Prevalence and pattern of comorbidities in chronic rheumatic and musculoskeletal diseases: the COMORD study.

Increased risk of comorbidities has been reported in Rheumatic and Musculoskeletal Diseases (RMD). We aimed to evaluate the prevalence and pattern of comorbidities in RMD patients nationwide, to identify multimorbidity clusters and to evaluate the gap between recommendations and real screening. Cross-sectional, multicentric nationwide study. Prevalence of comorbidities was calculated according to six EULAR axes. Latent Class Analysis identified multimorbidity clusters. Comorbidities’ screening was compared to international and local recommendations. In 769 patients (307 RA, 213 OA, 63 SLE, 103 axSpA, and 83 pSA), the most frequent comorbidities were cardiovascular risk factors and diseases (CVRFD) (hypertension 36.5%, hypercholesterolemia 30.7%, obesity 22.7%, smoking 22.1%, diabetes 10.4%, myocardial infarction 6.6%), osteoporosis (20.7%) and depression (18.1%). Three clusters of multimorbidity were identified: OA, RA and axSpA. The most optimal screening was found for CVRF (> = 93%) and osteoporosis (53%). For malignancies, mammograms were the most optimally prescribed (56%) followed by pap smears (32%) and colonoscopy (21%). Optimal influenza and pneumococcus vaccination were found in 22% and 17%, respectively. Comorbidities were prevalent in RMD and followed specific multimorbidity patterns. Optimal screening was adequate for CVRFD but suboptimal for malignant neoplasms, osteoporosis, and vaccination. The current study identified health priorities, serving as a framework for the implementation of future comorbidity management standardized programs, led by the rheumatologist and coordinated by specialized health care professionals.


comorbidities and risk factors.
• Cardiovascular: ischemic cardiovascular diseases, stroke, aneurysm, cardiac dysrhythmia, heart failure, thrombophlebitis; latest hypertension, diabetes, hyperlipidemia and renal function (serum creatinine) screening dates and results; cardiovascular risk score calculation (retrieved when present in the medical file); treatment with anti-platelets, anti-hypertensive, anti-diabetic and hypolipemic drugs. • Malignancy history (according to clinical reporting, by type), latest screening date by mammography, Pap smear, colonoscopy, Guaiac test, Prostate Specific Antigen (PSA), dermatology visit. • Infections: history of tuberculosis (active, latent, PPD, interferon gamma test), bacterial, viral (in particular hepatitis B and C, HIV), parasitic and fungal infection; latest vaccination status for influenza, pneumococcus, herpes zoster and HPV as well as up-to-date vaccination for poliomyelitis, diphtheria, tetanus and hepatitis B. • Gastrointestinal diseases: gastro-duodenal ulcer, Helicobacter pylori infection, previous gastroscopy. • Osteoporosis: low bone mineral density (BMD) (T-score = <−2.5 DS), osteoporotic fracture and location, anti-osteoporotic treatment, calcium and vitamin D supplementation and FRAX-score. A FRAX risk of > = 10% was considered as high as per the pharmacological treatment threshold recommended by the Lebanese osteoporosis societies. • Depression: diagnosed depression, anti-depressant therapy and screening for depression. Screening was done during the medical interview using the Patient Health Assessment Questionnaire (PHQ4).

Screening of comorbidities.
For each patient and each comorbidity/risk factor, optimal screening was calculated and documented as a score of yes (1) or no (0). Optimal screening for each comorbidity/ risk factor was the sum of the cases with a score of 1 divided by the total eligible patients for this screening. Identification of optimal screening according to local (when available) or international recommendations is listed in Table 1 16,[38][39][40][41][42][43][44][45][46][47][48][49][50][51][52] . The CRF was completed by a medical intern and a study nurse using a review of the medical record and an interview at the study visit. The data was entered in a Microsoft Excel database.

Statistical analysis.
A descriptive analysis of all patients was performed and a comparison between the five rheumatologists' data profile was done. The prevalence (and 95% CI) of each comorbidity and risk factor were estimated (Wald method). The number of comorbidities was correlated with predictive factors using Poisson Regression. Latent Class Analysis (LCA) was used to identify clusters of multimorbidity, which included the RMD and the most frequent comorbidities in the model. It consists of a measurement model in which individuals can be classified into mutually exclusive and exhaustive types, or latent classes, based on their pattern of categorical variables. The percentage of optimally screened patients according to the recommendations was calculated. Optimal screening (binary) was correlated with predictive factors using binary logistic regression. Analyses were conducted using the statistical software IBM SPSS Statistics 25 and XLSTAT 18.07 (LCA analysis).
Sample size calculation. The sample size was based on the width of the 95% CI of the proportion of expected events (the prevalence of comorbidities), assuming that a 753-patient sample would allow an observed 2% prevalence of a comorbidity to be estimated with a precision of 1% (95% CI 1% to 3%). (http://sampsize. sourceforge.net/iface/). ethical considerations. The study was approved by the ethical committee of the Saint-Joseph University (Approval number TFEM2016/48), Beirut, which acts in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. All physicians gave their written informed consent to interview the patients and to access the data on file and to interview the patients, all patients gave their oral informed consent to answer the study questionnaire.

Results
population characteristics. We recruited 769 patients between 2016 and 2018 (307 RA, 213 OA, 63 SLE, 103 axSpA, 83 pSpA). Significant differences in patients' demographics were found across the diseases ( Table 2). Mean age was 55.8 years (SD 13.9), (lowest in SLE 42.8 yo, highest in OA 63.6 yo, p < 0.001). Female gender was 76.9% (lowest in axSpA 43.7%, highest in SLE 93.7%, p < 0.001). 42.9% were professionally active (lowest in OA 26.1%, highest in axSpA 79.5%, p < 0.001, probably reflecting the age and gender differences). Most of the patients had a partial or total social coverage (77.1%), the rest being covered by the ministry of health (p = 0.595). Disease duration ranged from 84.5 months in OA to 103.5 months in RA (p = 0.013). 29.9% were previous or current smokers, 22.1% were current smokers (11.1% in SLE to 28.2% in axSpA, p < 0.001) and 2.9% drank alcohol regularly (1.6% in RA to 8.2% in axSpA, p = 0.562). Mean BMI was in the overweight range 27.02 kg/ m 2 (SD 4.71). There was no difference in the disease distribution between the five rheumatology clinics for the demographic and disease characteristics (p = 0.301).
A history of malignancies was found in 4.2% of patients, with the highest prevalence for breast (2.4%), prostate (1.2%) and colon (0.6%) cancers. Hematologic cancers were found in 0.4% of patients.
Tuberculosis history (latent and active) was found in 1.8% of patients. Recent history of bacterial infections (Salmonella, Pneumococcus, Brucella, E coli, Klebsiella, Pseudomonas) was found in 10% of patients. History of viral, parasitic and fungal infections (Hepatitis, CMV, Herpes, Isospora, Toxoplasma, Candida) was found in 3.2% of patients.
Gastric ulcer history was found in 5.5% of patients, Helicobacter pylori infection in 1.8% and Inflammatory Bowel Diseases (IBD) in 4.3%, highest in axial (23.3%) and peripheral (7.5%) SpA as it would be expected (p < 0.001).
The total number of comorbidities per patient was highest for OA (1.8) and lowest for axSpA (0.8), p < 0.001. In multivariate analysis, age (p < 0.001), higher BMI (p < 0.001) and biologic therapies (p = 0.05) were significantly associated with the number of comorbidities.   www.nature.com/scientificreports www.nature.com/scientificreports/ DXA was prescribed in 53% of correct indications, FRAX was available in 49% of patients, 74% of patients were supplemented with vitamin D.
Correct vaccination (influenza and pneumococcal) was found only in 22% and 17% respectively. Childhood vaccinations were vaguely remembered and were not included because of high recall bias and the absence of a vaccination record. No vaccinations were found for HPV and Herpes zoster virus.
Depression was very poorly screened according to the patient's file (1%).
Optimal screening was associated with the patient's age and the physician's university setting for all comorbidities and risk factors (p < 0.001). Moreover, it was associated with disease duration (p 0.017) and social coverage (p 0.036) for gynecologic cancer screening, with female gender for DXA prescription (p < 0.001) and with biological treatment (p < 0.001) and corticosteroid therapy (p 0.008) for adequate influenza and pneumococcus vaccination.

Discussion
This is the first prevalence study assessing comorbidities and risk factors and their screening in a nationwide population of RMD (CIRD and osteoarthritis simultaneously) in Lebanon.
prevalence of comorbidities. In our study, the main risk factors and comorbidities followed three main prevalence axes: cardiovascular, osteoporosis and depression.
Cardiovascular risk factors were the most prevalent with 36.5% of hypertension, 30.7% of hypercholesterolemia, 22.7% of obesity, 22.1% of smoking, and 10.4% of diabetes. These risk factors were statistically different across diseases, with the highest prevalence in the osteoarthritis arm, due to the age difference most evidently. Our numbers were consistent with the WHO and Lebanese Ministry of Health (MOH) reports for hypertension (28.8% to 41.3%), hyperlipidemia (32%), but slightly lower for smoking (32% (WHO) to 38.8% (MOH)), diabetes (18%) and obesity (27.4%) than the general Lebanese population aged 50 years and above 2,53,54 . We had a similar rate of myocardial infarction and stroke 53,55 .
Osteoporosis was the second most prevalent comorbidity axe, with a prevalence of 20.7% as identified by a FRAX score >10% and of 10.8% as identified by osteoporotic fractures. This number is slightly higher than the reported 13% osteoporotic bone mineral density at the femoral neck reported in the Lebanese patients aged 50-79 yo 56 .
The third axe of comorbidity was depression, with 18.1% of patients detected by the questionnaire, slightly more than the patients treated with anti-depressant drugs (14.4%). Depression was higher in SLE, and lowest in axSpA. It seemed to be higher than the previously reported 9.9% prevalence in a Lebanese cross-sectional study 57 and was associated with female gender as in the Portugese National Health multimorbidity Survey 6 .
Malignant neoplasms were identified in around 4% of patients and seemed to be lower than in the general Lebanese population, although direct comparison cannot be made. However, malignant neoplasms site proportions seemed to be respected 58 .
High percentage of smokers in our COMORD population could reflect the high smoking prevalence found in the Lebanese general population. On the other hand, higher depression rates could be attributed to the legacy of several brutal wars in Lebanon which has led to chronic psychological distress. These two preventable conditions should be considered seriously, as rheumatologists should actively promote smoking cessation and depression screening, and ideally plan automatic referral pathways when these conditions are detected.
Fibromyalgia was not included in the EULAR form. Although it's prevalent in RMDs and is usually recognized by the rheumatologist, it is often not noted in the medical file and is difficult to diagnose with confidence by the medical intern and the study nurse who performed the interview. Thus, due to the difficulty of screening of such a complex disease, we preferred not to add it in our form.
Multimorbidity patterns with RA resembles the fourth pattern identified in the Portuguese National Health Survey 6 , with an clustering of RMD, osteoporosis, depression and cardiovascular risk factors (such as hypertension and diabetes), suggesting a potential synergistic negative effect on outcomes.
The higher comorbidities prevalence and multimorbidity burden found in OA compared to the other diseases, particularly axSpA, can be explained by the large age differences between the two groups. In fact, the mean age Scientific RepoRtS | (2020) 10:7683 | https://doi.org/10.1038/s41598-020-64732-8 www.nature.com/scientificreports www.nature.com/scientificreports/ was 63.6 years in OA, whereas it was 46.7 in axSpA, which is probably the main factor driving the difference in multimorbidity.
The incomplete vaccination rates are almost universally found, as well as suboptimal malignant neoplasms screening 23,35,60 . In this case, the contribution of rheumatology nurses may improve the vaccination rates and comorbidities management, as it was suggested in the nurse-led program of the COMEDRA study 61 .
Several types of bias are inherent in our study. The prevalence of some comorbidities may be overestimated because of diagnostic bias, in that patients with RMD closely followed at a rheumatology practice may be offered more screening for comorbidities known to be associated with their RMD. On the other hand, the prevalence of some other comorbidities may be underestimated because of selection bias, as patients with life-threatening conditions as malignant neoplasms or severe myocardial infarction or stroke, may have been unable to participate in the study. Another bias is the lack of direct comparison general population arm. This comparison arm however is difficult to find, since patients recruited at primary care clinics may be offered higher screening than the general population, just for the reason of being medically followed. Finally, we didn't use common validated comorbidity indices, because they are usually validated only in one type of RMD, mostly RA 62,63 . Instead, we preferred using the EULAR points to consider as they cover all RMD simultaneously.
A main limitation of our study is the small sample size, which reflects the small general population size (around 4 million inhabitants) and the frequent management of RMDs by non-rheumatology specialists, i.e. orthopedic surgeons, internists, family physicians… Despite these limitations, this is the first study in the region that gives an estimate of the prevalence of major comorbidities associated with RMD, using a EULAR-recommended questionnaire, and that identifies the gaps between recommended screening and actual practice. Including an OA arm in our study is considered as an added value as it reflects the local large spectrum rheumatology practice covering both mechanical and inflammatory RMDs. It can be viewed as a control arm for inflammatory RMDs,. Moreover, the inclusion of the polyarticular form is recommended by the EULAR comorbidity taskforce. We had no differences between the different Lebanese practices, and the estimated prevalence were relatively consistent with the figures published internationally, which gives the study good external validity, despite the low sample size. Since the file review was complemented by a face-to-face interview, there was no missing data, except for childhood vaccination which was vaguely remembered and impossible to retrieve due to the absence of childhood vaccination records.

conclusion
Comorbidities have a high prevalence in RMD and should be considered in the patient's routine work-up and integrated in a holistic approach to the patient. Although cardiovascular screening seems satisfactory, major efforts should be made to promote smoking cessation, improve osteoporosis, malignant neoplasms and depression screening and implement adequate vaccination. The current study identified health priorities, serving as a framework for the implementation of future comorbidity management standardized programs, led by the rheumatologist and coordinated by specialized health care professionals.