Validation of a Polish-language version of the 9-item Hill–Bone Medication Adherence Scale in patients with noncommunicable diseases
CC BY 4.0
Validation of a Polish-language version of the 9-item Hill–Bone Medication Adherence Scale in patients with noncommunicable diseases
CC BY 4.0
Introduction
Noncommunicable diseases (NCDs), also known as chronic diseases, tend to be of long duration and result from a combination of genetic, physiological, environmental, and behavioral factors.1 Insufficient patient adherence to treatment recommendations constitutes a significant problem in the pharmacotherapy of NCDs. It is essential to note that medication adherence is crucial for effective progression of the therapeutic process.2 In accordance with the definition promulgated by the World Health Organization, adherence is defined as “the extent to which a person’s behavior—taking medication, following a diet, and / or executing lifestyle changes—corresponds with agreed recommendations from a health care provider.”3
The efficacy of pharmacotherapy, as prescribed by a medical practitioner, depends on the patient’s adherence to the treatment regimen. Evidence suggests that the level of medication adherence in chronic diseases ranges from 36% to 93% in diabetes mellitus, 47% to 82% in atrial fibrillation, and is about 53.8% in hypertension. One in 10 patients does not fill their prescription for a new type of drug treatment; a phenomenon referred to as “primary nonadherence.”4
Based on the objective criterion of medication adherence, at least 80% of the planned medication doses within specified time intervals should be taken to achieve effective treatment.5 In reality, for chronic diseases, this level only reaches about 50%. For example, the average nonadherence rate to hypertension therapy in Poland is around 60%.6,7 Similarly, a study of the Polish population showed that patients with type 2 diabetes tend not to adhere to medical recommendations.8 According to a report by the National Health Fund, in Poland, 14.7% of patients aged 18 to 40 years filled their prescriptions for reimbursed asthma control medications, while in the age groups of 41 to 55 and 56 to 70 years, the respective rates were 30.7% and 42%.9
The tools used to assess the level of medication adherence comprise, among others, the 9‑item Hill–Bone Medication Adherence Scale (9‑item HB‑MAS) and the 14‑item Hill–Bone Compliance to High Blood Pressure Therapy Scale (14‑item HBCS). While the latter is an evaluation tool designed for the assessment of patients with hypertension, the former can be used in patients with a range of NCDs, including those with multimorbidity.10-13 To date, the 9‑item HB‑MAS has not been validated in a Polish‑language version. This is in contrast to the 14‑item HBCS, which has been validated; however, the availability of its Polish‑language version is limited.14
The objective of this study was to translate, adapt, and validate the 9‑item HB‑MAS in patients diagnosed with NCDs, with a view to making it accessible, open, and freely available to researchers and clinicians in Poland. This type of tool, currently not available in Polish, could be used by health care workers to provide a consistent follow‑up service for patients treated in clinics and hospitals.
Patients and methods
This is a substudy of a multicenter, cross‑sectional MADAM (Medical Adherence in Diabetes and Multimorbidity) project, which investigated the level of medical adherence in patients with diabetes and comorbidities. A cross‑sectional research design was used to examine a convenience sample of individuals treated on an outpatient or inpatient basis. Following approval from the Bioethics Committee of the Medical University of Silesia (BNW/NWN/0052/KB1/1/25), the study was conducted between August 2024 and March 2025. It involved 122 patients with chronic diseases, from the primary health care setting and the Department of Pneumonology in Katowice, Poland. The participants were informed about the purpose of the study and assured of the anonymity and confidentiality of the data. At any stage of the study, they were free to withdraw from participation. Prior to the interview, the patients were requested to provide verbal consent for the study.
Sample size
The criterion applied for establishing the sample size was 10 individuals per 1 instrument item for conducting factor analysis (the higher the ratio, the better). Furthermore, it has been previously suggested that a sample size of over 100 participants is adequate for the purposes of assessing internal consistency, reliability, construct validity, and other relevant analyses.15
Inclusion and exclusion criteria
We recruited adult patients (≥18 years old) with stable NCDs who had their diagnosis confirmed by a physician at least 12 months prior to enrolment in the study.
The inclusion criteria encompassed the following: 1) regular pharmacotherapy, prescribed by the attending physician, lasting for at least 6 months; 2) absence of any impediments related to reading, writing, and comprehension; 3) availability of survey data collected at the initial (T1) and subsequent (T2) time points.
Patients younger than 18 years, those suffering from mental illness or cognitive impairment causing communication difficulties, and those who did not complete the survey at the second time point (T2) were excluded from the study. All data were collected following the principles of patient privacy and anonymity.
Questionnaire
The survey consisted of a set of questions related to sociodemographic factors, such as age, sex, and employment status, the list of diseases for which the patient is taking medication(s), and the 9‑item HB‑MAS by Johns Hopkins School of Nursing.
A 4‑point Likert scale ranging from 1 to 4 (with 1 point indicating that the event occurred all of the time; 2 points, most of the time; 3 points, some of the time; and 4 points, none of the time) was used in the 9‑item HB‑MAS. The overall score ranged from 9 to 36, with 36 points indicating perfect adherence. The Polish translation, cultural adaptation, and validation of the tool were based on the English‑language version, using standard methodology.15-17
The Polish version of the 9‑item HB‑MAS is included in Supplementary material, Table S1. The scale was translated to Polish by 2 independent translators; the results were synthesized, and the final Polish‑language version was back‑translated to English by 2 independent bilingual translators not involved in the project. Thereafter, a comparison between the translated version and the original was made by team members. Face validity was conducted on a sample of 12 patients with chronic diseases who were selected at random from among individuals treated at a primary health care facility. It was reported that no difficulties were encountered in comprehending or responding to the questions. The procedure was carried out by health care workers who had received consistent training and were following a standardized work protocol. In order to assess the test–retest reliability, data were collected at 2 time points (T1 and T2), with a 1‑hour interval between them. No specific guidelines are available concerning the optimal time between measurements to assess the test–retest reliability. Nonetheless, it is crucial that the interval is not too short, as this would result in the study participants recalling their responses from the preceding survey. On the other hand, over a longer period of time, there could be significant changes in the examined feature. The 1‑hour interval between the 2 assessments has been determined as the optimal duration, balancing the need to mitigate the “carry‑on” effect and short enough to allow for test–retest reliability assessment.18
Statistical analysis
Data were analyzed using the Statistica 13.3 package (TIBCO Software Inc., Palo Alto, California, United States). Quantitative variables are presented using descriptive statistics (mean, SD, median, lower quartile, upper quartile, minimum, and maximum). Normality of the distribution was verified using the Shapiro–Wilk test. Categorical variables are expressed as the number of observations and percentages. Additionally, the McNemar test was used to assess the differences in paired categorical data. The internal consistency and reliability of the Polish adaptation of the scale were analyzed using the Cronbach α coefficient, calculated for the entire scale and for the scale with each item excluded in turn. A Cronbach α value above 0.7 is generally considered to reflect good internal consistency. The relationship between an individual item and the scale as a total was tested using the item‑total correlation, which indicates how effectively a particular item corresponds to the overall construct. Typically, a value of 0.3 or higher is considered acceptable, suggesting that the item makes a significant contribution to the overall reliability of the scale. Finally, Cohen κ (with 95% CI) was used to test the inter‑rater reliability between items measured at T1 and T2. Generally, values between 0.61 and 0.8 indicate substantial, and between 0.81 and 1, almost perfect agreement. Moreover, the repeatability of the questionnaire was assessed using the Bland–Altman procedure, applied to the variables on their original scales and, additionally, following logarithmic transformation to ensure the precision of measurement according to non‑normal distribution. In all analyses, the level of statistical significance was set at a P value below 0.05.
Results
Sample characteristics
The study sample consisted of 122 participants (52 men [42.6%] and 70 women [57.4%]) at a mean (SD) age of 58.5 (17.6) years (range, 18–92 years). Of those, 40.2% (n = 49) reported a high school education level and 57.4% (n = 70) were married. A majority of the participants (63.1%; n = 77) reported no current occupational activity and were living in urban areas (70.5%; n = 86). Detailed characteristics are presented in Supplementary material, Table S2.
The 9‑item Hill–Bone Medication Adherence Scale scores
Analysis of overall adherence showed mean (SD) total scores of 33.15 (2.97) at T1 and 33.27 (2.97) at T2, and median total scores of 34 (32–36) at both T1 and T2. The mean individual item scores ranged from 3.48 to 3.93, indicating generally high levels of medication adherence. Similar mean values were observed at both time points (Supplementary material, Table S3).
A comparison of the level of medication adherence at T1 and T2 was also conducted (Supplementary material, Table S4). At T1, 27.9% of the participants demonstrated perfect adherence, whereas 72.1% were classified as nonadherent. By T2, the proportion of the participants with perfect adherence increased slightly to 31.1%, while the proportion of nonadherent individuals decreased to 68.9%. However, this change was not significant (P = 0.13).
Reliability of the scale
The Cronbach α score for the entire scale was 0.79 at T1 and 0.81 at T2, which indicates good internal consistency. Item‑total correlation coefficients ranged from 0.34 to 0.67 at T1 and from 0.3 to 0.71 at T2, reflecting acceptable internal reliability of the scale. The average inter‑item correlation was 0.32 at T1 and 0.34 at T2. Following exclusion of individual items, the Cronbach α values ranged from 0.75 to 0.82 across both time points. The absence of a substantial increase in Cronbach α values upon removal of any item confirms internal consistency of the scale. The values of Cohen κ coefficients ranged from 0.87 to 1, indicating almost perfect agreement over time (Table 1).
Item | Item‑total correlation | Cronbach α if the item was removed | Cohen κ (95% CI) | ||
T1 | T2 | T1 | T2 | ||
Abbreviations: T1, time of the first assessment; T2, time of the second assessment | |||||
| 0.45 | 0.49 | 0.79 | 0.81 | 0.92 (0.85–0.98) |
| 0.47 | 0.51 | 0.78 | 0.8 | 0.88 (0.78–0.96) |
| 0.67 | 0.71 | 0.75 | 0.78 | 0.88 (0.78–0.98) |
| 0.54 | 0.55 | 0.77 | 0.8 | 0.9 (0.82–0.98) |
| 0.53 | 0.57 | 0.77 | 0.8 | 0.95 (0.87–1) |
| 0.58 | 0.6 | 0.76 | 0.79 | 0.88 (0.79–0.96) |
| 0.34 | 0.3 | 0.79 | 0.82 | 0.98 (0.92–1) |
| 0.41 | 0.42 | 0.79 | 0.81 | 1 (1–1) |
| 0.5 | 0.53 | 0.77 | 0.8 | 0.87 (0.78–0.95) |
Finally, the Bland–Altman analysis (Supplementary material, Figure S1) also indicated very good repeatability of the 9‑item HB‑MAS overall score at T1 and T2. The mean difference between the 2 measurements was –0.123 (95% CI, –0.23 to –0.0159). A similar pattern of variability was observed after logarithmic transformation of the variables, with a mean difference of –0.003 (95% CI, –0.007 to –0.0004).
Discussion
The objective of this study was to translate, adapt, and validate the 9‑item HB‑MAS in patients diagnosed with NCDs in Poland. There are 2 primary methods of assessing medication adherence—direct and indirect. Direct methods include the measurement of drug or metabolite concentrations in body fluids, whereas indirect methods include, among others, self‑report adherence scales, such as the 9‑item HB‑MAS.19 We demonstrated satisfactory internal consistency of the Polish adaptation of the 9‑item HB‑MAS. Cronbach α coefficients were 0.79 at T1 and 0.81 at T2, indicating good reliability across both observation points. These values are consistent with those reported for other geographical, cultural, and linguistic versions of the scale, including the Chinese, Amharic, and Romanian adaptations.16,20,21 The overall level of medical adherence remained stable over time, with mean scores of 33.15 at T1 and 33.27 at T2. Additionally, the results of the Bland–Altman procedure and the values of Cohen κ (range, 0.78–1) confirmed very good repeatability and agreement over time. This stability is also supported by the comparable proportions of participants classified as perfectly adherent at both time points, with no significant differences observed (P = 0.13).
To the best of our knowledge, this is the first validation study to examine the utility of the 9‑item HB‑MAS among patients with NCDs in Poland. Admittedly, Uchmanowicz et al14 performed the Polish adaptation and validation of the 14‑item HBCS questionnaire. Still, this tool can only be applied to patients with high blood pressure. The number of items, as well as the target patient group and types of diseases, distinguish these 2 scales.
As an indirect method of assessing medication adherence, the 9‑item HB‑MAS is a cheap and useful tool for primary and hospital care workers. This self‑reporting method allows for information to be obtained from a large group of patients in a short amount of time. Another important advantage is that the test can be adapted to different groups of patients with chronic diseases, such as diabetes, chronic obstructive pulmonary disease, and cardiovascular disease. This tool is noninvasive, and can be quickly and easily interpreted by medical professionals. The popularity of this scale in studies and clinical practice stems from its high predictive validity. A higher level of adherence identified during the test significantly improves disease compensation. This specific instrument could be sufficient for consistent follow‑up services for patients treated in Polish clinics and hospitals, and support health care professionals in identifying and managing nonadherent individuals.
From a formal point of view, this is a single‑center study, but validation was performed in 2 clinical centers with different patient profiles. This makes it possible to extrapolate the results to a wider population.
A potential limitation of our study is a lack of external validity, particularly with regard to the relationship between the questionnaire results and other clinical and adherence parameters. Furthermore, we did not evaluate sensitivity to change, defined by an instrument’s responsiveness to detect a change. To achieve a comprehensive and accurate assessment, it is essential to establish a correlation between the observed scores and other metrics that are expected to undergo change.
The present research covers a broad range of chronic diseases, which makes it useful for specialists across various health care settings. Furthermore, it involves a diverse sample of participants, varying in age (18–92 y), educational level, and occupational and marital statuses.
The Polish‑language version of the 9‑item HB‑MAS was prepared using the back‑translation methodology.17 Such an approach ensured linguistic correctness and reliability of the scale. Of note, the survey was conducted in the same groups of patients (n = 122) at 2 consecutive time points (T1 and T2), which influenced comparability of the results. The Polish version of the 9‑item HB‑MAS complies with all the required criteria for a validation process and is thus suitable for use by researchers and clinicians.
Conclusions
The Polish‑language version of the 9‑item HB‑MAS questionnaire has been shown to be a reliable and valid tool for the assessment of medication adherence in patients with NCDs. This version of the questionnaire can be used by researchers and health care professionals for the evaluation of clinical practice in Poland.
- World Health Organization. Definition of noncommunicable diseases. https://www.who.int/news‑room/fact‑sheets/detail/noncommunicable‑diseases. Accessed April 2, 2025.
- Burnier M. The role of adherence in patients with chronic diseases. Eur J Intern Med. 2024; 119: 1‑5. | Crossref
- Ley C, Putz P. Efficacy of interventions and techniques on adherence to physiotherapy in adults: an overview of systematic reviews and panoramic meta‑analysis. Syst Rev. 2024; 21: 137. | Crossref
- Hershman DL, Neugut AI, Moseley A, et al. Patient‑reported outcomes and long‑term nonadherence to aromatase inhibitors. J Natl Cancer Inst. 2021; 2: 989‑996. | Crossref
- Parikh MA, Ramachandran S, Nsiah I, et al. Estimating optimal thresholds for adherence to RASA medications among older adults with hypertension. J Manag Care Spec Pharm. 2025; 31: 25‑32. | Crossref
SUPPLEMENTARY MATERIAL
ARTICLE INFORMATION