• 
• 

Three Part Question

[In elderly patients over 65 years old, presenting to the ED] what [clinical screening tool] leads [to more accurate identification of cognitive impairment] by emergency care providers]?

Clinical Scenario

Mr. Roberts is a 70 year old male that presents to the ED from a nursing home after feeling short of breath and unwell. A note from the nursing home states that Mr. Roberts has been quite fatigued and abnormal with his interactions with the nursing staff. Lab investigations reveals a urinary tract infection and CXR reveals a possible pneumonia. Talking to Mr. Roberts, he is unable to answer basic orientation related questions and appears quite withdrawn. There is no history of Dementia on his medical record, however you suspect a possible cognitive impairment. You wonder which clinical screening tool would be most sensitive, accurate and convenient to use on Mr. Roberts before attending to the many other patients in the waiting room.

Search Strategy

Ovid Medline (R) 1948 to November Week 3 2011
EMBASE 1980 to 2011 Week 52
PSYC info1967 to December Week 4 2011
CINAHL 1998 to 2011


Ovid Medline (R) 1948 to November Week 3 2011
EMBASE 1980 to 2011 Week 52
PSYC info1967 to December Week 4 2011

Elderly.mp. or exp Aged/ or exp Aging/ or seniors.mp. or exp Adult/ or geriatric.mp. or exp Geriatrics/
AND
exp emergency health service/ or exp emergency ward/ or emergency medicine.mp. or exp Emergency Medicine/ or exp Emergency Medical Services/ OR exp Emergency Service, Hospital/ or emergency department.mp. OR ED.mp. [mp=ti, ab, sh, hw, tn, ot, dm, mf, dv, kw, ps, rs, nm, an, ui, tc, id, tm] OR acute care hospital.mp. or exp Emergency Service, Hospital {Including Related Terms}/
AND
exp Questionnaires/ or screening tool.mp. or exp Nursing Assessment/ or assessment tool.mp. or exp Risk Assessment/ OR screening test.mp. [mp=ti, ab, sh, hw, tn, ot, dm, mf, dv, kw, ps, rs, nm, an, ui, tc, id, tm]
AND
cognitive impairment.mp. [mp=ti, ab, sh, hw, tn, ot, dm, mf, dv, kw, ps, rs, nm, an, ui, tc, id, tm] OR cognitive dysfunction.mp. [mp=ti, ab, sh, hw, tn, ot, dm, mf, dv, kw, ps, rs, nm, an, ui, tc, id, tm] OR cognition.mp. [mp=ti, ab, sh, hw, tn, ot, dm, mf, dv, kw, ps, rs, nm, an, ui, tc, id, tm] OR delirium.mp. [mp=ti, ab, sh, hw, tn, ot, dm, mf, dv, kw, ps, rs, nm, an, ui, tc, id, tm] OR dementia.mp. [mp=ti, ab, sh, hw, tn, ot, dm, mf, dv, kw, ps, rs, nm, an, ui, tc, id, tm] OR impaired mental status.mp. [mp=ti, ab, sh, hw, tn, ot, dm, mf, dv, kw, ps, rs, nm, an, ui, tc, id, tm]

CINAHL 1998-2011:
(MH "Delirium, Dementia, Amnestic, Cognitive Disorders") OR "cognitive impairment"
AND (MM "Emergency Medicine") OR (MM "Psychiatric Emergencies") OR (MM "Emergency Nurse Practitioners") OR "emergency department" AND"screening tool" OR (MH "Questionnaires") OR "questionnaire" AND (MH "Aged") OR (MH "Aged, 80 and Over") OR "elderly"
Applied related words and search within the full text of the articles. Age Group 65+ years.

Pubmed: Hand searched “exp Delirium/ or delirium screening.mp”

Several articles were found from scanning the references of relevant papers.

Search Outcome

685 articles were found on all databases, of which 7 were relevant for inclusion.

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Carpenter et al. 2011 USA	- Elderly≥ 65;.ED physician discretion, English speaking, able to comply with data gathering, patient consent; excluded patients receiving mental status altering medications (antiemetic, benzodiazepines, narcotics) prior or during period, discharged prior to enrollment, critically ill, previously enrolled. - 169 enrolled, 163 included in analysis due to lack of data collection or refusal to participate further. - Mean age 78+/- 8 SD; 61% female, 49% were African Americans and 51% were Caucasian.	Prospective, cross-sectional	Measured sensitivity of screening tools, MMSE used as gold standard	- MMSE (gold standard ≤ 23) identified 60/163 (37%) patients with cognitive dysfunction - Delirium was identified in 9 – only by use of the CAM-ICU - Each screening instrument identified more patients with cognitive dysfunction than the MMSE. BAS, 65%; cAD8, 55%, O3DY, 66% and SBT, 43%. - Sensitivity of SBT, BAS, O3DY was 95%. SBT showed the largest agreement with the MMSE compared to the other instruments. - Each instrument had low specificities and would not be clinically helpful if positive.	- Convenience sample used – may reduce external validity - CAM-ICU tool has not been validated in the ED. - Screening tools were administered consecutively – risk of test-retest bias and test fatigue. MMSE (gold standard) was administered last. - MMSE suboptimal gold standard: sensitivity for mild cognitive impairment as low as 18%.
Carpenter et. al. 2011 USA	319 Elderly patients >65; non-critically ill, English speaking, receiving care in the ED for any reason. Excluded patients who received medication that may affect their mental status (narcotics, benzodiazepines, antiemetic)	Observational cross sectional cohort study	Sensitivity and specificity (95%CI)	SIS 74% (CI:68-80), 77% (74-80); cAD8 63% (53-72), 79%(73-85); pAD8 37% (28-46), 82% (77-86).	- Nonconsecutive convenience sampling. - Order of test administration not randomized or recorded - AD8 was administered to patients in 59% of cases and to caregivers 41% of cases. - MMSE not an ideal gold standard, also not sensitive for mild cognitive impairment - Unable to discriminate dementia from delirium with tests used - Did not assess reliability of SIS, AD8 and did not assess for selection bias between research assistants. - Cannot exclude a ‘learning phenomenon’ for the 3 item recall for the SIS and MMSE. Also 30 minute delay could have potentially allowed cognitive improvement or decline. - The SIS although superior only evaluates 2 realms (recall and orientation) of cognitive dysfunction. ED screening tools that evaluate different or additional domains may prove to be more accurate in cognitive dysfunction detection
			Combined abnormal SIS and cAD8	sensitivity 89% (80-95), specificity 70% (63-73)
			Abnormal SIS and cAD8 +LR	19.9 (CI: 9.8-74.4)
			ROC AUC	SIS: 0.83 (0.78-0.87); cAD8 0.74 (0.65-0.81); pAD8 0.67 (0.60-0.74)
Carpenter et. al. 2011 USA	Patients ≥ 65, English speaking and not prior use of sedating medications, abnormal MMSE scores.	Prospective cross sectional study	Sensitivity, specificity	- BAS Sen 72%(95% CI: 57-66), Spec 65% (44-82) - SBT: Sen 63% spec 88% (68-97) - cAD8: Sen 40% (34-41) Spec 89% (60-98)	- Study only looked at MCI, excluded patients with abnormal MMSE
Monette et. al. 2001 Canada	110 elderly patients, ≥66 years old. Exclusion criteria were as follows: blind, deaf, mute, aphasic, hospitalized the week prior to the interview, did not speak English or French, lived in a nursing home, too ill to be interviewed. Patients were interviewed within at least 6 hours of arriving to ED.	Reliability Study	Comparing CAM results obtained by lay interviewers to those obtained by geriatricians. Sensitivity, specificity	CAM administered by lay interviewers (compared to geriatricians) were 86% sensitive and 100% specific with excellent inter-observer reliability (kappa= 0.91)	- Focused on delirium ?applicable to other types of cognitive impairment. - Patients were recruited on weekdays, during the day and early evening shifts – this may have lead to the inclusion of patients that were generally lower acuity and less sick (spectrum bias). -Exclusion of patients who resided in a nursing home decreases this study’s external validity as an increasing number of patients that come into the ED will live in an assisted living or nursing facility.
			Kappa statistics	Kappa statistics for CAM vs. DSM IV: 0.97 (0.78-1.16), CAM vs. DSM-IIIR: 0.86 (0.43-1.12) and CAM vs. clinical impression: 0.94 (0.76-1.13)
Schofield et. al 2009 UK	601 Patients ≥65 presenting to an inner city ED. Excluded patients unable to communicate, with learning disabilities, severe hearing disability, non-English speaking (with no interpreter).	Prospective comparison	Sensitivity, specificity, PPV,NPV: AMT4	sensitivity 80% (CI: 0.75-0.85), specificity88% (CI 0.84-0.91) PPV 84 (CI 0.77-0.88) NPV 85 (CI 0.81-0.89)	- Used MMSE as gold standard. - Convenience sample may result in selection bias. - Carried out in United Kingdom, possible patient differences that may limit external validity to Canada. - Exclusion criteria limit external validity. - AMT4 scores were deduced from AMT – not separately obtained. - The research assistant administering the MMSE and AMT were not blinded to the results of the previous assessment, which may contribute to observer bias, threatening this study’s internal validity
			Sensitiviy/specificity,PPV,NPV: AMT	sensitivity 76% (CI 0.69-0.81), specificity 93% (CI: 0.90-0.96)PPV 90% (CI 0.84-0.93) NPV 83% (CI 0.79-0.87)
			Sensitivity/Specificity: Judgment of admitting nurse	sensitivity 50.5% (CI 0.44-0.57) and specificity 98.6% (CI 0.96-1.00)
Wilber et. al 2008 USA	352 Elderly ED patients ≥65. Excluded non English speaking, received medication that affect cognition, critically ill, uncooperative, and previously enrolled patients.	Prospective, cross sectional study	Sensitivity and specificity of SIS compared to MMSE. PPV and NPV	Sensitivity:63% (95% CI: 53-72%) Specificity: 81% (95% CI: 75-85%) PPV:60 (95% CI:50-69) NPV:83% (95% CI:77-87)	- Convenience sample – based on research assistant availability– risk of selection bias - 22% of patients refused to participate- risk of selection of unknown bias into study - Exclusion of non-English speaking patients reduces external validity. - Study conduced at 3 academic inner city ED, this limits external validity to non-urban centers. - Possible learning phenomenon with 3-item recall may account for differing sensitivity and specificities at different locations. - Study did not comment whether the 3 item recall list differed between the SIS and MMSE - SIS only assesses for temporal orientation and recall – may be unable to adequately detect cognitive impairment. - MMSE not ideal gold standard.
Wilber et. al 2005 USA	149 ED patients ≥65, excluded non-English speaking, un-cooperative or unwilling patients, medically unstable and those who received medications affecting mental status; average age 75.	Prospective randomized cross-sectional study.	- Compared sensitivity, specificity of SIS and Mini-Cog. Gold standard MMSE	- 23% of patients were cognitively impaired according to the MMSE - SIS sensitivity 94% (CI 73-100%) and specificity 86% (CI 74-94%) PPV 68% NPV: 98% - Mini-Cog sensitivity 75% (CI 48-93%) and specificity 85% (CI 73-93%) PPV 57% NPV 93%	- 21% of eligible patients refused to participate – may introduce selection bias and/or exclusion of severely cognitively impaired. - Exclusion of non-English speaking patients may limit external validity to all patient types. - Study conducted during convenient times when investigators were available – this may augment selection bias by limiting the number of mild or severely cognitively impaired that may during busy times in the ED. Introduces spectrum bias. - Conducted at community teaching hospital – may limit external validity to larger academic facilities. - MMSE administered after either SIS or Mini-Cog. - Wide confidence interval for SIS sensitivity due to small sample sizes. - Investigators administering the MMSE were not blinded to the results of the SIS or Mini-Cog. This undeniably introduces observer bias into the study.
			- Kappa statistics and ROC AUC calculated.	- Kappa statistic for the SIS and MMSE was 0.7. and Mini-Cog 0.5 - SIS agreed with MMSE 88% of the time and Mini-Cog 83% of the time - AUC for the ROC curve for SIS was 0.96 indicating excellent accuracy of this test.

Comment(s)

Exclusion criteria were similar for all studies. Non-English speaking elderly patients were commonly excluded, decreasing the external validity of these screening tests given that emergency physicians will commonly encounter non-English speaking seniors and require the use of an interpreter. Further studies making the use of interpreters are necessary. Several studies varied in the way each cognitive screening test was administered in terms of who was administrating each test and in what order tests were given. Not all studies analyzed their screening tools using the same statistical method, which limits the ability to compare them. One major study weakness that was commonly encountered was the use of the MMSE as the criterion standard in detecting cognitive impairment. The MMSE is not ideal, as it has shown to have low sensitivity and overall accuracy towards mild cognitive impairment detection, and it is influenced by the education of the patient [5]. One out of the seven studies intentionally looked for mild cognitive impairment (MCI) using the MoCA as a criterion standard. The SBT was found to be most specific and useful clinically to rule in the possibility of MCI. Most of these tools are unable to distinguish between delirium and dementia; however their primary purpose in the ED would be to alert the emergency department physician or nurse to a possible cognitive dysfunction when considering treatment, management and disposition.

Clinical Bottom Line

There is a noticeable need to develop a cognitive screening tool that is highly sensitive, specific, brief and suitable for the ED. Until such a screening tool is created and validated, the evidence shows that it is reasonable to use tools such as the SBT, O3DY or the SIS to screen for cognitive impairment in the ED. These screening tools were able to display moderate sensitivity and specificity, and are brief, simple assessment tools suitable for any healthcare provider to administer in a fast paced environment. The CAM is the only screening tool available, validated in the ED, that distinguishes and detects delirium from other forms of cognitive impairment.

References

1. Carpenter CR, Bassett ER, Fischer GM et al. Four sensitive screening tools to detect cognitive dysfunction in geriatric emergency department patients: brief Alzheimer's Screen, Short Blessed Test, Ottawa 3DY, and the caregiver-completed AD8. Acad Emerg Med 2011;18(4):374-383
2. Carpenter CR, DesPain B, Keeling TN et al. The Six-Item Screener and AD8 for the Detection of Cognitive Impairment in Geriatric Emergency Department Patients Ann Emerg Med 2011; 57(6):653-61
3. Carpenter CR, Palatnik M, Koch S, Skebba T, Twaalfhoven E, Godbout A, Morris JC. Detecting mild cognitive impairment in geriatric emergency department patients: prevalence and diagnostic accuracy of three brief screening instruments (SAEM Abstracts) Acad Emerg Med 2011;18:S95-235.
4. Monette J, Galbaud du Fort G, Fung SH et al. Evaluation of the Confusion Assessment Method (CAM) as a screening tool for delirium in the emergency room Gen Hosp Psychiatry 2001;23(1):20-5.
5. Schofield I, Stott DJ, Tolson D et al. Screening for cognitive impairment in older people attending accident and emergency using the 4-item abbreviated mental test. European J Emerg Med 2010;17:340-2.
6. Wilber ST, Carpenter CR, Hustey FM. The Six-item Screener to Detect Cognitive Impairment in Older Emergency Department Patients Acad Emerg Med 2008;15(7):613-616.
7. Wilber ST, Lofgren SD, Mager TG et al. An Evaluation of Two Screening Tools for Cognitive Impairment in Older Emergency Department Patients Acad Emerg Med 2005;12(7):612-6.