Adapting and Translating Instruments

Overview

Translation changes words from one language to another. Cross-cultural adaptation seeks to preserve the construct, item intent, response task and score interpretation in a new linguistic and cultural context. A grammatically correct translation can still fail when examples, healthcare pathways, idioms, time frames or social assumptions do not apply [1–4].

A defensible process commonly includes permission from the rights holder, independent forward translations, reconciliation, back translation or another quality-control comparison, multidisciplinary review, cognitive debriefing with target respondents, proofreading, pilot implementation and documentation. The exact workflow can vary, but no single step—especially back translation—guarantees equivalence.

Adaptation produces a new language-context version that requires evidence. Original measurement properties do not transfer automatically. Content relevance and comprehensibility should be reconsidered, and structural or cross-cultural validity, reliability and other properties may require evaluation according to the intended use.

Why this matters

Health concepts are expressed through culture, healthcare systems and everyday language. Literal translation can introduce unfamiliar terms, alter politeness, change the time frame or make response categories unnatural. An item about “calling your primary care provider” may not match a system where appointments are arranged through a clinic counter or mobile application. A question about “out-of-pocket insurance costs” may measure a different experience in another financing system.

Poor adaptation threatens validity and fairness. Respondents may appear to have different levels of a construct because the items function differently, not because the underlying experience differs. Transparent adaptation helps preserve comparability while respecting local meaning.

Equivalence is multidimensional

Semantic equivalence

Words and sentences should convey comparable meaning. One source term may have several translations with different levels of formality or clinical connotation. The most literal option is not always the most comprehensible.

Idiomatic equivalence

Idioms rarely translate word for word. Phrases such as “felt under the weather” or “jumped through hoops” require a conceptually equivalent expression or direct wording.

Experiential equivalence

The situation described should exist in the target context. An item about driving to appointments may exclude participants who depend on public transport or family. Adaptation may require a broader phrase such as “travelling to the facility”.

Conceptual equivalence

The construct itself should have comparable meaning. “Self-management”, “family support” or “privacy” may be understood differently across settings. Qualitative work can determine whether the domain and item boundaries remain appropriate.

Operational equivalence

The administration mode, instructions and response task should function similarly. A visual analogue scale used on paper may not transfer directly to a telephone interview. Changes in mode can require separate evaluation.

Permission and preparation

Before translation, researchers should:

• identify the rights holder and current authorised version;
• obtain permission for translation, adaptation and intended administration modes;
• clarify ownership and sharing of the translated version;
• obtain the scoring manual and item definitions;
• review existing translations and harmonisation requirements; and
• define the target population, language variety, literacy range and context.

Some developers require a specific linguistic-validation workflow or review. Even when no fee is charged, permission and attribution requirements may apply.

Forward translation

Forward translation is usually conducted by at least two translators working independently before reconciliation [1,2]. Useful diversity includes:

• a translator familiar with the construct and clinical terminology; and
• a translator closer to natural everyday language and less influenced by the source wording.

Translators should receive item intent and construct definitions, not only isolated sentences. They should document alternatives, uncertainty and culturally problematic concepts.

The purpose of multiple translations is not majority voting. Differences reveal ambiguous source wording and possible interpretations.

Reconciliation

A reconciliation meeting develops one provisional target version. Decisions should consider conceptual accuracy, comprehensibility, natural language, response consistency and reading level. The team should avoid creating a hybrid sentence that is grammatically awkward merely because it combines parts of both translations.

Illustrative reconciliation log

illustrative

Table 1. Documenting adaptation decisions

Source concept	Initial alternatives	Decision	Rationale
primary care provider	doktor penjagaan primer / klinik biasa / petugas kesihatan utama	klinik atau petugas kesihatan yang biasanya mengurus rawatan anda	covers different service pathways without limiting the respondent to a doctor
co-pay	bayaran bersama / bayaran sendiri	kos yang perlu anda bayar sendiri	conceptually understandable across financing arrangements
patient portal	portal pesakit / aplikasi temu janji	sistem atau aplikasi untuk membuat temu janji	retains function rather than one technology

Note: The Malay phrases are illustrative and require expert and participant review before use.

Back translation and quality comparison

Back translation converts the reconciled target version into the source language, ideally by someone blinded to the original wording. It can reveal omissions, additions and shifts in meaning [1,2]. It should not be used as a mechanical test that rewards wording closest to the original.

A natural target-language item may back-translate differently while preserving the concept. Conversely, a literal target version can back-translate perfectly yet remain incomprehensible to respondents. Back translation is one quality-control input, not proof of validity.

Expert review and harmonisation

A multidisciplinary committee may include translators, clinicians, methodologists, language experts, instrument developers and representatives of the target population. The group reviews:

• equivalence of construct and item intent;
• clinical and factual accuracy;
• cultural relevance and sensitivity;
• response options and score direction;
• reading level and administration mode; and
• consistency with other language versions.

Harmonisation compares translations across languages or settings to avoid unnecessary divergence. It should not force identical solutions when cultural contexts genuinely require different expressions.

APPRAISAL CHECKLIST

Review an adapted item

• Is the intended concept preserved?
• Is the wording natural and understandable?
• Does the experience exist in the target setting?
• Are the recall period and response options interpreted consistently?
• Does the item avoid stigma, ambiguity and unnecessary complexity?
• Has any change to the construct or scope been documented?

Cognitive debriefing

Cognitive debriefing tests the provisional version with members of the target population. Participants answer items and explain how they understood the question, recalled information, formed a judgement and selected a response. Sampling should include variation relevant to comprehension, such as education, age, service pathway and language use.

The aim is not to obtain a favourable satisfaction rating. It is to identify misunderstandings, missing response options, unfamiliar terms and unintended emotional or cultural effects. Several iterative rounds may be more useful than one large final pilot.

Response options and language direction

Response categories require adaptation as carefully as item stems. Agreement or frequency terms may not have evenly perceived order. Negative wording can be especially difficult when grammar and response direction interact. Electronic forms should be tested for font, character display, line breaks and right-to-left direction where applicable.

Bilingual participants may switch languages within an interview. The protocol should specify whether respondents choose one version, whether interviewers may clarify, and how mixed-language administration is documented. Ad hoc interviewer translation creates uncontrolled versions.

Adaptation across literacy levels and administration modes

A translated item may be understandable when self-administered by highly literate respondents but difficult in interviewer administration or among people unfamiliar with rating scales. Adaptation should therefore test the intended mode. Interviewers need standard prompts that clarify procedure without reinterpreting the item.

Audio, telephone and electronic versions change what respondents can see and review. A grid of repeated response options may work on paper but encourage omission on a small screen. Telephone administration removes visual anchors. If modes will be combined, test whether instructions, ordering and response mapping remain comparable.

Reading level is not solved by replacing every medical term with a simpler word. Some concepts require a short explanation, while oversimplification can change meaning. Use plain language, short sentences and familiar examples, then confirm comprehension directly with target respondents.

When the target population uses more than one language in daily life, researchers should decide whether the instrument measures the construct independently of language proficiency. Participants may understand a clinical term in English but describe the experience in Malay or another language. The protocol can allow a chosen authorised version while recording language and mode; ad hoc switching within items should be avoided unless studied explicitly.

Adaptation can reach a point where the construct changes. Replacing system-specific items, splitting items or adding domains may be necessary for relevance, but the result may no longer be a translated version of the original scale. COSMIN guidance requires renewed attention to content and other measurement properties when versions differ materially [5,6]. At that point, the team should describe the work as adaptation or new development rather than implying direct equivalence.

Field testing the adapted version

After qualitative finalisation, field testing evaluates whether the intended score structure and measurement properties are supported. The sample should include relevant language, education and care-pathway diversity and should use the intended administration mode.

Analyse item non-response, category use, floor and ceiling effects, structural validity, reliability and prespecified construct relationships. When source-language data are available from comparable participants, examine cross-cultural validity rather than assuming equivalence from similar averages.

Unexpected differences require investigation. They may reflect translation, different service experience, sampling or genuine construct variation. Statistical adjustment should not be used to erase a translation problem. Revisit cognitive and expert evidence before deciding whether an item should be revised, interpreted differently or excluded from cross-language comparisons.

Finalisation, proofreading and version control

After cognitive debriefing, the team revises and proofreads the full instrument, instructions, response options and scoring information. Developers may review the final version. The project should retain:

• source and target files;
• translator qualifications and declarations;
• reconciliation and review logs;
• back translations;
• cognitive-interview findings;
• final decisions and dates;
• version number; and
• permission correspondence.

PROTOCOL CHECKLIST

Document the adaptation workflow

• rights holder and permission terms;
• source version and target language variety;
• construct and item-intent materials;
• forward translators and independent drafts;
• reconciliation procedure;
• back translation or alternative quality comparison;
• committee composition and decisions;
• cognitive-debriefing sampling and results;
• proofreading, developer review and finalisation; and
• version control and planned measurement-property evaluation.

Measurement evaluation after adaptation

Adaptation should be followed by evidence proportionate to the intended use. Content validity and comprehensibility are central. Further evaluation may include:

• structural validity of the target-language scores;
• internal consistency for supported domains;
• test-retest reliability;
• construct-validity hypotheses;
• cross-cultural validity or measurement invariance;
• measurement error and interpretability; and
• responsiveness when change will be measured.

A small pretest can identify obvious problems but cannot establish all measurement properties. Direct comparison of language groups is interpretable only when the construct and sampling contexts are comparable.

DECISION GUIDE

Decide whether an item can be adapted

1. Is the source concept relevant in the target setting?
2. Can wording preserve the same item intent?
3. Would a broader or different expression change the construct?
4. Can the developer approve a modification?
5. Should the item be retained, adapted, replaced or the instrument rejected?
6. What new evidence is required after the change?

Cross-cultural validity and differential item functioningMMed extension

Even after careful qualitative adaptation, an item may relate differently to the underlying construct across language groups. Multi-group confirmatory factor analysis and differential item functioning methods can evaluate whether item thresholds, loadings or response probabilities differ beyond expected construct differences. Statistical findings should be interpreted with qualitative and contextual evidence rather than used as automatic deletion rules.

Practical interpretation note

An adapted version should be named, dated and cited precisely so later studies can identify which wording and scoring system were used.

Interface with statistical analysis

The target-language version should retain the authorised scoring direction and domain structure unless evidence and permission support modification. Researchers should not pool scores across languages merely because totals have the same numerical range. Cross-language comparisons may require measurement-invariance evaluation.

Translation-stage samples are selected for information about comprehension, not for estimating population parameters. Larger field-test samples address structural and other measurement properties. The methodology and analysis purposes should remain distinct.

Worked example

WORKED EXAMPLE

Adapting an appointment-access domain

Source item: “It was difficult to book an appointment through my provider’s patient portal.”

Problem: The target population uses several pathways, including mobile applications, telephone calls, clinic counters and family assistance. A literal portal reference would be irrelevant to many participants.

Provisional adaptation: “It was difficult to use the usual system for arranging my appointment, such as an application, telephone call or clinic counter.”

Cognitive finding: Participants focused on using technology rather than whether appointment slots were available.

Final approach: Split the content into two items: difficulty contacting or using the appointment system, and difficulty obtaining a suitable appointment time. The change is discussed with the developer, documented as a substantive adaptation and scheduled for new content and structural evaluation.

Suggested protocol or reporting wording

Common mistakes

Knowledge check

Why can a grammatically accurate translation still be invalid?

Answer and reasoning

The wording may describe an irrelevant experience, alter the construct, use unfamiliar concepts or lead respondents through a different response process. Linguistic accuracy alone does not establish conceptual or cultural equivalence.

What is the main purpose of back translation?

Answer and reasoning

It is a quality-comparison step that can reveal omissions or shifts in meaning. It does not prove comprehensibility or validity and should be interpreted with item intent, expert review and target-respondent evidence.

Why are cognitive interviews needed after expert review?

Answer and reasoning

Experts cannot fully predict how target respondents understand, recall, judge and map answers to response categories. Cognitive interviews provide direct evidence about actual response processes and culturally problematic wording.

Can scores from the source and translated versions be compared automatically?

Answer and reasoning

No. Equivalent numerical ranges do not establish equivalent item functioning or score meaning. Cross-cultural validity or measurement-invariance evidence may be needed, especially for direct group comparisons.

Key take-home messages

SUMMARY

Key take-home messages

• Cultural adaptation preserves construct and response meaning, not identical wording.
• Permission, item-intent information and version control should precede translation.
• Multiple translation and review steps reveal different kinds of problems; no single step proves equivalence.
• Cognitive debriefing with target respondents is essential for comprehensibility.
• Adapted instruments require proportionate evaluation of measurement properties before substantive use.

References

1. Beaton DE, Bombardier C, Guillemin F, Ferraz MB. Guidelines for the process of cross-cultural adaptation of self-report measures. Spine. 2000;25(24):3186–3191. https://doi.org/10.1097/00007632-200012150-00014
2. Wild D, Grove A, Martin M, Eremenco S, McElroy S, Verjee-Lorenz A, et al. Principles of good practice for the translation and cultural adaptation process for patient-reported outcomes measures: report of the ISPOR Task Force for Translation and Cultural Adaptation. Value Health. 2005;8(2):94–104. https://doi.org/10.1111/j.1524-4733.2005.04054.x
3. Sousa VD, Rojjanasrirat W. Translation, adaptation and validation of instruments or scales for use in cross-cultural health care research: a clear and user-friendly guideline. J Eval Clin Pract. 2011;17(2):268–274. https://doi.org/10.1111/j.1365-2753.2010.01434.x
4. McKown S, Acquadro C, Anfray C, Arnold B, Eremenco S, Giroudet C, et al. Good practices for the translation, cultural adaptation, and linguistic validation of clinician-reported outcome, observer-reported outcome, and performance outcome measures. J Patient Rep Outcomes. 2020;4:89. https://doi.org/10.1186/s41687-020-00248-z
5. Mokkink LB, Prinsen CAC, Patrick DL, Alonso J, Bouter LM, de Vet HCW, et al. COSMIN manual for systematic reviews of PROMs. Version 2.0. Amsterdam: COSMIN; 2024. https://www.cosmin.nl/wp-content/uploads/COSMIN-manual-V2_final.pdf
6. Terwee CB, Prinsen CAC, Chiarotto A, Westerman MJ, Patrick DL, Alonso J, et al. COSMIN methodology for evaluating the content validity of patient-reported outcome measures: a Delphi study. Qual Life Res. 2018;27(5):1159–1170. https://doi.org/10.1007/s11136-018-1829-0