• unlimited access with print and download
    $ 37 00
  • read full document, no print or download, expires after 72 hours
    $ 4 99
More info
Unlimited access including download and printing, plus availability for reading and annotating in your in your Udini library.
  • Access to this article in your Udini library for 72 hours from purchase.
  • The article will not be available for download or print.
  • Upgrade to the full version of this document at a reduced price.
  • Your trial access payment is credited when purchasing the full version.
Buy
Continue searching

Effect of package label characteristics on pharmacists' visual perception of drug names

ProQuest Dissertations and Theses, 2009
Dissertation
Author: Todd Chermak
Abstract:
To highlight portions of a drug name that might be confused with other similar drug names, manufacturers use a variety of different stylistic label formats (e.g., bolding, italics, highlighting, underling, capitalization) on primary package labels of prescription drugs. Two experiments were conducted with pharmacists as participants to determine whether two high occurrence label formats (letter bolding and the use of mixed case text) had an impact on pharmacists' accuracy and reaction time in a speeded same-different judgment task. Both experiments examined the effect of orthographic similarity between drug names and stylistic label format (bolding and mixed case text) on accuracy and reaction time. Experiment 2 also examined the effect on accuracy and reaction time of the specific location of the stylistic format within the drug name. The results of both experiments highlighted the strong effect of orthographic similarity on accuracy and reaction time of pharmacists. Pharmacists were less accurate and exhibited longer reaction times when the drug names were orthographically similar. Mixed case text did not significantly impact accuracy or reaction time in either experiment despite the growing belief that these text enhancements reduce dispensing errors. The placement of the text enhancement within the drug name did not significantly effect accuracy or reaction time of pharmacists. Given the relatively high frequency various label formats are recommended by regulators and safety related institutions, it is important the pharmaceutical manufacturers and regulators implement evidence based models to establish regulatory policy related to package labeling. Clear regulatory policy based on rigorous scientific data will help ensure the use of any package label format to mitigate name confusion will provide the greatest patient safety benefit.

TABLE OF CONTENTS I. INTRODUCTION 1 A. BACKGROUND AND PROBLEM STATEMENT 1 B. APPROVAL OF DRUG NAMES - A REGULATORY OVERVIEW 6 C. STANDARDS 11 D. PURPOSE OF THE STUDY 12 II. CONCEPTUAL FRAMEWORK 13 A. INTERACTIVE ACTIVATION MODEL 13 B. THE EFFECTS OF FREQUENCY AND NEIGHBORHOOD CHARACTERISTICS ON WORD PERCEPTION 18 1. The Effects of Frequency on Word Perception 18 2. The Effect of Similar Neighborhoods on Word Perception 18 C. DRUG NAME CONFUSION AS A SOURCE OF MEDICATION ERRORS 21 D. EMPIRICAL STUDIES LINKING NAME SIMILARITY AND CONFUSION ERRORS 27 E. DRUG NAME CONFUSION MITIGATION STRATEGIES 32 F. RECOMMENDATIONS FOR MINIMIZING DRUG NAME CONFUSION 48 G. RESEARCH QUESTIONS 53 H. STUDY HYPOTHESES 54 HI. METHODS 56 A. PARTICIPANTS 56 B. SPEEDED SAME-DIFFERENT TASK 58 C. STUDY ONE EXPERIMENTAL DESIGN 59 1. Different Drug Name Trials 60 2. Same Drug Name Trials 71 D. STUDY TWO EXPERIMENTAL DESIGN 71 E. EXPERIMENTAL PROCEDURES 73 1. Experimental Software and Hardware 73 2. Setting for Data Collection 75 3. Data Collection 77 F. DATA ANALYSIS 78 G. POWER ANALYSIS 82 IV. RESULTS 84 A. EXPERIMENT 1 84 1. Descriptive Statistics 84 2. Hypothesis Tests 86 B. EXPERIMENT 2 97 1. Descriptive Statistics 97 2. Hypothesis Testing 98 V. DISCUSSION 114 A. DISCUSSION OF TEXT ENHANCEMENTS 115 B. DISCUSSION OF ORTHOGRAPHIC SIMILARITY 119 v

C. CONTRIBUTION OF THE NEW EXPERIMENTAL DATA 119 D. REGULATORY POLICY IMPLICATIONS 121 E. STUDY LIMITATIONS 123 F. FUTURE RESEARCH 124 G. CONCLUSIONS 125 CITED LITERATURE 126 APPENDICES 131 APPENDIX A - IRB APPROVAL 131 APPENDIX B - PHARMACIST RECRUITING LETTER 135 APPENDIX C - INFORMED CONSENT 137 APPENDIX D - PHARMACIST DEMOGRAPHIC QUESTIONNAIRE 138 APPENDIX E - DRUG NAMES INCORPORATED INTO EXPERIMENT 1 142 APPENDIX F - DRUG NAMES INCORPORATED INTO EXPERIMENT 2 146 VITA 148 VI

LIST OF TABLES TABLE I FDA TALL MAN LETTER PAIRS 33 TABLE II 3X3 COMPLETELY WITHIN PARTICIPANT FACTORIAL DESIGN 62 TABLE III EXAMPLE STUDY DRUG NAMES WITH JARO-WINKLER DISTANCE SCORES 63 TABLE IV DRUGNAME LISTS 65 TABLE V REPEATED MEASURES FACTORIAL DESIGN 72 TABLE VI EXAMPLE OF EXPORTED SUPER LAB DATA TO MICROSOFT EXCEL 79 TABLE VII CATEGORICAL DEMOGRAPHIC DATA 84 TABLE VIII QUANTITATIVE DEMOGRAPHIC DATA 85 TABLE IX MEANS, STANDARD ERRORS AND 95% CONFIDENCE INTERVALS FOR MAIN EFFECTS ON ACCURACY 87 TABLE X MEANS AND STANDARD ERRORS AND CONFIDENCE INTERVALS FOR INTERACTION EFFECTS OF STYLISTIC FEATURE AND ORTHOGRAPHIC SIMILARITY ON ACCURACY 90 TABLE XI REPEATED MEASURES ANOVA FOR REACTION TIME, MAIN EFFECTS 92 TABLE XII INTERACTION EFFECTS OF FORMAT FEATURE AND SIMILARITY ON REACTION TIME 95 TABLE XIII CATEGORICAL DEMOGRAPHIC DATA 97 TABLE XIV QUANTITATIVE DEMOGRAPHIC DATA 98 TABLE XV MEANS, STANDARD ERRORS AND CONFIDENCE INTERVALS FOR MAIN EFFECTS OF SIMILARITY, FORMAT FEATURE AND STYLISTIC PLACEMENT ON ACCURACY 100 TABLE XVI MEANS AND STANDARD ERRORS AND CONFIDENCE INTERVALS FOR INTERACTION EFFECTS OF STYLISTIC FEATURE, ORTHOGRAPHIC SIMILARITY AND FORMA PLACEMENT ON ACCURACY 103 TABLE XVII REPEATED MEASURES ANOVA FOR REACTION TIME, MAIN EFFECTS 107 TABLE XVIII MEANS, STANDARD ERRORS AND CONFIDENCE INTERVALS FOR FORMAT FEATURE, SIMILARITY, AND FORMAT PLACEMENT FOR REACTION TIME 110 vii

LIST OF FIGURES FIGURE 1: INTERACTIVE ACTIVATION MODEL (IAM) 14 FIGURE 2: BOTTOM-UP ACTIVATION 16 FIGURE 3 PRESCRIPTION DRUG LABEL FOR GLYBURIDE 34 FIGURE 4 TEXT FORMAT ALGORITHM AND CONVERSION RULES 71 FIGURE 5 CONFERENCE ROOM LAYOUT 76 FIGURE 6: PERCENT ACCURACY OF PARTICIPANTS 81 FIGURE 7 DISTRIBUTION OF PHARMACY PRACTICE SETTINGS 85 FIGURE 8 EFFECT OF ORTHOGRAPHIC SIMILARITY ON ACCURACY 88 FIGURE 9 EFFECT OF STYLISTIC FORMAT FEATURE ON ACCURACY 89 FIGURE 10 INTERACTION EFFECTS OF LABEL FEATURE AND ORTHOGRAPHIC SIMILARITY ON ACCURACY 90 FIGURE 11 MAIN EFFECT PLOT FOR SIMILARITY 93 FIGURE 12 MAIN EFFECT PLOT FOR STYLISTIC LABEL FEATURE 94 FIGURE 13 INTERACTION EFFECTS OF LABEL FEATURE X ORTHOGRAPHIC SIMILARITY 95 FIGURE 14 EFFECT OF ORTHOGRAPHIC SIMILARITY ON ACCURACY 100 FIGURE 15 EFFECT OF STYLISTIC LABEL FORMAT ON ACCURACY 101 FIGURE 16 MAW EFFECT PLOT FOR LABEL FORMAT PLACEMENT 102 FIGURE 17 INTERACTION EFFECTS OF LABEL FEATURE AND SIMILARITY ON ACCURACY 104 FIGURE 18 INTERACTION EFFECT OF SIMILARITY AND FEATURE PLACEMENT ON ACCURACY .... 104 FIGURE 19 INTERACTION EFFECT OF FEATURE PLACEMENT AND FEATURE FORMAT ON ACCURACY 105 FIGURE 20 EFFECT OF ORTHOGRAPHIC SIMILARITY ON REACTION TIME 108 FIGURE 21 EFFECT OF LABEL FEATURE ON REACTION TIME 109 FIGURE 22 EFFECT OF FEATURE PLACEMENT ON REACTION TIME 109 FIGURE 23 INTERACTION EFFECTS OF ORTHOGRAPHIC SIMILARITY AND STYLISTIC LABEL FORMAT ON REACTION TIME I l l FIGURE 24 INTERACTION EFFECTS OF ORTHOGRAPHIC SIMILARITY AND LABEL FORMAT PLACEMENT ON REACTION TIME 112 FIGURE 25 INTERACTION EFFECTS OF STYLISTIC LABEL PLACEMENT AND STYLISTIC LABEL FORMAT ON REACTION TIME 112 Vl l l

LIST OF ABBREVIATIONS ASTM CBER CDER CFR FDA FDAAA FMEA IAM ICH IRB ISMP NCC-MERP NDA NPSF OND OTC PDUFA USAN USP International Standards Worldwide Center for Biologies Evaluation and Research Center for Drug Evaluation and Research Code of Federal Regulations Food and Drug Administration Food and Drug Administration Amendments Act Failure Modes and Effects Analysis Interactive Activation Model International Council on Harmonization Institutional Review Board Institute for Safe Medication Practices National Coordinating Council for Medication Error Reporting and Prevention New Drug Application National Patient Safety Foundation Office of New Drugs Over the Counter Prescription Drug User Fee Act United States Adopted Name United States Pharmacopeia ix

SUMMARY To highlight portions of a drug name that might be confused with other similar drug names, manufacturers use a variety of different stylistic label formats (e.g., holding, italics, highlighting, underling, capitalization) on primary package labels of prescription drugs. Two experiments were conducted with pharmacists as participants to determine whether two high occurrence label formats (letter holding and the use of mixed case text) had an impact on pharmacists' accuracy and reaction time in a speeded same-different judgment task. Both experiments examined the effect of orthographic similarity between drug names and stylistic label format (holding and mixed case text) on accuracy and reaction time. Experiment 2 also examined the effect on accuracy and reaction time of the specific location of the stylistic format within the drug name. The results of both experiments highlighted the strong effect of orthographic similarity on accuracy and reaction time of pharmacists. Pharmacists were less accurate and exhibited longer reaction times when the drug names were orthographically similar. Mixed case text did not significantly impact accuracy or reaction time in either experiment despite the growing belief that these text enhancements reduce dispensing errors. The placement of the text enhancement within the drug name did not significantly impact accuracy or reaction time of pharmacists. Given the relatively high frequency various label formats are recommended by regulators and safety related institutions, it is important the pharmaceutical manufacturers and regulators implement evidence based models to establish regulatory policy related to package labeling. Clear regulatory policy based on

rigorous scientific data will help ensure the use of any package label format to mitigate name confusion will provide the greatest patient safety benefit.

I. INTRODUCTION A. Background and Problem Statement Medication errors are common, costly and harmful.1 The broad definition of a medication error is "any error occurring in the medication-use process."1 Medication errors include items such as administering the wrong drug or dosage form as well as failure by the caregiver to provide the medication or failure by the patient to take the medication. The Food and Drug Administration (FDA) and the Pharmaceutical Industry have considerable interest in the topic of medication errors2"12. Their strong interest is in part derived from some of the limitations associated with generating complete safety data during clinical trials. During the clinical trial phase of development, the number of human subjects exposed to the new drug generally increases from phase I to phase III of development as a drug sponsor (e.g., a pharmaceutical company) and regulators evaluate safety and efficacy data. However, there are some limitations associated with clinical trials, one of which includes the relatively small number of subjects exposed to a drug (-3000-4000 subjects) compared to the population of individuals who might use the drug after FDA approval (e.g., millions of individuals). As a result, post marketing surveillance systems have been established during the commercialization phase of a drug to continually monitor safety data to ensure the drug can be safely used in the broader patient population. One aspect of the post marketing surveillance system includes monitoring of medication errors. 1

Safety problems associated with a drug are often reported using adverse event reporting systems. In 2008, FDA received more than one half million adverse events, which have been steadily increasing since 1999. In the United States, adverse event reporting is voluntary and can be reported in a variety of ways. Common mechanisms for reporting adverse events by a patient or caregiver, is via the manufacturer (e.g., a medication error hotline) or directly to the FDA as part of their MED Watch program. Once a new safety concern is detected, the sponsor and FDA work closely to evaluate options that might improve the safe use of the drug (e.g., adding new safety information to the package insert portion of the labeling). A number of private organizations have been established over the last several years to help address the issue of medication errors. Their efforts have largely taken the form of identifying new safety signals and communicating that new safety information to healthcare providers and patients. Examples of these organizations include Institute for Safe Medication Practices (ISMP), National Patient Safety Foundation (NPSF) and the National Coordinating Council for Medication Error Reporting and Prevention (NCC- MERP). An important objective of each organization involves improving the safe use of drugs. Cohen outlined several sources of medication error in his book Medication Errors. The primary causes outlined by Cohen include the following (in bold text) along with some typical examples in each category:13'14 2

• Poor handwriting on written prescriptions: Illegible writing such that drugs like Keppra and Kaletra may be confused • Drugs with similar names: drugs with similar names for example the drugs Cozaar and Colase • Errors associated with trailing zeros and decimal points: confusing .1 mg for lmg. • Substitution of several dose systems: confusing 1/200 grain (0.3 mg) with 2 times 1/100 grain tablets (1.2 mg dose) • Abbreviations: BD (bedtime) mistaken for BID (twice daily) • Ambiguous or incomplete orders: Drugs written for administration over 4 days are provided everyday for 4 days as was reported for cyclophosphamide. • Poor distribution practices: Lack of a unit dose process involving redundant order-dispensing verification in many pharmacy practice settings • Dose miscalculations: small doses in pediatric patients are often cited as prone to miscalculation • Confusing and/or similar labeling: Similar package labeling, doses, drug names • Incorrect route of administration: Ear drops (written as AD, AS or AU) dispensed into the eye (confused as OD, OS, OU) • Lack of patient education: Ensuring patients understand their medication and know how to take each medication can minimize errors 3

One specific area of medication errors outlined by Cohen that has received increased attention in recent years involves administration of the wrong drug, specifically as a result of drug name confusion. In fact, it has received so much recent attention that FDA established a pilot program as part of the funding associated with the most recent Prescription Drug User Fee Act (PDUFAIV), which was reauthorized under the Food and Drug Administration Amendments Act (FDAAA) of 2007.15 The pilot program outlines key studies and possible strategies for understanding how new drug names may interact and be confused with existing prescription drug names. There is strong evidence in the literature drug name confusion is a significant source of medication errors and warrants further research. Confusion involving drug names, packages and labels has historically been implicated in 25% to 33% of all spontaneously reported adverse events (i.e., those errors reported voluntarily by clinicians, patients, risk managers and others).14"16 More recently, an analysis of USP's MedMarx database suggested that name confusion may account for 1.4% of all error reports in hospitals.17 Observational studies of dispensing in outpatient pharmacies suggest the rate of actual wrong drug errors—the type most likely to be the result of name, label and package confusion—is roughly 0.13%. With 3.8 billion prescriptions dispensed in 2007, that translates to 4.9 million wrong drug errors per year in the U.S.19 Not surprisingly, several ongoing programs of research have sought to better understand how improved design of drug names, labels and packages might reduce the rate of confusion errors.1"18'20"40 4

The strong research emphasis associated with drug name confusion is justified due to the potential harm to patients. There have been several cases reported to FDA in recent years that highlight the potential consequences of drug name confusion. For example, a 50 year old women was hospitalized after receiving Flomax, used to treat prostate enlargement, rather than Volmax to treat her bronchospasms.4 In the same FDA publication, it was reported that an 8-year-old boy died after a suspected treatment with methadone rather than methylphenidate. More recently, there has been reported confusion between two topical creams: Kuric (ketoconazole) and Carac (fluorouracil). Kuric is used to treat fungal infections and seborrheic dermatitis. Carac is used to treat multiple actinic or solar keratoses of the face and anterior scalp.7 The specific case involved a patient receiving Carac rather than Kuric for their fungal infection. The patient developed severe rash, peeling of the skin and secondary infection at the application site and surrounding areas. There has been considerable interest by the Food and Drug Administration (FDA) and the pharmaceutical industry over the last several years in the possibility of implementing various package label (the part of the labelling directly affixed to the primary container of the drug that includes the drug name and other required information) format changes to minimize drug name confusion during the drug dispensing process.21'25'26'40 Some of these techniques include the use of color, holding and capital letters to underscore portions of the drug name to avoid confusion and help identify the correct drug during dispensing. The type setting and font size has been the subject of historic research, because these attributes can impact the legibility of the package information.41'42 The use 5

of mixed upper and lower case letters ("Tall Man Letters") to improve their effectiveness in accurate name identification is an example of a recent research study aimed at evaluating package label features that can reduce medication errors.25'26 It stands to reason better designed names, labels and packages have the potential to significantly reduce medication errors. Unfortunately, empirical evidence is lacking about the relationship between specific label attributes (e.g., color, boldface type, italics, capitalization, etc.) and dispensing accuracy. Consequently, it is impossible to know which characteristic, or combination of characteristics, will most effectively minimize the risk of confusion. This dearth of empirical data not only precludes an understanding of the characteristics or combination of characteristics that would be most helpful in reducing medication errors, it also prevents identification of any of the characteristics that may actually result in more confusion or patient harm. B. Approval of Drug Names - A Regulatory Overview There are regulatory requirements associated with obtaining a drug name and subsequently using that name on a prescription drug label. The FDA Office of New Drugs (OND) reviews proprietary names with consultation from other areas within the Center for Drug Evaluation and Research (CDER). Other CDER offices include the Office of Drug Safety, Office of Medical Policy, Division of Drug Marketing, Advertising and Communication (DDMAC) and perhaps review with the Center for Biologies Review (CBER), if necessary. The primary areas of focus during the review of a new drug name include drug safety and the prevention of medication errors and promotion to avoid false or misleading claims (per requirements in 21CFR 210.10(c)). 6

There are no other specific regulatory requirements published to provide guidance in this area. The FDA relies on internal processes to help guide the review of a drug name to avoid confusion in the marketplace (i.e., medication errors) as well as prevent false or misleading claims. After a series of public meetings to solicit input on the approach for a proprietary name approval process, FDA published a concept paper outlining a pilot program planned to begin at the end of 2009 (part of PDUFAIV). Under PDUFA IV, there is a focus to both broaden and strengthen FDA's drug safety program. FDA's goals as part of PDUFA IV involve reducing medication errors associated with look-alike and sound-alike drug names, unclear label abbreviations, acronyms, dose designations and error-prone label and package designs.5 PDUFA programs will continue to be funded via user fees collected from industry sponsors. The key benefits of the pilot program involve moving the assessment of drug names to an earlier stage in the development process and transitioning the name approval process from one, which is primarily qualitative to a more quantitative approach based on the latest science. Drug names entered in the pilot can be submitted to FDA as early as the end of phase 2 of drug development. Currently, drug names are approved as part of the new drug application (NDA) review process and often a drug name approval is one of the last discussion points between FDA and the application sponsor. If a name is rejected at this late stage in the approval process, it can result in millions of dollars in wasted packaging and launch materials, not to mention delays in the approval timeline. 7

FDA's proposed review of names includes a promotional review as well as a safety review. The safety review for prescription drugs consists of a preliminary screening, a United States Adopted Name (USAN) stem search evaluation, an assessment of orthographic (written text) and phonologic (sound) similarity to other approved drugs, a review of medication error data, name similarity studies and failure mode and effects analysis.5 The preliminary screening consists of the following areas: 1. Dosing interval: Verify that the dosing interval is not incorporated into the name of the drug (e.g., drug name QD) 2. Dosage form/route of administration: Verify that the dosage form is not incorporated into the name of the drug (e.g., Lortabs) 3. Medication and/or product name abbreviations: Verify that common prescription abbreviations are not incorporated into the name of the drug (e.g., Name OD for right eye may be mistaken for once daily or QD). 4. Names that suggest the composition of the product: Verify that components of the formulation are not included in the name suggesting content or a particular action. The United States Adopted name is often referred to as the generic or nonproprietary name.43 Names are designated by USAN to indicate pharmacologic or chemical traits of the drug. For example, Angiotensin-Converting Enzyme (ACE) inhibitors are a class of antihypertensive agents with USAN names that end in "pril." USAN names such as enalapril and lisinopril are compounds that fall into this class of drugs. Because a portion 8

of the US AN name communicates information to the prescriber, FDA wants to avoid using these stems in a proprietary name to avoid potentially confusing several classes of drugs. The proposed FDA process includes a step to assess the orthographic and phonologic similarity to other FDA approved drags.5 This includes a more qualitative assessment by comparing handwriting of the drug name to other drugs with similar names as well as quantitative methodologies that may involve computational applications (e.g., bigram, trigram similarity methods). Medication error data are also part of the proposed safety review of a drug name.5'44 FDA uses existing databases to search for relevant error reports that can help make an informed decision about proposed proprietary names. This may be particularly useful when a new drug name is being introduced, which contains an existing active ingredient. Error reports related to the approved product can be used to determine potential issues associated with the proposed product containing the same active ingredient. The database may also be useful for a new drug within a particular therapeutic class. The therapeutic class may be beneficial in identifying common errors that should be avoided. One of the key new components involved in proposing a new proprietary name includes a proposal for application sponsors to run name simulation studies. Simulation studies are intended to test a variety of practice conditions (e.g., inpatient pharmacy written order, 9

fax order). FDA is proposing a minimum of 20 scenarios representing various prescribing conditions be included in the simulations. Finally, as part of the safety evaluation, FDA is proposing that failure mode and effects analysis (FMEA) be incorporated into the proposed design of drug names. FMEA involves examining possible ways a failure or in this case an error can occur.5'44 Steps in the process include identifying failure modes, failure effects and perhaps a list of mitigating steps that can help prevent a failure from occurring. With regard to regulatory requirements related to how drug names are presented on labels, the Code of Federal Regulations (CFR) provides a series of labeling requirements for prescription drugs in 21CFR201. These requirements include basic information that must be present on the primary package label of a prescription drug. Although the basic requirements are outlined in the regulations, detailed, specific guidance about the prescription drug primary package label format is lacking. For example, the Code of Federal Regulation specifies in 21CFR201.10(g) the established name of the drug must be at least half the size of the proprietary name, but other common textual format features are not specified. In summary, there is currently very little in the way of specific regulatory requirements associated with establishing a drug name and then presenting the drug name on a label, though there is recent evidence this paradigm is changing.5 The research associated with identifying confusing drug names and label format techniques used to manage risk 10

associated with existing confusable drug names is not guided by regulatory requirements; however, scientific research can play a role in shaping future requirements. FDA's more recent PDUFA IV pilot program is evidence science may play a greater role in the drug name approval process in the future. C. Standards International standards have been proposed to help enhance the identification of drug names on labels (ASTM D6398-01, "Standard Practice to Enhance Identification of Drug Names on Labels," 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959). The standards outlined in ASTM D6398-01 include information related to label color, layout and contrast between the drug name and the background (e.g., black text on a white background and blue text on a yellow background). The international standard also includes suggestions related to the use of upper and lower case letters, use of boldface fonts in the drug name to add emphasis, and overall spacing of the drug name relative to the remaining label information to improve legibility. The published standards reference general experimental work on human visual processing, but cite no evidence that adherence to these standards would impact the rate of pharmacy dispensing errors. 11

D. Purpose of the Study The purpose of this dissertation involves: (1) gaining an understand of the various label format features used to differentiate prescription drug names have on primary package labels and (2) evaluating the impact these label format features on pharmacists' ability to distinguish between a pair of drug names with varying levels of visual similarity. 12

II. CONCEPTUAL FRAMEWORK A. Interactive Activation Model The research described in this dissertation is based on concepts drawn from basic research in visual word recognition. Conceptual theory related to how individuals perceive letters and words can influence decision-making in drug dispensing activities. To better understand how to minimize drug name confusion, it is important to understand the theoretical basis for how words and letters are perceived. Grainger describes word recognition as "a set of operations that are repeated in the brain as a skilled reader scans a set of text to make sense of letter sequences and spaces."45 The operational steps in word recognition involve selecting a best word match by comparing physical signals (i.e., text) with abstract representations stored in long-term memory.45'46 Rumelhart and McClelland's Interactive Activation Model (IAM) is a highly influential theoretical model in the field of visual word recognition.47'48 A simplistic diagram of the model, which was reproduced from Rumelhart and McClelland is illustrated in Figure l.47 13

Top-Down Visual Processing d Word Level c Letter Level c Feature Level i Excitatory connection Inhibitory connection Bottom Up Visual Processing Figure 1: Interactive Activation Model (IAM) Adopted from Rumelhart and McClelland. 14

The model describes two mechanisms of word perception: bottom-up recognition of words and top-down recognition of words.47'48 A bottom-up mechanism, recognition of words starts with activation at the feature level of letters (i.e., the characteristic lines and curves that form the letter). Nodes at the feature level are activated when a letter node consistent with that feature is excited, and conversely, nodes at the feature level are inhibited when letter notes are inconsistent with the feature presented. This is illustrated in Figure 1 as arrows representing excitatory effects and lines with circles represent inhibitory nodes. At the letter level, nodes for letters compete with each other based on their serial position within a word. Each letter node competes with other nodes for those word nodes consistent with the serial position of the letter and inhibits nodes inconsistent with the letter serial position. Finally, word nodes compete with other word nodes and are either excited or inhibited based on letters consistent with the particular word. Essentially, the bottom-up model of word recognition starts with recognizing the feature of a letter, then letters and finally words consistent with letters in a particular serial position. A simple example is presented in Figure 2. 15

Word Level Letter Level Excitatory connection Inhibitory connection Figure 2: Bottom-Up Activation A T In the top-down mechanism of word recognition, letter level nodes can be activated from excitatory links from word nodes. Grainger describes this phenomena as the word superiority effect (i.e., letters can be detected more accurately in a word compared to a letter in isolation or in a non-word/letter string).46 Within the I AM model there are two models that can help account for the word superiority effect. The first involves cascade processing and the second involves interactive processing.46 Both of these models can help explain the word superiority effect. 16

In the cascade processing model, word level activation builds up faster than letter level activation because the word itself (assuming it is a real and recognized word) helps identify the component letters via orthographic (written text) representation stored in long-term memory. This is true even if letter representations receive some level of bottom-up activation first. There is a greater amount of activation at the word level when many inputs come together compared to separated inputs at the letter level of input.46 This theoretical component was supported through experimentation with real words and pseudowords where letters in real words were identified faster than with pseudowords.46 A second method of describing word superiority involves interactive processing. Similar to the cascade model, interactive processing involves top-down feedback from words to letters.46 Letters presented in words benefit from the additional activation of the word level nodes stored in long-term memory. Over time, letter level nodes are reinforced from word level nodes through excitatory links and word level nodes are reinforced from letter level notes through excitatory links. This theoretical component was supported through experimentation with words, non-words and pseudowords. Word recognition theory helps to provide a framework to better understand text enhancements used in prescription drug labels. In bottom-up word recognition, text enhancements may be beneficial in correct drug name recognition because individuals are reading letters singularly to form words. Therefore, text enhancement may provide additional information for individuals to correctly identify drug names. However, if individuals process drug names from a top-down model, drug name recognition may not 17

benefit from text enhancement such as capitalization because the overall shape of the letters is changed. Thus figural information may be eliminated from the word. B. The Effects of Frequency and Neighborhood Characteristics on Word Perception 1. The Effects of Frequency on Word Perception It has been well established in the psycholinguistics literature words high in printed frequency are recognized more efficiently than words low in printed frequency.30 That is, as the printed frequency of a word increases, the time required to identify the word decreases. A variety of mechanisms have been presented by researchers to explain this phenomenon. One argument put forward involves the resting activation level of word nodes. Higher frequency word nodes are thought to have a higher resting level of activation compared to lower frequency word nodes. A higher level of resting activation results in a more accurate and quicker reaction time to reach the recognition threshold. ' 46 The second argument is that the decision process is somehow biased by frequency although the exact mechanism is not clear. 2. The Effect of Similar Neighborhoods on Word Perception Lambert et al. investigated the effects of frequency and similarity neighborhoods on drug names by designing a visual perception task using pharmacists as experimental subjects. A "neighborhood" in this context was a set of words that were similar to the target word of interest given some set of criteria. There were three hypotheses associated with the study: (i) high frequency drug names (prescribing frequency) would be identified more 18

Full document contains 162 pages
Abstract: To highlight portions of a drug name that might be confused with other similar drug names, manufacturers use a variety of different stylistic label formats (e.g., bolding, italics, highlighting, underling, capitalization) on primary package labels of prescription drugs. Two experiments were conducted with pharmacists as participants to determine whether two high occurrence label formats (letter bolding and the use of mixed case text) had an impact on pharmacists' accuracy and reaction time in a speeded same-different judgment task. Both experiments examined the effect of orthographic similarity between drug names and stylistic label format (bolding and mixed case text) on accuracy and reaction time. Experiment 2 also examined the effect on accuracy and reaction time of the specific location of the stylistic format within the drug name. The results of both experiments highlighted the strong effect of orthographic similarity on accuracy and reaction time of pharmacists. Pharmacists were less accurate and exhibited longer reaction times when the drug names were orthographically similar. Mixed case text did not significantly impact accuracy or reaction time in either experiment despite the growing belief that these text enhancements reduce dispensing errors. The placement of the text enhancement within the drug name did not significantly effect accuracy or reaction time of pharmacists. Given the relatively high frequency various label formats are recommended by regulators and safety related institutions, it is important the pharmaceutical manufacturers and regulators implement evidence based models to establish regulatory policy related to package labeling. Clear regulatory policy based on rigorous scientific data will help ensure the use of any package label format to mitigate name confusion will provide the greatest patient safety benefit.