Psychiatric rating scales in Urdu: a systematic review

With an estimated population of about 165 million [1] Pakistan is the sixth most populous nation in the world. Although only about 8% of Pakistanis speak Urdu (the national language of Pakistan) as their first language, most people in Pakistan are bilingual speaking their regional language and Urdu almost equally easily [2]. English is spoken mostly by the educated classes and used for official correspondence. With a literacy ratio of 44% and about 50% of the population receiving only primary or below primary education [3] there are many Pakistanis who are unable to read and understand English.

Most psychiatric research involves use of questionnaires of one sort of another. Most of these questionnaires have been developed in the English language and Western culture. There are questions as to how applicable or relevant these questionnaires would be in a primarily non-English speaking Eastern nation like Pakistan. We, therefore, need questionnaires that are in a language that can be understood by majority of the Pakistani people, like Urdu, and are relevant to their culture. In the absence of such questionnaires the two options available are either to create a new questionnaire in Urdu, or to translate and adapt an already established questionnaire from English.

If the latter route of translating and adapting an established questionnaire is taken, which is more often taken by virtue of being a less daunting task than creating a new questionnaire, there are five major domains of cross-cultural validity that need to be considered [4, 5].

1. Content validity. The content of the instrument should be relevant in the culture into which the instrument is being translated.

2. Semantic validity. The words in the original instrument and the translated instrument should have the same meaning.

3. Technical validity. The method of assessment is comparable in each culture e.g. self-rated instruments assume literacy which is not very high in Pakistan.

4. Criterion validity. The interpretation of responses to similar items in source and target languages should remain the same when compared with the norm of each culture studied.

5. Conceptual validity. The instrument is measuring the same theoretical construct within each culture.

Whether indigenously developed or translated, all new questionnaires also need to have their criterion validity established against an existing gold standard in an appropriate group of respondents to be declared clinically useful [6]. In this manner their validity coefficients such as sensitivity, specificity, positive predictive value and negative predictive value can be established to make them comparable with other similar questionnaires. An instrument is valid if it correctly identifies most people with the disorder (high sensitivity) and correctly excludes most people without the disorder (high specificity).

We were able to find only one review "Clinicians' Compendium Of Assessment Tools for Mental Health Clients from Culturally and Linguistically Diverse Backgrounds [7]" done in Australia that has reviewed validation status of questionnaires available in languages other than English. While the Compendium does list a few assessment tools that have been translated into Urdu, Urdu was not a search term in that review and all instruments in Urdu were included in the category of Instruments in Languages Other Than English (LOTE) for which insufficient published information was available and accessible. In this review we have therefore, tried to explore how many questionnaires measuring psychiatric symptoms are available in Urdu, whether indigenous or translated, that have undergone some degree of validation. We have also tried to assess to what extent these questionnaires have undergone either criterion validation (applicable to all questionnaires) or cross-cultural validation (applicable to translated questionnaires only).

Our initial databases search yielded 29 studies. Of these 15 were found to be relevant. Our secondary search yielded 42 more studies. Three of these were found through the experts we had contacted. Of these 17 were found to be relevant. Thus a total of 32 studies, reporting either cross-cultural or clinical validation of 19 psychiatric questionnaires in Urdu were included in the review. Details of validation of translation of Edinburgh Postnatal Depression Scale were found in the abstract of validation study of Harvard Trauma Questionnaire [14].

To our knowledge our study is the first of its kind looking at the validation status of all the psychiatric rating scales available in Urdu. We found 19 rating scales, 6 indigenous and 13 translated from English, which have undergone some degree of validation in Urdu. Among the six indigenous scales, the BSI has been most extensively validated both in urban and rural settings. Among the other indigenous scales AKUADS, PADQ and SSDS were validated in reasonably large samples. ASR-Q did not go through a criterion validation study while the PTSD-Q validation study had a very small sample size.

Among the 13 translated scales only the How I Feel scale, the SDQ and SRQ were evaluated for both cross-cultural and criterion validation, the SRQ being the most extensively evaluated and validated. Rest of the translated scales underwent either only cross-cultural (CIS-R, EAT, GHQ-28, HSCL, HTQ, ITG, WHO-QOL) or criterion (GHQ-12, PHQ) validation. The HADS scale underwent both cross-cultural and criterion validity evaluation but these were two different translations one undergoing the former and the other the latter.

The Australian "Clinicians' Compendium Of Assessment Tools for Mental Health Clients from Culturally and Linguistically Diverse Backgrounds [7]" shows that BDI-II, HADS, EPDS and GHQ have all been translated and undergone some degree of validation in Arabic; EPDS, GHQ and HADS in Italian; BDI, EPDS, GHQ and HADS in Chinese/Cantonese; and BDI-II and HADS in Spanish. As explained above Urdu was not a search term in this review.

On one hand it was rather surprising and encouraging to find 19 questionnaires measuring psychiatric symptoms in Urdu which had undergone some degree of either cross-cultural or criterion validation. On the other hand most of these are screening tools for anxiety, depression or general psychiatric morbidity. The very commonly used research tools like HRSD, MADRS, BDI, PANSS etc, and the definitive diagnostic instruments like SCID have not undergone any sort of validation in Urdu.

Bhui et al. [44] have suggested that even within a broad ethnic group expressions of distress may vary between different sub-groups and may change as a result of acculturation. The GHQ-12 performed better than the ADI (Amritsar Depression Inventory, developed in the Punjab in India) in detecting depression even in the Punjabi population settled in UK. This suggests that even instruments developed in one language may not be equally valid for all sub-groups speaking that language depending on the culture they are living in. In that sense language and culture are not one and the same where validation of instruments is concerned.

So what does one do when one wants to do research in a language other than English and there is no fit-for-purpose tool that has been validated in that language? In their review of cross-cultural adaptation of health-related quality of life measures Guillemin et al. [10] have stated that there are two possible options. The first is to develop a new measure using culturally defined, within-group variables that have been developed and described in terms of the language and customs of a particular culture at a particular time, called the Emic approach [4]. The second approach is to use a measure from another language and culture applying the concepts of behaviour and techniques of measuring that behaviour from the so-called source culture to the target culture, called the Etic approach.

The problem with an exclusively emic approach is that it does not allow quantitative comparison across times and between cultures. The problem with an exclusively etic approach is that manifestations and expressions of a universal phenomenon, for example depression, may be different in different cultures, and thus may be missed if concepts and measures from one culture are applied blindly to another culture [4]. The first is time, labour and expertise intensive because of the need to conceptualise a new measure and select its items, while the second is fraught with the difficulties of the relevance and validity of a measure developed in one language and culture being used in another language and culture.

In Urdu it seems like both approaches have been used, with most scales being translated from English and a few being developed indigenously from complaints of Pakistani patients later diagnosed as suffering from Depression and Anxiety. However, since even the latter were validated against etic constructs like ICD and DSM diagnoses it is difficult to say if there are any purely emic instruments in Urdu. This raises the question whether there should be a different set of criteria for diagnosing depression in Pakistan if people suffering from depression in Pakistan present with different expressions of distress compared to patients in the West? If the diagnostic criteria are different should we call this syndrome something other than depression? Questions like these would only be answered after a lot more cultre-centred research than has been carried out as yet.

Nineteen questionnaires measuring psychiatric symptoms have so far been evaluated for cross-cultural and/or criterion validity in Urdu. Six of these have been developed indigenously while thirteen have been translated from English. The BSI, SRQ and AKUADS are the questionnaires that have been most thoroughly evaluated in Urdu.