Patients eligible for the study were males and females aged ≥40 years with at least 1 year of being diagnosed with OA of the knee. According with the criteria of the American College of Rheumatology, OA was confirmed with physical exploration diagnosis of knee pain plus one of the following: bone crackling with movement, morning stiffness ≤15 min, age > 50 years or articular hypertrophy. And grade II-IV knee OA was confirmed by radiology according to the Kellgren-Lawrence grading system [12,13].
Patients diagnosed with grade I were referred for a less invasive treatment and patients with grade IV were included to evaluate the treatment effect in OA most advanced conditions. Three physicians examined and diagnosed the recruited patients, two were orthopedic specialist and one was surgeon specialized in radiology. Exclusion criteria included: An intra-articular injection of any substance administered within the last 3 months, joint inflammatory diseases, microcrystalline arthropathies, current pregnancy, uncontrolled hypertension, active infection, undergone surgery/arthroscopy within the last 3 months and diagnosis by radiography of knee OA Kellgren and Lawrence grade I, coagulation or platelet disorders or any concomitant disease that could interfere with the evaluation,. Patients were enrolled by public advertising and were studied at the San José Hospital in Queretaro, México.
The study was conducted in compliance with ICH (International Committee of Harmonization) Good Clinical Practices and the Declaration of Helsinki, and its applicable amendments. It was approved by the institutional review board for human research of the University of Querétaro and all subjects voluntarily signed informed consent before being enrolled in the study.
A total of 26 patients per experimental group were necessary to detect a clinically significant change of 20% within experimental treatments, from baseline to post treatment evaluations in the global score of Lequesne Index and Western Ontario and McMastern University Index (WOMAC) indexes considering an estimated standard deviation of 35%, a two-sided alpha level of 0.05 and a statistical power of 0.8. Considering a possible drop-out rate of 35%, 72 patients had to be recruited. In addition, the mentioned sample size can find a significant difference of 30% between two experimental groups’ changes in the global score of Lequesne and WOMAC. The parameters used to calculate the sample size were estimated from the results of a pilot study previously carried out with 18 participants.
The study was a randomized, double-blind, parallel-design clinical trial. Patients were followed up during 12 months with treatments and for 6 months after intervention to evaluate a longer term effect and safety of the treatment. Subjects who met the selection criteria were randomly assigned to receive one of two treatments in both knees (except for 2 patients with prosthesis only one knee was intervened): 1) Sodium bicarbonate and calcium gluconate (SBCG1) 2) Sodium bicarbonate and a double dose of calcium gluconate (SBCG2). The randomization method was based on a list of randomly assorted treatments, generated by using an online computing program . One researcher that had no direct contact with patients created the randomization list and delivered it to each physician that examined the patients who were assigned to each physician in a systematic order. The treatment was administered to each patient every 30 days by one of the 3 trained and experienced physicians who followed the same patients throughout the study. In the baseline evaluation and in the monthly scheduled visits during intervention period and post-intervention follow-up, patients were clinically evaluated with the WOMAC and Lequesne questionnaires and also monitored for adverse events. All the fieldwork personnel were blinded to the treatments.
Knee radiographs were taken to each patient at baseline, after 3 months, after 12 months of intervention and also after 6 months of post-intervention follow up. Patients were not permitted to receive concomitant treatment with analgesic or systemic corticosteroids.
The treatments were identical in packaging, labeling, schedule of administration and appearance. Both treatments, SBCG1 and SBCG2 were pharmaceutical compositions in aqueous solution (5 mL) ready for intra-articular infrapatellar injection. They were prepared at the pilot plant of the Center for Research and Technological Development in Chronic Disease (Cindetec A.C.). The SBCG1 treatment was a solution with sodium bicarbonate and calcium gluconate, both at a concentration of 7.5% (w/v); and SBCG2 had the same concentration of sodium bicarbonate than SBCG1, while the concentration of calcium gluconate was 15% (w/v).
The primary efficacy outcome variables were the changes from baseline to post-intervention assessments in the WOMAC, and Lequesne’s functional indexes and joint space width.
The WOMAC OA index is a validated, multidimensional, disease specific, health status measure. It provides with clinically important patient relevant symptoms in the areas of pain, stiffness and physical function in patients with hip and/or knee OA. It consists of 24 questions in three separated subscales: pain, physical function and stiffness. Each subscale score weighed 10 points, and the WOMAC global scale is the sum of the three subscales and ranges from 0 to 30 . The Lequesne index is a 10-question interview-style survey that includes evaluation of pain or discomfort, maximum walking distance and daily activities performance. The total questionnaire is scored on a 24 points scale .
Joint space width changes were assessed on radiographs that followed standardized techniques . At each time point 4 knee radiographs were obtained per patient, to get from each knee a standing anterior-posterior and lateral views. All radiographs were taken by the same technician and were interpreted by 2 independent physicians who were blinded to treatment, both with certifications in orthopedics and traumatology. The outcome measure was the progression of joint space narrowing as suggested by Abadie et al. . One independent collaborator, blinded to treatment measured joint space width in radiographs by visual reading according to a validated method  at the joint’s narrowest point using an X10 magnifying lens graduated at 0.1 mm intervals.
The safety of experimental treatments was assessed with reported adverse events, defined as any unfavorable and unintended sign, symptom, or disease temporary present during the intervention period, whether or not related to the treatment. Adverse events were either reported by the patient or discovered by physicians during visits. Adverse events were classified into light, moderate, serious and lethal according to the Mexican Official Norm for Drug Surveillance . In addition patients were monitored with laboratory tests (hematology, biochemistry and urine analysis) at baseline, 12 and 18 months.
Baseline demographic variables and treatments’ compliance were analyzed with the chi square test. The collected data were analyzed in an intention-to-treat basis with subjects that received the allocated intervention and attended to at least one post-treatment evaluation. The WOMAC, and Lequesnes´s functional indexes’ mean values were calculated for each evaluation and a Student T test was used to evaluate the significance of changes at each post-treatment evaluation compared with baseline evaluation. Mean changes between both treatments at each evaluation were compared with ANOVA.
In the radiographies' analysis, the narrowest joint space from both measurements in each knee was selected for analysis. Mean values were calculated for both knees joint spaces and a paired T-test was performed to evaluate joint space change for each treatment. To evaluate the response between treatments, a Generalized Estimating Equation model was performed to evaluate joint space changes, this model considered the correlation between each subject’s knees and the baseline values as a covariate.
To evaluate safety of the treatment, adverse events were classified by seriousness and were quantified to get the incidence of each adverse event.
All statistical models were tested at the 0.05 level of significance. Analyses were performed with SPSS® for Windows version 18.0 (IBM®).