Psychosomatics 44:382-387, October 2003
© 2003 The Academy of Psychosomatic Medicine
The Cornell Psychiatric Screen: A Brief Psychiatric Scale for Hospitalized Medical Patients
Carla Boutin-Foster, M.D., M.S.,
Stephen J. Ferrando, M.D., and
Mary E. Charlson, M.D.
Presented in part at the fourth annual meeting of the Society for Clinical Epidemiology and Health Care Research, San Diego, May 5, 2001. Received May 2, 2002; revision received Nov. 12, 2002; accepted Dec. 2, 2002. From the New York-Presbyterian Hospital Weill Medical College of Cornell University. Address reprint requests to Dr. Boutin-Foster, Division of General Internal Medicine, Weill Medical College of Cornell University, 525 E. 68th St., Box 46-Baker Tower 14, New York, NY 10021; Cboutin{at}mail.med.cornell.edu (e-mail).
|
ABSTRACT
|
Psychiatric comorbidity among medicine inpatients is prevalent, yet no reliable screen for identifying need for psychiatric referrals exists. The Cornell Psychiatric Screen was developed as a brief, reliable, and valid screening tool. The items on the Cornell Psychiatric Screen assess a range of psychiatric conditions, including cognition and behavior, depressive symptoms, anxiety, drug and alcohol history, and the patient's desire to see a psychiatrist. Of the patients whose Cornell Psychiatric Screen results indicated possible psychopathology, 89% had documented psychiatric comorbidity according to DSM-IV criteria. On the basis of preliminary validation, the Cornell Psychiatric Screen appears to be a useful tool for identifying patients who require a psychiatric evaluation.
|
INTRODUCTION
|
Psychiatric comorbidity is common among general medicine inpatients, with prevalence rates ranging from 15% to 50%.1,2 Patients with psychiatric comorbidity receive more diagnostic tests, have longer lengths of stay, and have higher hospital costs compared with patients who do not. In this context, there has been considerable interest in the impact of psychiatric consultation-liaison services on these outcomes.3–5 Studies of psychiatric screening and consultation-liaison interventions have shown that such interventions are associated with improved psychiatric outcomes, earlier discharge, and cost savings to the hospital. The earlier the consultation, the better the results.5–7 Unfortunately, in everyday practice, fewer than 5% of medical inpatients receive psychiatric intervention. This is likely due to a number of factors, including underidentification of psychiatric symptoms and stigmatization of psychiatric intervention by physicians and patients.8
Such low rates of psychiatric referrals underscore the need for practical screening instruments to identify patients at risk for psychiatric comorbidity, especially given the difficulty of detecting subtle psychological symptoms in a medical population. Numerous instruments have been developed to screen for psychiatric comorbidity in the medical setting.9–12 Most scales are self-report, cover primarily depression and anxiety symptoms, and are intended for use in the ambulatory care setting. The objective of the study was to develop a brief screening instrument that could be administered by non-mental-health professionals to screen for psychiatric comorbidity among medical inpatients.
|
METHOD
|
The questionnaire was developed and tested in four phases: selection of candidate items, test-retest reliability, selection of final items, and validation. After the scale was developed, its properties were tested among three independent cohorts of patients (N=106, N=299, N=206). All were adult patients admitted to the internal medicine service of the New York-Presbyterian Hospital. The institution's institutional review board on research approved the study protocol.
Phase 1: Selection of Potential Candidate Items
In order to generate a comprehensive list of candidate screening items, psychiatrists with expertise in consultation-liaison psychiatry, mood disorders, schizophrenia, and geriatric psychiatry were convened. This team selected items that pertained to psychiatric symptoms or disorders commonly encountered among medical inpatients. In addition, items associated with adverse medical outcomes (e.g., extended length of stay, physical and psychological decline) were also identified.13
Phase 2: Test-Retest Reliability
To test for reliability, a closed-format questionnaire was developed from the items generated and was administered on two separate occasions to 106 patients who were admitted to the medical service. Kappa was employed as the statistical test of agreement to evaluate test-retest reliability.
Phase 3: Selection of Final Items
To establish face validity and limit the final screen to the most germane items regarding need for psychiatric evaluation, a second independent group of 299 patients was recruited. Within 24 hours of admission, eligible patients were identified with a computer program that the medical house staff use to follow their patients (SIGNOUT, J. Hollenberg, Roslyn, N.Y.). Once informed consent was obtained, trained non-mental-health professional research assistants administered the questionnaire items. A group of psychiatrists, blinded to whether or not the patient was seen by the psychiatry service during their hospitalization, were given an abstract of the patient's chart for review. This abstract described each patient's demographic and clinical characteristics, including their reason for admission, living situation, medical history, comorbidity,14 medications, recent hospitalizations and operations, and recent surgical procedures. Data on their cognitive status were provided by the Short Portable Mental Status Questionnaire.15 Their functional history was documented by using the physical component summary scale of the Medical Outcomes Study 12-Item Short-Form Health Survey.16 Embedded in these abstracts were the items that had been generated as potential candidates. The candidate items were not identified as such.
On the basis of this information and using criteria routinely used by the consultation-liaison service, the psychiatrists rated whether or not the patient needed a psychiatric evaluation on a 5-point Likert scale: definitely needed, probably needed, uncertain, probably not needed, and definitely not needed. The following operational definition of "need for psychiatric evaluation" was used: those patients who may be at risk for complicated or prolonged hospitalization because of a possible underlying psychiatric condition. In total, 15% of patients were rated as definitely needing a psychiatric evaluation (probably needed: 16%; uncertain: 14%; probably not needed: 21%; definitely not needed: 34%). For the purpose of multivariate analysis, the ratings were dichotomized, with the 15% categorized as definitely needing an evaluation being contrasted with the 85% who did not. The questionnaire items were analyzed by using logistic regression to determine which were significantly associated with definitely needing psychiatric evaluation. Items that were significant at the 0.01 level in bivariate analyses were entered as independent predictors in multivariate models that adjusted for potential confounders such as age, ethnicity, gender, and comorbidity. Adjusted odds ratios were derived from the regression coefficients and used as weights in scoring the final screen.
Receiver operator characteristic curves were constructed to establish an optimal cutoff score for the screen at which point both sensitivity and specificity for assessing need for a psychiatric evaluation would be maximized. The area under the receiver operator characteristic curve was calculated to determine the ability of this cutoff score to discriminate between patients who needed and those who did not need a psychiatric evaluation. The method described by Cantor and Kattan was used for calculating the area under the receiver operator characteristic curve.17
Phase 4: Validation
As a further validation effort, the extent to which psychiatric comorbidity was present in patients whose screen results indicated possible psychopathology was determined among a separate group of 206 patients. The external criterion used was DSM-IV criteria for psychiatric disease. An independent psychiatrist from the consultation-liaison service who was also blinded to the outcomes of the study evaluated all patients whose score on the screen was 2 or greater (which indicated possible psychopathology). Using DSM-IV criteria, the psychiatrist determined whether the patient had a comorbid psychiatric diagnosis. To determine the proportion of patients that would not be captured by the screen, patients who received a consultation during the routine process of care but who scored below the cutoff on the screen were also identified.
Finally, the ability of the screen to predict length of hospital stay was determined. The rationale for selecting length of stay is that patients who are admitted to nonpsychiatric services with concomitant psychiatric comorbidity are known to have greater lengths of hospital stay compared with those with no comorbidity.13 To determine whether higher scores on the psychiatric scale were associated with longer in-hospital stays, a series of multivariate analyses were conducted with length of hospital stay as the dependent variable. Adjustments were made for demographic characteristics as well as severity of illness and medical comorbidity. Length of hospital stay was calculated by using the hospital's administrative database (Transition Systems, Inc.). The APACHE II and Charlson Comorbidity Index were used to measure severity of illness and medical comorbidity.14,18 All analyses were performed with SAS software (SAS Institute, Cary, N.C.).
|
RESULTS
|
Selection of Potential Candidate Items
Table 1 shows the initial 18 items that were generated through consensus as well as their respective kappa values. The first three items involved alcohol and substance abuse. Three additional items asked patients to assess their level of energy, sadness, and anxiety. Other items evaluated the patient's orientation, presence of delusional thinking, rambling or unconnected speech, inappropriate dialogue, and whether the patient had wanted to see or had been recommended to see a psychiatrist.
Test-Retest Reliability
Patients who participated in this phase of screen development were an average of 57 years of age (SD=18); approximately 50% were women and Caucasian. As shown in Table 1, three items were dropped from further analysis because they had a kappa value of <0.50. A decision was made to keep the item on use of recreational drugs and enrollment in a drug treatment program, despite its low kappa. The consensus of the psychiatric liaison team was that this item had important clinical relevance.19 An additional item assessing whether or not the patient was confused on admission was also added for similar reasons.
Selection of Final Items on the Scale
Two hundred ninety-nine patients participated in this phase; their mean age was 63 years, and the majority were female and Caucasian. In total, 45 patients were characterized as definitely needing a psychiatric evaluation. Demographic factors such as age, ethnicity, and level of education were not associated with the need for psychiatric evaluation. A history of alcohol and drug use, however, was associated with the need for psychiatric evaluation, as well as current mood and anxiety level. In addition, patients who were confused or who had potential indications of psychosis were also rated as needing a psychiatric evaluation.
On multivariate analysis, seven items remained significant after adjusting for age, gender, ethnicity, and comorbidity. Specifically, patients who were downhearted, blue, or nervous most or all of the time were rated as needing a psychiatric evaluation. Patients who used drugs and were not enrolled in treatment programs, and patients currently drinking more than three drinks a day were also rated as needing a psychiatric evaluation. Patients with evidence of hallucinations, bizarre thoughts or rambling, and tangential speech as well as those with confusion on admission were also rated as needing a psychiatric evaluation. Finally, patients who indicated that they wanted to see a psychiatrist were also rated as needing a psychiatric evaluation. Table 2 shows the final seven items on the psychiatric scale, their odds ratios in predicting definite need to see a psychiatrist (determined through multivariate analysis), and the weight assigned to the item.
Summing the weights shown in Table 2 derives the final score for the Cornell Psychiatric Screen. The mean Cornell Psychiatric Screen score was 1.0 (SD=1.4, range=0–8); higher scores indicated a greater need for psychiatric evaluation. Receiver operator characteristic curves were constructed to establish the cutoff score of 2, which had a sensitivity of 90% and a specificity of 87% in discriminating between those who need an evaluation and those patients who do not.20 The area under the receiver operator characteristic curve at this cutoff was 0.88, which indicates good discriminative ability.
Validation of the Cornell Psychiatric Screen
Participants in this phase were a mean age of 70.3 (SD=16.8). Sixty-one percent were male; 30% were African American or Latino American. An independent psychiatrist from the consultation-liaison service evaluated the charts of the 44 patients who scored  2 on the scale. Of these patients, 39 (89%) had a documented DSM-IV diagnosis in the consultation note. The range of diagnoses included the following: depression (36%), anxiety (9%), dementia (18%), delirium (14%), alcohol dependence/ withdrawal (7%), and personality disorder and insomnia (5%). The remaining 11% did not meet DSM-IV criteria for psychiatric diagnosis.
The charts of 22 patients who had also received a psychiatric consultation during the study period were reviewed. Of these patients, 15 (68%) were captured by the screen. Those who were not captured by the screen had been referred for suicidal ideation (29%), refusing medical recommendations (29%), and overt psychosis (14%); the remaining 28% did not have a reason for psychiatric referral written in the chart.
Finally, higher scores on the Cornell Psychiatric Screen were associated with longer length of hospital stay (F=2.67, df=56, 147, p=0.02). In multivariate analysis, after adjusting for age, gender, severity of illness, and comorbidity, the score on the scale was found to be a significant predictor of length of stay (r=0.32, df=10, p=0.03).
|
DISCUSSION
|
Designed as a brief screening instrument to be administered by nonpsychiatrists, the Cornell Psychiatric Screen can quickly and reliably identify hospitalized patients with underlying psychiatric comorbidity who are in need of a psychiatric evaluation. The Cornell Psychiatric Screen consists of seven items requiring information that is readily available upon admission. The Cornell Psychiatric Screen was shown to be a reliable scale with kappa values ranging from 0.50 to 1.00. The one item that did not meet this criterion was the item concerning recreational drug use. In a study by Stein and colleagues,19 the prevalence of recreational drug use among hospitalized patients on medical or surgical wards was reported as 11%; only 18% of these patients had been asked about drug use. Since drug use may be underreported among medical patients, this item was felt to have clinical relevance. Patients whose Cornell Psychiatric Screen results indicated possible psychopathology were found to have documented evidence of psychiatric comorbidity: 89% had evidence of psychiatric comorbidity according to DSM-IV criteria.
There are established scales that have been developed for diagnosing psychiatric comorbidity.9–11 The Patient Health Questionnaire from the Primary Care Evaluation of Mental Disorders (PRIME-MD)21 was developed to diagnose mental disorders in primary care. Although it is an abbreviated version of the original PRIME-MD questionnaire,22 its three-page format may be less practical in the inpatient setting where patients may be fatigued from the admission process or sedated due to medications. Furthermore, scoring of this instrument requires the use of a diagnostic algorithm, which may make its use less favorable. The shorter number of items on the Cornell Psychiatric Screen may offer some advantages over existing instruments. However, future studies are needed to test the discriminative ability of the Cornell Psychiatric Screen against other measures.
The Cornell Psychiatric Screen also has potential use in identifying patients at risk for increased length of stay. As the score on Cornell Psychiatric Screen increased from 0 to 8, there was an incremental increase in the length of stay from 7 to 12 days. This association with length of stay was independent of the patient's severity of illness or other medical comorbidity. These results agree with prior studies that have demonstrated an association between psychiatric comorbidity and increased utilization during hospitalization.1,3,8 In the current health care environment, this association of length of stay is important as hospitals and third-party payers are developing strategies to cut costs. While there is evidence to support the early involvement of consultation-liaison teams, two large studies did not show a reduction in hospital cost nor an improvement in patient outcome.23,24 The Cornell Psychiatric Screen may be a useful guide in developing interventions to reduce length of stay.
There are some limitations to the development of the Cornell Psychiatric Screen. First, it relies on patient self-report and not on clinician findings. However, several other valid instruments routinely used in psychiatry rely upon this method of data collection.9–11,21 Second, the Cornell Psychiatric Screen was developed on the basis of judgments about referral need by psychiatrists at only one institution. Since the nature of referrals may vary among different institutions, the screen may need to be supplemented with additional items that better reflect the conditions that are prevalent in other settings. Third, the Cornell Psychiatric Screen was designed to supplement clinical observation by aiding in the identification of patients with more subtle psychiatric conditions that may otherwise go unrecognized; it does not include an exhaustive list of items.
Given the brevity of the scale, conditions such as adjustment disorders, panic disorder, and suicide attempts may not be detected. In spite of this limitation, of the patients who were actually seen by the consultation-liaison service, the screen identified 68%. The patients who were not captured by the screen were evaluated for reasons such as suicidal ideation, refusing medical recommendations, and overt psychosis. These conditions would most likely initiate a referral in any event. Replication studies may need to broaden the screen by including these items. Finally, in an ideal situation, the external psychiatric evaluation using established criteria should have occurred concurrently with the administration of the Cornell Psychiatric Screen. However given the resources of this study, this was not done.
Preliminary attempts to validate the Cornell Psychiatric Screen have been described. While there is a need to further refine and validate the Cornell Psychiatric Screen, it has important clinical value. The Cornell Psychiatric Screen has the potential to identify patients with psychiatric comorbidity. In the absence of such a screen, patients who are at risk for poor outcomes may go unrecognized and untreated. The Cornell Psychiatric Screen has the potential to assist in allocating resources to those patients who exhibit the greatest need. Finally, the Cornell Psychiatric Screen has the potential for use in testing interventions that may improve outcomes among medical inpatients who have underlying psychiatric comorbidity.
|
ACKNOWLEDGMENTS
|
Supported by funding from the New York-Presbyterian Hospital.
|
REFERENCES
|
- Levenson JL, Hamer R, Silverman JJ, Rossiter LF: Psychopathology in medical inpatients and its relationship to length of hospital stay: a pilot study. Int J Psychiatry Med 1986; 16:231–236[Medline]
- Shevitz SA, Silberfarb PM, Lipowski ZJ: Psychiatric consultations in a general hospital: a report on 1,000 referrals. Dis Nerv Syst 1976; 37:295–300[Medline]
- Hall RC, Frankel BL: The value of consultation-liaison interventions to the general hospital. Psychiatr Serv 1996; 47:418–420[Abstract/Free Full Text]
- Wallen J, Pincus HA, Goldman HH, Marcus SE: Psychiatric consultations in short-term general hospitals. Arch Gen Psychiatry 1987; 44:163–168[Abstract]
- Strain JJ, Lyons JS, Hammer JS, Fahs M, Lebovits A, Paddison PL, Snyder S, Strauss E, Burton R, Nuber G, Abernathy T, Sacks H, Nordlie J, Sacks C: Cost offset from a psychiatric consultation-liaison intervention with elderly hip fracture patients. Am J Psychiatry 1991; 148:1044–1049[Abstract/Free Full Text]
- Lyons JS, Hammer JS, Strain JJ, Fulop G: The timing of psychiatric consultation in the general hospital and length of hospital stay. Gen Hosp Psychiatry 1986; 8:159–162[CrossRef][Medline]
- Slaets JP, Kauffmann RH, Duivenvoorden HJ, Pelemans W, Schudel WJ: A randomized trial of geriatric liaison intervention in elderly medical inpatients. Psychosom Med 1997; 59:585–591[Abstract/Free Full Text]
- Karasu TB, Waltzman SA, Lindenmayer JP, Buckley PJ: The medical care of patients with psychiatric illness. Hosp Community Psychiatry 1980; 31:463–472[Abstract/Free Full Text]
- Herrmann C: International experiences with the Hospital Anxiety and Depression Scale—a review of validation data and clinical results. J Psychosom Res 1997; 42:17–41[CrossRef][Medline]
- Hamilton M: A rating scale for depression. J Neurol Neurosurg Psychiatry 1960; 23:56–62
- Lobo A, Campos R, Perez-Echeverria MJ, Izuzquiza J, Garcia-Campayo J, Saz P, Marcos G: A new interview for the multiaxial assessment of psychiatric morbidity in medical settings. Psychol Med 1993; 23:505–510[Medline]
- McDowell I, Newell C: A Guide to Rating Scales and Questionnaires. New York, Oxford University Press, 1996
- Saravay SM, Steinberg MD, Weinschel B, Pollack S, Alovis N: Psychological comorbidity and length of stay in the general hospital. Am J Psychiatry 1991; 148:324–329[Abstract/Free Full Text]
- Charlson ME, Pompei P, Ales KL, MacKenzie CR: A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987; 40:373–383[CrossRef][Medline]
- Pfeiffer E: A short portable mental status questionnaire for the assessment of organic brain deficit in elderly patients. J Am Geriatr Soc 1975; 23:433–441[Medline]
- Ware J Jr, Kosinski M, Keller SD: A 12-item short-form health survey: construction of scales and preliminary tests of reliability and validity. Med Care 1996; 34:220–233[CrossRef][Medline]
- Cantor SB, Kattan MW: Determining the area under the ROC curve for a binary diagnostic test. Med Decis Making 2000; 20:468–470[Abstract/Free Full Text]
- Knaus WA, Draper EA, Wagner DP, Zimmerman JE: APACHE II: a severity of disease classification system. Crit Care Med 1985; 13:818–829[Medline]
- Stein MD, Wilkinson J, Berglas N, O'Sullivan P: Prevalence and detection of illicit drug disorders among hospitalized patients. Am J Drug Alcohol Abuse 1996; 22:463–471[Medline]
- Hanley JA, McNeil BJ: The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982; 143:29–36[Abstract/Free Full Text]
- Spitzer RL, Kroenke K, Williams JB: Validation and utility of a self-report version of PRIME-MD: the PHQ Primary Care Study. JAMA 1999; 282:1737–1744[Abstract/Free Full Text]
- Spitzer RL, Williams JB, Kroenke K, Linzer M, deGruy FV III, Hahn SR, Brody D, Johnson JG: Utility of a new procedure for diagnosing mental disorders in primary care: the PRIME-MD 1000 study. JAMA 1994; 272:1749–1756[Abstract]
- Levenson JL, Hamer RM, Rossiter LF: A randomized controlled study of psychiatric consultation guided by screening in general medical inpatients. Am J Psychiatry 1992; 149:631–637[Abstract/Free Full Text]
- Gater RA, Goldberg DP, Evanson JM, Lowson K, McGrath G, Tantam D, Million L: Detection and treatment of psychiatric illness in a general medical ward: a modified cost-benefit analysis. J Psychosom Res 1998; 45:437–448[CrossRef][Medline]
This article has been cited by other articles:
|
|
|
|
 
K. M. Gebbie, R. M. Larkin, S. J. Klein, L. Wright, J. Satriano, J. J. Culkin, and B. S. Devore
Improving Access to Mental Health Services for New York State Prison Inmates
Journal of Correctional Health Care,
April 1, 2008;
14(2):
122 - 135.
[Abstract]
[PDF]
|
|
|
|
|
|
|
J. D. Ford, R. L. Trestman, V. Wiesbrock, and Wanli Zhang
Development and Validation of a Brief Mental Health Screening Instrument for Newly Incarcerated Adults
Assessment,
September 1, 2007;
14(3):
279 - 299.
[Abstract]
[PDF]
|
|
|
|
|
|
|
M. Moran
Scale Detects Psychiatric Comorbidity In Hospitalized Medical Patients
Psychiatr News,
December 5, 2003;
38(23):
24 - 59.
[Full Text]
|
|
|
Get information about faster international access.
Privacy Policy
Copyright © 2003
Academy of Psychosomatic Medicine.
All rights reserved.
Home
| Search
| Current Issue
| Past Issues
| Subscribe
| All APPI Journals
| Help
| Contact Us
|
TOP
ABSTRACT
INTRODUCTION
