clinical pr acticehavioral Research, 2 and in the judgment <strong>of</strong> manyexperts, it reflects how courts actually deal withthese cases.Approaches <strong>to</strong> <strong>Assessment</strong>Given the requirement <strong>of</strong> competence for validinformed consent, the assessment <strong>of</strong> the patient’scapacity <strong>to</strong> make decisions is an intrinsic aspect<strong>of</strong> every physician–patient interaction. Usually, theassessment will be implicit, since in the absence<strong>of</strong> a reason <strong>to</strong> question a patient’s decision making,the presumption <strong>of</strong> competence will prevail. Whenan explicit competence evaluation is required, physiciansshould be aware <strong>of</strong> the relevant criteria andshould be encouraged <strong>to</strong> use a structured approach<strong>to</strong> assessment. In one set <strong>of</strong> studies, simply providingphysicians with specific legal standards <strong>to</strong>guide their judgments, similar <strong>to</strong> the criteria inTable 1, significantly increased interrater agreement(the kappa statistic for agreement increasedfrom 0.14 <strong>to</strong> 0.46). 30,37 Another research groupfound that asking physicians and nurses <strong>to</strong> use asystematic set <strong>of</strong> questions for competence assessmentled <strong>to</strong> a high rate <strong>of</strong> agreement with expertjudgments. 38 Published question sets with goodface validity are readily available, and they shouldbe used <strong>to</strong> guide clinical assessments. 31,38 Table1 includes sample questions.Any physician who is aware <strong>of</strong> the relevant criteriashould be able <strong>to</strong> assess a patient’s competence.Indeed, treating physicians may have theadvantage <strong>of</strong> greater familiarity with the patientand with available treatment options. Psychiatricconsultation may be helpful in particularly complexcases or when mental illness is present. Althougha simple instrument <strong>to</strong> screen patients forimpaired capacity would facilitate the identification<strong>of</strong> patients who may require more detailedassessment, <strong>to</strong> date the quest for a brief neuropsychologicalscreening instrument has not yieldedconsistent findings. However, the Mini–MentalState Examination (MMSE) has been found <strong>to</strong> correlatewith clinical judgments <strong>of</strong> incapacity, 11 andit may have some use in identifying patients at thehigh and low ends <strong>of</strong> the range <strong>of</strong> capacity, especiallyamong elderly persons with some degree <strong>of</strong>cognitive impairment. 39,40 MMSE scores rangefrom 0 <strong>to</strong> 30, with lower scores indicating decreasingcognitive function. No single cut<strong>of</strong>f scoreyields both high sensitivity and high specificity.MMSE scores <strong>of</strong> less than 19 are highly likely <strong>to</strong>be associated with incompetence 39,40 ; studies varyin suggesting that scores <strong>of</strong> 23 <strong>to</strong> 26 or higherare strongly indicative <strong>of</strong> competence. 11,38-40In an effort <strong>to</strong> further standardize and henceincrease the reliability and validity <strong>of</strong> competenceevaluations, several more formal assessment instrumentshave been developed. Their characteristicsand psychometric properties have been describedelsewhere. 41,42 The most widely used <strong>of</strong>these instruments is the MacArthur <strong>Competence</strong><strong>Assessment</strong> Tool for <strong>Treatment</strong>, a structured interviewthat, unlike many other assessment instruments,incorporates information specific <strong>to</strong> a givenpatient’s decision-making situation. 43 Quantitativescores are generated for all four criteria related <strong>to</strong>decision-making capacity, but evalua<strong>to</strong>rs must integratethe results with other data in order <strong>to</strong> reacha judgment about competence. The high interrateragreement on these scores that has been reportedby a number <strong>of</strong> research groups 11,41,44 is usuallygreater than that reported in studies <strong>of</strong> systematicclinical assessment.The MacArthur test takes approximately 20minutes <strong>to</strong> administer and score, assuming thatthe person who administers and scores the test hasexperience with the format and scoring criteria.Given the extra time associated with the use <strong>of</strong>assessment instruments, they would appear <strong>to</strong>have particular value when assessment is especiallydifficult or when a case is likely <strong>to</strong> be resolvedin court, where the availability <strong>of</strong> systematicdata collected in a standard format may beuseful <strong>to</strong> a nonmedical fact finder. However, evenif scores are not generated, the use <strong>of</strong> a structuredinstrument can help guide the clinical assessmentprocess.Whatever approach <strong>to</strong> assessment is used, examinersshould first ensure that patients have beengiven the information that is relevant <strong>to</strong> makingan informed decision about their treatment. Typically,such disclosure includes the nature <strong>of</strong> thepatient’s condition, the nature and purpose <strong>of</strong> theproposed treatment, and the risks and benefits <strong>of</strong>the proposed treatment and <strong>of</strong> alternative treatments,including the option <strong>of</strong> no treatment at all. 1Since such disclosure cannot be presumed, eitherthe evalua<strong>to</strong>r should ask a physician responsiblefor the patient’s care <strong>to</strong> disclose the relevant informationagain in the evalua<strong>to</strong>r’s presence or then engl j med 357;18 www.nejm.org november 1, 2007 1837Downloaded from www.nejm.org at CHILDRENS HOSPITAL-LIB FAGAN PLAZA on November 2, 2007 .Copyright © 2007 Massachusetts Medical Society. All rights reserved.