Development of an explicit tool assessing potentially inappropriate medication use in Hong Kong elder patients

The first step in developing the Hong Kong-specific PIM criteria was to establish a preliminary drug list based on existing criteria in the literature. A literature search was conducted to identify explicit tools assessing PIM use on the PubMed database from January 1991 to April 2019 internationally and consequently, 40 sets of validated explicit criteria were identified. Explicit criteria that 1) applied to the general population aged 65 years and older, 2) used the literature review on safety and efficacy of drug use as an evidence base, and 3) described the methods of development and validation for the criteria, were taken into account. Regarding the criteria that were updated occasionally, only the latest versions of PIM lists were considered and the early versions of the criteria were eliminated from the study. Consequently, the following nine sets of explicit criteria were included as reference criteria in the current study: the McLeod criteria [22], the Rancourt criteria [23], the Lindblad criteria [24], the Laroche criteria [17], the Winit-Watjana criteria [25], the Norwegian General Practice [26], the PRISCUS criteria [12], the version 2 of the Screening Tool of Older Person’s Prescription (STOPP) [27], and the 2015 version of the Beers criteria [28].

The structures of the nine sets of reference criteria were heterogeneous, with the classification system of criteria statements as disease-oriented or/and drug-oriented. The evaluated aspects of the reference criteria are summarized in Table 1. The most common evaluated aspects in the nine sets of reference criteria included generally avoided drugs independent of diagnoses (7/9), dosage (6/9), disease-drug interactions (6/9), and drug-drug interactions (7/9). After reviewing the structures of the reference criteria, the evaluation of dosage was found to be rarely classified as a separate category of PIM use in the reference criteria. Furthermore, the lists of drug-drug interactions in the reference criteria were mostly selective and not comprehensive. Therefore, the current study was intended to exclude drug-drug interactions and develop a disease-oriented preliminary PIM list. The structure of the Hong Kong-specific preliminary PIM list was formulated using two common categories of PIMs from the reference criteria, namely, generally avoided drugs independent of diagnoses, and disease-drug interactions.

Based on the structure of the preliminary PIM list, the operational definitions of PIM use in the current study are described as follows: 1) medications or medication classes that should be generally avoided independent of diagnosis or conditions and 2) medications or medication classes that should not be used in older adults known to have specific medical conditions. Other evaluated aspects of PIM use including dosage, duration of therapy, alternative therapies and special considerations of use were supplemented by experts’ comments during the Delphi process. Following the operational definition of PIM use in this study, seven sets of reference criteria included generally avoided drugs independent of diagnoses and six sets of criteria included disease-drug interactions. After extracting all related statements from the reference criteria, 279 medications/classes under generally avoided drugs independent of diagnoses and 115 medications/classes considering specific medical conditions were identified.

To prepare a manageable amount of PIM statements for further evaluations in the Delphi process, any medication or medication class appeared in at least two sets of the reference criteria (if appeared in two sets of criteria, one of which must be the Beers criteria) and its related medical conditions were selected from relevant reference criteria as candidate PIMs for each category of PIM use. We used the Beers criteria to select medications that appeared in two sets of criteria because the Beers criteria were considered as the golden standard for PIM assessment and many country-specific explicit criteria were developed based on the Beers criteria [19]. The availability of candidate PIMs was examined by the Hospital Authority (HA) drug formulary issued in April 2018. The candidates not covered by the HA drug formulary were excluded from the preliminary PIM list. Meanwhile, all the individual medications belonging to each medication class were also identified from the HA drug formulary. The medications/classes were identified by the Anatomical Therapeutic Chemical (ATC) classification system codes to enhance the standardization of PIM candidates. After applying the inclusion and exclusion criteria, the final Hong Kong-specific preliminary PIM list included 168 candidates, 78 of which were developed under PIMs independent of diagnoses, and 90 were developed under PIMs considering specific medical conditions for further questioning in the Delphi process.

The preliminary PIM list was validated by a two-round modified Delphi process. Experts from different specialties were recruited by invitation emails and personal communication, all involved with practice or research in medication issues for elder patients for at least five years. A face-to-face meeting was also convened with six potential experts one by one to brief them on the study background and objectives. After the briefing meeting, five experts decided to participate in the project. Invitation emails were also disseminated to the Hong Kong Geriatric Society (HKGS). Four geriatric physicians from the HKGS replied for a further in-person meeting and three of them finally decided to take part in the study after the meeting. Overall, a total of eight (80%) experts agreed in participating in the project seeking to develop the Hong Kong-specific PIM list. On the basis of the preliminary PIM list, a five-point Likert scale was designed to construct the questionnaire for collecting experts’ ratings and comments during the Delphi process. The questionnaire was examined for wording and content validity by three qualified experts.

The first round of the Delphi process began in August, 2019. The experts were asked to rate the degree of inappropriateness of each candidate statement from a score of 1 (strongly disagree) to 5 (strongly agree). A score of 3 was neutral or undecided. Furthermore, the experts were encouraged to provide comments on any exception or restriction in terms of indication, dose, route of administration, or therapy duration to make each candidate statement more accurate. They were also asked to propose therapeutic alternatives to the PIMs independent of diagnoses and suggest new PIM candidates not included in the questionnaire.

After the first round of questioning, the experts’ answers were presented as median with interquartile ranges (IQR) for each candidate PIM. The statements with medians ≥3.5 and IQR ≤ 1.5 were included in PIM use, whereas the statements with medians < 3.0 were excluded from PIM use [25]. The statements with 3.0 ≤ medians < 3.5 or IQR > 1.5 were regarded as questionable PIMs that required further evaluations in the second round along with the new candidate PIMs suggested by the expert panel [25]. The second round of the Delphi process was conducted in October, 2019. All the experts received the second-round questionnaire together with the panel’s ratings and comments from the first round. The experts’ feedbacks in the second round were evaluated using the same procedure.