Be on TRAQ – Cross-cultural adaptation of the Transition Readiness Assessment Questionnaire (TRAQ 5.0) and pilot testing of the German Version (TRAQ-GV-15)

Description of the original instrument

The TRAQ 5.0 [38] is a disease-neutral, self-administered questionnaire assessing transition readiness in adolescents and young adults (AYA) aged 12–26 years with chronic conditions and therefore with special health care needs. It comprises 20 items subdivided in 5 subscales (“Managing Medication”, “Appointment Keeping”, “Tracking Health Issues”, “Talking With Providers” and “Managing Daily Activities”). The response set of the English TRAQ is derived from the Transtheoretical Model [44]. Nevertheless, the items score in a range from 1 (“No, I do not know how”) to 5 (“Yes, I always do this when I need to”) with a maximum total score of 100 and a minimum total score of 20. A lower overall score indicates less transition readiness.

Translation and cross-cultural adaptation of the TRAQ

After approval of the authors of the TRAQ 5.0, the cross-cultural adaptation of the questionnaire was conducted according to recommendations from the pertinent literature [45]. Initially, two bilingual health care professionals from the two study sites (Vienna, Austria and Freiburg, Germany) familiar with the TRAQ 5.0 independently translated the original 20-item TRAQ 5.0 to German. Then, two English native speakers (US American and Canadian, fluent in German) both blinded to the TRAQ and not otherwise involved in the study, independently translated it back to English. In a third step, we discussed our back-translated English versions with the authors of the TRAQ 5.0 via skype. Equivalents and discrepancies due to different health systems needed some considerations. Areas that lacked consistency or clarity were discussed. We looked for ways to translate English phrases and colloquialisms in German. These edits were incorporated into the German version. In the original American TRAQ 5.0 five items (item 14 and item 15 as well as item 18, item 19 and item 20, all three items assigned to the “managing daily activities” scale) did not add statistically relevant information in interpreting the TRAQ results: item 14 and item 15 showed low factor loadings, and furthermore, the “daily activities” subscale showed a lower Cronbach’s alpha than the other subscales. Therefore, the 5 items were omitted in the German Version of the TRAQ.

Hence, we obtained an instrument with 15 items where scores fall between maximum 75 and minimum 15. Finally, we pilot-tested the questionnaire for comprehensibility in a convenience sample of ten YSHCN aged 14–19 at the outpatient clinic of the Medical University of Vienna. Minimal changes led to our definite version, TRAQ-GV-15.

Survey sample

Our study population comprised AYA under medical care at two outpatient clinics, the Department of Pediatrics and Adolescent Medicine, Medical University of Vienna and the Center for Pediatrics, Medical Centre – University of Freiburg, Germany. Both clinics provide comprehensive primary care services for a broad variety of chronic conditions. The University of Freiburg is linked to a nationwide transition program [46], but no standardized transition readiness assessment has been implemented so far. In both clinics care providers can extend medical treatment of patients beyond the age of 18 if they think that a transfer to adult care in the present state would pose a health risk to the patients. Hence, the resulting convenience sample from the two study sites consisted of 194 YSHCN aged from minimum 14 to maximum 23 years of following outpatient clinics: Gastroenterology, Nephrology, Endocrinology (Diabetes, Growth and Bones), and Pulmonology. Exclusion criteria were severe cognitive impairment or poor German language skills.

Questionnaire administration

The questionnaire was administered to YSHCN at the two study sites from March 2017 to October 2017. The responses to the self-administered paper-and-pencil survey were collected during routine clinical care. Participants were instructed to answer the items on the survey on their own. However, a study investigator assisted whenever support was needed. Clinical data were obtained from medical records. Demographic data such as socioeconomic and family background, first language and educational level of the patients, were obtained by means of an additional anamnestic questionnaire. Supplementary information related to the feasibility of the TRAQ like “help needed”, “fill-in time” and “consulting time” were documented by the study investigators.

Statistics

We tested the TRAQ-GV-15 for reliability, construct validity, and scale-internal consistency. We applied the class mean imputation method for corrected mean sums to compensate for missing data. However, less than 0.5% of the values of any given item were missing. We conducted standard descriptive statistics for single items and overall TRAQ-GV-15 score description including frequency distribution, means, standard deviation, minimum, maximum and median [47]. To test validity for age and sex we conducted an ANOVA with Bonferroni adjustment for multiple comparisons [48]. A p-value of ¡0.01 was considered as statistically significant.

Because we eliminated five items from the original TRAQ 5.0 there was no evidence for a beforehand structure of our adapted 15 items instrument. Therefore, we conducted an exploratory factor analysis (EFA). We verified adequacy of the sampling with Kaiser-Meyer-Olkin- and Bartlett’s-Test. We conducted Principal Component Analysis and Orthogonal Varimax Rotation with Kaiser Normalization. Internal consistency of the German questionnaire was estimated using Cronbach’s alpha [47]. Statistical analysis was performed with IBM® SPSS® software Version 22.

Sociodemographic characteristics of study sample

The final sample comprised the data sets of participants with given informed consent (n = 172). We included patients with following conditions: Cystic Fibrosis (n = 24), Type 1 Diabetes Mellitus (n = 76), Juvenile Idiopathic Arthritis (n = 49), Chronic Renal Diseases (n = 12) and various rare conditions such as Turner syndrome or Glycogen Storage Disease (n = 11). Mean age of our population was 16.9 years (SD ±1.8) ranging from 14 to 23 with almost two thirds female respondents (60.5%). More than half of our respondents (62.2%) reported that German language was their first or main language, all participants however were fluent in German. Information on insurance status was not included in our survey since general social insurance applies to all our patients. Data on educational level and school type were either lacking (26.7%), inaccurate (6.4%) or grouped as middle school level (8.7%), as vocational training (7.6%) or as high school level (50.6%). Sociodemographic characteristics are listed in Table 1.

Feasibility

Feasibility was tested by assessing if whether or not respondents could answer the questionnaire without help, and whether or not the questionnaire was completed within 10 min or if participants needed extra explanation in addition to a brief introduction. The majority (76.7%) completed the TRAQ-GV-15 within 5 min, about 20% within 10 min, respectively. Only 24 respondents (13.9%) needed help more than once. It became apparent that 15 min of explanation and introduction were sufficient for most participants (78.5%). Data shown in Table 2.

Exploratory factor analysis

EFA was conducted on our sample of 172 respondents to assess a factor structure of the TRAQ-GV-15. Model adequacy was verified with Kaiser-Meyer-Olkin (0.823) and Bartlett’s-Test (p < 0.001), indicating suitability for factor analysis. In the first round of EFA (data not shown) using Orthogonal Varimax Rotation, four factors demonstrated an Eigenvalue ≥1 explaining 54.94% of the total variance. The scree plot showed a clear elbow with only the first factor, demonstrating an Eigenvalue of 4.38. Furthermore, factor 4 comprised only two items. Analysis of reliability is not meaningful with solely two items [47]. Therefore, we conducted another round of EFA extracting three factors accounting for 47.87% of the total variance. The new factor structure led to three clinically meaningful domains for the TRAQ-GV-15: domain 1 “Autonomy” (items 01, 04, 05, 07, 12, 13, 15), domain 2 “Health Literacy” (items 02, 09, 10, 11) and domain 3 “Adherence” (items 03, 06, 08, 14,). All items loaded on at least one factor at a level of more than 0.40 except item15 (“Do you answer questions that are asked by the doctor, nurse or clinic staff”, factor loading 0.388). Factor loadings are demonstrated in Table 3.

Internal consistency

We calculated Cronbach’s alpha Coefficient for internal consistency. The TRAQ-GV-15 presented globally a Cronbach’s alpha = 0.824 which indicates a good reliability [47]. For the three subscale domains, the values of Cronbach’s alpha appeared as follows: Domain 1 Cronbach’s alpha = 0.779, Domain 2 Cronbach’s alpha = 0.721 and Domain 3 Cronbach’s alpha = 0.507. Results are presented in Table 3.

TRAQ-GV item values and overall score

In line with the original TRAQ, the items of the TRAQ-GV-15 score in a range from 1 to 5. Of the expected 2580 (172 responses to 15 Items) values only 12 values were missing which corresponds to 0.47% missing values. We calculated the TRAQ-GV-15 mean overall item score (54.77 ± 10.17) (Table 4). Similarly, we analyzed the 15 items individually, see Table 5. Item11 (“Do you manage your money and budget household expenses – for example: use checking/debit card?”) demonstrated the lowest mean score (2.28 ± 1.50) whereas item03 (“Do you take medications correctly and on your own?”) demonstrated the highest mean score (4.75 ± 0.56).

The data followed a normal distribution, no skew was found. The two-way (2 × 3) ANOVA showed no significant difference for sex. However, age had a significant (p < 0.001 after Bonferroni adjustment for multiple comparisons) effect on the overall TRAQ score. Adolescents of the youngest age group (14–15.99 years) scored significantly lower than YSHCN of the two other age groups comprising adolescents and young adults from 16.00 to 17.99 and 18.00–23.99 years. TRAQ overall mean scores were 48.52 vs. 55.86 and 60.27, respectively (Table 4).

Limitations

The TRAQ-GV-15 is a self-reporting questionnaire which always holds the risk of socially-desirable answers. Moreover, the possibility of underestimation or rather overestimation of one’s own competencies and skills is a frequent phenomenon. External observation like expert-rating or caregiver’s assessment can compensate for these limitations. However, we did not include an external validation in our study design because we aimed at cross-cultural adaptation and validation of the TRAQ-GV structure.

Non-response was primarily caused by the fact that parents did not give informed consent. Therefore, we could not include data collected from YSHCN (n = 20) who managed their hospital appointments without parents or otherwise caregivers. Consequently, we had to exclude AYA who were ready to meet their doctor’s on their own before the age of 18. This might attribute to a bias in our data.