FAQs for Voils Domains of Subjective Extent of Nonadherence (DOSE-Nonadherence) Measure

This two-domain scale was designed to assess missed doses and reasons for missed doses. There are many ways in which a patient can be nonadherent to their medications (e.g., missed doses, failing to fill prescriptions), each of which can be measured with different methods. For a full discussion of this issue, please see:

Kronish, I. M., Thorpe, C. T., & Voils, C. I. (2019). Measuring the multiple domains of medication nonadherence: Findings from a Delphi survey of adherence experts. Translational Behavioral Medicine. Published October 03, 2019.

We purposefully include “nonadherence” instead of “adherence” in the measure name. This decision is based on cognitive interviews in which participants reported that they think of medication-taking in terms of doses missed—rather than doses taken–because missed doses are less frequent and the exception rather than the rule.

The currently licensed version is provided in the following publication. Of note, Duke University owns the copyright, so a license must be obtained before using the measure. Please contact David Chang Villacrese (david.chang.villacreses@duke.edu) at the Duke Office of Licensing and Ventures (https://olv.duke.edu/) to obtain a license.

Voils, C. I., King, H., Thorpe, C. T., Blalock, D. V., Kronish, I., Reeve, B. B., Boatright, C., & Gellad, Z. F. (2019). Content validity and reliability of a self-report measure of medication nonadherence in hepatitis C treatment. Digestive Diseases and Sciences, 64(10), 2784-2797.

Although this paper presents validation in patients with Hepatitis C, the scale was designed to be applied to different populations and/or people with various medical conditions. The Extent of Nonadherence items are universal and can be administered for any medical condition, whereas the Reasons for Nonadherence must be tailored to the medical condition/population of interest (for more information, see How do I tailor the Reasons for Nonadherence scale for my population?) The first version of the scale, with slightly different wording of the Extent of Nonadherence items, has been used in patients with hypertension, dyslipidemia, and type 2 diabetes.

Voils, C. I., King, H. A., Neelon, B., Hoyle, R. H., Reeve, B. B., Maciejewski, M. L., & Yancy Jr., W. S. (2014). Characterizing weekly self-reported antihypertensive medication nonadherence across repeated occasions. Patient Preference and Adherence, 8, 643-650. PMID: 24855340.

Blalock, D., Zullig, L. L., Bosworth, H. B., Taylor, S., & Voils, C. I. (2019). Self-reported medication nonadherence predicts cholesterol values over time. Journal of Psychosomatic Research, 118, 49-55.

Sagalla, N., Yancy Jr., W. S., Edelman, D., Jeffreys, A., Coffman, C. J., Voils, C. I., Alexopoulos, A.-S., Maciejewski, M. L., & Crowley, M. (in press). Factors associated with non-adherence to insulin and non-insulin medications in patients with poorly controlled diabetes. Chronic Illness.

Duke University owns the copyright to this measure, so a license must be obtained for each project. Please contact Galo Mejia (galo.mejia@duke.edu) at the Duke Office of Licensing and Ventures (https://olv.duke.edu/) to obtain a license. After completing the agreement, Duke will send you the measure in the requested language(s) and a user’s guide with scoring instructions.

To expedite the license drafting process, please answer the following questions in your email:

• What is the title of the study or project?
• How many clinics and/or physical locations will be carrying out this research?
• What is your estimated number of uses (patients)?
• What is your estimated effective date and termination date (start & end date of the study or project)?
• What is the research purpose (academic/government/industry)?
• In which languages do you need to use the measure (if other than English)?
• What is the funding source, if any?

The measure is free to use for research studies and for implementation in healthcare settings. For commercial use, please contact David Chang Villacreses (david.chang.villacreses@duke.edu) at the Duke Office of Licensing and Ventures (https://olv.duke.edu/).

Please refer to it as the “Voils DOSE-Nonadherence measure” and reference the currently licensed version:

Voils, C. I., King, H., Thorpe, C. T., Blalock, D. V., Kronish, I., Reeve, B. B., Boatright, C., & Gellad, Z. F. (2019). Content validity and reliability of a self-report measure of medication nonadherence in hepatitis C treatment. Digestive Diseases and Sciences, 64(10), 2784-2797.

The Extent of Nonadherence scale includes multiple items so that measurement error associated with individual items is averaged out when individual item scores are combined into a total score. The items are similar because they capture a single construct (i.e., missed doses). Removing items may reduce reliability and affect validity. Single items cannot have their reliability estimated with internal consistency (i.e., alpha), generally have reduced content validity compared to multi-item scales, and are limited in their ability to discriminate. Of note, respondents generally interpret “I skipped a dose of my medicine” as being intentional nonadherence and “I missed my medicine” as unintentional; “I did not take a dose of my medicine” has been interpreted both ways. Because these questions are interpreted differently, and because nonadherence can be intentional or unintentional, it is necessary to administer all questions. It may be useful to explain these concepts to staff administering the scale so they can respond to questions from respondents about why the items seem similar.

The measure was designed to capture missed doses rather than late doses. Whether and when a dose is considered “late”—and what to do in that case (i.e., skip a dose vs. take it)—is medication-dependent. Therefore, we recommend including an optional instruction, “If you took the medicine later than usual, (investigator may choose: do/ do not) count it as a missed dose,” that can be included if appropriate.

The lists included in our publications are neither exhaustive nor fully inclusive of the possible reasons for nonadherence in various contexts. We expect that a subset of the reasons will be relevant to medications for other condition but that additional reasons may need to be added to be relevant to the target condition/population. Furthermore, investigators or clinicians might have time constraints that preclude administration of an exhaustive set of reasons. Adding or deleting Reasons for Nonadherence items is expected and does not constitute modification of the measure. Users of the Reasons for Nonadherence measure should be aware that additions or deletions of items assessing reasons may increase or decrease content validity, respectively.

If the Reasons for Nonadherence measure is to be used for a new medical condition, we recommend conducting literature review and, if possible, further qualitative evaluation (i.e., interviews or focus groups) to determine a priori 1) which reasons in the current list are appropriate and inappropriate; and 2) which reasons not listed in the current version might be appropriate to add. The instructions, formatting, response scales, and recall period should not be modified because they were developed and evaluated through cognitive interviews. If any reason items are added, we recommend pre-testing them (e.g., via cognitive interviews) prior to fielding them in your study. Please consider sending any newly developed reasons to Dr. Voils for dissemination with the measure so that other researchers can use them.

Studies evaluating cognitive processes in survey response have not supported the superiority of fully labeled vs. end-labeled response scales. We use end labeling to reduce cognitive load (i.e., because it’s easier to hold the endpoints in mind, particularly as there are so many items).

• Extent of Nonadherence scale: Users of the Extent of Nonadherence scale should be aware that even minor changes to the wording of items or response scale may impact reliability and validity. Therefore, modifications to the wording of items or response scale for the Extent of Nonadherence scale are not permitted per the licensing agreement.
  • Reasons for Nonadherence measure: Additions or subtractions to the Reasons for Nonadherence domain are expected and permitted. If any reason questions are modified, we recommend pre-testing the items (e.g., via cognitive interviewing) prior to fielding them in your study. The Reasons for Nonadherence response measure should not be changed as it has been evaluated in cognitive interviews in various medical conditions. Having a common response scale will enable comparisons medical conditions.

Cognitive interviews in different medical conditions have identified 7 days as an optimal recall period for daily medications that balances recall accuracy with response burden. Participants generally cannot recall longer time periods accurately. Investigators interested in assessing medication nonadherence over longer time periods are advised to conduct repeated assessments (e.g., weekly survey of adherence). To reduce participants’ desire to adjust their responses to reflect longer time periods, we include a sentence in the instructions advising them that 7 days may not represent what they do over longer periods.

In cognitive interviews with patients with three common cardiometabolic diseases (i.e., type 2 diabetes, hypertension, and dyslipidemia), we compared administration of a single, universal version asking participants to consider extent of and reasons for nonadherence to medications for all of these diseases vs. three sequential, disease-specific versions. Patients reported they would be able to generate more accurate answers with sequential, disease-specific versions (manuscript in preparation). This finding needs to be confirmed in larger quantitative studies. Another advantage to administering disease-specific versions is that each can have a tailored list of reasons for nonadherence.

In cognitive interviews with patients with type 2 diabetes taking both oral and injectable medications, we compared administration of a single version asking participants to consider oral and injectable medications vs. separate versions assessing nonadherence to oral and injectable medications. Participants reported that they would be able to generate more accurate answers with separate versions (manuscript in preparation). This finding needs to be confirmed in larger quantitative studies. Another advantage to administering separate versions is that each can have a tailored list of reasons for nonadherence (e.g., the injectable version can have reasons such as fear of needles).

Ideally, one should validate the measure in a new population to ensure that it will retain ideal psychometric properties. We make the following recommendations:

• Extent of Nonadherence scale: We designed these items to apply to any medical condition. If resources permit, consider enrolling a small sample to evaluate the internal consistency reliability (alpha) of the Extent of Nonadherence items. Coefficient alpha is based on inter-item correlations, and the inter-item correlations should be positive and large in magnitude, so a sample size as few as 30 patients would provide meaningful information.
• Reasons for Nonadherence measure: We realize that researchers may have few resources to carry out a full-scale validation before including the measure in a study. At minimum, to determine whether additional reasons for nonadherence need to be added, consider conducting 8-12 cognitive interviews or 1-2 focus groups with (1) patients of the target population representing diverse backgrounds (e.g., race/ethnicity, literacy levels, age), and/or (2) clinicians with appropriate expertise/knowledge. For brevity’s sake, you may also evaluate whether some items could be dropped or a higher-order reason could be used in place of two or more reasons.

Our goal is to have a single translated version, rather than multiple, similar versions, for each language. Before embarking on translation, please check Dr. Voils’ website and contact the Duke Office of Licensing and Ventures to make sure the measure has not already been, or is not currently being, translated to your desired language.

All translations are considered derivatives of the scale and, as such, the copyright for translations belongs to Duke University. When translating the measure, the key is not to have a linguistically equivalent translation, but for the items to capture the concepts reflected in the items. There can be flexibility in the translation to capture those concepts. For questions about the intent of each item, please contact Dr. Voils. Users may wish to refer to the following papers for guidance on scale translation:

• Eremenco, S. K., Cella, D., & Arnold, B. J. (2005). A comprehensive method for the translation and cross-cultural validation of health status questionnaires. Evaluation & the Health Professions, 28, 212-232.
• Wild, D., Grove, A., Martin, M., Eremenco, S., McElroy, S., Verjee-Lorenz, A., & Erikson, P. (2005). Principles of good practice for the translation and cultural adaptation process for patient-reported outcomes (PRO) measures: Report of the ISPOR Task Force for Translation and Cultural Adaptation. Value in Health, 8, 94-104.
• Dewolf, L., Koller, M., Velikova, G., Johnson, C., Scott, N., Bottomley, A. (2009). EORTC Quality of Life Group Translation Procedure (3rd ed.). Brussels, Belgium: EORTC Quality of Life Group.