QfQiosrptchsH

D

C5

©0

Plain-Radiographic Image Labeling:A Process to Improve Clinical

OutcomesKenneth T. Aakre, BS, RT(R), C. Daniel Johnson, MD

Purpose: To determine the rate of film-labeling errors and to describe a process for improved plain-film imagelabeling and the clinical outcomes from this process improvement.

Methods: Image-labeling errors (absent or incorrectly assigned left or right lateral identifier marker, absent orincorrect patient-identifying number, absent or incorrect examination date, incorrect marker placement, absenttechnologist initial marker, or incorrect body-part order) were measured among 2,536 consecutive plain-filmradiographs over a 2-week period. Following a process improvement initiative based on failure mode effective-ness analysis, left-side and right-side indicator markers, patient demographics, and date labels were identified asthe most common sources of error. An improvement initiative using larger and colored left and right lateralindicator markers, an automated process to label patient demographics, and direct patient verification ofidentification was begun. The numbers of labeling errors were again assessed in 2,421 consecutive plainradiographs over a 2-week period. The error rates before and after the improvement initiatives were compared.

Results: Plain-radiographic labeling errors occurred in 62 of 2,536 (2.4%) images before the improvementinitiative. Labeling errors were reduced to 17 of 2,421 (0.70%; 95% exact binomial confidence interval, 0.4%to 1.1%; P � .001, chi-square test) by using the improvement tools.

Conclusions: Plain radiographic image labeling can be improved using bar-code reading of patient demo-graphic information, linked to patient Digital Imaging and Communications in Medicine modality work listsand image printing. Patient verification of demographic information is key and can be electronically managed.Lateral marker identification can be improved with larger (more easily read) and color-coded indicators.

Key Words: Quality improvement, image-labeling error, patient verification, lateral marker identification,plain radiographs

J Am Coll Radiol 2006;3:949-953. Copyright © 2006 American College of Radiology

lafip

M

OdowoesprD

uality and safety issues in medicine have come to theorefront since the Institute of Medicine’s [1] Crossing theuality Chasm was published in 2001. National quality

nitiatives are gaining momentum, including in radiol-gy. Many safety issues identified by the Joint Commis-ion on Accreditation of Healthcare Organizations areelevant to the practice of radiology. In particular, correctatient identification and universal protocols are impor-ant topics in radiology and would translate broadly intoorrect image labeling. The potential for patient harm isigh if the wrong patient is identified or if the incorrectide is marked, leading to a wrong-side procedure.ighly reliable methods are needed to ensure the proper

epartment of Radiology, Mayo Clinic, Rochester, Minn.

Corresponding author and reprints: Kenneth T. Aakre, RT(R), Mayolinic, Department of Radiology, 200 SW First Street 55905, Rochester, MN

t5905; e-mail: aakre.kenneth@mayo.edu.

2006 American College of Radiology091-2182/06/$32.00 ● DOI 10.1016/j.jacr.2006.07.005

abeling of radiologic examinations. This paper describesstudy designed to assess a process for improved plain-lm image labeling and the clinical outcomes of thisrocess improvement.

ETHODS

ver a 2-week period, 2,536 consecutive plain-film ra-iographs were evaluated from a musculoskeletal radiol-gy section (Table 1, group 1). An image-labeling erroras defined if any of the following were identified: absentr incorrectly assigned left or right lateral identifier mark-r; absent or incorrect patient-identifying number; ab-ent or incorrect examination date; incorrect markerlacement; lack of technologist initial marker; or incor-ect body-part order (eg, left knee instead of right knee).uring normal clinical practice work flow, a quality con-

rol technologist reviewed all of the completed examina-

949

tetc

uip(odfobddaiaarws

qamsctf1mca

fiootsc2aisc

L

CstcsrIlitsl

P

Anrswp

cqcdtilped

crgrtri

950 Journal of the American College of Radiology/Vol. 3 No. 12 December 2006

ions before interpretation and identified any labelingrrors. Besides correcting any errors, the quality controlechnologist counted the number of images obtained andompared that with the number of errors tallied.

A failure mode effectiveness analysis (FMEA) grid wassed by the improvement team to determine which steps

n the image-labeling process were most detrimental toatient outcomes and the delivery of a quality productTable 2). Failure mode effectiveness analysis is a meth-dology for identifying reliability problems in a processuring the development process. Its aims are to detect theailure modes of a process, determine their effects on theperation, and identify actions to mitigate the failuresefore they occur. In essence, it is an engineering tool toesign failure out of a project. The FMEA procedureescribes the process step by step and the failure modesnd the downstream effects of each step, assigns a numer-cal score for each step, and determines the actions toddress potential failures. Usually, the numerical scoresre the products of the severity, the probability of occur-ence, and the likelihood of detection. In this study, theyere the products of probability and severity. These

cores were used to prioritize the corrective actions.The improvement team, composed of a radiologist, a

uality control technologist, an imaging technologist,nd a medical imaging technical service employee, deter-ined each image-labeling process step. Within each

tep, they identified possible errors (failure modes) thatould occur at each process step. Once the team agreedhat all possible error scenarios (failure modes) were listedor each process step, the team assigned a ranking value of

to 5 for the probability and the severity of the failureodes when they occurred. An assigned value of 1 indi-

ated a low probability of a failure mode occurring, and

Table 1. Summary of Labeling ErrorsError Group 1 Group 2

✓ LT or RT 11 2✓ Demographic data 2 6*✓ Date 6 0✓ No marker in image 8 1✓ Marker placement 11 4✓ Body part order 3 3✓ Other 21 1✓ Total Errors 62 1✓ Images 2536 2421✓ error rate 2.4% 0.70%✓ TOTAL ERROR REDUCTION OF 74%

��At the time of data collection, portable exam images wereusing the old method (lead number tapes as per group 1) forpatient demographics and date, which accounted for 4 of theseerrors.

n assigned value of 5 indicated a high probability of the p

ailure mode occurring. Similarly, an assigned value of 1ndicated a low level of severity when the failure modeccurred, and an assigned value of 5 indicated a high levelf severity when the failure mode occurred. The teamhen calculated the FMEA score by multiplying the as-igned value for each failure mode in the probabilityolumn by the value assigned in the severity column. Themost common errors were analyzed separately to devise

n error correction plan for each. However, after processmprovements, the technologist repeated the quality as-essment score, and the percentages of errors were recal-ulated collectively (Table 1, group 2).

eft-Side and Right-Side Indicator Markers

onventional left and right lateral (side) markers aremall lead letters fixed to a piece of tape placed adjacent tohe examination date. The letters were sometimes diffi-ult to read at a glance. Two changes were made in theide indicators, including enlargement of the left andight markers with 1-inch-high “L” and “R” indicators.n addition, each side was color coded blue and red foreft and right, respectively (Figure 1). According to ourmprovement results and interviews conducted withechnologists who used the enlarged colored markers,ide verification was improved by more easily reading theetters and confirming the correct side color.

atient Demographic and Date Labels

manual labeling device had been used in the past (leadumbers adhered to tape that was fixed to the computedadiographic cassette) and was identified as a commonource or error. To eliminate the human error associatedith generating these labels, the following automatedrocess was devised.First, patients’ demographics were selected via a bar-

ode reader by scanning the patients’ examination re-uest paperwork, which contained bar codes. This pro-ess would automatically generate the patients’emographic information and examination dates ontohe computed radiography modality via a Digital Imag-ng and Communications in Medicine modality workist. This step eliminated the manual production andlacement of the lead markers, because the modalitylectronically inserted the demographic information andates on the final images.Second, to ensure that an examination belonged to the

orrect patient, a computer monitor was installed in theadiographic examination room, so that the computer-enerated patient demographic information could be di-ectly verified by the patient. If the patient indicated thathe demographic information on the monitor was cor-ect, the technologist would initial a paper form indicat-ng that the process had been completed properly and

roceed with performing the examination. If the patient

Aakre, Johnson/Improved Radiology Image Labeling 951

Table 2. Image Labeling - Failure Modes Effect Analysis (FMEA)

Process Step Failure ModeProbabilityto Occur

Severity Whenit Happens FMEA��

Order is created/passed toRIMS�

1. wrong exam 1 1 12. absence of LT/RT†

indicator1 1 1

3. no indication of exam 1 1 1Desk personnel enters

indications, LT/RT1. enter wrong side LT/RT or

no side1 2 2

2. enter wrong examindication

1 2 2

Desk personnel makes leadnumber tape(demographics)

1. make the wrong date 2 5 102. make the wrong lead

number tape(demographics)

3 5 15

Technologist calls patient,verifies:

PatientExam orderDateLead tape

(demographics)Affected side

Technologist does not verify1. lead tape (demographics)

and date4 5 20

2. side 2 5 103. exam 2 5 104. patient 2 5 10

Tech wands RIMS examorder sheet

1. technologist forgets towand RIMS exam ordersheet

1 5 5

2. wands the incorrect RIMSexam order sheet

1 5 5

Technologist places leadtape (demographics)/side markers oncassette

1. incorrect placement 4 2 82. does not use copper when

needed2 2 4

3. does not place lead tape(demographics)/markers oncassette

3 5 15

Technologist labels imagewith LT/RT lateralmarker

1. no LT/RT marker on at all 3 5 15

2. incorrect one 4 5 20

Image Stacker QCs‡image, verifies correct:

LT/RTLead number

(demographics), anddate to image header

PatientRIMS exam order sheetImage order/orientationFolder

Does not verify:1. LT/RT 2 5 102. lead number

(demographics)/date2 5 10

3. patient 2 5 104. RIMS exam order sheet 2 5 105. image order/orientation 2 5 106. folder 2 5 10

Probability to occurUnlikely 1 - 5 LikelySeverity when it happensNot severe 1-5 Very severe�RIMS� Radiology Information Management System; †LT� Left, RT� Right; ‡ QCs� Quality Controls; ��FMEA�Failure Modes

Effects Analysis. In this case, a numerical score derived from the product of the severity score and the probability to occur score.

inpib

cvw

R

G

Sctraos

vmemwit

D

Tihietfiwpc

smsitst[v

acparp

tmofBtttsg

bfiib

Fn

952 Journal of the American College of Radiology/Vol. 3 No. 12 December 2006

ndicated that the demographics were incorrect, the tech-ologist would correct the error and repeat the process ofatient validation. If patients were too ill to verify theirdentities, technologists would verify this information onehalf of the patients.

This process takes less than or equal to 1 minute toomplete but engages patients directly by having themiew and authenticate their demographic information,hich becomes the image label.

ESULTS

roup 1 and 2 Data Collection and FMEA

ixty-two errors among 2,536 images (2.4%) were en-ountered in a 2-week period. The FMEA demonstratedhat errors involving the nonuse or misuse of the left andight lateral indicators (markers), patient demographics,nd examination dates were most detrimental to patientutcomes (Tables 1 and 2). The “other” category repre-

ig 1. Top: old right and left initial markers; bottom:ew right and left initial markers.

ents errors associated with cone-clipped and hard-to- w

isualize lateral indicator markers. A 2-week postimple-entation data collection was performed. Seventeen

rrors among 2,421 images (0.070%; 95% exact bino-ial confidence interval, 0.4% to 1.1%) were found in 2eeks. This difference in the error rates comparing pre-

mprovement and postimprovement procedures was sta-istically significant (P � .001, chi-square test).

ISCUSSION

he manual placement of critical image-labeling markerss associated with frequent errors. We found that in aigh-volume, plain-radiographic practice, critical label-

ng errors occurred in 2.4% of all images. This level ofrror is concerning, because major decisions regardinghe care of patients are made on the basis of radiographicndings. A finding attributed to the wrong patient, therong side, or the wrong date on an examination couldotentially have catastrophic implications for patientare.

The importance of these critical safety issues is empha-ized by their inclusion in many national safety improve-ent guidelines and requirements. The Joint Commis-

ion on Accreditation of Healthcare Organizations [2]ncludes correct patient identification and the elimina-ion of wrong-site, wrong-patient, and wrong-procedureurgery as key components of its 2005 and 2006 Na-ional Patient Safety Goals. The National Quality Forum3] specifically endorses standardized protocols to pre-ent the mislabeling of radiographs.

Our study indicates that the reliability of correctlyffixing critical image labels can be improved using color-oded, enlarged left and right lateral markers and directatient verification combined with a computer-gener-ted work list that is electronically linked to a computedadiographic system. This process enabled a 10-fold im-rovement (a 2� to 3� improvement in reliability).Other identification procedures have been established

o identify mislabeled images, but these have been imple-ented after errors have occurred. In these cases, radiol-

gists compare mislabeled images with other radiographsrom several patients to establish patients’ identities.halla et al [4] reported (in decreasing order of reliability)

he evaluation of “the first costotransverse articulations,he spinous processes of the lower cervical and upperhoracic spine, and the scapular wings” as key anatomictructures to assess for identifying mislabeled radio-raphs.

Fabian [5] reported on the use of an “identificationlocker” containing the phrase “exposure side” that isxed on one side of the imaging screen to cause a clear

mprint on an image. The identification blocker woulde used on standard image projections in conjunction

ith the traditional lead left and right lateral markers and

“a

aektoDittto

asrcgttscdDcipgitta

qtia

psfrtioppfel

R

1

2

3

4

5

6

Aakre, Johnson/Improved Radiology Image Labeling 953

can serve as a back up or cross check” to the current leftnd right lateral marking method.

Image-labeling improvements can also occur by usingDICOM modality work list in place of the manual

ntry of patient demographics into the modality via aeyboard. The use of a DICOM modality work list vir-ually eliminates typographical errors, which are prone toccur with human intervention. However, when using aICOM modality work list, a different error can be

ntroduced, in which the modality work list that displayshe patient demographics is incorrectly selected, placinghe incorrect patient demographics on the electronic pa-ient record [6]. In our study, 2 such demographic errorsccurred.

Further improvements in correct film identificationre needed. This study evaluated only a single musculo-keletal, plain-radiographic practice. Other types of er-ors that were not identified in this study may exist withross-sectional images and interventional and angio-raphic practices. Reducing manual processes seems keyo our reported improvement and a requirement for fur-her reductions in image-labeling errors. Although 2teps in our routine process were replaced by both bar-ode reading of patients’ examination paperwork andownloading patients’ demographic information from aICOM modality work list, additional errors could oc-

ur upstream in the work process. These errors couldnclude the incorrect generation of patients’ examinationaperwork or DICOM modality work lists. Technolo-ists must recognize the importance of verifying patients’dentification numbers and names (demographics) andake the necessary time to explain the process to patientso obtain informed responses. In cases in which patients

re unable to respond or the integrity of their responses is

uestionable, technologists must be willing to take theime and effort to reverify. Electronic solutions that elim-nate these human steps will likely result in further im-ge-labeling improvements.

As radiology improvement projects become common-lace, we hope that this information is helpful as ourpecialty begins to benchmark best practices. We lookorward to reading about improved results over thoseeported here; critical evidence is needed today to raisehe standards of our entire specialty. Plain-radiographicmage labeling can be improved using bar-code readingf patient demographic information, which is linked toatient DICOM modality work lists and final imagerinting. Patients’ verification of their demographic in-ormation is key and can be electronically managed. Lat-ral (side) marker identification can be improved witharger (more easily read) and color-coded indicators.

EFERENCES

. Committee on Quality of Health Care in America. Institute of Medicine,Crossing the quality chasm: a new health system for the 21st century.Washington, DC: National Academy Press; 2001.

. Joint Commission on Accreditation of Healthcare Organizations. 2005and 2006 national patient safety goals. Available at: http://www.jointcommission.org. Accessed February 9, 2006.

. National Quality Forum. Safe practices for better healthcare: a consensusreport. Available at: http://www.qualityforum.org. Accessed February 9,2006.

. Bhalla M, Noble ER, McLean FM, Norris DM, Hicklin OM, Henschke C.Radiographic features useful for establishing patient identity from improp-erly labeled portable chest radiographs. J Thorac Imaging 1994;9:35-40.

. Fabian CE. Knowing right from left on x-rays: a way to minimize errors oflaterality. J Appl Radiol 2005;34:19-24.

. Kuzmak PM, Dayhoff RE. Minimizing Digital Imaging and Communica-tions in Medicine (DICOM) modality worklist patient/study selection

errors. J Dig Imag 2001;14(2 suppl 1):153-7.