The local Medical Ethics Committee of Fasa University of Medical Sciences (FUMS) approved the protocol of this clinical trial (Code: IR. FUMS.REC.1394.39). Additionally, the study protocol was registered on the Iranian Registry of Clinical Trials website (IRCT2017081711341N7). This investigation was carried out in accordance with the guidelines of the Declaration of Helsinki (1989 revision) and the IAHAIO Tokyo Declaration. Patients were informed about the trial procedure, and if there was consent to participate in the study, they signed the relevant form.

Patients diagnosed with OA considering the criteria of the American College of Rheumatology (Altman et al., 1986) and Kellgren-Lawrence Grading Scale (Kellgren and Lawrence, 1957) (grade I to III), who were between 30 and 70 years old., and had experienced knee pain for 3 months or more were included in this research. Exclusion criteria were pregnancy, lactation, obesity (BMI≥35 kg/m^2), and secondary OA (e.g., metabolic arthritis, infected arthritis, gout, pseudo gout, and traumatic arthritis). Individuals with a past medical history of joint replacement surgery, inflammatory bowel disease, peptic ulcer, knee skin lesions, or drug abuse were also excluded. Additionally, patients who had used topical or oral analgesic medication within 3 days prior to the study, used glucosamine or chondroitin sulfate, received corticosteroids or hyaluronic acid injections in their knees during the last 3 months, consumed oral or topical corticosteroids within 14 days before enrollment, or underwent any physical modalities (such as transcutaneous electrical nerve stimulator, acupuncture, and physiotherapy) within 2 weeks prior to evaluation were not eligible to participate.

To prepare the traditional formulation of basil oil, the required herbs including fresh basil leaves and sesame seeds were purchased from an herbal shop. The samples of them were transferred to the Department of Pharmacognosy, Faculty of Pharmacy, Kerman University of Medical Sciences and voucher codes was assigned for basil (KF1368) and sesame (KF1674).

Basil leaves were pounded and pressed them to extract the juice. The juice extracted from basil leaves was combined with cold press sesame (Sesamum indicum L.) oil in a weight proportion of 2:1. It was heated gently to evaporate the water content, leaving behind the oil.

The basil oil fatty acid compound was analyzed via gas chromatography (GC)-flame ionization detector (FID) after methyl ester derivatization. Methyl ester derivatives of fatty acids were obtained based on the guideline provided in ISO 12966-2:2017 (https://www.iso.org/standard/72142.html). To identify the fatty acids, we compared their retention times with those of standard compounds injected into the GC–FID system immediately after sample injection with the same protocol.

This prospective randomized double-arm, double-blind active-controlled clinical trial was carried out to compare the efficacy of basil oil versus diclofenac sodium gel (%1) in individuals suffering from knee OA. The allocation ratio was considered 1:1, and the study method remained unchanged after the approval of the proposal.

This study was carried out in the orthopedic clinic affiliated with FUMS, Fasa. Patients were initially evaluated by an orthopedic specialist and then introduced to the researchers upon clinical diagnosis of knee OA. Eligible cases were invited to take part in the study. Patient recruitment occurred between September 2018 and April 2019.

All patients who were diagnosed with osteoporosis based on the diagnosis of an orthopedic specialist and visited the aforementioned clinic underwent simple sampling. Given the inclusion and exclusion criteria, 100 knee OA subjects were selected, randomized and allocated to either the control or intervention groups. A biostatistician used the blocked randomization method (non-stratified, with four participants in each block) by Microsoft Excel^® software to generate the randomization list for this trial. To ensure allocation concealment, opaque and sealed packets containing numbers “1” and “2” were prepared. A clinic secretary allocated the participants to the study groups by sequentially selecting packets with the corresponding numbers.

Both basil oil and diclofenac gel were prepared in identical containers with similar characteristics. The contents of the drugs were undisclosed to the patients, physicians, and outcome assessors. A third person, who had no role in the study, gave the drugs to the participants.

At first, the researchers obtained the written informed consent from the participants; then, they were randomized to either the diclofenac sodium gel 1% (control) or basil oil (intervention) groups. They received basil oil or diclofenac gel (Sobhan Darou) to apply on the knee (1.5 cc or one fingertip unit, respectively) thrice daily over a period of 4 weeks. All participants were provided with capsules containing 100 mg of celecoxib (Tehran Darou); to the maximum allowed daily dose of celecoxib for pain relief was 200 mg.

In our clinical trial, compliance was assessed through a combination of patient self-reporting and monitoring during follow-up visits. Patients were provided with a daily diary to record their usage. During follow-up visits, patients were asked about their adherence to the treatment regimen, and any deviations or issues with compliance were documented. Additionally, any unused product was collected and accounted for at each visit to ensure that patients were not using the product more frequently than recommended. This approach allowed us to monitor and track patient compliance throughout the study duration. Additionally, participants were advised to maintain their usual lifestyle and dietary habits to minimize potential fluctuations in weight that could impact the study outcomes.

The primary outcome of this investigation was measured using a validated questionnaire known as the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) score (Bellamy et al., 1988).

The secondary outcomes were assessed as follows.

- Measurement of knee stiffness and physical function using the WOMAC questionnaire

Additionally, adverse events were elicited and assessed through a structured approach, during follow-up visits. The monitoring process involved the use of a checklist comprising a series of inquiries pertaining to different physiological systems, encompassing the skin, cardiovascular, nervous, respiratory, and gastrointestinal systems.

Based on a study by Lei et al. (Lei et al., 2017), knee problems measured by the WOMAC scale decreased from 43.7 ± 12.7 to 22.5 ± 10.3 over 6 months. Given a power of 90% and a 5% error rate, a minimum of 43 individuals in each group were needed to obtain an 8-unit reduction in the WOMAC index. Taking into account the possibility of participant dropouts, we initiated the trial with a 20% larger sample size of 50 individuals in each group. n = s 1 2 + s 2 2 x 1 ¯ − x 2 ¯ 2 Z 1 − α 2 + Z 1 − β 2 s1 = 12.7, s2 = 10.3, α = 0.05, Z1-α/2 = 1.96, β = 0.1, Z1-β = 1.28.

The mean and standard deviation (SD) were used to present the data. The chi-square test (sex, occupation) and t-test (age, BMI, disease duration) were applied to compare the baseline variable between the two groups.

Several factors were evaluated for each patient, including two WOMAC sub-scores (physical function and stiffness), walking speed, and analgesic consumption. The comparison of these factors at three different time points was conducted using Repeated Measurement ANOVA, followed by Bonferroni post hoc test within each group. The variables were compared between the study groups using t-tests at baseline and ANCOVA at 2 and 4 weeks. The latter analysis was adjusted for baseline measurements as a covariate. For variables assessed individually for each knee (such as knee flexion angle and VAS), as well as the total score of the WOMAC questionnaire and pain, a generalized estimate equation (GEE) model with an unstructured correlation matrix was employed for data analysis. GEE is an approach that accounts for correlated data without fully specifying the distribution of responses within each patient as a cluster (Hanley et al., 2003). The analysis focused on examining the variation of the mentioned variables separately for each group across three different evaluation times. In this analysis, the correlation between the three evaluation times and the two knees of each patient was taken into account. Another GEE model was used to compare the baseline characteristics between the study groups, treating the two knees as the correlated and the study group as the fixed factors. Two additional GEE models were employed to compare the study groups at 2 and 4 weeks after the intervention. In these models, the correlation between the knees was considered, with baseline measurements as the covariates and the study group as the fixed factor.

All of the statistical analyses were done using IBM SPSS 24 (IBM Corp. Released 2016. IBM SPSS Statistics for Windows, Version 24.0. Armonk, NY: IBM Corp.). For drawing the charts Microsoft Excel (Microsoft Co, Redmond, USA) was used. A p-value less than 0.05 was considered as the significant level.

Following the methylation of the basil oil, GC–FID was employed to determine and identify the oil fatty acids compounds. The identified fatty acids are presented in Table 1. The predominant fatty acid detected was linoleic acid (42.42%), followed by oleic acid (40.83%) and palmitic acid (11.98%).

The eligibility of a total of 146 patients was evaluated from September 2018 to April 2019. After applying the inclusion criteria, we selected 100 eligible subjects for this trial and randomly allocated them to the intervention or control groups, with 50 cases in each group. Throughout the study, seven patients from the control group and seven cases from the intervention group were excluded because of such reasons as loss of follow-up, application of topical corticosteroids, or improper use of diclofenac gel or basil oil. Ultimately, 86 participants successfully completed the trial. The baseline demographic information is indicated in Table 2, which shows no significant differences in the measured clinical or demographic factors between the two groups at the beginning of the trial. Figure 1 displays the trial flowchart.

The participants in each group were followed for a duration of 4 weeks starting from the initiation of the intervention. The outcomes were assessed at beginning of the study, 2, and 4 weeks after starting the intervention. No statistically significant differences were seen between the intervention and control groups regarding pain severity based on the VAS scale (p values: 0.619 and 0.880), knee flexion angle (p values: 0.993 and 0.473), 8-m walking test (p values: 0.436 and 0.555), number of analgesic consumption (p values: 0.127 and 0.117), pain severity based on WOMAC questionnaire score (p values: 0.642 and 0.887), physical function based on WOMAC questionnaire score (p values: 0.744 and 0.239), knee stiffness (p values: 0.202 and 0.658), and total WOMAC scores (p values: 0.970 and 0.354) at the 2-week and 4-week intervals, respectively.

However, the analysis within each group showed significant differences in most variables, including pain severity based on the VAS scale or WOMAC questionnaire scores, physical function, knee stiffness, and total WOMAC scores when comparing baseline and the second week. Similarly, within-group analysis showed significant differences in the outcomes such as pain severity based on the VAS scale or WOMAC questionnaire scores, physical function, knee stiffness, total WOMAC scores, and knee flexion angle when comparing baseline and the fourth week (Figure 2; Table 3, 4).

All patients tolerated basil oil well, and the oil did not induce any adverse event in any of the participants.

The current trial had certain methodological limitations. The lack of a placebo control group is indeed a limitation of our study. While our study design focused on comparing the efficacy of topical basil oil to an active control group, we recognize the value of including a placebo group for a more comprehensive assessment of treatment effects. Future studies could benefit from the inclusion of a placebo control group to provide a clearer understanding of the specific benefits attributed to the treatment intervention versus the natural course of the disease. Furthermore, the inclusion of sesame oil as the carrier for basil oil raises valid concerns regarding the potential influence of sesame oil on the observed outcomes. Recent research has highlighted the beneficial effects of sesame oil on osteoarthritis, particularly due to the chondroprotective properties of sesamine, the primary phytochemical compound found in sesame (Phitak et al., 2012; Askari et al., 2019). Nevertheless, the primary objective of our trial was to evaluate the efficacy of the traditional basil formulation as prescribed in Persian medicine texts for knee osteoarthritis. Therefore, sesame oil was chosen as the carrier for basil oil in order to maintain the authenticity of the traditional formula being studied.

Another limitation is the age range of the individuals who participated in the investigation as most of them were middle-aged or elderly. Polypharmacy, which is common in this age group, may have affected the results. Ethical concerns prevented us from forbidding the elderly participants from taking their medications, which is another limitation. One more limitation of this study is the potential for unblinding due to the differing consistencies and aromas of the basil oil and diclofenac gel treatments. Despite efforts to maintain blinding with opaque containers and participant instructions, the distinct sensory characteristics of the treatments may have influenced patient perceptions and treatment expectations. Patient expectations are known to impact treatment outcomes, and the potential for unblinding could have introduced bias in the results. While steps were taken to minimize this effect, the inherent differences between the treatments remain a limitation of the study. Furthermore, OA is a chronic disorder, and the short follow-up duration in this trial can be considered a limitation. Long-term medication is typically required for patients with OA, so conducting studies with longer follow-up durations may yield different results regarding efficacy and safety. The gender proportion of participants is also worth noting as a limitation. Approximately 84% of the subjects were female, so the applicability of the results to male patients with knee OA may be restricted.