic, hormonal, and renal abnormalities as the etiology of their hypertension, and, by exclusion, were found to have essential hypertension. In this particular cohort, normal renal function was defined as serum creatinine <1.4 mg/dL and no underlying renal pathology by a comprehensive laboratory and imaging evaluation. They were all stable on their chronic medical therapy (no interval antihypertensive medication changes >3 months). BP was measured in accordance with the 2013 European Society of Hypertension/European Society of Cardiology Clinical Practice Guidelines1 in a sitting position at rest by a nurse using an oscillometric automated BP meter (Omron M4‐I; Omron Healthcare B.V., Hoofddrop, The Netherlands). Patients' bioimpedance indices were measured using a regularly calibrated BCM multifrequency bioimpedance apparatus (BCM—Body Composition Monitor; software version 3.2; Fresenius Medical Care, Bad Homburg, Germany) in a manner previously described.17 Briefly, the patients were asked to lay flat on an examination table having measured their height and weight when undressed to underwear. Manufacturer‐provided electrodes were then placed on the patients' distal and metacarpal bones and wrists, and distal and metatarsal bones and ankles, over the dorsal surfaces of these joints. The BCM apparatus returns measured total body water, extracellular water in liters, intracellular water in liters, and lean tissue mass in kg and calculated values such as overhydration (OH) in liters, which is a part of the extracellular volume that is calculated on measurement values. The BCM measurements are based on the principle developed by Hanai20; namely, that certain high electric frequencies conduct through all tissue compartments including the intracellular and extracellular fluid spaces and other frequencies can only travel through the extracellular compartment. The various algorithms built in the software21 of the bioimpedance monitor were based on early fluid compartment measurements using isotopes and other tracers.

Data were tabulated in a computer spreadsheet and dichotomized based on various clinical indices such as sex (male or female), age (>65 or ≤65 years), presence or absence of diabetes as a chronic diagnosis, number of antihypertensive medications (>2 or ≤2 antihypertensive medications), presence or absence of diuretics as part of the therapeutic regimen, self‐reported daily urine volume(>1 or ≤1 L/d), body mass index (BMI) (>30 or ≤30 kg/m²), and systolic BP (>140 mm Hg or ≤140 mm Hg. Data preparations and analysis were performed with Microsoft Excel (Redmond, WA) software and additional statistics were performed with SPSS Statistics version 22 (IBM Corporation, Armonk, NY). Statistical comparisons between variables were calculated with t test for continuous variables and with nonparametric testing (chi‐square) for categorical variables. Linear regression analysis was performed with dependent variables of age, sex, presence of diabetes mellitus, number of antihypertensive medications, BMI, presence of (thiazide‐type) diuretics, percentage of OH of ECV, fat mass (percentage of body weight), and creatinine clearance (according to Cockroft‐Gault22) to examine predictors of ICV:ECV ratio.

Results

The study population included 60 patients, aged 62.1±13.2 years, with 63.3% men and 30% diabetic (Table 1). Serum creatinine for the cohort was 1.08±0.24 mg/dL, with a calculated creatinine clearance according to the Cockcroft‐Gault equation of 82.3±32.5 mL/min. Fourteen (23.3%) of the patients had calculated creatinine clearance <60 mL/min, with a mean of 52.3±5 mL/min. Serum potassium was well preserved at 4.3±0.4 mEq/L. When the population was dichotomized according to sex (female vs male: 22 vs 38), it was clear that systolic BP was higher in men (140.7 mm Hg vs 129.7 mm Hg; P=.029) and that men were much more overhydrated than women (1.56 L, 6.4% vs 0.17 L, 1.11%; P=.02 vs .03, respectively). There was also a greater proportion of fat content in women when expressed as fat percentage (43.2% vs 31.7%, P<.0001), likely contributing to the greater degree of OH in men, with muscle retaining more water than adipose tissue. There were no other differences among the other indicators in this grouping.

Table 1.

Patient Characteristics for the Entire Cohort

	Overall Cohort (N=60)	No Diuretics (n=29)	Diuretics (n=31)	P Value
Age, y	62.1 (13.2)	57.2 (12.1)	66.7 (12.6)	.004
Male sex, %	63.3	66.5	61.3	(χ2) .734
Diabetes mellitus, %	30	27.6	32.3	(χ2) .693
Urine volume per d, mL	1378 (676)	1301.7 (436)	1450 (842)	.4
Antihypertensive medications, No.	2.3 (1.3)	1.3 (1)	3.2 (0.9)	<.0001
Systolic blood pressure, mm Hg	136.7 (19)	135.6 (15.7)	137.7 (21.7)	.663
Diastolic blood pressure, mm Hg	79 (13.9)	81.5 (12.8)	76.7 (14.7)	.181
Body weight, kg	84.7 (19.8)	90.5 (19.6)	79.2 (18.6)	.26
Body mass index, kg/m2	30.2 (6.2)	28.9 (6.7)	31.4 (5.3)	.117
OH, L	1 (2.2)	0.7 (1.4)	1.35 (2.7)	.26
OH % of ECV	4.5 (9.3)	2.9 (6.4)	5.9 (11.3)	.217
Fat mass, kg	30.9 (12.6)	30 (15.1)	31.7 (10.7)	.598
Fat mass, % of body weight	35.9 (9.8)	35.4 (11.1)	36.3 (8.6)	.703
ECV, L	19.1 (4.8)	18.5 (4.8)	19.7 (4.8)	.352
ICV, L	21 (5.8)	21.3 (5.4)	20.8 (6.1)	.748
ICV:ECV ratio	1.1 (0.2)	1.15 (0.13)	1.06 (0.16)	.017
Creatinine clearance, Cockroft‐Gault; mL/min	82.3 (32.5)	89.9 (39.4)	75.6 (23.7)	.096
Potassium, mEq/L	4.3 (0.4)	4.3 (0.4)	4.4 (0.4)	.352
Sodium, mEq/L	139.7 (3)	139.7 (3.4)	139.7 (2.7)	.967

Abbreviations: ECV, extracellular fluid volume; ICV, intracellular fluid volume; OH%, overhydration relative to the extracellular fluid volume; OH, overhydration as measured by bioimpedance. Values are expressed as means (±standard deviations). P values by Student t test between the “no diuretics” and the “diuretics” groups.

Ranking patients by age revealed that those younger than 65 years (n=36) were taking fewer antihypertensive medications (1.9 vs 2.8; P=.01), and the use of diuretics was much more common among older patients than younger patients (75% vs 36.1%; P=.002). Of the patients taking diuretics, 100% were taking a thiazide‐type diuretic (hydrochlorothiazide or indapamide) while one patient took both hydrochlorothiazide and spironolactone. There was no statistically significant association between diuretic use and reduced (<60 mL/min) creatinine clearance (P=.35), overall creatinine clearance (P=.096), or serum potassium (P=.347). Younger patients' diastolic BP was also significantly higher (82.5 mm Hg vs 73.8 mm Hg; P=.02) but no other difference could rise to the level of statistical significance in bioimpedance‐measured indicators of BMI or the presence of diabetes. As expected, patients with diabetes were more likely to have a greater BMI (34.0 kg/m² vs 28.6 kg/m²; P=.001) and a larger fat content (fat percentage, 40.0% vs 34.1% P=.03); however, the BPs were not statistically different. No other parameter was significantly different in this grouping.

While BMI >30 kg/m² is associated with a greater fat content (39.4% vs 33.3%; P=.01), it is also associated with a greater number of liters of ECV fluid (22.3 L vs 16.8 L; P<.0001) and intracellular (ICV) fluid (23.5 L vs 19.2 L; P=.004). While the number of liters of OH is greater in obese patients (1.7 L vs 0.58 L; P=.05), this degree of OH relative to the extracellular fluid (6.7% vs 21.9%; P=.11) is not significant. Obese patients take diuretics significantly more often (68% vs 40%; P=.03) than nonobese patients and are more likely to be diabetic (44% vs 20%; P=.04). BMI does not seem to relate to BP, as this dichotomy revealed a minor difference in systolic and diastolic BPs (138/80 mm Hg vs 135/78 mm Hg; P=.51).

The use of diuretics (n=31) had no correlation with achieved BP (diuretics in treated vs untreated patients: 137/76 [±21.7/14.7] mm Hg vs 135/81 [±15.7/12.8] mm Hg, repectively; P=.66) (Table 1). Diuretic therapy was strongly associated with a more complex medical regimen (total number of antihypertensive medications: 3.2±0.9 with diuretics vs 1.3±1 without; P<.0001) and advancing age (66.7±12.1 years vs 57.2±12.6 years; P=.004). OH was increased to a nonsignificant degree in the group that used diuretics (5.9±11.3% vs 2.9±6.4%; P=.21). In keeping with the presumed steady‐state of the study's patients, diuretics had no association with self‐reported daily urine output (1450 mL/d vs 1301 mL/d; P=.40). Total body water did not statistically differ in the two groups (41.8 L vs 40.5 L; P=.64) in the values of either ICV or ECV (Figure 1). The ratio of ICV:ECV, however, seems to have been significantly affected, with a ratio of 1.15±0.13 among those not taking diuretics but 1.05±0.16 among those taking diuretics (P=.017) (Figure 2). Q‐Q plots of ICV:ECV distribution were in keeping with normal distribution for the entire cohort as well as for the subcohorts (treated vs untreated with diuretics). Linear regression was performed with stepwise selection, with multiple dependent variables included (Table 2). The final model (step 5) selected OH percentage, fat content (percentage of body weight), presence of thiazide diuretics, BMI, and age as independent variables statistically associated, with ICV:ECV ratio as an independent outcome. The model's overall R ² improved significantly from step 1 (0.557) to step 5 (0.921); thus, the final model showed approximately 92% variation of ICV:ECV ratio.

Figure 1.

Fluid spaces in diuretic‐untreated (n=29; D−) and diuretic‐treated (n=31; D+) groups. ECV indicates extracellular volume; ICV, intracellular volume; NS, not statistically significant (by Student t test). Column heights indicate means and bars indicate standard deviations.

Figure 2.

Intracellular volume/extracellular volume (ICV/ECV) ratios in diuretic‐untreated (n=29; D−) and diuretic‐treated (n=31; D+) groups. P indicates significance (by Student t test). Column heights indicate means and bars indicate standard deviations.

Table 2.

Linear Regression Modeling of Predictors, ICV:ECV Ratio (R ² 0.921)

Variables	Standardized Beta Coefficient	t	Significance
Constant		33.623	<.0001
OH % of ECV	−0.799	−19.198	<.0001
Fat mass, % of body weight	−0.623	−13.863	<.0001
Diuretics, thiazide‐type (yes/no)	−0.143	−3.356	.001
Body mass index, kg/m2	0.135	2.953	.005
Age	−0.090	−2.052	.045

Abbreviations: ECV, extracellular fluid volume; ICV, intracellular fluid volume; OH%, overhydration relative to the extracellular fluid volume; OH, overhydration as measured by bioimpedance. Result of linear regression modeling, with ICV:ECV as an independent variable in linear regression modeling with stepwise selection. Overall adjusted R ² for the final model: 0.914. Excluded variables from the model (step 5) are sex, presence of diabetes mellitus, number of antihypertensive medications, and creatinine clearance (Cockroft‐Gault formula).

Discussion

In a dialysis population, we recently showed23 that a single session ultrafiltration during a dialysis treatment reduced the excessively enlarged ECV space without significantly changing ICV. This differential effect of ultrafiltration influenced the ICV:ECV ratio by normalizing it toward the 55%:45% (1.22) of total body water distribution ratio from 1.00 to 1.11, with a relative drop of the ECV by 11.6% and ICV by 1.9%, respectively.23 This dialysis cohort, however, had an intervention—ultrafiltration—with the deliberate goal to reduce accumulated fluid excess during renal dialysis. On the other hand, the present study of hypertensive patients with predominantly normal renal function had no particular intervention to change the various fluid spaces, and we can only speculate on a potential “diuretic effect” on the volume of fluid spaces. In this study we found no significant association between diuretic use and fluid compartments. Neither the ECV nor the ICV are significantly different among persons taking and not taking a diuretic (Figure 1). There seemed to be, however, a significant difference in ICV:ECV ratios (Figure 2). This ratio in the diuretics group of 1.05 is further from the standard ratio of 1.22 (45%:55%). This ratio distortion is the result of a 6.3% higher ECV in the diuretic group and a 2.3% lower ICV, albeit not to a statistically significant degree by themselves. Thus, patients taking diuretics have a decreased ICV (20.8 L) compared with those not taking diuretics (21.3 L; P=.78) despite a (not significantly) greater BMI (31.4 vs 28.9; P=.11) and a higher ECV (19.7 L vs 18.5 L; P=.35). This raises the question of whether diuretics really deplete the exchangeable extracellular sodium ions only or perhaps have an influence on the intracellular, osmotically active electrolytes such as potassium as well. The two (fluid)‐compartment model describes a homeostatic state where Na⁺ is restricted to the extracellular fluid space and K⁺ to the intracellular space. Each electrolyte retains water in the respective fluid space through its osmotic activity.24 According to this model, “body water appears to be passively distributed in proportion to osmotic activity, and all or almost all of body potassium is osmotically active.”25 This model ascribes the extracellular water and sodium regulation to the kidneys. Additionally, there are mechanisms through which there may be a third osmotically inaccessible pool of Na⁺ stored and an omotically inactive Na⁺/K⁺ exchange mechanism.25, 26 Our study, however, which is not a sodium study in the classical sense, but a fluid compartment study, seems to raise the possibility to challenge the notion that these compartments would be constant or inaccessible to renal function and fluid loss by diuresis.

Study Limitations

Our study also has several limitations, including the single‐center design, relatively small number of participants, and the purely observational nature of data; thus, causation cannot be ascertained, only association. The effect of aging may have confounded the association of decreasing ICV/ECV ratio observed with diuretic use. We can only speculate that the fluid space alterations in the presence of diuretics are occurring as a result of their pharmacologic activity rather than a confounder. Interestingly, a study by Cianci27 has also observed that hypertensive patients have a different fluid compartment distribution than controls, with an elevated ICV. Thus, it is possible that this association of fluid compartment differences simply confirm that hypertensive patients have a different fluid distribution and our observation is not a diuretic effect per se.

Conclusions

The diuretic‐related distortion of ICV:ECV ratio indicates potential fluid redistribution in hypertensive patients, with the intracellular fluid space participating in the process. It is unclear whether this minor but statistically significant change in ICV:ECV ratio is clinically significant or even beneficial. Prospective, interventional studies could answer the question of whether this decrease of ICV:ECV ratio as a result of diuretic use is caused by total body sodium loss, perhaps in the osmotically inactive Na⁺ reserve, or another mechanism.

J Clin Hypertens (Greenwich). 2014;16:895–899. DOI: 10.1111/jch.12428. © 2014 Wiley Periodicals, Inc.