Developing a novel tool for assessing olfactory dysfunction: The parosmia, phantosmia, and anosmia test (PARPHAIT)

Abstract

Background Symptoms of olfactory dysfunction (OD) apply to different conditions, one of which is COVID-19. OD may manifest as a loss or reduction of smell, called quantitative OD, or as a change in the perception of odours, also known as qualitative OD. The symptomatic picture may look different across cases, and impact on the quality of life to varying extents. The methods for measuring symptoms also differ across studies, and involve clinical interviews, objective tests, and questionnaires that may or may not be standardised. When we ask patients about their symptoms, the formulation of items and response options vary across tools, which makes room for interpretation and heterogeneous results. For quantitative OD, objective, standardised tests have been developed and evaluated, whilst for qualitative OD, less is known, and measurement tools could be improved. Based on this unmet need, the current PhD project has focused on the measurement of parosmia and phantosmia, mainly in the context of COVID-19. The overall aim was to better understand the measurement of qualitative OD, and to improve the methods of capturing these symptoms. This aim motivated the following objectives: To provide an overview of the tools that exist in capturing qualitative OD, to explore variations, and to identify discrepancies in the formulations of items and response options in these tools. Based on this overview, we identified a need for a standardised, specific, and clear tool to capture both qualitative and quantitative OD, and this led to developing and testing the novel Parosmia, phantosmia, and anosmia test (PARPHAIT). Methodology This PhD project followed a sequential mixed methods design, consisting of a systematic literature review, quantitative, and qualitative methods. First, a systematic literature review was carried out. The aim was to provide an overview of the tools used to measure qualitative OD, as well as to identify discrepancies in the presentation of items and response options. The primary outcomes were the objective tests and questionnaires, and the item and response formulations. Secondary outcomes involved the number of items, response design, scoring protocol, validation and reliability testing, and demographic variables like age and sex. This process followed a modified PRISMA guideline, and involved searching MEDLINE, Web of Science, and EMBASE, to which the search was restricted to a COVID-19 population. Results were interpreted and presented narratively. Based on the findings of the review, the second study involved developing an initial pool of items, testing its content validity, factor structure, and reliability. The study was primarily quantitative but did also include qualitative methods. An initial 93-item version of PARPHAIT was distributed digitally to a sample of subjects with OD primarily due to COVID-19. They were asked to complete the questionnaire and to rate the clarity, length, and response design, among other aspects, on a scale from 1 to 5, where 5 correspond to “highly suitable”. Following this content validation, correlations between items were assessed before carrying out an exploratory factor analysis (EFA) and reliability test. To further validate PARPHAIT’s content and applicability, the third study involved inviting international experts to review and evaluate the questionnaire. This was done qualitatively through individual digital interviews, after which transcribed data were interpreted using a directed content analysis. Coding was done using Nvivo and was primarily deductive according to the interview guide. However, some codes were generated inductively. Results The first study included 72 articles, and these identified four objective tests, seven widely used questionnaires, some of which were standardised, and 43 self-developed questionnaires. The self-developed questionnaires were used mostly for descriptive purposes. Tools were generally heterogeneous, and some were unclear with regards to how symptoms were defined and presented. For instance, a typical question assumed symptoms were constant and unpleasant (e.g. “I always have a bad smell in my nose”). Importantly, parosmia is generally accounted for to a greater extent than phantosmia, which has not been the centre of attention in the measurement of qualitative OD. In the second study, PARPHAIT was completed by 165 participants, ratings were provided by 164, and 85 had additional comments. Some suggested including triggers like peanut butter and egg whites, and others would like a list of descriptors or a free-text entry to explain symptoms in more detail. Some did not consider the parosmia and phantosmia subscales applicable because they had anosmia, and others regarded the formulations of questions and introductory texts as unclear. Inspecting the inter-item correlations led to reducing the total items to 56. The EFA further reduced these to 34, and suggested four factors that were named “Loss of smell”, “Presence, valence, frequency, and intensity of phantosmia”, “Triggers of parosmia”, and “Frequency of parosmia”. All factors had very good to excellent internal consistency (the lowest being α=.88 and the highest α=.95). The third study both confirmed and challenged previous findings. For instance, items specific to a change in the pleasantness and quality of odours did not appear to contribute to capturing symptoms after the EFA and were removed. However, as one of the experts noted: “… there is also the issue of hedonic vs. qualitative changes in parosmia. My suspicion is that these can be independent.” The distinction between these and accounting for both aspects is important, as they make up the very nature of parosmia. Moreover, an odour may be perceived as more or less pleasant, but not necessarily different, or it may be different, but not necessarily more or less pleasant. In addition, some suggested applying response designs according to the type and nature of questions. For instance, one would respond to questions about intensity differently from questions about food triggers. In line with patients’ feedback, formulations could be clearer and other important triggers should be included, like cucumber and toothpaste. A longer time frame was emphasised as well as shortening the total length of PARPHAIT. Conclusions In summary, this PhD project aimed to gain a deeper understanding of how qualitative OD is assessed and to improve such methods. Results identified the need for a standardised, specific tool to capture qualitative OD, and the development of such a tool involved multiple methods and perspectives. Developing PARPHAIT started from addressing patients’ viewpoints, which have contributed to a sound fundament for refining the tool. Statistical tests and experts’ perspectives were then integrated, and changes were applied, leading to the 32-item version of PARPHAIT. Introducing PARPHAIT to a clinical setting may benefit the communication between the patient and clinician, ensure accurate diagnoses and treatment, and improve patient’s own insight into their own condition. In the context of research, PARPHAIT may promote a greater consensus around the measurement of OD and the use of terms. It provides a specific and detailed account of symptoms, and focuses on phantosmia, which previously has not been the centre of attention. PARPHAIT still needs to be refined, validated, and tested in larger samples and different populations. However, PARPHAIT is a promising contribution that adds value both to the clinical context and the scientific community.