Setting and study population
The Early Arthritis Clinic of the Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán was established in February 2004. Patients referred to the clinic had suspicion of RA and symptom duration of less than 12 months. Patients with confirmed recent-onset RA also attend the clinic; in such patients, the time from diagnosis to first evaluation is ≤5 weeks.
Patients received treat-to-target-oriented treatment. Traditional disease-modifying anti-rheumatic drugs (DMARDs) were used in 99% of our population, with or without corticosteroids (20 to 40%).
At baseline, a complete medical history and sociodemographic characteristics were obtained. Blood samples after 9 to 12 hours fast included at least serum glucose (GLU), total cholesterol, triglycerides (TG), high-density lipoprotein-cholesterol (C-HDL), low-density lipoprotein-cholesterol, uric acid, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), rheumatoid factor and antibodies to cyclic citrullinated proteins (ACCP). Urine analysis was also performed. For serum GLU and lipid determinations, commercially available reagents from Beckman Co. (Limerick, Ireland) for the Synchron analyzer (Laboratorios Clínicos del Instituto Nacional de Ciencias Médicas y Nutrición Salvador Zubirán, Mexico City, Distrito Federal, México) were used. The laboratory obtained ISO 9001:2008 certification (since 30 April 2004, valid from 29 April 2013 until 28 April 2016).
Standard baseline and consecutive rheumatic evaluations included, height, weight and blood pressure measurements, extended joint counts, physician and patient-reported outcomes [25,26], disease activity score evaluated in 28 joints (DAS28) , adverse events and comorbidity. Blood pressure was measured twice in each case in which initial measurement was ≥130/90 mmHg unless the patient had a previous diagnosis of systemic arterial hypertension. Measurements were performed after a 5-minute resting period. Body weight and height were performed by a trained nurse, usually on the same equipment that was regularly calibrated according to the manufacturer’s recommendations.
Patients were evaluated every 2 months during the first 2 years of follow-up, and thereafter every 2, 4 or 6 months (fixed and scheduled for all patients). At fixed 6-month intervals, analysis in order to measure serum GLU, total cholesterol, TG, C-HDL, low-density lipoprotein-cholesterol, uric acid, ESR, CRP and urine analysis were scheduled. Medical evaluations were performed by the same rheumatologist and the information was recorded on standardized formats.
RA treatment records included previous treatment (defined as prescribed during the month prior to the baseline evaluation) and current treatment (at baseline and consecutive evaluations), which included names, doses, schedule, order start and stop dates of DMARDs, of corticosteroids and of nonsteroid anti-inflammatory drugs. In addition, any other drug indicated for a comorbid condition was also recorded; for the present study, names, doses, schedule, order start and stop dates of antihypertensive, antidiabetic and lipid-lowering medications were obtained. Information regarding treatment was recorded after a direct interview on a drug record registry applied at scheduled evaluations .
Up to March 2014, charts from 162 consecutive patients with early RA were reviewed. In order to accomplish the first objective, only RA patients with completed baseline blood pressure, weight and height, and serum TG, C-HDL and GLU levels were included (two patients were excluded); there were 160 patients left.
Of the 160 patients with complete baseline data, 19 were lost to follow-up soon after the baseline evaluation (and their baseline characteristics did not differ from those patients with follow-up) and 28 additional patients had prevalent MetS. Accordingly, there were 113 patients with at least 6 months of follow-up and MetS-free baseline, in whom incidental MetS was investigated.
Finally, to accomplish the second objective related to MetS and its impact on remission, data from patients with at least 24 months of follow-up were analyzed (n = 133). This lag time was chosen based on previous reports that defined median time to sustained remission in our cohort as 13.6 (±8.8) months .
To achieve the first objective, one control for each RA patient was randomly selected from a local database that included data from more than 10,000 Mexican Mestizo adults without either known medical condition or treatment, in whom serum samples were obtained in order to investigate the prevalence of the MetS in Central Mexico. Controls were matched according to age (±5 years), gender, smoking habit (see definition below), years of education, menopause status and urban residency.
To achieve the second objective, a case–control study nested within a cohort was designed; cases consisted of RA patients with incidental MetS and controls consisted of RA patients who never developed MetS during their follow-up. Controls were matched according to gender, age, index date (as defined when MetS was diagnosed or equivalent visit in MetS-free RA patients), menopause status, presence of rheumatoid factor or ACCP, and follow-up.
Variables and definitions
Body mass index (BMI) was calculated as weight (kg)/height (m2).
Smoking status was self-reported at the baseline evaluation and categorized as never smoker, former smoker (at least one cigarette per day for at least 3 months during their lifetime but who do not currently smoke) or current smoker (at least one cigarette/day for at least 3 months).
Postmenopausal women were self-reported and defined as women ≥48 years of age who have been amenorrheic for at least 2 years, or females with bilateral oophorectomy or females with documented hysterectomy aged ≥47 years .
Fasting GLU, uric acid, total cholesterol, low-density lipoprotein-cholesterol, C-HDL and TG were measured in serum and reported in milligrams per deciliter.
Sustained remission was defined based on the DAS28 if score ≤2.6  and according to 2012 American College of Rheumatology/European League Against Rheumatism criteria  and if maintained for at least 6 months of follow-up.
Hypertension was defined if recorded on the charts, or antihypertensive medication was used, or a diastolic blood pressure ≥90 mmHg was detected or a systolic blood pressure ≥140 mmHg was detected.
Diabetes mellitus was defined if a physician diagnosis was recorded on the charts, if antidiabetic medication was recorded or if a fasting blood sugar level ≥126 mg/dl was detected.
MetS was defined according to three sets of criteria (Table 1) [22,33,34]. In all of these sets, BMI ≥30 kg/m2 was considered a surrogate of waist circumference ≥102 cm in males and ≥88 cm in females.
The study was approved by the Ethics Committee of the Instituto Nacional de Ciencias Médicas y Nutrición. Written informed consent was obtained in order to have the patient’s charts reviewed and data presented in scientific forums or published.
Distribution of each variable was analyzed. Student’s t test and the chi-square test were used for normally distributed variables and the Mann–Whitney U test for non-normally distributed variables.
To summarize cumulative outcomes (disease activity, disability) or variables (CRP, ESR), the mean of consecutive values from corresponding evaluations was obtained.
Follow-up missing data varied from 3% (for BMI) to 20% (for serum GLU levels). Imputation was calculated for linear regression method, considering an arbitrary pattern of missing values.
Prevalence of MetS was determined based on different sets of criteria applied at baseline or within 2 months from baseline evaluation. The prevalence of the MetS was compared between early RA and matched controls using Fisher’s exact test.
Logistic and Cox regression’s models were used: to evaluate the risk of MetS (defined at the baseline evaluation); to achieve remission; to identify predictors of incidental MetS; and to investigate incidental MetS as a predictor of sustained remission. In all cases, selection of variables to be included was based on their statistical significance in the bivariate analysis; a cutoff point was established based on the number of variables a priori included in order to avoid overfitting of the models; based on the number of outcomes of interest, four or five variables were included. Correlation between variables was also examined, and when significant only one variable was selected. Selection was based on variables’ clinical relevance; in all cases, correlated variables were switched into the models to test the integrity of the results. Finally, if a variable was deemed to be clinical significant, its inclusion was forced into the model (cumulative treatment).
In the first model, there were no variables with statistical significance in the univariate analysis and the covariates included a priori were age, gender and MetS at baseline evaluation.
In the second analysis, variables selected were age (correlated to menopause, r = 0.5 and to years of formal education, r = 0.3), BMI, cumulative DAS28 (correlated to cumulative ESR, r = 0.75, to cumulative CRP, r = 0.4 and to cumulative HAQ, r = 0.67) and follow-up (correlated to erosions at index date, r = 0.26), P ≤0.001 for all correlations.
In the third analysis, variables included were baseline DAS28 (correlated to ESR, r = 0.6, to HAQ, r = 0.67 and to CRP, r = 0.6), incidental MetS (correlated to BMI, r = 0.4 and to obesity, r = 0.3), disease duration at baseline, ACCP and age (P ≤0.001 for correlations); cumulative treatment was also forced into models.
All statistical tests were two-sided and evaluated at the 0.05 significance level. Statistical analysis was performed using the SPSS/PC program (v.17.0; IBM SPSS Statistics, Chicago, IL, USA).