Seasonality Affects Fluid Intake Behaviors among Young Adults in Hebei, China

Background: Evidence on the association between environmental factors and fluid intake behavior remains limited. The current study aims to explore seasonal variations in fluid intake behaviors among young adults in China. Methods: A prospective cohort of 79 healthy young adults (43 males and 36 females) aged 19–21 in Hebei, China, was assessed for fluid intake behaviors for four seasons. For each assessment, the participants’ anthropometric measurements were collected. Temperature and humidity on survey days were measured. Participants’ total drinking fluid (TDF) was recorded using a self-administrative 7 d, 24 h fluid intake questionnaire. To calculate water from food (WFF), we weighed all foods consumed by participants. Duplicates of consumed food samples were collected to measure the water content via the drying method. Results: The mean total water intake (TWI) was 2761 ± 881, 2551 ± 845, 2210 ± 551, and 1989 ± 579 for spring, summer, fall, and winter, respectively (F(2.37) = 42.29, p < 0.001). The volume and proportion of TWI from TDF and WFF varied across the four seasons. The volume of WFF in spring (1361 ± 281, F(2.61) = 17.21, p < 0.001) and TDF in summer (1218 ± 502, F(2.62) = 9.36, p < 0.001) was among the highest, while participants’ fluid intake behaviors in spring and summer were less distinct than the other pairwise comparisons. A moderate association was found between outdoor temperature and TDF (r = 0.53, p < 0.01). Different general estimating equations suggested that gender, seasonality, outdoor temperature, differences in indoor and outdoor temperature, and mean temperature were independent factors of TDF. An interactive effect was found for gender and temperature, showing that the expected TDF of males may increase more as the temperature climbs. Conclusions: Gender, seasonality, and air temperature could significantly affect fluid intake behaviors, including the amount and type of fluid intake. However, the independent effect of BMI and humidity remains unclear.


Introduction
The survival and well-being of humans rely heavily on water, a precious natural resource and essential nutrient.Water is a fundamental building material of the human body, accounting for approximately 60% of an adult's body mass, and widely distributed as body fluid in diverse cells, tissues, and organs [1].In addition to regulating the body temperature and maintaining electrolyte balance, water is also deeply involved in human metabolism.Any disruption in the water homeostasis within the body, either dehydration or hyperhydration, may lead to adverse health effects [1].
Therefore, the brain closely monitors and regulates one's plasma and serum osmolality in collaboration with the kidney and arginine vasopressin (AVP), a hormone released by the hypothalamus and posterior pituitary.When the body is dehydrated, either caused on environmental factors and to evaluate its intensity and direction of influence on fluid intake behaviors.

Study Designs
A prospective cohort study on fluid intake behaviors throughout the four seasons was implemented in 2021 in Baoding, Hebei, China.Each assessment lasted seven consecutive days and followed the same procedure.The four assessments were conducted once each season from 12 to 27 April, 5 to 11 June, 21 to 27 October, and 9 to 15 December, respectively.The division of seasons follows the national standard (GB/T 42074-2022), in which the 5 d moving average of temperatures in spring, summer, fall, and winter are 10-22 • C, ≥22 • C, 10-22 • C, and <10 • C, respectively [12].

Ethics
This study has been registered with the Chinese Clinical Trial Registry.The registration number is ChiCTR2100045268.
The Peking University Institutional Review Committee has reviewed and approved the study protocol.The ethical approval project identification code is IRB00001052-21013.

Participants
Healthy male and female college students aged 18-25 from a university in Baoding, Hebei, China, were recruited by convenient and snowball sampling.Participants who were smokers, habitually consumed alcohol (>20 g/d), and who had gastrointestinal, oral, and other chronic diseases were excluded from this study.All participants gave informed consent.

Anthropometric Measurements
The height and weight of each participant were measured by trained investigators using a height-weight meter (HDM-300; Huaju, Zhejiang, China).Participants were required to wear light clothing and be barefoot.The height and weight were measured twice, and the mean was recorded to the nearest 0.1 cm and 0.1 kg, respectively.Each participant's body mass index (BMI) was then calculated with the following equation: BMI = weight (kg)/height 2 (m).The classification of normal (18.5 ≤ BMI < 24.0), underweight (<18.5),overweight (24.0 ≤ BMI < 28.0), and obese (≥28.0) was based on the criteria suggested by the Working Group on Obesity in China [15].

Temperature and Humidity
Temperature hygrometers (No.8813, Deli, Zhejiang, China) were placed on the sports ground, classroom, and dormitory.The indoor and outdoor measurements were recorded to the nearest 0.1 • C and 1% by trained investigators thrice daily at 10:00, 14:00, and 20:00.

Assessment of Fluid Intake
Total fluid intake is composed of drinking fluid and water from food.Drinking fluid intake was assessed for 7 d, and water from food was assessed for 3 d.

Drinking Fluid
For each assessment, participants completed a self-administrative 7 d, 24 h fluid intake questionnaire previously used and tested for reliability and validity [16,17].The questionnaire records the type, volume, time, and place of drinking fluid as described in a previous study [13].TDF was classified into plain water and other beverages, including tea, sugar-sweetened beverages (SSBs), and alcohol.The classification of drinking fluid follows the General Standard for Beverages of China (GB/T 1-789-2015) [18].The volume of drinking fluid was measured to the nearest 5 mL with a graduated water bottle.Each participant was asked to use a 5-mL-scaled bottle designed for the study when they drank any fluid to ensure accuracy.

Water from Food
Each assessment measured and recorded water from food (WFF) for three consecutive days.The exact amount of each food item consumed was weighed to assess WFF.All consumed foods were weighed before and after participants consumed them.Duplicates of consumed food items were collected.Moisture in collected food samples was measured via the drying method, according to the national standard of GB 5009.3-2016[19].WFF was then calculated using the amount of food consumed and the moisture content of each food item.Types of foods were classified as follows: staple food, dishes, porridge, soup, dairy, and snacks.The volume of dairy consumed (milk and yogurt) was recorded in the 7 d, 24 h fluid intake questionnaire.Moisture in dairy, fruits, and snacks was evaluated according to the China Food Composition Table (6th edition) [20].

Statistics
SPSS 26.0 was used for statistical analysis of baseline characteristics and fluid intake.All statistical tests were two-sided; a significance level of <0.05 represents a statistically significant difference.Continuous variables were analyzed using the t-test, one-way analysis of variance (ANOVA), and the non-parametric Kruskal-Wallis H-test.Results are presented as mean ± standard deviation (x ± S).
Repeated-measure analysis of variance (RMANOVA) was implemented to assess the impact of seasons on repeatedly measured variables that were normally or approximately normally distributed.For variables that did not follow a normal distribution, Friedman's M test, the non-parametric RMANOVA, was implemented to assess the between-group differences by ranks.The Bonferroni-adjusted p-value was applied to post hoc comparisons.
To assess the association between environmental factors and total drinking fluid (TDF), we calculated the daily average of TDF, frequency, volume consumed each time, plain water intake, and beverage intake for the overall, male, and female participants based on data collected from the 4 × 7 d assessment on drinking fluid intake.Spearman rank correlations were then conducted between the collected 4 × 7 d of environmental measurements and the variables computed above.Scatter plots were also built.The general estimating equations (GEEs) with the working correlation matrix set to be the exchangeable mode were implemented to access the exact direction and the strength of the association between the environmental factors and TDF of participants while controlling for confounders.

Characteristics of Participants
Table 1 summarizes the anthropometric measurements of participants in each season.Among 84 recruited participants aged 19 to 21, 94% (n = 79, 43 males and 36 females) completed all four assessments on fluid intake behavior from April to December 2021.For the 79 participants, the average age was 19.9 years old.Anthropometric measurements were taken each season.The mean height, weight, and BMI were 169.6 cm, 66.4 kg, and 23.0 kg/m 2 , respectively.In contrast to height and weight, BMI only exhibited significant gender differences in fall and winter, though the marginal p-value in spring and summer suggested a trend of statistical significance.Results of MANOVA showed significant variation in participants' height, weight, and BMI in different seasons.Pairwise comparisons suggested that the height in fall significantly differed from the rest of the seasons.No significant difference between fall and winter was found in terms of weight.BMI in each season was significantly different from the other season.Appendix A summarizes the significance levels for all pairwise comparisons among anthropometric measurements in different seasons.

Total Fluid Intake
Table 2 shows temperature and humidity during each survey period.Participant's TFI, including TDF and WFF, is summarized in Table 3.We found significant impacts of seasonality, gender, and BMI on TFI, TDF, and WFF.We found that TFI, TDF, and WFF significantly differed between males and females every season.However, no significant impact of BMI on TDF in fall and WFF in spring was found, and the significance levels of pairwise comparisons for different BMIs are shown in Table A1.Seasonal variations of TWI (F(2.37)= 42.29,p < 0.001), TDF (F(2.62)= 9.36, p < 0.001), and WFF (F(2.61)= 17.21, p < 0.001) were found.The significance levels for all pairwise comparisons are summarized in Appendix A.  Seasonal variations in the composition of TWI were also analyzed.Figure 1 shows the TWI and its composition among different seasons.The volume of each fluid source consumed in different seasons is shown in Table 4, and the proportions are shown in Table 5.We found that the compositions of fluid sources significantly varied across the year.However, seasonal variations were found in the proportion but not the volume of water from staple foods.Neither the volume nor the proportion of dairy significantly varied across the year.We summarize the significance levels for all pairwise comparisons in Appendix A.
the TWI and its composition among different seasons.The volume of each fluid source consumed in different seasons is shown in Table 4, and the proportions are shown in Table 5.We found that the compositions of fluid sources significantly varied across the year.However, seasonal variations were found in the proportion but not the volume of water from staple foods.Neither the volume nor the proportion of dairy significantly varied across the year.We summarize the significance levels for all pairwise comparisons in Appendix A.

Environmental Factors and TDF
Correlational coefficients of drinking fluid and environmental factors derived from Spearman rank correlation are listed in Table 6.We found that outdoor and indoor temperatures were significantly associated with TDF, volume consumed each time, and plain water intake.MT and ∆T were also associated with TDF.No significant correlation between temperatures and the frequency of fluid intake was found for the overall population, but moderate correlations were found when analyzing the average frequency of male and female participants separately.However, beverage intake had a stronger association with indoor temperature than outdoor temperature.Most of the correlational coefficients exhibited gender differences.Participants' TDF, frequency of fluid intake, volume consumed each time, and plain water intake tended to increase as indoor and outdoor temperatures climbed.Scatter plots of temperature and total drinking fluid, as shown in Figure 2, were also built to visualize the correlation between environmental factors and the fluid intake behavior of participants.

Environmental Factors and TDF
Correlational coefficients of drinking fluid and environmental factors derived from Spearman rank correlation are listed in Table 6.We found that outdoor and indoor temperatures were significantly associated with TDF, volume consumed each time, and plain water intake.MT and ΔT were also associated with TDF.No significant correlation between temperatures and the frequency of fluid intake was found for the overall population, but moderate correlations were found when analyzing the average frequency of male and female participants separately.However, beverage intake had a stronger association with indoor temperature than outdoor temperature.Most of the correlational coefficients exhibited gender differences.Participants' TDF, frequency of fluid intake, volume consumed each time, and plain water intake tended to increase as indoor and outdoor temperatures climbed.Scatter plots of temperature and total drinking fluid, as shown in Figure 2, were also built to visualize the correlation between environmental factors and the fluid intake behavior of participants.GEEs were built based on 4 × 7 d data for seasons, outdoor temperature, ∆T, and MT, with TDF as the dependent variable, as shown in Table 7.The models were built separately to avoid collinearity.We found that gender possibly confounded the association between BMI and TDF among the study population.Moreover, the temperature was the potential confounder of the association between humidity and TDF (β = 4.6, p = 0.002).∆T (β = 4.8, p = 0.008), and MT (β = 15.4,p < 0.001) were all positively associated with TDF.TDF increased as the temperature climbed.The results of the GEE analysis also suggested that the interactions between gender and outdoor temperature, as well as the differences in temperature, and that the association between temperature and TDF might be stronger among males than females.Seasons were also associated with TDF, among which summer (β = 236.1,p < 0.001) has the leading β, followed by spring (β = 178.7,p < 0.001) and summer (β = 79.0,p = 0.018), respectively.

Discussion
In the current study, we assessed the fluid intake behaviors of the same group of participants four times throughout 2021, one in each season.Recruited participants were young adults aged 19-21 at a university in Hebei, China.
In each assessment, participants' anthropometric measurements were also obtained.When analyzing participants' anthropometric measurements, we found significant seasonal variations in participants' weight, height, and BMI as consequences.The variations in participants' height and weight could also be due to minor deviations in the timing and instrumental reading while measuring in different seasons.Significant variations in fluid intake behaviors of participants with different weight statuses were found when conducting one-way ANOVA, and obese participants tended to have higher TDF and TWI than the rest.However, the association between BMI and TDF could be potentially confounded by gender due to the significant differences in the BMI of male and female participants and the strong collinearity of gender and BMI while we conducted the GEE analysis.
Participants' fluid intake behaviors, including the TDF, WFF, and TWI, exhibited significant gender differences in all four seasons.Male participants in our study tended to have higher TDF and WFF than females, resulting in a higher TWI.In line with the results from the present study, a recent survey among 2233 Chinese residents in 27 cities reported a significantly higher TFI in males than females regardless of age group [21].Wu et al. [22] also found a significant difference in daily TWI between US males and females based on the National Health and Nutrition Examination Survey 2011-2014 data.According to the results of the GEE analysis, gender was significantly associated with TDF.Males exhibited higher coefficients than females, ranging from 221.0 ± 87.8 to 428.2 ± 105.0 depending on the other variable in the model built.
The higher muscle contents in the body composition of males might contribute to the gender differences in fluid intake behaviors and requirements.Established guidelines of fluid intake worldwide also embody the gender difference.For instance, the recommended TWI by the CNS is 3000 mL/d for males and 2700 mL/d for females.The recommended TDF by the CNS were 1700 mL/d and 1400 mL/d for males and females [23].European Food Safety Authority (EFSA) recommended a TWI of 2500 mL/d for males and 2000 mL/d for females [24].However, neither the mean TDF of males nor females across the four seasons in the current study met the recommendation of CNS and EFSA.The mean TWI for both male and female participants was also lower than that recommended by CNS throughout the year, urging attention to the importance of adequate fluid intake and establishing an optimal fluid intake habit among the general population.
Regarding environmental factors, we measured and recorded indoor and outdoor temperature and humidity of all 28 survey days.The mean and differences between indoor and outdoor temperature and humidity were also calculated.The statistical analysis suggested that all measurements of temperature significantly varied across different seasons but not that of humidity.Only the difference in indoor and outdoor humidity in summer significantly differed from the rest of the seasons.That could partially explain the negligible influence of humidity on TDF in our study.
According to the results of correlational analysis among daily temperature, humidity, and TDF, the temperature significantly impacted participants' fluid intake behaviors, including the volume, frequency, and type of fluid consumed.Outdoor temperature, indoor temperature, and MT were all positively correlated with TDF.The GEE analysis indicated that TDF would increase as outdoor temperature, ∆T, or MT climbed.However, only indoor humidity was found to have a weak association with the volume consumed each time among all participants and the plain water intake among females.∆H was also weakly correlated with the volume consumed each time and plain water intake among males.
Similarly, a previous study in Japan found that the temperature but not humidity were associated with water intake, that for every 1 • C increase in the mean air temperature, the expected TDF would increase by 8.4 g/d when controlled for BMI and gender [7].The correlational analysis suggested a correlation between the temperature and humidity.For instance, indoor humidity correlated with indoor temperature (r = 0.375, p = 0.049) and ∆T (r = −0.460,p = 0.014).Even though the correlations were weak, summing up with the strong collinearity we discovered later in the GEE analysis, temperature might still confound the correlation between indoor humidity and TDF.
Fluid intake behaviors may also vary according to seasons, especially the volume and the composition of TWI.The results of MANOVA showed significant differences in TDF (F(2.62)= 9.36, p < 0.001), WFF (F(2.61)= 17.21, p < 0.001), and TWI (F(2.37)= 42.29,p < 0.001) across the four seasons.TDF in summer (1219 ± 502) was significantly higher than in fall (1082 ± 475, p = 0.015) and winter (1082 ± 475, p = 0.009) but did not significantly differ from that in spring (1182 ± 594, p = 1.000).As results from GEE analysis showed, when controlled for gender, summer (β = 236.1,p < 0.001) had the leading β among all seasons with winter as the reference.The β of spring and fall were 178.7 (p < 0.001) and 79.0 (p = 0.018), respectively.Consistent with our findings, Malisova et al. [25] concluded that there was no significant difference in the water balance between summer and winter among Greek residents, while TDF was significantly higher in summer than in winter, and the proportion of WFF in winter outran that in summer.
We also analyzed the volume and proportion of specific sources of fluid intake.The results indicated that the proportion of TWI from TDF was significantly higher in summer than in fall and winter.The volume but not the proportion of WFF in spring, specifically the volume of water from dishes and porridge, was significantly higher than in the resting seasons.However, Tani et al. [7] concluded that for every 1 • C increase in the mean air temperature, the expected WFF would decrease by 3.1 g/d.The contradicting results might be attributed to the significant increase in other beverage intake in summer.Correspondingly, a survey conducted in Korea showed that plain water and beverage intake increases by 19.7% and 7.6% in summer compared to spring, though we did not find a significant difference in the volume of plain water intake between spring and summer [26].Possibly due to differences in dietary habits and culture, the proportion of WFF in TWI in the present study was also higher across the four seasons in our study than the data from previous assessments worldwide.For instance, only 18% of the TWI intake of US citizens were WFF based on 1999-2006 NHANES data [27].The proportion of WFF in TWI among UK and French citizens was 27% and 36%, respectively [28].
Previous studies exploring the impact of seasonality, temperature, and humidity on fluid intake behaviors rarely conducted assessments in all four seasons.Otherwise, data from national-based dietary surveys were utilized and analyzed.Limitations and imprecisions could be found in measuring and estimating the exact amount of fluid intake.In the present study, composed of four assessments of the same group of participants, we employed a specifically designed 7 d, 24 h fluid intake questionnaire that had been proven reliable and valid in our past study.Participants were also trained to record the type, volume, time, and place of drinking fluid immediately after.The exact amount of WFF was calculated by weighting the consumed amount of each food item and measuring moisture in the duplicate of each food item via the drying method, significantly improving the accuracy and precision of TWI measurement.In addition, the current study analyzed the seasonal differences of TWI, TDF, and WFF and further investigated the association between participants' fluid intake behaviors and environmental factors, including indoor and outdoor temperature and humidity.Limitations of the present study are that participants were prescribed to only participate in light physical activity for quality control.However, physical activity is essential in fluid intake behaviors and might be significantly influenced by the environment.By minimizing the confounding effect of physical activity, we also lose the chance to explore the impact of physical activity on participants' fluid intake behaviors and the underlying interaction of environmental factors and physical activities, as little evidence has been established on the interactive effects of physical activity and environmental factors on one's fluid intake behavior and overall hydration state.

Conclusions
The current study verified the association between fluid intake behaviors and environmental factors.Even though it was possibly due to the relatively steady humidity in our survey site across the year, no significant association was found between humidity and fluid intake behaviors, and seasonality and temperature could still significantly impact the total volume of fluid intake and the proportion of different fluid sources in TWI.Gender was the other factor significantly associated with fluid intake behaviors.Even though previous research suggested that overweight and obesity were associated with higher TWI, urine osmolality, and risk of dehydration, gender possibly confounded the correlation between BMI and fluid intake behaviors in the current study.
In subsequent studies, an in-depth exploration of one's anthropometric measurements, including height, weight, BMI, body surface area, and blood pressure, should be completed, along with analyses of their association and water homeostasis within the body.TDF increases as temperature elevates and is the highest in summer.Males tended to have higher TWI, WFF, and TDF across all four seasons.Their TDF could also be impacted by temperature to a greater extent than that of females.
Future studies could incorporate biomarkers of hydration status into the analysis to obtain a more comprehensive and systematic understanding of how environmental factors may affect one's hydration status and water homeostasis, thereby finding implications in refining the recommendation of fluid intake in different seasons.

Figure 1 .
Figure 1.Participants' total water intake among different seasons.

Figure 1 .
Figure 1.Participants' total water intake among different seasons.

Figure 2 .
Figure 2. (a) Correlational plot between outdoor temperature and total fluid intake.(b) Correlational plot between indoor temperature and total fluid intake.

Figure 2 .
Figure 2. (a) Correlational plot between outdoor temperature and total fluid intake.(b) Correlational plot between indoor temperature and total fluid intake.

Table 1 .
Seasonal variations in characteristics of subjects.
Statistics were derived from MANOVA.BMI, body mass index; df, degree of freedom.

Table 2 .
Air temperature and humidity in each season.

Table 3 .
Seasonal variations in total fluid intake.
a , one-way ANOVA b , and t-test c .TDF: total drinking fluid; WFF: water from food; TWI: total water intake; BMI: body mass index; df : degree of freedom.

Table 4 .
Sources of fluid intake in different seasons (volume).
Statistics were derived from RMANOVA.TDF, total drinking fluid; WFF, water from food; df, degree of freedom.

Table 4 .
Sources of fluid intake in different seasons (volume).
Statistics were derived from RMANOVA.TDF, total drinking fluid; WFF, water from food; df, degree of freedom.

Table 5 .
Sources of fluid intake in different seasons (proportions).
Statistics were derived from Friedman's M test.TDF, total drinking fluid; WFF, water from food.

Table 6 .
Correlation among daily temperature, humidity, and total drinking fluid.Statistics were derived from Spearman rank correlation.TDF, total drinking fluid ∆ = outdoor − indoor.* The significance level is <0.05; ** the significance level is <0.01.

Table 7 .
Variables associated with total drinking fluid.