Assessment and validation of three spot urine assay methods for the estimation of 24‐hour urinary sodium excretion in Chinese Tibetan adults living in the mountains

Abstract Twenty‐four‐hour urine collection is the gold standard method for the evaluation of salt intake, but it is often impractical in large‐scale investigations, especially in resource‐poor areas. Methods for the estimation of 24‐hour urinary sodium excretion (USE) using a spot urine sample have been established, but have not been validated in Chinese Tibetans. Therefore, the authors aimed to evaluate the Kawasaki, Tanaka, and the International Cooperative Study on Salt, Other Factors, and Blood Pressure (INTERSALT) formulas for the prediction of 24‐hour USE in Chinese Tibetan adults. The authors analyzed the bias, correlation, agreements between estimated values and measured values, and the relative and absolute differences and misclassification at the individual level for the three methods in 323 Tibetan participants from the Ganzi Tibetan Autonomous Prefecture of Sichuan Province, China. The mean biases between the measured values and the estimated 24‐hour USE using the Kawasaki, Tanaka, and INTERSALT methods were 5.4 mmol/day (95% confidence interval [CI]: 0.8–10.1 mmol/day), −40.8 mmol/day (95% CI: −44.6 to −36.9 mmol/day), and −57.1 mmol/day (95% CI: −61.9 to −52.4 mmol/day), respectively. The Pearson correlation coefficients for the relationships between the measured values and the estimated 24‐hour USE were 0.43 (Kawasaki), 0.38 (Tanaka), and 0.27 (INTERSALT), respectively (all p < .01). The intraclass correlation coefficients showed similar patterns to the correlation data: 0.47 for Kawasaki, 0.40 for Tanaka, and 0.27 for INTERSALT (all p < .01). The upper and lower limits of agreement between the measured values and the estimated 24‐hour USE were −92.6 and 81.8 mmol/day for the Kawasaki method, −28.5 and 110.0 mmol/day for the Tanaka method, and −28.4 and 142.7 mmol/day for the INTERSALT method. Compared with the other two methods, the percentage of individuals that were misclassified by using the Kawasaki method was 48.2%, while those for the Tanaka and INTERSAL methods was 72.1% and 75.5%, respectively. However, when an individual's salt intake was higher than 12.8 g/day, the misclassification rates of the Kawasaki, Tanaka, and INTERSALT methods were 20%, 90%, and 97.5%, respectively. Thus, the authors found that the Kawasaki equation may have performed better than the other equations at Chinese Tibetan population level assessment, but none of these equations are suitable for use or perform well at the individual level. A more accurate method of using a spot urine sample to evaluate individual 24‐hour USE for Tibetans is needed.


INTRODUCTION
High salt intake is an important and preventable risk factor for hypertension. The International Cooperative Study on Salt, Other Factors, and Blood Pressure (INTERSALT) confirmed that individual salt intake was closely related to blood pressure. Specifically, for every 100 mmol increase in 24-hour urinary sodium excretion (USE), systolic and diastolic blood pressure increased by 6.0 and 2.5 mmHg, respectively. 1 In addition, high salt intake can cause target organ damage, 2,3 including left ventricular hypertrophy, renal functional impairment, arterial stiffness, and high urinary albumin/creatinine ratio, which significantly increase the risk of cardiovascular events. 4 Of the global number of deaths owing to cardiovascular diseases, approximately 1.65 million per year have been attributed to excessive salt intake. 5 Tibetans are the eighth largest ethnic minority in China, with a population of more than 6 million. A previous epidemiologic survey demonstrated that the prevalence of hypertension in Tibetans ranked front among the 56 ethnic groups in China. 6 This heavy burden of hypertension has been related to the high altitude at which they live, with low atmospheric pressure predisposing toward hypoxia 7 ; genetic factors 8 ; and the special lifestyle of Tibetan residents, such as the consumption of salty beverages and cured yak meat. Therefore, a survey of salt intake in Tibetans may be particularly important for the prevention and treatment of hypertension in this area.
The accurate assessment of salt intake is an essential prerequisite for the design of strategies aimed at reducing salt consumption.
The existing methods of measuring salt intake include dietary recall, weighted diet records, food frequency questionnaires, 24-hour urine collection, and spot urine sample collection. 9 Of these, 24-hour urine collection is recognized as the "gold standard" method for the evaluation of salt intake, 9 but the collection process is time-consuming, expensive, troublesome, and complex for participants. 10 In addition, incomplete emptying of the bladder, poor storage of the samples, and incorrect or incomplete collection might generate errors that significantly affect its accuracy. 10 Therefore, 24-hour urine collection is challenging in large-scale epidemiological investigations, and especially in those conducted in remote and/or resource-poor areas. In comparison, spot urine sample-based methods are relatively simple and inexpensive, with the collection and storage of samples being easier. Therefore, they may be practical and affordable alternatives to 24-hour USE for use in population surveys. 11 The spot urine methods for estimation of USE that commonly used are the Kawasaki formula, 12 Tanaka formula, 13 and INTERSALT formula. 14 However, the results of recent studies on the validity and reliability of spot urine sample-based methods for the estimation of 24-hour USE have yielded contradictory findings. 15,16 Moreover, the use of these formulas has not been validated in Chinese Tibetans. Therefore, the aim of the present study was to compare the use of the Kawasaki, Tanaka, and INTERSALT methods for the estimation of 24-hour USE in a general population of Chinese Tibetan adults living in the mountains. We anticipate that this study would lay a foundation for estimating and monitoring salt intake in Chinese Tibetans.

Study participants
We performed a cross-sectional study of adults over 18

2.3
Collection and measurement of urine samples

Methods of collection
All the participants were asked to maintain their normal diet and liquid intake, and to avoid a high-protein diet and strenuous physical activity. Urine collection was performed on 2 consecutive days (days

Methods of evaluating urine sodium excretion
The methods for the estimation of 24-hour USE by using spot urine samples were Kawasaki method, 12 Tanaka method, 13 and INTER-SALT method. 14 The estimation formulas used are listed in Table 1.

Related definition
Hypertension was defined by the average systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg at the physical examination or who were receiving the antihypertensive therapy or who had a previous diagnosis of systemic hypertension before this investigation. 21 Diabetes mellitus (DM) was defined by a previous diagnosis of type 1 and 2 DM with or without hypoglycemic therapy or fasting venous blood glucose ≥ 7.0 mmol/L this time. 22 The formula of body mass index was as follow: weight (kg)/height (m 2 ).

Characteristics of the participants
Three hundred fifty participants were initially recruited, of whom 13 with incomplete 24-hour urine collection, 11 with suspected urine sample contamination, and 3 whose 24-hour urinary creatinine > 3 times SD of the population mean were excluded ( Figure 1). Therefore, data for 323 participants were included in the analyses. The baseline characteristics of the participants are presented in Table 2. Of the participants, 61% were females, and 54.5% and 5% had been diagnosed with hypertension and DM, respectively. They were 51.2 ± 15.1 years old.
The mean ± SD sodium concentration of their 24-hour urine samples was 125.8 ± 21.5 mmol/L and their mean ± SD 24-hour urine volume was 1539.7 ± 195.6 ml. The mean concentrations of sodium, potassium, and creatinine in their spot urine samples were 151.5, 55.6, and 11.2 mmol/L, respectively.

Measured versus estimated 24-hour USEs
The mean ± SD measured 24-hour USE of the participants was 198.8

Relationships between the measured and estimated 24-hour USE values
The Pearson correlation coefficients for the relationships between  ( Table 4).

Agreements between the measured and estimated 24-hour USE values
Bland-Altman plots were used to evaluate the agreement between the values estimated using the three methods and the measured 24-hour USE values, and the results are shown in Figure 2

Relative and absolute differences between the measured and estimated 24-hour USE values
When using the measured 24-hour USE as the reference, the proportions of the samples for which the relative differences were within ±10% for the Kawasaki

3.6
The percentage of individuals that were misclassified using the three methods for the assessment of salt intake We also compared the individual salt intake classification group according to their measured salt intake compared with estimated value by using three formulas (Table 5). According to the quartile of mea-

DISCUSSION
To the best of our knowledge, the present study is the first to compare The inconsistency of these findings implies that ethnicity and dietary pattern might affect the choice of the optimal means of estimating 24-hour USE. In addition, the Kawasaki and Tanaka methods were established during the studies of the Japanese population, which has a high salt intake, 12,13 and therefore these methods might potentially overestimate the salt intake of individuals consuming a western diet.
Conversely, the INTERSALT method was developed using the relatively low salt intake of the European and American populations, 14 and therefore might underestimate the 24-hour USE of individuals consuming a relatively large amount of salt. In particular, salt intake is high in Asia, and especially in east Asia countries, such as China, Japan, and Korea, 33 as well as in Tibet, China. 34 Moreover, a study conducted in the general Brazilian adult population found that the use of the Kawasaki formula was appropriate only when the salt consumption was between 12 and 18 g/day. 35 In the present study, the mean salt intake of the Tibetan participants was 12.6 g/day, which is significantly higher than the global mean, which might explain the relatively small differences between the measured values and estimated values using the Kawasaki

CONCLUSIONS
We found that the Kawasaki equation may have performed better than the other equations at Chinese Tibetan population level assessment, but none of these equations are suitable for use or perform well at the individual level. A more accurate method of using a spot urine sample to evaluate individual 24-hour USE for Tibetans is needed.

ACKNOWLEDGMENTS
The authors sincerely thank the doctors, nurses, and medical staff from