Sleep disturbances and psychological disorders including stress, anxiety, aggression and depression are common complications in diabetic patients. Diabetic patients have poor sleep quality, with poorer sleep efficiency and sleep disorders compared with non-diabetics(Reference Sridhar and Madhu1–Reference Budhiraja, Roth and Hudgel3). Moreover, anxiety and depression are more prevalent in diabetic patients than individuals with normal glucose tolerance(Reference Shinkov, Borissova and Kovatcheva4). Approximately one-third of type 2 diabetic patients have sub-threshold depression(Reference Bot, Pouwer and Ormel5) and over 40 % present with minor or major depression with anxiety disorders(Reference Golden, Shah and Naqibuddin6).
Diet is an important and modifiable environmental factor, which can affect glycaemic control as well as psychological disorders and sleep disturbances. For example, low glycaemic index foods which contain high dietary fibre are associated with lower blood glucose(Reference Valachovicova, Krajcovicova-Kudlackova and Blazicek7). Moreover, Zn, Mg and B-vitamins, which are abundant in vegetables, are associated with decreased risk of depression(Reference Murakami, Mizoue and Sasaki8,Reference Jacka, Mykletun and Berk9) . Based on previous studies, these micronutrients can also lead to an improved sleep quality(Reference Ji and Liu10,Reference Rondanelli, Opizzi and Monteferrario11) . Today, according to interaction and combination of various macro- and micronutrients in a whole diet, epidemiological studies are conducted to determine the association of dietary patterns and dietary quality indices and various diseases. Protective dietary patterns with the content of nuts, fruits and vegetables are associated with reduced risk of depression(12). Plant-based diet index (PDI) and dietary acid load (DAL) are two indices to assess whole diet quality. As nutrients which present in vegetable and fruits have indicated good effects on sleep and psychological status, the present study hypothesised that whether a whole plant diet has a reduced association with sleep disturbances and psychological disorders. Based on previous studies in diabetic patients, DAL was positively associated with type 2 diabetes risk(Reference Fagherazzi, Vilier and Bonnet13) and the metabolic syndrome(Reference Iwase, Tanaka and Kobayashi14). Also, women with higher DAL scores had increased risk of gestational diabetes mellitus(Reference Saraf-Bank, Tehrani and Haghighatdoost15). Furthermore, metabolic acidosis leads to reduced insulin secretion and induces insulin resistance(Reference Saraf-Bank, Tehrani and Haghighatdoost15).
Therefore, the aim of this cross-sectional study was to determine the association of a PDI and DAL with sleep status, anxiety, stress and depression in type 2 diabetic women.
Methods
This cross-sectional study was carried out in 230 type 2 diabetic women who were referred to diabetes research or health centres in Tehran, Iran. The sample size was determined based on the mean of psychological distress score in patients with lowest (15·98 (sd 9·72)) and highest (12·38 (sd 9·95)) adherence to the PDI, respectively: α 0·05 and β 0·2 (power 80 %). Therefore, in the present study, the sample size was determined about 115 individuals in each group(Reference Zamani, Daneshzad and Siassi16). Participants were randomly selected and provided informed written consent. Only women with type 2 diabetes without other medical complications were included in the present study. Women who had other chronic diseases such as thyroid problems, cancers, CVD and kidney disease were excluded. In addition, individuals who reported a total energy intake of <3347·2 and >17 572·8 kJ were excluded. The present study was approved by the ethical committee of Tehran University of Medical Sciences (96-03-161-36923).
Assessment of anthropometric measures
Body weight was measured in minimal clothing using a calibrated digital scale (SECA 803). Height was measured while participants were in standing position using an unstretched tape measure to the nearest to 0·1 mm. BMI was calculated as weight divided by height squared (kg/m2). Waist circumference was measured using an unstretched tape measure to the nearest to 0·1 cm at the narrowest cite of the waist in light clothing.
Assessment of dietary intake
A 168-item semi-quantitative FFQ, which was validated and reliable for the Iranian population, was used to determine past year dietary intakes(Reference Azadbakht and Esmaillzadeh17). All participants filled the amount and frequency of consumption of each food item on a daily, weekly or monthly basis during the past year. The reported portion sizes of consumed foods converted to g/d. Nutritionist IV software (version 7.0; N-Squared Computing) which was adapted for Iranian foods was used for nutrients analysis.
Plant-based diet indices
We created a whole PDI, healthful PDI (hPDI) and unhealthful PDI (uPDI) as reported in previous studies(Reference Martinez-Gonzalez, Sanchez-Tainta and Corella18,Reference Satija, Bhupathiraju and Rimm19) . Briefly, eighteen food groups are created and classified into three main categories: animal foods, and healthy and unhealthy plant foods. Healthy food groups included fruits, vegetables, whole grains, legumes, vegetable oils, nuts, tea and coffee, whereas less healthy food groups included sugar-sweetened beverages, refined grains, fruit juices, potatoes, sweets and desserts. Animal food groups included dairy products, eggs, animal fats, fish and seafood, poultry and red meat, and miscellaneous animal-based foods. Food grouping details are shown in online Supplementary Table S1. These eighteen food groups were ranked into quintiles and given scores between 1 and 10. For creating PDI, the highest decile of a food group received a score of 10 and the lowest decile received a score of 1.
Participants in the highest decile of animal food groups received a score of 1 and to the lowest decile received a score of 10. For hPDI, healthy plant food groups received a positive score, whereas scores were reversed for unhealthy plant food groups and animal food groups. For uPDI, a score of 1 was given to the lowest decile of less healthy plant food groups and 10 for the highest decile, whereas reverse scores were attributed to healthy plant and animal food groups (online Supplementary Table S1)(Reference Satija, Bhupathiraju and Rimm19). Finally, all eighteen food group scores were summed (range 18–180) to attain the indices score, which indicates that a higher intake of all three indices reflected lower animal food intake. Finally, these three indices were categorised into two groups by median-split to assess their association with dependent measures.
Assessment of dietary acid load
DAL was calculated based on potential renal acid load (PRAL)(Reference Remer and Manz20,Reference Remer, Dimitriou and Manz21) and net endogenous acid production (NEAP) method(Reference Frassetto, Todd and Morris22).
$${\rm{PRAL }}\;\left( {{\rm{mEq/d}}} \right)\, = \,\left( {{\rm{protein }}\;\left( {{\rm{g/d}}} \right)\, \times \,0\!\cdot\!49} \right)\, + \,\left( {{\rm{P }}\;\left( {{\rm{mg/d}}} \right)\, \times \,0\!\cdot\!037} \right)\,-\,\left( {{\rm{K }}\;\left( {{\rm{mg}}/{\rm{d}}} \right)\, \times \,0\!\cdot\!021} \right)\,-\,\left( {{\rm{Ca }}\;\left( {{\rm{mg}}/{\rm{d}}} \right)\, \times \,0\!\cdot\!013} \right)\,-\,\left( {{\rm{Mg }}\;\left( {{\rm{mg}}/{\rm{d}}} \right)\, \times \,0\!\cdot\!026} \right).$$
$${\rm{Estimated \ NEAP }\;}\left( {{\rm{mEq/d}}} \right)\, = \,\left( {54\!\cdot\!5\, \times \,{\rm{protein \;intake }}\;\left( {{\rm{g/d}}} \right)/{\rm{K \;intake }}\;\left( {{\rm{mEq/d}}} \right)} \right)\,-\,10\!\cdot\!2.$$Assessment of sleep
The Pittsburgh Sleep Quality Index is a self-report sleep instrument which has been validated in several studies(Reference Akbarzadeh, Khezri and Mahmudi23–Reference Farrahi Moghaddam, Nakhaee and Sheibani25). This questionnaire measures the quality and pattern of sleep over the past month and consists of nine items, differentiating from poor to good on a 0–3 scale (0, not in the past month; 1, less than once per week; 2, once or twice per week and 3, three or more times per week). These items explain sleep latency, duration and efficiency, use of sleep medication, sleep disturbances and daytime dysfunction. Pittsburgh Sleep Quality Index scores are between 0 and 21. A score of 5 and above indicates poor sleep quality.
Assessment of stress, anxiety and depression
The twenty-one-item Depression, Anxiety and Stress Scale is a self-reported questionnaire which contains twenty-one items to assess the severity of negative emotional states and symptoms of depression, anxiety and stress in the last week. These subscales include seven questions with a rating scale between 0 (never) and 3 (always). For depression, total score between 0 and 9 is considered normal, whereas scores above 9 indicate increasing severity of depression. For anxiety, a total score between 0 and 7 is normal, whereas scores above 7 indicate an increase for stress, a total score of 0–14 is normal, and greater is determined as having stress. The validity and reliability of twenty-one-item Depression, Anxiety and Stress Scale have been investigated in Iran(Reference Samani and Jokar26,Reference Elyansb27) .
Assessment of other variables
Socio-demographic information including age, education level and occupation, income, smoking habits, medical history and current medication and supplement use were assessed by the questionnaire. Physical activity (PA) levels were recorded over 7 d and expressed as metabolic equivalent h/week(Reference Ainsworth, Haskell and Whitt28). Blood pressure was measured in duplicate using a sphygmomanometer, and the mean of both measured was used as the participants blood pressure. Biochemical markers including fasting blood sugar, 2-h postprandial blood sugar, Hb A1C, total cholesterol, HDL-cholesterol, LDL-cholesterol and TAG were obtained from the participants’ medical files.
Statistical analysis
Participant characteristics were compared by ANOVA or χ 2 tests and reported as the mean values and standard deviations or percentages. Dietary intakes were reported by median-split for PDI and DAL, adjusted for energy intake using ANCOVA. Also, energy-adjusted dietary intakes in poor and good sleepers and healthy participants or participants with mental health disorders were reported using ANCOVA. The association of sleep and mental health status by media-split for PDI and DAL was determined using the χ 2 test. Moreover, having a mental health disorder and poor sleep in diabetic patients was presented in different models using binary logistic regression. In model 1, adjustment was performed for age, BMI, PA, socio-economic status, supplements consumption, vitamin D and energy intake. Further statistical control was performed in model 2 for medications, lipid profile, blood pressure, sleep duration at night and nap time. Finally, linear regression as a continuous statistical method was used to present the association between DAL score and the score of PDI with mental health disorders and having poor sleep through the fully adjusted model. SPSS version 16 was used to analyse the data. P < 0·05 was considered statistically significant.
Results
Table 1 shows the general participant characteristics. Mean age was 59·9 years. There was a significant association between both weight and BMI with uPDI (P < 0·05). A significant difference was observed in nap time and sleep duration at night and PRAL as well as uPDI (P < 0·05). There was a significant difference for SES across groups of the PDI, hPDI, uPDI and PRAL scores. Participants did not report consuming alcohol or smoking.
Consumption of carbohydrate, Na, organ meats, processed meats, high-fat dairy products, starchy vegetables and refined grains was greater in the top uPDI group. Consumption of carbohydrate, fibre, K, Fe, Mg, Cu, P, Mn, vitamins A, K, E, C, B6, and folate was lower in the top PRAL group. Table 2 presents energy-adjusted dietary intakes among PDI, NEAP and PRAL.
Table 1. Participant characteristics by median-split plant-based indices and dietary acid load (Mean values and standard deviations)
PDI, Plant-based diet index; hPDI, healthy PDI; uPDI, unhealthy PDI; NEAP, net endogenous acid production; PRAL, potential renal acid load; WC, waist circumference; SBP, systolic blood pressure; DBP, diastolic blood pressure; TC, total cholesterol; PA, physical activity; SES; socio-economic status; BGR, blood glucose reducers; BPR, blood pressure reducers; BLR, blood lipids reducers.
* Calculated by the χ 2 and t tests for qualitative and quantitative variables, respectively.
† 1: lower than median, 2: higher than median.
Table 2. Dietary intakes by median-split plant-based diet indices (PDI) and dietary acid load (Mean values with their standard errors)
hPDI, healthy PDI; uPDI, unhealthy PDI; NEAP, net endogenous acid production; PRAL, potential renal acid load; CHO, carbohydrate; RAE, retinol activity equivalents.
* Calculated by t test for energy intake and multivariate ANCOVA for other variables. All the variables, except energy, adjusted for energy intake.
† 1: lower than median, 2: higher than median.
Energy-adjusted dietary intakes among good and poor sleepers and patients with and without mental health disorders are described in Table 3. Protein, fibre, K, Fe, Ca, Mg, P, Zn, Cu, Mn, folate, vitamins B2, B3, B6, A and C intake and low-fat dairy products, vegetables and fruits consumption in poor sleepers and patients with mental health disorders were lower than participants without these problems (P < 0·05). Starchy vegetables, refined grains and high-fat dairy consumption were greater in participants with mental health disorders and poor sleepers (P < 0·05).
Table 3. Dietary intakes among poor and good sleepers and healthy participants or participants with mental disorders (Mean values with their standard errors)
CHO, carbohydrate.
* Calculated by t test for energy intake and multivariate ANCOVA for other variables. All the variables, except energy, adjusted for energy intake.
According to the χ 2 test, there was a significant association between sleep status and hPDI and uPDI (P < 0·0001). There was also a significant association between sleep status and NEAP and PRAL (P < 0·0001). Table 4 shows the OR and 95 % CI for having mental health disorders and poor sleep in crude model and adjusted models across DAL scores and PDI groups. Participants in the top group of uPDI, NEAP and PRAL scores were more likely to be poor sleepers in crude and adjusted models. Participants in the top group of hPDI had a 72 % decreased risk of poor sleep. Participants in the top group of hPDI had 74 and 76 % decreased risk of anxiety and stress, respectively. Participants in the top group of uPDI had more than nine, four and four times increased risk of depression, anxiety and stress, respectively. Table 5 presents the association between DAL score and the score of PDI with mental health disorders and having poor sleep using linear regression as a continuous statistical method.
Table 4. Mental disorders and having poor sleep by median-split dietary acid load and plant-based diet indices (PDI) (Odds ratios and 95 % confidence intervals)
hPDI, healthy PDI; uPDI, unhealthy PDI; NEAP, net endogenous acid production; PRAL, potential renal acid load.
* Calculated by logistic regression.
‡ 1: lower than median, 2: higher than median.
§ Model 1: All the variables adjusted for age, BMI, socio-economic status, physical activity, supplement intake, and vitamin D and energy intake.
|| Model 2: All the variables adjusted further for medications, lipid profile, blood pressure, sleep duration at night and nap time in addition to adjusted variables in model 1.
Table 5. Association between dietary acid load score and the score of plant-based diet indices (PDI) with mental disorders and having poor sleep using linear regression* (β-Coefficients and P values)
hPDI, healthy PDI; uPDI, unhealthy PDI; NEAP, net endogenous acid production; PRAL, potential renal acid load; PSQI, Pittsburgh Sleep Quality Index.
* Calculated by linear regression. Full-adjusted model: all the variables adjusted for age, BMI, socio-economic status, physical activity, supplement intake, vitamin D and energy intake, medications, lipid profile, blood pressure, sleep duration at night and nap time.
Discussion
This cross-sectional study revealed that higher DAL scores and adherence to uPDI were associated with poor sleep. Also, greater adherence to uPDI increased the risk of depression, anxiety and stress. Participants in the top group of the hPDI had decreased risk of poor sleep, depression, anxiety and stress. To the best of our knowledge, this is the first study that has assessed the association between DAL and PDI and psychological disorders and sleep status in diabetic patients.
It was reported in a prospective cohort study that DAL was associated with the development of type 2 diabetes(Reference Fagherazzi, Vilier and Bonnet13) and gestational diabetes mellitus(Reference Saraf-Bank, Tehrani and Haghighatdoost15). Also, some vegetarian diets are associated with a reduction in the incidence of diabetes(Reference Tonstad, Stewart and Oda29). However, there is no study that has assessed the association between DAL and PDI with diabetes complications. PDI emphasise greater intakes of whole grains, fruits, vegetables, nuts, vegetable oils, nuts and legumes and lower consumption of animal fats, fish, meat and animal-based foods (online Supplementary Table S1). The cumulative and synergic effects of these healthy food groups within dietary patterns can have protective effects on depressive symptoms(Reference Opie, Itsiopoulos and Parletta30). A meta-analysis of thirteen studies represented that high intakes of fruits, vegetables, nuts and whole grains were associated with a reduced odds of depression(Reference Lai, Hiles and Bisquera31). Another systematic review found that vegetarian diets have an inverse association with depressive symptoms(Reference Molendijk, Molero and Ortuno Sanchez-Pedreno32). Moreover, dietary patterns rich in fruits and vegetables, which have a high content of fibre, antioxidants and polyphenols, have been positively associated with mental health outcomes in adolescents(Reference Oddy, Robinson and Ambrosini33,Reference McMartin, Jacka and Colman34) .
One of the biological mechanisms that is involved in reducing mental disorders is a low level of inflammation status(Reference Schmitz and Ecker35). Inflammation can trigger melancholic symptoms through activation of inflammatory pathways in the brain(Reference Slavich and Irwin36). Low intake of whole grains, fruits and vegetables is associated with increased inflammatory markers(Reference Lopez-Garcia, Schulze and Fung37). Plant-based diets contain alkali-rich food groups such as whole grains, vegetables and fruits, while containing less of animal products, high-protein, and high-phosphorus foods(Reference Remer and Manz38). Plant-based diets increase bicarbonate and bicarbonate precursors, while animal products increase potential anorganic acid precursors(Reference Remer39). Hence, vegetarian diets significantly have lower DAL(Reference Deriemaeker, Aerenhouts and Hebbelinck40). Even moderate increases in DAL stimulate secretion and activity of glucocorticoids which leads to renal acid excretion(Reference Esche, Shi and Sanchez-Guijo41,Reference Buehlmeier, Remer and Frings-Meuthen42) . On the other hand, glucocorticoids modulate emotion and behaviour through changes in limbic areas of the brain(Reference Hamm, Ambuhl and Alpern43,Reference Mora, Segovia and Del Arco44) . Systematic acid–base balance modifies blood-brain turnover and glutamate turnover in the brain(Reference Ang, Hoop and Kazemi45). Glucocorticoids alter the expression and activation of vesicular proteins which are involved in glutamate neurotransmission(Reference Popoli, Yan and McEwen46). Moreover, glutamatergic neurons regulate brain activity and sleep stages(Reference Shi and Yu47).
To the best of our knowledge, there is only one publication that has assessed the association between DAL and mental disorders in, which showed that participants with higher PRAL had more hyperactivity and emotional problems(Reference Buhlmeier and Harris48). However, this study was conducted in adolescents, and psychological disorders were determined using the Strengths and Difficulties Questionnaire.
The main analysis indicates that higher adherence to hPDI is associated with lower odds of mental disorders in crude and all adjusted models. Also, higher adherence to the hPDI was associated with a 90 % lower odds of sleep disorders. Consistent with the foregoing, a recent review suggests plant-based diets may have the potential to improve overall health status through the effect on improving sleep quality(Reference St-Onge, Crawford and Aggarwal49). Cao et al. have showed that there was a significant inverse association between isoflavone intakes and sleep duration(Reference Cao, Taylor and Zhen50). Besides lower DAL and related mechanisms in plant-based diets, high amounts of antioxidants, phytochemicals, flavonoids, vitamins and minerals are related to beneficial effects on mental(Reference Bell, Lamport and Butler51) and sleep disorders through suppression of inflammation and reducing oxidative stress(Reference Hermsdorff, Zulet and Puchau52). Legumes and beans are high in tryptophan which is a precursor of melatonin and serotonin, which play a role in sleep regulation(Reference Bravo, Matito and Cubero53). Beezhold et al. reported that greater adherence to a vegetarian diet and even less animal food intake was associated with better mood(Reference Beezhold, Radnitz and Rinne54).
In the present study, outcome measures were adjusted for several confounders such as age, BMI, energy intake and PA. Although we adjusted for several known confounders, there are possible residual effects which can have effect on the outcome variables of interest. Increased body weight and BMI increase the risk of diabetes, insulin resistance and higher blood glucose concentrations. Furthermore, there is an association between obesity, sleep quality and depressive symptoms(Reference Wise55,Reference Fisher, Law and Dudeney56) . Moreover, body fatness is associated with both depression and sleep quality(Reference Haidar, de Vries and Karavetian57). A study in Swiss adolescents showed that higher levels of PA were related to more favourable sleep quality and lower insomnia scores, but participants tend to overestimate their level of PA(Reference Lang, Brand and Feldmeth58). However, PA has been shown to have favourable effects on sleep quality(Reference Dworak, Wiater and Alfer59), by modulating symptoms of anxiety, stress and depression(Reference Brand, Gerber and Beck60). Totally, PA was not significantly associated with poor sleep and psychological symptoms in the present study.
While this is the first study that has examined the association between DAL and PDI with sleep and psychological status among diabetic patients, there are several limitations. We randomly included diabetic women from different socio-economic status which could be a representative sample of diabetic women in Tehran. However, the relationship is not generalisable to other populations with different sex and health conditions. Moreover, the present study was conducted on a diabetic population from Tehran, and the generalisability of results to the other cities in Iran is uncertain. Due to the observational nature of the present study, cause and effect cannot be established. Furthermore, while adjustment for confounding was performed, there are possible residual effects which may have affected the outcome variables. In addition, laboratory markers of acid–base balance were not collected in the present study. While the limitations of FFQ have been widely reported for dietary assessment in large studies, there is limited validation data for anions. The twenty-one-item Depression, Anxiety and Stress Scale questionnaire is a self-reported scale which may lead to misclassification of participants and is therefore not suitable scale for clinical diagnosis of depression and anxiety.
In conclusion, the present study has shown that there was a positive association between psychological and sleep disorders with DAL, while plant-based diets had a protective effect. Prospective cohorts or intervention studies of vegetarian diets are needed to confirm our findings.
Acknowledgements
The present study is supported by Tehran University of Medical Sciences (grant number 96-03-161-36923). We would like to express our gratitude to Dr Nasli, secretary of the Diabetes Research Center of Tehran University of Medical Sciences, and all staff member for their help with the present study. The present study was approved by the ethical committee of Tehran University of Medical Sciences.
E. D., B. L. and L. A. designed and L. A. supervised the study. E. D. conducted the study. E. D., M. Q. and L. A. performed the statistical analyses. E. D. prepared a first draft of the manuscript and L. A., S. A. K. and B. A. finalised it. N. B. reviewed and edited the manuscript.
The authors declare that they have no conflicts of interest.
Supplementary material
For supplementary material referred to in this article, please visit https://doi.org/10.1017/S0007114519003179
