Akin-Akinyosoye, Kehinde
(2020)
A clinical assessment tool to improve the use of pain relief treatments in knee osteoarthritis.
PhD thesis, University of Nottingham.
Abstract
Background: In the UK, approximately 25% of individuals aged over 55 have chronic knee pain, often due to osteoarthritis (OA). Knee pain originates from the joint due to structural changes or inflammation (peripheral mechanisms), and is often intensified by processing of afferent signals by the central nervous system (central mechanisms). Imaging and psychophysical approaches could inform the presence of underlying mechanisms within individuals with knee pain but lack feasibility within clinical settings. Feasible and validated self-report approaches that can aid identification of knee OA pain mechanisms are currently unavailable.
Objectives: [1] to generate a shortlist of self-report items which reflect traits associated with underlying pain mechanisms; [2] to select a valid set of self-report items that measure a phenotypic trait associated with pain mechanisms; [3] to investigate the ability of the newly identified items to predict 1-year pain outcomes; [4] to understand participants’ interpretation of items included within the developing questionnaire to inform item revision where necessary; [5] to evaluate the psychometric properties of a newly developed mechanism-based questionnaire.
Methods: Item generation and selection was based on exploratory analysis of responses to shortlisted items by individuals reporting knee pain (n=2152) included within the ‘Knee Pain in the Community (KPIC)’ cohort study. A subset of these participants (knee pain n=322, no knee pain n=98) undertook Pressure Pain Detection Thresholds (PPT) assessments at baseline. Items measuring specific traits related to pain mechanisms were selected from the survey based on expert consensus, face validity, item association with underlying phenotypes measured by originating host questionnaires, adequate targeting, and PPT correlations. An underlying trait was sought by factor analysis of the selected items.
To examine the predictive validity of baseline scores for the identified trait, logistic and linear regression models assessed associations with 1-year follow-up pain outcomes. Receiver-operator-characteristic (ROC) curves and areas-under-the-curve (AUC) compared the predictive strength of the identified trait to other predictors of pain outcome.
Selected items were rewritten and included within the Central Aspects of Pain in the Knee (CAP-Knee) questionnaire. Cognitive interviews across individuals with knee pain (n=22) participating within the ’CAP-Knee study’ assessed participant interpretation of CAP-Knee items. Thematic analysis of participants’ discussions for each item was used to identify emergent themes which were categorised according to whether or not they were aligned to the intended interpretation of the item. Content analysis across interview transcripts allowed coding of participant responses following Tourangeau’s question response model: comprehension (completely-, partially or not completely aligned), retrieval (no-, partial- and complete- retrieval difficulty), judgement (certain initial or uncertain initial judgement) and response formulation (consistent or inconsistent).
Items were rewritten and retested in another group of interviews if (i) a mixture of aligned and not aligned themes emerged from discussions for an item, and ii) >15% of participants provided responses related to codes of poor item function, including complete non-alignment, complete retrieval difficulty, uncertain initial response and no response consistency.
Psychometric properties of the CAP-Knee were assessed in 250 community-based individuals with knee pain, of whom 76 completed the CAP-Knee twice over one month to measure repeatability.
Results:
Item generation and selection: Eight self-report items measuring traits of anxiety, depression, catastrophizing, neuropathic-like pain, fatigue, sleep disturbance, pain distribution, and cognitive impact were identified as likely indices of central pain mechanisms. PPTs were associated with items representing each trait and with their originating questionnaires. A single factor, interpreted as “central mechanisms trait” was identified across the 8 selected items and explained variation in PPT (R2 = 0.17) better than did any originating questionnaire (R2 = 0.10-0.13).
Predictive Validity: The central mechanisms trait score significantly predicted year 1 pain outcomes, even after adjustment for age, sex, BMI, radiographic OA severity and symptom duration (Pain persistence: RR=2.14, n=204, p=0.001; Persistent pain severity: β=0.47, n=118; p<0.002). The central mechanisms trait score showed good discrimination power in distinguishing pain persistence cases from resolved pain cases (AUC = 0.70; n=1471). The discrimination power of other predictors, including radiographic OA (AUC = 0.62; n=204), age, sex and BMI (AUC range = 0.51 to 0.64; n=1471), improved significantly (p<0.04) when the central mechanisms trait was included in each logistic regression model (AUC range = 0.69 to 0.74).
Interpretation of CAP-Knee items: Participant interpretation of the final version of the CAP-Knee items was closely aligned to their intended meaning. Overall, 15 key themes were discussed by participants for items included within the CAP-Knee {One Anxiety theme = Fear; two Depression themes = Social function, Physical limitation; two Catastrophizing themes = Causes and consequences, Avoidance behaviours; two Cognitive impact themes = Task distraction, and Hypervigilance; two Sleep themes = Sleep disturbance and Use of sleeping aids; two Fatigue themes = Source of fatigue, Fatigue relief; one Pain distribution theme = Painful sites and three Neuropathic-like pain themes = Thermal allodynia, Weather induced pain and Thermotherapy. A mixture of aligned and not aligned themes emerged from discussions about the Neuropathic-like pain- and depression- items. More than 15% of participants provided responses indicative of poor item performance for the Neuropathic-like pain item only, but not the depression item.
The rewritten version of the neuropathic-like pain item was considered to work well.
Psychometric properties of the CAP-Knee: CAP-Knee displayed a wide range of scores across the study population (median 8, range 0-24). Internal consistency was acceptable (α = 0.75) and test–retest reproducibility excellent (ICC=0.91, 95% CI, 0.86-0.94). All CAP-Knee items contributed significantly (item loading range = 0.21-0.92; p<0.01) to one distinct factor (CFI = 0.99; TLI= 0.98; X2(df)=37(20); RMSEA= 0.06). The CAP-Knee targeted the knee pain population well and constituted a unidimensional measure. Fit to the Rasch model was improved by item rescoring.
Conclusion: The CAP-Knee is a simple and valid self-report questionnaire, consisting of the 8 selected items which measure a single latent trait (‘central mechanisms’) in individuals with knee pain, and may help identify and target treatments that aim to reduce central sensitisation. No items associated with peripheral mechanisms of knee OA pain were identified in this project. Future research should seek to clinically validate the stratification and prognostic characteristics of the CAP-Knee.
Actions (Archive Staff Only)
 |
Edit View |
Tools
Tools
![[thumbnail of Compilation_repository.pdf]](https://eprints.nottingham.ac.uk/style/images/fileicons/application_pdf.png "Compilation_repository.pdf")