The Use of a Brief Depression Screen in OlderEmergency Department Patients

Fredric M. Hustey, MD

AbstractObjectives: To determine the sensitivity and specificity of abrief two-question depression screen for the detection ofdepression in older emergency department (ED) patients,and to determine the prevalence of depression in thispopulation. Methods: This was a prospective, observationalstudy. Participants included a convenience sampling of EDpatients 70 years and older presenting to an urban teachinghospital over a 17-month period. Exclusions were refusalto participate, inability to communicate, and critical illness.Subjects were screened for depression with the previouslyvalidated Short Form Geriatric Depression Scale (SFGDS).Standardized scores on the SFGDS were used to determinethe prevalence of depression. Patients were then given apreviously published two-question depression screen, andresults were compared with the SFGDS as the criterionstandard. Sensitivity, specificity, and prevalence data arereported as proportions with 95% confidence intervals

(95% CIs). Results: Two hundred sixty-seven of 327 eligiblepatients were enrolled. Forty-four (17%; 95% CI = 12% to21%) scored positive for depression on the SFGDS. Thesensitivity of the brief two-question depression screen was84% (37/44; 95% CI = 70% to 93%), with a specificity of 61%(136/223; 95% CI = 55% to 67%) using a cutoff score of atleast one of two positive responses. Conclusions: Depres-sion is fairly prevalent in older ED patients. The brief two-question depression screen, using a cutoff score of at leastone positive response, is promising for ED use. However,given lower specificity, patients scoring positive for depres-sion should be followed up with a more specific tool such asthe self-administered SFGDS prior to referral for furtherevaluation and treatment. Key words: depression; detection;prevalence; emergency; elderly; screening. ACADEMICEMERGENCY MEDICINE 2005; 12:905–908.

Depression is one of the most common psychiatricdisorders in older people.1 Approximately 30% of allolder patients seen in emergency departments (EDs)may suffer from depression.1,2 These patients are athigher risk for morbidity and mortality, includingmore frequent ED visits, prolonged hospital staysafter admission, and higher rates of health careresource utilization in general.1,3–5

It has been recommended that social, psychological,and functional status issues be routinely addressedas part of the older patient’s emergency encounter.However, routine screening in the ED can be time-consuming and difficult to implement. Recently, abrief two-question depression screen was reported tohave a high sensitivity and specificity for the detec-tion of depression in patients in the primary caresetting.6 We designed a study with the objective ofdetermining the sensitivity and specificity of this

two-question depression screen for the detection ofdepression in older ED patients compared with thepreviously validated Short Form Geriatric DepressionScale (SFGDS) as the criterion standard. The SFGDSis a 15-item scale derived from the original 30-itemGeriatric Depression Scale. It has been shown toperform similarly to the original scale, with sensitivityof 91% using a cutoff score of 5 or more for thedetection of depression.7 A secondary objective wasto determine the prevalence of depression in olderED patients.

METHODS

Study Design. This was a preplanned subanalysisfrom a larger prospective study attempting to deter-mine the effect of depression screening, using theSFGDS, on the care of older ED patients. This studywas reviewed and approved by the hospital institu-tional review board (IRB). The requirement for writteninformed consent was waived by the IRB, and verbalconsent was obtained from all participants.

Study Setting and Population. This study wasconducted between July 2000 and November 2001 atan urban teaching hospital with approximately 55,000ED visits per year. All patients aged 70 years or olderpresenting to the ED during the study period wereeligible for enrollment. Patients were excluded if they

From the Department of Emergency Medicine, The Cleveland ClinicFoundation (FMH), Cleveland, OH.Received February 14, 2005; revisions received March 22, 2005, andApril 6, 2005; accepted April 6, 2005.Presented in part at the American College of Emergency Physiciansannual meeting, San Francisco, CA, October 2004.Address for correspondence and reprints: Fredric M. Hustey, MD,Department of Emergency Medicine-E-19, The Cleveland ClinicFoundation, 9500 Euclid Avenue, Cleveland, OH 44195. Fax:216-444-1703; e-mail: husteyf@ccf.org.doi:10.1197/j.aem.2005.04.009

ACAD EMERG MED d September 2005, Vol. 12, No. 9 d www.aemj.org 905

refused to participate, were critically ill, were unableto communicate or cooperate with data acquisition, ordid not speak English in the absence of a translator.

Study Protocol. A convenience sample of older EDpatients was enrolled. Sampling was done during 91six- to eight-hour shifts that were varied to includedays, nights, weekdays, and weekends with frequen-cies based on the usual ED presentation times of olderpatients. Only the initial visit was included for pa-tients presenting to the ED more than once. Eligibleand consenting patients were evaluated for depres-sion using the SFGDS and then presented with a brieftwo-question depression screen asking the followingquestions: During the past month, have you oftenbeen bothered by feeling down, depressed, or hope-less? During the past month, have you often beenbothered by little interest or pleasure in doing things?Research assistants were given a brief lecture on theuse of the screening tools, and the primary investiga-tor observed the first ten screens for each assistantto ensure reliability. Patients were handed sheets ofpaper with the surveys to complete. Research assis-tants aided patients who requested help completingthe surveys. Patients were also screened for cognitiveimpairment using the previously validated Orienta-tion–Memory–Concentration test.8 Standardized cut-off scores for cognitive impairment screening wereused. As part of the larger study, patients were alsoscreened for alcoholism using a modified CAGEquestionnaire. Demographic data were obtained onchart review. Research assistants were blinded tostudy objectives, and tests were scored only afterpatients had completed both surveys.

Standardized cutoff scores on the SFGDS were usedto determine the prevalence of depression. Results ofthe two-question depression screen were comparedagainst the SFGDS as criterion standard. Sensitivityand specificity are reported as proportions with 95%confidence intervals (95% CIs).

RESULTS

Three hundred twenty-seven eligible patients werescreened, of whom 60 were subsequently excluded(Table 1). Of the remaining 267 patients, one hundredforty (52%) were white, 125 (47%) African American,and two (1%) Hispanic. One hundred fifty-one (57%)were female. The age range was 70 to 102 years; themean age (6SD) was 77.82 years (65.71).

The prevalence of depression was 17% (44/267;95% CI = 12% to 21%). Eight (18%; 95% CI = 8% to33%) of these 44 patients had a prior history ofdepression, and ten of 44 (23%; 95% CI = 12% to38%) were taking antidepressants at the time of theED visit.

One hundred twenty-four of 267 patients had atleast one positive response on the two-question de-

pression screen (46%; 95% CI = 40% to 52%), while 49were positive on both items (18%; 95% CI = 14% to23%). The sensitivity of the two-question depressionscreen for the detection of depression is summarizedin Table 2.

The prevalence of cognitive impairment in theoverall study population was 26% (69/267; 95%CI = 21% to 31%). The prevalence of cognitive impair-ment in patients screening positive for depressionon the SFGDS was much higher (19/44, 43%; 95%CI = 28% to 59%) than in the general study popula-tion. Sensitivity of the two-question depression screenfor the detection of depression in patients with cog-nitive impairment was 79% (15/19; 95% CI = 54% to94%) versus 22 of 25 patients without impairment,(88%; 95% CI = 69% to 98%). Sensitivities were alsosimilar across gender (16/18 males, 89%; 95%CI = 65% to 99%; vs. 21/26 females, 81%; 95%CI = 61% to 94%) and race (24/29 white, 83%; vs.13/15 African American, 87%).

All eight patients with a past medical history ofdepression and a positive score for depression on theSFGDS were positive on the two-question depressionscreen (8/8, 100%; 95% CI = 63% to 100%). In addi-tion, all ten patients with a positive score on theSFGDS for depression who were taking antidepres-sants at the time of the ED visit were also positive(10/10, 100%; 95% CI = 65% to 100%).

DISCUSSION

Consistent with the prior report,6 this brief two-question depression screen has good sensitivity and

TABLE 1. Patient Participation

Participation No. (%)

Ineligible 60/327 (18%)Prior enrollment 5Incomplete data 10Refusal 21Inability to cooperate with survey 15Non-English-speaking 5Critical illness 4

Completed interview 267/327 (82%)

TABLE 2. Sensitivity of the Two-question DepressionScreen for the Detection of Depression in OlderEmergency Department Patients (Patients withDepression: n = 44)

Scoring Method Sensitivity Specificity

At least one of thetwo questions ispositive fordepression

84% (37/44; 95%CI = 70% to 93%)

61% (136/223; 95%CI = 55% to 67%)

Both questions mustbe positive fordepression

46% (20/44; 95%CI = 30% to 61%)

87% (194/223; 95%CI = 83% to 91%)

906 Hustey d BRIEF DEPRESSION SCREEN FOR OLDER ED PATIENTS

fair specificity for the detection of depression using acutoff score of at least one positive response. A priorstudy in an adult primary care population usingthis cutoff score reported a sensitivity of 96% (95%CI = 90% to 99%) and a specificity of 57% (95%CI = 53% to 62%) for the detection of depressionwhen compared with the National Institute of MentalHealth Quick Diagnostic Interview Schedule.6 Thesensitivity in our population was slightly lower (84%),but not within statistical significance. The specificitieswere nearly identical. Studies evaluating shorterscreening tools for depression in older ED patientsare rare. A brief three-question screen was shown tohave a sensitivity of 79% with a specificity of 66%9

when used in older ED patients. We are aware of noother studies to evaluate shorter screening tools fordepression in this population.Depression is prevalent in older ED patients and

poorly recognized by emergency physicians.1,2 Inaddition, depression in ED patients may contributeto a lack of health care compliance and adverseoutcomes.3–5 Identifying depression in these patientscould lead to targeted interventions aimed at reducingthese problems. The SFGDS has been shown to haveboth good sensitivity and good specificity for thedetection of depression in older patients.7,10 The needfor studies to develop brief screening instruments forspecific conditions in older ED patients has beenpreviously identified through the Geriatrics for Spe-cialists project of the American Geriatrics Society. Thisbrief two-question screen helps address this need.While these two questions contain several elementseach (some may consider it a five-question screen), itis still significantly shorter than the 15-item SFGDS.While this two-question screen is sensitive for the

detection of depression in older ED patients, it stillhas limitations in the ED setting. The high rate offalse-positive results suggests that further testingwould be warranted prior to determining a need forreferral and further treatment. Given the time con-straints of the ED, one strategy might be to screen allolder patients with this survey and then rescreenthose with at least one positive response using a morespecific tool such as the SFGDS. Patients scoringpositive on the SFGDS could then be referred forfurther evaluation and treatment. In addition, thelower limit of the 95% CI for sensitivity was 70%,which might be considered low for a screening test.Larger studies may be useful in further clarifying thisissue.This study also supports existing data that depres-

sion is fairly prevalent in older ED patients. Ourprevalence of 17% is slightly lower than that previ-ously reported, although those studies did not usethe SFGDS, but rather a modified version of theKoenig Scale.1,2 In addition, current state of healthand presenting problems related to the ED visit, whichmay have contributed to depression, may vary among

these study populations. The prevalence of depres-sion may also have been underestimated. A signifi-cant proportion of the 11% of patients who refusedto participate or were unable to cooperate with datacollection may have been depressed or cognitivelyimpaired. This may have led to some bias in the studypopulation.

Of the 44 patients scoring positive for depression onthe SFGDS, more were taking antidepressants thanhad a prior history of depression. Some of thesepatients may have been taking antidepressants forother reasons, such as chronic pain.

LIMITATIONS

The order of administration of the SFGDS and two-question depression screen was not randomized. Thismay have significantly biased results, and sensitivityand specificity may actually be lower than reported.Our population was a convenience sampling, andsampling bias cannot be excluded. In addition, theSFGDS was used as the criterion standard instead offormal psychiatric interview focusing on Diagnosticand Statistical Manual IV (DSM IV) criteria for thediagnosis of depression. Data were not collectedregarding socioeconomic status, which could be aconfounder for patients with depression. Our rela-tively small sample size also contributed to limitedstatistical significance in some subgroup analyses.

CONCLUSIONS

Depression is fairly prevalent in older ED patients.The brief two-question depression screen, using acutoff score of at least one positive response, ispromising for ED use. However, given suboptimalspecificity of this screen, patients scoring positive fordepression should be followed up with a more spe-cific tool such as the SFGDS prior to referral forfurther evaluation and treatment.

References

1. Meldon SW, Emerman CL, Schubert DS. Recognition ofdepression in geriatric ED patients by emergency physicians.Ann Emerg Med. 1997; 30:442–7.

2. Meldon SW, Emerman CL, Schubert DS, Moffa DA, EtheartRG. Depression in geriatric ED patients: prevalence andrecognition. Ann Emerg Med. 1997; 30:141–5.

3. Dahlen I, Janson C. Anxiety and depression are related to theoutcome of emergency treatment in patients with obstructivepulmonary disease. Chest. 2002; 122:1633–7.

4. Ganzini L, Smith DM, Fenn DS, Lee MA. Depression andmortality in medically ill older adults. J Am Geriatr Soc. 1997;45:307–12.

5. Fischer LR, Wei F, Rolnick SJ, et al. Geriatric depression,antidepressant treatment, and healthcare utilization in a healthmaintenance organization. J Am Geriatr Soc. 2002; 50:307–12.

6. Whooley MA, Avins AL, Miranda J, Browner WS. Case-findinginstruments for depression. Two questions are as good asmany. J Gen Intern Med. 1997; 12:439–45.

ACAD EMERG MED d September 2005, Vol. 12, No. 9 d www.aemj.org 907

7. Lesher EL, Berryhill JS. Validation of the Geriatric DepressionScale–Short Form among inpatients. J Clin Psychol. 1994; 50:256–60.

8. Katzman R, Brown T, Fuld P, Peck A, Schechter R, SchimmelH. Validation of a short Orientation–Memory–ConcentrationTest of cognitive impairment. Am J Psychiatr. 1983; 140:734–9.

9. Fabacher DA, Raccio-Robak N, McErlean MA, Milano PM,Verdile VP. Validation of a brief screening tool to detectdepression in elderly ED patients. Am J Emerg Med. 2002; 20:99–102.

10. Burke WJ, Roccaforte WH, Wengel SP. The short form of theGeriatric Depression Scale: a comparison with the 30-itemform. J Geriatr Psychiatr Neurol. 1991; 4:173–8.

908 Hustey d BRIEF DEPRESSION SCREEN FOR OLDER ED PATIENTS