A 20-month-old girl is brought to the emergency department (ED) after 2 days of vomiting and diarrhea. Her father reports that she has not eaten normally since the illness began and now will not drink. She has had 8 stools so far today, but he does not think there were any diapers with urine in them. The child appears mildly ill but does make tears while crying. Her respiratory rate and quality are normal, along with her other vital signs. Her mouth is somewhat dry, capillary refill time is 1.5 seconds, and skin turgor is normal. Her serum (blood) urea nitrogen concentration (BUN) is 12 mg/dL, and bicarbonate concentration is 19 mEq/L.
A 5-month-old boy presents to a health care clinic in a developing country. The child lives in a rural area, and there is no running water in the family home. The child began having nonbloody, profuse, watery stools approximately 7 days ago. The family has World Health Organization (WHO) oral rehydration packets at home that the child has eagerly consumed. He seemed less interested in drinking this morning so his parents began the trip to the clinic. The child is now quiet and hyperpneic. He has sunken eyes and a dry mouth. His capillary refill time is 3 seconds, and his skin turgor is prolonged.
Why Is the Clinical Examination Important?
Dehydration is one of the leading causes of morbidity and mortality in children throughout the world.1, 2 Diarrheal disease and dehydration account for as much as 30% of worldwide deaths among infants and toddlers; 8000 children younger than 5 years die each day because of gastroenteritis and dehydration.2-4 In the United States, children younger than 5 years have an average of 2 episodes of gastroenteritis per year, leading to 2 to 3 million office visits and 10% of all pediatric hospital admissions.1, 5, 6 The direct costs of outpatient and hospital visits are more than $2 billion per year, not including indirect costs to families and society.4 Despite aggressive medical care, as many as 300 US children still die each year as a result of gastroenteritis and associated dehydration.1, 6
Many other childhood illnesses in addition to gastroenteritis are associated with dehydration. Gingivostomatitis, bronchiolitis, pyloric stenosis, and focal bacterial infections such as pneumonia, meningitis, and urinary tract infections can all lead to dehydration. For this reason, the morbidity and mortality related to dehydration are actually much higher than that associated solely with gastroenteritis. Dehydration is such a common concern in pediatrics that clinicians in primary care offices, EDs, and hospital settings all assess volume status as part of their evaluation. This assessment helps guide decisions about therapy and patient disposition.
The American Academy of Pediatrics (AAP), Centers for Disease Control and Prevention (CDC), and WHO have all developed treatment guidelines for gastroenteritis according to the clinical assessment of dehydration. The AAP guideline states that “the treatment of a child with diarrhea is directed primarily by the degree of dehydration present.”4 They recommend clinically deciding whether a patient is mildly (3%-5%), moderately (6%-9%), or severely (≥10%) dehydrated and then treating according to that classification. The CDC uses a similar assessment and scale in its recommendations on the initial management of diarrhea.1, 3 WHO has also incorporated signs of dehydration into the Integrated Management of Childhood Illness Scale, which assists practitioners in developing countries in making treatment and referral decisions.7
Inaccurate assessment of dehydration can have important consequences. Unrecognized and untreated fluid deficits can create electrolyte disturbances, acidosis, and end-organ damage, including cardiovascular instability, renal insufficiency, and lethargy. These complications can produce devastating results, including permanent injury or death. Conversely, unnecessary interventions can occur after erroneous assessment that a child has moderate or severe dehydration when he or she is actually euvolemic or only mildly dehydrated.5 Despite recommendations for oral rehydration in mild or moderate dehydration, this therapy is used in less than 30% of the cases of diarrhea in the United States for which it is indicated.8 Clinicians may rely on the more invasive intravenous rehydration in part because they overestimate the degree of dehydration. Both overestimating and underestimating the degree of dehydration can increase health care costs and cause unnecessary morbidity.
Pediatricians generally use the terms dehydration, volume depletion, and hypovolemia interchangeably to represent fluid loss in outpatient settings. Literature that focuses on physiologic changes caused by different types of fluid loss differentiates among these terms.9 Because this discrimination can have unclear clinical implications and to simplify discussion, much of the clinical literature combines terminology.10 Herein, we follow this convention and use the term dehydration to represent all fluid deficits except in circumstances such as whole blood loss or significant sodium alteration, in which important clinical implications are evident.
The quantification of dehydration is an important and commonly used skill for assessment of pediatric patients. Despite this importance, the utility of the clinical history, physical examination, and laboratory tests to assess dehydration in children has not been systematically reviewed. Most teaching regarding the assessment of dehydration is based on clinical experience and medical tradition. We conducted a systematic review of the literature on the precision and accuracy of medical history, physical examination, and laboratory tests in identifying dehydration in children between 1 month and 5 years old.
Anatomic/Physiologic Origins of Dehydration Signs
Many signs in pediatric assessment are attributed to the fluid and electrolyte shifts caused by dehydration. Early work to understand dehydration in children focused on intracellular and extracellular physiologic changes associated with fluid loss. Researchers have fastidiously documented fluid and electrolyte losses in dehydration and have even performed biopsies of the muscle of children with severe diarrhea to understand intracellular fluid and electrolyte shifts.11 Particularly instructive experiments used radiolabeled albumin to demonstrate that the percentage of body weight lost was directly proportional to the percentage of plasma volume lost.12 For example, children who had lost 5% of their body weight lost approximately 5% of their plasma volume. Because plasma volume is only a small percentage of total body water, this experiment indirectly demonstrated that the majority of fluid lost in childhood dehydration actually comes from either interstitial or intracellular sources.
The correlation of losses from specific fluid compartments to corresponding physical signs has not been clearly documented. The signs of dehydration appear to represent an actual desiccation of tissue (eg, dry mucous membranes), a compensatory reaction of the body to maintain vital perfusion (eg, tachycardia), or some combination of both (eg, capillary refill time). Although some authors offer more specific explanations of theoretic fluid compartments and their examination correlates, these 3 principles should be sufficient for clinical assessment of patients.
How to Elicit Symptoms and Signs
Pediatrics practitioners often elicit historical points from adult caregivers instead of directly from the patient. When assessing volume status in infants, physicians may ask about number of wet diapers (surrogate for urine output), presence or absence of vomiting and diarrhea, and amount and type of oral intake. Caregivers also frequently report their interpretation of examination signs by clarifying whether the child is active, whether the eyes appear sunken, and whether the child drinks vigorously. Clinicians should ask parents whether they have given a successful trial of clear fluids at home, whether the child has been treated by another medical practitioner during the illness, and the date and value of the child's most recent weight measurement.1, 13
The ability to elicit some examination signs is impaired when pediatric patients are crying and uncooperative. Therefore, assessment of hydration status should progress from the least to the most invasive maneuvers. The examination should begin with the child across the room in a position of comfort (eg, in the parent's lap). Overall appearance, activity, and response of the child to stimulation should be observed. Evaluating the respiratory pattern is important for assessment of dehydration and all other acute illnesses. Respiratory rate should be measured for 60 seconds by observing chest wall movements. The precise measurement requires a quiet and comfortable child. The rate should be compared with age-based norms.14 In a potentially dehydrated child, the examiner should specifically look for hyperpnea (deep, rapid breathing without other signs of respiratory distress), suggestive of an acidosis.1 Other vital signs, including temperature, pulse, and blood pressure, should also be evaluated while the child is comfortable.1
Next, the clinician should assess skin turgor and capillary refill time. Skin turgor has been used to diagnose dehydration for more than 50 years and, when abnormal, is also called “tenting” or “inelastic skin.”15, 16 To elicit the sign, the examiner should use the thumb and index finger to pinch a small skin fold on the lateral abdominal wall at the level of the umbilicus.15 The fold should be promptly released, and then the time is measured for the skin fold's return to normal form.15 Clear norms for this time have not been published, and most clinicians simply qualify skin turgor as immediate, slightly delayed, or prolonged.
Excess subcutaneous fat and hypernatremia may falsely normalize the turgor in dehydrated children, whereas malnutrition may falsely prolong the recoil time.15, 17-21 Primary skin disorders complicate the interpretation of skin turgor.19
To assess capillary refill time, the examiner compresses a superficial capillary bed and estimates the time it takes for normal color to return after the pressure is released. Capillary refill time varies as a function of ambient temperature, site of application, lighting, medications, and primary (eg, reflex sympathetic dystrophy) or secondary (eg, cardiogenic shock) autonomic changes.16, 18, 22-24 Extremes in patient temperature may also affect the capillary refill time; for example, capillary refill times are markedly prolonged after cold immersion.25 However, Gorelick et al22 found that fever did not affect the test characteristics in children with vomiting, diarrhea, or poor oral intake. According to the available studies, and to standardize examination techniques, we recommend assessing capillary refill time on a finger with the arm at the level of the heart in a warm ambient temperature. Pressure should be gradually increased on the palmar surface of the distal fingertip and then released immediately after the capillary bed blanches. The time elapsed until restoration of normal color should be estimated. Although many practitioners use other sites to measure capillary refill time, most studies of this sign use the palmar surface of the distal fingertip.22-26 Using this approach, values for nondehydrated children are less than 1.5 to 2 seconds.25
Search Strategy and Quality Review
We identified articles by direct searches of the MEDLINE database via the PubMed search engine. The first and broadest search strategy used “dehydration” and “diagnosis,” “hypovolemia” and “diagnosis,” or “intravascular volume depletion” and “diagnosis.” All searches were limited by age (all children: 0-18 years) and publication date (January 1966 to April 2003). These searches produced 1537 articles. We supplemented this preliminary search with the standardized search technique used in The Rational Clinical Examination series (available from the authors). This second search produced 24 additional articles.
Each of the authors reviewed the titles and available abstracts from the 1561 articles, selecting for further review those that appeared to address the evaluation of dehydration in children aged 1 month to 5 years. We did not exclude articles if the study enrolled some children outside that age range. Through consensus, we identified 68 articles as potential sources of primary data or reviews with potential background information and thorough reference lists.
To ensure a comprehensive literature review, we used additional techniques to identify articles (Figure 25-1). One author (M.J.S.) searched for individual symptoms and signs associated with the diagnosis of dehydration in children. These terms included “capillary refill,” “skin turgor,” “dry cry,” “tears,” “mucous membrane,” “sunken eyes,” “fontanelle” and “dehydration,” “urine specific gravity,” “urine” and “dehydration,” “hemoconcentration,” “BUN,” “urine,” “blood pressure,” “bioimpedance,” “orthostasis,” “respiration,” “parent” and “dehydration,” “pulse,” and “heart rate” (all limit: aged 0-18 years, human, NOT “dehydration” and “diagnosis”). The Cochrane Library, reference lists of pediatric and physical examination textbooks,27-32 reference lists of all included articles, and articles from the collections of experts in the field were reviewed. Forty-two potential articles were identified from the supplemental searches.
Selection Process for Studies Included in Review
We performed a full review of the 110 retained articles to identify those with primary data comparing dehydration with a symptom, sign, or laboratory value in pediatric patients. Twenty-six articles met these criteria and underwent a full quality assessment with an established methodologic filter that has been consistently used and described in The Rational Clinical Examination series (see Table 1-7).33 A second author then checked the initial quality review. The group always arrived at a consensus on the final evidence quality level assigned.
Nine of the 110 articles that underwent a full-text review were written in languages other than English. Medical school faculty, residents, or students at our institution who were primary speakers of the written language read each of these articles. Six of these 9 articles did not meet inclusion criteria and were excluded, whereas 3 were assigned an evidence quality level according to a translation of the article.
No studies on physical examination signs, symptoms, or laboratory results in childhood dehydration demonstrated evidence quality criteria for level 1 or 2. Four studies were assigned to level 3, but one of these was eventually excluded because the study population overlapped with that in another included study.22 Twelve studies were initially assigned to level 4, although one was excluded because of methodologic flaws12 and another was excluded because of its retrospective design and restriction to children with pyloric stenosis.34
We chose the difference between the rehydration weight and the acute weight divided by the rehydration weight as the best available gold standard of percentage of volume lost.35 Ten articles used gold standards based solely on examination signs or a general dehydration assessment. These were assigned an evidence quality level of 5 and were subsequently excluded. Figure 25-1 shows a schematic representation of the methods, and Table 25-1 summarizes the 13 included studies.
Table 25-1.Summary of Included Studies |Favorite Table|Download (.pdf) Table 25-1. Summary of Included Studies
|Source, y ||Evidence Quality Level ||Country ||Setting ||No. of Participants ||Age Range ||Inclusion Criteria |
|Porter et al,13 2003 ||3 ||United States ||Emergency department ||71 ||1 mo-5 y ||Chief complaint of vomiting, diarrhea, or poor oral intake |
|Laron,15 1957 ||4 ||United States ||Hospital ||21 ||1 mo-3.5 y ||Admitted with diarrhea |
|Saavedra et al,16 1991 ||4 ||United States ||Hospital ||32 ||2-24 mo ||Admitted with diarrhea |
|Duggan et al,18 1996 ||4 ||Egypt ||Gastroenteritis clinic ||135 ||3-18 mo ||Acute diarrhea and dehydrated |
|Gorelick et al,35 1997 ||3 ||United States ||Emergency department ||225 ||1 mo-5 y ||Chief complaint of vomiting, diarrhea, or poor oral intake |
|Duggan et al,36 1997 (precision only) ||3 ||Egypt ||Gastroenteritis clinic ||100 ||2 mo-4 y ||>5 Stools in last 24 h |
|MacKenzie et al,37 1989 ||4 ||Australia ||Hospital ||102 ||<4 y ||Admitted with gastroenteritis and dehydration |
|English et al,38 1997 ||3 ||Kenya ||Hospital ||119 ||>1 mo ||Admitted with malaria and coma, respiratory distress, or prostration |
|Plata Rueda and Diaz Cruz,39 1974 ||4 ||Columbia ||Hospital ||100 ||<73 mo ||Admitted with diarrhea and dehydration |
|Vega and Avner,40 1997 ||4 ||United States ||Emergency department ||97 ||2 wk-15 y ||Dehydrated and needed intravenous fluids |
|Amin et al,41 1980 ||4 ||Indonesia ||Hospital ||36 ||<24 mo ||Admitted with diarrhea and dehydration |
|Teach et al,42 1997 ||4 ||United States ||Emergency department ||40 ||2 wk-12 y ||Dehydrated and needed intravenous fluids |
|Yilmaz et al,43 2002 ||4 ||Turkey ||Emergency department ||168 ||1-21 mo ||Received intravenous fluids and hospitalized for gastroenteritis and dehydration |
We report precision data as a range of κ values obtained directly from the published results. Two-by-two tables were created from the published information regarding accuracy and were used to calculate point estimates and 95% confidence intervals (CIs) for the sensitivity, specificity, and likelihood ratios (LRs) for each test.44 One author provided original data to calculate these values because they were not calculable from the original publication.18 We created these 2 × 2 tables for detecting both 5% and 10% dehydration when data were available. A range of values was provided when only 2 studies evaluated an individual diagnostic test. If more than 2 studies evaluated a test, then we combined the results with a random-effects model. Data for meta-analysis were not weighted according to the quality of included studies. Statistical tests were performed with STATA software, version 7.0 (StataCorp, College Station, Texas).
We performed tests of heterogeneity for data used in all meta-analyses and found significant heterogeneity for most signs. Analysis of data with a random-effects model is complicated by the presence of heterogeneity. However, combining data in this manner allows clinicians to make general summary “best estimates” of utility according to all of the included studies. Furthermore, the degree of uncertainty between LRs of summary estimates was more obvious with the broad range of 95% CIs as opposed to the narrower range for the individual point estimates. Thus, the summary LRs lower the risk of clinicians being overly confident about the utility of clinical findings.
Precision of Symptoms and Signs
Porter et al13 evaluated the agreement between parental observation of examination signs and the signs elicited by trained ED nurses. The κ value demonstrated substantial agreement beyond chance when assessing for a sunken anterior fontanelle (κ = 0.73) and presence of cool extremities (κ = 0.70). There was moderate agreement on general appearance (κ = 0.46), presence of sunken eyes (κ = 0.49), absence of tears (κ = 0.57), and presence of dry mouth (κ = 0.52).
Three included studies reported interrater agreement among clinicians, ranging from chance to good agreement (Table 25-2).16, 35, 36 Agreement on respiratory rate and pattern may be no better than that which occurs by chance. The other signs had higher levels of agreement, although the range of κ levels for these findings was broad.
Table 25-2.Precision of Examination Signs for Dehydration |Favorite Table|Download (.pdf) Table 25-2. Precision of Examination Signs for Dehydration
|Finding ||Reference ||Total No. of Participants ||Range of κ Values |
|Prolonged capillary refill ||16, 35, 36 ||216 ||0.01 to 0.65 |
|Abnormal skin turgor ||35, 36 ||184 ||0.36 to 0.55 |
|Abnormal respiratory pattern ||35, 36 ||184 ||–0.04 to 0.40 |
|Extremity perfusion ||35 ||100 ||0.23 to 0.66 |
|Absent tears ||35, 36 ||184 ||0.12 to 0.75 |
|Sunken fontanelle ||36 ||100 ||0.10 to 0.27 |
|Sunken eyes ||35, 36 ||184 ||0.06 to 0.59 |
|Dry mucous membranes ||35, 36 ||184 ||0.28 to 0.59 |
|Weak pulse ||35, 36 ||184 ||0.15 to 0.50 |
|Poor overall appearance ||35, 36 ||184 ||0.18 to 0.61 |
Accuracy of Symptoms, Signs, and Laboratory Studies
Three studies evaluated the accuracy of history-taking in assessing dehydration.13, 35, 37 All 3 of these studies evaluated history of low urine output as a test for dehydration. In the pooled analysis, low urine output did not increase the likelihood of 5% dehydration (LR, 1.3; 95% CI, 0.9-1.9). Porter et al13 showed that a history of vomiting, diarrhea, decreased oral intake, reported low urine output, a previous trial of clear liquids, and having seen another clinician during the illness before presenting to the ED yielded LRs that lacked utility in the assessment of dehydration. However, their data did suggest that children who had not been previously evaluated by a physician during the illness might be less likely to be dehydrated on presentation (LR, 0.09; 95% CI, 0.01-1.4). Similarly, parental report of a normal urine output decreases the likelihood of dehydration (Gorelick et al35 reported an LR of 0.27 [95% CI, 0.14-0.51] and Porter et al13 reported an LR of 0.16 [95% CI, 0.01-2.5]).
Table 25-3 is a comprehensive list of individual physical examination signs and their test characteristics in evaluating children for 5% dehydration. Signs were included when they were evaluated in 2 or more studies, and calculations based on pooled results were performed when evaluated in 3 or more studies.
Table 25-3.Summary Test Characteristics for Clinical Findings to Detect 5% Dehydration |Favorite Table|Download (.pdf) Table 25-3. Summary Test Characteristics for Clinical Findings to Detect 5% Dehydration
|Finding ||Reference ||Total No. of Participants ||LR Summary, Value (95% CI) or Range ||Sensitivity (95% CI) ||Specificity (95% CI) |
| || || ||Present ||Absent || || |
|Prolonged capillary refill ||16, 35, 37, 38 ||478 ||4.1 (1.7-9.8) ||0.57 (0.39-0.82) ||0.60 (0.29-0.91) ||0.85 (0.72-0.98) |
|Abnormal skin turgor ||15, 18, 35, 37, 38 ||602 ||2.5 (1.5-4.2) ||0.66 (0.57-0.75) ||0.58 (0.40-0.75) ||0.76 (0.59-0.93) |
|Abnormal respiratory pattern ||18, 35, 37, 38 ||581 ||2.0 (1.5-2.7) ||0.76 (0.62-0.88) ||0.43 (0.31-0.55) ||0.79 (0.72-0.86) |
|Sunken eyes ||13, 18, 35, 37 ||533 ||1.7 (1.1-2.5) ||0.49 (0.38-0.63) ||0.75 (0.62-0.88) ||0.52 (0.22-0.81) |
|Dry mucous membranes ||13, 18, 35, 37 ||533 ||1.7 (1.1-2.6) ||0.41 (0.21-0.79) ||0.86 (0.80-0.92) ||0.44 (0.13-0.74) |
|Cool extremity ||13, 18 ||206 ||1.5-19 ||0.89-0.97 ||0.10-0.11 ||0.93-1.0 |
|Weak pulse ||18, 35 ||360 ||3.1-7.2 ||0.66-0.96 ||0.04-0.25 ||0.86-1.0 |
|Absent tears ||13, 35, 37 ||398 ||2.3 (0.9-5.8) ||0.54 (0.26-1.1) ||0.63 (0.42-0.84) ||0.68 (0.43-0.94) |
|Increased heart rate ||18, 35, 37 ||462 ||1.3 (0.8-2.0) ||0.82 (0.64-1.0) ||0.52 (0.44-0.60) ||0.58 (0.33-0.82) |
|Sunken fontanelle ||13, 18, 37 ||308 ||0.9 (0.6-1.3) ||1.12 (0.82-1.5) ||0.49 (0.37-0.60) ||0.54 (0.22-0.87) |
|Poor overall appearance ||13, 35, 37 ||398 ||1.9 (0.97-3.8) ||0.46 (0.34-0.61) ||0.80 (0.57-1.0) ||0.45 (–0.1 to 1.0) |
Three signs were evaluated in multiple studies, had a clinically helpful pooled LR in detecting 5% dehydration, and had 95% CIs wholly above 1.0. Capillary refill time was evaluated in 4 studies, and the pooled sensitivity of prolonged capillary refill time was 0.60 (95% CI, 0.29-0.91), with a specificity of 0.85 (95% CI, 0.72-0.98), for detecting 5% dehydration.16, 35, 37, 38 The LR for abnormal capillary refill time was 4.1 (95% CI, 1.7-9.8). This was the highest value among examination signs with pooled results. Abnormal skin turgor had a pooled LR of 2.5 (95% CI, 1.5-4.2)15, 18, 35, 37, 38 and abnormal respiratory pattern had a pooled LR of 2.0 (95% CI, 1.5-2.7).18, 35, 37, 38
Presence of cool extremities or a weak pulse or absence of tears also may be helpful tests for dehydration. Absence of tears had a pooled LR of 2.3 (95% CI, 0.9-5.8), but the potential utility is limited by a wide 95% CI that crosses 1.0.13, 35, 37 Two studies examined a weak pulse quality as a test for dehydration. One study found a reasonably precise LR for weak pulse of 3.1 (95% CI, 1.8-5.4),35 but in the other study, the 95% CI was too wide to make a reasonable estimate (LR, 7.2; 95% CI, 0.4-150).18 The 2 studies that evaluated cool extremities as a test of dehydration found imprecise point estimates for the positive likelihood ratio (LR+) in detecting 5% dehydration (LR, 19; 95% CI,18 1.1-330 and LR, 1.5; 95% CI,13 0.2-12).
Sunken eyes and dry mucous membranes offer little help clinically; both had narrow 95% CIs but pooled LRs of 1.7. An increased heart rate, a sunken fontanelle in young infants, and an overall poor appearance are frequently taught as good tests for dehydration. However, the objective evidence reveals that all have summary LRs of less than 2.0 and 95% CIs that cross 1.0.
Some tests may be clinically useful in decreasing the likelihood of dehydration. Absence of dry mucous membranes (LR, 0.41; 95% CI, 0.21-0.79), a normal overall appearance (LR, 0.46; 95% CI, 0.34-0.61), and absence of sunken eyes (LR, 0.49; 95% CI, 0.38-0.63) had pooled LRs of less than 0.5. Most clinical scenarios require lower LRs than these to rule out dehydration effectively.
Four studies evaluated clinical prediction models or groups of signs.18, 35, 39, 40 Vega and Avner40 evaluated a table similar to that used in many pediatric textbooks and also commonly taught to medical students as the best evaluation tool for dehydration.30 This scale, displayed in Table 25-4, is similar to the one used by the AAP and CDC in their recommendations for the management of acute gastroenteritis.1, 3, 4 The tool uses the assessment of 9 physical examination findings to classify children as mildly (4%-5%), moderately (6%-9%), or severely (≥10%) dehydrated. In 97 children presenting to the ED with dehydration requiring intravenous fluids, a classification of severe on the scale had an LR of 3.4 (95% CI, 1.5-7.7) for the presence of at least 5% dehydration. Classification of severe dehydration also yielded an LR of 4.3 (95% CI, 2.4-7.8) for at least 10% dehydration. A moderate classification by examination was less useful to diagnose 5% dehydration (LR, 2.1; 95% CI, 0.9-4.8).40
Table 25-4.Example of a Commonly Taught Dehydration Assessment Scalea |Favorite Table|Download (.pdf) Table 25-4. Example of a Commonly Taught Dehydration Assessment Scalea
|Variable/Sign ||Dehydration |
| ||Mild (4%-5%) ||Moderate (6%-9%) ||Severe (≥10%) |
|General appearance ||Thirsty, restless, alert ||Thirsty, drowsy, postural hypotension ||Drowsy, limp, cold, sweaty, cyanotic extremities |
|Radial pulse ||Normal rate and strength ||Rapid and weak ||Rapid, thready, sometimes impalpable |
|Respirations ||Normal ||Deep, may be rapid ||Deep and rapid |
|Anterior fontanelle ||Normal ||Sunken ||Very sunken |
|Systolic blood pressure ||Normal ||Normal or low ||Low |
|Skin elasticity ||Pinch retracts immediately ||Pinch retracts slowly ||Pinch retracts very slowly |
|Eyes ||Normal ||Sunken ||Grossly sunken |
|Tears ||Present ||Absent ||Absent |
|Mucous membranes ||Moist ||Dry ||Very dry |
Duggan et al18 evaluated 2 dehydration assessment scales that classified children as mild, moderate, or severe according to the number of dehydration examination signs present. The authors reported the final mean percentage of dehydration within each group, and these averages increased significantly as the severity assessment increased,18 which suggests that as more signs of dehydration appear, children tend to be more dehydrated. Plata Rueda and Diaz Cruz39 also presented groupings of signs and symptoms that attempted to stratify children into different degrees of dehydration. Minor physical examination changes did not significantly change the likelihood of dehydration; however, the presence of abnormal skin turgor on the abdomen, thorax, extremities, and face, combined with sunken eyes, dry mucous membranes, and a sunken fontanelle, did increase the likelihood of 10% dehydration (LR, 3.7; 95% CI, 1.6-8.1).39
Gorelick et al35 created a scale giving equal weight to 10 commonly elicited signs: decreased skin elasticity, capillary refill time greater than 2 seconds, general appearance, absence of tears, abnormal respirations, dry mucous membranes, sunken eyes, abnormal radial pulse, tachycardia (heart rate > 150/min), and decreased urine output. The presence of at least 3 of the 10 signs had a sensitivity of 0.87 and a specificity of 0.82 in detecting 5% dehydration (LR+, 4.9; 95% CI, 3.3-7.2, and negative LR, 0.15; 95% CI, 0.08-0.30). Similarly, 7 of 10 signs had an LR+ of 8.4 (95% CI, 5.0-14) in diagnosing 10% dehydration. A logistic regression analysis performed by Gorelick et al35 showed that capillary refill time, dry mucous membranes, absence of tears, and abnormal overall appearance contained most of the predictive power. A simplified assessment tool using the presence of 2 of these 4 signs yielded an LR+ of 6.1 (95% CI, 3.8-9.8) for diagnosing 5% dehydration.35
Six studies evaluated the utility of laboratory tests in assessing dehydration (Table 25-5).37, 38, 40-43 Five studies evaluated BUN concentration or BUN/serum creatinine ratio as a test for dehydration.37, 38, 41-43 BUN cutoffs of 8, 18, and 27 mg/dL produced LRs ranging from 1.4 to 2.9. Yilmaz et al43 found that in a group of hospitalized children with gastroenteritis, BUN greater than 45 mg/dL was specific for at least 5% dehydration (specificity of 1.0). However, this was a small study and the estimated 95% CI for an LR+ was 3 to 730.
Table 25-5.Summary Test Characteristics for Laboratory Tests Assessing Dehydration |Favorite Table|Download (.pdf) Table 25-5. Summary Test Characteristics for Laboratory Tests Assessing Dehydration
|Laboratory Value ||Reference ||Total No. of Participants ||Sensitivity, Value (95% CI) or Range ||Specificity, Value (95% CI) or Range ||LR Summary, Value (95% CI) or Range |
| || || || || ||Present ||Absent |
|Blood urea nitrogen, mg/dL |
| >8 ||37, 38 || ||0.38-0.71 ||0.71-0.82 ||2.1-2.4 ||0.41-0.76 |
| >18 ||41, 43 || ||0.63-0.90 ||0.55-0.57 ||1.4-2.1 ||0.17-0.68 |
| >27 ||41 ||36 ||0.44 (0.19-0.68) ||0.85 (0.69-1.0) ||2.9 (0.9-9.5) ||0.66 (0.41-1.1) |
| >45 ||43 ||168 ||0.43 (0.34-0.52) ||0.99 (0.96-1.0) ||46.1 (2.9-733) ||0.58 (0.49-0.68) |
|Blood urea nitrogen/creatinine ratio > 40 ||42 ||40 ||0.23 (0.01-0.46) ||0.89 (0.77-1.0) ||2.1 (0.5-8.9) ||0.87 (0.62-1.2) |
|Bicarbonate, mEq/L |
| <17 ||40 ||97 ||0.83 (0.72-0.94) ||0.76 (0.64-0.88) ||3.5 (2.1-5.8) ||0.22 (0.12-0.43) |
| <15 ||43 ||168 ||0.93 (0.88-0.98) ||0.40 (0.26-0.53) ||1.5 (1.2-1.9) ||0.18 (0.08-0.37) |
|Base deficit > 7 mEq/L ||37, 38 || ||0.67-0.75 ||0.52-0.59 ||1.4-1.8 ||0.42-0.68 |
|pH < 7.35 ||37 ||102 ||0.43 (0.28-0.58) ||0.80 (0.70-0.91) ||2.2 (1.2-4.1) ||0.71 (0.53-0.95) |
|Anion gap > 20 mEq/L ||42 ||40 ||0.46 (0.19-0.73) ||0.74 (0.58-0.91) ||1.8 (0.8-4.2) ||0.73 (0.42-1.3) |
|Uric acid > 10 mg/dL ||42 ||40 ||0.23 (0.01-0.46) ||0.78 (0.62-0.93) ||1.0 (0.3-3.5) ||0.99 (0.69-1.4) |
Four studies evaluated acidosis as a test for dehydration.37, 38, 40, 43 Most patients enrolled in these studies had acute diarrhea, a potential cause of acidosis. Mackenzie et al37 and English et al38 used a base deficit of greater than 7 as the measure of acidosis. (Base deficit estimates the severity of metabolic acidosis by comparing the patient's bicarbonate concentration to historical norms for a given pH and Pco2.) In both studies, the LR+ was less than 2.0. Although Yilmaz et al43 found that an absolute serum bicarbonate concentration of less than 15 mEq/L was not helpful (LR for low serum bicarbonate, 1.5; 95% CI, 1.2-1.9), Vega and Avner40 found that an absolute bicarbonate concentration of less than 17 mEq/L offered some help in diagnosing children with 5% dehydration (LR, 3.5; 95% CI, 2.1-5.8). Teach et al42 evaluated serum uric acid and an increased anion gap as tests for dehydration but found that abnormal results were not helpful. Urine specific gravity was evaluated by English et al38 but was not found to be significantly correlated with dehydration. The only laboratory measurement that appears to be valuable in decreasing the likelihood of 5% dehydration is serum bicarbonate. A serum bicarbonate concentration of more than 15 or 17 mEq/L has an LR range of 0.18 to 0.22, reducing the likelihood of dehydration if the child has gastroenteritis.40, 43
The published literature on assessment of dehydration has significant limitations affecting both internal and external validity. As discussed in the “Methods” section, none of the identified studies met the criteria for high-quality (level 1 or level 2) evidence according to the established methodologic filter. The best available studies had modest sample sizes, used nonconsecutive patients, and did not compare the included children with those excluded from the study populations. The most common bias in level 4 evidence studies was that they enrolled children already thought to be dehydrated and to need intravenous fluids or who were admitted to the hospital. The diagnostic tests may perform better in children who are thought to be dehydrated compared with children solely at risk of dehydration. Thus, there may be limitations to the generalizability of these results when applied to an unselected group of children simply at risk of dehydration.
The results of the study by Gorelick et al35 differed from those of the other included studies. Gorelick et al35 evaluated the interrater reliability for 10 physical examination signs. The κ values ranged from 0.40 to 0.75, which were clearly better than those found in the other studies on precision by Saavedra et al16 and Duggan et al.36 The accuracy of signs was also generally better in the study by Gorelick et al35 than in other included studies. The LRs of positive tests were all statistically significant and ranged from 1.8 to 12. All 10 of the signs evaluated by Gorelick et al35 were assessed in other studies. For 9 of the 10 signs, the results by Gorelick et al35 produced the highest LRs of any included study, which is difficult to explain. The study by Gorelick et al35 is of high methodologic quality in comparison with the other included studies. It achieved an evidence quality level 3 according to nonconsecutive patient selection that did not introduce a clear systematic bias. They enrolled a relatively large group of patients and followed them meticulously. The sensitivity values of the tests were generally similar to those found in other studies, but the specificity was often much higher. The high percentage of true-negative test results may have been affected by a patient population with a relatively low incidence of disease in comparison with patients enrolled in the other studies.35
Ten of the 26 articles that met initial inclusion criteria were later found to have a methodologic flaw with the diagnostic standard and were excluded from the final analysis. These studies used a gold standard for dehydration according to examination signs or clinical assessment, which represents a circular flaw in assessing the utility of the history taking or examination in establishing dehydration. Conversely, the difference between an ill weight and a rehydrated weight (after illness) appears to be the best pragmatic diagnostic standard for dehydration that has been validated in the literature.35 However, problems can be introduced by the timing of the rehydration weight. For example, if it is obtained too early, children may still be dehydrated or may actually be overhydrated because of aggressive intravenous fluid administration. The timing of the rehydration weight varied among the included studies, and most studies used additional assessments to validate their perception of a true rehydration weight. For example, Teach et al42 used the weight when the physical examination findings had normalized and the urine-specific gravity level was low. Incorporating other assessments that were not based on weight into the gold standard could bias the results. Some studies avoided this problem by documenting the rehydration weight when measured weight remained unchanged over time.35 Another criticism of a weight-based gold standard is that infants may “gain” a significant percentage of their body weight if they have a full bladder and colon, which they may then “lose” when they void.20 In studies of large sample size, the weight contribution of a full bladder would be unlikely to have a major effect on the LRs for clinical findings. Additionally, the number of children with weight “gained” or “lost” because of impending or recent voids should balance.
Pediatricians are taught that hypernatremia may alter the test characteristics of signs in dehydration.30 For example, prolonged skin turgor is less sensitive in detecting significant dehydration in children with diabetes insipidus and pure water loss than in children with diarrhea.15 Because of this clinical experience, some studies excluded children with significant hypernatremia.35, 38 Other studies used subgroup analysis to demonstrate that assessment had not been affected by hypernatremia.37, 43 Because tests of dehydration are usually applied without any knowledge of the serum sodium level in the patient, it seems appropriate to structure studies without excluding hypernatremic children.
Dehydration is an important cause of morbidity and mortality as a complication of pediatric illness. However, the literature evaluating the symptoms, signs, and laboratory values for assessing dehydration is limited. We found few high-quality studies with accurate gold standards and minimal systematic bias.
The evidence shows that tests of dehydration are imprecise, generally showing only fair to moderate agreement among examiners. Historical points have moderate sensitivity as a screening test for dehydration. However, parental reports of dehydration symptoms are so nonspecific that they may not be clinically useful. The best 3 individual examination signs for assessing dehydration are prolonged capillary refill time, abnormal skin turgor, and abnormal respiratory pattern. Groups of signs or use of clinical scales improves diagnostic characteristics. Commonly obtained laboratory tests such as BUN and bicarbonate concentrations generally are only helpful when results are markedly abnormal. A normal bicarbonate concentration helps somewhat to reduce the likelihood of dehydration. These laboratory tests should not be considered definitive for dehydration.
The literature reports more than 30 potential tests for detecting dehydration. This large number should not distract clinicians from focusing on signs and symptoms with proven diagnostic utility. Unfortunately, the data also suggest that signs of dehydration can be imprecise and inaccurate, making clinicians unable to predict the exact degree of dehydration. For this reason, we agree with the WHO and other groups that recommend using the physical examination to classify dehydration as “none,” “some,” or “severe.”1, 45 This general assessment can then be used to guide clinical management.
Author Affiliations at the Time of the Original Publication
Departments of Pediatrics (Drs Steiner, DeWalt, and Byerley) and Internal Medicine (Drs Steiner and DeWalt), University of North Carolina School of Medicine, and Robert Wood Johnson Clinical Scholars Program, University of North Carolina (Dr DeWalt), Chapel Hill. Dr Steiner is now at the Clinica Sierra Vista, Bakersfield, California.
Dr DeWalt received salary support from the Robert Wood Johnson Clinical Scholars Program.
We thank Kenneth B. Roberts, MD, Lee S. Benjamin, MD, Peter Margolis, MD, Anita K. Ying, MD, and William Miller, MD, PhD, for their thoughtful comments on drafts of the manuscript. We also thank Marcos Mestre, MD, Myriam F. Bauer, MD, Sandro Pinheiro, PhD, Kaori Longphre, MD, and Pia J. Hauk, MD, for their assistance in translating articles written in languages other than English.
The historical clues provided by the father are minimally helpful in assessing the child's dehydration. There are no signs present that increase the likelihood of dehydration. The negative LRs associated with the absence of multiple examination signs and the serum bicarbonate concentration of 19 mEq/L make significant dehydration much less likely. This child probably has “no” dehydration instead of “some” or “severe” dehydration.
The hyperpnea, prolonged capillary refill time, and delayed skin turgor all increase the likelihood of dehydration. Because there are multiple signs of dehydration, the possibility of severe dehydration should be considered and treated appropriately.
The following disclosures were reported at the time this original article was first published in JAMA.
Funding/Support: Dr DeWalt received salary support from the Robert Wood Johnson Clinical Scholars Program.
Role of the Sponsor: The sponsor played no role in the data collection and analysis, manuscript preparation, or authorization for publication.
C. Managing acute gastroenteritis among children. MMWR Recomm Rep. 2003;52(RR-16):1–16.
J. Where and why are 10 million children dying every year? Lancet.
RI. The management of acute diarrhea in children. MMWR Recomm Rep. 1992;41(RR-16):1–20.
American Academy of Pediatrics, Provisional Committee on Quality Improvement
, Subcommittee on Acute Gastroenteritis. Practice parameter: the management of acute gastroenteritis in young children. Pediatrics.
MS. Estimates of morbidity and mortality rates for diarrheal diseases in American children. J Pediatr. 1991;118(4 pt 2):S27-S33.
S. Integrated management of childhood illness by outpatient health workers: technical basis and overview. Bull World Health Organ. 2003;75:7–24.
et al.. Oral rehydration therapy for diarrhea: an example of reverse transfer of technology. Pediatrics. 1997;100(5):E10.
et al.. Language guiding therapy: the case of dehydration versus volume depletion. Ann Intern Med. 1997;127(9):848–853.
DC. The retention of electrolyte during recovery from severe dehydration due to diarrhea. Pediatrics. 1949;3:129–156.
JW. Plasma volume as an index of total fluid loss. Am J Dis Child. 1961;102:190–193.
K. The value of parental report for diagnosis and management of dehydration in the emergency department. Ann Emerg Med.
Z. Skin turgor as a quantitative index of dehydration in children. Pediatrics.
L. Capillary refilling (skin turgor) in the assessment of dehydration. Am J Dis Child.
Jr. Therapy of infants with hypertonic dehydration due to diarrhea. Am J Dis Child.
et al.. How valid are clinical signs of dehydration in infants? J Pediatr Gastroenterol Nutr.
KL. Skin turgor: do we understand the clinical sign
W. Clinical signs of dehydration in children. Lancet.
Dehydration and fat babies. BMJ. 1971;1(5741):125.
MD. Effect of fever on capillary refill time. Pediatr Emerg Care.
M. Effect of ambient temperature on capillary refill in healthy children. Pediatrics.
T. Adverse lighting condition effects on the assessment of capillary refill. Am J Emerg Med.
L. Defining normal capillary refill: variation with age, sex, and temperature. Ann Emerg Med.
I. Capillary refill and core-peripheral temperature gap as indicators of haemodynamic status in paediatric intensive care patients. Arch Dis Child.
R. DeGowin & DeGowin's Diagnostic Examination. 6th ed. New York, NY: McGraw-Hill Co; 1994.
C. Harriet Lane Pediatric Handbook. 16th ed. St Louis, MO: Elsevier Science; 2002.
J. The Art and Science of Bedside Diagnosis. Baltimore, MD: Urban & Schwarzenberg; 1990.
H. Nelson Textbook of Pediatrics. 16th ed. Philadelphia, PA: WB Saunders Co; 2000.
R. A Guide to Physical Examination and History Taking. 6th ed. Philadelphia, PA: JB Lippincott Co; 1995.
H. Atlas of Pediatric Physical Diagnosis. 4th ed. St Louis, MO: Mosby; 2002.
D. The assessment of dehydration in congenital pyloric stenosis. N Z Med J.
of clinical signs in the diagnosis of dehydration in children. Pediatrics.
et al.. Interrater agreement in the assessment of dehydration in infants. J Trop Pediatr.
F. Clinical signs of dehydration in children. Lancet.
K. Signs of dehydration in severe childhood malaria
. Trop Doct.
E, Diaz Cruz
G. Clinical and biochemical evaluation of the degree of dehydration in children with acute diarrhea [in Spanish]. Bol Med Hosp Infant Mex. 1974;31(3):561–576.
JR. A prospective study
of the usefulness of clinical and laboratory parameters for predicting percentage of dehydration in children. Pediatr Emerg Care.
Suharjono. Blood urea nitrogen (B.U. N.) in gastroenteritis with dehydration. Paediatr Indones. 1980;20(3-4):77–82.
LG. Laboratory predictors of fluid deficit in acutely dehydrated children. Clin Pediatr (Phila).
N. Evaluation of laboratory tests in dehydrated children with acute gastroenteritis. J Paediatr Child Health.
DB. Likelihood ratios with confidence. J Clin Epidemiol.
World Health Organization. A Manual for the Treatment of Diarrhoea. Geneva, Switzerland: World Health Organization Program for Control of Diarrhoeal Diseases; 1990.