Psychiatric Quarterly

, Volume 85, Issue 1, pp 111–120

Elevated Blood Urea Nitrogen and Medical Outcome of Psychiatric Inpatients

Authors

    • Zucker Hillside HospitalNorth Shore – Long Island Jewish Health System
    • Hofstra North Shore – Long Island Jewish School of Medicine at Hofstra University
  • Zainab Al-Dhaher
    • Zucker Hillside HospitalNorth Shore – Long Island Jewish Health System
  • Sameer Khan
    • Zucker Hillside HospitalNorth Shore – Long Island Jewish Health System
  • John M. Kane
    • Zucker Hillside HospitalNorth Shore – Long Island Jewish Health System
    • Hofstra North Shore – Long Island Jewish School of Medicine at Hofstra University
  • Christoph U. Correll
    • Zucker Hillside HospitalNorth Shore – Long Island Jewish Health System
    • Hofstra North Shore – Long Island Jewish School of Medicine at Hofstra University
Original Paper

DOI: 10.1007/s11126-013-9274-2

Cite this article as:
Manu, P., Al-Dhaher, Z., Khan, S. et al. Psychiatr Q (2014) 85: 111. doi:10.1007/s11126-013-9274-2

Abstract

Elevated blood urea nitrogen (BUN) is associated with increased severity of illness and mortality, but its predictive value has not been studied in patients admitted to free-standing psychiatric hospitals. To determine the clinical outcome of psychiatric inpatients with elevated BUN on admission and to create a quantitative method of using BUN for predicting deteriorations requiring transfers of psychiatric inpatients to a general hospital we conducted a retrospective cohort study of 939 adults consecutively admitted to a free-standing psychiatric hospital in 2010. Transfer to a general hospital was used as a proxy marker for poor medical outcome. The score Age (years) plus BUN (mg/dL) was used in sensitivity analyses to identify patients with medical deterioration in derivation (N = 523) and validation (N = 414) samples. Fifty-two (5.5 %) patients had admission azotemia (BUN >25 mg/dL). Medical deteriorations requiring emergency transfer to a general hospital occurred in 24 (46.2 %; 95 % confidence interval = 32.6–49.8 %) of azotemic patients and 112 (12.6 %; 95 % confidence interval = 10.4–14.8 %) of those with normal BUN (p < 0.0001). Age + BUN ≥90 identified 51 transferred patients and had positive and negative predictive values of 39.8 and 89.5 %, respectively, in the entire sample. We conclude that psychiatric inpatients with BUN >25 mg/dL or Age + BUN ≥90 are at risk for medical deterioration. Free-standing psychiatric hospitals should develop models of care requiring frequent, scheduled medical follow-up and enhanced monitoring for this vulnerable populations.

Keywords

Medical outcomePsychiatric inpatientsBlood urea nitrogenAgePredictive value

Introduction

Higher levels of blood urea nitrogen (BUN) correlate with the hospital outcome of a large number of medical conditions, such as acute coronary syndromes [1], congestive heart failure [2, 3], acute dyspnea [4, 5], community-acquired pneumonia [68], acute viral hepatitis A [9], spontaneous bacterial peritonitis [10], acute intracerebral hemorrhage [11], and the development of bacteremia among febrile adults [12]. The laboratory abnormality is also predictive of hospital severity of surgical conditions or procedures, including acute pancreatitis [1315] and infected pancreatic necrosis [16], pyogenic liver abscesses [17], hemorrhagic episodes in patients with coronary artery disease [18] and esophagectomy [19]. Among critically ill patients [20, 21], as well as in patients with generally favorable prognosis, such as acute coronary syndrome, an elevated BUN is associated with mortality independent of creatinine-derived estimates of the glomerular filtration rates [1].

The value of azotemia for risk stratification of the medical outcome of psychiatric patients has not been the object of a systematic evaluation, but instruments have been developed to predict the severity of hospital course for a large number of medical and surgical conditions. As a general rule, these tools use the predictive abilities of demographic, clinical and laboratory characteristics recorded on admission [22] and contain specific and nonspecific items. The common denominators of the nonspecific items are age and the admission level of BUN. Some of these tools are quite complex and require individual weighting for age and BUN [18], while others [5] use simple additions of age (in years) and BUN (in mg/dL) to create the predictive score. For clinical application, a prediction tool for the identification of psychiatric patients at high risk for medical deterioration must be simple to use, based on routinely collected data, and have a high precision rate.

In the United States, BUN values are almost always available at the time of psychiatric admission because patients are screened with a basic metabolic panel for the presence of electrolyte imbalance and renal dysfunction, which may lead to serious complications during treatment with psychotropic drugs. The goals of this study were to evaluate that characteristics of patients with admission azotemia and to test the score created by the sum of Age + BUN as a predictor of medical deteriorations of psychiatric inpatients admitted to a free-standing hospital.

Methods

Setting and Patients

This is a secondary analysis of data generated by 1,000 adult patients consecutively admitted to a 208-bed free-standing teaching hospital located in New York city from August through December 2010. The data retrieval was performed according to a protocol approved by the Institutional Review Board, North Shore – Long Island Jewish Health System, Manhasset, NY, USA.

Hospital policy requires that all consenting patients receive a medical evaluation on the day of admission. The evaluation includes psychiatric and medical history, physical examination and laboratory testing (i.e., at least a basic metabolic panel, complete blood count and measurement of electrolytes, BUN, creatinine and glucose). Medical consultations are provided by on-site, salaried, board-certified physicians. The patients with medical conditions who cannot be managed in the psychiatric setting are transferred to the general hospital.

Data Collection

Data extracted for this report included psychiatric and medical diagnoses, medications prescribed on admission, results of mandatory admission laboratory testing, length of stay and reason for transfer to the general hospital.

Identification of Azotemic and Non-Azotemic Patients

All BUN levels were measured with the standard enzymatic conductivity rate method in the same laboratory. Azotemia was defined as a plasma BUN level of 25 mg/dL or greater. Two-tailed analyses of variance and Chi square tests were used to compare continuous and categorical variables, respectively, in patients with normal and above normal BUN and for azotemic patients transferred and non-transferred to the general hospital.

Age + BUN as Predictor of Transfers

The first step was a univariate comparison of age, gender, psychiatric diagnoses, number of active somatic disorders, BUN, serum creatinine, BUN/creatinine ratio and Age + BUN in the transferred and non-transferred patients, which was followed by logistic regression to determine whether the score of Age + BUN was independently correlated with the primary outcome. In a second step, the means of the sums Age + BUN in the transferred and non-transferred patients were used to guide the assessment of the frequency distribution of medical deteriorations by Age + BUN score in order to identify the category of patients most likely to suffer such deterioration.

In the third step, the alphabetical list of the cohort was divided into derivation and validation groups comprising a similar number of transferred patients. We determined the test characteristics (sensitivity, specificity, positive and negative predictive values) of the high risk score in the entire sample. We then performed sensitivity analyses of these characteristics in the validation group and applied them to the derivation group. Sensitivity represented the proportion of patients from the transferred group who met or exceeded the cut-off, specificity indicated the proportion of those below the cut-off who were not transferred, the positive predictive value was defined as the proportion of true positives, the percentage of patients transferred from among those with a score above a given cut-off, and the negative predictive value was defined as the proportion of patients with scores below a given cut-off who did not require a transfer.

Results

Incidence of Admission Azotemia (BUN 25 mg/dL or Greater)

Of the 1,000 psychiatric inpatients, 939 consented to blood draw for routine laboratory work on the day of admission. Fifty-two patients (5.6 %) had plasma BUN levels of 25 mg/dL or greater.

Clinical Characteristics of Psychiatric Inpatients with Admission Azotemia

Compared with non-azotemic patients, those with elevated BUN on admission were older (p < 0.0001) and more likely to be diagnosed with dementia (Table 1).
Table 1

Demographic and psychiatric characteristics

Characteristic

Patients with BUN ≥25 mg/dL (N = 52)

Patients with normal BUN (N = 887)

p value

Age (years SD)

66.5 ± 16.9

44.1 ± 19.0

<0.0001

Gender (males N, %)

27 (51.9 %)

472 (53.2 %)

0.86

Psychiatric diagnoses (N, %)

 Psychotic disorders

   

 Schizophrenia or schizoaffective disorder

13 (25.0 %)

252 (28.4 %)

0.59

 Psychosis NOS

3 (5.8 %)

73 (8.2 %)

0.52

Affective disorders

 Major depressive disorder or depression NOS

9 (17.3 %)

282 (31.8 %)

0.028

 Bipolar disorder

13 (25.0 %)

162 (18.3 %)

0.23

 Anxiety disorders

2 (3.9 %)

55 (6.2 %)

0.49

 Substance use disorder

7 (13.5 %)

213 (24.0 %)

0.08

 Dementia

18 (34.6 %)

50 (5.6 %)

<0.0001

Bolded values are statistically significant after correcting for multiple comparisons

Azotemic patients were also more likely to have arterial hypertension (p < 0.0001), dyslipidemia (p < 0.0001), stroke or transient ischemic attacks (p < 0.0001), diabetes mellitus (p < 0.0001), and coronary artery disease (p = 0.0004) (Table 2). The mean number of active somatic disorders was significantly greater in the azotemic group (3.9 ± 1.8 vs. 2.8 ± 1.9, p < 0.0001). A total of 62 patients were receiving diuretics at the time of admissions; 27 patients were treated with furosemide and 35 with hydrochlorothiazide. The proportion of patients taking one of these two drugs was significantly higher in the azotemic group (26.9 vs. 5.4 %, p < 0.0001). None of the azotemic patients were treated with lithium, which was among the medications taken by 69 (7.8 %) of the patients with normal BUN (p = 0.036).
Table 2

Medical history

Characteristic

Patients with BUN ≥25 mg/dL (N = 52)

Patients with normal BUN (N = 887)

p value

Arterial hypertension

39 (75.0 %)

235 (26.5 %)

<0.0001

Dyslipidemia

22 (42.3 %)

169 (19.05 %)

<0.0001

Coronary artery disease

8 (15.4 %)

39 (4.4 %)

0.0004

Congestive heart failure

3 (5.8 %)

15 (1.7 %)

0.037

Diabetes mellitus

17 (32.7 %)

100 (11.3 %)

<0.0001

Cerebrovascular accident

4 (7.7 %)

9 (1.0 %)

<0.0001

Hypothyroidism

7 (13.5 %)

52 (5.9 %)

0.028

Asthma/chronic obstructive pulmonary disease

6 (11.5 %)

85 (9.6 %)

0.64

Parkinson’s disease

1 (1.9 %)

7 (0.8 %)

0.39

Bolded values are statistically significant after correcting for multiple comparisons

Compared with patients with normal BUN on admission, a significantly higher number of azotemic patients had a BUN/creatinine ratio <20, suggestive of pre-renal disorder, rather than intrinsic renal disease (Table 3).
Table 3

Admission laboratory data

Characteristic

Patients with BUN ≥25 mg/dL (N = 52)

Patients with normal BUN (N = 887)

p value

Renal function

 Blood urea nitrogen (mg/dL)

34.2 ± 13.3

13.6 ± 4.5

<0.001

 Creatinine (mg/dL)

1.5 ± 0.72

0.89 ± 0.21

<0.0001

 BUN/Cr >20 (N, %)

38 (73.1 %)

160 (18.1 %)

<0.0001

Electrolytes (meq/L)

 Sodium

138.9 ± 3.2

139.4 ± 2.8

0.23

 Potassium

3.99 ± 0.5

3.97 ± 0.4

0.72

 Chloride

102.8 ± 4.0

102.9 ± 3.3

0.83

 Carbon dioxide

24.5 ± 3.1

25.3 ± 2.8

0.031

 Calcium, total

9.3 ± 0.6

9.3 ± 0.5

0.93

Glucose, random (mg/dL)

123.4 ± 52.5

101.0 ± 38.5

<0.0001

Bolded values are statistically significant after correcting for multiple comparisons

Medical Outcome of Patients with Admission Azotemia

Medical deteriorations requiring transfer from the free-standing psychiatric hospital to the general hospital were significantly more common in the azotemia group (46.2 vs. 12.7 %, p < 0.0001). Ten (41.2 %) transferred azotemic patients had a medical deterioration due to a febrile illness. Fall/head trauma, pulmonary/cardiovascular instability and acute neurological changes were each the reason for transfers in 3 (12.5 %) patients. Renal insufficiency was the specified reason for transfer in only two (3.8 %) of the transferred azotemia patients.

The 24 azotemia patients transferred for medical deterioration and the 28 patients who remained medically stable had similar demographic and clinical characteristics.

Distribution of Transfers According to Age + BUN Score

The means ± standard deviations of Age + BUN scores were 76.1 ± 26.9 in the transferred group and 57.4 ± 21.5 in the non-transferred patients (p < 0.0001). Given these values, patients were grouped according to scores less than 50, 50–74, 75–99, and 100 or greater (Fig. 1). Among patients with Age + BUN ≥100, the proportion of patients transferred was 44.3 %, which was significantly higher than in groups with scores of 75–99 (22.8 %, p = 0.001), 50–74 (11.4 %, p < 0.0001), and <50 (8.0 %, p < 0.0001), respectively (Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs11126-013-9274-2/MediaObjects/11126_2013_9274_Fig1_HTML.gif
Fig. 1

Distribution of transferred and not transferred patients by Age + BUN score

Predictive Value of the Age + BUN Score in the Derivation and Validation Sample

The derivation and validation groups used to determine the predictive value of the Age + BUN ≥100 score for medical transfer were similar with regard to all of the demographic, clinical and laboratory features (Table 4).
Table 4

Characteristics of derivation and validation samples

Characteristic

Derivation sample (N = 523)

Validation sample (N = 414)

 

Patients not transferred (N = 455)

Patients transferred (N = 68)

Patients not transferred (N = 346)

Patients transferred (N = 68)

Age (years ± SD)

42.9 ± 18.5

61.2 ± 20.2*

43.8 ± 18.4

54.5 ± 22.6*

Gender, male (N, %)

253 (55.6)

27 (39.7)

183 (52.9)

35 (51.5)

Psychiatric diagnosis (N, %)a

 Schizophrenia/psychosis NOS

168 (36.9)

19 (27.9)

115 (33.2)

32 (49.1)

 Bipolar disorder

83 (18.2)

15 (22.1)

67 (19.4)

10 (14.7)

 Major depression/depression NOS

145 (31.9)

22 (32.3)

109 (31.5)

14 (20.6)

 Anxiety disorders

35 (7.7)

3 (4.4)

17 (4.9)

2 (2.9)

 Substance use disorders

114 (25.1)

11 (16.8)

81 (23.1)

14 (20.6)

 Dementia

23 (5.1)

17 (25.0)*

16 (4.1)

12 (17.6)*

Somatic disorders (N ± SD)

2.7 ± 1.9

3.6 ± 2.1**

2.9 ± 1.8

3.7 ± 2.2**

BUN (mg/dL ± SD)

14.0 ± 6.2

18.1 ± 8.8*

14.2 ± 7.2

18.4 ± 9.2*

Creatinine (mg/dL ± SD)

0.9 ± 0.3

1.0 ± 0.3****

0.9 ± 0.3

1.0 ± 0.4***

BUN/Creatinine (ratio ± SD)

15.6 ± 5.4

18.2 ± 8.8**

16.2 ± 7.4

18.3 ± 6.2****

Age + BUN (sum ± SD)

56.9 ± 21.5

79.3 ± 24.9*

57.9 ± 21.4

72.9 ± 28.5*

NOS not otherwise specified, SD standard deviation, BUN blood urea nitrogen

p < 0.0001; ** p < 0.0005; *** p < 0.01; **** p < 0.05 for difference between transferred and not transferred groups

asome patients had more than one admission psychiatric diagnosis

Sensitivity analyses performed in the derivation and validation samples for Age + BUN scores of 80, 90, 100 and 110 indicated that the score ≥100 had the highest positive predictive values, but identified, overall, only 31 transferred patients. A score ≥90 identified 51 transferred patients. The negative predictive value of these two scores was essentially similar (Table 5).
Table 5

Test characteristics and sensitivity analysis of different Age + BUN score thresholds

Test characteristics

Derivation sample (N = 523)

Validation sample (N = 414)

Total (N = 937)

Age + BUN ≥80

 Sensitivity (true positives)

40/68 = 58.8 %

25/68 = 36.8 %

65/136 = 47.8 %

 Specificity (true negatives)

385/455 = 84.6 %

295/346 = 85.3 %

680/801 = 84.9 %

 Positive predictive value

40/110 = 36.6 %

25/76 = 32.9 %

65/186 = 35.0 %

 Negative predictive value

385/413 = 93.2

295/338 = 87.3 %

680/751 = 90.6 %

Age + BUN ≥90

 Sensitivity (true positives)

27/68 = 39.7 %

24/68 = 35.3 %

51/136 = 37.5 %

 Specificity (true negatives)

409/455 = 89.9 %

315/364 = 91.0 %

724/801 = 90.4 %

 Positive predictive value

27/73 = 36.9 %

24/55 = 43.6

51/128 = 39.8 %

 Negative predictive value

409/450 = 90.1

315/359 = 87.7

724/809 = 89.5 %

Age + BUN ≥100

 Sensitivity (true positives)

15/68 = 22.1 %

16/68 = 23.5 %

31/136 = 22.8 %

 Specificity (true negatives)

434/455 = 95.4 %

328/346 = 94.8 %

762/801 = 95.1 %

 Positive predictive value

15/36 = 41.7 %

16/34 = 47.1 %

31/70 = 44.3 %

 Negative predictive value

434/487 = 89.1 %

328/380 = 86.3 %

762/867 = 87.9 %

Age + BUN ≥110

 Sensitivity (true positives)

6/68 = 8.8 %

7/68 = 10.3 %

13/136 = 9.6 %

 Specificity (true negatives)

445/455 = 97.8 %

338/346 = 97.7 %

783/801 = 97.8 %

 Positive predictive value

6/16 = 37.5 %

7/15 = 46.7 %

13/31 = 41.9 %

 Negative predictive value

445/507 = 87.8 %

338/399 = 84.7 %

783/906 = 86.4 %

Discussion

Free-standing psychiatric facilities provide the bulk of hospital care for patients with severe mental illnesses and their scope of practice is regulated by state and local policies [23], but the manpower and equipment used in these institutions to address medical issues are highly variable [24]. Most psychiatric hospitals employ consultants to evaluate and manage outpatient-level conditions, but do not provide inpatient-level medical or surgical services [25]. Patients suffering significant medical deteriorations are usually transferred to a nearby general hospital [25, 26], a process that is highly distressing for most psychiatric patients, particularly those with cognitive deficits or psychotic symptoms. Furthermore, medical deteriorations may interrupt psychopharmacological and behavioral interventions and prolong the length of stay.

BUN has significant prognostic value across the clinical spectrum that is not seen with other measures of renal function [27]. The observed correlation is not due to a direct toxic effect of urea nitrogen, but to the fact that elevated BUN is a biomarker of enhanced production of proinflammatory cytokines [1, 4, 13, 28]; increased protein catabolic rate, negative nitrogen balance, sarcopenia [20], nonosmotic release of arginine vasopressin, and activation of the renin-angiotensin-aldosterone system [27], with the potential to increase vulnerability to infection, decrease cardiopulmonary performance, and produce muscle weakness and falls.

Overall, BUN levels in the azotemic range were present on admission in 52 patients and identified 24 transfers from the free standing hospital to the general hospital. The addition of age increased substantially the identification of transferred patients. For example, an Age + BUN score ≥80 identified 65 patients, while a score of score ≥90 predicted correctly the hospital outcome of 51 patients. Advancing age correlates with biological frailty, a state that reflects the concomitant presence of malnutrition, decreased endurance and muscle strength, impaired balance and gait, and cognitive and functional decline [29]. The importance of age as a predictor of severity of illness is related to its association with “geriatric syndromes”, which include falls, delirium, and syncope [30, 31]. The frailty phenotype includes lower income, poorer health and high rates of comorbid chronic disease and disability, all of which are common in older psychiatric inpatients [32].

Although the independent predicting value of the BUN + Age score was determined by multivariate logistic regression analyses and the threshold score was determined in a derivation sample and validated in a replication sample, our study has the limitations inherent to the fact that the data were collected in only 937 patients and in a single hospital, which may have different admission restrictions and medical monitoring standards than other free-standing psychiatric institutions in the United States and elsewhere.

The findings indicate that the clinical utility of Age + BUN as a predictive instrument requires a thorough understanding of its costs and benefits. In our sample, the benefit of using a score ≥80 is the identification of 65 patients likely to deteriorate, while the potential cost is enhancing care for a total of 186 patients. With a score ≥90, care will be enhanced for 128 patients, of whom 51 would be truly at risk. At the present time, medical consultations in most self-standing psychiatric hospitals occur as acute, unscheduled visits. Assuming that the score will be used toward selecting patients for a model of inpatient care based on frequent, scheduled interdisciplinary assessments, expanded medical expertise of the psychiatric nursing staff, and increased availability of bedside consultations with medical subspecialists, decisions should take into consideration the resources used for the care patients who end up being false positives. Future studies should explore other patient samples and settings in order to assess generalizability and utility of the proposed prediction tool.

Disclosures

Drs. Manu, Al-Dhaher and Khan have nothing to disclose. Dr. Kane has been a consultant to or has received honoraria from Astra-Zeneca, Bristol-Myers Squibb, Cephalon, Eli Lilly, Janssen Pharmaceutica, Johnson and Johnson, Lundbeck, Otsuka, Pfizer Inc, PgXHealth, Proteus, Vanda and Wyeth, has served on the speaker’s bureau of AstraZeneca, Bristol-Myers Squibb/Otsuka and Eli Lilly, and is a share holder of MedAvante. Dr. Correll has been a consultant and/or advisor to or has received honoraria from: Actelion, Alexza; Bristol-Myers Squibb, Cephalon, Eli Lilly, Genentech, Gerson Lehrman Group, IntraCellular Therapies, Lundbeck, Medavante, Medscape, Merck, Janssen/J&J, Otsuka, Pfizer, ProPhase, Roche, Sunovion, Takeda, Teva, and Vanda. He has received grant support from BMS, Janssen/J&J, and Otsuka.

Copyright information

© Springer Science+Business Media New York 2013