Knee osteoarthritis (OA) is a common and disabling condition driven by chronic systemic inflammation (1), leading to joint pain (2), functional limitations (3), and reduced health-related quality of life (HRQOL) (4). It is often accompanied by impaired sleep quality (5), fatigue (6) and depression (7), which exacerbate the overall disease burden. While current treatments focus primarily on symptom management, emerging evidence suggests that dietary interventions may offer modest benefits on physical HRQOL (8). Knee OA outcomes also differ by sex, with females generally reporting worse symptoms and lower HRQOL (9), while males report greater cognitive and mental decline at similar disease stages (10). These disparities may be partly explained by differences in inflammatory responses, as females tend to have higher levels of pro-inflammatory markers and estrogen receptor activity, contributing to increased inflammation and disability (9).

Diet can modulate systemic inflammation by exerting either anti-inflammatory or pro-inflammatory effects (11–13). Anti-inflammatory diets have been linked with greater HRQOL in healthy adults (14), type 2 diabetes (15, 16) and rheumatoid arthritis (17). However, whether similar benefits extend to individuals with knee OA remains unclear. Likewise, while dietary inflammatory potential has been associated with psychological distress (18) and poor sleep quality in healthy adults (19), evidence in chronic conditions, including knee OA, is limited. Exploring these connections in those with knee OA may provide valuable insights into the broader impact of dietary inflammation on wellbeing and inform targeted dietary strategies to enhance quality of life beyond joint health.

The inflammatory potential of diets can be assessed using the dietary inflammatory index (DII^®) (20) and the more recently developed energy-adjusted DII (E-DII™) (21), which accounts for the confounding effects of total energy intake (20). Our previous work found no associations between diet-related inflammation and knee pain characteristics (22). However, the link between dietary inflammation potential and HRQOL in knee OA remains largely unexplored (8).

Given the high burden of knee OA and its impact beyond the joint, it is critical to understand whether diet-associated inflammation influences these outcomes and if these relationships differ by sex, which may have implications for treatment approaches. This study aimed to assess the sex-specific associations between dietary inflammatory potential and HRQOL, sleep quality, energy and fatigue levels, and psychological distress in adults with symptomatic knee OA.

Of the 144 participants, 92 (64%) were female, and 52 (36%) were male. Females had mean ± SD age 65 ± 8 years, BMI 30.2 ± 7.2 kg/m², and 80% were classified as overweight/obese (Table 1). Male participants had mean age 67 ± 8 years and BMI 30.4 ± 3.8 kg/m², with the majority (96%) classified as overweight/obese. The mean E-DII was −0.35 ± 1.34 in females and −0.26 ± 1.52 in males, while mean DII was 0.72 ± 1.49 and 0.33 ± 1.45, respectively.

Higher E-DII scores were associated with lower energy levels in males (unadjusted β = −3.47, 95% CI −7.41 to 0.46; adjusted β = −4.01, 95% CI −7.91 to −0.30) (Table 2). Higher E-DII scores were also associated with lower EQ-5D utility scores in females (unadjusted β = −0.03, 95% CI −0.05 to −0.001), but not after adjustment (β = −0.02, 95% CI −0.04 to 0.01). DII scores were negatively associated with energy levels in males (unadjusted β = −4.34, 95% CI −8.44 to −0.23), and this association remained after adjustment (β = 0.76, 95% CI −2.12 to 3.64) (Table 3). No associations were found between either the E-DII or DII and EQ-5D VAS, self-reported sleep quality, fatigue levels, or psychological distress (Supplementary Figures 1, 2).

This cross-sectional study is the first to explore the association between dietary inflammatory potential and key quality of life outcomes in males and females with symptomatic knee OA. Higher E-DII scores (reflecting more pro-inflammatory diets) were associated with lower energy levels in males. Similarly, higher DII scores were associated with lower energy levels in males, but given the paucity of previous research in this area, the clinical significance of these findings are unknown. No associations were found between DII and HRQOL, sleep quality, fatigue, or psychological distress in either sex. These findings are partly consistent with our previous work (22), which found no evidence of an association between dietary inflammatory potential and knee-specific outcomes, medical comorbidities, and symptomatic musculoskeletal sites (22). Notably, the inverse association between E-DII and EQ-5D utility scores observed in females was attenuated after adjusting for age and BMI. This highlights the importance of considering demographic and anthropometric factors when interpreting diet-health relationships and may partly explain the lack of consistent sex-specific findings.

There is limited evidence in knee OA populations for direct comparison with our findings. A study by Toopchizadeh et al. (34) conducted in a slightly larger cohort of knee OA patients (n = 220) reported an association between higher DII tertiles and lower HRQOL as assessed by the SF-36 instrument. A cross-sectional analysis conducted in a healthy population found that a pro-inflammatory diet (mean E-DII −1.28 ± 1.51) was associated with greater risk of worse mental health outcomes, including anxiety, psychological wellbeing and depressive symptoms (18), particularly in females, whereas no significant associations were found in males (18). The overall diets in that study were more inflammatory than the FEAST cohort, which had a mean E-DII score −0.31 ± 1.40. Our recent meta-analysis concluded that anti-inflammatory dietary interventions result in small improvements in physical component scores of HRQOL but not mental component scores of HRQOL (8). While we used the Kessler instrument to assess psychological distress, our current study aligns in part with our review as we found no relationship between dietary inflammatory potential and mental health.

While chronic systemic inflammation is strongly tied to sleep disturbances (35, 36), evidence specifically linking dietary inflammation to sleep remains inconsistent (37). Higher DII scores have been linked with poorer sleep in females with overweight and obesity (38), while other studies conducted in healthy adults (39–41) and individuals with sleep apnoea (42) reported no significant association between DII scores and sleep. A larger cohort study reported links between higher DII and poorer sleep quality, particularly in individuals with overweight or obesity, a group characterized by chronic low-grade inflammation (43). Several factors may explain the lack of associations observed in this study. Sleep is regulated by complex physiological mechanisms, with inflammation being only one of many contributing factors. Pain and depressive symptoms—both highly prevalent in those living with knee OA—may have a more pronounced impact on sleep quality than dietary inflammation (5, 7). Additionally, the use of self-reported sleep and dietary intake data introduces the potential for measurement bias (44). Importantly, while the E-DII is a validated tool for estimating dietary inflammatory potential, it may not capture other aspects of dietary behavior that could influence sleep, energy, and fatigue, such as meal timing (45) and specific nutrient profiles (46). The narrow range of DII and E-DII scores observed in the FEAST cohort may also have limited our ability to detect associations, particularly when compared with studies that included individuals with more proinflammatory diets. Furthermore, other factors such as physical limitations (47) and mental health status (48) appear to be strongly linked to fatigue rather than systemic inflammation alone in knee OA. This supports the view that fatigue is a multifactorial construct that may not be fully captured by single-item scales like a VAS (49).

In addition to improved disease-specific outcomes, a reported benefit of lowering dietary inflammatory potential is improved energy levels and less fatigue (50). We found that higher dietary inflammatory potential was linked to lower energy levels in male participants, but no associations with fatigue levels in either sex. This sex-specific association between a more proinflammatory diet and reduced energy levels in males is consistent with previous research conducted in healthy Australian adults, which reported associations between diet quality (assessed using the dietary guideline index and recommended food score) and energy levels in men (51). While these indices reflect overall diet quality rather than dietary inflammatory potential specifically, the overlap suggests that proinflammatory diets may lack key nutrients involved in brain function and energy metabolism, such as omega-3 fatty acids and B vitamins (52). While this suggests that anti-inflammatory dietary patterns may play a role in supporting energy levels in men with knee OA, no studies to date have established whether such changes in self-reported energy levels (as linked to dietary inflammatory potential) translate to meaningful impacts on function or quality of life. Therefore, the clinical significance of this association remains unclear. Further research is needed to confirm these sex-specific effects and understand the underlying mechanisms.

Despite well-documented sex differences in knee OA severity and symptoms (9, 53), we found few differences in the relationships with dietary inflammatory potential in males and females. Studies consistently show that females exhibit higher levels of proinflammatory markers (e.g., interleukin-6 and tumor necrosis factor-α) compared to males (54). It is possible that the effects of diet-related inflammation are overshadowed by other dominant drivers of sex differences in knee OA outcomes. For example, hormonal changes—particularly post-menopausal declines in estrogen—have been linked to knee OA incidence, pain sensitivity and inflammatory responses in females (9), partly due to estrogen’s regulatory role in immune modulation and cytokine expression. Males and females also differ in inflammatory metabolism, with females exhibiting higher circulating levels of proinflammatory cytokines, such as IL-6 and TNF-α (54). These sex-specific immune and hormonal differences may influence the body’s response to dietary inflammation, potentially explaining why associations with energy levels were observed in males but not females. Sex differences in body composition and central pain processing could also contribute to variations in knee OA severity (9).

Our results should be interpreted with caution given the exploratory nature of this analysis from baseline FEAST trial data. Limitations include the potential for our analyses to be underpowered, as the sample size calculation was based on the main trial outcome (12-week change in knee symptoms) and may lead to insufficient statistical power for the sex-specific associations. While the EQ-5D and Kessler 10 are valid tools for assessing HRQOL and psychological distress, respectively, we used a single VAS question to assess sleep quality and levels of energy and fatigue. While validated instruments for these measures exist [e.g., the Pittsburgh Sleep Quality Index (55), Multidimensional Fatigue Inventory (56)], we were constrained to a more simple measure to minimize participant burden in the larger trial. The use of validated multi-item instruments (e.g., the Pittsburgh Sleep Quality Index or Multidimensional Fatigue Inventory) in future research may improve measurement precision and construct validity. It is well-established that self-report diet data is notoriously difficult to validate (57), but all 3-day food diary data in the FEAST trial was diligently checked and clarified with participants to minimize misreporting and missing data. We were also only able to include 28 and 29 of the total 45 parameters that make up the E-DII and DII score, respectively, due to constraints of the nutrient composition databases in Foodworks, which did not contain data for several DII components. Many of the missing parameters—such as turmeric and flavonoids–were not available, limiting our ability to compute the full DII/E-DII. However, research shows that there is no change in predictive ability when at least 25 parameters are used (58). The 16 missing components are also all anti-inflammatory, and some may have a beneficial effect on knee OA symptoms, which may influence the associations observed (59, 60). In light of these shortcomings, it is important to note that the ranges of DII and E-DII scores (−2.72 to 3.75 and −3.16 to 2.90, respectively) are very narrow compared to other studies (18, 61). It is known in epidemiologic studies that a very narrow range of exposure will bias the results toward the null (62, 63). Lastly, while BMI was used as a measure of adiposity in this study, we acknowledge that assessing body composition, particularly fat mass and fat-free mass, could have provided more precise insights into the link between diet, systemic inflammation, and knee osteoarthritis. Future studies incorporating detailed body composition measures may help clarify these relationships further.

Dietary inflammatory potential, measured by the E-DII and DII, was generally not associated with HRQOL, self-perceived health indicators of sleep, fatigue, or psychological distress in males or females with symptomatic knee OA. A higher proinflammatory diet score was associated with lower energy levels in males but given the lack of an established minimum clinically important difference (MCID) for single-item energy VAS measures in knee OA populations, the clinical significance of this association is currently uncertain. The results of the FEAST trial will reveal if reducing inflammatory potential of diet improves these important health outcomes longitudinally.