We analyzed data from EpiReumaPt, a national cross-sectional, population-based study with a representative sample of the Portuguese population that aimed to analyze the burden of rheumatic and musculoskeletal diseases (RMDs) in Portugal. As described in detail elsewhere (16), participant recruitment was conducted between September 2011 and December 2013 using a random selection of private households in Portugal stratified by administrative territorial units (NUTS II: Norte, Centro, Lisboa and Vale do Tejo, Alentejo, Algarve, Azores, and Madeira) and the size of the population within each locality. In each household, an individual ≥18 years old with permanent residence and the most recently celebrated birthday was selected to participate in the study. In total, 28,502 households were contacted, 8,041 individuals refused to participate, and 10,661 were included in the study. The EpiReumaPt population was similar to the Portuguese population (Census 2011) in age strata, sex, and NUTII distribution (16).
EpireumaPt data collection was performed using a three-staged approach. In the first stage, participants completed a face-to-face interview to collect sociodemographic and health-related information and to screen for RMDs. Interviews were conducted by a team of non-medical healthcare professionals trained for this purpose, and data were collected using a computer-assisted personal interview system. Screening was considered positive if a participant mentioned a previously known RMD, if any algorithm in the screening questionnaires was positive, or if the participant reported muscle, vertebral, or peripheral joint pain in the previous 4 weeks. When the overall performance of the RDM screening algorithm was evaluated using final diagnosis after the third stage as the gold standard, its sensitivity and specificity were 98% and 22% and positive and negative predictive value were 85% and 71%, respectively.
In the second stage, participants who screened positive for at least one RMD (n=7,451) and approximately 20% of participants who screened negative for RMDs (n=701) were invited to a clinical appointment at the primary care center of the participant’s neighborhood. Participants were seen by a multidisciplinary team consisting of a rheumatologist, X-ray technician, and nurse. Clinical assessment consisted of a structured evaluation, laboratory tests, and imaging exams, if needed, to establish a diagnosis and evaluate disease-related information. According to participants’ complaints, simple X-rays were performed in 122 hips and 479 knees, among other joints. Rheumatologists were blind to prior health-related data. Of the participants in the second stage, 4,275 did not attend the clinical appointment. Therefore, at the end of the second stage, there were 3,877 clinical observations: 3,198 participants received an RMD diagnosis, and 679 did not receive an RMD diagnosis.
In the third stage, three experienced rheumatologists reviewed all data and validated the RMD diagnosis. Diagnostic agreement among the three rheumatologists was 98.3%, with a Cohen’s K coefficient of 10.87 (95% confidence interval (CI): 0.83, 0.91) (16). When data were insufficient to fulfill international classification criteria for an RMD, five rheumatologists met to reach agreement on the final diagnosis. When doubts persisted, the opinion of the rheumatologist who performed the clinical assessment in the second stage prevailed. A total of 1,087 participants had a validated diagnosis of HKOA, 199 had a validated diagnosis of hip OA, and 981 had a validated diagnosis of knee OA (Figure 1).
--------------------------------------Figure 1 should be placed here----------------------------------
This study included participants of EpiReumaPt with validated a diagnosis of HKOA according to American College of Rheumatology criteria (17,18).
Case definition and measurement
Mean pain intensity in the previous week, measured on a 11-point numeric pain rating scale in the second stage of EpiReumaPt, was used to categorize participants with HKOA into APR (<5 points) and IPR (≥5 points), which was validated by Zelman et al. (2003) using the question 5 of Brief Pain Inventory scale, as the average pain in the previous week on an 11-point NPRS. The optimal cut-off point found for manageable day pain in OA was 5 (F(7, 9)=7.08, p<0.001)(19). When both the hip and knee were affected, the worst score was considered.
Sociodemographic, clinical, and lifestyle variables
Sociodemographic, clinical, and lifestyle variables were collected during the first and second phases of EpiReumaPt. To assure better clinical interpretation, some variables were subjected to categorical transformation.
Sociodemographic and anthropometric variables
Sociodemographic variables were age, sex, and geographic location according to NUTS II territorial units. Madeira and Azores were merged in the analysis as the Islands region. Marital status was categorized as “partner” (married or consensual union) or “no partner” (single, widowed, or divorced). Education level was categorized according to years of education completed: <4 years (less than primary education), 4-9 years (primary or secondary education), or ≥10 years (secondary or higher education).
Body mass index (BMI) was categorized as underweight (≤18.49 kg/m2), healthy weight (≥18.5 and ≤24.99 kg/m2), overweight (≥25 and ≤29.99 kg/m2), or obese (≥30 kg/m2).
Lifestyle and clinical variables
Lifestyle variables were alcohol intake (“no” or “occasionally or daily”), smoking habits (“never” and “occasionally or daily”), and regular exercise/sports (“yes” or “no”).
The number of chronic non-communicable diseases was calculated as the numeric count of the following self-reported conditions: high blood pressure, high cholesterol, cardiac disease, diabetes mellitus, chronic lung disease, problems in the digestive tract, renal colic, neurological disease, allergies, mental or psychiatric illness, cancer, thyroid or parathyroid problems, hypogonadism, and hyperuricemia. Multimorbidity was defined as having two or more chronic non-communicable diseases (20).
In addition to pain intensity, other clinical variables were considered: performance in ADL, QoL, and the presence of depression and/or anxiety symptoms. Performance in ADL and QoL related to HKOA were evaluated with the Portuguese version of the Knee Injury and Osteoarthritis Outcome Scale (KOOS) (21) and Hip Disability and Osteoarthritis Outcome Scale (HOOS) (22). These self-reported outcome measures evaluate short- and long-term consequences of HKOA in five dimensions: pain, symptoms, ADL, sports and leisure, and QoL. For this study, we used only the HOOS/KOOS ADL and HOOS/KOOS QoL subscales. Scores for each dimension were transformed on a 0-100 scale, with 0 representing extreme hip/knee problems and 100 representing no hip/knee problems (21,22). For both subscales, if more than one joint was affected, the worst score was considered.
Anxiety and depression symptoms were evaluated using the Hospital Anxiety and Depression Scale subscales for depression (HADS-D) and anxiety (HADS-A). Both scales have a range of 0 to 21, with higher values representing more severe symptoms of anxiety or depression. Final HADS-A and HADS-D scores were categorized using validated cut-offs as: “with anxiety” (HADS-A ≥11) or “without anxiety” (HADS-A <11) and “with depression” (HADS-D ≥11) or “without depression” (HADS-D <11) (23).
Use of therapies
Information on pharmacological therapies, defined as daily medications taken in the previous month, was collected in the first-stage interviews. Medication for pain relief was classified according to the Anatomical Therapeutic Chemical Classification System as: glucosamine (M01AX05); analgesics/antipyretics (N02B), specifically paracetamol (N02BE01); simple (N02A) and combined (N02AJ) opioids; non-steroidal anti-inflammatory drugs (NSAIDs; M01A); and topical agents (M02A). Information on physiotherapy attendance in the previous 12 months, was also collected in the first-stage interviews. Information on previous hip or knee surgery was collected during the second-stage clinical appointments, which occurred no more than 15 days after the first stage.
After participants were categorized into IPR and APR subgroups, weighted proportions of participants with IPR were computed taking sampling design into account as described elsewhere (16). The logit transformation method was used to calculate 95% CIs. Analysis of the proportion of participants with IPR and APR (relative and absolute frequencies) was conducted separately for participants with hip OA and those with knee OA.
Descriptive statistics were used to characterize all participants and, separately, the APR and IPR subgroups, according to sociodemographic, lifestyle, and health-related variables as well as use of therapy. Differences between subgroups were analyzed using independent samples t-tests for continuous variables and Chi-square tests for categorical variables.
We first analyzed associations between sociodemographic, lifestyle, and health-related variables and pain relief. Variables with p<0.25 were included in a univariate logistic regression model in a forward selection process (24) to avoid early exclusion of potentially important variables (Additional File 1). Variables with p<0.05 were then kept in a backward selection process to construct a multivariable model.
We next analyzed associations between IPR and clinical outcomes. Associations between IPR and HOOS/KOOS ADL and QoL subscale scores were analyzed using linear regression models adjusted for the variables retained in the multivariable model. Associations between IPR and the presence of anxiety and depression symptoms were analyzed using logistic regression models adjusted for the same variables.
Given the scarcity of data, normal and underweight BMI categories were merged into a single category (<25 kg/m2).
All analyses were weighted and performed with SPSS 26 complex samples for MacOS (IBM Corp., Armonk, NY, USA). Statistical significance was defined as p<0.05.