The Influence of Symbols on the Short-Term Recall of Pharmacy-Generated Prescription Medication Information in a Low Health Literate Sample

Abstract

The effect of symbols on conveying health information has been studied in a variety of contexts and populations. This investigation assessed whether short-term recall of prescription medication information could be enhanced in a low health literate population through the use of symbols. Participants recruited from various nonpharmacy educational service locations were randomly assigned to receive a prescription medication information leaflet in either a text-only, text with symbols, or symbols-only format and then administered the Rapid Estimate of Adult Literacy in Medicine (REALM). Only those with REALM scores indicative of a sixth grade or lower reading level were included in data analysis. An 8-item, interviewer-administered questionnaire was used to assess participants' recall of information contained in the leaflet. A fictitious medication was used to control for participants' past experiences with medications. Results of statistical testing revealed that the three groups did not differ with respect to their ability to recall the medication information, F(2, 157) = 0.943, p = .392. The symbols used in this investigation were not found to enhance the short-term recall of the medication information presented. These findings should not be considered a wholesale indictment of the use of symbols within health care and in pharmacies. Rather, they should serve as a caution to their widespread use without careful consideration.

Health literacy is “the degree to which individuals have the capacity to obtain, process, and understand basic health information and services needed to make appropriate health decisions” (Ratzan & Parker, 2000, p. vi). It is estimated that more than 90 million U.S. adults may lack the reading and computational skills necessary to function adequately in today's health care environment (Nichols-English, 2000). This is especially true as it relates to navigating through the ever-changing and complex nature of the U.S. health care system (Mika, Kelly, Price, Franquiz, & Villarreal, 2005). Some argue that health care information in general can be difficult even for the most educated of patients to understand (Hardin, 2005). This may prohibit patients from adequately obtaining, processing, and understanding the basic health information they need to make appropriate health decisions.

Patients are required regularly to act upon medical information provided by health care professionals (e.g., calculating dosage amounts, what to do if they miss a medication dose). It is unfortunate that differences exist between what patients comprehend and what health care professionals expect them to know or believe they know (AMA Ad Hoc Committee, 1999). Patients with low literacy are less likely to comprehend and to remember medication advice (AMA Ad Hoc Committee, 1999), and they have difficulty understanding what the majority of health care professionals would consider the most basic of medication instructions (Youmans and Schillinger, 2003). Further complicating matters for those with poor health-literacy skills is the fact that the majority of printed health care–related materials are written at a 10th-grade reading level or higher (Safeer and Keenan, 2005). This may pose a problem given that the majority of adults are estimated to read at an eighth- or ninth-grade level (Kirsch, Jungeblut, Jenkins & Kolstad, 1993).

The influence of symbols on the conveyance of health information may be valuable to those possessing limited literacy skills (Sojourner & Wogalter, 1997). However, we were unable to find published evidence where this influence was tested adequately within a sample of patients possessing inadequate health literacy as defined by a commonly accepted health-literacy assessment instrument. Therefore, the objective of this investigation was to assess whether short-term recall of pharmacy-generated prescription medication information could be enhanced in a low health literate sample through the use of symbols.

Evidence exists in the literature demonstrating the effect that inadequate health-literacy skills may have on the medication-taking behaviors of patients (Hardin, 2005). For example, patients with limited health literacy may take inappropriate dosages, take medications at incorrect times, and even use inappropriate routes of administration. The consequences of limited health literacy may worsen in the future as new medications are introduced to the market that increase the complexity of drug regimens as well as the possibility of drug-drug interactions and adverse effects (Hardin, 2005). Written health information is becoming more prevalent as health care professionals attempt to decrease medication errors and increase patient understanding of their medical conditions and treatments (Nichols-English, 2000). Without appropriate counseling by health care professionals, including the provision of appropriately compiled written materials, inadequate health literacy will continue to place patients at an increased health risk resulting from medication-related problems (Hardin, 2005).

The effect of symbols and pictures on conveying health information has been studied in a variety of contexts and in several different populations (e.g., Delp & Jones, 1996; Dowse & Ehlers, 2003a, 2003b; Dowse & Ehlers, 2005; Hameen-Antila, Kemppainen, Enlund, Patricia, & Marja, 2004; Hanson, 1995; Hanson & Hartzema, 1995; Mansoor & Dowse, 2003; Morrow, Hier, Menard, & Leirer, 1998; Morrow et al., 2005; Ngoh & Shepherd, 1997; Sansgiry, Cady, & Adamcik, 1997; Sojourner & Wogalter, 1997, 1998). For example, studies show that symbols may improve recall (Hanson & Hartzema, 1995; Houts et al., 1998; Houts, Witmer, Egeth, Loscalzo, & Zabora, 2001) and comprehension of (Delp & Jones, 1996; Dowse & Ehlers, 2004, 2005; Hanson & Hartzema, 1995; Hardie, Gagnon, & Eckel, 1979; Mansoor & Dowse, 2003; Ngoh & Shepherd, 1997; Schillinger et al., 2005) as well as adherence to important medical instructions among patients with limited literacy skills (Delp & Jones, 1996; Dowse & Ehlers, 2005; Hanson, 1995; Hardie et al., 1979; Ngoh & Shepherd, 1997; Schillinger et al., 2005). As with other studies that examined various counseling strategies on recall, this investigation defined recall as the ability to recollect information that was contained within the medication-information leaflet (Houts et al., 1998; Houts et al., 2001). The literature suggests patients recall between 29% and 72% of what is communicated to them by health care providers (Ley, 1982). An inverse relation between the amount of information provided and the recall rate of patients has also been reported (Ley, 1982). Given the amount of information that must be considered when consuming prescription medications (e.g., side effects, possible food and drug interactions, dosage instructions), improving the recall of patients can play an integral role to their health. This is especially true when patients are prescribed multiple medications. In addition, evidence exists demonstrating that patients prefer symbol-enhanced health information across all literacy levels (Hanson, 1995; Mansoor & Dowse, 2003; Sojourner & Wogalter, 1997, 1998), and symbols used in conjunction with more traditional modes of instructions are believed to be useful in assisting patients with taking medications appropriately (Katz, Kripalani, & Weiss, 2006).

Of the studies evaluating the use of symbols in health care, few used reliable and valid measures to assess the health literacy status of their participants prior to the study's execution (e.g., Hardie et al., 1979; Schillinger et al., 2005). Others have used surrogate measures of health literacy (e.g., education level) to classify participants. While education is important, it is known that education level alone cannot appropriately determine an individual's health-literacy skill (Hardin, 2005). Furthermore, few published studies attempt to provide a theoretical explanation for the value of using symbols to communicate health information, including prescription medication information, to patient populations, including those with low health-literacy (Ngoh & Shepherd, 1997; Sojourner & Wogalter, 1997). These limitations served as the impetus for this research.

Theoretical Framework

Empirical evidence suggests that pictures aid in facilitating recall, making the material more specific and providing an alternative to which text information can be cognitively represented. Paivio's dual coding theory (DCT; Paivio, 1986, 1991; Sadoski & Paivio, 2001) may be useful in this regard in that it argues that equal weight should be given to verbal and nonverbal processing, because presenting information in both visual and verbal form enhances recognition and recall.

According to DCT, information is processed by two separate conceptual-coding systems. The first verbal system processes and stores linguistic elements of a message (logogens). The second system, the nonverbal system, processes nonverbal elements (imagens; Simpson, 1995). While the verbal and nonverbal systems are thought to be functionally independent in that one system can be active without the activation of the other, these two systems are believed to be active in parallel. One system triggers activity in the other system rather than information actually flowing from one system to the other, hence dual coding (Paivio, 1991). The activation of an image or description is situation dependent for each individual and occurs by simply recognizing or thinking about words or objects. The functional strength of its association is determined by prior experiences with referent class members (i.e., expertise or prior knowledge of the task) and the verbal descriptions associated with them. As a result, the application of this theory to image use in health care is dependent on the image being used and patients' firm understandings of its context.

According to the DCT, information is much easier to retain and retrieve when dual coded because of the multiple mental representations being available, rather than only one (Sadoski & Pavio, 2001). Paivio suggests that images are more likely to activate both coding systems than are words and as such the image code is superior to the verbal code for evoking recall; images are easier to remember than are words (Sadoski & Pavio, 2001). Because counseling is a critical and valued component of the patient-pharmacist encounter, a systematic evaluation of the effect of symbol use in a pharmacy context is warranted. Can the recall of prescription medication information be enhanced in a low health-literate population through the use of symbols? It is hypothesized that the recall of patients receiving written prescription medication information combined with symbols will be greater than that of those receiving the information in text or symbol form only. It is also hypothesized that the recall of patients receiving written prescription information in symbol form only will be greater than that of those receiving the information in text form only.

Method

Participants

Permission to recruit potential participants was received from a local literacy council, adult basic education programs, Workforce Investment Network Job Centers, and a Job Corp Center. All data were collected by the same researcher over a 3-month period from the various sites. All potential participants who visited the aforementioned sites and reported that English was their first language were eligible for inclusion in the investigation. However, similar to previous studies assessing the health literacy skills of patients (e.g., Parker, Baker, Williams, & Nurss, 1995), patients were excluded if they (a) were under 18 years of age, (b) were too ill to participate, (c) had unintelligible speech, (d) had impaired hearing or vision, (e) were under police custody, (f) had severe cognitive impairment or an overt psychiatric illness, or (g) had the smell of alcohol on their breath. Those agreeing to participate were directed one at a time to a private area where they were given an explanation of the purpose of the investigation and the opportunity to ask questions. This study was approved by the University of Mississippi's institutional review board.

Because the instrument used to assess the participants' ability to recall the information provided was developed specifically for this research, information concerning variance was not available. In addition, other published studies did not have as their exclusive samples patients with low health-literacy and, as a result, the statistics from those studies were of limited value when attempting to determine a sample-size estimate for the present study. Therefore, the calculation of an appropriate sample size for this project was based on a three-group experimental design (see the subsequent section for a description of the three groups) assuming a standardized medium effect size (Cohen's f = 0.25), a power of 0.8, and an alpha level of .05. This power analysis indicated that 53 participants per group were necessary (N = 159).

Because the focus of the study was the low-literate population, all who agreed to participate in the investigation were administered the Rapid Estimate of Adult Literacy in Medicine (REALM) to assess the health literacy skills of potential participants. The REALM was chosen for this study because it has shown to be a rapid, reliable measure of reading ability in a medical setting. The REALM has demonstrated high criterion validity, correlating well with the Peabody Individual Achievement Test–Revised (0.97), Wide Range Achievement Test–Revised (0.88), and the Test of Functional Health Literacy in Adults (0.84; Davis et al., 2005). The test–retest reliability of the REALM was found to be 0.97 (Davis et al., 1993). The instrument consists of three columns of 22 medical words each, arranged by number of syllables and level of difficulty, beginning with simple, one-syllable words (e.g., flu and fat), progressing toward complex, multiple-syllable words (e.g., osteoporosis and potassium). The word items assess literacy at a third-grade through high-school level. Scores on the REALM range from 0 (no words pronounced correctly) to 66 (all words pronounced correctly). The raw scores are converted to grade ranges of four categories: (1) 0–18 indicating a third-grade equivalent reading level and below, (2) 19–44 indicating a fourth- to sixth-grade reading level, (3) 45–60 indicating a seventh- to eighth-grade reading level, and (4) 61–66 indicating a high-school reading level. Only those participants scoring in categories 1 or 2 were considered in the present study.

Experimental Manipulation: Medication Information Leaflet

Three versions of a patient medication information leaflet for a fictitious medication were developed. Each version contained the same eight medication directives (Figure 1) that specified the instructions and warnings associated with the medication's use. The medication leaflets created were (a) text only, (b) text enhanced with symbols, and (c) symbols only. The medication information leaflets used in this study were modeled after those developed by Sojourner and Wogalter (1998) and were consistent with those developed by pharmaceutical companies. The leaflets were printed on 8.5 in. × 11 in. (216 mm × 279 mm) white bond paper with the medication name and eight medication directives printed on one side (the other was side blank). A fictitious medication was used to control for prior knowledge or personal experiences with the medication. The medication name and indication, also fictitious, were printed at the top of each leaflet. This information was followed by the eight medication directives that specified the instructions and warnings associated with the use of the medication, such as take three times a day, take until gone, and so forth. The medication directives were printed in list format with their corresponding symbol located to the left of the text. The symbols used in this study were developed and standardized by the United States Pharmacopeia (1997).

Figure 1 Medication directives with symbols.

In the text enhanced with symbols version of the leaflet, text and the corresponding symbol were shown for each directive. In the text-only version, the symbols were not shown, and in the symbols-only version, the text was not shown. The text-only version served as the control as it represented the most commonly used method in present-day pharmacy practice.

Procedures

After written consent to participate was obtained, participants were randomly assigned to one of the three study groups—text-only group (Group 1), text with symbols group (Group 2), and symbols-only group (Group 3)—and were administered the REALM. The REALM was scored and used as a screening instrument to identify those patients possessing inadequate health literacy skills. As mentioned earlier, only those scoring between 0 and 44 on the REALM, indicating a sixth-grade or lower reading level, were included in data analysis. All individuals who agreed to participate in the investigation were randomized and completed all aspects of the study, including completion the REALM, reading of the information leaflet, and completion of the recall task. This was done for logistical reasons but also to avoid potentially stigmatizing participants with low REALM scores (our university's institutional review board requested that we not exclude individuals with high REALM scores until data analysis).

Before being provided a group-specific leaflet, each participant was informed that they would be given 1 minute to review the leaflet; after that 1 minute the leaflet would be returned to the interviewer, and participants would be asked to respond to questions concerning the medication directives on the leaflet. The dependent variable (recall) was assessed using an 8-item, open-ended interviewer-administered questionnaire. The questions were based on the eight medication directives for the fictitious prescription medication as listed in Figure 1. One question corresponded to each of the directives on the group-specific information leaflet (see Table 1).

Table 1. Recall assessment

1. How many times a day must you take this medication?

2. How long before eating will you take this medication?

3. Must you take this medication with water?

4. Can you take this medication with milk?

5. Must you take all of this medication?

6. Where must this medication be kept or stored?

7. Can you drink alcohol while taking this medicine?

8. If this medication makes you sleepy, what can you NOT do?

All participants were asked by the interviewer to assume (pretend) that they were receiving a prescription for an unfamiliar medication to treat a personal medical condition and also to assume that the leaflet had been provided by their pharmacist. After 1 minute, the leaflet was retrieved and the eight questions were read aloud to the participants and responses were recorded. After the interview was completed and the subject departed, responses were scored as correct or incorrect, and a “total correct” score was calculated. Unclear responses were reviewed by the research team until consensus was achieved. The entire contact period, from obtaining consent through debriefing, lasted approximately 15 minutes per participant.

The readability and face validity of the instructions read by the interviewer to the patients, the three versions of the patient information leaflets, and the recall measure were established by a panel, which included pharmacy-trained academicians and researchers in addition to practicing pharmacists. Before the start of fieldwork, all materials were pretested among individuals across all levels of health literacy, as determined by the REALM, to ensure proper understanding. The symbols used to indicate the medication directives and the time allowed to the participants for reviewing the leaflets (1 minute) were also chosen based on the results of this pretest. The questions used to assess recall were selected based on feedback from the pretest and are consistent with those posed in similar studies (e.g., Dowse & Ehlers, 2005).

Results

Sample Description

Individual and group characteristics of respondents are provided in Table 2. A total of 239 individuals met the study's initial inclusion criteria and were interviewed during the three months of data collection. Of the 239 interviewed, 161 scored 44 or below on the REALM and were used in subsequent data analysis. Of those 161 meeting the REALM score inclusion criteria, 54 had been assigned to the text-only group (Group 1), 54 had been assigned to the text with symbols group (Group 2), and 53 had been assigned to the symbols-only (Group 3). The mean REALM score for the analysis sample was 24.48. More than half of those in the analysis sample were female (52.8%), and the majority were African American (85.7%). More than one third reported having no health insurance (37.8%), while an additional one third reported having either Medicare or Medicaid (34.0%). The mean age of participants was 33.8 years, and more than three fourths had not earned a high school diploma (75.8%). With the exception of gender (Pearson chi square = 13.872, df = 2, p = .001), individual characteristics were consistent across groups.

Table 2. Group characteristics

	Text-only group (n = 54)	Text with symbols group (n = 54)	Symbols-only group (n = 53)	Total (N = 161)	p
Gender
Male (%)	17 (31.5)	36 (66.7)	23 (43.4)	76 (47.2)	.001*
Female (%)	37 (68.5)	18 (33.3)	30 (56.6)	85 (52.8)
Race
Caucasian (%)	9 (16.7)	8 (14.8)	6 (11.3)	23 (14.3)	.725
African American (%)	45 (83.3)	46 (85.2)	47 (88.7)	138 (85.7)
Age in years	33.8	35.8	31.6	33.8	.440
REALM score	27.06	22.00	24.40	24.48	.245
Insurance status
Private insurance	11.3%	9.6%	5.9%	9.0%	.304
Employer insurance	15.1%	19.2%	17.6%	17.3%
Medicaid	37.7%	15.4%	25.5%	26.3%
Medicare	7.4%	11.5%	3.9%	7.7%
Other government	1.9%	1.9%	0.0%	1.3%
Unknown	0.0%	0.0%	2.0%	0.6%
None	26.4%	42.3%	45.1%	37.8%
Education level
Less than high school degree	72.2%	74.1%	81.1%	75.8%	.955
High school degree or GED	22.1%	22.1%	17.0%	20.5%
More than high school	1.9%	1.9%	0.0%	1.2%
Unknown	1.9%	1.9%	1.9%	1.9%
Refused	1.9%	0.0%	0.0%	0.6%

Examination of Hypotheses

Cell means, standard deviations, and results of statistical testing appear in Table 3. As discussed earlier, the dependent variable in this investigation was assessed as the number of questions out of eight that participants answered correctly. Of the demographic variables examined, only gender was found to differ significantly among the three groups. Therefore, gender was statistically controlled as a covariate in the subsequent analysis of group differences. We conducted an analysis of covariance at the a priori alpha level of .05 to examine the differences between the recall levels of the three groups after exposure to the pharmacy-generated prescription medication information leaflet. Analysis revealed that the three groups did not statistically differ, F(2, 157) = 0.943, p = .392. The mean scores on the recall assessment (minimum score = 0, and maximum score = 8) for the three groups were 6.54 (text-only), 6.65 (text with symbols), and 6.36 (symbols-only).

Table 3. Comparison of groups on dependent variable

	Text-only group (n = 54), M (SD)	Text with symbols group (n = 54), M (SD)	Symbols-only group (n = 53), M (SD)	p
Recall assessment	6.54 (1.40)	6.65 (1.40)	6.36 (1.43)	.392

Discussion

It was expected for this experiment to demonstrate that participants who received the pharmacy-generated prescription medication information in standard text form combined with symbols would be able to recall the information significantly better than those receiving the same information in text form only. Such a relation has been suggested in the health literacy literature (Hanson & Hartzems, 1995; Houts et al., 1998; Mika et al., 2005). On the basis of DCT, it was expected that those receiving the pharmacy-generated prescription medication information in symbol form only would recall the information significantly better than those receiving the information in text form only (Paivio, 1991; Sadoski & Pavio, 2001). However, these hypotheses were not supported. Regardless of presentation format, the ability to recall medication-related information, as presented in this investigation, was similar.

There are two plausible explanations for the lack of significant findings in the present study. First, the possibility exists that the text used in the treatment manipulations was such that even those with inadequate health literacy could interpret the language; of the 47 words contained within the eight medication directives, 35 (74%) were monosyllabic. It is possible that most of our subjects could read and understand the medication directive text. The health literacy of participants in this study, according to the REALM, corresponded to reading abilities up to the sixth-grade level. It is possible that symbols will enhance immediate recall of pharmacy-generated medication information only when the complexity of that verbal information exceeds the patient's reading level. Future research examining the influence of symbols on the recall of medication directives must be conducted with varying degrees of complexity and should include directives that exceed the reading ability of subjects.

Second, the nature of this investigation required a great deal of involvement on the part of the participants in that they were asked to in explain how the fictitious medication was to be used. Therefore, the teachback method was, in essence, used as the basis for assessing the dependent measure. The teachback method is one in which the health care provider asks the patient to explain in their own words what they are being asked to do, and this method has been recommended strongly by many authors (Davis et al., 1993; Slosson, 1990; Williams, Davis, Parker, & Weiss, 2002). For example, pharmacists may ask patients, “How many times a day must you take this medication?” Here, patients would respond according to the directions that they were given concerning their prescription medications, as participants of this study were expected to do. Given the involvement of the participants, the use of this reliable method of assisting patients to comprehend medication instructions and the aforementioned possibility of language simplicity's effect on the study's findings, it is possible that the value of symbols and DCT may only be realized hours or days later, rather than immediately, when patients are removed from health care provider environments and are attempting to recall what had been communicated to them earlier. Future research should assess recall of medication information at various points in time of a medication regimen (e.g., immediately, 8 hours later, 1 day later, 1 week later). Such research would assist in determining the true influence of symbols on the recall of medication directions.

There are several potential limitations to our study. First, participants who participated in the investigation were provided information concerning a fictitious medication. This method was necessary in order to control for participants' experience with specific medications. Also, a fictitious drug was used in a contrived (nonpharmacy environment) situation in which participants were asked to recall information within a relatively short period of time following viewing the leaflets. Further research is needed to assess the recall deterioration in a similar intervention as the one used in this study. This study aimed to assess the influence of symbols on recall of printed medication information. As such, the effectiveness of symbols as part of printed information that supplements pharmacist counseling cannot be determined. In addition, participants may perform differently when facing an actual drug-use situation and when the participants may be “ill.” Second, participants and the researcher were isolated in a private room without the distractions and typical surroundings of a pharmacy setting. Future research in this area should strive to emulate more closely a true pharmacy setting and its associated distractions. Third, study participants knew they were being tested, and this may have affected the way they viewed the materials. The possibility also exists that individuals believing they would perform better on the recall assessment were more likely to enroll into the study. This raises concerns regarding selection bias, which may have affected the results of this investigation. However, refusal by participants to take part was not a problem in this investigation. The vast majority of participants were excited to contribute to the study. Fourth, there exists a possibility that the instrument used to assess recall failed to discriminate appropriately between the three groups. Limiting the instrument to only eight questions may have negatively affected its ability to differentiate between the three groups. Additional research using this instrument is needed to further establish construct validity. This investigation was also limited in that instructions were provided for only one medication. It would be of interest for future research to assess the influence of symbols when participants are asked to recall information regarding multiple medications.

Fifth, we do not know the generalizability of our findings given that data were collected in various settings from predominately indigent individuals. The sample for this investigation consisted primarily of African Americans, and published evidence suggests that cultural background may influence the interpretation of symbols (Kassam, Vaillancourt, and Collins, 2004). However, inadequate health literacy encompasses society (Kirsch et al., 1993), and this investigation provides evidence that functional health literacy of patients is an important element of the medication-use process. The findings of this investigation may also be attributed to the symbols themselves. It is possible that the symbols used may not have been adequately designed or were misunderstood by the participants of this study. This possibility should be examined in a similar sample during future research. Last, the number of males and females was disproportionate in the three groups. It is difficult to discern whether gender differences among the groups affected the results. However, analyses comparing the treatment groups were conducted under conditions where gender was both controlled (i.e., a multivariable model) and not controlled; similar results were found (i.e., no significant differences among the treatment groups), suggesting that differences among the groups in gender composition did not meaningfully affect study findings.

Helping to ensure that patients recall and comprehend important health information is a challenging task perhaps overlooked by health care professionals, including pharmacists. However, it can be argued that it is also a task as equally important as making a correct diagnosis and selecting an appropriate treatment. This research demonstrates that incorporating symbols into health information is not a panacea, at least not for low health literate patients in a low reading complexity situation. This study contributes to the existing literature in health literacy in several important ways. As best as can be determined, this was the first attempt with an adequately sized sample to assess the influence of symbols or pictorial aids on the recall of health information in an English-speaking, true low health literate population in the United States. In addition, this study experimentally tested the DCT within the context of health care for the first time. Although no effect was found, this study serves as one of few studies that attempted to provide a theoretical explanation for the value of using symbols to enhance the recall of health information (prescription medication information) by patient populations, including those who are low health literate (Ngoh & Shepherd, 1997; Sojourner & Wolgalter, 1997). Future research should build on this investigation to explore further the value of DCT in understanding and ultimately enhancing recall of medication information.

Conclusions

We observed no indication that medical symbols enhance recall among patients with low health literacy after a 15-minute delay. While this study did not specifically address the issue of symbol development or dissemination, it may be that the full benefit of symbol containment within health information cannot be realized unless a large portion of the population becomes accustomed to their meaning and their use. Symbols need to be simplified and evaluated in a broad array of patient populations. Comprehension of symbols ultimately needs to be evaluated when teaching about medications, and many people will need to be taught the meaning and implications in order for the system of medication symbols to be effective. Symbols must convey the appropriate information and do so in a consistent manner across different populations. However, what remains is that it may be far easier for pharmacists to teach the meaning of a symbol to a functionally low health literate individual during in-depth patient counseling sessions than it would be to teach that person to read. The results of this study should not be considered a wholesale indictment of the use of symbols within health care. Rather, they should serve as a caution to their widespread use without careful consideration of many issues.

Acknowledgments

This work was presented at the American Pharmacists Association Annual Meeting, San Antonio, Texas, April 6, 2009.

The authors acknowledge Mr. Phillip Schwab; the Memphis, Tennessee, Literacy Council; the Grenada, Mississippi, Education Center; the Batesville, Mississippi, Job Corp Center; and all the Workforce Investment Network Job Centers for assisting in the recruitment of the sample for this investigation. The authors also thank the United States Pharmacopeia for use of their symbols in this research and the anonymous reviewers whose thoughtful comments improved the quality of the article.

Notes

Note. Answers to recall assessment were scored as correct or incorrect.

Note. Chi-square was used to compare the demographic variables gender, insurance status, and education level. An analysis of variance was conducted to compare groups on age and REALM score.

*Significant.

Note. The dependent variable (recall) was assessed using an open-ended interviewer administered questionnaire. Scores ranged from 0 to 8. A score of 8 indicated the highest level of recall.