Dietary acid load and oxidative stress in relation to sleep quality and duration among Iranian adults

Abstract

Few studies assessed the relation between dietary acid load and sleep quality and duration in Middle Eastern populations. Moreover, no study evaluated the potential relationship between oxidative stress biomarkers and sleep parameters. We conducted the present cross-sectional investigation to comprehensively assess this issue. Dietary acid load was assessed using a validated semi-quantitative 168-item food frequency questionnaire. Oxidative stress biomarkers were evaluated in participants’ blood samples after a 12-h fast. Sleep quality and duration were assessed using the validated Pittsburgh Sleep Quality Index (PSQI). A total of 536 Iranian adults (289 men) with a mean age of 42.59 year were included. Participants in the second and third tertiles of potential renal acid load (PRAL) ([OR_{T2 vs. T1}=1.79; 95%CI: 1.03–3.15], [OR_{T3 vs. T1}=2.47; 95%CI: 1.44–4.32]) and net endogenous acid production (NEAP) ([OR_{T2 vs. T1}=2.07; 95%CI: 1.20–3.61], [OR_{T3 vs. T1}=2.58; 95%CI: 1.49–4.53]) were more likely to have short sleep duration, compared to the reference. Moreover, participants in the third tertile of PRAL (OR_{T3 vs. T1}=1.72; 95%CI: 1.02–2.90), and NEAP (OR_{T3 vs. T1}=1.77; 95%CI: 1.05-3.00) were more likely to have poor-quality sleep than those in the first tertile. Higher malondialdehyde levels (MDA > 185 nmol/mL) and lower glutathione peroxidase (GPx < 0.70 mU/mL) were associated with higher odds of poor sleep quality. All mentioned results were based on fully adjusted model. In conclusion, there was a dose-related direct association between dietary acid load and odds of disrupted sleep quality and duration, especially in overweight/obese adults. Higher oxidative stress levels (higher MDA and lower GPx) were non-linearly associated with higher odds of poor sleep quality.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1038/s41598-026-59257-5.

Introduction

Sleep is a crucial behavior that strongly influences both mental and physical well-being, and overall quality of life depends on both its duration and quality¹. Recent investigations present a widespread distribution of sleep disorders^2,3. Their prevalence is also considerable among Iranian adults⁴. There is well-established evidence suggesting an association between sleep inadequacy and different inflammatory-mediated metabolic disorders, including metabolic syndrome, cardiovascular diseases, and type two diabetes mellitus^5,6. Alongside metabolic disorientations, sustained sleep shortages can easily cause psychological and mood disorders, affecting one’s personal life⁷. Moreover, sleep complaints may cause personal injury, reduced performance quality, and consequently all-cause mortality, indicating the crucial effect of sleep on global public health^3,8.

Sleep duration and quality may be affected by modifiable factors, including behavioral and sociodemographic, physical activity, and dietary factors, as well as fixed factors such as sex, age, and genetics^9,10. The acidity or alkalinity of the diet can easily be associated with one’s metabolic and psychological state through endocrine changes it may be related to, resulting in changes in sleep quality and quantity¹¹. Dietary acid load is assessed via four scoring approaches, including assessing the potential renal acid load (PRAL), net endogenous acid production (NEAP), dietary acid load (DAL) itself, and protein/potassium ratio^12,13. As stated earlier, there is a thin line between metabolic status and sleep parameters, in that they affect one another^14,15. Oxidative stress, as a metabolic disorientation, may influence sleep duration and quality by affecting brain regions related to mood and sleep regulation¹⁶.

A cross-sectional study on 398 Iranian elders suggested that individuals in the third tertile of PRAL, NEAP, and DAL had poorer sleep quality than those in the first tertile¹⁷. Another cross-sectional investigation showed a significant association between PRAL and NEAP and poor sleep quality among 230 Iranian diabetic women¹⁸. In contrast, another cross-sectional study by Katagiri et al. among middle-aged Japanese female workers found no significant association between carbohydrate intake (as an alkaline diet component) and poor sleep quality¹⁹. No previous study had evaluated the potential association between dietary acid load and sleep duration, or oxidative stress and sleep parameters.

Given the point that previous studies were conducted on a limited group of the population with certain conditions, their results would not be translated to the total population. Moreover, their results were inconsistent regarding sleep parameters, making it difficult to generalize their findings on sleep quality. Alongside that, none of those studies assessed the association between dietary acid load and sleep duration, as a crucial sleep parameter. Moreover, as our literature review suggested, no human research was available on the association between oxidative stress and sleep quality or duration. Thus, in this cross-sectional investigation, we aimed to comprehensively assess the potential relationship between dietary acid load, oxidative stress biomarkers, and sleep parameters (including quality and duration).

Methods

Study design and participants

The current cross-sectional study was implemented in 2021 in Isfahan (a large city in the center of Iran) on a somewhat representative sample of adults²⁰. Considering a 95% confidence interval (type I error of 5%), a precision (d) of 4%, and a 35% prevalence of sleep disorders among Iranian adults, a minimum of 519 participants were required for the current investigation⁴. Based on a multistage cluster-randomized sampling design, 3–4 schools were chosen from each of the 6 educational districts of Isfahan (285 private and public schools in total). Twenty schools, varying from preschools through high schools, were selected for the study during the process. Initially, because of the prevalence of COVID-19 and potential dropouts, 600 men and women aged 20–65 years were invited from these twenty schools. Adults with all job roles in schools (including teachers, crews, managers, assistants, and workers) were included in these invitations to build a representative sample of adults across different socioeconomic statuses. The following non-inclusion criteria were considered before participants were included: (1) adherence to a special weight loss or weight gain diet, (2) being in a pregnancy or lactation state, or (3) a history of cardiovascular diseases (CVDs), stroke, neoplasms, or type 1 diabetes mellitus. Among those who were invited, 543 accepted participation (response rate of 90.5%), of whom seven were omitted with the following reasons: (1) incomplete food frequency questionnaire (n = 4) or (2) a reported total energy intake out of the range of 800 to 4200 kcal/d (n = 3). Ultimately, 536 adults participated in the analyses of the current investigation. Moreover, data on biomarkers were available for 527 of them; thus, the analyses regarding biomarkers were based on this number of participants, and those with missing information were excluded from analyses on the biomarkers (n = 9). The study protocol was assessed ethically and approved by the local Ethics Committee of Isfahan University of Medical Sciences. The study was conducted in accordance with the Helsinki and STROBE guidelines, and all participants signed a written informed consent form.

Assessment of dietary acid load

Usual dietary intake of the participants in the past year was evaluated via a semi-quantitative 168-item FFQ, which had been previously validated among Iranian adults²¹. Before handing the questionnaires, a trained dietitian explained to participants how to report their usual dietary intakes in daily, weekly, or monthly frequencies. Dietary items were converted to g/d using standard household portion sizes²², and intakes in gram were entered into the “Nutritionist IV” software to obtain micro- and macro-nutrient and total energy intakes for each participant.

Dietary intakes can change urinary net acid excretion. Since direct assessment of urinary net acid excretion is difficult, four dietary indices, including PRAL, NEAP, DAL, and the protein/potassium ratio, have been introduced to calculate body acid or dietary acid load. Remer et al. have validated PRAL and NEAP by comparing the scores with the urinary acid load of a 24-hour urine sample²³. Protein, phosphorus, potassium, magnesium, and calcium intakes were adjusted by energy intake via the residual method, and the following formulas were applied to calculate the four acid load indices^12,23–27:

• A)PRAL [mEq/d] = (0.4888 × protein intake [g/d]) + (0.0366 × phosphorus [mg/d]) – (0.0205 × potassium [mg/d]) – (0.0125 × calcium [mg/d]) – (0.0263 × magnesium [mg/d])
• B)NEAP [mEq/d] = (54.5 × protein intake [g/d] ÷ potassium intake [mEq/d]) − 10.2.
• C)DAL [mEq/d] = PRAL [mEq/d] + (body surface area [m²] ×41 [mEq/d]/1.73 [m²])
  1. Body surface area [m²] = 0.007184 × height [cm]^0.725 × weight [kg]^0.425.
• D)The ratio of protein/potassium = Pro/K.

Assessment of sleep duration and quality

The Pittsburgh Sleep Quality Index (PSQI), a widely used self-report questionnaire for sleep evaluation, validated among Iranian adults (with a sensitivity of 93.6 and a specificity of 72.2), was used to assess sleep quality and duration^28,29. PSQI contains 19 questions, which are categorized into seven domains: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficacy, sleep disturbances, use of sleep medications, and daytime dysfunction. Each domain was scored from 0 to 3 using a four-point grading system, with zero indicating a normal state and three a problematic state. The scores from all domains were then summed to obtain an overall score for each participant. The final score could range from 0 to 21; a higher score indicated poorer sleep. An overall score of 6 or higher was considered “poor sleep quality”²⁸. The fourth item of the PSQI asked, “How many hours during the night do you sleep?” The answer to this question was used to calculate sleep duration. Based on a meta-analysis, previous studies have indicated that sleeping less than 6 h can be problematic and may be associated with mortality, CVDs, obesity, and diabetes²⁸. Thus, sleeping less than 6 h overnight was considered “short sleeping”. Moreover, a sensitivity analysis for the odds of sleeping less than 7 h overnight was implemented.

Assessment of oxidative stress biomarkers

Participants were asked to fast overnight before blood sampling. Blood samples were centrifuged, and their serum was collected. Serum samples were stored in a -80-degree centigrade freezer for an average of 3 months. Samples were defrosted in a 4-degree fridge for 24 h before tests. Using commercial enzyme-linked immunosorbent assay (ELISA) kits, we measured serum concentrations of malonaldehyde (MDA) (Kiazist Co, Hamedan, Iran), superoxide dismutase (SOD) (Kiazist Co, Hamedan, Iran), and glutathione peroxidase (GPx) (Kiazist Co, Hamedan, Iran). The detection range was 20–100 nmol/mL for MDA. All intra- and inter-assay CVs were below 10%, and the measurement procedure was handled once for each participant.

Assessment of other variables

Participants’ height and weight were measured while they were standing barefoot and wearing minimal clothing. A body composition analyzer (Tanita MC-780 MA, Tokyo, Japan) was used to measure participants’ body weight with a precision of 0.1 kg. A wall-mounted tape (with an accuracy of 0.1 cm) was used to measure participants’ height. Weight (in kg) and height (in m) were used to calculate each participant’s body mass index (BMI). Waist circumference (WC) was measured for each person using a nonelastic tape (precision 0.1 cm). A self-reported questionnaire was administered to gather information on additional influential confounders, including sex, age, education, marital status, smoking status, socioeconomic status (SES), antidepressant use (nortriptyline, amitriptyline, imipramine, fluoxetine, citalopram, and fluvoxamine), and coffee or tea consumption after 8:00 PM. Obstructive sleep apnea (OSA) was assessed via the STOP-Bang questionnaire (a self-reported eight-item yes-no questionnaire)³⁰. Items involved fatigue, snoring, BMI over 35 kg/m², age over 50 years, male sex, neck circumference over 40 cm, high blood pressure, and observed apnea^30,31. Moreover, physical activity was evaluated using a Persian-validated version of the International Physical Activity Questionnaire (IPAQ)³².

Statistical methods

Normality of continuous variables was evaluated using the Kolmogorov-Smirnov test. Data were presented as percentages and means ± SD (standard deviation) / SE (standard error) for categorical and continuous variables, respectively. Participants were classified into tertiles based on their energy-adjusted, protein/potassium ratio, PRAL, NEAP, and DAL; the first tertile of each index was considered the reference for statistical analyses. Categorical and continuous variables were compared across tertiles using chi-square and one-way analysis of variance (ANOVA), respectively. Moreover, analysis of covariance (ANCOVA) was used to assess dietary intakes (adjusted for sex, age, and total energy intake) across tertiles of the indices. To obtain the odds ratio (OR) for each outcome (sleep duration, quality, or quality domains), binary logistic regression was used. The test was repeated in crude and three adjusted models: model 1 with adjustments for age, sex, and total energy intake; model 2 with further adjustments for marital status, education level, physical activity (HEPA), smoking status, antidepressants usage (including nortriptyline, amitriptyline, imipramine, fluoxetine, citalopram, and fluvoxamine), SES, OSA, and drinking tea or coffee after 8:00 PM; model 3 or fully adjusted-model with further adjustment for BMI. An exploratory subgroup analysis based on weight status was performed based on crude and the first two models. Moreover, another exploratory analysis was carried out for the association between PRAL and NEAP with each PSQI item, in the crude and fully-adjusted model. The non-linear association between serum oxidative stress biomarkers (including MDA, GPx, and SOD) and adjusted odds (adjusted for age, sex, marital status, education level, physical activity (HEPA), smoking status, antidepressants usage, SES, OSA, drinking tea or coffee after 8:00 PM and BMI) of short sleeping and poor sleep quality was evaluated using restricted cubic splines (RCS) test with 4 knots (at 5%, 35%, 65%, and 95% percentiles), using rms (version 8.1) and rmsMD (version 1.0.0) packages. Median values were considered as references in RCS analyses. Sensitivity analyses, excluding extreme biomarker levels (out of the range of 2–98 percentiles), were performed to assess whether results were dependent on the extreme values. All statistical analyses were carried out via R (version 4.5.2).

Results

In total, 536 participants (289 men and 247 women) with a mean age of 42.59 ± 11.14 years and a mean BMI of 26.92 ± 4.43 kg/m² were included in the present investigation. As Supplemental Table 1 presents, there were enormous significant correlations between dietary acid load indices (all P < 0.05). Table 1 presents general characteristics of participants across tertiles of PRAL and NEAP. Mean age of participants was significantly higher in the first tertile of both PRAL and NEAP. In contrast, mean values of weight and height were higher in the third tertile of PRAL and NEAP. Mean MDA concentration was significantly higher in the second tertile of PRAL, and WC was smaller in the first tertile of NEAP. Finally, distributions of sex and having OSA were significantly different across tertiles of both PRAL and NEAP. No other significant differences were observed across the tertiles of PRAL and NEAP in general characteristics or biomarkers.

Table 1.

General characteristics of study participants across tertiles of PRAL and NEAP1.

	PRAL				NEAP
	T1 (n = 179)	T2 (n = 178)	T3 (n = 179)	P2	T1 (n = 179)	T2 (n = 178)	T3 (n = 179)	P2
Index	< −18.9	−18.9 to −2.9	> −2.9		< 31.1	31.1 to 41.8	> 41.8
Age (year)	44.50 ± 11.01	43.36 ± 11.25	39.92 ± 10.69	< 0.001	44.37 ± 11.63	42.68 ± 10.30	40.73 ± 11.19	0.01
Sex [n, (%)]				< 0.001				< 0.001
Male	75 (41.90)	96 (53.93)	118 (65.92)		77 (43.02)	92 (51.69)	120 (67.04)
Female	104 (58.10)	82 (46.07)	61 (34.08)		102 (56.98)	86 (48.31)	59 (32.96)
Weight (kg)	74.41 ± 13.52	74.67 ± 13.68	78.20 ± 16.02	0.02	73.36 ± 11.89	75.55 ± 15.69	78.38 ± 15.36	0.01
Height (cm)	165.50 ± 7.59	167.17 ± 8.68	170.13 ± 8.67	< 0.001	165.69 ± 7.97	167.40 ± 8.21	169.70 ± 8.94	< 0.001
BMI (kg/m2)	27.16 ± 4.49	26.70 ± 4.21	26.91 ± 4.59	0.61	26.72 ± 3.91	26.93 ± 4.93	27.12 ± 4.41	0.69
Waist circumference	91.79 ± 10.72	92.63 ± 10.87	93.57 ± 12.69	0.34	90.92 ± 9.73	93.03 ± 12.09	94.03 ± 12.26	0.03
Serum malondialdehyde (nmol/mL)	168.04 ± 90.77	183.72 ± 97.47	157.92 ± 84.95	0.03	172.00 ± 93.93	167.79 ± 88.72	169.44 ± 92.42	0.91
Serum glutathione peroxidase (mU/mL)	1.79 ± 3.11	2.22 ± 4.69	1.91 ± 3.39	0.55	1.82 ± 3.11	2.24 ± 4.72	1.86 ± 3.34	0.53
Serum superoxide dismutase (Unit)	1.08 ± 0.45	1.07 ± 0.48	1.22 ± 0.54	0.01	1.09 ± 0.46	1.07 ± 0.46	1.20 ± 0.55	0.30
Marital status [n, (%)]				0.71				0.67
Single	30 (16.76)	28 (15.73)	29 (16.20)		30 (16.76)	30 (16.85)	27 (15.08)
Married	145 (81.01)	148 (83.15)	149 (83.24)		145 (81.01)	146 (82.02)	151 (84.36)
Divorced or widowed	4 (2.23)	2 (1.12)	1 (0.56)		4 (2.23)	2 (1.12)	1 (0.56)
Smoking status [n, (%)]				0.16				0.50
	172 (36.09)	164 (92.13)	170 (94.97)		172 (96.09)	165 (92.70)	169 (94.41)
	5 (2.79)	8 (4.49)	2 (1.12)		5 (2.79)	6 (3.37)	4 (2.23)
	2 (1.12)	6 (3.37)	7 (3.91)		2 (1.12)	7 (3.93)	6 (3.35)
Education level [n, (%)]				0.87				0.98
Diploma or lower	21 (11.73)	20 (11.24)	18 (10.06)		20 (11.17)	20 (11.24)	19 (10.61)
Higher than diploma	158 (88.27)	158 (88.76)	161 (89.94)		159 (88.83)	158 (88.76)	160 (89.39)
Physical activity (HEPA) [n, (%)]				0.65				0.72
Inactive	102 (56.98)	104 (58.43)	98 (54.75)		100 (55.87)	105 (58.99)	99 (55.31)
Minimally active	63 (35.20)	64 (35.96)	63 (35.20)		64 (35.75)	63 (35.39)	63 (35.20)
Active	14 (7.82)	10 (5.62)	18 (10.06)		15 (8.38)	10 (5.62)	17 (9.50)
SES [n, (%)]				0.35				0.75
Low	61 (34.08)	58 (32.58)	64 (35.75)		59 (32.96)	57 (32.02)	67 (37.43)
Moderate	54 (30.17)	69 (38.76)	54 (30.17)		58 (32.40)	64 (35.96)	55 (30.73)
High	64 (35.75)	51 (28.65)	61 (34.08)		62 (34.64)	57 (32.02)	57 (31.84)
OSA [n, (%)]				0.03				0.04
No	132 (73.74)	118 (66.29)	106 (59.22)		133 (74.30)	120 (67.42)	105 (58.66)
Borderline	36 (20.11)	54 (30.34)	62 (34.64)		37 (20.67)	51 (28.65)	63 (35.20)
Yes	11 (6.15)	6 (3.37)	11 (6.14)		9 (5.03)	7 (3.93)	11 (6.14)
Tea or coffee drinking after 8:00 PM [n, (%)]				0.69				0.55
No	99 (55.31)	102 (57.30)	107 (59.78)		97 (54.19)	106 (59.55)	105 (58.66)
Yes	80 (44.69)	76 (42.70)	72 (40.22)		82 (45.81)	72 (40.45)	74 (41.34)
Antidepressant3 usage [n, (%)]				0.06				0.26
Non-user	165 (92.18)	167 (93.82)	175 (97.77)		166 (92.74)	168 (94.38)	173 (96.65)
User	14 (7.82)	11 (6.18)	4 (2.23)		13 (7.26)	10 (5.62)	6 (3.35)

¹Continuous variables are reported as mean ± SD, categorical variables are presented as percent,

²Obtained from one-way ANOVA and the Pearson chi-square test for continuous and categorical variables, respectively,

³Including nortriptyline, amitriptyline, imipramine, fluoxetine, citalopram, and fluvoxamine,

BMI, body mass index; SES, socio-economic status; OSA, obstructive sleep apnea; HEPA, Health-Enhancing Physical Activity; n, number; T, tertile.

Dietary intakes of study participants across tertiles of PRAL and NEAP are illustrated in Table 2. As presented, total energy intake was significantly lower in the second tertile of PRAL and NEAP. Participants in the third tertile of PRAL had higher intakes of protein, fat, and phosphorus. In contrast, participants in the first tertile of PRAL had higher intakes of carbohydrates, potassium, and magnesium. Regarding NEAP, protein and fat intakes were higher in the third tertile, and magnesium and potassium intakes were higher in the first tertile. Considering food groups, participants in the third tertile of both PRAL and NEAP consumed more whole and refined grains and red, processed, and white meat; in contrast, participants in the first tertile of both indices had higher consumption of fruits and vegetables. Moreover, consumption of nuts, soy, and legumes was significantly lower in the first PRAL tertile than in the other two.

Table 2.

Dietary intakes (energy, macro/micronutrients, and food groups) of study participants across tertiles of PRAL and NEAP1.

	PRAL				NEAP
	T1 (n = 179)	T2 (n = 178)	T3 (n = 179)	P2	T1 (n = 179)	T2 (n = 178)	T3 (n = 179)	P2
Index	< −18.9	−18.9 to −2.9	> −2.9		< 31.1	31.1 to 41.8	> 41.8
Total energy intake (kcal)	2407.97 ± 49.79	2030.87 ± 49.44	2355.12 ± 50.84	< 0.001	2387.68 ± 50.32	2087.65 ± 50.01	2319.35 ± 51.30	< 0.001
Macronutrients (% of energy)
Protein	13.08 ± 0.20	13.99 ± 0.20	15.82 ± 0.20	< 0.001	12.61 ± 0.19	14.42 ± 0.19	15.86 ± 0.20	< 0.001
Carbohydrate	65.06 ± 0.56	60.58 ± 0.55	56.14 ± 0.57	< 0.001	65.00 ± 0.57	59.62 ± 0.56	57.19 ± 0.58	< 0.001
Fat	24.74 ± 0.49	27.22 ± 0.49	29.27 ± 0.50	< 0.001	25.29 ± 0.50	27.80 ± 0.50	28.11 ± 0.51	< 0.001
Micronutrients (mg/day)
Potassium	4691.37 ± 61.92	3575.72 ± 62.41	3045.86 ± 62.94	< 0.001	4601.22 ± 63.35	3708.45 ± 63.45	3006.65 ± 64.31	< 0.001
Calcium	957.06 ± 28.16	881.21 ± 28.38	924.28 ± 28.62	0.17	929.28 ± 28.07	944.56 ± 28.12	889.05 ± 28.50	0.36
Magnesium	339.04 ± 7.38	245.41 ± 7.33	266.10 ± 7.53	< 0.001	318.09 ± 4.58	285.06 ± 4.59	250.29 ± 4.65	< 0.001
Phosphorus	1180.72 ± 25.46	1170.83 ± 25.66	1261.77 ± 25.88	0.03	1180.66 ± 25.44	1241.48 ± 25.48	1190.09 ± 25.83	0.20
Food groups (g/day)
Whole grains	95.35 ± 5.92	102.92 ± 5.97	135.07 ± 6.02	< 0.001	89.59 ± 5.75	98.80 ± 5.76	145.30 ± 5.84	< 0.001
Refined grains	233.57 ± 12.01	282.27 ± 12.11	294.20 ± 12.21	0.01	240.60 ± 11.98	269.09 ± 12.00	300.19 ± 12.16	0.01
Fruits	800.25 ± 19.72	523.12 ± 19.88	339.19 ± 20.05	< 0.001	790.41 ± 20.04	520.55 ± 20.07	352.44 ± 20.34	< 0.001
Vegetables	495.47 ± 15.50	315.47 ± 15.62	229.16 ± 15.75	< 0.001	469.05 ± 15.89	341.93 ± 15.91	230.10 ± 16.13	< 0.001
Red and processed meat	53.40 ± 3.27	65.21 ± 3.29	86.21 ± 3.32	< 0.001	48.24 ± 3.19	69.01 ± 3.19	87.74 ± 3.24	< 0.001
White meat	30.68 ± 2.50	39.48 ± 2.52	48.21 ± 2.54	< 0.001	28.54 ± 2.46	39.75 ± 2.46	50.16 ± 2.50	< 0.001
Dairy	312.81 ± 20.03	293.76 ± 20.19	337.05 ± 20.36	0.32	296.88 ± 19.93	343.40 ± 19.96	303.27 ± 20.23	0.21
Nut, soy, and legumes	44.78 ± 2.84	54.43 ± 2.86	53.91 ± 2.89	0.03	48.45 ± 2.84	53.61 ± 2.84	50.89 ± 2.88	0.44

¹Values are presented as mean ± SE. Intakes of energy and macronutrients were adjusted for age and sex, all other values were adjusted for age, sex, and energy intake.

²Obtained from ANCOVA.

n, number; T, tertile; g, grams; PRAL, potential renal acid load; NEAP, net endogenous acid production.

Figure 1 illustrates prevalence of having short sleeping and poor sleep quality across tertiles of PRAL and NEAP. The prevalence of short sleeping was marginally higher in the third tertile of NEAP and PRAL (P = 0.05), but no differences were observed across the tertiles of the indices for poor sleep quality (P > 0.05).

Fig. 1.

Prevalence of poor sleep quality and short sleeping across tertiles of PRAL and NEAP. P-values were obtained via chi-square test.

Multivariable-adjusted results for PRAL and NEAP in relation to odds of short sleeping and poor sleep quality are presented in Table 3. In addition, Supplemental Tables 2 and 3, respectively, present multivariable-adjusted results for DAL and protein/potassium ratio in relation to odds of short sleeping and poor sleep quality. Based on fully-adjusted model, participants in the third tertile of PRAL had 72% higher odds of poor sleep quality than those in the first tertile (OR = 1.72; 95% CI: 1.02, 2.90). Moreover, each tertile increase in PRAL was associated with a 31% higher likelihood of poor sleep quality (OR = 1.31; 95% CI: 1.01, 1.70). Similar results were obtained for NEAP; participants in the third tertile were 77% more likely to have poor sleep quality (OR = 1.77; 95% CI: 1.05, 3.00), and each tertile increase was associated with a 33% increase in the odds (OR = 1.33; 95% CI: 1.02, 1.72), based on fully-adjusted model. Participants in the third tertile of DAL and the protein/potassium ratio were, respectively, 95% (OR = 1.95; 95% CI: 1.13, 3.40) and 77% (OR = 1.77; 95% CI: 1.05, 3.00) more likely to have inappropriate sleep quality than those in their reference tertile, based on fully-adjusted analysis. Each tertile increase in DAL and protein/potassium ratio was additionally associated with 40% (OR = 1.40; 95% CI: 1.06, 1.85) and 33% (OR = 1.33; 95% CI: 1.02, 1.72) higher odds of poor sleep quality, respectively.

Table 3.

Multivariable adjusted odds ratio (OR) and 95% confidence interval (CI) for sleep poor quality and short duration across tertiles of PRAL and NEAP1.

	PRAL					NEAP
	T1 (n = 179)	T2 (n = 178)	T3 (n = 179)	Each tertile increase2	Ptrend2	T1 (n = 179)	T2 (n = 178)	T3 (n = 179)	Each tertile increase2	Ptrend2
Index	< −18.9	−18.9 to −2.9	> −2.9			< 31.1	31.1 to 41.8	> 41.8
Poor sleep quality
Cases (n)	44	49	60			41	53	59
Crude	1.00	1.17 (0.73, 1.87)	1.55 (0.98, 2.46)	1.25 (0.99, 1.57)	0.06	1.00	1.43 (0.89, 2.30)	1.65 (1.04, 2.65)	1.28 (1.02, 1.62)	0.04
Model 1	1.00	1.30 (0.79, 2.13)	1.80 (1.11, 2.95)	1.34 (1.06, 1.72)	0.02	1.00	1.57 (0.96, 2.57)	1.94 (1.19, 3.18)	1.39 (1.09, 1.77)	0.01
Model 2	1.00	1.42 (0.84, 2.40)	1.71 (1.02, 2.90)	1.31 (1.01, 1.70)	0.04	1.00	1.55 (0.94, 2.60)	1.77 (1.05, 3.00)	1.33 (1.02, 1.72)	0.03
Model 3	1.00	1.42 (0.85, 2.40)	1.72 (1.02, 2.90)	1.31 (1.01, 1.70)	0.04	1.00	1.55 (0.93, 2.60)	1.77 (1.05, 3.00)	1.33 (1.02, 1.72)	0.03
Short sleeping
Cases (n)	36	44	56			35	46	55
Crude	1.00	1.30 (0.79, 2.16)	1.81 (1.12, 2.95)	1.35 (1.06, 1.72)	0.02	1.00	1.43 (0.87, 2.37)	1.82 (1.13, 2.99)	1.35 (1.06, 1.72)	0.02
Model 1	1.00	1.56 (0.92, 2.64)	2.08 (1.25, 3.48)	1.44 (1.12, 1.85)	0.01	1.00	1.67 (1.00, 2.82)	2.06 (1.25, 3.46)	1.43 (1.11, 1.84)	0.01
Model 2	1.00	1.77 (1.02, 3.10)	2.43 (1.42, 4.24)	1.55 (1.19, 2.04)	0.001	1.00	2.08 (1.21, 3.63)	2.58 (1.49, 4.53)	1.58 (1.21, 2.08)	< 0.001
Model 3	1.00	1.79 (1.03, 3.15)	2.47 (1.44, 4.32)	1.56 (1.19, 2.06)	0.001	1.00	2.07 (1.20, 3.61)	2.58 (1.49, 4.53)	1.58 (1.21, 2.08)	< 0.001

¹All values are odds ratios and 95% confidence intervals. Boldness indicates statistical significance (P < 0.05).

Model 1: Adjusted for age, sex, and total energy intake.

Model 2: Additionally adjusted for marital status, education level, physical activity (HEPA), smoking status, antidepressants (nortriptyline, amitriptyline, imipramine, fluoxetine, citalopram, and fluvoxamine) usage, and socio-economic status, obstructive sleep apnea, and tea or coffee drinking after 8:00 PM.

Model 3: Additionally adjusted for body mass index (BMI).

²Odds ratios and P_trend were obtained by treating tertiles as a continuous variable.

n, number; T, tertile; PRAL, potential renal acid load; NEAP, net endogenous acid production.

Based on fully adjusted model, participants in the second tertile of PRAL were 79% (OR = 1.79; 95% CI: 1.03, 3.15) more likely to experience short sleeping (< 6 h of night sleep) than those in the first tertile. Those in the third tertile were even more likely to be short sleepers, with an OR of 2.47 (95% CI: 1.44, 4.32) for sleeping less than six hours overnight. Each PRAL tertile increase was associated with 56% higher odds of short sleeping (OR = 1.56; 95% CI: 1.19, 2.06), as well. Similarly, participants in the second and third tertiles of NEAP and protein/potassium were 107% (OR = 2.07; 95% CI: 1.20, 3.61) and 158% (OR = 2.58; 95% CI: 1.49, 4.53) more likely to be short sleepers, based on fully-adjusted model, compared to those in the first tertile. Each tertile increase in NEAP and protein/potassium ratio was associated with a 58% increase in odds of short sleep (OR = 1.58; 95% CI: 1.21, 2.08). Participants in the second and third tertiles of DAL had 76% (OR = 1.76; 95% CI: 1.02, 3.09) and 149% (OR = 2.49; 95% CI: 1.41, 4.45) higher odds of short sleeping, with a trend OR of 1.54 (95% CI: 1.19, 2.09), based on fully-adjusted model.

Supplemental Table 4 presents the results for a sensitivity analysis of the association between the indices and odds of sleeping shorter than 7 h. As shown, though the results weakened based on this nighttime sleep duration cut point, participants in the third tertiles of PRAL (OR = 1.56; 95% CI: 0.94, 2.60), NEAP (OR = 1.64; 95% CI: 1.00, 2.71), DAL (OR = 1.70; 95% CI: 1.00, 2.92), and protein/potassium (OR = 1.64; 95% CI: 1.00, 2.71) had marginally higher odds of short sleeping, compared the ones in reference. Moreover, those in the second tertile of NEAP and protein/potassium had a significantly higher odds of short sleeping (less than 7 h) (OR = 1.71; 95% CI: 1.06, 2.79). There was also a significant trend for NEAP and protein/potassium (P _trend = 0.04).

BMI-stratified multivariable adjusted OR and 95% CI for inappropriate quality and duration of sleep, across tertiles of PRAL and NEAP, are shown in Table 4. There was no significant association between PRAL and NEAP with poor sleep quality among participants with a normal weight (P > 0.05). Participants with overweight/obesity in the third tertile of NEAP were 110% more likely to have poor sleep quality than those in the first tertile, based on fully-adjusted model (OR = 2.10; 95% CI: 1.12, 4.00). Moreover, each tertile increase in NEAP was associated with a 45% higher odds of short sleeping (OR = 1.45; 95% CI: 1.06, 1.99). With regard to PRAL index, although adjusting for sex, age, and energy intake resulted in a significant association between this index and odds of poor sleep quality (OR = 1.79; 95% CI: 1.01, 3.23) in adults with overweight/obesity in the third tertile, this result did not remain significant after adding other variables included in the second model (OR = 1.69; 95% CI: 0.91, 3.18). Based on fully-adjusted model, adults in the third tertile of both PRAL (OR = 3.62; 95% CI: 1.20, 12.0) and NEAP (OR = 4.66; 95% CI: 1.46, 16.7), with normal weight, were more likely to have insufficient sleep compared to those in the first tertile. The same pattern was evident among those in the second tertile of NEAP (OR = 3.96; 95% CI: 1.25, 14.0) compared to the first tertile. Each PRAL and NEAP tertile increase was associated with 88% (OR = 1.88; 95% CI: 1.09, 3.33) and 99% (OR = 1.99; 95% CI: 1.15, 3.57) higher odds of having short sleeping, respectively, among participants with normal weight. Moreover, participants with overweight/obesity in the third tertile of PRAL (OR = 2.27; 95% CI: 1.19, 4.41) and NEAP (OR = 2.29; 95% CI: 1.21, 4.46) were more likely to have insufficient sleep than those in the first tertile. Ultimately, each tertile increase in PRAL (OR = 1.51; 95% CI: 1.09, 2.10) and NEAP (OR = 1.51; 95% CI: 1.10, 2.10) was associated with a 51% increase in odds of short sleeping among adults with overweight/obesity.

Table 4.

Multivariable adjusted odds ratio (OR) and 95% confidence interval (CI) for sleep poor quality and short duration, stratified by BMI, across tertiles of PRAL and NEAP1.

	PRAL					NEAP
	T1	T2	T3	Each tertile increase2	Ptrend2	T1	T2	T3	Each tertile increase2	Ptrend2
Index	< −18.9	−18.9 to −2.9	> −2.9			< 31.1	31.1 to 41.8	> 41.8
Poor sleep quality
Normal-weight
Case/participants (n)	10/51	16/60	18/61			12/55	16/59	16/58
Crude	1.00	1.49 (0.61, 3.76)	1.72 (0.72, 4.28)	1.30 (0.84, 2.01)	0.24	1.00	1.33 (0.57, 3.20)	1.37 (0.58, 3.28)	1.16 (0.76, 1.79)	0.49
Model 1	1.00	1.59 (0.63, 4.16)	1.91 (0.78, 4.90)	1.37 (0.88, 2.15)	0.17	1.00	1.40 (0.57, 3.47)	1.50 (0.62, 3.72)	1.22 (0.78, 1.90)	0.38
Model 2	1.00	1.96 (0.71, 5.56)	2.07 (0.75, 5.94)	1.41 (0.86, 2.35)	0.19	1.00	1.39 (0.53, 3.75)	1.30 (0.47, 3.60)	1.13 (0.69, 1.87)	0.62
Overweight/obese
Case/participants (n)	34/128	33/118	41/118			29/124	37/119	43/121
Crude	1.00	1.07 (0.61, 1.88)	1.53 (0.89, 2.64)	1.24 (0.94, 1.63)	0.13	1.00	1.48 (0.84, 2.62)	1.81 (1.04, 3.18)	1.34 (1.02, 1.77)	0.04
Model 1	1.00	1.19 (0.66, 2.15)	1.79 (1.01, 3.23)	1.34 (1.00, 1.80)	0.05	1.00	1.61 (0.90, 2.93)	2.13 (1.18, 3.90)	1.46 (1.09, 1.96)	0.01
Model 2	1.00	1.22 (0.65, 2.28)	1.69 (0.91, 3.18)	1.30 (0.95, 1.79)	0.10	1.00	1.65 (0.89, 3.08)	2.10 (1.12, 4.00)	1.45 (1.06, 1.99)	0.02
Short sleeping
Normal-weight
Case/participants (n)	7/51	14/60	18/61			7/56	15/59	17/58
Crude	1.00	1.91 (0.72, 5.46)	2.63 (1.03, 7.35)	1.59 (1.01, 2.56)	0.05	1.00	2.34 (0.90, 6.62)	2.84 (1.11, 7.98)	1.62 (1.03, 2.60)	0.04
Model 1	1.00	2.30 (0.83, 6.91)	3.03 (1.14, 8.85)	1.67 (1.05, 2.74)	0.03	1.00	2.84 (1.04, 8.45)	3.26 (1.22, 9.53)	1.70 (1.07, 2.77)	0.03
Model 2	1.00	2.18 (0.69, 7.37)	3.62 (1.20, 12.0)	1.88 (1.09, 3.33)	0.03	1.00	3.96 (1.25, 14.0)	4.66 (1.46, 16.7)	1.99 (1.15, 3.57)	0.02
Overweight/obese
Case/participants (n)	29/128	30/118	38/118			28/124	31/119	38/121
Crude	1.00	1.16 (0.65, 2.10)	1.62 (0.92, 2.87)	1.28 (0.96, 1.70)	0.09	1.00	1.21 (0.67, 2.18)	1.57 (0.89, 2.79)	1.25 (0.94, 1.67)	0.12
Model 1	1.00	1.37 (0.75, 2.54)	1.87 (1.03, 3.44)	1.37 (1.01, 1.85)	0.04	1.00	1.38 (0.76, 2.55)	1.78 (0.98, 3.28)	1.33 (0.99, 1.81)	0.06
Model 2	1.00	1.60 (0.84, 3.08)	2.27 (1.19, 4.41)	1.51 (1.09, 2.10)	0.01	1.00	1.71 (0.91, 3.26)	2.29 (1.21, 4.46)	1.51 (1.10, 2.10)	0.01

¹All values are odds ratios and 95% confidence intervals. Boldness indicates statistical significance (P < 0.05).

Model 1: Adjusted for age, sex, and total energy intake.

Model 2: Additionally adjusted for marital status, education level, physical activity (HEPA), smoking status, antidepressants (nortriptyline, amitriptyline, imipramine, fluoxetine, citalopram, and fluvoxamine) usage, socio-economic status, obstructive sleep apnea, and tea or coffee drinking after 8:00 PM.

² Odds ratios and P_trend were obtained by treating tertiles as a continuous variable.

n, number; T, tertile; PRAL, potential renal acid load; NEAP, net endogenous acid production.

Supplemental Table 5 presents interactions between PRAL and NEAP with BMI in relation to poor sleep quality and short sleeping. Interactions were significant between PRAL and NEAP with BMI in relation to odds of short sleeping (P _interaction = 0.03), but not in relation to poor sleep quality (P _interaction > 0.05).

Results from multivariable-adjusted analyses comparing odds of sleep quality domain across tertiles of PRAL and NEAP are shown in Table 5. Participants in the second and third tertiles of PRAL had, respectively, 155% (OR = 2.55; 95% CI: 1.22, 5.49) and 268% (OR = 3.68; 95% CI: 1.74, 8.11) higher odds of sleep latency than those in the first tertile, according to the fully-adjusted model. Moreover, each tertile increase in PRAL was associated with 88% higher odds of sleep latency (OR = 1.88; 95% CI: 1.31, 2.76). Also, participants in the third tertile of NEAP had significantly higher odds of sleep latency than those in the first tertile (OR = 3.39; 95% CI: 1.67, 7.13). Similar to PRAL, each increase in NEAP tertile was associated with 88% higher odds of sleep latency (OR = 1.88; 95% CI: 1.31, 2.75). Participants in the second tertile of PRAL and NEAP were, respectively, 65% (OR = 1.65; 95% CI: 1.03, 2.68) and 75% (OR = 1.75; 95% CI: 1.09, 2.83) more likely to have low sleep duration than the respective reference. Each PRAL and NEAP tertile increase was associated with 29% (OR = 1.29; 95% CI: 1.01, 1.64) and 32% (OR = 1.32; 95% CI: 1.04, 1.68) higher odds of low sleep duration, respectively. Ultimately, crude analysis showed a significant association between daytime dysfunction for participants in the second tertile of NEAP (OR = 1.85; 95% CI: 1.01, 3.45); however, the association did not remain significant after adjustments for potential confounders (OR = 1.58; 95% CI: 0.82, 3.09). No other significant associations were observed between PRAL or NEAP and sleep quality domains.

Table 5.

Multivariable-adjusted odds ratio for individual domains of sleep quality across tertiles of PRAL and NEAP1.

	PRAL					NEAP
	T1 (n = 179)	T2 (n = 178)	T3 (n = 179)	Each tertile increase2	Ptrend2	T1 (n = 179)	T2 (n = 178)	T3 (n = 179)	Each tertile increase2	Ptrend2
Index	< −18.9	−18.9 to −2.9	> −2.9			< 31.1	31.1 to 41.8	> 41.8
Subjective sleep quality
Cases (n)	15	21	22			15	22	21
Crude	1.00	1.46 (0.73, 2.99)	1.53 (0.77, 3.11)	1.23 (0.88, 1.72)	0.23	1.00	1.54 (0.78, 3.13)	1.45 (0.73, 2.97)	1.19 (0.85, 1.67)	0.31
Fully-adjusted model	1.00	1.48 (0.69, 3.22)	1.66 (0.76, 3.70)	1.28 (0.87, 1.90)	0.63	1.00	1.36 (0.65, 2.90)	1.41 (0.65, 3.10)	1.18 (0.81, 1.74)	0.31
Sleep latency
Cases (n)	16	25	30			18	20	33
Crude	1.00	1.66 (0.86, 3.29)	2.05 (1.09, 4.00)	1.41 (1.04, 1.94)	0.03	1.00	1.13 (0.58, 2.24)	2.02 (1.10, 3.81)	1.45 (1.06, 1.99)	0.02
Fully-adjusted model	1.00	2.55 (1.22, 5.49)	3.68 (1.74, 8.11)	1.88 (1.31, 2.76)	< 0.001	1.00	1.29 (0.62, 2.72)	3.39 (1.67, 7.13)	1.88 (1.31, 2.75)	< 0.001
Sleep duration
Cases (n)	64	71	78			62	73	78
Crude	1.00	1.19 (0.78, 1.83)	1.39 (0.91, 2.13)	1.18 (0.95, 1.46)	0.13	1.00	1.31 (0.85, 2.02)	1.46 (0.95, 2.24)	1.21 (0.98, 1.49)	0.08
Fully-adjusted model	1.00	1.33 (0.83, 2.14)	1.65 (1.03, 2.68)	1.29 (1.01, 1.64)	0.04	1.00	1.55 (0.97, 2.47)	1.75 (1.09, 2.83)	1.32 (1.04, 1.68)	0.02
Habitual sleep efficiency 3
Cases (n)	1	1	0			0	2	0
Crude	1.00	–	–	–	–	1.00	–	–	–	–
Fully-adjusted model	1.00	–	–	–	–	1.00	–	–	–	–
Sleep disturbances
Cases (n)	18	18	20			16	21	19
Crude	1.00	1.01 (0.50, 2.01)	1.13 (0.57, 2.22)	1.06 (0.76, 1.49)	0.73	1.00	1.36 (0.69, 2.74)	1.21 (0.60, 2.46)	1.09 (0.78, 1.54)	0.60
Fully-adjusted model	1.00	1.18 (0.55, 2.55)	1.17 (0.54, 2.56)	1.08 (0.73, 1.60)	0.69	1.00	1.53 (0.73, 3.28)	1.14 (0.52, 2.56)	1.07 (0.73, 1.57)	0.73
Use of sleep medications
Cases (n)	6	6	6			8	4	6
Crude	1.00	1.01 (0.31, 3.27)	1.00 (0.31, 3.25)	1.00 (0.56, 1.79)	0.99	1.00	0.49 (0.13, 1.59)	0.74 (0.24, 2.18)	0.84 (0.46, 1.50)	0.56
Fully-adjusted model	1.00	1.97 (0.50, 8.25)	1.08 (0.24, 4.69)	1.04 (0.53, 2.06)	0.90	1.00	0.63 (0.15, 2.33)	0.73 (0.17, 2.79)	0.84 (0.41, 1.67)	0.61
Daytime dysfunction
Cases (n)	22	30	21			19	32	22
Crude	1.00	1.45 (0.80, 2.65)	0.95 (0.50, 1.80)	0.98 (0.72, 1.32)	0.88	1.00	1.85 (1.01, 3.45)	1.18 (0.61, 2.28)	1.07 (0.79, 1.46)	0.64
Fully-adjusted model	1.00	1.40 (0.73, 2.70)	0.78 (0.37, 1.61)	0.90 (0.63, 1.27)	0.24	1.00	1.58 (0.82, 3.09)	1.06 (0.52, 2.20)	1.03 (0.73, 1.46)	0.88

¹All values are odds ratios and 95% confidence intervals. Boldness indicates statistical significance (P < 0.05).

Fully-adjusted model: Adjusted for age, sex, and total energy intake, marital status, education level, physical activity (HEPA), smoking status, antidepressants (nortriptyline, amitriptyline, imipramine, fluoxetine, citalopram, and fluvoxamine) usage, socio-economic status, obstructive sleep apnea, and tea or coffee drinking after 8:00 PM, and body mass index (BMI).

² Odds ratios and P_trend were obtained by treating tertiles as a continuous variable.

³OR (95% CI) could not be calculated, due to lack of cases in a tertile.

n, number; T, tertile; PRAL, potential renal acid load; NEAP, net endogenous acid production.

Figure 2 illustrates non-linear association between oxidative stress biomarkers (MDA, GPx, and SOD) and adjusted odds of poor sleep quality and short sleep duration, based on fully-adjusted model. The knots were at 37.280, 124.194, 207.404, and 295.799 nmol/mL for MDA; 0.438, 0.848, 1.187, and 1.997 Units for SOD; and 0.130, 0.530, 1.070, and 9.581 mU/mL for GPx. There was a significant non-linear association between serum MDA and multivariable-adjusted odds of poor sleep quality, such that an MDA in the range of 115–180 nmol/mL was marginally associated with lower odds of poor sleep quality. In contrast, an MDA level more than 180 nmol/mL was significantly associated with higher odds of poor sleep quality. A GPx level lower than 0.70 mU/mL was significantly associated with higher odds of poor sleep quality. There was no statistically significant association between SOD level and poor sleep quality (P > 0.05). Moreover, no significant associations were observed between the adjusted odds of short sleeping and MDA, GPx, or SOD (P > 0.05). As the observations at two ends were fewer, CIs were wider at both ends of all analyses.

Fig. 2.

Non-linear association between oxidative stress biomarkers (MDA, GPx, and SOD) and poor sleep quality and short sleeping. Values are adjusted odds ratios and 95% confidence intervals for poor sleep quality and short sleeping; adjustments included age, sex, marital status, education level, physical activity (HEPA), smoking status, antidepressants usage, SES, OSA, drinking tea or coffee after 8:00 PM, and BMI. P values present overall associations.

Supplemental Fig. 1 visualizes results for a sensitivity analysis on the non-linear association between MDA, GPx, and CRP with odds of abnormal sleep parameters, excluding those with extreme values. In total, 30 extreme values for MDA, 20 for GPx, and 28 for SOD (out of the range of 2–98 percentiles) were excluded. No significant difference occurred when excluding these values, suggesting the earlier results were independent of outliers.

Discussion

Our findings demonstrated that adults with higher dietary acid load indices (NEAP, PRAL, DAL, and protein/potassium ratio) were more likely to have short sleeping and to experience low-quality sleep. Moreover, serum MDA and GPx were non-linearly associated with poor sleep quality; such that individuals with higher MDA levels and those with low GPx levels had higher odds of inappropriate sleep quality. These findings were independent of potential confounders.

A sensitivity analysis suggested that the association between dietary acid load may be weaker with odds of sleeping less than 7 h (as compared to the relation with less than 6 h overnight). It should be noted that PSQI asks for overnight sleep and does not consider daytime naps. Previous studies suggest that the sleeping less than 6 h overnight may be better related to abnormalities than 7 h²⁸.

Classifying the participants into two groups of those with a normal weight (BMI < 25 kg/m²) and those with overweight/obesity (BMI ≥ 25 kg/m²), suggested that higher PRAL and NEAP values were only associated with poor sleep quality in those with overweight/obesity, not in those with a normal weight. Excessive adipose tissue, especially in the abdominal region, can apply pressure to the diaphragm and impair normal respiratory function, leading to poor-quality sleep³³. Moreover, metabolic changes related to adiposity (including inflammation and insulin resistance) may be related to one’s usual circadian rhythm and, in turn, sleep quality³³. Crucially, these adiposity-related factors may synergistically act with dietary acid load. In individuals with overweight/obesity, low-grade adipose tissue inflammation activates immune cells in a way that metabolic acidosis induced by dietary acid load triggers a larger inflammatory response³⁴. Similarly, insulin resistance impairs renal net acid excretion, leading to higher systemic acid retention, amplifying cortisol dysregulation and circadian disruption³⁵. These mechanisms likely explain why the association between dietary acid load and poor sleep quality was only significant in the overweight/obese subgroup. Higher levels of PRAL and NEAP were associated with short sleeping in adults with both normal weight and overweight/obesity. This association appeared to be stronger in normal-weight individuals. Although this stronger association was accompanied by a wide confidence interval due to fewer normal-weight participants, it calls for caution in interpreting the results. Moderate and high versus low PRAL values were associated with a higher probability of sleep latency. Participants with high NEAP were also more likely to experience sleep latency than those with a low NEAP. Moreover, the odds of low sleep duration were higher among adults with high PRAL and NEAP values. This suggested that a more acidic diet may be associated with higher odds of sleep latency and short beneficial sleep duration.

As our findings suggest that a higher acid load may be strongly associated with short, low-quality sleep, dietitians could lower the acidity of their clients’ diets, especially for those experiencing sleep problems or those with daytime activities that require high concentration. But these findings are purely from a cross-sectional investigation and naturally unable to detect causal relations. Therefore these findings should be interpreted with caution, and future prospective studies should evaluate the findings before making recommendations. Dietary acid load could be reduced by increasing fruit and vegetable consumption and total carbohydrate intake conservatively.

Mahjourian et al. evaluated the association between dietary acid load indices and sleep quality in a cross-sectional study among Iranian elders¹⁷. Their results showed that PRAL, NEAP, and DAL were significantly associated with odds of poor sleep quality in elderly individuals, similar to our results, with similar effect sizes in most cases. Except for NEAP in relation to sleep quality, where the effect size was larger (with also wider CIs) compared to ours, possibly because of different adjustments or fewer cases. The mentioned study did not assess short sleeping as a crucial sleep parameter. Moreover, the study did not evaluate the potential association between dietary acid load indices and each sleep quality domain. It is also worth noting that their study sample differed from our research, and their results were not generalizable to the total population. Similarly, another cross-sectional study by Daneshzad et al. found a significant association between PRAL and NEAP with poor sleep quality among Iranian diabetic women¹⁸. Though their effect size was larger, it came with significantly wider confidence intervals, possibly because of their smaller sample. Again, generalization was limited because the sample was restricted to diabetic women. Again, in the mentioned investigation, dietary acid load was not assessed in relation to sleep duration or domains of sleep quality. In contrast, another cross-sectional study by Katagiri et al. among middle-aged Japanese female workers found no significant association between total carbohydrate intake (as percentage of energy and as an alkaline diet component) and odds of poor sleep quality¹⁹. This contradiction likely stems from the fact that the last-mentioned study assessed each dietary item against odds of poor sleep quality. Hence, it did not consider the total effect that diet could have on sleep quality. There was no study assessing the association between oxidative stress biomarkers and sleep quality or duration in a population similar to that of the present study.

A potential mechanism by which dietary acid load may impair sleep involves the induction of mild metabolic acidosis³⁶. Diets high in acid load can lower systemic pH, triggering compensatory responses including activation of the hypothalamic-pituitary-adrenal (HPA) axis and elevated cortisol secretion³⁷. Cortisol dysregulation is a known disruptor of circadian rhythms and sleep architecture³⁴. In parallel to that, diets with higher acid load are usually high in processed foods, meat, and low in fiber, causing systemic inflammation by elevating inflammatory biomarkers such as tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), and C-reactive protein (CRP)^13,38–40. Oxidative stress is also interconnected with systemic inflammation as it can activate immune cells (such as neutrophils, monocytes, and macrophages), leading to the release of proinflammatory molecules⁴¹. This systemic inflammation may compromise the blood-brain barrier (BBB) and allow immune cells and cytokines to activate microglia and other glial cells in the brain, leading to neuroinflammation and neurotoxicity, disrupting sleep⁴².

The current investigation was the first study to comprehensively examine the associations of all four dietary acid load indices and oxidative stress biomarkers with sleep parameters among the general population of Iranian adults. However, as a cross-sectional study by nature cannot detect causal relations, future cohort studies are required to evaluate our findings prospectively. In parallel, clinical trials involving alkali-rich diets’ effects on sleep parameters can even better evaluate the findings in a clinical setting. Moreover, the study sample was limited to a large but single-city school staff. Although all sets of school jobs were included, generalizing the findings to the total population may not be fully possible. Sample size was calculated for the primary outcome; hence, the study did not have enough power to detect interaction. Besides, the number of cases was too few in the PSQI singular domains and BMI-stratified analyses, resulting in wide CIs in those analyses. We adjusted for multiple confounders, though unknown and residual confounders (e.g., depression, anxiety, perceived stress, shift work, chronotype, renal function, type 2 diabetes, hypertension, inflammatory markers, work shift, screen time, and exact macronutrient quality) were not adjusted for, due to the limitation in the number of confounders or lack of measurement. Although a validated FFQ was used for dietary assessment of participants, recall bias was not entirely unavoidable with this method. The same case was present for PSQI; though it had been validated among Iranians, it is still a self-report questionnaire, and future studies should apply objective tools, such as actigraphy or polysomnography, to better evaluate sleep quality or duration. And finally, DAL indices were FFQ-driven and not validated in this sample against urinary net acid excretion or urine pH (as we did not have access to participants’ 24-h urinary samples); it would be highly beneficial for the findings, and future studies should consider validating these indices in their samples if calculating DAL indices using dietary data.

In conclusion, individuals with a higher dietary acid load had higher odds of low sleep quality and short sleeping in a dose-response manner. In those with overweight/obesity, particularly, dietary acid load was even more associated with poor sleep quality. Moreover, higher MDA and lower GPx (as oxidative stress biomarkers) were non-linearly associated with higher odds of poor sleep quality. It should also be noted that reverse causation cannot be ruled out, inherent to a cross-sectional study, and altered sleep parameters, elevated oxidative stress biomarkers, and increased consumption of acidogenic comfort foods can be both cause and consequence. Future prospective studies should evaluate the findings to detect causal relations.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ELISA

Enzyme-linked immunosorbent assay

IPAQ

International Physical Activity Questionnaire

ANOVA

Analysis of variance

ANCOVA

Analysis of covariance

GPx

Glutathione peroxidase

MDA

Malondialdehyde

SOD

Superoxide dismutase

RCS

Restricted cubic splines

PRAL

Potential renal acid load

NEAP

Net endogenous acid production

DAL

Dietary acid load

Author contributions

M.M., Z.M., P.R., F.S., and P.S. contributed to conception, design, data collection and interpretation, manuscript drafting, data analyses, approval of the manuscript, and agreed on all aspects of the work.

Funding

The financial support for this study is provided by the Nutrition and Food Security Research Center, Isfahan University of Medical Sciences, Isfahan, Iran.

Data availability

The data supporting this study’s findings are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki and the STROBE protocols. The local Ethics Committee of Isfahan University of Medical Sciences approved the study protocol. All study participants were asked to provide written informed consent.

Consent for publication

All the authors agreed to submit the manuscript to the journal. No third-party material requiring approval was used in the preparation of the current paper.