Translation, cultural adaptation and validation of the Chinese Multimorbidity Treatment Burden Questionnaire(C-MTBQ): a study of older hospital patients

The data was collected from one hospital in Zhengzhou City, Henan province, China between August and October 2019. Patients were invited to take part if they met the following inclusion criteria: (1) complete hospital records and diagnosis at time of discharge; (2) two or more long term conditions; (3) age ≥ 60 years. Patients were excluded if they had a serious physical or mental illness. The sample size of 156 was calculated using the Consensus-based Standards for the Selection of Health Measurement Instruments (COSMIN) [11]. This guideline advises 5–10 participants per item of the questionnaire. We slightly increased the sample size to account for incorrect filling of questionnaires.

The Multimorbidity Treatment Burden Questionnaire (MTBQ) has 10 questions and three additional optional questions (questions which did not apply in a UK context but may apply to other populations). For the C-MTBQ, -these three additional questions were included. For each question, answers are ranked using a 5-point Likert scale, ranging from 0 (does not apply or not difficult),1 (a little difficult), 2(quite difficult), 3(very difficult) and 4 (extremely difficult). The global MTBQ score is calculated as the mean score, multiplied by 25, giving a score of 0 to 100. Global treatment burden scores can be categorized as: no-treatment burden (0), low treatment burden (< 10), medium treatment burden (10–22), high treatment burden (≥22).

The Treatment Burden Questionnaire (TBQ) was originally developed by Tran [12] and measures the perceived treatment burden of patients with long term conditions. It has been translated into many countries [12, 13]. It includes 15 questions and four dimensions: medication(1A ~ 1D), seeing doctors and subsequent visit (2A ~ 2E), medical related and lifestyle(3A ~ 3E) and health problem(4A), A 10-point Likert scale is used, ranging from 0 (no burden) to 10 (high burden). The global score of the TBQ is calculated as the sum of the answers to each item and ranges from 0 to 150, with higher scores indicating a higher level of treatment burden.

Researchers followed the Brislin model to translate the MTBQ into Chinese [14]. Step 1: The MTBQ was independently translated into Chinese by two researchers with Chinese linguistic backgrounds(T1, T2). T1 has a Master’s degree and is medically trained and T2 has a postgraduate degree in English with no medical training. Step 2: A third researcher, who has a Master’s degree in nursing and 6 years of IELTS training experience, reviewed and synthesized the translated versions created by T1 and T2, referring to the original English version. Step 3: two bilingual translators(One is a Master’s degree student in English, the other is a Nursing PhD)back-translated the synthesized version into English to highlight conceptual errors in the translations. Step 4: All translators and team members compared the original MTBQ and the two back-translation versions to form a comprehensive back-translation version to confirm accuracy [15]. The comprehensive back-translation was reviewed by the original MTBQ author(PD), and minor revisions were made until the comprehensive back-translation version and the original questionnaire had a semantic consistency rate of over 90%.

Six experts conducted the cross-cultural adaptation: two general medical doctors; two hospital nurses who specialize in long-term conditions; a nurse who specializes in care of the elderly and the director of the care of the elderly hospital department.

They first evaluated the Chinese version of MTBQ in terms of accuracy, simplicity of the text, grammar, use of proper terms and syntax. The cultural relevance (language clarity, language habits, cultural background conformity and content relevance) and content validity were evaluated using a 4-point rating scale ranging from 1 (not relevant) to 4 (very relevant) to ensure the cultural applicability and content equivalence of the questionnaire [16]. The content validity index (CVI) (the proportion of questions rated by experts as either 3 or 4) was calculated.

The final version of the MTBQ was piloted in 30 elderly patients who met the inclusion criteria. We found that they could understand the items easily and that they required an average of 5 minutes to finish the questionnaire.

The questionnaires were administered face-to-face by three post-graduate students who were trained by the research team. Participants were identified by nursing staff working on the hospital wards. Data from the patient’s hospital records was collected with written consent.

Data were analysed using SPSS version 21.0. Descriptive statistics were generated to describe the participants’ characteristics. The distribution of responses for each question, proportion of missing data, proportion of ‘does not apply’ responses and floor and ceiling effects were examined. An exploratory factor analysis (EFA) was conducted to evaluate the dimensionality of the questionnaire [17], and the number of extracted factors was determined using the principal-component analysis (PCA) and varimax rotation. Factor loadings (λ) > 0.40 or < − 0.40 were considered acceptable [18].

To assess internal consistency reliability, we examined the inter-item correlation matrix and calculated Cronbach’s alpha (0.7–0.95 was deemed acceptable) [11, 19]. Test-retest reliability was assessed by comparing the total C-MTBQ scores in a subset of patients who completed the questionnaire on both day 0 and day 14. The intraclass correlation coefficient (ICC) was calculated and interpreted as follows: excellent (> 0.8), good (0.61–0.80), moderate (0.41–0.60), fair (0.21–0.40) and poor (≤0.20) [20]. To assess criterion validity, we examined the relationships between C-MTBQ and the Chinese version of the TBQ.

We examined the distribution of scores for each question. Ceiling and floor effects were considered to be present if more than 15% of respondents achieved the lowest (0) or highest (4) score, respectively.

The study was approved by the ethics committee of People’s Hospital of Zhengzhou. Written consent was taken for all the participants.

“Health professionals” (questions 6, 7, and 8) was replaced by the traditional Chinese word for “medical staff” as the experts felt this was, the more commonly used expression. During interviews with participants, for the “appointments” in questions 6 and 8, the participants reports that there was no need to make an appointment to see a doctor in hospital mostly, so we removed the word “appointments”. The item CVIs ranged from 0.83 ~ 1.00, while the scale CVI was 0.89, indicating good content validity of the C-MTBQ.

One hundred and fifty-six participants completed the study. There characteristics are shown in Table 1. The average age was 73.5 years (60–99 years) with slightly more males (54.5%). Less than a fifth had a college education. Two thirds of patients had three or more long-term conditions.

The proportion of missing data for each question was 0% (see Table 2). For the optional question about “Getting help from community health services (eg, physical therapy, health services provided by community nurses, etc.)”, 64% of patients responded “does not apply”. As this was greater than 40%, this question was removed from the questionnaire. High floor effects (the proportion of participants who responded ‘not difficult’ or ‘does not apply’) were found for all questions. The range of skewness was between 0.453 and 2.093, and the range of kurtosis was between 0.040 and 3.721, indicating that the items of the C-MTBQ were non-normally distributed. The global C-MTBQ scores were skewed and varied from 2 to 60. None of participants had a global C-MTBQ score of 0, and no treatment burden(0), low treatment burden(< 10), medium treatment burden(10–22) and high treatment burden (≥22) accounted for 0, 14.7, 39.7 and 45.5%, respectively. The median C-MTBQ total score was 20.8 (interquartile range 12.5–29.2).

The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy(0.776) and the Bartlett test of sphericity (p < 0.001) showed that factor analysis of the data was appropriate [21, 22]. The scree plot is shown in Fig. 1. Three common factors were obtained, which explained the total variance of 54%, loadings of all items ranged from 0.505 to 0.816. The eigenvalue of the three factors were -2.494, 2.022 and 1.960, respectively. The original English MTBQ had only one dimension. According to the content characteristics, factor 1 (1,2,6,7) was named as the medication and treatment dimension (4 items); factor 2 (3,4,8,9,13) was named as the medical related dimension (5 items), and factor 3(5、11、12) was named as the daily self-health management dimension. The three dimensions represented are shown in Table 3.

The Cronbach’s alpha and the corrected item-total correlations of all 12 questions are shown in Table 3. Item-total correlations ranged from 0.403 (item 9) to 0.701 (item 4), and all items met the recommended minimum of 0.20. Removing any item of the C-MTBQ didn’t result in severely changing in the value of Cronbach’s alpha. The internal consistency coefficient for the total score of the C-MTBQ were showed with a Cronbach’s alpha of 0.755. Thirty of the 156 participants were randomly selected to evaluate the test-retest reliability. The mean scores of the first and second measurements were 19.79 (interquartile range 19.79–29.17) and 21.00 (interquartile range 14.75–25.50), respectively. The test-retest reliability was satisfactory with an ICC of 0.944.

Correlations of C-MTBQ with TBQ and TBQ subscales scores are displayed in Table 4. The results suggested moderate to high correlations of C-MTBQ with TBQ score (r = 0.875) and TBQ subscales scores (r: 0.495 ~ 0.740).