Development and validation of a Chinese medication literacy measure

Abstract Background Despite the impact of medication literacy (ML) on patients’ safe use of medications, existing instruments are mostly for general health literacy measurement or designed for specific disease populations, with few specifically designed for ML. Objective To develop and validate the first Chinese medication literacy measure (ChMLM). Methods The ChMLM was developed by a multidisciplinary and bilingual expert panel and subsequently pilot‐tested. The final version had 17 questions in four sections: vocabulary, non‐prescription drug, prescription drug and drug advertisement. Face‐to‐face interviews were administered in a convenience sample of adults with diverse sociodemographic characteristics. Internal consistency was assessed by Cronbach's alpha. Content validity was confirmed by the expert panel, and hypothesis testing was performed to assess construct validity. Results A total of 634 adults were interviewed. The mean (SD) total ChMLM score was 13.0 (2.8). The internal validity was acceptable (Cronbach's alpha=0.72). Nine of the ten a priori hypotheses were fulfilled. Younger age, higher income and higher education levels were significantly associated with a higher ChMLM score. Furthermore, higher scores on the ChMLM were associated with higher confidence or less difficulty in writing, reading, speaking and listening abilities in a health‐care encounter. No association was found between ChMLM total scores and frequency of doctor's visits. Conclusion The ChMLM is a valid and reliable ML measure. It may help pharmacists and other health‐care providers to target patients and problem areas that need interventions with the ultimate goal of preventing medication errors and harm.


| INTRODUCTION
Health literacy (HL) is defined as the degree to which individuals have the capacity to obtain, process and understand basic health information and services needed to make appropriate health decisions. 1 Health literacy has increasingly gained interest in the field of public health and health care because of its influence on patients' health. Low HL is found to be associated with adverse health outcomes, including higher risk of emergency care use 2-7 and hospitalization, 3,7,8 poorer overall health status 3,[9][10][11][12][13] and higher mortality rate. 9,14,15 In addition to general health outcomes, a few studies have found an association between inadequate HL and medication-related skills, including dosing errors, 16 misunderstanding of prescription labels, 17 poorer ability to take medication appropriately, [17][18][19][20][21][22][23][24][25][26][27] use of nonstandardized dosing tools 27 and lack of knowledge of weight-based dosing. 27 Although medication literacy (ML), or the ability to read, understand and process medication-related information, is assumed to be related to HL, ML may not be fully and adequately captured by general HL assessments. A lack of adequate ML could result in poor medication adherence and the misunderstanding of medication-related information or instructions, which could in turn make patients more prone to medication errors that adversely affect their health.
Despite the impact of ML on patients' safe use of medications, existing instruments are mostly for general HL measurement or designed for specific disease populations, with few specifically designed for ML. Stilley et al. 28 developed a medication health literacy screen as a measure of use and understanding of information on prescription labels. The instrument contained two labels: one for an immunosuppressant medication and one for a diabetes medication. Another ML measure, the Medication Literacy Assessment in Spanish and English (MedLitRxSE), developed by Sauceda et al., 29 aimed to assess individuals' ability to access, understand and act on medication information.
Three interrelated constructs were tested in this instrument, namely prose literacy, document literacy and numeracy. Both instruments have demonstrated good psychometric properties.
To assess individuals' ML levels and subsequently create a supportive environment that encourages correct use of medications, it is essential to develop a valid and reliable ML tool. The main goal of this study was to develop and validate the first Chinese ML measure for the general adult population in Taiwan.

| Instrument development
The initial list of ChMLM items was generated by a pharmacist researcher (HWL) adapting previously validated medication-related instruments found in the literature. [28][29][30][31][32][33] The first version of the instrument, which contained 25 items, was reviewed by a panel of 11 multidisciplinary and bilingual experts, including pharmacists, health literacy experts and researchers with backgrounds in psychometrics, communication, education, health literacy and clinical pharmacy. Half of the experts also had experience in the translation and validation of patient-reported measures. The experts were asked to rate each item's importance and appropriateness/relevance on a 5-point scale, ranging from "not important/adequate at all" to "very important/ adequate." In addition, the experts were encouraged to modify and/or comment on the items and explain their rationales. Several iterations of feedback and discussion among the experts generated the revised version. Pilot testing was performed in a convenience sample of 35 individuals with diverse demographic characteristics. These participants were asked to identify ambiguous or unclear questions and suggest an alternative wording. As a result, minor changes were made to enhance clarity and comprehension.

| Study setting and participants
The final version of the ChMLM was administered through face-toface interviews in a convenience sample of the general population in Taiwan from September 2015 to November 2015. Thirty-four pharmacy undergraduate students and research assistants were trained as interviewers by standardized procedures and multiple rehearsals.
The interviewers were reminded to be non-judgemental, avoid overinterpreting the questions and encourage interviewees to try their best on the test and avoid guessing. The interviewers' friends, relatives, neighbours and the customers/members of participating pharmacies and organizations were approached as potential participants.
Potential participants were referred by the pharmacists in the participating community pharmacies, and the interviews were conducted in front of the pharmacies or in nearby areas. The participants were not only necessarily customers of the pharmacies but also included community residents who were acquaintances of the pharmacists. To be eligible, participants needed to be at least 20 years old and able to speak Mandarin or Taiwanese. Exclusion criteria were having speaking, hearing or cognitive impairment that precluded the participants from adequately interacting with the interviewer. The questionnaire was placed on the Internet by Survey Monkey, and the participants' responses were collected either directly using an electronic device (eg, mobile phone, iPad) or indirectly via a hard copy of the questionnaire that was later transferred to the online version by the interviewer.
Participants were offered a choice of self-administration or verbal administration by the interviewer. There was no set time limit for completion of the interview, and each participant was given an NTD$50 convenience store voucher after completing the interview. The study was approved by the institutional review board of the China Medical University & Hospital Research Ethics Center.

| Interview instrument
In addition to the ChMLM items, the interview instrument also collected respondents' sociodemographic information, such as age, gender, education level and personal annual income. For validation purposes, we also asked respondents how frequently they visit a doctor and to self-report their confidence or difficulty in writing, reading, speaking and listening abilities in a health-care encounter. Specifically, we asked how confident they were when filling out medical forms in hospitals, how confident they were in their ability to read and under-

| Psychometric evaluation
The ChMLM total score was calculated by the number of questions answered correctly (a score of 1 was assigned for each correct answer). Several psychometric properties of the ChMLM were assessed, including the percentage of correct responses and the correlation between each item and the total score. Internal consistency was tested by Cronbach's alpha. A Cronbach's alpha coefficient equal to or larger than 0.70 is considered acceptable. 38 Content validity was confirmed by the expert panel. Factor analysis and hypothesis testing were performed to assess the construct validity of the scale. Spearman's correlation was used to test a priori hypotheses that a higher ChMLM total score is associated with higher education levels, higher income, higher frequency of doctor visit and respondents' self-reported higher confidence or less difficulty in their writing, reading, speaking and listening abilities in a health-care encounter. One-way ANOVA with Scheffe's test was performed to test the hypotheses that a higher total score was associated with living in the northern residential areas of Taiwan (ie, the most urbanized and populous metropolitan area in Taiwan) and with speaking Mandarin more commonly. In addition, Pearson's correlation analysis was used to examine the association between age and performance on the ChMLM.
All analyses were performed using PASW Statistics 18 (PASW Statistics for Windows, SPSS Inc., Chicago, IL, USA). The level of significance was set at probability (P)<.05.

| RESULTS
A total of 634 eligible adults with diverse sociodemographic characteristics gave their consent and were enrolled. Among the 634 en- The total number of ChMLM items answered correctly ranged from 0 to 17, with a mean±SD of 13.0±2.8 (Figure 1). The scores were negatively skewed, but the ceiling effect was not evident (2.8% had a full score). The internal consistency of the ChMLM was acceptable with a Cronbach's alpha value of 0.72. As shown in Table 2 P=.001). ANOVA with post hoc comparison results indicated that the respondents in northern residential areas had a higher ChMLM total score than those in the southern residential areas. Moreover, the mean ChMLM total scores were the highest in those who more commonly spoke Mandarin, followed by those who spoke both Mandarin and Taiwanese and then those who spoke Taiwanese more commonly (P<.001). Nevertheless, there was no association between ChMLM total scores and frequency of doctor visit (r s =0.03; P=.53).

| DISCUSSION
In this study, an ML instrument, the ChMLM, was developed to assess Moreover, it is one of the few tools that focus on medication literacy,  This study has a few limitations. First, due to the concern of response burden, the type of medication props and the number of ques- T A B L E 2 Psychometric properties of the individual items information than those who were not. Lastly, medication props, instead of actual medication labels or packages, were used, and it is unknown whether respondents' comprehension and behaviours would differed with actual materials.

| CONCLUSION
This study demonstrated that the developed ChMLM is a valid and reliable performance-based ML measure. It may help pharmacists and other health-care providers to target patients and problem areas that need interventions with the ultimate goal of preventing medication errors and harm. The practical use of the ChMLM needs to be further examined in a representative sample.