The Translation Process of Pharmaceutical Care Network Europe v9.00 to Bahasa Indonesia: An Instrument to Detect Drug-Related Problem

Background Drug-related problems (DRPs) remain a major health challenge in tertiary health services such as hospitals in Indonesia. These problems are detected and solved using classification systems such as Pharmaceutical Care Network Europe (PCNE). Therefore, this study aims to obtain a valid and reliable Bahasa Indonesia version of the PCNE. Methods A draft of the Bahasa Indonesia version of the PCNE v9.00 was discussed by four experts from May to August 2020 using the Delphi method. Furthermore, the instrument was assessed for its readability, clarity and comprehensiveness by 46 hospital pharmacists throughout Indonesia. In October 2020, two pharmacists from Haji General Hospital, Makassar, Indonesia carried out the inter-rater agreement to assess 20 cases where the proportion of coding matches between both raters were observed. Results The instrument was found to be valid after passing the face and content validity, and the Scale Content Validity Index (S-CVI) value for each PCNE domain was 0.91, 0.89, 0.93, 0.97 and 0.93, respectively. Moreover, there was a fair agreement between the two raters that ranged between 40%–90%. Also, kappa statistics showed a substantial agreement on the ‘Problems’ and ‘Causes’ domains. Conclusion The Bahasa Indonesia version of the PCNE v9.00 instrument passed face and content validity as well as inter-agreement to be used in hospital settings.


Introduction
Drug-related problems (DRP) are events or circumstances involving drug therapy that occur or potentially interfere with the achievement of desired health outcomes (1). Some of the factors that contribute to the emergence of DRP in patients include inappropriate prescription, ineffective treatment, underdose, noncompliance, etc (2). In Indonesia, DRP occurred in several chronic diseases such as diabetes (3), kidney (4) and heart failure (5). Therefore, there is an urgent need for understanding the importance of the pharmacist's role in identifying, solving and reducing the incidence of DRP in patients (5).
Moreover, the documentation and classification of DRP can help pharmacists to identify and resolve DRP in a patient. Several classification systems, such as APS-Doc, Cipolle,

Forward Translation
The study began with a forward translation of PCNE draft version 9.00 (English version) into Bahasa Indonesia by two independent sworn translators and the results were separately discussed. Furthermore, the translations were combined by paying attention to the excellent choice of words and pharmaceutical terms.

Backward Translation
The instrument was again re-translated from Bahasa Indonesia to English by other translators. Similarly, the results were separately discussed.

Face Validity
The combined draft was then given to four experts, including hospital pharmacists and academicians with master and doctoral qualifications. This assisted in the critical review of the translation results and suggestions for improvements to make the instrument easier to use. The process to reach consensus among experts was carried out using the Delphi method (15), which ensures that the expert panels do not know each other and report only to the researcher.
A draft of the Bahasa Indonesia version of the PCNE was sent to the expert panels in parallel and they were given time to provide a critical review of the translation. All the critical reviews from each expert panel were then combined and the instrument was refined. The draft was then returned to each expert panel and the process was repeated until they reached a consensus.

Content Validity
This process involved a minimum of 20 clinical pharmacists who work in the hospital as respondents. They were asked to rate the criteria of readability, clarity and comprehensiveness of the instrument using a 5-point Likert scale. Furthermore, the Item Content Validity Index (I-CVI) and the Scale Content Validity Index (S-CVI) were calculated. The I-CVI compares the number of respondents that gave ratings of 3 and above with the total number of respondents. In contrast, the S-CVI is the average of the I-CVI values (16). The following formulae were used to calculate the I-CVI and S-CVI: DOCUMENT, Consensus of Granada, Strand Classification and the Pharmaceutical Care Network Europe (PCNE) system, are applied (6-7). A DRP classification system needs to have an open hierarchical structure with clear definitions for each category described in the instrument to reduce ambiguity or multiple interpretations when carrying out the coding process (8). Furthermore, the ease of use is also a specific requirement of the DRP classification system, which must be acceptable. Therefore, the DRP classification system needs to be validated before it is widely used (8)(9).
Besides having a good validity, an instrument also needs to have a good level of inter-rater reliability, which is a measure of the degree of agreement between two or more raters (10). This is required to determine the extent to which the raters consistently assign a precise value to each rated item (11). It is essential because the raters need to give the same value in the same conditions and cases.
In this study, the PCNE instrument was selected as the starting point because it is structured, detailed and also identifies the patient's DRP status based on their problems, causes and interventions. Furthermore, the instrument has several translations including Spanish, Turkish, Croatian, French (12), Slovenia (7) and German (13). This study, therefore, aims to obtain, validate and determine the inter-rater agreement (percentage agreement and kappa statistics) of Bahasa Indonesia version of the PCNE version 9.00.

PCNE Instrument Usage Permit
There was difficulty in downloading the PCNE instrument from their official website; therefore, permission to translate the instrument into Bahasa Indonesia was requested from the PCNE organisation in the Netherlands. The instrument was then downloaded via the PCNE website. Version 9.00 of the PCNE instrument served as the starting point for the translation and it consisted of five main domains (problem, causes, planned intervention, intervention acceptance, and status of DRP), 24 primary domains and 84 secondary domains (14). of the number of cases in which both raters gave the same code to the total number of cases. The formula below is used to calculate percentage agreement (18) The kappa statistics were carried out using IBM SPSS ® version 24 software and its interpretation is showed in Table 1 (18).

Results
The PCNE classification version 9.00 was translated into Bahasa Indonesia by two sworn translators that did not meet. Furthermore, a reconciliation process was conducted with each translator regarding the translation results, which were then combined. The draft of the translated instrument as shown in Appendix 1 were submitted to the experts for a critical review. After two sessions of discussion with the expert panels, the following changes were incorporated: A total of 46 hospital pharmacists ( Table 2)  The cutoff value for I-CVI and S-CVI are set based on Shrotryia and Dhanda, which is ≥ 0.78 for I-CVI and ≥ 0.8 for S-CVI (17).

Inter-Rater Agreement
The inter-rater agreement involved two pharmacists from the Haji Regional General Hospital, Makassar, who conducted a DRP assessment on 20 selected patient cases using validated instruments. These cases were taken from patient medical records using consecutive sampling methods, which met the following eligibility criteria: i) Has undergone inpatient care at the hospital (recovered or died) ii) Patients aged ≥ 18 years old iii) Medical records are well documented Before the test, the two raters were trained separately using five practice cases to familiarise them with the instrument. They were asked to provide a code in the 'Problem' and 'Cause' domains following the given patient's DRP case using the Bahasa Indonesia version of PCNE. The coding consistency and chance agreement between the two raters were determined by calculating percentage agreement and kappa statistics. The percentage agreement is the ratio However, they are not familiar with PCNE instruments, therefore, they were trained using five practice cases.
After assessing the DRP cases of 20 patients, the percentage agreement was 90% higher in the 'problems' domain for both the primary and secondary domains, respectively. While in the 'causes' domain, it was much lower by 60% and 40% on the primary and secondary domains, respectively. Furthermore, kappa statistics were performed to calculate the chance agreement of two raters when identifying the DRP on a case using PCNE. The result showed a significant agreement between the two raters on

Discussion
This study is the first to translate, validate and determine the inter-rater agreement of the translated PCNE into Bahasa Indonesia. After the forward (English-Bahasa Indonesia) and backward (Bahasa Indonesia-English) translations, some differences were noticed. These include the changes in word structure, especially in the 'intervention acceptance' domain, which is different from the original version because that of Bahasa Indonesia uses conjunctions to make the domain easier for users to understand. According to the suggestions from experts and respondents during the validation process, the changes in the number of word 5-point Likert scale, followed by 5, and a few filled 1 and 2. Moreover, the respondents assessed the instrument using four aspects including readability, clarity, ambiguity and comprehensiveness of the instrument. Also, the I-CVI and S-CVI values of each domain's instrument ranged between 0.85-0.98 and 0.89-0.97, respectively ( Table 3). The final version of PCNE after conducting the face and content validity is the Indonesia version shown in Appendix 2.
In addition, the inter-rater agreement involved two pharmacists that worked at Haji General Regional Hospital, Makassar, as raters. The first had 15 years of working experience in the hospital, while the second had 8 years. The kappa statistics showed a high degree of agreement on both 'problem' and 'causes' domains. The value was higher compared to others such as DOCUMENT (0.53 versus 0.615) (19) and GSASA V2 (0.52 versus 0.615) (20), lower than APS-Doc (0.68 versus 0.615) (21) and similar with the classification developed by the Pharmaceutical Society of Singapore (ILTC DRP Classification System) (0.614 versus 0.615) (8). Furthermore, it is believed that the low value of kappa in this study is because the raters are not too familiar with the instrument. Therefore, using the instrument frequently may increase the value of kappa. This is influenced by the raters' level of knowledge and experience.
This study has certain limitations. First, the instrument does not assess the inter-rater agreement on the 'planned intervention,' 'intervention acceptance,' and 'status of DRP' domains because only secondary data were used. Second, the inter-rater agreement is still limited to only two assessors due to the unfamiliarity of this instrument in daily pharmacy practices in Indonesian hospitals. Furthermore, construct validity, such as convergence to see the instrument's reliability under different conditions (22), was not performed. Therefore, further studies are suggested to focus mainly on reliability testing by involving more pharmacists and performing the construct validity. There is also a high consistency in the 'problems' domain of the instrument on an interrater agreement study. However, the 'causes' domain has low consistency, which differs from the results of Koubaity et al. in 2019 and Schindler et al. in 2020 (12-13). Furthermore, these two studies yielded a percentage agreement between 59%-100% and 57.4%-77.3%, respectively. Several factors resulted in the low consistency between the two raters of the 'causes' domain. First, this study used a small sample of pharmacists compared to Schindler et al. (13) which considered a total of 32 pharmacists. Second, the variety of codes and the ability to code the case summaries led to different perspectives between the two raters, causing the domain to have low consistency (12). Finally, the raters admitted that it was quite challenging to choose the correct code for a patient's case,

Conclusion
The PCNE v9.00, Bahasa Indonesia version has passed content validity and interagreement for use in pharmacy practice in both hospitals and academic settings. Further study is suggested to focus mainly on interrater reliability tests using more pharmacists to measure the validity of the instruments in various conditions in hospital settings.

Acknowledgements
Researchers would like to thank the experts, respondents, and pharmacists in Haji General Regional Hospital, Makassar, who already participate and pour out their thoughts, especially on the instrument's validation process and inter-rater agreement. Any grant sources did not fund this research.

Ethics of Study
This study was ethically approved by the Health Research Ethics Committee of the Faculty of Medicine, University of Indonesia [KET-516/ UN2.F1/ETIK/PPM.00.02.2020] and it was used to collect data for the inter-rater agreement study. To protect the patient's data and privacy, anonymity was maintained while collecting data and presenting them to raters during the interrater agreement study.

Conflict of Interest
None.

Funds
None.