Background

In 2004, the NIH launched a program of research called the Patient Reported Outcome Measurement Information System (PROMIS®) [1]. The goal of PROMIS was to provide clinicians and researchers access to short, precise, valid, and responsive adult- and child-reported measures of health (for more information see www.healthmeasures.net; for pediatric applications see [2]) via advanced measurement methods (Item Response Theory – IRT [3]) as well as computerized adaptive testing (CAT [4]). PROMIS was carried out by a network of research and health care centers across the US with collaborations worldwide aiming at creating standardized questionnaires, which can be used for national and international comparisons of clinician’s and study results [5, 6]. It provides a conceptualization of health for adults and children and developed, validated, and translated item banks for short and long fixed-length forms and CATs. The advantage of utilizing IRT methods and CATs is that item banks can be established from which either a CAT or a specific short-form can be drawn, thus minimizing the number of items displayed to the child while maintaining measurement precision and responsiveness [7,8,9,10]. Since 2013 there has been a steady increase in the translation of PROMIS measures including into Chinese [11, 12], Dutch-Flemish [13, 14], Spanish, German [15,16,17], Nepali [18] and other languages.

The PROMIS conceptual framework of pediatric health subsumes the domains of physical, mental, and social health under the general concept of global health. Each domain comprises a set of subdomains such as mobility, fatigue, and pain (physical health), anxiety, depressive symptoms, and life satisfaction (mental health) or family and peer relationships (social health; see Fig. 1).

Fig. 1
figure 1

PROMIS Conceptualization of pediatric health

Full size image

Due to our interest in the above mentioned methodological potentials and our experience in developing pediatric health-related quality of life (HRQoL) measures such as the KIDSCREEN [20] and the German Kids-CAT [19], we first focused on developing and empirically testing the PROMIS pediatric subjective well-being domains (SWB) including the subdomains positive affect, life satisfaction, meaning, and purpose [21,22,23]. Since 2012 our team has also carried out the first large-scale translation of PROMIS pediatric item banks to German (and Spanish) to improve international comparability of health screenings of children and adolescents [15]. We first translated and cross-culturally adapted eight PROMIS pediatric item banks for use in the BELLA study [24], which extended beyond SWB to physical activity, experience of stress, and family relations ([15, 16], see Fig. 1 with previously translated item banks framed in blue). Subsequently, we translated six further pediatric PROMIS item banks (see Fig. 1 item banks framed in red) to German, which is the focus of this article. Specifically, we translated the item banks (version 1) and short forms (version 2) of the PROMIS pediatric subdomains anger (ANG, [25]), anxiety (ANX, [26]), depressive symptoms (DEP, [26]), fatigue (FAT, [27]), pain interference (P, [28, 29]), and peer relationships (PR, [30]) both as self-report and proxy-versions [31], to a universal German version. The objective was to attain semantically and linguistically cross-cultural as well as content- and conceptually equivalent translated item banks and short forms, which can be used in Germany, Austria, Switzerland, and other German-speaking countries like Luxembourg for example.

Methods

Translation approach

A total of 164 items with 82 self and 82 proxy item versions were translated: ANG (5 items), ANX (13 items), DEP (13 items), FAT (23 items), PAIN (13 items), and PR (15 items), plus 3 divergent items of proxy versions (for ANG and DEP, FAT), and a Likert scale with 5 response options. Higher scores on those PROMIS items correspond to higher levels of the concept name (e.g., greater fatigue or better peer relationships). We translated the item banks to be used with German-speaking children in Germany, Austria, and Switzerland (one universal German version). To account for variations in wording between countries using the same language, our translation team was comprised of two native translators from each country (six translators in total), who were external and recruited for the purpose of this project. In addition to those translators, two experts acted as translation process managers (TPM), who oversaw and managed the process (see Appendix 1).

We applied the “universal translation approach” following a widely used, multi-step forward–backward translation technique with cognitive interviews complying to the recommendations of the FACIT translation methodology (see Fig. 2 [32, 33]), which is the current PROMIS translation standard and in line with the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) PRO Translation Task Force [34, 35] and PRO Consortium recommendations [36].

Fig. 2
figure 2

FACIT translation methodology

Full size image

The same methodology was applied to the translation of 8 PROMIS pediatric item banks in 2018 [16] as well as to other pediatric translations processes [11, 13] and may be the best choice as long as no gold standard of translation exists [37].

Translation guidelines

First, translation guidelines were written based on a review of the literature, translation guidelines, and PROMIS standards (see Protocol, Appendix 2). Items were to be translated as accurately as possible, keeping the intended item definition/content in mind and producing easy, child-friendly item translations avoiding region-specific or colloquial words.

Original items

Then PROMIS provided the original item versions and conceptual and technical item definitions, based on which a comprehensive item history template was developed to guide the translators, facilitate accurate translation, enable thorough documentation, and ensure conceptual/cross-cultural equivalence (see Table 1 as an example).

Table 1 Exemplary item definitions provided by the U.S. authors and forward translation template developed for the translation process
Full size table

Item translation

The TPMs informed the translators about the translation process, provided the items, and instructed the translators. Two independent native translators specializing in pediatric research from each country (Germany, Austria, Switzerland) were asked to review the items for translatability, that is to comment on the translation difficulty issues such as unclear idioms, ambiguous words, unclear concepts, etc. The TPMs summarized those comments and reported them back to the PROMIS test developers to clarify and/or improve the item definitions, if needed, so that concept validity was ensured. Subsequently, the items were forward translated (FTs) to German. In each country, translators worked independently from one another following the translation guidelines (Appendix 2) to create the translations. They rated the translation difficulty of the items using a three-point Likert scale (1: easy, 2: feasible, and 3: difficult to translate).

Reconciliation process

A two-step reconciliation process was conducted: The translators first reconciled their versions within each country agreeing upon a good version of the FT. Then all translators and experts from each country held a teleconference to harmonize the versions to a universal German version, the final product of which was a Reconciled Universal Forward translation (RUFT).

Back translation and consented universal translation

The German versions were back translated (BT) by an independent professional translator not involved previously, who had no knowledge of the original English source items or the item definitions. An expert group consisting of the back translator, forward translators and the TPMs reviewed the BTs and the FTs, and the translations were revised where necessary, based on the equivalence between BTs and the source versions. In this way a consented universal translation (CUT) was created. The CUTs underwent a quality review, were formatted, and proof-read by an independent PROMIS translation expert.

Cognitive interviews

Cognitive Interviews (CIs) were conducted in Germany, Austria, and Switzerland to ensure the feasibility, relevance, and equivalence in comprehension of the CUT [32, 38,39,40]. Prior to that, ethical approval for the translation projects had been given by the Ethics Committee of the Chamber of Psychotherapists (Psychotherapeutenkammer) of Hamburg, Germany. For the CIs, children from the general population, residing in the specific countries (Germany, Austria, Switzerland) were included under the condition that they were able to speak and read German and had no mental or physical condition that interfered with their ability to be interviewed. Children and parents/caregivers were recruited and informed about the study via email and gave their informed consent before participation. CIs of the self-report items were performed with n = 58 children and adolescents aged 8–17 years (Germany: n = 16, Austria: n = 22, Switzerland: n = 20). For practical reasons, focus groups with 2–8 participants were chosen for the cognitive debriefing in Austria (Vienna/Linz) and Switzerland (Bern), whereas in Germany (Hamburg) participants were mainly interviewed individually and focus groups were only conducted if siblings or close friends participated. CIs were performed by trained staff.

To ensure that parents found the items easy to understand and understood them the same way as intended, CIs were also performed with n = 42 parents or other caregivers (of 8–17-year-old children/adolescents; Germany: n = 12, Austria: n = 17, Switzerland: n = 13).

The interviews lasted about 60 min and started with the interviewer giving the 32 children/adolescents and parents a questionnaire containing a selection of 25 PROMIS items and asking them to rate them on a three-point Likert scale (1 = difficult, 2 = potentially problematic, 3 = easy to understand). Then the interviewers asked questions about the cognitive understanding of the items using a mix of cognitive techniques, such as “general probing” and “paraphrasing” [40]. Interviews were recorded and pilot testing reports were written, which were fed back to the translators and TPM, in order for the items to be optimized when needed. Children and adolescents received a 10 Euro gift voucher for their participation.

The CI results were reviewed by the TPMs and feedback was compiled. The translation team then discussed the results and agreed on final changes based on the translation issues identified during the CIs. The final versions were finally reviewed again and approved by the PROMIS Statistical Center.

Note that the described steps were performed mainly for the self-report versions of the items, because the proxy versions have the exact same wordings except that the perspective is changed (i.e., “I” was replaced by the words “my child”).

Results

Translatability review

Using the three-point Likert scale, translators found most items “easy” to translate (69%), 26% of the items slightly more challenging, but “feasible to translate”, and 4% of the items were categorized as “difficult to translate”. For example, difficult items used ambiguous words (“feel mad” = could either mean “angry” or “insane”), idioms (“it hurt all over my body” = my whole body hurts) or unclear words, which led the translation manager to clarify their content with the US developers. This was also the case for the item “I felt upset” (ANG items), the item word “worried” (ANX item; the translation “anxious” was considered too strong a translation by the developers, instead preoccupied/concerned were suggested as translations) and the DEP items “I felt alone” versus “I felt lonely” (the first item implying that no one is physically present, the second being possible if others are present). The translation which communicated emotional loneliness best, regardless of physical presence was kept, i.e., “Ich fühlte mich einsam”.

In the FAT item bank, in 10 out of 23 items the translation expert raised an issue with the words “too tired (to…)”. This wording does not work well in German because using the literal translation “zu müde” in German implies that things cannot be done, whereas in English the items imply that things may still get done although with difficulty. However, to capture the English original content as much as possible the literal translation was kept after a thorough discussion.

In the PAIN item bank, the item “It was hard to have fun when I had pain” the TPM needed to ask the developers whether “have fun” referred to the feeling of enjoyment, or to participating in joyful activities (the former was the case).

Figure 3 summarizes the translation difficulty ratings across the item banks. On average German translators rated the items more difficult to translate (average = 1.5; standard deviation (SD) = 0.20) than the Austrian (average = 1.3; SD = 0.16) and the Swiss translators (average = 1.2; SD = 0.14).

Fig. 3
figure 3

Translation difficulty ratings

Full size image

The ANG item bank was the most difficult to translate due to item words such as “I felt fed up” where item definitions were multifaceted, covering feeling impatient, disgusted, and bored beyond endurance. The ANX item bank was rated the easiest to translate. Six out of 13 items used the word “worried”, which is easily translatable to “besorgt/sich Sorgen machen”, and items using words such as “scared or afraid” seemed easy to translate as there are similar anxiety words in German (Angst haben, fürchten, erschrecken).

For the DEP item bank, translation issues arose with the item “I didn’t care about anything”, because it is, combined with the response option “never”, a double negative. The items using the word “thing/s” posed some challenges to the translators, too (e.g., “I had trouble finishing things”, German: “Dinge”/ “Aufgaben”).

In the PAIN item bank, the word “run” and “(walk a) block” were difficult to translate, due to different possible translations of run: “laufen” (run fast), “gehen” (walk), “rennen” (sprint), “joggen” (jog), and due to the concept of a block not being a common distance measure in German. The latter was solved by describing “a block” as a short distance (“kurze Strecke (100 m))” as recommended by the test developers.

For the PR item bank, translation issues were that the item “I was good at making friends” was hard to translate as it originally captures how easy it is for someone to make new friends “in the last week”. The underlying American concept of making friends is different from the German concept; there, making friends is usually considered to be a process of several weeks instead of one week. In addition, there is no natural, literal good way of translating “I felt good about my friendships” in German. Translators finally decided on “habe ich mich mit meinen Freundschaften wohl gefühlt”, but this still does not sound very natural to a native German child. Further, the concept of “I (played alone and) kept to myself” was not that easy to translate to German, because translators doubted that young children understand the content easily. They finally decided upon “(Ich habe allein gespielt und) bin für mich selbst geblieben”.

Reconciliation process

The reconciliation process took about 2.5 h and the translators and experts carefully discussed most items to ensure cross-cultural equivalence. Cross-cultural differences in wordings became apparent. For instance, the item “I felt fed up” was translated “Ich hatte die Nase voll” (literally: “my nose was full/I was fed up”) by German and Swiss translators, while the Austrian translators suggested “Ich war angefressen” (literal “I was eaten up/fed up/full”). The latter was not used, because in Swiss German it would have meant that the child is very interested in something.

Also, the word “angry/mad” was discussed among translators and translation options included “verärgert”, “wütend”, “sauer” (colloquial German) or “zornig” (preferred by Austrian translators). In the end, “wütend” as a clear moderate emotional word was agreed upon for two items and “zornig” for one item.

For some item banks, translators discussed that US items often included the word “feeling” (e.g., “I felt nervous”), while in German the natural equivalent is “being” (e.g., “I was nervous”). To improve item understanding for German children the word “felt” was replaced by “was”.

Also, in the reconciliation process it was found that the letter sharp “ß” is not used in Switzerland. It was suggested to create different language versions or write the word without the “ß” in brackets in universal items (e.g., “Ich habe mir Sorgen darüber gemacht, was mir zustoßen (zustossen) könnte”, “Es ist mir schwergefallen Spaß (Spass) zu haben”).

For the PR item bank, translators discussed whether words like “kids” were age appropriate for children and adolescents up to 17 years of age, and it was suggested to translate “kids” as “Kinder/Jugendliche” (i.e., add youths to the item). In addition, test developers were asked to reword the word “play” for adolescents, but in the end those items included other activities than playing that were age-appropriate for youths, so the word “play” remained as one activity next to the others in the item. Also, it was discussed that while it may take longer to make friends the older a child becomes (PR item “it was easy for me to make friends”), in times of digitalization the German concept of making “friends” may have changed as German children/adolescents tend to refer to peers as “friends” more quickly than a decade ago (when friends were only called friends after a very substantial amount of time). This issue could not be resolved in the translation process. Another age-related translation issue occurred with the item “other kids wanted to be with me”, of which the most common translation was “andere Kinder wollten mit mir zusammen sein”, however translators discussed that teenagers may misunderstand the phrase “zusammen sein” as “dating”, which would distort the meaning. Hence the universal translation “andere Kinder/Jugendliche wollten mit mir Zeit verbringen” (“other kids wanted to spend time with me”) was agreed upon.

Back translation (BT)

The back translations showed some small wording differences. For instance, for the ANG bank, words like “(being) furious” and “annoyed” were used, which were not part of the original items, but fit the content. For anxiety, “anxious” was synonymously used as “scared” and “afraid”. For depression, the word “felt (sad)” in English was often replaced by the word “was (sad)” – replicating the issue discussed above. For FAT, the BT showed that the word “bed” had been mistakenly omitted from one item in the translation process and was again inserted to ensure literal equivalence. For PAIN, the BT showed that the duration of standing still was implicit in the original item version but needed to be specified in the translation to avoid misunderstandings. In total, the review of the BTs led to six minor changes in item wording.

Cognitive interviews (CI)

Fifty-eight children and adolescents and 42 of their caregivers took part in the cognitive interviews (see Table 2).

Table 2 Age and gender of the sample, who underwent cognitive interviewing
Full size table

The CIs took up to two hours, with parent interviews taking longer than those with younger children. Overall, the motivation of the children and parents was very high and children and parents liked the CI as the material displayed was perceived as child-friendly (e.g., with smileys etc.). The CIs showed detailed insights into the comprehension of the item words by children and parents. After the recorded sessions were summarized and reported to the TPM, the CI results led to minor revisions of seven items, which are displayed in Table 3.

Table 3 Results of the cognitive interviews, which were particularly interesting and resolved in the cross-cultural harmonization process
Full size table

Discussion

The PROMIS pediatric self- and proxy report items of the item banks (version 1) and short forms (version 2) covering the subdomains anger, anxiety, depressive symptoms, fatigue, pain interference, and peer relationships were translated well to German (for exemplary items see Appendix 3). The translation process proved most items easy or feasible to translate and most items were understood as intended in pretesting. Only 14 out of 82 items of the self-report and 15 out of 82 items of the proxy report items required minor rewording during the translation process. Our study shows that it is important to avoid ambiguous and multifaceted words when developing a questionnaire because these words will create translational difficulties later on. In general, the translation of some ANG items proved to be the most difficult one as some of the items were multidimensional. For example, the item word “upset” evoked translation issues at several stages of the translation. These were similar to translation difficulties found in translations of this item to German by Jakob et al. ([17] in a depression measure) and to Spanish ([41] in the Neuro-QoL). Interestingly this item was also different in the psychometric analyses performed by Kaman et al. [42] in that it was the only item which did not show floor effects on the anger scale.

In addition, the study found cross-cultural differences in concepts, e.g., between the German and the US culture regarding the speed with which participants make friends, or regarding the comprehension of words “zornig” vs. “wütend” between the German, Austrian-German, and Swiss-German languages.

Further, we would like to point out that questionnaires should always keep the target participants in mind. We found that teenagers may not feel addressed when using words such as “children” in the instruction/item texts requiring us to add the word “adolescents” (even though this was not part of the original item text).

Similar to previous studies [12,13,14, 16], translating units of measurement/metrics (blocks, meters etc.) usually pose some challenges cross-culturally and were solved accordingly. Differential item functioning analyses are needed to explore whether cross-culturally different measurement units (e.g., metric vs. imperial measures; colloquial terms such as a “block”) may result in different item difficulties and may need language-specific item calibrations.

Further, similar to our previous study, the translation of some items needed two (synonymous) words to capture the intended concept in a way that all children/adolescents of varying ages and of all German-speaking countries could understand (see also [16] or [18]).

The strengths of this translation are its rigorous PROMIS translation methodology and the large sample size of particularly Swiss and Austrian children and parents interviewed in cognitive debriefings – exceeding the CI sample sizes of our previous translation project [16]. Limitations of this study were that many of the children and parents participating in the CIs were overall healthy and only age and gender was assessed for the participants. While this study shows very good comprehensibility of the translated items, it may be that children or parents with a lower socioeconomic or educational background, those coming from an ethnic minority background or those with chronic diseases or learning disabilities may have difficulties in understanding the items. This needs to be investigated in future studies with more heterogeneous samples. Further, the CIs did not include questions about the relevance of the items as indicators of the domains assessed in Germany. In addition, this study did not include psychometric analysis or norming of the translated items (for US norms of the PROMIS source measures see [43]), thus criterion and technical equivalence of the translations need to be checked by psychometric validation studies (e.g., [44] or [45]) to provide additional information of the degree of cross-cultural equivalence achieved in the present study.

In 2019 our research group performed the first psychometric analyses of a German PROMIS pediatric translation [42]. The translated pediatric anger item bank demonstrated good psychometric properties, including satisfactory distribution characteristics, unidimensionality, good internal consistency, and congruent validity. Also, German normative data for this item bank was established. In 2014–2017, the BELLA study administered the German PROMIS pediatric translations of the subjective and family well-being item banks, physical activity, and peer relationship item banks as well as the PROMIS Global Health scale [46], and for adolescents aged older than 18, the Profile 29 [47]. Thus, German representative data is now available for psychometric analysis and norming of the translated tools.

Conclusion

Six pediatric PROMIS® item banks (anxiety, anger, depressive symptoms, fatigue, pain interference and peer relationships) were successfully translated into German and adapted for use in Germany, Austria, and Switzerland with relatively minor changes to the content of the original English items. The translated item banks cover the items of the corresponding current PROMIS pediatric short forms (version 2) and also allow for computerized adaptive testing (CAT) for some of the translated dimensions (e.g., fatigue). Both the item banks and short forms are available at https://www.healthmeasures.net/search-view-measures. They are ready to be used in clinical studies and for pediatric public health research offering clinicians and researchers nationally and internationally cross-culturally comparable standardized assessments, which are of high psychometric quality.