Patient-Reported Questionnaires to Identify Adverse Drug Reactions: A Systematic Review

Background: This systematic review aims to summarise available patient-reported questionnaires to detect adverse drug reactions (ADRs) that can be utilised by healthcare professionals in clinical practice and to summarise the psychometric properties (validity, reliability, and responsiveness) of the questionnaires. Methods: A systematic literature search was conducted using Medline, Pubmed, Embase, and Emcare databases to screen for articles published between January 2000 and July 2020. Data items regarding validity, reliability, and responsiveness were extracted independently by two authors. The methodological quality was assessed using the COSMIN (Consensus-Based Standards for the Selection of Health Measurement Instruments) checklist. Results: A total of 1563 unique article titles were identified after removing duplicates. Following shortlisting of relevant articles, 19 patient-reported ADR questionnaires were identified. Questionnaires most commonly focused on mental health medications (42.1%, n = 8), followed by general questionnaires applicable to any medication (21.1%, n = 4). Many questionnaires did not report assessing the validity and reliability of the measurement tool. For example, only 11 questionnaires (58%) mentioned assessing content validity, in addition to criterion or construct testing. Conclusion: This systematic review summarised the available patient-reported questionnaires that can be used in research and clinical practice to identify ADRs. Results of this systematic review highlight the need for more robust validity and reliability testing when developing patient-reported ADR questionnaires.


Introduction
It has been estimated that the global costs of medication-related harms exceed 40 billion USD annually [1], with half of the harm considered preventable [2]. Adverse drug reactions (ADRs) are unintended and unexpected effects caused by administration of a medication [3]. A 2021 systematic review which included 33 studies estimated that between 8% and 20% persons receiving care in the primary care setting experience ADRs [4]. In Australia, approximately one in five people are likely to be suffering from an ADR at the time they receive a home medicine review [2], which is a government-funded service in Australia where a pharmacist visits the patient in the home to undertake a thorough review of all medications [5].
At the core of preventing or reducing the burden of ADRs is good communication between patients and healthcare professionals regarding any medication-related sideeffects being experienced. However, previous studies have shown that patients do not report the adverse reactions that were potentially due to a medication to their doctors, or that there is a mismatch between information provided by the healthcare professionals and information wanted by the patients [6,7]. From a patient's perspective, the type and severity of ADRs are considered more important than the degree of benefit from the treatment [8]. If the ADRs are not addressed by the healthcare professionals, patients end up lacking knowledge or understanding of the side-effects, do not report the ADRs, and may stop using the medicines [8]. Barriers to patients reporting ADRs to healthcare professionals or regulatory bodies include busy clinical settings, lack of a relationship between patients and practitioners, and issues with pharmacovigilance systems such as lengthy and complicated reporting forms [9][10][11][12]. When ADRs are not detected and not managed properly, patients are likely to stop taking the medications altogether [13].
Patient-reported side-effect questionnaires can be useful to identify ADRs and can contribute to an increased reporting of A26DRs [7,14]. A number of patient-reported sideeffect questionnaires such as the Maudsley Side-Effects (MSE) measure [15] and Patient-Reported Adverse Drug Event Questionnaire (PROMISE) [16] have been developed for patients to report side-effects of medications to researchers and healthcare professionals, but the questionnaires are not routinely used in clinical practice. A summary of characteristics and the validity and reliability of existing patient-reported questionnaires will be useful for healthcare professionals to determine the most suitable questionnaire to identify medication harms in practice. Thus, the aims of this systematic review were (i) to identify published patient-reported side-effect questionnaires that can be utilised by healthcare professionals in clinical practice, and (ii) to summarise the psychometric properties (validity, reliability, and responsiveness) of the questionnaires.

Materials and Methods
The reporting of this systematic review conforms to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement [17,18]. The PROSPERO International Prospective Register of Systematic Reviews registration number for this systematic review is CRD42020198412.

Search Strategy and Study Selection
The Medline, Pubmed, Embase, and Emcare databases were searched for articles published between January 2000 and July 2020, in humans and in the English language. Databases and search terms were selected by the study investigators, in consultation with an academic librarian specialising in health-related database searches. The search strategy (Supplement File 1) included three main components: "adverse drug reaction", "patientreport", and "questionnaire". Two independent authors (R.L. and N.M.) screened all titles and abstracts using the Covidence software [19]. A third reviewer (L.K.E.) was consulted when there was disagreement about the inclusion of a study. Disagreement was resolved by consensus among the three reviewers. Reference lists of shortlisted studies were screened to identify other relevant articles. If articles referred to existing questionnaires but were not the original article describing the development of the questionnaire, the original article describing the questionnaire was retrieved and included in the review.
Studies were included if they developed general questionnaires which were applicable to medications used for any medical condition or questionnaires targeting medications for common conditions, as guided by a list of common chronic conditions which include cardiovascular disease, chronic obstructive pulmonary disease, asthma, diabetes, arthritis, back pain, cancer, chronic kidney disease, mental health conditions, and osteoporosis [20,21]. Studies focusing on medications used in rare medical conditions were excluded. Studies were also excluded if they translated existing questionnaires in English to other languages; the study describing the original (English) version of the questionnaire was included in-stead. Questionnaires were excluded if they were for proxy reporting (i.e., not self-reported). Another exclusion criterion was questionnaires focusing only on specific side-effects of medications, such as questionnaires developed to assess peripheral neuropathy from chemotherapy or to assess extrapyramidal side-effects of antipsychotic medications.

Data Extraction
Data extraction items were discussed and agreed upon by two authors (R.L. and N.M.). Data items regarding questionnaire characteristics included (i) the questionnaire name, (ii) author, year, and participant demographics for the primary study developing the questionnaire, (iii) whether questionnaires were general or focused on specific medication classes, (iv) number of items and domains, (v) use of a scoring system, (vi) whether patients can nominate bothersome side-effects, (vii) presence of open-ended questions, and (viii) whether patients were asked if they thought their symptoms were likely medication-related.
Data items regarding validity, reliability, and responsiveness were extracted independently by two authors (R.L. and N.M.) [14,22]. Data items for acceptable development and validity included (i) whether a literature review was undertaken during questionnaire development, (ii) whether a Delphi process or an expert panel was used, (iii) whether the development included patient input, and (iv) criterion validity (comparative testing of a new questionnaire against an established 'gold-standard' questionnaire) or construct validity (testing whether the questionnaire assesses the skills and abilities it intends to) [23]. Reliability was assessed as internal consistency (whether the different items in the questionnaire measure the overall general purpose) and test-retest reliability (ability of the questionnaire to produce consistent and very similar results when applied to the same person repeatedly) [23,24]. Responsiveness was measured as the ability of the questionnaire to detect change over time [23,25].

Assessment of Methodological Quality
The methodological quality of all included studies was assessed by one author (R.L.) using the COSMIN (Consensus-Based Standards for the Selection of Health Measurement Instruments) risk of bias checklist (Table S1) [26]. The COSMIN checklist consists of 10 measure properties: PROM development, content validity, structural validity, internal consistency, cross-cultural validity, reliability, measurement error, criterion validity, hypothesis testing for construct validity, and responsiveness. Each measurement property was assessed using a four-point scale: "very good", "adequate", "doubtful", and "inadequate" [26]. The overall rating of the quality was determined using the lowest rating under each measure property [26].

General vs. Focusing on Specific Medications or Disease/Health Conditions Name of Questionnaire
Mental health medications, n = 8 [15,27,31,35,36,39,41,42] My Medicines and Me (M3Q) Approaches to Schizophrenia Communication Self-Report Checklist (ASC-SR) Subjects' Response to Antipsychotics (SRA)  Antiepileptics, n = 2 [30,40] Side-Effect Checklist Assessment of Side-Effects in AED treatment (SIDAED) Asthma and chronic obstructive pulmonary disease, n = 2 [29,33] Satisfaction with Asthma Treatment Questionnaire Inhaled Corticosteroids Questionnaire (ICQ) Diabetes, n = 1 [34] Patient's Qualitative Assessment of Treatment (PQAT) Chemotherapy, n = 1 [37] Common Terminology Criteria for Adverse Events Side-Effects Questionnaire Triptans, n = 1 [32] Triptans Questionnaire   Only nine questionnaires (47%) used a scoring system, with six out of the nine questionnaires having a scoring system with a minimum and a maximum score. Out of these questionnaires mentioning the range of potential scores, the GASS questionnaire was the only tool to provide explicit advice regarding interpretation of scores.
Out of the 19 questionnaires, 11 (58%) included open-ended questions allowing patients to mention any other potential side-effects not explicitly covered by the questionnaire. Six (32%) questionnaires asked patients whether potential side-effects or symptoms were likely medication-related. Six (32%) questionnaires allowed the patients to specifically identify which side-effects were bothersome to them.
Questionnaires differed in their level of detail regarding side effects. For example, the TSQM has a list of 14 questions with five questions relating to side-effects, such as "As a result of taking this medication, do you currently experience any side-effects at all?" and "To what extent do the side-effects interfere with your physical health and ability to function (i.e., strength, energy levels, etc.)?". Questionnaires such as the SRA had an extensive list of side-effects for patients to indicate whether they have experienced the side-effects or not. Table 3 shows the results of the validity, reliability, and responsiveness of each questionnaire. With regard to development, 12 (63%) questionnaires explicitly mentioned undertaking a literature review when developing the tool. Eleven (58%) studies mentioned including expert input for the content validity, while 15 (79%) questionnaires involved patients in their development. Criterion or construct testing was undertaken for 11 (58%) of the questionnaires. Out of these questionnaires with criterion or construct testing, four (36%) questionnaires undertook criterion testing by comparing the new questionnaire to a previously published questionnaire, and seven (64%) undertook construct validity testing by employing methods such as exploratory factor analysis using eigen values and accounting for the percentage of total variance. With regard to reliability testing, eight (42%) questionnaires were tested for internal consistency, measured using Cronbach's alpha. Another aspect of reliability testing is test-retest reliability, which was tested in seven (37%) questionnaires using intraclass correlation coefficients, Cohen's kappa statistic, and Pearson's correlation testing. With regard to testing the responsiveness of questionnaires, three (16%) questionnaires assessed the ability of the questionnaires to measure change over time.

Discussion
When developing questionnaires to identify medication-related side-effects, it is important to consider whether the questionnaire will be general (i.e., focusing on all medications) or specific (i.e., focusing on specific medication classes), involving both patients and health professionals in their development, the presence of scoring systems to quantify the burden of ADRs, and use of open-ended questions to elicit additional information [14,26]. Previous literature has also outlined the importance of considering the validity, reliability, and responsiveness of patient-reported questionnaires when developing and using these tools in clinical practice [14,22]. In this systematic review, we found that different types of patient self-reported questionnaires to identify medication sideeffects have been developed, with questionnaires most commonly focusing on identifying side-effects due to mental health medications. There were fewer medication side-effect questionnaires applicable to any medication. The majority of questionnaires (n = 15/19) involved patients in their development.
Questionnaires to identify medication side-effects need to be thorough enough to detect the range of ADRs patients experience, as well as provide the opportunity to identify other ADRs which have been missed [14,43]. The challenge in designing patientreported ADR questionnaires lies in ensuring a comprehensive list of questions to obtain necessary ADR-related information and being practical enough for patients to complete prior to their appointment with their clinician, as well as discussion with their clinician in busy practice settings [44]. We found that the level of detail was variable between the questionnaires; general questionnaires simply asked whether patients had experienced sideeffects, whereas condition-or medication-specific questionnaires provided a thorough list of potential ADRs. Questionnaires containing a list of potential ADRs have the advantage of asking about side-effects that patients may not think of; however, the disadvantage is that only known ADRs are listed. Whilst the use of open-ended questions allows patients to mention ADRs which may not already be listed or which they have missed due to not understanding the particular language used, only around 50% of questionnaires included in this systematic review included open-ended questions which allow free-text responses. The lack of time to discuss ADRs during busy practice settings has been identified as a barrier to reporting ADRs [9][10][11]; thus, future patient-reported questionnaires could consider asking patients to list additional side-effects by asking an open-ended question. Open-ended questions do increase response burden and time to complete the survey; therefore, it will be important to assess the duration of time required to complete the questionnaire to reflect the practicality of use.
Previous studies have shown that patients and clinicians may place emphasis on different aspects of treatment [6,8]. The choice to continue medications may depend on the type and severity of side-effects for patients, whereas clinicians' decision to continue medications may be driven by mortality and morbidity benefits [8]. In this systematic review, we found that less than half of the questionnaires explicitly asked patients to indicate whether the side-effects are bothersome. Considering that the severity of sideeffects has an effect on patients' decision to continue a treatment, it would be important for patients to be able to indicate how severe or bothersome the side-effects actually are for the patients. Additionally, less than half of the questionnaires asked the patients whether the symptom or side-effect is likely medication-related. Uncertainty by patients in whether side-effects are medication-related or not has been identified as a barrier to patients discussing them during appointments [9][10][11]14]. Other barriers to reporting of side-effects are low awareness of patients and healthcare professionals regarding side-effect reporting systems, as well as lack of integration of these questionnaires with electronic health record systems.
The validity of questionnaires is a critical aspect of questionnaire development; however, many questionnaires included in this systematic review did not report assessing the different components of validity. Only around 50% of the questionnaires mentioned assessing content validity, as well as criterion or construct testing. Previous studies have stated the importance of firstly assessing content validity prior to testing other aspects such as construct validity, reliability, and responsiveness [44,45]. Approximately 80% of questionnaires had patient input in their development or validation; it is recommended that all patient-reported questionnaires consider having end-user input to ensure the questionnaires resonate with the target audience [14,26,46].
In terms of reliability of questionnaires, less than 50% of the questionnaires included in this review were tested for internal consistency and test-retest reliability, which fall under the broad category of reliability. A previous literature review of patient-reported side-effect questionnaires undertaken in 2008 argued that it is unclear if internal consistency is a good measure for side-effect questionnaires [14]. This is because there may be a wide range of side-effects which may be unrelated to each other, lowering the internal consistency of the questionnaire, whereas the questionnaire may in fact be appropriate in measuring the different types of potential side-effects [14]. Of the questionnaires included in our systematic review that were tested for internal consistency, the high Cronbach's α of more than 0.7 for all the questionnaires tested suggests that internal consistency can be used a measure of reliability for patient side-effect questionnaires.
Factors that can affect choice of questionnaire for use in research or practice include the population in which the tool will be used, the time taken to complete the tool, psychometric properties of the questionnaires, and presence of a scoring system for practicality reasons. Findings of our systematic review suggest that the Glasgow Antipsychotic Side-Effect Scale (GASS) [41], Patient Assessment Questionnaire (PAQ) [36], and Maudsley Side-Effects (MSE) [15], which have adequate psychometric properties and a scoring system, are suitable for use in clinical practice to identify side-effects of antipsychotic medications. The Patient-Reported Adverse Drug Event Questionnaire [16], a general questionnaire which consists of a comprehensive list of adverse drug events and has good psychometric properties, is considered suitable for use in clinical practice.
To the best of our knowledge, this is the first systematic review that summarised the range of patient-reported questionnaires to identify side-effects of medications. Previous systematic reviews focused on specific types of medications, such as the 2015 systematic review which included rating scales to measure the side-effects of antipsychotic medications [47]. A limitation of this systematic review is the exclusion of non-English studies which may have subsequently removed questionnaires in other languages. The focus of our systematic review was to identify questionnaires that can be used in clinical practice to detect side-effects of medications used for any condition or common chronic conditions. As a result, there may be questionnaires for rare conditions which employ novel ways of obtaining side-effect-related information that were excluded in our systematic review. While we used a comprehensive search strategy and covered several databases, we may not have identified all questionnaires available that meet our inclusion criteria. There may be a range of terms which studies may use to refer to the concept of "patient report" such as "patient perception" which may have resulted in some questionnaires being missed.

Conclusions
This systematic review summarised the available patient-reported questionnaires that can be used in research and clinical practice to identify ADRs. Questionnaires that have been developed to date most commonly focused on identifying side-effects due to mental health medications. Patient-reported side-effect questionnaires can be a useful tool to identify ADRs that may not otherwise be reported by patients, potentially facilitating improved medication adherence and patient outcomes. Results of this systematic review highlight the need for more robust validity and reliability testing when developing patientreported ADR questionnaires. Future studies could consider assessing the applicability and effectiveness of these tools in clinical practice to improve patient care and outcomes.

Data Availability Statement:
No new data were created or analyzed in this study. Data sharing is not applicable to this article.