Hydration Status Assessment in Older Patients

Abstract

Background

Hydration disturbances are common in old age: the reported prevalence of dehydration in elderly patients ranges from 19% to 89%, depending on the definition and the population in question. However, the clinical assessment of patients’ hydration status is difficult. In this review, we discuss the diagnostic value of currently used methods that may or may not be suitable for assessing older patients’ hydration status.

Methods

We conducted a selective literature search for relevant studies concerning patients aged 65 and above. Of the 355 articles retrieved by the initial search, a multistep selection process yielded 30 that were suitable for inclusion in this review.

Results

107 different methods for the diagnostic assessment of dehydration in older persons were evaluated on the basis of the reviewed publications. High diagnostic value, especially for the determination of hyperosmolar dehydration, was found for serum osmolality, serum sodium concentration, inferior vena cava ultrasonography, a history (from the patient or another informant) of not drinking between meals, and axillary dryness. On the other hand, a variety of clinical signs such as a positive skin turgor test, sunken eyes, dry mouth, tachycardia, orthostatic dysregulation, and dark urine were found to be of inadequate diagnostic value.

Conclusion

Only five of the 107 methods considered appear to be suitable for determining that a patient is dehydrated. Thus, the available scientific evidence indicates that all clinicians should critically reconsider their own techniques for assessing hydration status in elderly patients. To optimize the clinical assessment of patients’ hydration status, there seems to be a need for the rejection of unsuitable methods in favor of either newly developed criteria or of a combination of the best criteria already in use.

The terms dehydration (hypohydration), euhydration and hyperhydration describe a negative, normal or positive water balance, respectively (e1). For all three states, no standard definitions are in place (e1–e3). Dehydration, which is a common clinical condition, can be further subdivided based on sodium concentration and/or serum osmolality. Hypertonic dehydration is caused by water loss, occasionally related to fever or sweating but in most cases due to insufficient intake of pure water. Hypotonic dehydration (possibly with hyponatremia) is frequently caused by the use of diuretics, heavy sweating with replacement of the fluid loss with low-sodium drinks (for example, drinking water), or losses from the gastrointestinal tract when electrolyte loss exceeds water loss. Isotonic dehydration is frequently secondary to the loss of isotonic fluids such as blood, but may also be associated with vomiting or diarrhea (e2).

The rather vague term volume status is commonly used instead of the broad term hydration, which describes the water content in the entire body. Especially in the clinical context, volume status describes the intravascular volume. The shift of water and electrolytes is usually in the same direction (hypovolemia/euvolemia/hypervolemia).

Thus, in the following, we use the term hydration status (HS) that includes the intravascular volume status.

Assessment of the hydration status in older patients

Little has changed in recent decades in terms of the challenges associated with assessing the HS of older patients (1, 2). Isotope dilution analysis is considered the gold standard for assessing HS. However, it is not a feasible method in clinical practice, given the significant amount of time and resources required.

Older persons are at an increased risk of developing hydration disorders (e4). This is primarily due to reduced renal function, resulting from decreased renin and aldosterone activity and age-related decline in thirst in people of older age (e2). In addition, fluid intake may be hampered by functional and cognitive impairments. This may be compounded by the effects of medications, for example diuretics or laxatives (e2).

The prevalence of dehydration in older patients varies widely, depending on the definition used and the population studied. In the context of admission examinations of older patients, 19% of patients were diagnosed with dehydration, based on the blood urea nitrogen to creatinine (BUN/Cr) ratio (3). In nursing home residents, hypertonic dehydration was determined based on elevated serum osmolality in 89% of the residents (4). During the 8-year follow-up, patients with increased plasma osmolality were found to be at an increased relative mortality risk of 1.4 (95% confidence interval: [1.0; 1.9]) compared to patients with normal plasma osmolality, as well as an increased relative risk of 2.3 [1.2; 4.3] for impairments of instrumental activities of daily living (IADL) (e4).

Because of this, and also because of known complications, such as decreased vigilance, falls, urinary tract infections, respiratory tract infections, and renal impairment, a quick and easy assessment of the hydration status of older patients appears to be of critical importance.

The aim of this study is to provide an overview of the diagnostic accuracy of current methods and to propose a procedure that is suitable for use in clinical practice.

Methods

This review is based on pertinent publications retrieved from a selective literature search. The search terms were selected based on the PICO framework and fall into the following three main categories (eFigure 1):

eFigure 1.

Literature search according to the PICO framework

PICO, patient, intervention, comparison, outcome

• Hydration status

• Patient group

• Diagnostic methods.

All results published in PubMed up to and including September 2021 with full-text articles from medical journals and with a study population aged 65 years or older were included in this analysis.

The resulting 355 articles were subjected to a multi-step selection process with predefined inclusion criteria (eTable). This left 24 articles, from whose references a further six articles were selected, using the same procedure (so-called reverse search). This resulted in a total of 30 full-text articles from medical journals being included in the assessment. These comprised 29 individual studies and one meta-analysis.

eTable. Overview of all studies included in our analysis.

Reference method for determining HS	Publications/study design	Study setting and study population	N	Methods assessed	Key messages
Hydration status: Hyperosmolal dehydration
Directly measured serum osmolality, calculated serum osmolarity, or rapid weight loss over a period of 7 days	Hooper et al. (2015) (5) meta-analysis	Hospital patients, residents of nursing facilities, persons living independently → Age: ≥ 65 years	3412	67 different clinical signs/symptoms, hemodynamic parameters, lab tests, BIA; among others: skin turgor; dryness of mouth, mucous membranes, axilla; systolic blood pressure, Schellong test, urine color, U-stix, BIA resistance, and more.	3 of the 67 methods studied were assessed as suitable: “Missed drinks between meals“, expressing fatigue, BIA resistance at 50 Hz; potentially suitable: axillary dryness, among others.
Directly measured serum osmolality	Bunn & Hooper (2019) (6) Cross-sectional study	Long-term residents of nursing facilities: nursing homes, assisted living, special facilities for dementia patients → Age: ≥ 65 years → DRIE cohort	188	49 different clinical methods, incl. skin turgor, dry mouth, axillary moisture, capillary refill time, peripheral vein filling, among others.	There is a complete lack of diagnostic quality for determining dehydration.
	Siervo et al. (2014) (7) Cross-sectional study		186	36 equations for calculating serum osmolality	One formula (Khajuria & Krahn [2005] [31] Clin Biochem) was found to have utility in a geriatric population.
	Hooper et al. (2015) (8) Cross-sectional study	Hospitalized and non-hospitalized persons 5 different cohort studies → Age: ≥ 65 years → DRIE, NU-AGE and potentially Fortes, Sjöstrand, and Pfortmüller cohorts	595		Recommendation to use directly measured serum osmolality or a formula adjusted to the population.
	Hooper et al. (2016) (9) Cross-sectional study			13 different urine parameters, including urine color, urine specific gravity, volume, urinalysis, urine osmolality	Urine color, urine specific gravity and osmolality show only low diagnostic quality for identifying hypertonic dehydration.
Calculated serum osmolarity	Kinoshita et al. (2013) (10) Cross-sectional study	Patients of an acute care hospital → age ≥ 65 years	29	Four clinical signs of dehydration: reduced skin turgor, sunken eyes, axillary dryness, device-based measurement of skin moisture in axilla (stated in %)	Axillary moisture levels > 50% can virtually rule out dehydration.. Axillary moisture of <30% is suggestive of hypertonic dehydr. (spec. = 0.91; sens. = 0.12).
	Shimizu et al. (2012) (11) Cross-sectional study		27	Six clinical signs of dehydration: impaired consciousness, dryness of axilla, mucous membranes and mouth, sunken eyes, skin turgor, capillary refill time, laboratory parameters	Incident dry axilla (Sens. = 0.44; Spec. = 0.89) shows high specificity in detecting hyperosmolal dehydration.
	Wojzel (2020) (12) Cross-sectional study		358	Serum parameters: sodium, potassium, urea, glucose	Best results (ROC analysis) for sodium with a cut-off value of 140 mmol/L sensitivity 0.80; specificity 0.83.
	Sanson et al. (2021) (2) Retrospective cohort study		4613	Serum and urine parameters: Serum sodium, potassium, urea, creatinine; BUN/Cr, hematocrit, urine specific gravity	Only serum sodium and urea showed a significant progressive increase with an increase in serum osmolarity.
Hydration status: hyperosmolal and iso-osmolal dehydration
Directly measured serum osmolality and BUN/Cr	Fortes et al. (2015) (3) Cross-sectional study	Patients of an acute care hospital → Age: ≥ 60 years	178	7 clinical and hemodynamic signs of dehydration: tachycardia, systolic blood pressure<100 mm hg, dryness of mucous membranes and axilla, skin turgor, capillary refill time, sunken eyes	All clinical signs were found to have low sensitivity (0.0–0.44). Salivary osmolality had a sensitivity of 0.7 and a specificity of 0.68 to diagnose both types of dehydration.
	Eaton et al. (1994) (13) Cross-sectional study	Geriatric acute patients → Age: ≥ 70 years	100	Axillary moisture: manually palpated sweat excretion and sweat excretion measured by placing tissue paper in the axilla	Palpation of a moist axilla was found to have a high negative predictive value (84%). Potential utility as an exclusion criterion.
Hydration status: dehydration
BUN/Cr	Orso et al. (2016) (14) Cross-sectional study	Patients in emergency department → Age: ≥70 years	270	Ultrasonography of inferior vena cava: absolute expiratory diameter, caval index, shock index, among others	Significantly lower values of expiratory diameters in dehydrated patients (1.3±0.5 cm vs. 1.7±0.4 cm; p<0.001); higher caval index in dehydrated patients (75 vs. 40%; p = 0.01); caval index with cut-off values of 48%:: sens.= 0.993, spec.= 1.0 for diagnosing dehydration; cut-off value of 1.55 cm for the expiratory diameter: sens.= 0.855, spec.= 1.00.
Clinical synopsis based on systolic blood pressure, serum sodium, BUN/Cr, serum osmolality	Gross et al. (1992) (15) Cross-sectional study	Acute patients in an emergency department → Age: ≥ 60 years	55	38 clinical signs: vigilance, sunken eyes, skin turgor, mucosal dryness, axillary moisture, and more.	Some methods showed a significant correlation with patient age (e.g., skin turgor) but not with hydration status.
Clinical synopsis: Medical history, physical examination, fluid intake, weight changes	Vivanti et al. (2008) (17) Cross-sectional study	Inpatients in a geriatric rehabilitation facility → Age: ≥ 60 years	43	Hematological and urine parameter, hemodynamic measurements, clinical signs: oral mucosal dryness, skin turgor.	No statistically significant differences in urine parameters, hematological and serological parameters between patients classified as dehydrated or euhydrated.
	Vivanti et al. (2013) (16) Cross-sectional study		10	Hourly weight fluctuations within 9 h for three consecutive days.	Weight fluctuations between 0.3% and 2.8% observed in euhydrated patients. Weight fluctuations < 3% should not be used as an indicator of dehydration.
Serum uric acid	Yoshihara et al. (2007) (18) Cross-sectional study	Outpatients of a national screening program → Age: 76 years	403	Salivary samples: salivary flow rate, salivary spinnability	A statistically significant difference in salivary spinnability between subjects with uric acid concentrations ≥/< 7 mg/dl was found.
None	Johnson & Hahn (2018) (4) Cross-sectional study	Nursing home residents → Age: 64–103	60	Correlation between hematological parameters, urine parameters (for example: FRI = urine specific gravity, urine osmolality, urine creatinine and urine color), clinical signs, hemodynamic parameters, sense of thirst, medical history	No statistically significant correlation between FRI and measured serum osmolality or serum sodium; patients with increased FRI reported a less pronounced sense of thirst; clinical examinations did not correlate with other markers of dehydration.
	Ekman, L. et al. (2020) (219) Cross-sectional study	Patients of a geriatric rehabilitation facility with status post hip fracture → Age: ≥ 65 years	38		Low fluid intake correlated strongly with increased urine specific gravity. Patients with low fluid intake reported a less pronounced sense of thirst. Patients with a fluid intake of < 800 ml/d were found to have more concentrated urine and had lost weight in the previous month. urine specific gravity proposed as a potential marker of low-intake dehydration.
Hydration status: global HS
TBW, ECW using isotope dilution methods	Olde Rikkert et al. (1997) (1) Prospective observational study	Geriatric hospital inpatients → Age: ≥ 70 years	53	MF-BIA, weight changes	Utility of individual BIA measurements for diagnostic purposes questionable; serial measurements useful for evaluation of treatment response.
	Olde Rikkert et al. (1998) (20) Prospective observational study			Hematological parameters: serum sodium, potassium, urea, creatinine, hematocrit; BUN/Cr; observation for 16 months	In euvolemic patients, the median of all serially measured laboratory parameters was within the population-based normal ranges. One-time testing of the parameters showed high interindividual variability; intraindividual variability much less pronounced.
	Powers et al. (2009) (21)Cohort study	Patients of an acute care hospital → Age ≥ 65 years	32	TBW estimation using BIA, anthropometric measurements	BIA measurements did not differ significantly when compared to the isotope dilution analysis.
	Ritz (2001) (22) Cohort study		169		TBW using BIA (50 kHz): Difference of 0.48 ± 2.3 L compared to isotope dilution analysis;BIA measurement are useful for monitoring patients with shifts in fluid and volume status.
	Hoyle et al. (2011) (23)Cohort study	Geriatric hospital inpatients with hyponatremia → Age: ≥ 65 years	22	BIA measurements: TBW, ECW Clinical examination: mucosal dryness, axillary dryness, skin turgor, peripheral edema, among others	Clinical assessment showed only moderate inter-individual agreement (κ = 0.59; p<0,001).bia-tbw showed a clear linear relationship with tbw measurements by isotope dilution (r = 0.69; p<0.001).
Clinical assessment of HS based on weight, HR, sunken eyes, mucosal dryness, skin turgor, peripheral venous filling, among others	Rosher & Robinson (2004) (24) Uncontrolled, non-randomized interventional study	Nursing home residents → Age: ≥ 70 years	51	Intervention: 9-week hydration program with minimum intake of 470 mL in addition to previous oral fluid intake. Measurements: BIA, physical examination	Stat. significant increase in TBW, measured using BIA at the end of the program; improvement in filling of veins on the dorsum of the foot; no effect on other clinical signs
	Rösler et al. (2009) (25) Cohort study	Geriatric hospital inpatients → Age: ≥ 65 years	103	BIA: resistance, reactance, vector analysis, anthropometric measurements.	Agreement of expert assessment with BIA measurements in 40/60 cases in euvolemic patients, but no agreement with regard to diagnosis of dehydration or hyperhydration (0/43)
	Cumming et al. (2014) (26) Cohort study	Hyponatremic inpatients with fragility fractures→ Age: ≥ 70 years	125	BIA measurements & obtaining clinical signs, e.g. skin turgor, dry mouth, among others	BIA measurements have the potential to improve the assessment of volume status. In 23 of 33 cases with hyponatremia, the expert assessment of HS was consistent with the classification based on BIA measurements.
None	Wakefield et al. (2002) (27) Cohort study	Nursing home residents → Age: ≥ 65 years	89	Correlations between various parameters, including urine color, osmolality, specific gravity, and hematological parameters	Significant correlations between urine color and urine specific gravity (r = 0.57; p<0.001) and urine osmolality (r = 0.57; p<0.001);sign. correlations between urine osmolality and serum osmolality (r = 0.37; p<0.01) and bun/cr ratio (r = 0.39; p<0.01); no correlation between urine color and hematological & serological parameters
	Mentes et al. (2006) (28) Cohort study	Geriatric hospital inpatients → Age: ≥ 65 years	92	Correlation between urine specific gravity and urine color.	Correlations between urine color and urine specific gravity range from r = 0.3 to r = 0.7, depending on patient subgroup.
	Powers et al. (2012) (29) Cohort study	Hospital outpatients and inpatients → Age: ≥ 65 years	63	Correlation between various BIA (TBW, ECW) and urine parameters	No statistically significant correlations
Hydration status: Hyperhydration
Clinical synopsis based on peripheral edema, pulmonary congestion, jugular venous congestion etc.	Coodley et al. (1995) (30) prospective cohort study	Patients with congestive heart failure and IV diuretic therapy → Age: ≥65 years	12	BIA resistance, reactance, TBW	Single measurements using BIA seem to have limited usefulness for monitoring diuretic therapy; consecutive measurements for intraindividual assessment may be considered.

BIA, bioelectrical impedance analysis; BUN/Cr: blood urea nitrogen to creatinine ratio; ECW, extracellular water; FRI, Fluid Retention Index; HS, hydration status; IV, intravenous; MF-BIA, multi-frequency bioelectrical impedance analysis; pts, patients; CRT, capillary refill time; ROC analysis, receiver operating characteristic analysis; Sens., sensitivity; Spec., specificity; TBW, total body water

A detailed description of the methodological approach can be found in the eMethods section, supplemented by eBox, eFigure 1, eFigure 2, and eTable.

eMethods. The methodological approach for the selective search of the literature.

A selective search of the literature was performed, for which the search vocabulary was defined according to the PICO framework and divided into three main categories: hydration status, patient group and diagnostic methods. Keywords were selected using available MeSH terms and the PubMed keyword database. All keywords were linked to form a search string, using a Venn diagram (eFigure 1). All results published in PubMed up to and including September 2021 with full-text articles from medical journals were included in the analysis.

Using the search string, the database search yielded 355 articles that were subjected to a multistep selection process (eFigure 2), employing inclusion criteria previously defined by the author group (eBox). According to the criteria of the German Society for Geriatrics (DGG) (e5), we searched for data of patients aged 65 years and older, or, in the absence of age information, for existing need for nursing care, treatment in a geriatric facility, frailty or multimorbidity.

This left 24 articles, from whose references a further six articles were selected by reverse search, using the same procedure. This involved reviewing the references of the studies already included for suitability, using the selection process described above. In this step, six additional studies were included. A total of 30 full-text articles from medical journals were included in the analysis—29 individual studies and one meta-analysis.

First, we arranged the studies retrieved in a summary table (eTable) and then discussed the utility and clinical applicability of the described diagnostic procedures with all co-authors in two consecutive conferences. This positive selection was presented in smaller tables (Tables 1–4).

eBox. Inclusion criteria of the screening process.

• Language

  • English

  • German

• Methods

  • Primary literature and meta-analyses

  • Studies on diagnostic quality

  • Studies comparing methods for hydration status assessment with each other or with a reference method

• Study population

  • Patients aged 65 years or older

  • Groups of persons who, because of their frailty, represent a geriatric population (e.g. nursing facility residents, multimorbidity, frailty, nursing care needs)

eFigure 2.

Literature selection process

Results

Seventeen individual studies and one meta-analysis addressed the topic of dehydration (2–19); in 12 cases, the directly measured serum osmolality, the calculated serum osmolarity and/or the blood urea nitrogen to creatinine (BUN/Cr) ratio were used as the standard method for diagnosing dehydration (2, 3, 5–14). In the remaining six studies, the diagnosis dehydration was based on the clinical synopsis, various isotope dilution methods or other methods.

Eleven of the reviewed publications investigated the patients’ global hydration status (1, 20–29). In five of these studies, an isotope dilution technique was used as the standard method for assessing the hydration status (1, 20–23). Three working groups assessed the hydration status using the clinical synopsis, based on preexisting conditions, systolic blood pressure, general condition, and fluid intake, among other factors (24–26). The remaining studies used laboratory chemistry methods or compared various analyses with each other.

Only one study addressed the diagnosis of hyperhydration in geriatric patients (30). This was a prospective cohort study, reviewing the utility of bioelectrical impedance analysis (BIA) for monitoring diuretic therapy.

Differences in methods, definitions and levels of evidence make it challenging to compare results. The studies included cover 107 different clinical signs as well as various examination methods, including laboratory parameters, urine tests as well as diagnostic devices.

A large number of clinical signs did not show sufficient diagnostic accuracy, especially with regard to the detection of hypertonic dehydration (3, 5, 6, 9, 11, 15–18). These included, among others:

• Reduced skin turgor

• Dryness of the mouth and mucous membranes

• Sunken eyes

• Hemodynamic parameters (for example, Schellong test)

• Urine tests (for example, color of urine).

None of these tests, which are commonly used in clinical practice, achieved a sensitivity of >0.60 and a specificity of >0.75 (5, 6).

We outline below methods that, based on the studies included in this review, may be useful for assessing the hydration status in geriatric patients. In the Tables 1–4, the included publications, key statistical figures as well as the reference methods applied in the individual studies are presented.

Table 1. Recommended laboratory methods.

Publication	Reference standard	Statistical key figures	Evaluation	Comments/ notes
Method assessed: calculated serum osmolality
Siervo et al. (2014) (7)	Classification by directly measured S-Osmolality 1. No dehydr. 2. Impending dehydr. (≥ 295 mOsm/kg) 3. Manifest dehydr. (≥ 300 mOsm/kg)	Cut-off value: 295–300 mOsml/L Sens. = 0.79; Spec.= 0.89 PLR: 7.53, NLR = 0.23; OR = 32.4 Cut-off value: ≥ 300 mOsml/L Sens. = 0.64; Spec. = 0.93 PLR*1: 8.85, NLR*1 = 0.39; OR = 22.6	If direct measurement of S-Osmolality is not available → use age-validated formula for calculation, cut-off value: 295 mOsm/L	Age-validated formula*2 according to Khajuria & Krahn (2005) (31)
Hooper et al. (2015) (8)		Cut-off value: ≥296 mOsml/LSens. = 0,80; Spec. = 0,66 PLR* 1 = 2,36; NLR* 1 = 0,30
Method assessed: Serum sodium
Wojszel (2020) (12)	Calculated S-Osmolarity ≥ 295 mOsml/L	Serum sodium: Cut-off value: 140 mmol/L Sens.= 0.80 (95% CI [0.74; 0.86]) Spec.= 0.83 (95% CI [0.75; 0.88]) PLR*1 = 4.58; NLR*1 = 0.24	Serum sodium may be used to detect hypertonic dehydr. . Cut-off value: 140 mmol/L
Sanson et al. (2021) (2)		Serum sodium: Correlation with calculated serum osmolality: r = 0.634; p<0.001

*¹ positive or negative likelihood ratio describes the change in the probability of disease when positive or negative test results occur; Example: PLR>1 → Probability of disease increases

*² While the age-validated formula does not include patient age in the calculation, it shows the highest diagnostic quality for a geriatric population:

S-Osmolarity [mOsm/L] = 1.86 x (Na+ + K+) + 1.15 x glucose + urea + 14 (all measurements in [mmol/L])

Dehydr., dehydration; NLR, negative likelihood ratio; OR, odds ratio; PLR, positive likelihood ratio; S-Osmol, serum osmolality; Sens., sensitivity;

Spec., specificity; 95% CI, 95% confidence interval

Table 4. Potential diagnostic utility of technology-based methods.

Publication		Reference standard	Key statistical figures	Evaluation	Comments/ notes
Method assessed: Ultrasonography of the inferior vena cava
Orso et al. (2016) (14)		BUN/Cr ratio ≥ 20	Exsp. IVC diameter: Cut-off value: 1.55 cm Sens. = 0.86 (95% CI [0.79; 0.90]) Spec. = 1.00 (95% CI [0.97; 1.00]) PPV = 1.00; NPV= 0.83 Caval index: Cut-off value: 0.48 Sens. = 0.99 (95% CI [0.96; 0.99]) Spec. = 1.00 (95% CI [0.97; 1,00]) PPV = 1.00; NPV = 0.99	Exceptional diagnostic quality of ultrasound → Re-evaluation advised, using populations with different definitions of dehydration	Caval index = [(maximum diameter – minimum diameter)/maximum diameter]
Method assessed: BIA resistance at 50 kHz (cut-off: 550 kOhm)
Hooper et al. (2015) (5)	Allison et al. (2005)* Kafri et al. (2013)* Powers et al. (2012)* Stookey et al. (2005)*	S-Osmol ≥ 295 mOsmol/kg	Sens. = 0.18–0.80 Spez. = 0.61–1.00 PPW = 0.03–1.00 NPW = 0.26–0.96	Wide dispersion of sensitivity and specificity data for resistance upon re-analysis of the data in this meta-analysis	Data from small individual studies → BIA better for intra-individual monitoring than for interindividual comparison
		S-Osmol. ≥ 300 mOsmol/kg	Sens. = 0.17–1.00 Spec. = 0.61–0.91 PPV = 0.00–0.75 NPV = 0.56–1.00

* The studies mentioned are included in the meta-analysis by Hooper et al. (2015) (5). The raw data of the mentioned working groups were obtained and re-analyzed according to the research question of Hooper et al. The data analyzed are based on both previously published studies and unpublished data from various working groups. Due to this re-analysis, some of the results of the individual studies differ from those of the meta-analysis.

BIA, bioelectrical impedance analysis; BUN/Cr, blood urea nitrogen to creatinine ratio; exsp., exspiratory; IVC, inferior vena cava; NPV, negative predictive value; PPV, positive predictive value; S-Osmol, serum osmolality; Sens., sensitivity; Spec., specificity

Laboratory parameters

Serum osmolality and serum sodium

While several studies used directly measured serum osmolality as the standard method to diagnose hyperosmolal dehydration or dehydration due to lack of water, it was shown that the diagnostic quality of calculated serum osmolarity was also sufficient to detect hypertonic dehydration (7, 8). Preferably, an age-validated equation by Khajuria and Krahn (2005) should be used for the calculation (31).

Furthermore, Wojzel proposed a serum sodium cut-off level of 140 mmol/L to diagnose hypertonic dehydration (defined by a calculated serum osmolarity ≥ 295 mOsmol/L) (12). Table 1 provides an overview of key statistical figures.

Medial history clues and clinical examination

Missed drinks between meals and expressing fatigue

A meta-analysis published in 2015 evaluated the use of clinical and low-tech methods for their utility to diagnose hypertonic dehydration (defined by increased calculated serum osmolarity or directly measured serum osmolality). It was found that both a history of missing some drinks between meals (stated by the patient or a third party) and expressing fatigue have high sensitivity and specificity (Table 2) (5).

Table 2. Potential diagnostic utility of (third-party) medical history clues.

Publication		Reference standard	Statistical key figures	Evaluation	Comments/ notes
Method assessed: Missed drinks between meals
Hooper et al. (2015) (5)	Kajii et al. (2006)*	Impending hypertonic dehydr.: S-Osmol ≥ 295 mOsmol/kg	Sens. = 1.00 (95% CI [0.59;1.00]) Spec. = 0.77 (95% CI [0.64; 0.86]) PPV = 0.32; NPV = 1.00	In our opinion, potential benefit; further evaluation needed; difficult to obtain information	So far, data obtained only by one working group
		Overt hypertonic dehydr.: S-Osmol. ≥ 300 mOsmol/kg	Sens. = 1,00 (95% CI [0.16;1.00]] Spec. = 0.71 (95% CI [0.59; 0.81]) PPV = 0.09; NPV = 1.00
Method assessed: Self-reported fatigue
Hooper et al. (2015) (5)	Kajii et al. (2006)* Sjöstrand, (2013)* Sjöstrand Healthy (2013)*	S-Osmol ≥ 295 mOsmol/kg	Sens. = 0.30–0.71 Spec. = 0.75–1.00 PPV = 0.24–1.00 NPV = 0.21–0.96	Wide dispersion of key statistical figures; “some potential” in meta-analysis; in our opinion not suitable for frail old patients (Fried frailty criteria, including weakness and self-reported exhaustion)

* The studies mentioned are included in the meta-analysis by Hooper et al. (2015) (5). The raw data of the mentioned working groups were obtained and re-analyzed according to the research question of Hooper et al. The data analyzed are based on both previously published studies and unpublished data from various working groups. Dehydr., dehydration; NPV, negative predictive value; PPV, positive predictive value; s-Osmol, serum osmolality; Sens., sensitivity; Spec., specificity; 95% CI, 95% confidence interval

Axillary moisture

Several studies analyzed the palpation of axillary moisture (3, 5, 6, 10, 11, 13), the measurement of axillary skin moisture using a skin moisture meter (10) or the measurement of axillary sweat by placing tissue paper in the axilla (13). Here, axillary dryness scored high in specificity (Table 3).

Table 3. Potential diagnostic utility of clinical examination.

Publication		Reference standard	Key statistical figures	Evaluation	Comments/ notes
Method assessed: Incident axillary dryness
Bunn & Hooper (2019) (6)		S-Osmol ≥ 295 mOsmol/kg	Sens. = 0.32 (95% CI [0.17; 0.48]) Spec. = 0.73 (95% CI [0.66; 0.81]) OR: 1.33 (95% CI [0.61; 2.90)		Axillary moisture, palpated manually
Hooper et al. (2015) (5)	Eaton et al. (1994)*1		Sens. = 0.50 (95% CI [0.27; 0.73]) Spec. = 0.82 (95% CI [0.70; 0.90])	“some potential” to diagnose hyperosmolal dehydration (4)
	Shimizu et al. (2012)*1		Sens. = 0.36 (95% CI [0.11; 0.69]) Spec. = 0.83 (95% CI [0.59; 0.96])
Eaton et. al (1994)*2 (13)		BUN/Cr ≥ 1:10 plasma osmolality ≥ 295 mOsmol/kg	Sens. = 0.50 (10/20 subjects) Spec. = 0.82 (54/66 subjects) PPV = 0.45; NPV =0.84	High NPV & good reproducibility support the use of this method.
Fortes et al. (2014) (3)		normal hydr., WDD (S-Osmol ≥ 295 mOsmol/kg) and WEDD (BUN/Cr ≥ 20 & normal S-Osmol)	Both types of dehydration OR = 1.4 (95% CI [0.7; 3.0]) WDD: OR = 1.3 (95% CI [0.5; 3.1]) WEDD: OR = 1.7 (95% CI [0.7; 4.2])	Dry axilla increases the chance of diagnosing dehydration.
Gross et al. (1992) (15)		Assessed normal hydr. (= 0) to severely dehydr. (= +3) (RR, S-Na & S-Osmol, BUN/Cr)	Kendall’s Tau correlation = 0.19 p ≥ 0.05 (n = 53)	Small study population (n = 53)
Shimizu et al. (2012)*3 (11)		Calculated S-Osmolarity ≥ 295 mOsmol/L	Sens. = 0.44 Spec. = 0.89 OR = 4.0	Axillary dryness as potential sign of dehydration
Kinoshita et al. (2013) (10)			Cut-off <30% skin moisture: Sens. = 0.12; Spec. = 0.91 Cut-off 40% skin moisture Sens. = 0.59; Spec. = 0.09	Axillary dryness useful to confirm suspected diagnosis of dehydration	Device-based measurement of skin moisture with skin moisture meter

*¹ The studies mentioned are included in the meta-analysis by Hooper et al. (2015) (5). The raw data of the mentioned working groups were obtained and re-analyzed according to the research question of Hooper, L. et al. The data analyzed are based on both previously published studies and unpublished data from various working groups. Due to this re-analysis, some of the results of the individual studies differ from those of the meta-analysis.

*² Results of the individual study by Eaton, D. et al. (1994) (13)

*³ Results of the individual study by Shimizu, M. et al. (2012) (11) BUN/Cr, blood urea nitrogen to creatinine ratio; dehydr., dehydrated; hydr., hydrated; OR, odds ratio; S-Osmol, serum osmolality; S-Na, serum sodium;

WDD, water-deficiency dehydration; WEMD, water- and electrolyte-deficiency dehydration.; Sens., sensitivity; Spec., specificity; 95% CI, 95% confidence interval

Technology-based diagnostic assessment

Ultrasonography of the inferior vena cava

Orso et al. assessed the feasibility of using ultrasound of the inferior vena cava to diagnose dehydration (defined by a BUN/Cr>20). A high diagnostic accuracy was found for both the parameters of expiratory diameter of the inferior vena cava with a cut-off of 1.55 cm and of the caval index with a cut-off of 0.48 (Table 4) (14).

Bioelectrical impedance analysis

The utility of bioelectrical impedance analysis (BIA) for determining the hydration status was evaluated in several studies (1, 5, 21–26, 29, 30). The use of calculated parameters, such as “total body water” (TBW), “extracellular water” (ECW) and others, was not convincing in any of these studies. However, there is evidence that directly measured resistance (R) (5) and the use of calculated parameters are of utility for intraindividual follow-up measurements (1, 30) (Table 4).

Discussion

Our review shows that many clinical signs used to assess the hydration status in older patients are not reliable. This includes clinical signs, such as positive skin turgor test, dry mucous membranes, sunken eyes, peripheral venous filling status, as well as pulse and blood pressure. Consequently, a clinical synopsis based on these criteria is of questionable utility, despite its frequent use in studies. Rosi et al. evaluated in 2022 a diagnostic approach based on the Geriatric Dehydration Screening Tool (GDST-M) which includes a questionnaire on drinking behavior, pain and mobility and also enables the assessment of various clinical signs, such as axillary dryness, body mass index (BMI) and dry mouth. In comparison with calculated serum osmolarity, this tool showed a sensitivity of 0.62 and a specificity of 0.47 (32). While this tool offers a higher diagnostic quality than any single method, this development does not seem to provide a breakthrough in the difficult assessment of the hydration status.

Furthermore, it becomes apparent that the absence of a gold standard, inconsistent definitions, and different reference methods not only affect the comparability of research results, but also make it difficult to translate them into practical recommendations for everyday clinical practice.

In many studies, directly measured serum osmolality is used as the reference method to diagnose hypertonic dehydration and this method is also recommended in the current ESPEN guideline on monitoring of the hydration status in geriatric patients (33). If this test is not available, calculated serum osmolarity or serum sodium levels can be used (7, 8), in which case care should be taken that an age-validated formula (by Khajuria & Krahn, 2005 [31]) and potentially reduced cut-off values (12) are applied.

In everyday clinical practice, easy-to-use, non-invasive methods are needed. The described (third-party) history of missing some drinks between meals can be easily obtained in patients without cognitive deficits. For this parameter, a negative predictive value of 1.00 was reported (5). However, data on this parameter are only available from one working group. Another study provides an argument for paying attention to drinking behavior. It showed that geriatric patients with a lower fluid intake also reported a less intense feeling of thirst; consequently, they were more likely to have elevated serum sodium levels (19).

Axillary moisture is a parameter that was investigated in numerous studies. Axillary dryness seems to indicate (hypertonic) dehydration with high specificity. In case of mere palpation-based findings, moisture quantification appears to be problematic.

To our surprise, one meta-analysis mentioned incident fatigue as a useful diagnostic criterion (5). It is noticeable in the cited studies that there is a large spread in the statistical values. Furthermore, weakness and exhaustion are classical criteria of frailty and difficult to quantify; thus, our team of authors does not believe it is appropriate to apply this criterion to geriatric patients.

The use of inferior vena cava (IVC) ultrasonography to assess the hydration status is an established method; however, only few studies on the geriatric population are available.

In our literature selection, Orso et al. proposed to diagnose dehydration (defined as an increased BUN/Cr ratio) by determining both the expiratory IVC diameter with a cut-off value of 1.55 cm and the caval index with a cut-off value of 48%, because of the high diagnostic accuracy achieved with this approach (14). Diederich and Burkhardt recommend to diagnose dehydration based on an inspiratory diameter of ≤ 0.4 cm, with a specificity of 0.80 (34). Boccardi et al. categorized the hydration status of geriatric patients based on the BUN/Cr ratio. They were unable to detect any differences between dehydrated, normally hydrated and hyperhydrated patients based on IVC ultrasound findings. However, this needs to be qualified by noting that rather unusual cut-off values were applied for the classification based on the BUN/Cr ratio (35).

One must not forget that the ability of IVC ultrasound to provide information about volume status is limited in patients with pulmonary hypertension and/or severe tricuspid regurgitation. The prevalence of pulmonary hypertension in persons aged 65 years and older is reported to be 5% to 10% (36).

According to our literature search, studies on other modalities, such as chest X-ray, measurement of central venous pressure, measurement of pulmonary artery occlusion pressure, and thermodilution or pulse contour measurements, are not available for older patients.

The role of bioelectrical impedance analysis in the assessment of HS in old age remains undetermined; the ESPEN guideline recommends against using this method (33).

It is possible that electrolyte concentrations are decisive for the result of total body water measurements using bioelectrical impedance analysis: In a study by Berneis et al., the natural water loss in young healthy subjects was replaced by infusion of hypertonic saline solution. Consequently, the actual TBW remained unchanged and plasma sodium levels increased from 142 to 148 mmol/L. As the result, bioelectrical impedance analysis showed an increase in TBW by 6.8% (37). However, studies comparing bioelectrical impedance analysis with isotope dilution methods come to more positive conclusions. This also applies to studies involving intra-individual controls (1, 21, 22); thus, it is currently not possible to predict the future significance of BIA.

Based on the analyzed literature, we currently consider the following parameters to be best supported by evidence with regard to the diagnosis of dehydration in older patients: directly measured serum osmolality, calculated serum osmolarity and serum sodium levels, as well as inferior vena cava ultrasound. To a lesser extent, a (third-party) history of missing some drinks between meals as well as axillary dryness can contribute to the diagnosis.

Before this paper went to press, we re-applied the developed search string. Other working groups confirmed that the existing diagnostic tools are inadequate for diagnosing dehydration in older age (38). Combining conventional methods also failed to achieve a significant improvement in diagnostic accuracy (32). This could be remedied by introducing innovative technology. Rodin et al. used a dehydration body monitor device—a polymer film-based photoplethysmographic device affixed to the case back-glass of a smartwatch—to measure the fluid loss in younger participants during a period of physical exercise. An advancement of this technology so that in can be used in geriatric patients is conceivable (39). Further technological innovations, such as smart containers or camera-based observations of the surrounding, could improve compliance with recommended fluid intake (40).

The available data indicate that all clinicians should critically reconsider their own approach for assessing the hydration status in older patients. To optimize the clinical assessment of patients’ hydration status, there seems to be a need for the rejection of unsuitable methods in favor of either newly developed criteria or of a combination of the best criteria already in use. In everyday clinical practice, the combination of the above four criteria could be a start. It has to be prospectively evaluated in future studies.