|Home > RNJ > 2009 > September/October > Postadmission Dehydration: Risk Factors, Indicators, and Outcomes|
Postadmission Dehydration: Risk Factors, Indicators, and Outcomes
Detecting and treating dehydration in hospitalized patients is critical because of the adverse outcomes associated with this condition. Using a case-control design, this study estimated the incidence, risk factors, and outcomes of dehydration in hospitalized adults. The overall incidence rate for developing one of three ICD-9 codes for dehydration during a hospital stay was 3.5%. Cases and controls differed significantly on a number of clinical variables on admission; a large percentage of patients may have had dehydration on admission to the hospital. Mortality rates at 30 and 180 days postdischarge were significantly higher when dehydration was present. Patients may be discharged to rehabilitation settings in a dehydrated state, which prolongs recovery. Despite the increased risk for dehydration and higher rates of hospitalization in older populations, little systematic research has addressed the risk factors for and indicators of dehydration in hospitalized patients.
Dehydration is a common water and electrolyte disorder among elderly patients. In 1991, 6.7% (731,695) of all Medicare patients admitted to hospitals had dehydration as one of five diagnoses (Warren et al., 1994). Recent estimates place the potential economic savings from avoiding dehydration admissions at more than $1 billion (Xiao, Barber, & Campbell, 2004). People age 65 and older have more hospital stays than any other age group and are at high risk for dehydration.
Detecting and treating dehydration in hospitalized patients is critical because of the adverse outcomes associated with dehydration, such as delirium, falls, incontinence, and mortality (Weinberg & Minaker, 1995). Knowledge of admission risk factors for the subsequent development of dehydration will help focus efforts on patients at highest risk. Most work to date has focused on dehydration in long-term-care settings (Culp et al., 2004; Kayser-Jones, 2006; Kayser-Jones, Schell, Porter, Barbaccia, & Shaw, 1999; Mentes, 2006; Mentes & Culp, 2003). The purpose of this study was to describe incidence estimates and patient level factors (predisposing factors, signs and symptoms, and mortality) for patients who developed one of three diagnoses of three dehydration codes: 276.0 hyperosmolality and/or hypernatremia, 276.1 hypoosmolality and/or hyponatremia, and 276.5 volume depletion (Weinberger et al., 1995) after hospital admission.
Older people, especially those with multiple chronic illnesses, are at high risk for dehydration because of the combined effects of disease and normal age-related changes. Age-related changes predisposing elders to dehydration include a decrease in total body water, decreased kidney function, and an increased risk for functional limitations (Feinsod et al., 2004). However, dehydration can be difficult to diagnose (Robinson & Weber, 2004; Thomas, Tariq, Makhdomm, Haddad, & Moinuddin, 2004) because medication side effects and symptoms associated with acute illnesses such as infection may cause symptoms similar to dehydration (Feinsod et al.). Clinicians often use the terms dehydration and volume depletion interchangeably, and nurses tend to describe these terms as fluid volume deficit. In a study of 102 hospitalized adults older than age 65 with a clinical diagnosis of dehydration, only 17% had a serum osmolality higher than 295, and 11% had a serum sodium higher than 145. A BUN/creatinine ratio higher than 20 was present in 68% of the subjects reviewed. Furthermore, at least one-third of the diagnoses of intravascular volume depletion were incorrect based on lab data, suggesting physicians rely more on physical signs rather than laboratory data (Thomas et al.). To date, no single measure has been identified to clearly discriminate between people who are and are not dehydrated (Stookey, Pieper, & Cohen, 2005).
During the past decade, nurses have increasingly focused on measuring quality of care. Efforts include the development of standardized languages to define nursing practice (Dochterman & Bulechek, 2003; Moorhead, Johnson, & Maas, 2003) and development and use of large databases to analyze nursing care (Aydin et al., 2004; Doran et al., 2006). A relationship between nurse staffing in hospitals and patient outcomes has been demonstrated in several studies (Haberfelde, Bedecarre, & Buffum, 2005). Typically, the outcomes assessed in these studies include patient falls, pressure ulcers, medication errors/adverse drug events, postoperative and urinary tract infections, pneumonia, wound infections, and patient satisfaction. Development of dehydration after hospital admission may be a measure of hospital care quality, but almost no data exist describing the incidence, risk factors, and outcomes of dehydration in hospitalized adults.
The study employed a case-control design. In this design, people with the disease or condition of interest (cases) and those without the disease or condition (controls) are selected. The proportions of subjects who have risk factors and outcomes of interest in the two groups are determined and compared (Kelsey, Thompson, & Evans, 1986).
Subjects and Setting
Before data collection, the study was approved by the University of Iowa Institutional Review Board (IRB), which serves as the IRB for the Iowa City VA Medical Center (VAMC). The Iowa City VAMC provides outpatient and inpatient medical, surgical, psychiatric, and neurological care to more than 36,000 veterans residing in eastern Iowa and western Illinois. The Medical Center serves as a referral center for several specialty services.
Medical records of patients admitted to the Iowa City VAMC between calendar years 1997 and 2000 were eligible for inclusion. The VA Patient Treatment File (PTF) was used to generate a list of records. The PTF is a VA database containing records of all hospital admissions to VA facilities. Cases were defined as patients with one of three ICD-9-CM codes representing dehydration diagnoses treated during the index hospitalization (but not the reason for admission). The three principal diagnosis ICD-9-CM codes were 276.0 hyperosmolality and/or hypernatremia, 276.1 hypoosmolality and/or hyponatremia, and 276.5 volume depletion (Weinberg & Minaker, 1995).
In hypernatremic dehydration (ICD code 276.0), water losses are greater than sodium losses. Characteristics include hypernatremia (serum sodium levels >145 mmol/L) and hyperosmolality (serum osmolality >300 mmol/kg). This type of dehydration commonly is attributable to fever and inability to increase oral intake. In hypotonic/hyponatremia (ICD code 276.1), sodium loss exceeds water loss reflected by serum sodium levels <135 mmol/L and hyperosmolality (serum osmolality <280 mmol/kg). This type of dehydration may result from overuse of diuretics, causing excess sodium loss, for example, in patients with heart failure. The third type of dehydration, hypovolemia or volume depletion (ICD code 276.5), results from balanced loss of water and sodium. Hypovolemia commonly is associated with a complete fast or prolonged vomiting and diarrhea. Combinations of these three types of dehydration also can occur (e.g., continuing diuretics while a person has diarrhea).
Control patient records were randomly selected from all admissions without one of these three ICD-9-CM codes listed (either at admission or during hospital stay). Control patient records were matched to cases using three variables: age within 5 years, ward location, and admission month (within the same year).
Two abstractors, one physician, and one clinical pharmacist reviewed all study records using an investigator-developed chart abstraction tool that included 77 items reflecting demographic data, predisposing factors, and signs and symptoms of dehydration (Figure 1). Items on the chart abstraction tool were based on a conceptual model (Figure 2) derived from a review of literature (Mentes et al., 2000) and clinical knowledge of dehydration.
The time period selected for the record review overlapped with the initial implementation of the electronic medical record in the VA; consequently, both paper and electronic records were used in the review. Reviewers had access to the complete medical record, but the data for this study focused on information available for up to 72 hours before admission (data recorded in the admission history) and for the first 24 hours after the index admission to the hospital. In the case of multiple data points (e.g., multiple laboratory values in the first 24 hours after admission), the data point closest to hospital admission time was selected.
After initial training on the abstraction tool, interrater reliability between the two reviewers was established by having each abstractor individually review 10 records. These ratings were compared and each item was discussed (e.g., where to find the data when data were recorded multiple times and the clarified meaning of each item). The abstractors then reviewed 50 records and the investigators again met to discuss ratings and modify the instrument to improve clarity. Periodic interrater reliability checks were conducted throughout data collection.
After all data were collected, the data set was cleaned to check for out-of-range values and missing data. For the laboratory data, ranges were reviewed by a medical technologist and pathologist. Out-of-range values were corrected by reexamining the medical record. To the extent they were available, missing data were identified by a re-review of the record. Data for deaths were obtained by merging the data set with data files at the VA Austin Automation Center using the Beneficiary Identification Records Locator Subsystem, a Veterans Benefits Administration database containing records of all beneficiaries, including veterans whose survivors applied for death benefits.
All data were analyzed using Statistical Analysis Software Version 9.1. Student’s t test was used to compare continuous variables (and Levine’s test for unequal variances where appropriate); Pearson’s chi-square was used to compare categorical variables. Records were grouped under each ICD-9 code and each group was compared separately to the control group data. Because the medical records did not contain complete information for every variable of interest, variables for which data were missing for more than 25% of subjects were eliminated. For example, the amount of urine output in the 72 hours before admission generally was not available. Consequently, results are reported for a subset of variables from the record review.
During the 4-year study time frame, there were 15,146 hospital admissions. During this time, 533 patients were treated for one of three study ICD-9-CM codes during the hospital admission, resulting in an overall postadmission incidence rate of 3.5%. Among the 533 patients, data were obtained for 355 cases (67%). Records not reviewed included those that had been transferred elsewhere, could not be located, or were restricted. Data were obtained on 334 control patients. The majority of the patients (79%) were cared for on medicine units, with 11% on surgical units, 3% in intensive care, 3% in psychiatry, and 3% in the palliative care unit (unit location was missing for 4 subjects). Of the 355 cases reviewed, 5 (1.4%) of the patients had ICD code 276, 110 (31%) had ICD code 276.1, 224 (63.1%) had ICD code 276.5, and 16 (4.5%) had both ICD codes 276.5 and 276.1. The analyses presented in this paper focus on patients with codes 276.1 and 276.5 (patients with only code 276.0 and those with multiple codes were excluded from further analysis).
Overall the mean age of the sample was 63.6 years, and, consistent with the population, 99% of patients were men. Demographic data for the three analyzed groups (controls and subjects with admission codes 276.1 and 276.5) are shown in Table 1. Reflecting the demographics of the population treated at this hospital, most subjects were Caucasian. Most patients in the control group were admitted for the major diagnostic categories of circulatory system (31%), respiratory (15%), and neoplasms (10%). In contrast, the cases were admitted for neoplasms (18%), respiratory (17%), and circulatory (13%) problems (Table 2).
Hyponatremic Group (ICD code 276.1)
Compared to control patients, patients with code 276.1 had significantly lower body mass index (BMI; 26.2 compared to 28 in controls), sodium (126.1 mEq/L compared to 136.6 mEq/L in controls), chloride (89.9 mmol/L compared to 98.6 mmol/L in controls), calcium (8.6 compared to 8.8 in controls), plasma osmo (258.0 compared to 279.8 in controls), urine-specific gravity (1.012 compared to 1.016 in controls), and significantly higher pulse (93 compared to 83 in controls) and potassium (4.4 compared to 4.1 in controls; Table 3). The cases were significantly less likely to be independent in activity (36.5% compared to 51.1% in controls), but more likely to have generalized weakness, paraplegia, or hemiplegia (21.5% compared to 11.4% in controls); be incontinent (22.6% compared to 12.6 % in controls); be NPO (19.2% compared to 10.5% in controls); and be taking potassium-sparing diuretics (14.3% compared to 3.4% of controls) and hyperosmotic laxatives (13.2% compared to 3.1%) before admission (Table 4). The mean sodium and chloride values in the cases were below-normal values (using sodium 135–145 mEq/L and chloride 95–107 mmol/L as normal ranges). Mortality rates at 30 and 180 days after discharge were significantly higher than control patients (Table 5).
Volume Depletion Group (ICD â€¨Code 276.5)
Compared to control patients, patients with code 276.5 had significantly lower BMI (26.2 compared to 28.0 in controls), systolic blood pressure (129 compared to 134 in controls), and sodium (134.6 mEq/L compared to 136.6 mEq/L in controls). Cases had a significantly higher pulse rate (88 compared to 83 in controls); potassium (4.4 compared to 4.1 in controls); chloride (96.5 mmol/L compared to 98.6 mmol/L in controls); calcium (9.1 compared to 8.8 in controls); glucose (184.8 compared to 146.2 in controls); BUN, creatinine, and BUN/creatinine ratio (20.3 compared to 18.2 in controls); and white blood cell count (10.7 compared to 9.7 in controls; Table 3). Cases were significantly more likely to have both constipation (16.4% compared to 8.3%) and diarrhea (23.5% compared to 5.6% in controls); generalized weakness, paraplegia, or hemiplegia (33.7% compared to 11.6% in controls); need help eating (13.4% compared to 6.6%); be NPO prior to admission (37.3% compared to 10.5% in controls); have either increased or decreased urine output (18.9% compared to 9.9% in controls); and vomiting (29.1% compared to 7% in controls). Cases also were more likely to be taking hyperosmotic laxatives (7.2% compared to 3.1% in controls) and stool softeners (33.5% compared to 23.0% in controls) at time of admission to the hospital (Table 4). Cases were less likely to be independent in activity (37% compared to 51.1% in controls) or have edema (22.6% compared to 39.5% in controls; Table 3). Mortality rates at 30 and 180 days after discharge were similar to patients with code 276.1 and significantly higher than control patients (Table 5).
Patients with code 276.1 differed from controls on a number of variables including BMI, pulse, sodium, potassium, chloride, calcium, BUN, plasma osmolality, urine-specific gravity, independence in activity, generalized weakness and hemiplegia, incontinence, food and fluid intake before admission, and use of potassium-sparing diuretics and hyperosmotic laxatives. Patients with code 276.5 differed from controls on a number of variables including BMI, pulse, systolic blood pressure, sodium, potassium, chloride, calcium, glucose, BUN, creatinine, BUN/creatinine ratio, white blood count, independence in activity, constipation, diarrhea, generalized weakness and hemiplegia, edema, assistance with eating, food and fluid intake before admission, urine output, vomiting, and use of hyperosmotic laxatives and stool softeners.
Mortality rates were significantly higher among patients who developed dehydration during their hospital stay. Patients with either dehydration code were almost twice as likely to die after hospital discharge.
Between 1997 and 2000, the overall incidence of having one of the three ICD codes for dehydration increased from 2.3% in 1997 to 4.6% in 2000. While the percentage increased in a linear fashion over time, the total number of hospital admissions decreased during this same period. The dehydration rate increase may reflect the VA’s efforts to shift from a system focused on inpatient care to one focused on improved management in outpatient care. This shift resulted in only the sickest patients being admitted, as opposed to a true increase in the incidence of dehydration.
In the group of cases, about 40% had one of the study ICD codes listed second on the list of problems treated during hospital stay. Codes are listed in the order of the amount of treatment provided during hospital stay. Risk factors for this group of patients are similar to those found in patients who were admitted for dehydration (Wakefield, Mentes, Holman, & Culp, 2008). It is possible that a large number of patients had dehydration on admission that may have been associated with an admitting diagnosis such as neoplasm.
There are several limitations to this study. The most salient limitation is use of existing medical records as the data source. In particular, missing data are a concern. It is unknown whether data are missing because a patient did not exhibit a characteristic, the patient did not report a symptom (e.g., a history of alcohol use), or the clinician did not recognize or record the symptom. Furthermore, it is not possible to validate the information contained in the record (Krowchuk, Moore, & Richardson, 1995). It is possible there were unrecognized cases of dehydration in the group of control patients (misclassification bias). All data were collected from records in one medical center, so findings may be biased by the admission polices and medical-practice approach in that setting. While we matched cases and controls on several variables, we did not employ a severity-of-illness measure. Finally, although our conceptual model posits that dehydration is associated with functional decline, we were unable to determine whether functional decline occurred due to the cross-sectional nature of the study design.
Implications for Rehabilitation Nurses
Despite increased dehydration risk in older populations, little systematic research has addressed the risk factors for and indicators of dehydration in hospitalized patients. In recent years, a growing body of research has addressed functional decline in hospitalized patients, but this literature tends to focus on physical and cognitive decline (Covinsky et al., 2003; Fortinsky, Covinsky, Palmer, & Landefeld, 1999; Wakefield & Holman, 2007). Iatrogenic dehydration in hospitalized patients has not been addressed previously even though it is as “nurse-sensitive” as outcomes such as postoperative and urinary tract infections, pneumonia, and wound infections. This lack of attention may be due in part to the difficulty in defining and diagnosing dehydration or the belief that it is a problem confined to long-term-care settings.
Dehydration is associated with a number of adverse outcomes such as delirium and death, as well as adverse effects on medication metabolism and homeostasis. What is not known is how many patients are discharged from hospitals with continuing dehydration in postacute settings. Patients hospitalized for problems such as hip fracture and stroke commonly are discharged to rehabilitation settings (Wells, Seabrook, Stolee, Borrie, & Knoefel, 2003) and are at high risk for the development or exacerbation of existing dehydration during a hospital stay. As noted earlier, there is no single measure that will identify dehydration (Stookey et al., 2005). It is critical that complete baseline data be collected when rehabilitation services are initiated after hospitalization because dehydration can affect rehabilitation outcomes (Mukand, Cai, Zielinski, Danish, & Berman, 2003). Hospital discharge diagnoses should be reviewed to determine if one of the three ICD-9 codes described earlier are present during hospital admission. There currently is no standardized dehydration assessment tool; recognizing dehydration depends on skilled patient assessment. The data needed to diagnose dehydration (e.g., vital signs, fluid intake and output, mental status, and medications) routinely are collected. The conceptual model in Figure 2 identifies key factors to consider in the assessment process—factors that predispose patients to the clinical indicators of dehydration. To facilitate recognition and treatment, future studies should address practical ways for nurses to conduct the assessment and package the data into a dehydration risk assessment (e.g., decision support tools embedded in electronic patient records). Future studies also need to systematically evaluate the effectiveness of nursing interventions to prevent and treat dehydration; these studies will validate dehydration as a nurse-sensitive outcome measure.
The contents are solely the responsibility of the authors and do not necessarily represent the official views of the Department of Veterans Affairs. Dr. Wakefield was supported by a Department of Veterans Affairs Health Services Research and Development Career Development Award.
About the Authors
Bonnie J. Wakefield, PhD RN, is director of Health Services Research and Development at Harry S. Truman Memorial Veterans Hospital and research associate professor at Sinclair School of Nursing, University of Missouri-Columbia. Address correspondence to her at email@example.com.
Janet Mentes, PhD RN, is an assistant professor at UCLA School of Nursing in Los Angeles, CA.
John E. Holman, MA, is a research associate at the Center for Research in Implementation of Innovative Strategies in Practice at Iowa City VA Medical Center in Iowa City, IA.
Kennith Culp, PhD RN FAAN FGSA, is a professor at the University of Iowa College of Nursing in Iowa City, IA.
Aydin, C., Bolton, L., Donaldson, N., Brown, D., Buffum, M., Elashoff, J., et al. (2004). Creating and analyzing a statewide nursing quality measurement database. Journal of Nursing Scholarship, 36(4), 371–378.
Covinsky, K. E., Palmer, R. M., Fortinsky, R. H., Counsell, S. R., Stewart, A. L., Kresevic, D., et al. (2003). Loss of independence in activities of daily living in older adults hospitalized with medical illness: Increased vulnerability with age. Journal of the American Geriatrics Society, 51, 451–458.
Culp, K. R., Wakefield, B., Dyck, M. J., Cacchione, P. Z., DeCrane, S., & Decker, S. (2004). Bioelectrical impedance analysis and other hydration parameters as risk factors for delirium in rural nursing home residents. The Journals of Gerontology, Series A, Biological Sciences & Medical Sciences, 59A(8), 813–817.
Dochterman, J., & Bulechek, G. (2003). Nursing interventions classification. St. Louis: Mosby.
Doran, D. M., Harrison, M. B., Laschinger, H. S., Hirdes, J. P., Rukholm, E., Sidani, S., et al. (2006). Nursing-sensitive outcomes data collection in acute care and long-term care settings. Nursing Research, 55(Suppl. 2), S75–S81.
Feinsod, F. M., Levenson, S. A., Rapp, K., Rapp, M. P., Beechinor, E., & Liebmann, L. (2004). Dehydration in frail, older residents in long-term care facilities. Journal of the American Medical Directors Association, 5(Suppl. 2), S35–S41.
Fortinsky, R. H., Covinsky, K. E., Palmer, R. M., & Landefeld, C. S. (1999). Effects of functional status changes before and during hospitalization on nursing home admission of older adults. The Journals of Gerontology, Series A, Biological Sciences & Medical Sciences, 54(10), M521–M526.
Haberfelde, M., Bedecarre, D., & Buffum, M. (2005). Nurse-sensitive patient outcomes. Journal of Nursing Administration, 35(6), 293–299.
Kayser-Jones, J. (2006). Preventable causes of dehydration: Nursing home residents are especially vulnerable. American Journal of Nursing, 106(6), 45.
Kayser-Jones, J., Schell, E. S., Porter, C., Barbaccia, J. C., & Shaw, H. (1999). Factors contributing to dehydration in nursing homes: Inadequate staffing and lack of professional supervision. Journal of the American Geriatrics Society, 47(10), 1187–1194.
Kelsey, J., Thompson, D., & Evans, A. (1986). Methods in observational epidemiology. New York: Oxford University Press.
Krowchuk, H., Moore, M. L., & Richardson, L. (1995). Using health care records as sources of data for research. Journal of Nursing Measurement, 3(1), 3–12.
Mentes, J. C. (2006). A typology of oral hydration problems exhibited by frail nursing home residents. Journal of Gerontological Nursing, 32(1), 13–19; quiz 20–21.
Mentes, J. C., & Culp, K. (2003). Reducing hydration-linked events in nursing home residents. Clinical Nursing Research, 12(3), 210–225; discussion 226–228.
Mentes, J. C., Rapp, C. G., Wakefield, B., Culp, K., Tripp-Reimer, T., Mobily, P., et al. (2000). Hydration management protocol. Journal of Gerontological Nursing, 26(10), 6–15.
Moorhead, S., Johnson, M., & Maas, M. (2003). Nursing outcomes classification. St. Louis: Mosby.
Mukand, J. A., Cai, C., Zielinski, A., Danish, M., & Berman, J. (2003). The effects of dehydration on rehabilitation outcomes of elderly orthopedic patients. Archives of Physical Medicine & Rehabilitation, 84(1), 58–61.
Robinson, B. E., & Weber, H. (2004). Dehydration despite drinking: Beyond the BUN/creatinine ratio. Journal of the American Medical Directors Association, 5(Suppl. 2), S67–S71.
Stookey, J. D., Pieper, C. F., & Cohen, H. J. (2005). Is the prevalence of dehydration among community-dwelling older adults really low? Informing current debate over the fluid recommendation for adults aged 70+ years. Public Health Nutrition, 8(8), 1275–1285.
Thomas, D. R., Tariq, S. H., Makhdomm, S., Haddad, R., & Moinuddin, A. (2004). Physician misdiagnosis of dehydration in older adults. Journal of the American Medical Directors Association, 5(Suppl. 2), S30–S34.
Wakefield, B., & Holman, J. E. (2007). Functional trajectories associated with hospitalization in older adults. Western Journal of Nursing Research, 29(2), 161–182.
Wakefield, B., Mentes, J., Holman, J., & Culp, K. (2008). Risk factors and outcomes associated with hospital admission for dehydration. Rehabilitation Nursing, 33(6), 233–241.
Warren, J. L., Bacon, W. E., Harris, T., McBean, A. M., Foley, D. J., & Phillips, C. (1994). The burden and outcomes associated with dehydration among US elderly, 1991. American Journal of Public Health, 84(8), 1265–1269.
Weinberg, A. D., & Minaker, K. L. (1995). Dehydration. Evaluation and management in older adults. Journal of the American Medical Association, 274(19), 1552–1556.
Weinberger, M., Kirkman, M. S., Samsa, G. P., Shortliffe, E. A., Landsman, P. B., Cowper, P. A., et al. (1995). A nurse-coordinated intervention for primary care patients with non-insulin-dependent diabetes mellitus: Impact on glycemic control and health-related quality of life. Journal of General Internal Medicine, 10(2), 59–66.
Wells, J. L., Seabrook, J. A., Stolee, P., Borrie, M. J., & Knoefel, F. (2003). State of the art in geriatric rehabilitation. Part II: Clinical challenges. Archives of Physical Medicine and Rehabilitation, 84(6), 898–903.
Xiao, H., Barber, J., & Campbell, E. S. (2004). Economic burden of dehydration among hospitalized elderly patients. American Journal of Health-System Pharmacy, 61(23), 2534–2540.