|Home | About | Journals | Submit | Contact Us | Français|
Patient misunderstanding of instructions on prescription drug labels is common and a likely cause of medication error and less effective treatment.
To test whether the use of more explicit language to describe dose and frequency of use for prescribed drugs could improve comprehension, especially among patients with limited literacy.
Cross-sectional study using in-person, structured interviews.
Three hundred and fifty-nine adults waiting for an appointment in two hospital-based primary care clinics and one federally qualified health center in Shreveport, Louisiana; Chicago, Illinois; and New York, New York, respectively.
Correct understanding of each of ten label instructions as determined by a blinded panel review of patients’ verbatim responses.
Patient understanding of prescription label instructions ranged from 53% for the least understood to 89% for the most commonly understood label. Patients were significantly more likely to understand instructions with explicit times periods (i.e., morning) or precise times of day compared to instructions stating times per day (i.e., twice) or hourly intervals (89%, 77%, 61%, and 53%, respectively, <0.001). In multivariate analyses, dosage instructions with specific times or time periods were significantly more likely to be understood compared to instructions stating times per day (time periods — adjusted relative risk ratio (ARR) 0.42, 95% Confidence Interval (CI) 0.34–0.52; specific times — ARR 0.60, 95% CI 0.49–0.74). Low and marginal literacy remained statistically significant independent predictors of misinterpreting instructions (low - ARR 2.70, 95% CI 1.81–4.03; marginal -ARR 1.66, 95% CI 1.18–2.32).
Use of precise wording on prescription drug label instructions can improve patient comprehension. However, patients with limited literacy were more likely to misinterpret instructions despite use of more explicit language.
Patient misunderstanding of instructions on prescription drug labels is a medication safety and health literacy concern1–3. The 2006 Institute of Medicine Report, Preventing Medication Errors, cited poor patient comprehension and subsequent unintentional misuse of prescription drugs as a root cause of medication error, poor adherence, and worse health outcomes3. A recent study by our research team found nearly half of primary care patients misunderstood common dosage instructions on prescription container labels4. Patients with limited literacy and those taking more medications were at greatest risk. As patients, particularly the elderly, are taking an increasing number of prescription drugs, the ability to accurately interpret medication instructions becomes even more critical for ensuring proper and safe use5,6.
While limited literacy may impede patient comprehension of medication dosage instructions, the instructions also may not be written in the most clear and precise manner 7–10. There is little evidence supporting best practices for writing prescription medication dosage instructions to promote patients’ understanding of use. Data from our previous study and earlier cognitive factors research suggest that less complex and more explicit dosage instructions might improve patient understanding4,7,11–15. The purpose of this study was to evaluate whether the use of more explicit language to describe the dose and frequency of prescribed drugs could improve comprehension, especially among patients with limited literacy. We hypothesized that more explicit instructions would improve patient interpretation, and the association between literacy and understanding how to take prescribed drugs would be reduced.
Study participants were adult patients who attended one of three outpatient primary care clinics in Shreveport, Louisiana, Chicago, Illinois and New York, New York. All of these study clinics provide care for a large number of indigent patients. Subject recruitment took place from May to December 2006. The Shreveport and New York clinics were within a public university hospital while the Chicago study clinic was a Federally Qualified Health Center. Institutional Review Boards at the affiliated institutions (Louisiana State University Health Sciences Center at Shreveport, Northwestern University, Mount Sinai School of Medicine) approved this study.
Patients at the three clinics were eligible if they were 18 years of age or older. Research assistants (RA) approached consecutive patients in each clinic while they were waiting to see physicians. Patients were excluded from participation if they reported they had severely impaired vision, hearing problems, were acutely ill, or did not speak English. A total of 401 patients were approached and 373 consented to the study. Nine individuals were excluded based on language barriers, and three were ineligible due to visual impairment. In all, 359 consented to the study (90% response rate).
We studied instruction labels for three commonly prescribed medications: glyburide, metformin, and atenolol. Three physicians and one pharmacist identified a typical dose for each medication, along with variations in the frequency of use for the drug’s daily administration. Atenolol was written to be taken once a day, while glyburide and metformin were written for twice a day. A minimum of three variations of the dosage instructions were used per drug, ranging from vague to most explicit. Specifically, frequency of use for the prescribed drug was presented either as 1) number of times per day (“twice daily”), 2) hourly intervals (“every 12 hours”), 3) time periods (“morning”, “evening”), or 4) specific times (“8 a.m.”, “5 p.m.”; Table 2). A total of ten mock pill bottles were developed based on these different presentations of dose (number of pills) and frequency of use (number of times to be taken per day) for the three drugs.
After obtaining informed consent, a trained RA administered a structured interview that lasted approximately 20 minutes and included a self-report of sociodemographic information (age, gender, race/ethnicity, and education, number of prescription medicines taken daily) and a brief literacy assessment. The RA then showed each patient the ten prescription bottles one at a time and asked “How would you take this medicine?” The RA documented patients’ verbatim responses. All patients viewed the pill bottles in the same order, which was determined by random assignment. This procedure has been widely used by this research team to assess patients’ functional understanding of prescription drug instructions and warnings 4,7,11.
Patient responses were independently rated as either correct or incorrect by three general internal medicine attending physicians from two academic medical centers. Physicians were blinded to patient information and were trained to follow stringent coding guidelines previously agreed upon by the research team7. Correct scores were given only if patients’ responses included both the proper dose (number of pills to be taken at one time) and frequency of use (number of times drug is to be taken daily) as stated on the label. For label instructions that detailed a drug’s frequency using hourly intervals or time periods, raters followed a predetermined range of acceptable responses for coding purposes to allow for some variability in interpretation. Instructions that included specific times for taking the medicine had to be precise or give a very close approximation to be correct. If frequency was stated using the number of times per day, responses were correct if either the exact number was reported back, or if appropriate specific times or time periods (i.e., 8 a.m., noon, 5 p.m.; breakfast, lunch, dinner) were described. If patients’ responses were inaccurate or incomplete in their interpretation they were scored as incorrect.
Inter-rater reliability between the three physicians coding the patient responses was very high (Kappa=0.84). Responses that received discordant ratings between the three reviewers (=252) were scored by a panel of one primary care physician and two behavioral scientists with expertise in health literacy. Each panel member, blinded to patient information, independently coded the responses as correct or incorrect. A consensus ruling was achieved for 91% of responses. A majority rule was used for the remaining 24 responses.
Patient literacy was assessed using the Rapid Estimate of Adult Literacy in Medicine (REALM), a reading recognition test comprised of 66 health related words 16,17. The REALM is the most commonly used test of patient literacy in medical settings18. Raw scores can be converted into one of three reading levels: sixth grade or less (0–46), seventh to eighth (45–60), ninth grade and above (61–66). The REALM is highly correlated with other standardized reading tests and the Test of Functional Health Literacy in Adults (TOFHLA)18,19.
All statistical analyses were performed using SAS software version 9.1 (Cary, NC). Descriptive statistics (percentage, mean and standard deviation) were calculated for each variable. Chi-square tests were used to evaluate the association between sociodemographic characteristics and patient understanding of each of the ten prescription label instructions. In multivariate analysis, the ten binary repeated responses of correct vs. incorrect understanding per subject were modeled using a generalized linear model with a complementary log–log link function. A generalized estimating equation (GEE) approach was used to adjust model coefficients and standard errors for within-patient correlation using PROC GENMOD (SAS Institute, Cary, NC). Wald 95% confidence intervals were calculated for adjusted relative risk ratios using the robust estimate of the standard error as detailed by Liang and Zeger 20,21. The final multivariate model included the variables age, gender, race (white vs. African American), education, site, and number of medications currently taken daily. The type of language used to state frequency of use (times per day, hourly intervals, time periods, specific times) was entered in the model as the primary independent variable of interest. The complexity of the instruction (one tablet a day vs. two tablets twice daily) was considered to be a potential risk factor to patient understanding and also entered in the analysis as a covariate. Patient literacy was classified either as low (6th grade and below), marginal (7th–8th grade) or adequate (9th grade and higher). In order to examine whether more explicit instructions could overcome the barrier of limited literacy on patient understanding, an interaction term for literacy and type of language used in the instruction was included in the final model.
The mean age of patients was 48.4 years (SD=13.7; range 20 to 80 years); 72% were female and 61% were African-American. Approximately half of patients were recruited in Shreveport (56%), 25% in New York City, and 19% in Chicago. Twenty percent of respondents had less than a high school education; 15% were identified as reading at or below a 6th grade level (low literacy), and 30% were reading at the 7th–8th grade level (marginal literacy). Patients were currently taking an average of 2.8 prescription medications (SD=2.9). Low literacy was associated with older age (<0.001), African American race (<0.001), and less education (<0.001; Table 1).
Each patient provided interpretations for ten different instructions for a total of 3,590 responses for the three drugs. Of these 839 (23%) were coded as incorrect. Seventy-eight percent of patients misunderstood one or more instructions, with 37% misunderstanding a minimum of three labels. The prevalence of incorrectly interpreting one or more label instructions among patients with adequate, marginal and low literacy was 71%, 84%, and 93%, respectively (<0.001). Rates of correct interpretation were lowest for instructions that depicted frequency in hourly intervals or the number of times of day (“Take 1 pill by mouth every 12 hours with a meal”, “Take two tablets by mouth twice daily”; 53% and 61%, respectively) and highest for those that used time periods (“Take 2 pills in the morning and 2 pills in the evening”, “Take 1 pill by mouth every day. Take in the morning”; 89% for both labels).
Patients with low literacy were more likely to misinterpret seven of the ten instructions compared to those with adequate literacy (Table 2). Two of three label instructions where literacy was not significantly associated with correct interpretations were for atenolol, which had the most basic frequency schedule (1 tablet a day). No statistically significant differences in rates of understanding the medication labels were noted by either age or number of prescription medications currently taken by patients.
In multivariate analyses, prescription instructions that gave time periods (morning, evening) or specific times (8 a.m. and 5 p.m.) were significantly less likely to be misinterpreted compared to those using the number of times per day [twice daily] (time period — adjusted relative risk ratio (ARR) 0.42, 95% confidence interval (CI) 0.34–0.52; specific times — ARR 0.60, 95% CI 0.49–0.74; Table 3). Frequency of use stated in hourly intervals (i.e., every 12 hours) was significantly more likely to be misinterpreted compared to writing frequency as the number of times per day (ARR 2.87, 95% CI 2.29–3.60). The reference group was then altered from the previous times per day to time periods in order to determine if this latter format significantly improved patient comprehension compared to the use of specific times. Misinterpretation of instructions was higher with the use of specific times compared to time periods (ARR 1.43, 95% CI 1.19–2.71).
Low and marginal literacy were also statistically significant independent predictors of misinterpreting instructions (low — ARR 2.70, 95% CI 1.81–4.03; marginal —ARR 1.66, 95% CI 1.18–2.32). Fewer years of education (< high school, ARR 1.36, 95% CI 1.03–1.77) and greater dose complexity (four tablets taken per day [glyburide]); ARR 1.47, 95% CI 1.20–1.83) were also found to be significantly and independently associated with misinterpretation. The interaction term for literacy and type of language used to depict drug frequency of use was included in the final multivariate model; it approached but did not reach statistical significance (ARR 0.91, 95% CI 0.85–1.01; =0.079).
Physicians may assume patients can interpret prescription drug label instructions, yet four out of five patients (79%) in this study misinterpreted one or more of the ten common prescription label instructions they encountered. Although the instructions were brief and of minimal reading difficulty, rates of patient understanding varied widely across all literacy levels. More explicit language instructing patients when to take the medicine using time periods were better understood compared to instructions that more vaguely stated the number of times per day or hourly intervals. This finding is supported by prior research demonstrating that older adults have greater difficulty interpreting medication instructions that do not explicitly detail how and when to take a prescribed medicine13–15.
Labels that instruct patients to take medications “twice daily” or “every 12 hours” require patients to make additional mental steps to infer when to take a medicine. For patients with limited literacy, this adds an unnecessary cognitive burden, resulting in poorer comprehension12. Despite the use of more precise instructions, however, comprehension among those with low literacy skills was still significantly lower than patients with marginal or adequate literacy skills. This is also not surprising, as earlier health literacy studies found that materials with low reading grade levels were likely to improve comprehension among patients with adequate literacy, but had only variable success in improving comprehension among patients with low literacy22.
Interestingly, identifying specific times each day (8 A.M., 5 P.M.) for administration was a more easily understood instruction format than stating times per day or hourly intervals. However, patients were significantly more likely to misinterpret these instructions compared to those stating time periods in the day (morning, evening). It is possible that patients do not need such precision when following medication instructions. Stating frequency using time periods of day rather than precise times may better reflect patients’ preference to tailor the implementation of their drug regimens to their daily schedule. Also of note, more complex dose regimens requiring patients to take more pills a day was a significant independent predictor of misinterpretation of instructions. A prescription requiring a patient to take four pills a day was 47% more likely to be misinterpreted than instructions for a ‘one-a-day’ regimen. Patients with low literacy did not differ significantly from those with adequate literacy in interpreting instructions to take one pill a day, or even understanding “Take 2 pills by mouth every day” and “Take 1 with breakfast and 1 with supper.” Although the latter instruction involved taking pills two times daily, the label broke down the instructions for dose and frequency and provided a context for the time of day.
The limitations to our study should be noted. First, we investigated patient understanding of different styles of writing instructions included on the primary label for prescription medications only. The association between misunderstanding of these instructions and medication error was not examined. We also did not study patients’ actual prescription drug-taking behaviors. Patients’ motivation, concentration and comprehension might have been greater if they were reporting on their own medicine given by their physician for conditions they or their children actually had14,23,24. Second, since the study design did not include a chart review, we did not have information on patients’ health information; in particular whether they had actual experience with the study medications. Third, we primarily manipulated the language for frequency of use; however there were more subtle differences in word choice and numeric presentation of dose on the various drug instructions that may also have altered patients’ understanding. Fourth, patients in our study were mostly socioeconomically disadvantaged individuals from three primary care clinics in diverse areas of the country. Our sample addresses those individuals disproportionately affected by poor health outcomes, and whose health care are targeted for improvement by Healthy People 201025. Finally, the generalizability of our findings are further limited by the fact that patients were predominantly female (an accurate depiction of the clinic patient populations), and that participation was limited to patients who spoke English. This was due in part to criteria for using the Rapid Estimate of Adult Literacy in Medicine (REALM) as our literacy assessment.
While further improvements might be made in the design of prescription drug labels, it is likely that patient counseling will also be needed to address health literacy deficits. Previous research has found physicians do not commonly review the instructions when prescribing medications, nor do pharmacists routinely verbally counsel patients when filling a prescription26–29. Both the American Medical Association and American Pharmacists Association recommend provider training in health literacy communication ‘best practices’29,30. A highly efficacious approach described in recent cognitive factors research, known as “implementation intention” might also aid provider training activities31. This could be a promising health literacy strategy at the provider level, as it refers to a process of helping patients visualize exactly how a prescribed medication will be self-administered within the context of their own daily routine. As minimal standards exist to guide physician and pharmacist best practices for writing and transcribing the dose and frequency of use on label instructions for patients, both professionals should make it their goal to be simple, clear and explicit in directing patients on how to self-administer their medication.
The authors are grateful to Mary Bocchini, Kat Davis, Sumati Jain, Jennifer Webb, Jessica Salazar and Silvia Skripkauskas. The study was supported in part by internal funding from the Health Literacy and Learning Program at Northwestern University.
Conflict Of Interest None disclosed.