OP0228 USE OF NON-STEROIDAL ANTI-INFLAMMATORY DRUGS AND RISK OF COMORBIDITIES IN PEOPLE WITH AND WITHOUT OSTEOARTHRITIS - A UK PRIMARY CARE DATABASE COHORT STUDY

Background

People with osteoarthritis (OA) are at higher risk of developing a wide array of comorbidities. Whether the use of non-steroidal anti-inflammatory drugs (NSAIDs) contributes to the increased risk of some incident comorbidities remains unknown.

Objectives

To examine the contribution of NSAIDs in the development of a wide range of comorbidities in people with and without OA.

Methods

This observational cohort study used the UK primary care Clinical Practice Research Datalink (CPRD) GOLD containing data on 20+ million people covering 937 practices. We identified 259,000 people with incident OA and 259,000 age (±2 years), sex, and practice matched controls at 1:1 ratio. Controls were assigned the same index date (the date of first diagnosis of OA) as cases for the start of follow-up. Both cases and controls were further divided into two groups according to NSAID prescriptions at any time after the index date. This allowed us to examine both the main effect of each exposure and interaction between OA and NSAID exposure after the index date. People with an NSAID prescription before the index date were excluded from the study. NSAID exposure was defined as at least two prescriptions within 90 days. Exposure status of each participant was assessed every six months as yes/no until the end of the study/outcome of interests/death/last data available, whichever came first. Comorbidities were grouped into 9 categories as cancer, cardiovascular disease (CVD), endocrine, psychological, renal, gastrointestinal (GI), genitourinary, hepatic, and neurological conditions. Propensity scores for the prescription of NSAIDs were calculated using a logistic regression model including age, sex, body mass index (BMI), musculoskeletal and pain related conditions covariates. The propensity score adjusted time varying exposure analysis was undertaken using a multivariate COX model and hazard ratio (HR) and 95% confidence intervals were calculated. Proportional hazard assumption was tested using Schoenfeld test. Smoking, alcohol, ever prescription of proton pump inhibitors (PPIs) and other comorbidities were included in the adjusted model. The additional contribution of NSAIDs and OA towards the incident comorbidity was estimated using addictive interaction methods. We also investigated the individual risk across non-selective, and COX-2 selective NSAIDs.

Results

The mean age was 59.4±12.8 years in people with OA and 60.2±12.8 years for controls with 57.7% being female. Nearly two thirds of people with OA were prescribed NSAIDs as defined, compared to one third in the control population. People with OA and exposed to NSAIDs had highest risk of developing psychological (1.51; 1.43,1.60), CVD (1.38; 1.33,1.43), cancer (1.34; 1.25,1.44), GI (1.25; 1.16,1.34) and renal (1.17; 1.11,1.24) comorbidities after adjusting for all the covariates and PPI drugs, compared to the non-OA and non-NSAID group. (Figure 1) Interaction between OA and NSAID was significant for cancer, GI, renal, hepatic, and neurological outcomes. Within people with OA, non-selective NSAIDs increased the risk of CVD (1.25; 1.20,1.30), cancer (1.11; 1.04,1.19), endocrine (1.15; 1.10,1.19), renal (1.19; 1.13,1.26) and psychological (1.21; 1.15,1.28) comorbidities, whereas COX-2 selective NSAIDs increased risk of incident CVD (1.34; 1.25,1.44), endocrine (1.13; 1.04,1.21), renal (1.25; 1.14,1.37), and psychological (1.21; 1.09,1.34) comorbidities.

Figure 1.

Hazard ratio of developing different comorbidities (reference group: no OA and no NSAIDs) OA- Osteoarthritis; NSAIDS- Non-steroidal anti-inflammatory drugs.

Conclusion

Use of NSAIDs among people with OA is associated with increased risk of a wide variety of comorbidities. Non-selective and COX-2 selective NSAIDs are both associated with increased risk of cardiovascular, renal, and psychological comorbidities.

Acknowledgements

We thank the Patient Research Participants (PRP) members Jenny Cockshull, Stevie Vanhegan, and Irene Pitsillidou for their involvement since the beginning of the project. We would like to thank the FOREUM for financially supporting the research.

Disclosure of Interests

Subhashisa Swain: None declared, Anne Kamps: None declared, Jos Runhaar: None declared, Andrea Dell’Isola: None declared, Aleksandra Turkiewicz: None declared, Danielle E Robinson: None declared, Victoria Y Strauss: None declared, Christian Mallen: None declared, Chang-Fu Kuo: None declared, Carol Coupland: None declared, Michael Doherty Consultant of: Consultant of: Advisory borads on gout for Grunenthal and Mallinckrodt, Grant/research support from: Michael Doherty Grant/research support from: AstraZeneca funded the Nottingham Sons of Gout study, Aliya Sarmanova: None declared, Daniel Prieto-Alhambra Speakers bureau: paid speaker services from Amgen and UCB Biopharma., Consultant of: His department has received advisory or consultancy fees from Amgen, Astellas, AstraZeneca, Johnson, and Johnson, and UCB Biopharma, Grant/research support from: Prof. Prieto-Alhambra’s research group has received grant support from Amgen, Chesi-Taylor, Novartis, and UCB Biopharma., Martin Englund: None declared, S.M.A. Bierma-Zeinstra: None declared, Weiya Zhang Speakers bureau: Speakers bureau: Bioiberica as an invited speaker for EULAR 2016 satellite symposium, Consultant of: Consultant of: Grunenthal for advice on gout management,

POS1124 EVALUATION OF COMORBIDITY PATTERNS AND IDENTIFICATION OF SUB-GROUPS IN PATIENTS DIAGNOSED WITH HIP OSTEOARTHRITIS IN 94,720 PATIENTS FROM SPAIN

Background

Osteoarthritis (OA) patients are more likely to have other comorbidities (Swain, Sarmanova et al. 2020). Improving the understanding of comorbidity profiles of OA patients may lead to improvement in their clinical care.

Objectives

To identify sub-groups in patients diagnosed with hip OA using patterns of comorbidity.

Methods

Routinely-collected data of individuals ≥18 years with an incident diagnosis of hip OA (baseline/time of diagnosis), with at least 1 year of follow-up in SIDIAP (Information System for Research in Primary Care, a primary case database from Spain) were collected from January 1st 2006 to June 31st 2020. Those with soft-tissue disorders or other bone/cartilage diseases at the same joint in the year prior/after baseline were excluded. Comorbidities associated with OA in the literature and present in ≥1% of the study population were included. Clusters of comorbidities were identified at baseline using latent class analysis (LCA), a soft clustering method that classifies individuals according to the distribution of their measured items. The number of clusters or sub-groups within the study population was decided by comparing goodness of fit parameters (CAIC, BIC, ABIC) and log-likelihood changes of models from 2 to 8 clusters. The selected model was externally evaluated by a survival analysis assessing 10 years mortality within each cluster, where the weight of the posterior probability was used as a probability of sampling weight.

Results

We identified 94,720 individuals with an incident diagnosis of hip OA, 56.3% women and 43.7% men, with a mean age (SD) of 67.2 (13.1) years. We selected the LCA model with 5 clusters that could be described as: healthier (lower prevalence of all comorbidities than average in the cohort), multimorbidity (higher prevalence of all comorbidities, multiple comorbidities), back/neck pain plus mental health (B/N-mental), cardiovascular disease (CVD), and metabolic syndrome (MetS) (Figure 1). Cox regression (HR [95CI%]) showed higher mortality risk for multimorbidity (3.76 [3.70-3.83]), CVD (1.56 [1.53-1.59]) and MetS (4.56 [4.35-4.78]), compared to healthy. No difference was observed for B/N-mental cluster.

Figure 1.

Distribution of comorbidities within each cluster using latent class analysis. Clusters were described as Healthier, Multimorbidity, B/N-mental, CVD and MetS. Black horizontal lines represent the prevalence of the comorbidity before the clusterization. Abbreviations: Healthier, lower prevalence of all comorbidities; Multimorbidity, higher prevalence of all comorbidities; B/N-mental, back/neck pain plus mental health disorders; CVD, cardiovascular disease; Met, metabolic syndrome; Bhp, benign prostate hypertrophy; Chd, chronic heart disease; Chf, chronic heart failure; Ckd, chronic kidney disease; Copd, chronic obstructive pulmonary disease; Gbs, gall bladder stone; Gerd, gastroesophageal reflux disease; Ibd, inflammatory bowel disease; Ovd, other vessel diseases; Substance, substance abuse.

Conclusion

Clustering of co-morbidities in hip OA patients at the time of diagnosis has the potential to detect sub-groups of hip OA patients who might require additional care.

References

[1]Swain, S., A. Sarmanova, C. Coupland, M. Doherty and W. Zhang (2020). “Comorbidities in Osteoarthritis: A Systematic Review and Meta-Analysis of Observational Studies.” Arthritis Care Res (Hoboken) 72(7): 991-1000.

Acknowledgements

We thank the Patient Research Participants (PRP) members Jenny Cockshull, Stevie Vanhegan, and Irene Pitsillidou for their involvement since the beginning of the project. We would like to thank the FOREUM for financially supporting the research.

Disclosure of Interests

None declared

Comparative efficacy and safety of acetaminophen, topical and oral non-steroidal anti-inflammatory drugs for knee osteoarthritis: evidence from a network meta-analysis of randomized controlled trials and real-world data

OBJECTIVE

Current global guidelines regarding the first-line analgesics (acetaminophen, topical or oral non-steroidal anti-inflammatory drugs [NSAIDs]) for knee osteoarthritis remain controversial and their comparative risk-benefit profiles have yet to be adequately assessed.

DESIGN

Pubmed, Embase, Cochrane Library, and Web of Science were searched from database inception to March 2021 for randomized controlled trials (RCTs) comparing acetaminophen, topical NSAIDs and oral NSAIDs directly or indirectly in knee osteoarthritis. Bayesian network meta-analyses were conducted. A propensity-score matched cohort study was also conducted among patients with knee osteoarthritis in The Health Improvement Network database.

RESULTS

122 RCTs (47,113 participants) were networked. Topical NSAIDs were superior to acetaminophen (standardized mean difference [SMD] = -0.29, 95% credible interval [CrI]: -0.52 to -0.06) and not statistically different from oral NSAIDs (SMD = 0.03, 95% CrI: -0.16 to 0.22) for function. It had lower risk of gastrointestinal adverse effects (AEs) than acetaminophen (risk ratio [RR] = 0.52, 95%CrI: 0.35 to 0.76) and oral NSAIDs (RR = 0.46, 95%CrI: 0.34 to 0.61) in RCTs. In real-world data, topical NSAIDs showed lower risks of all-cause mortality (hazard ratio [HR] = 0.59, 95% confidence interval [CI]: 0.52 to 0.68), cardiovascular diseases (HR = 0.73, 95%CI: 0.63 to 0.85) and gastrointestinal bleeding (HR = 0.53, 95%CI: 0.41 to 0.69) than acetaminophen during the one-year follow-up (n = 22,158 participants/group). A better safety profile was also observed for topical than oral NSAIDs (n = 14,218 participants/group).

CONCLUSIONS

Topical NSAIDs are more effective than acetaminophen but not oral NSAIDs for function improvement in people with knee osteoarthritis. Topical NSAIDs are safer than acetaminophen or oral NSAIDs in trials and real-world data.

OP0074 MULTIMORBIDITY CLUSTERS, DETERMINANTS AND TRAJECTORIES IN OSTEOARTHRITIS IN THE UK: FINDINGS FROM THE CLINICAL PRACTICE RESEARCH DATALINK

Background:

Multimorbidity (≥2 chronic conditions) escalates the risk of adverse health outcomes. However, its burden in people with osteoarthritis (OA) remains largely unknown.

Objectives:

To identify the clusters of patients with multimorbidity and associated factors in OA and non-OA populations and to estimate the risk of developing multimorbidity clusters after the index date (after diagnosis).

Methods:

The study used the Clinical Practice Research Datalink – a primary care database from the UK. Firstly, age, sex and practice matched OA and non-OA people aged 20+ were identified to explore patterns and associations of clusters of multimorbidity within each group. Non-OA controls were assigned with same index date as that of matched OA cases. Secondly, multimorbidity trajectories for 20 years after the index date were examined in people without any comorbidities at baseline in both OA and non-OA groups. Latent class analysis was used to identify clusters and latent class growth modelling was used for cluster trajectories. The associations between clusters and age, sex, body mass index (BMI), alcohol use, smoking habits at baseline were quantified through multinomial logistic regression.

Results:

In total, 47 long-term conditions were studied in 443,822 people (OA- 221922; non-OA- 221900), with a mean age of 62 years (standard deviation ± 13 years), and 58% being women. The prevalence of multimorbidity was 76.6% and 68.9% in the OA and non-OA groups, respectively. In the OA group five clusters were identified including relatively healthy (18%), ‘cardiovascular (CVD) and musculoskeletal (MSK)’ (12.3%), metabolic syndrome (28.2%), ‘pain and psychological (9.1%), and ‘musculoskeletal’ (32.4%). The non-OA group had similar patterns except that the ‘pain+ psychological’ cluster was replaced by ‘thyroid and psychological’. (Figure 1) Among people with OA, ‘CVD+MSK’ and metabolic syndrome clusters were strongly associated with obesity with a relative risk ratio (RRR) of 2.04 (95% CI 1.95-2.13) and 2.10 (95% CI 2.03-2.17), respectively. Women had four times higher risk of being in the ‘pain+ psychological’ cluster than men when compared to the gender ratio in the healthy cluster, (RRR 4.28; 95% CI 4.09-4.48). In the non-OA group, obesity was significantly associated with all the clusters.

Figure 1

: Posterior probability distribution of chronic conditions across the clusters in Osteoarthritis (OA, n=221922) and Non-Osteoarthritis (Non-OA, n=221900) group. COPD- Chronic Obstructive Pulmonary Disease; CVD- Cardiovascular; MSK- Musculoskeletal

OA (n=24139) and non-OA (n=24144) groups had five and four multimorbidity trajectory clusters, respectively. Among the OA population, 2.7% had rapid onset of multimorbidity, 9.5% had gradual onset and 11.6% had slow onset, whereas among the non-OA population, there was no rapid onset cluster, 4.6% had gradual onset and 14.3% had slow onset of multimorbidity. (Figure 2)

Figure 2

: Clusters of multimorbidity trajectories after index date in OA (n=24139) and Non-OA (n=24144)

Conclusion:

Distinct identified groups in OA and non-OA suggests further research for possible biological linkage within each cluster. The rapid onset of multimorbidity in OA should be considered for chronic disease management.

Supported by:

Acknowledgments:

We would like to thank the University of Nottingham, UK, Beijing Joint Care Foundation, China and Foundation for Research in Rheumatology (FOREUM) for supporting the study.

Disclosure of Interests:

Subhashisa Swain: None declared, Carol Coupland: None declared, Victoria Strauss: None declared, Christian Mallen Grant/research support from: My department has received financial grants from BMS for a cardiology trial., Chang-Fu Kuo: None declared, Aliya Sarmanova: None declared, Michael Doherty Grant/research support from: AstraZeneca funded the Nottingham Sons of Gout study, Consultant of: Advisory borads on gout for Grunenthal and Mallinckrodt, Weiya Zhang Consultant of: Grunenthal for advice on gout management, Speakers bureau: Bioiberica as an invited speaker for EULAR 2016 satellite symposium

Trends in incidence and prevalence of osteoarthritis in the United Kingdom: findings from the Clinical Practice Research Datalink (CPRD)

OBJECTIVE

This study aimed to explore the incidence and prevalence of OA in the UK in 2017 and their trends from 1997 to 2017 using a large nationally representative primary care database.

DESIGN

The UK Clinical Practice Research Datalink (CPRD) comprising data on nearly 17.5 million patients was used for the study. The incidence and prevalence of general practitioner diagnosed OA over a 20 years period (1997-2017) were estimated and age-sex and length of data contribution standardized using the 2017 CPRD population structure. Cohort effects were examined through Age-period-cohort analysis.

RESULTS

During 1997-2017, there were 494,716 incident OA cases aged ≥20 years. The standardised incidence of any OA in 2017 was 6.8 per 1000 person-years (95% CI 6.7 to 6.9) and prevalence was 10.7% (95% CI 10.7-10.8%). Both incidence and prevalence were higher in women than men. The incidence of any-OA decreased gradually in the past 20 years at an annual rate of -1.6% (95%CI -2.0 to -1.1%), and the reduction speeded up for people born after 1960. The prevalence of any-OA increased gradually at an annual rate of 1.4% (95% CI 1.3-1.6%). Although the prevalence was highest in Scotland and Northern Ireland, incidence was highest in the East Midlands. Both incidence and prevalence reported highest in the knee followed by hip, wrist/hand and ankle/foot.

CONCLUSION

In the UK approximately one in 10 adults have symptomatic clinically diagnosed OA, the knee being the commonest. While prevalence has increased and become static after 2008, incidence is slowly declining. Further research is required to understand these changes.

Baseline self-report 'central mechanisms' trait predicts persistent knee pain in the Knee Pain in the Community (KPIC) cohort

OBJECTIVES

We investigated whether baseline scores for a self-report trait linked to central mechanisms predict 1 year pain outcomes in the Knee Pain in the Community cohort.

METHOD

1471 participants reported knee pain at baseline and responded to a 1-year follow-up questionnaire, of whom 204 underwent pressure pain detection thresholds (PPTs) and radiographic assessment at baseline. Logistic and linear regression models estimated the relative risks (RRs) and associations (β) between self-report traits, PPTs and pain outcomes. Discriminative performance for each predictor was compared using receiver-operator characteristics (ROC) curves.

RESULTS

Baseline Central Mechanisms trait scores predicted pain persistence (Relative Risk, RR = 2.10, P = 0.001) and persistent pain severity (β = 0.47, P < 0.001), even after adjustment for age, sex, BMI, radiographic scores and symptom duration. Baseline joint-line PPTs also associated with pain persistence (RR range = 0.65 to 0.68, P < 0.02), but only in univariate models. Lower baseline medial joint-line PPT was associated with persistent pain severity (β = -0.29, P = 0.013) in a fully adjusted model. The Central Mechanisms trait model showed good discrimination of pain persistence cases from resolved pain cases (Area Under the Curve, AUC = 0.70). The discrimination power of other predictors (PPTs (AUC range = 0.51 to 0.59), radiographic OA (AUC = 0.62), age, sex and BMI (AUC range = 0.51 to 0.64), improved significantly (P < 0.05) when the central mechanisms trait was included in each logistic regression model (AUC range = 0.69 to 0.74).

CONCLUSION

A simple summary self-report Central Mechanisms trait score may indicate a contribution of central mechanisms to poor knee pain prognosis.

Thresholds of ultrasound synovial abnormalities for knee osteoarthritis - a cross sectional study in the general population

OBJECTIVE

To establish "normal" ranges for synovial thickness and effusion detected by ultrasound (US) and to determine cut-offs associated with knee pain (KP) and radiographic knee osteoarthritis (RKOA) in the community.

METHODS

147 women and 152 men ≥40 years old were randomly selected from the Nottingham KP and Related Health in the Community (KPIC) cohort (n = 9506). The "normal" range was established using the percentile method in 163 participants who had no KP and no RKOA. Optimal (maximum sensitivity and specificity) and high specificity (90%) cut-offs were established using receiver operating characteristic (ROC) curve analysis in a comparison between people with both KP and RKOA and normal controls.

RESULTS

Effusion and synovial hypertrophy differed by gender but not by age or laterality, therefore gender-specific reference limits were estimated. However, the "normal" ranges between men and women were similar for effusion (0-10.3 mm vs 0-9.8 mm), but different for synovial hypertrophy (0-6.8 mm vs 0-5.4 mm). Power Doppler Signal (PDS) in the healthy controls was uncommon (1.2% in men and 0.0% in women). The optimal cut-off was 7.4 mm for men and 5.3 mm for women for effusion, and 3.7 and 1.6 for hypertrophy respectively. The high specificity cut-off was 8.9 for men and 7.8 for women for effusion, and 5.8 and 4.2 for hypertrophy respectively.

CONCLUSIONS

US effusion and synovial hypertrophy but not PDS are common, but differ by gender, in community-derived people without painful knee OA. Currently used cut-offs for abnormality need reappraisal.

Contribution of central and peripheral risk factors to prevalence, incidence and progression of knee pain: a community-based cohort study

AIM

To explore risk factors that may influence knee pain (KP) through central or peripheral mechanisms.

METHODS

A questionnaire-based prospective community cohort study with KP defined as pain in or around a knee on most days for at least a month. Baseline prevalence, and one year incidence and progression (KP worsening) were examined. Central (e.g., Pain Catastrophizing Scale (PCS)) and peripheral (e.g., significant injury) risk factors were examined. Adjusted odds ratio (OR) and 95% confidence interval (CI) were calculated using logistic regression. Proportional risk contribution (PRC) was estimated using receiver-operator-characteristic (ROC) analysis.

RESULTS

Of 9506 baseline participants, 4288 (45%) had KP (men 1826; women, 2462). KP incidence was 12% (men 11%, women 13%), and KP progression 19% (men 16%, women 21%) at one year. While both central and peripheral factors contributed to prevalence, central factors contributed more to progression, and peripheral factors more to incidence of KP. For example, although PCS (OR 2.06, 95% CI 1.88-2.25) and injury (5.62, 4.92-6.42) associated with KP prevalence, PCS associated with progression (2.27, 1.83-2.83) but not incidence (1.14, 0.86-1.52), whereas injury more strongly associated with incidence (69.27, 24.15-198.7) than progression (2.52, 1.48-4.30). The PRC of central and peripheral factors were 19% and 23% for prevalence, 14% and 29% for incidence, and 29% and 5% for progression, respectively.

CONCLUSIONS

Both central and peripheral risk factors influence KP but relative contributions may differ in terms of development (mainly peripheral) and progression (mainly central). Further study of such relative contributions may inform primary and secondary prevention strategies.