Introduction
Overactive bladder (OAB) and nocturia are among the most prevalent lower urinary tract symptoms (LUTS), affecting individuals across all age groups [1, 2]. Their prevalence increases markedly with advancing age [3]. However, these symptoms are not limited to older adults; they are increasingly observed in younger populations, possibly driven by modern lifestyle factors such as heightened psychological stress, sedentary behavior, and irregular sleep patterns [4–6].
Clinically, OAB is defined by urinary urgency, typically accompanied by increased frequency and nocturia, with or without urgency urinary incontinence [2]. The worldwide prevalence of OAB is estimated to range from 12% to 17% in men and women, whereas nocturia—defined as waking at least twice per night to void—affects nearly one-third of adults [1, 3]. Both conditions markedly impair quality of life, disrupt sleep, and are associated with increased risks of depression, frailty, and falls in older adults [7–9].
Their management, meanwhile, imposes a significant socioeconomic burden, with annual healthcare expenditures exceeding billions of dollars in the United States alone [10].
Although OAB and nocturia frequently coexist and share overlapping risk factors, they may differ in certain underlying pathophysiological mechanisms [11]. OAB is primarily related to detrusor overactivity and altered afferent signalling [12], whereas nocturia commonly results from nocturnal polyuria, reduced bladder capacity, or impaired circadian regulation of urine production [13]. Both are strongly influenced by systemic metabolic and vascular disturbances, such as obesity, insulin resistance, and chronic inflammation, which may contribute to bladder dysfunction through oxidative stress, endothelial impairment, and autonomic dysregulation [14, 15].
These shared mechanisms have led to growing interest in modifiable lifestyle factors, particularly diet, as potential targets for prevention and management of these conditions.
Accumulating evidence suggests that dietary components can affect bladder function through multiple mechanisms. For instance, high caffeine or alcohol intake has been linked to urinary urgency and frequency [16], whereas excessive consumption of spicy foods or artificial sweeteners may exacerbate LUTS [17]. Conversely, dietary fibre and antioxidant-rich foods may alleviate urinary symptoms by improving metabolic and inflammatory profiles [18].
While such findings highlight the importance of diet, most existing studies have focused on individual nutrients or isolated foods, offering limited insight into the role of overall dietary patterns. Recent investigations of comprehensive dietary models, such as the Mediterranean or plant-based diets, have demonstrated protective effects against metabolic disorders and inflammation, which are key contributors to LUTS pathophysiology [19]. However, the relationship between other emerging dietary patterns and urinary symptoms remains insufficiently characterised.
The ketogenic diet (KD), characterised by high fat and very low carbohydrate intake, has gained global popularity as an effective intervention for weight reduction, glycaemic control, and neurological diseases [20, 21]. Indeed, KD was developed in the 1920s as a therapeutic approach for refractory epilepsy [22]. Over the decades, its use has been extended to other neurological disorders based on its ability to shift metabolism and neural excitability [23].
Ketone metabolism has been shown to enhance mitochondrial function, mitigate oxidative stress, and suppress systemic inflammation [21, 23]. These physiological adaptations may influence bladder function through enhanced endothelial stability, modulation of autonomic and afferent nerve activity, and improved metabolic resilience [24]. Nevertheless, the links between ketogenic dietary patterns and LUTS remain largely controversial, with limited direct evidence from human studies.
Excessive or prolonged adherence has been associated with elevated lipid levels, impaired renal function, and increased cardiovascular risk in susceptible individuals [25, 26]. Given these contrasting metabolic responses, it is still unclear how KD patterns influence the risk of OAB or nocturia, underscoring the need for further investigation.
To date, no large-scale population-based studies have comprehensively examined the associations between ketogenic dietary patterns and lower urinary tract symptoms, nor have they explored the potential mechanisms underlying these relationships. To address these gaps, the present study employs a comprehensive analytical framework integrating multivariate logistic regression, RCS for modeling nonlinear effects, and causal mediation analysis. In this context, we conducted a cross-sectional analysis using data from the National Health and Nutrition Examination Survey (NHANES) 2005–2018 to investigate both linear and nonlinear associations of the ketogenic diet ratio (KDR) with OAB and nocturia.
Furthermore, we assessed whether the frailty index (FI) and platelet-to-HDL-C ratio (PHR) mediated these associations, and additionally explored the relationship between KDR and nocturia across strata of physical activity level to determine whether physical activity acts as an effect modifier. This study thus aims to advance novel evidence on the metabolic and inflammatory pathways linking dietary composition to urinary symptoms, to provide lifestyle-based insights for LUTS prevention and management, and to inform more refined population-level risk stratification and targeted health guidance.
Methods
Data source and study population
The study population was derived from the National Health and Nutrition Examination Survey (NHANES), an ongoing, nationally representative, cross-sectional program conducted by the U.S. Centers for Disease Control and Prevention (CDC). NHANES employs a multistage, stratified probability design to collect standardized data on diet, health status, physical examinations, and laboratory measurements.
For this analysis, data from seven consecutive survey cycles (2005–2018) were combined. Participants were excluded if they were < 20 years old, pregnant, or lacked information on urinary outcomes, key dietary variables, or essential covariates (Fig. 1).
The final analytical sample included 22,249 adults.
All NHANES protocols were approved by the National Center for Health Statistics (NCHS) Ethics Review Board, and written informed consent was obtained from all participants. Because this study used de-identified, publicly available data, additional institutional review board approval was not required. All analyses incorporated survey weights, strata, and primary sampling units to ensure nationally representative estimates and correct variance calculations.
Variable definitions
Measurement of KDR
To quantify adherence to a ketogenic dietary pattern, KDR was calculated based on macronutrient intake derived from NHANES dietary recall data. Dietary intake information was obtained from two 24-hour dietary recall interviews administered by trained professionals, and the mean nutrient intake across the two recalls was used for analysis. The KDR was computed according to the equation originally proposed by Withrow and subsequently adopted in population-based nutritional studies: KDR = (0.9 × fat + 0.46 × protein) / (0.1 × fat + 0.58 × protein + carbohydrates), where fat, protein, and carbohydrate intake (grams per day) were obtained from the NHANES dietary component.
This ratio reflects the relative contribution of ketogenic (fat and part of protein) and anti-ketogenic (carbohydrates and the remaining protein) macronutrients. A higher KDR value indicates greater adherence to a ketogenic dietary pattern and a stronger potential to induce nutritional ketosis [27, 28].
Ascertainment of overactive bladder and nocturia
OAB and nocturia were assessed based on standardized urinary health questionnaires administered through structured, face-to-face interviews in NHANES. OAB was evaluated using the Overactive Bladder Symptom Score (OABSS), derived from three NHANES items: urinary urgency (KIQ044), urge urinary incontinence (KIQ450), and nocturia (KIQ480). Each item was scored according to validated criteria from the International Continence Society, and the total score ranged from 0 to 15.
Participants with a total OABSS score ≥ 3 were classified as having OAB, a feasible and validated approach used in prior population-based studies (Supplementary Table S1).
Specifically, urinary urgency was determined by asking, “During the past 12 months, have you had an urge to urinate that was difficult to defer?”, while urge urinary incontinence (UUI) was defined as a positive response to “During the past 12 months, have you leaked or lost control of even a small amount of urine with an urge or pressure to urinate and could not get to the toilet fast enough?”. Nocturia was assessed using the question, “During the past 30 days, how many times per night did you most typically get up to urinate, from the time you went to bed at night until you got up in the morning?”. Participants reporting two or more voids per night were defined as having nocturia.
In this analysis, OAB was treated as the primary urinary outcome, and nocturia was examined as an independent but pathophysiologically related condition.
Measurement of FI
Frailty was evaluated using an FI constructed from NHANES data based on the cumulative deficit model. The FI quantifies biological vulnerability as the proportion of health deficits relative to the total number of potential deficits, ranging from 0 to 1. In this study, a 49-item FI covering seven health domains—cognition, functional limitation, depressive symptoms, comorbidities, healthcare utilization, physical performance, and laboratory indicators—was developed following validated protocols (Supplementary Table S2).
Participants with valid data for at least 80% of items (≥ 40 variables) were included. An FI ≥ 0.21 indicated frailty, while values below this threshold denoted non-frailty.
Measurement of PHR
PHR was calculated as platelet count (10⁹/L) divided by serum HDL-C (mmol/L). This ratio serves as a composite marker of systemic inflammation and metabolic imbalance, reflecting the interplay between platelet activation and lipid metabolism. Higher PHR values indicate increased inflammatory and oxidative stress burden, which may contribute to bladder dysfunction.
Assessment of covariates
Covariates were determined based on prior literature, clinical judgment, and potential biological mechanisms that could confound the observed associations. These included sociodemographic, lifestyle, and clinical variables. Sociodemographic variables included age (< 40, 40–60, ≥ 60 years), sex (male or female), and race/ethnicity (non-Hispanic White, non-Hispanic Black, Mexican American/Other Hispanic, and Other, including multiracial).
Lifestyle factors included smoking status, alcohol consumption (≤ 12 or > 12 drinks per year), physical activity level (categorized as < 500 or ≥ 500 MET·min/week according to WHO guidelines), and sedentary time (< 7 or ≥ 7 h per day). Educational attainment was categorized as below high school, high school graduate, or above high school. Marital status was grouped as married/living with partner, widowed/divorced/separated, or never married.
Socioeconomic status was represented by the family poverty income ratio (PIR; <1.3, 1.3–3.5, or > 3.5).
Clinical covariates included body mass index (BMI; <25, 25–30, > 30 kg/m²), hypertension, dyslipidemia, diabetes mellitus, chronic kidney disease (CKD), and cardiovascular disease (CVD). Hypertension was defined as systolic blood pressure ≥ 140 mmHg, diastolic blood pressure ≥ 90 mmHg, current use of antihypertensive medication, or a self-reported physician diagnosis of hypertension. Diabetes mellitus was defined as an HbA1c level of ≥ 6.5%, a fasting plasma glucose level of ≥ 126 mg/dL, the use of antidiabetic medications, or a self-reported diagnosis.
CVD was determined based on self-reported physician diagnoses of coronary heart disease, angina pectoris, myocardial infarction, heart failure, or stroke.
Total energy intake was calculated as the mean value of two 24-hour dietary recalls, and plain water intake (PWI) was also derived from the two-day dietary data. Multicollinearity among all covariates was assessed using generalized variance inflation factors adjusted for degrees of freedom (GVIF^(1/(2*DF))). All adjusted GVIF values were below 2, suggesting minimal collinearity.
To minimize potential bias caused by missing information, multiple imputation was performed for covariates with missing rates below 6.5%, assuming data were missing at random. Missing values were imputed using linear and logistic regression models, with final estimates generated through nearest-neighbor matching under the predictive mean matching (PMM) framework.
Statistical analysis
All analyses were performed in accordance with the analytic guidelines of NHANES, accounting for the complex, multistage sampling design by incorporating sampling weights, strata (SDMVSTRA), and primary sampling units (SDMVPSU) to obtain nationally representative estimates. Weighted descriptive statistics were used to summarize participant characteristics. The normality of continuous variables was evaluated using the Shapiro–Wilk test and visual inspection of histograms.
Variables with approximately normal distributions were expressed as weighted means and standard errors (SE), while those with skewed distributions were presented as weighted medians and interquartile ranges (IQR). Categorical variables were summarized as weighted counts and percentages. Between-group comparisons were performed using the Student’s t-test or Mann–Whitney U test for continuous variables, and the Rao–Scott chi-square test for categorical variables.
Weighted multivariable logistic regression models were employed to investigate the associations between the KDR and the risk of OAB and nocturia, with results presented as odds ratios (ORs) and 95% confidence intervals (CIs). The regression models incorporated a comprehensive set of covariates selected based on prior research findings and clinical relevance. These included demographic factors (age, sex, race/ethnicity), socioeconomic indicators (education level, marital status, and poverty-to-income ratio), lifestyle behaviors (smoking status, alcohol consumption categorized as < 12 or ≥ 12 drinks per year, physical activity level, and sedentary time defined as < 7 or ≥ 7 h per day), and clinical conditions (BMI, hypertension, diabetes, dyslipidemia, cardiovascular disease, and chronic kidney disease).
In addition, mean total energy intake and mean drinking water intake, averaged over two 24-hour dietary recalls, were included as continuous covariates.
The NHANES complex survey design, including sampling weights, was accounted for in all analyses. Restricted cubic spline (RCS) analysis, adjusted for multivariate regression, was applied to flexibly examine the potential nonlinear dose–response relationships between KDR and LUTS, including OAB and nocturia. Threshold (piecewise) regression analyses were further performed to identify potential inflection points, with the optimal threshold determined using a maximum-likelihood recursive algorithm.
Likelihood ratio tests were used to compare the goodness of fit between linear and segmented models, verifying the presence of nonlinear effects. To examine the robustness of the associations, subgroup and interaction analyses were performed across major demographic and clinical characteristics, such as age group, sex, education level, physical activity, BMI category, marital status, smoking status, alcohol consumption, hypertension, diabetes, CKD, dyslipidemia, and sedentary time. Statistical interaction was assessed by introducing cross-product terms between KDR and each stratification variable into the regression models.
Finally, mediation analyses were conducted to determine whether the FI and PHR mediated the associations between KDR and OAB. All analyses were performed using R software (version 4.4.1) and EmpowerStats. A two-sided P value < 0.05 was considered statistically significant.
Results
Baseline characteristics of study participants
A total of 22,249 participants were included in this study. Among them, 4,404 individuals (19.79%) had OAB, and 6,838 individuals (30.73%) reported nocturia. Participants with OAB or nocturia were generally older and exhibited differences across multiple demographic, lifestyle, and clinical variables (all P < 0.05).
Specifically, significant differences were observed in age, energy intake, water consumption, alcohol use, education level, smoking status, physical activity, BMI, race/ethnicity, marital status, poverty-to-income ratio, hypertension, hyperlipidemia, chronic kidney disease, cardiovascular disease, diabetes status, and frailty index (Tables 1 and 2). Notably, sedentary time did not differ significantly between OAB and non-OAB groups (P > 0.05), whereas a significant difference was found between participants with and without nocturia (P < 0.05).
Table 1Baseline characteristics of participants with or without overactive bladderVariableTotal (n = 22249)Without OAB (n = 17845)With OAB (n = 4404)PAge, years47.46 (0.27)45.56 (0.27)57.89 (0.42)< 0.001Mean energy intake, kcal/day2110.33 (8.80)2140.28 (10.01)1946.54 (18.89)< 0.001PWI, ml/day855.00 (340.88, 1566.48)866.53 (355.20, 1587.00)762.85 (296.25, 1474.00)< 0.001KDR0.37 (0.0017)0.37 (0.0018)0.36 (0.0027)0.002Age group, n(%)< 0.001< 407425 (35.62)6849 (39.36)576 (15.18)40–607455 (38.72)6139 (39.26)1316 (35.75)> 607369 (25.66)4857 (21.38)2512 (49.06)Race/ethnicity, n(%)< 0.001Non-Hispanic White3346 (7.93)2701 (8.04)645 (7.36)Non-Hispanic Black10,157 (69.75)8305 (70.53)1852 (65.52)Mexican American / Hispanic4539 (10.40)3320 (9.22)1219 (16.86)Other Race4207 (11.91)3519 (12.21)688 (10.25)Gender, n(%)< 0.001male11,166 (49.2481)9327 (51.2722)1839 (38.1787)female11,083 (50.7519)8518 (48.7278)2565 (61.8213)Education level, n(%)< 0.001Less than high school5192 (15.06)3658 (13.16)1534 (25.42)High school or equivalent5083 (22.18)4043 (21.69)1040 (24.84)College or above11,974 (62.76)10,144 (65.14)1830 (49.74)Smoking status, n(%)< 0.001No12,098 (54.33)9960 (55.54)2138 (47.68)Yes10,151 (45.67)7885 (44.46)2266 (52.32)Alcohol intake, n(%)< 0.001No6077 (21.78)4584 (20.50)1493 (28.75)Yes16,172 (78.22)13,261 (79.50)2911 (71.25)Physical activity, n(%)< 0.001Low physical activity8487 (33.52)6217 (30.98)2270 (47.42)High physical activity13,762 (66.48)11,628 (69.02)2134 (52.58)BMI, n(%)< 0.001< 256321 (29.88)5435 (31.46)886 (21.24)25–307375 (33.21)6082 (33.84)1293 (29.80)> 308553 (36.91)6328 (34.70)2225 (48.96)Marital status, n(%)< 0.001Married/Living with partner13,320 (62.81)10,980 (63.69)2340 (57.98)Widowed/Divorced/Separated4915 (18.55)3399 (16.55)1516 (29.47)Never married4014 (18.64)3466 (19.76)548 (12.55)PIR group, n(%)< 0.001< 1.36939 (21.44)5205 (19.82)1734 (30.27)1.3–3.58323 (34.71)6578 (33.82)1745 (39.62)> 3.56987 (43.85)6062 (46.36)925 (30.10)Hypertension, n(%)< 0.001No12,736 (62.12)11,180 (65.98)1556 (41.02)Yes9513 (37.88)6665 (34.02)2848 (58.98)hyperlipemia, n(%)< 0.001No6269 (28.67)5434 (30.30)835 (19.75)Yes15,980 (71.33)12,411 (69.70)3569 (80.25)CKD, n(%)< 0.001No18,380 (86.58)15,349 (89.02)3031 (73.26)Yes3869 (13.42)2496 (10.98)1373 (26.74)CVD, n(%)< 0.001No19,851 (91.49)16,421 (93.51)3430 (80.49)Yes2398 (8.51)1424 (6.49)974 (19.51)Diabetes mellitus, n(%)< 0.001No10,310 (53.03)9002 (56.27)1308 (35.34)prediabetes7699 (32.56)6085 (31.84)1614 (36.48)diabetes4240 (14.40)2758 (11.89)1482 (28.18)ST, n(%)0.887< 713,739 (58.46)10,996 (58.44)2743 (58.62)≥ 78510 (41.54)6849 (41.56)1661 (41.38)Frailty status, n(%)< 0.001No17,601 (83.38)15,237 (88.01)2364 (58.07)Yes4648 (16.62)2608 (11.99)2040 (41.93)KDR four, n(%)0.019Q16234 (25.00)4911 (24.59)1323 (27.25)Q25759 (25.00)4633 (25.07)1126 (24.59)Q35228 (25.00)4243 (24.96)985 (25.20)Q45028 (25.00)4058 (25.38)970 (22.96)KDR, ketogenic diet ratio; OAB, overactive bladder; PHR, platelet-to-HDL-C ratio; FI, Frailty Index; PWI, plain water intake; BMI, body mass index; PIR, poverty-to-income ratio; CKD, chronic kidney disease; CVD, cardiovascular disease; ST, sedentary time
Table 2Baseline characteristics of participants with or without nocturiaVariableTotal (n = 22249)Without nocturia (n = 15411)With nocturia (n = 6838)PAge, years47.46 (0.27)44.86 (0.29)55.32 (0.36)< 0.001Mean energy intake, kcal/day2110.33 (8.80)2146.20 (10.87)2002.17 (15.40)< 0.001PWI, ml/day855.00 (340.88, 1566.48)870.00 (355.50, 1587.00)780.00 (300.00, 1507.50)< 0.001KDR0.37 (0.0017)0.37 (0.0018)0.37 (0.0024)0.019Age group, n(%)< 0.001< 407425 (35.62)6232 (41.07)1193 (19.19)40–607455 (38.72)5289 (38.87)2166 (38.26)> 607369 (25.66)3890 (20.06)3479 (42.55)Race/ethnicity, n(%)< 0.001Non-Hispanic White3346 (7.93)2288 (7.92)1058 (7.98)Non-Hispanic Black10,157 (69.75)7322 (71.08)2835 (65.74)Mexican American / Hispanic4539 (10.40)2708 (8.59)1831 (15.87)Other Race4207 (11.91)3093 (12.41)1114 (10.41)Gender, n(%)< 0.001male11,166 (49.25)7975 (50.88)3191 (44.33)female11,083 (50.75)7436 (49.12)3647 (55.67)Education level, n(%)< 0.001Less than high school5192 (15.06)2957 (12.51)2235 (22.75)High school or equivalent5083 (22.18)3397 (21.04)1686 (25.60)College or above11,974 (62.76)9057 (66.45)2917 (51.65)Smoking status, n(%)< 0.001No12,098 (54.33)8709 (56.43)3389 (47.99)Yes10,151 (45.67)6702 (43.57)3449 (52.01)Alcohol intake, n(%)< 0.001No6077 (21.78)3875 (19.89)2202 (27.47)Yes16,172 (78.22)11,536 (80.11)4636 (72.53)Physical activity, n(%)< 0.001Low physical activity8487 (33.52)5272 (30.46)3215 (42.75)High physical activity13,762 (66.48)10,139 (69.54)3623 (57.25)BMI, n(%)< 0.001< 256321 (29.88)4803 (31.99)1518 (23.50)25–307375 (33.21)5251 (33.83)2124 (31.35)> 308553 (36.91)5357 (34.17)3196 (45.15)Marital status, n(%)< 0.001Married/Living with partner13,320 (62.81)9485 (63.73)3835 (60.02)Widowed/Divorced/Separated4915 (18.55)2845 (15.97)2070 (26.32)Never married4014 (18.64)3081 (20.29)933 (13.66)PIR group, n(%)< 0.001< 1.36939 (21.44)4308 (19.08)2631 (28.55)1.3–3.58323 (34.71)5682 (33.83)2641 (37.37)> 3.56987 (43.85)5421 (47.09)1566 (34.08)Hypertension, n(%)< 0.001No12,736 (62.12)9986 (67.78)2750 (45.05)Yes9513 (37.88)5425 (32.22)4088 (54.95)hyperlipemia, n(%)< 0.001No6269 (28.67)4792 (30.69)1477 (22.55)Yes15,980 (71.33)10,619 (69.31)5361 (77.45)CKD, n(%)< 0.001No18,380 (86.58)13,417 (89.81)4963 (76.85)Yes3869 (13.42)1994 (10.19)1875 (23.15)CVD, n(%)< 0.001No19,851 (91.49)14,268 (93.98)5583 (84.01)Yes2398 (8.51)1143 (6.02)1255 (15.99)Diabetes mellitus, n(%)< 0.001No10,310 (53.03)7985 (57.44)2325 (39.73)prediabetes7699 (32.56)5213 (31.46)2486 (35.88)diabetes4240 (14.40)2213 (11.10)2027 (24.38)ST, n(%)0.020< 713,739 (58.46)9370 (57.84)4369 (60.35)≥ 78510 (41.54)6041 (42.16)2469 (39.65)Frailty status, n(%)< 0.001No17,601 (83.38)13,318 (88.97)4283 (66.52)Yes4648 (16.62)2093 (11.03)2555 (33.48)KDR four, n(%)0.008Q16234 (25.00)4175 (24.28)2059 (27.16)Q25759 (25.00)4043 (25.34)1716 (23.99)Q35228 (25.00)3702 (25.11)1526 (24.66)Q45028 (25.00)3491 (25.27)1537 (24.19)KDR, ketogenic diet ratio; OAB, overactive bladder; PHR, platelet-to-HDL-C ratio; FI, Frailty Index; PWI, PWI, plain water intake; BMI, body mass index; PIR, poverty-to-income ratio; CKD, chronic kidney disease; CVD, cardiovascular disease; ST, sedentary time
The relationship between KDR and OAB
Weighted multivariable logistic regression analyses revealed a significant inverse association between KDR and the risk of OAB across all models (Table 3). When KDR was treated as a continuous variable, higher KDR was consistently associated with a lower risk of OAB across all regression models. In the unadjusted model, each unit increase in KDR was significantly linked to reduced odds of OAB (OR = 0.53, 95% CI: 0.35–0.79, P = 0.002).
After adjustment for demographic factors (Model 2), the inverse association remained statistically significant (OR = 0.53, 95% CI: 0.33–0.84, P = 0.009). This relationship persisted and even strengthened in the fully adjusted model that included metabolic and lifestyle covariates (Model 3: OR = 0.48, 95% CI: 0.30–0.79, P = 0.004). When KDR was categorized into quartiles, a clear dose–response relationship was observed.
Participants in the highest quartile (Q4) had significantly lower odds of OAB compared with those in the lowest quartile (Q1) (OR = 0.83, 95% CI: 0.72–0.96, P = 0.011), and the linear trend was statistically significant (P for trend = 0.023). These results indicate a consistent and robust inverse relationship between KDR and the risk of OAB. To further characterize the dose–response pattern, RCS analyses were performed (Fig.
2). RCS analysis demonstrated a significant and primarily linear inverse association (P for overall < 0.001; P for nonlinearity = 0.211). Both multivariable and RCS analyses revealed a largely linear inverse association between KDR and the risk of OAB, with no strong evidence of nonlinearity.
However, to further explore whether subtle threshold effects might exist within this overall linear trend, a two-piecewise linear regression model was subsequently applied.
Table 3Weighted logistic regression analysis of KDR and the risk of OAB across multiple modelsVariablesModel 1Model 2Model 3OR (95%CI)POR (95%CI)POR (95%CI)PKDR continuous0.53 (0.35 ~ 0.79)0.0020.53 (0.33 ~ 0.84)0.0090.48 (0.30 ~ 0.79)0.004KDR quartilesQ 11.00 (Reference)1.00 (Reference)1.00 (Reference)Q 20.89 (0.79 ~ 0.99)0.0390.85 (0.75 ~ 0.97)0.0180.88 (0.77 ~ 1.01)0.084Q 30.91 (0.80 ~ 1.04)0.1850.91 (0.79 ~ 1.05)0.2060.92 (0.79 ~ 1.07)0.294Q 40.82 (0.72 ~ 0.92)0.0020.81 (0.70 ~ 0.94)0.0070.79 (0.68 ~ 0.92)0.004P for Trend0.46 (0.28 ~ 0.76)< 0.0010.47 (0.26 ~ 0.85)0.0140.40 (0.22 ~ 0.72)0.003Q: quartiles; OR: odds ratio; CI: confidence interval; Ref: referenceModel1: Crude: UnadjustedModel 2: Adjusted for age, sex, and raceModel 3: Adjusted for age, sex, race, education, alcohol consumption, smoking status, physical activity, BMI, marital status, poverty-to-income ratio, hypertension, hyperlipidemia, chronic kidney disease (CKD), cardiovascular disease, diabetes, and average daily water and energy intake (2-day average)
Fig. 1Flowchart of the sample selection from NHANES 2005–2018
Fig. 2Weighted restricted cubic spline regression of KDR with OAB (A) and nocturia (B). Notes: The adjusted restricted triple spline model was adjusted for age, sex, race, education, alcohol consumption, smoking status, physical activity, BMI, marital status, poverty-to-income ratio, hypertension, hyperlipidemia, chronic kidney disease (CKD), cardiovascular disease, diabetes, and average daily water and energy intake (2-day average).
KDR, ketogenic diet ratio; OAB, overactive bladder;
The relationship between KDR and nocturia
In the unadjusted model, a higher KDR was significantly associated with a reduced likelihood of nocturia (OR = 0.67, 95% CI: 0.48–0.94, P = 0.021). This inverse association persisted after adjusting for age, sex, and race (OR = 0.62, 95% CI: 0.43–0.90, P = 0.014). In the fully adjusted model, which additionally accounted for socioeconomic, behavioral, and clinical covariates, the association remained negative in direction but did not reach statistical significance (OR = 0.71, 95% CI: 0.48–1.04, P = 0.082).
The attenuation of significance in the fully adjusted model suggests that the relationship between KDR and nocturia may not follow a strictly linear pattern, warranting further exploration of potential nonlinear or threshold effects (Table 4).
Table 4Weighted logistic regression analysis of KDR and the risk of nocturia across multiple modelsVariablesModel1Model2Model3OR (95%CI)POR (95%CI)POR (95%CI)PKDR0.67 (0.48 ~ 0.94)0.0210.62 (0.43 ~ 0.90)0.0140.71 (0.48 ~ 1.04)0.082OR: Odds Ratio, CI: Confidence IntervalQ: quartiles; OR: odds ratio; CI: confidence interval; Ref: referenceModel 1: Crude modelModel 2: Adjusted for age, sex, and raceModel 3: Adjusted for age, sex, race, education, alcohol consumption, smoking status, physical activity, BMI, marital status, poverty-to-income ratio, hypertension, hyperlipidemia, chronic kidney disease (CKD), cardiovascular disease, diabetes, sedentary time, and average daily water and energy intake (2-day average)
To evaluate the possibility of nonlinearity, RCS analysis was conducted. The RCS analysis demonstrated a significant overall association (P for overall < 0.001) and clear evidence of nonlinearity (P for nonlinearity < 0.001). The spline curve revealed a decline in nocturia risk at lower KDR values, followed by a gradual rise at higher KDR levels (Fig.
2), suggesting a biphasic rather than a monotonic dose–response pattern.
Threshold effect analysis of KDR on OAB
In contrast, the threshold effects analyses provided a more refined interpretation of the relationship between KDR and OAB (Table 5). The likelihood ratio test indicated that the two-piecewise linear model significantly improved model fit compared with the single linear model (P = 0.021), suggesting the presence of a potential threshold effect. An inflection point was identified at KDR = 0.218.
Below this threshold, higher KDR was associated with markedly lower odds of OAB (OR = 0.01; 95% CI, 0.00–0.72; P = 0.035), whereas above this point, the inverse association persisted but became substantially weaker (OR = 0.68, 95% CI: 0.47–0.99, P = 0.044). These findings indicate that higher KDR is associated with lower prevalence of OAB, with the association appearing stronger at lower KDR levels and leveling off beyond the identified threshold.
Table 5Threshold effect analysis of KDR on OABOutcomeEffectPModel 1 Fitting model by standard linear regression0.56 (0.40–0.79)< 0.001Model 2 Fitting model by two-piecewise linear regressionInflection point (K)0.218< K-segment effect0.01 (0.00–0.72)0.035≥ K-segment effect0.68 (0.47–0.99)0.044P for likelihood test0.021
Threshold effect analysis of KDR on nocturia
To further quantify the potential threshold pattern between KDR and nocturia, a segmented (two-piecewise) linear regression model was employed (Table 6). Compared with the single linear model, the segmented model provided a significantly better fit to the data (likelihood ratio test, P < 0.001), supporting the presence of a non-uniform dose–response relationship. An inflection point was identified at KDR = 0.342.
Below this threshold, higher KDR values were markedly associated with a decreased likelihood of nocturia (OR = 0.14, 95% CI: 0.06–0.36, P < 0.001), suggesting a strong protective effect within the lower range of KDR. However, beyond this point, the inverse association substantially attenuated and became statistically non-significant (OR = 1.34, 95% CI: 0.84–2.14, P = 0.224).
Table 6Threshold effect analysis of KDR on nocturiaOutcomeEffectPModel 1 Fitting model by standard linear regression0.65 (0.49–0.87)0.004Model 2 Fitting model by two-piecewise linear regressionInflection point (K)0.342< 0.3420.14 (0.06–0.36)< 0.001≥ 0.3421.34 (0.84–2.14)0.224P for likelihood test< 0.001
Subgroup analysis
To assess the robustness of the observed association between KDR and OAB, stratified analyses were performed across multiple demographic and clinical subgroups. The results revealed a consistently inverse association between KDR and OAB across nearly all subgroups (ORs < 1), indicating that higher KDR levels were generally associated with lower odds of OAB. The only exception was the 40–60-year age group, in which the association was null (P = 0.954).
No significant interaction effects were detected in any subgroup, suggesting that the protective relationship between KDR and OAB was robust and broadly generalizable across different population characteristics (Fig. 3).
Fig. 3Weighted subgroup analyses showing the associations of KDR with OAB (A) and nocturia (B). Notes: The adjusted restricted triple spline model was adjusted for age, sex, race, education, alcohol consumption, smoking status, physical activity, BMI, marital status, poverty-to-income ratio, hypertension, hyperlipidemia, chronic kidney disease (CKD), cardiovascular disease, diabetes, and average daily water and energy intake (2-day average).
KDR, ketogenic diet ratio; OAB, overactive bladder; OR, odds ratio; CI, confidence interval
Stratified analyses demonstrated that the inverse association between KDR and nocturia was more pronounced in specific subgroups (Fig. 3). The association was particularly significant among Mexican American/Hispanic participants (OR = 0.48, 95% CI: 0.24–0.96, P = 0.043), older adults aged > 60 years (OR = 0.51, 95% CI: 0.28–0.92, P = 0.028), individuals with low physical activity (< 500 MET-min/week) (OR = 0.43, 95% CI: 0.24–0.79, P = 0.009), those with diabetes (OR = 0.40, 95% CI: 0.19–0.83, P = 0.017), and participants with a PIR between 1.3 and 3.5 (OR = 0.53, 95% CI: 0.28–0.98, P = 0.045).
These findings suggest that populations with greater metabolic vulnerability or limited energy expenditure may derive stronger protective benefits from higher KDR levels.
Although modest heterogeneity was observed across subgroups, the direction of association remained largely consistent, suggesting that the relationship between KDR and nocturia is generally robust but may vary according to physiological or behavioral contexts. Notably, a significant interaction with physical activity was identified (P for interaction < 0.05). Notably, stratified analyses revealed that in individuals with low activity levels, higher KDR was linearly associated with a reduced risk of nocturia, whereas in those with higher activity levels, the association followed a nonlinear U-shaped curve with an inflection point near KDR = 0.344 (Fig.
4; Supplementary Table S3). These results indicate a potential association between moderate ketogenic patterns and lower prevalence of nocturia. In contrast, higher levels of KDR combined with greater physical activity were associated with smaller or reversed differences, highlighting the complex interplay between macronutrient metabolism, energy balance, and nocturnal bladder function.
Fig. 4Weighted restricted cubic spline regression of KDR with nocturia in participants with < 500 MET-min/week (A) and ≥ 500 MET-min/week (B). Notes: The adjusted restricted triple spline model was adjusted for age, sex, race, education, alcohol consumption, smoking status, physical activity, BMI, marital status, poverty-to-income ratio, hypertension, hyperlipidemia, chronic kidney disease (CKD), cardiovascular disease, diabetes, and average daily water and energy intake (2-day average).
KDR, ketogenic diet ratio;
Mediation analysis of the relationship between KDR and OAB
In the mediation model illustrated in Figs. 5, KDR served as the independent variable, OAB was the dependent variable, and PHR and frailty index acted as the mediating variables. As shown in Table 7, after adjusting for potential confounders, both mediators exhibited significant indirect effects on the association between KDR and OAB.
Specifically, in the PHR-mediated model, the indirect effect of KDR on OAB was − 0.0042 (P = 0.040), accounting for 5.8% of the total effect, while the direct effect remained significant (β = −0.0672, P < 0.001). In the frailty index–mediated model, the indirect effect was − 0.0153 (P < 0.001), contributing to 21.6% of the total effect. Both models demonstrated that higher KDR was associated with a lower risk of OAB, partially mediated through improved platelet–lipid balance and reduced frailty levels.
Fig. 5Mediation analysis chart of the indirect effects of KDR on OAB via PHR (A) and FI (B). Analyses were adjusted for covariates according to Model 3.
KDR, ketogenic diet ratio; OAB, overactive bladder; PHR, platelet-to-HDL-C ratio; FI, Frailty Index
Table 7Mediation analysis of the FI and PHR in the association between KDR and OABEffectEstimate95% CI lower95% CI upperP-valueMediationFIIndirect effect-0.0153-0.0224-0.00750.000121.6420Direct effect-0.0521-0.0837-0.00560.040078.3580Total effect-0.0674-0.0964-0.02230.0001100.0000EffectEstimate95% CI lower95% CI upperP-valueMediationPHRIndirect effect-0.0042-0.0068-0.00050.04005.8461Direct effect-0.0672-0.1053-0.03230.000194.1539Total effect-0.0714-0.1114-0.03620.0001100.0000
Discussion
Using data from 22,249 participants in the NHANES, our analysis revealed two distinct association patterns between the KDR and LUTS. A higher KDR was consistently associated with a substantially lower risk of OAB, indicating a robust inverse relationship. In contrast, the association between KDR and nocturia exhibited a nonlinear, threshold-dependent pattern.
RCS and threshold effects analyses identified a turning point at approximately KDR = 0.342. Below this threshold, an increase in KDR was strongly protective against nocturia; however, beyond this level, the risk of nocturia began to rise progressively, suggesting that excessive ketogenic patterns may attenuate or even reverse the metabolic benefits. Moreover, physical activity significantly modified this relationship.
Among individuals with lower activity levels, the association between KDR and nocturia was nearly linear and inverse. In contrast, those with higher physical activity exhibited a U-shaped pattern, suggesting that dietary and behavioral factors may jointly influence nocturnal bladder regulation. Further mediation analysis revealed that both the FI and the PHR partially mediated the association between KDR and OAB.
The significance of dietary patterns in relation to bladder symptoms has been previously highlighted by studies in this domain. A comprehensive review emphasised that nutritional intake and physical activity constitute modifiable factors that influence storage symptoms such as OAB and nocturia [29, 30]. A cross-sectional investigation in an older U.S. cohort observed that greater flavonoid intake was inversely associated with nocturia [31].
Additionally, an observational case series reported improvement in urinary incontinence among obese women following a very-low-carbohydrate, ketogenic diet [31]. While these studies did not quantify KDR or explore dose response and lifestyle interaction, they provide supporting evidence for the concept that macronutrient balance and low-carbohydrate patterns may benefit bladder function.
Mechanistically, ketogenic diets have long been recognised for their neuro-metabolic and anti-inflammatory efficacy in neurological disorders, including epilepsy, Alzheimer’s disease, and Parkinson’s disease [32]. These effects include enhanced mitochondrial function, suppression of inflammasome activation, modulation of neurotransmitter balance (increasing GABA and reducing glutamate excitotoxicity), and alteration of gut-brain signaling [23, 33]. The relevance of these pathways to LUTS is increasingly recognised, as both OAB and nocturia involve dysregulated bladder-brain communication, heightened afferent signalling, and impaired central inhibition [12, 13].
In individuals with OAB, the observed inverse association with KDR aligns with prior evidence suggesting that neural metabolic status may influence detrusor activity and autonomic regulation [34]. Similarly, the consistent inverse pattern is compatible with emerging evidence linking neural metabolic processes to the control of visceral organs [35]. The OAB phenotype is increasingly understood as a state of “neuro-urothelial hyperexcitability,” sharing pathophysiological features with central nervous system excitability disorders like epilepsy [36, 37].
The ketogenic diet’s efficacy in drug-resistant epilepsy is well-established, with mechanisms extending beyond simple ketone body fueling. Crucially, β-hydroxybutyrate (BHB) has been shown to function as a signaling molecule, activating hydroxyl-carboxylic acid receptor 2 (HCAR2) and downstream pathways that enhance mitochondrial biogenesis and reduce oxidative stress in neurons [38, 39]. We hypothesize that a similar mechanism may be associated with stabilization of neural circuits within the pontine micturition center and sacral spinal cord, potentially relating to higher thresholds for bladder contractions [40].
This is consistent with our mediation analysis, where the Frailty Index may partially explain the observed association, suggesting KDR may be linked to lower symptom burden through pathways involving systemic physiological resilience [41, 42].
The identification of a U-shaped relationship with a threshold is a novel finding that challenges the “more is better” assumption. Nocturia represents a multifactorial disorder involving the interplay between sleep–wake regulation, circadian control of renal function, metabolic status, and nocturnal polyuria [15, 43]. We propose a dual-phase model to explain this nonlinearity.
At moderate KDR levels, observed patterns are consistent with aspects of improved metabolic health. For instance, prior studies have reported that a low-carbohydrate diet is associated with reduced nocturnal sodium excretion and more stable fluid balance in individuals with insulin resistance, a condition linked to nocturnal polyuria, without implying direct causation [44, 45]. Furthermore, ketone bodies, particularly BHB, can promote deeper, more consolidated sleep as reported in prior studies, potentially reducing arousals that precipitate voiding [46].
However, beyond a certain threshold, the elevated risk of nocturia observed at very high KDR levels may reflect physiological adaptations reported under marked carbohydrate restriction. Previous studies have shown that low-carbohydrate states are accompanied by reduced insulin concentrations, a metabolic condition associated with increased urinary sodium excretion and higher urine volume [20]. Moreover, under these conditions, nutritional ketosis may develop, with higher circulating and urinary ketone bodies whose excretion increases the renal solute load and may induce osmotic diuresis [47].
Together, these alterations in natriuresis, ketonuria, and renal handling of electrolytes could influence overall fluid and electrolyte balance [48], providing a plausible physiological context for the greater nocturnal urine output observed at the upper end of the KDR range. While the precise mechanisms remain incompletely understood, these patterns may be related to the higher nocturia risk observed at very high KDR levels. The profound modification by physical activity provides a critical clue: in highly active individuals, exercise may already optimize metabolic and stress pathways, leaving little room for additional benefit from an extreme diet, hence the observed plateau and potential for harm [49, 50].
An increasingly plausible, yet unexplored, pathway linking KDR to LUTS is the gut-bladder axis. Ketogenic or low-carbohydrate dietary patterns are associated with changes in gut microbial composition and function; in some contexts, these differences may be linked to higher microbial production of short-chain fatty acids (SCFAs) such as butyrate, which have been proposed to exert anti-inflammatory effects and modulate peripheral sensory signaling [51]. These metabolites exert anti-inflammatory effects and modulate peripheral sensory signaling [52].
Experimental studies have shown that gut-derived SCFAs can reduce the excitability of bladder afferent pathways, including those mediated by TRPV1 and P2 × 3 receptors [53, 54]. Accordingly, a moderate KDR may foster a more favorable microbial and metabolic environment, stabilizing urothelial–neural communication and thereby alleviating urgency and frequency. This hypothesis highlights a biologically plausible link between dietary composition, microbiota-derived metabolites, and bladder function, meriting validation in future mechanistic studies.
Mediation analyses further elucidated the systemic pathways linking KDR to bladder function. The FI mediated approximately 21.6% of the KDR–OAB association, indicating that enhanced physiological resilience may contribute to better bladder control, consistent with reports linking frailty and metabolic dysregulation to detrusor overactivity [55, 56]. The PHR accounted for 5.8% of the effect, suggesting that vascular–inflammatory processes are also involved [57, 58].
These findings imply that dietary patterns may influence bladder function not only through neural mechanisms but also via systemic metabolic and inflammatory modulation. Physical activity further modifies the diet–bladder link: in individuals with low activity, the metabolic shift induced by higher KDR may serve a compensatory role; whereas in already active individuals, metabolic pathways may already be optimised by exercise, and additional KDR benefit may plateau or even invert. This compensatory relationship emphasizes the integrated role of diet and physical activity in maintaining bladder health and preventing LUTS.
Our findings carry significant translational potential but should be interpreted with the caution warranted by an observational design. Rather than endorsing a one-size-fits-all dietary recommendation for LUTS, the results support a precision lifestyle approach that integrates dietary composition with physical activity level. For individuals with nocturia who are physically inactive, modest increases in KDR may confer benefit, whereas for highly active individuals, maintaining a moderate rather than extreme KDR appears more appropriate.
Importantly, these propositions are exploratory and should be regarded as hypothesis-generating rather than prescriptive.
Future studies, including randomized controlled trials, are warranted to evaluate whether targeting KDR near the empirically identified thresholds can meaningfully improve urinary symptoms in stratified populations. Such trials should ideally incorporate objective measures to verify ketosis, assess urinary patterns, and monitor sleep or other relevant physiological parameters. Complementary mechanistic studies, including controlled animal experiments, will be important to elucidate the underlying biological pathways and establish causality, particularly regarding ketone-mediated modulation of bladder function and central micturition circuits.
Integrative multi-omics approaches may further clarify potential interactions along the gut–bladder axis and other systemic mechanisms.
This study has several notable strengths. To our knowledge, it is among the first large-scale, nationally representative investigations to explore the relationship between KD and LUTS. By concurrently analyzing OAB and its key clinical manifestation, nocturia, the study offers an integrated perspective on bladder dysfunction from both dietary and metabolic dimensions.
The use of RCS modeling and threshold effect analysis enabled the identification of nonlinear and segmented associations, revealing a potential inflection point in the KDR–nocturia relationship. Furthermore, mediation and sensitivity analyses clarified potential biological pathways and confirmed the robustness of the findings. These analyses also demonstrated that physical activity modifies the metabolic effects of KDR on urinary outcomes, underscoring the interactive influence of diet and lifestyle on bladder regulation.
Collectively, these findings extend current understanding by illustrating that nutritional composition, metabolic health, and behavioral context jointly shape bladder function, thereby suggesting promising directions for lifestyle-based prevention and management of LUTS.
Nonetheless, several limitations should be acknowledged. First, the cross-sectional design precludes causal inference. Individuals with severe nocturia or OAB may modify their diet—for example, by reducing fluid intake or altering macronutrient composition—potentially introducing reverse causation.
Second, dietary intake was assessed using two 24-hour recalls, which may not fully capture habitual eating patterns, and inter-individual variability in macronutrient consumption was substantial. KDR reflects a relative ketogenic dietary tendency rather than fixed macronutrient distributions; therefore, the identified thresholds should be interpreted as indicators of dietary inclination rather than precise intakes of carbohydrate, protein, or fat. Third, the study lacked direct measurements of ketone levels, nocturnal urine volume, vasopressin rhythms, and sleep architecture.
Residual confounding from unmeasured factors, such as medication use, timing of fluid intake, or undiagnosed sleep disorders, cannot be excluded. Finally, because the analysis was based on the US population in NHANES, generalizability to other populations may be limited. Future studies, ideally longitudinal or interventional, are warranted to validate these associations and to define corresponding quantitative nutrient targets under controlled or standardized conditions.
Conclusion
In conclusion, we found that a ketogenic dietary tendency is associated with reduced odds of OAB and exhibits a threshold-dependent, nonlinear association with nocturia that is modified by physical activity. These data support an integrative framework that links diet, neural and urothelial signaling, systemic health, and lifestyle factors. They generate specific hypotheses for future experimental and clinical research aimed at personalised, non-pharmacological strategies to prevent and manage lower urinary tract symptoms.
However, these findings remain preliminary and should be interpreted with caution. Future longitudinal and interventional studies are needed to further clarify the underlying mechanisms and determine whether the observed associations reflect causal relationships.
Supplementary Information
Supplementary Material 1
Supplementary Material 2
Supplementary Material 3