膳食钠估算方法：新旧方法在心血管高危人群中的准确性和局限性

克里斯蒂亚娜·齐里米亚库 1 2 ， 加利奥皮·卡拉齐 3 4 ， 埃里尼·德·帕斯特吉 1 2 安东尼奥斯·阿格里斯 1 ， 西奥多·加·帕帕永 5 ， 玛丽亚·扬纳库利亚 2 年 阿塔纳斯·德·布罗托格罗 1 *

1 心血管预防与研究单位，临床和病理生理学实验室，医学系，雅典国立和卡波迪斯特里安大学，75，Mikras Asias Street，雅典 11527，希腊：希腊雅典哈罗科皮奥大学健康科学与教育学院营养与营养学系：雅典农业大学食品科学与人类营养系营养学和生活质量实验室， 希腊：希腊心血管健康与营养基金会，希腊雅典：希腊国立和卡波迪斯特里安大学医学系心脏病学第一系生物医学工程部

提交日期：2021 年 1 月 8 日： 收到最终修订版 2021 年 10 月 1 日： 接受日期：2021 年 10 月 20 日：2021 年 10 月 25 日首次在线发布

抽象

目的：缺乏准确易用的人群膳食钠摄入量估算方法。我们的目标是 （i） 使用各种饮食方法 （DM） 和泌尿方法 （UM） 估计组水平的平均 Na 摄入量，并将它们与 24 小时尿液收集 （24UCol） 相关联，以及 （ii） 提高现有 DM 的准确性。

设计：应用最常见的 DM（三次 24 小时饮食召回 （24DR） 和 FFQ）和 UM（使用通用方程的 24UCol 和现场尿液收集）。为了改进现有的：（i） 24DR，使用与盐相关的问题或在总 Na 摄入量中额外增加 15% 来量化酌情的 Na，以及 （ii） FFQ，在标准问卷 （NaFFQ） 中添加了富含 Na 的食物和与盐相关的问题。

周边环境：希腊雅典国立和卡波迪斯特里安大学。

参与者：总共 122 名高心血管风险受试者（56.0 ± 12.6 岁;55.7% 男性）。

结果：平均24 h钠排泄量（24UNa）为2810±1304 mg/d。点尿方法高估了24UNa（偏倚范围：-1781至-492 mg），并且与24UCol中等相关（r = 0·469–0·596，P ≤ 0·01）。DM 低估了 24UNa（偏倚范围：877 至 1212 mg），并且与 24UCol 的相关性较弱。改进后的DM低估了24UNa（偏倚范围：877至923 mg）。与所有其他方法一样，NaFFQ的偏差最小（-290±1336 mg），与24UCol的相关性最强（r = 0·497，P ≤ 0·01），但在Bland-Altman图中一致性范围很宽（-2909 mg;2329 mg）。

结论：现有方法的准确性较差。新开发的NaFFQ的进一步改进可能有助于在流行病学研究中更准确地估计平均膳食Na摄入量。需要更多的验证研究。

关键字

饮食 Na 评估 24 小时尿液收集 24 小时尿液收集

高钠摄入量是导致血压升高的重要因素，增加心血管疾病的风险和死亡率。尽管国际组织建议每日最大钠摄入量为 2000 毫克，但在全球范围内估计几乎翻了一番，达到 3950 毫克/天。在大规模的流行病学研究中，准确估计膳食钠摄入量对于检测实际消费量和识别食物项目非常重要

与钠摄入量相关的食物模式或饮食行为及其与疾病和治疗的关联。在临床环境中，Na摄入量的评估对于评估患者对建议的依从性和指导药物治疗决策至关重要。有多种尿法（UM）和饮食法（DM）可用于估算膳食钠摄入量;然而，其准确和精确的量化仍然难以捉摸。

公共卫生营养：25（4），866-878

doi：10.1017/S1368980021004390

*通讯作者：Email aprotog@med.uoa.gr © The Author（s）， 2021.由剑桥大学出版社代表营养学会出版。这是一篇开放获取文章，根据知识共享署名许可 （https://creativecommons.org/licenses/by/4.0/） 的条款分发，该许可允许在任何媒体上不受限制地重复使用、分发和复制，前提是正确引用原始作品。

https://doi.org/10.1017/S1368980021004390 Published online by Cambridge University Press

现尿样本、隔夜尿液收集和 24 小时尿液收集 （24UCol） 代表 UM。基于在24小时内消耗的约90%的Na通过尿液排泄的知识，24UCol被认为是金标准方法。然而，它是一种繁琐、耗时的方法，难以应用于大规模研究以及无并发症动脉高血压管理的日常临床实践中。通过专门设计的方程式（表1），现场尿液样本更方便地估计24小时尿钠排泄量（24UNa），该方程式已在多个人群中进行了评估。

另一方面，Na 估计最常见的 DM 包括 24 小时饮食召回 （24DR）、FFQ 和饮食记录。这些方法通常用于基于人群的研究中，因为它们可以有效地突出富含 Na 的食物;然而，存在许多方法上的缺点。一个主要问题是所有这些方法都无法量化盐的任意使用（食盐或烹饪过程中盐的使用），以前有报道称盐对总钠摄入量有很大贡献。

已经做出了一些努力来开发一种基于饮食的最佳工具，用于估计群体水平上的平均 Na 摄入量，其中大多数侧重于短 FFQ，与其他 DM 相比，这些 FFQ 简短、易于完成并估计更长的时间段的 Na 摄入量。然而，FFQ通常是为特定人群开发的，并根据他们的文化、饮食习惯和传统食谱进行设计，因此可能无法准确地应用于其他人群。

据我们所知，使用黄金标准 24UCol 同时评估不同 UM 和 DM 对 Na 估计的准确性的研究很少。此外，没有准确的 DM 用于定量可自由支配的盐，专为高 CVD 风险人群设计，对他们来说，识别 Na 摄入量至关重要。考虑到所有这些问题，本研究的目的是 （a） 使用各种 DM 和 UM 估计人口的平均 Na 摄入量;（b） 将这些方法与黄金标准 24UCol 相关联，以及 （c） 改进现有的 DM，以便更准确地估计人口水平的平均 Na 摄入量。

方法

研究设计和人群

2017 年 1 月至 2018 年 10 月进行了一项横断面研究。研究人群包括由于存在 CVD 危险因素（疑似或已确定治疗或未治疗的高血压、血脂异常、血脂异常）而处于高 CVD 风险的连续和同意参与个体

糖尿病和/或慢性炎症性疾病）。为了检测每种 Na 估计方法与 24UNa 之间每日 Na 摄入量的最小差异 500 mg （α = 0·05，功效 = 0·80），计算了每对方法的最小样本量 （n 60）。为了说明估计为 50% 的损耗（未参与、数据缺失或不完整的 24UCol），邀请了 120 人参加。该研究得到了伦理/科学委员会的批准。所有参与者均提供知情同意书，并同时接受饮食和尿液评估，评估在1个月内完成。

使用尿液方法评估膳食钠摄入量 24 小时尿液收集。参与者被要求按照书面和口头说明以及标准化协议保留一个 24UCol。指示是从周日醒来开始，在接下来的 24 小时内进行收集，丢弃第一天早上的空隙，没有：（a） 缺失的空隙和 （b） 他们的饮食或药物的任何变化（过去 1 个月）。为了验证完整性，在应用所有可用的 24 小时尿液完整性标准（统计分析部分）后进行了敏感性分析。使用以下公式计算从 24UCol 导出的 Na：

24UNa ðmg=dÞ 1/4 24-h Na 浓度ðmmol=lÞ

24小时尿量ðlþNa分子量ð23mg=mmolþ

尿斑。参与者还被要求将第一天早上排尿的单点尿液样本保存在适当的瓶子中。为了估计点尿标本的 24 小时 Na 排泄量，应用了最常见的转换方程（表 1）。

使用现有饮食方法评估膳食钠摄入量 24 小时膳食召回。由训练有素的营养师通过电话或面对面访谈，使用多次通过方法进行了 3 次 24DR（2 个工作日和 1 个周末，间隔 7 天）。参与者被要求报告他们在过去 24 小时内消耗的所有食物和饮料及其数量。通过使用相关的营养素分析软件（Nutritionist Pro，5.2版，Axxya Systems-Nutritionist Pro，Stafford，TX，USA），从宏量营养素和微量营养素摄入量的角度分析了来自24DR的食物数据。采用3 d的Na摄入量平均值。如果可用的 24DR 少于 3 个，则使用其余的平均值。

食物频率问卷。在

饮食评估，所有参与者都被要求完成半定量 FFQ，该 FFQ 可重复且有效，可用于有关能量和宏量营养素的营养评估。FFQ 由一份包含 69 个主要内容的清单组成

膳食钠评估：方法比较

org/10 https://doi.org/10.1017/S1368980021004390 Published online by Cambridge University Press

表1

用于估计单点尿液标本 24 小时尿 Na 排泄量的方程式

川崎

预计24小时

1/4 16 3

ffiffiff

预测 24 小时尿液 Cr

雄性 ： 预测 24 小时 Cr

重量

高度

女性 ： 预测 24 小时 Cr

4

零

重量

零

高度

预计 24 小时 Na、mmol/dspot Na、mmol/lspot Cr：mg/l预测 24 小时 Cr：mg/dage、年重、kgheight、cm

田中

预计24小时

21

X Na 0  392

现货 Na

点铬

预测 24 小时 Cr

预测 24 小时 Cr

阿拉伯数字

14

89重量

16

高度

估计 24 小时 Na，mmol/dspot Na，mmol/lspot Cr：mg/dl预测 24 小时 Cr：mg/dage，年重，kgheight，cm

盐间

使用 Spot K

雄性 ： 估计 24 小时 Na

25

46 þ

现货 Na

点铬

现货 K

女性：预计24小时

¼ 5  07

现货 Na

点铬

现货 K

无点 K

雄性 ： 估计 24 小时 Na

23

51 þ

现货 Na

点铬

女性：预计24小时

¼ 3 

现货 Na

点铬

估计 24 小时 Na、mmol/dspot Na、mmol/lspot Cr：mmol/lspot K：mmol/lBMI：kg/m

年龄、岁月

现货 Naspot Cr

预测 24 小时 Cr

男性：

预计24小时

X Na 0  345

预处理 24 小时铬

零

7

重量

高度

估计 24 小时 Na，mmol/dspot Na，mmol/lspot Cr：mg/dl预测 24 小时 Cr：mg/dage，年重，kgheight，cm

女性：

预计24小时

X Na 0  196

预测 24 小时 Cr

女性

重量

高度

预计24小时

现货 Naspot Cr

预测 24 小时 Cr

雄性 ： 预测 Cr 24 小时

00179

重量

高度

0  5 ð

雌性 ： 预测 Cr 24 小时

¼ 0 

00163

重量

高度

0  5

零

0

非裔美国人或黑人种族

1/其他种族

预计 24 小时 Na、mmol/dspot Na、mmol/lspot Cr：mg/dl预测 24 小时 Cr：mg/dBMI：kg/m

重量，公斤高，厘米

C Tsirimiagkou 等人。

https://doi.org/10.1017/S1368980021004390 Published online by Cambridge University Press

食物类别（即谷物和淀粉类食物、水果、蔬菜、乳制品、肉类、鱼类、豆类、添加脂肪、糖果和酒精饮料）以及与饮食行为和习惯有关的问题。参与者被要求报告上个月以六级量表（从从不/很少到超过 2 次/天）食用这些食物组的频率，以预先指定的食物量以克、毫升或其他常见措施表示。前面已经介绍了 FFQ 开发的更多细节。

每日食物消耗量计算如下

每日食物消耗量 1/4 份量

消费频率

其中消耗频率为：从不 = 0;1-3 次/月 = 0·07;1-2 次/周 = 0·21;3-6 次/周 = 0·64;1 次/天 = 1;≥ 2 次/天 = 2。

每个食物组的 Na 估计值计算为

每日食用的食物 食物的钠含量

源自美国农业部（USDA）和当地食品成分表。

使用改进的饮食方法评估膳食钠摄入量 24DR 加上酌情盐问题。为了估计可自由支配的盐，参与者被要求在24DR中的每一个人的早餐、午餐和晚餐中分别回答两个与盐相关的问题：

问题1：你在准备饭菜时用了多少盐？

a = 无，b = 一点点，c = 中等，d = 很多

问题2：你有没有在盘子里加额外的盐（食盐）？

A = 否，B = 是。

对于问题 1，每个答案应用了以下 Na 量：a = 无 = 0 毫克 Na，b = 每 100 克食物 50 毫克 Na，c = 中等 = 每 100 克食物 350 毫克 Na，d = 很多 = 每 100 克食物 600 毫克 Na。这些估计是基于希腊食品管理局（EFET）的相关声明/评估：“如果每100克食物含有超过0.6克的钠（或1.5克盐），那么它的钠/盐含量就很高。如果食物每 100 克含有 0·1 克或更少的钠，那么它的钠/盐含量很低。如果每 100 克的盐含量介于这些值之间，则食物中等水平的盐。24DR 的份量是根据食物当量和当地食物成分表以克计算的。

对于问题 2，答案“是”被定义为 2 少许盐，相当于 775 毫克 Naand，当答案为“否”时，没有添加 Na（0 毫克）

然后将问题 1 和 2 得出的平均 Na 添加到从 24DR 得出的 Na 中，计算公式为

24DR þ SQ 1/4 Na 来自 24DR

þ 表示早餐问题 1 中的 Nað Þ þ 表示午餐问题 1 中的 Nað Þ 表示晚餐问题 1 中的 Nað Þ 表示早餐问题 2 中的 Nað Þ 表示早餐问题 2 中的 Nað Þ 表示午餐问题 2 中的 Nað Þ 晚餐问题 2 中的 Nað Þ 表示晚餐问题的 Nað Þ

膳食（早餐、午餐和晚餐）的 Na 是根据问题 1 和 2 的 Na 摄入量估计值（三个 24DR 的平均值）计算的。

24DR 加 15%。应用了另一种方法来估计盐的酌情使用量。如前所述，我们根据以下假设计算了可自由支配的 Na：烹饪和餐桌中的 Na 占我们人口总 Na 摄入量的 15%。具体而言，总 Na 摄入量计算为

24DR þ 15% 1/4 Na 来自 24DR

þ ð15% Na 来自 24DRÞ

钠FFQ。为了改进Na的估计，前面提到的FFQ的食物清单增加了富含Na的食物和与酌情使用盐（NaFFQ）有关的饮食行为的问题。添加的食物组和问题在补充中列出（见在线补充表 1）。添加的食物包括咸黄油和人造黄油、几种富含 Na 奶酪（例如罗克福、帕尔马干酪、伊丹、豪达、格鲁耶尔等）、咸饼干/饼干、鱼罐头/海鲜和精制番茄汁。根据希腊食品管理局（EFET）的说法，为了估计熟食和沙拉中添加的Na，使用了NaFFQ的问题b（您在熟食和沙拉中使用了多少盐？见在线补充表1）。参与者的答案计算为无 = 0 毫克钠，一点 = 50 毫克钠/100 克食物，适量 = 350 毫克钠/100 克食物，多 = 600 毫克钠/100 克食物，非常多 = 900 毫克钠/100 克食物。然后，将参与者在问题 b 中的回答中得出的量化 Na 添加到每 100 克 NaFFQ 食物的每顿熟食和沙拉中。然后从现有的FFQ中计算NaFFQ 1/4 Na

þ Na 来自食品中添加的富含 Na þ Na 添加到熟食和沙拉中

熟食包括米饭，土豆，红肉和白肉，鱼和海鲜，豆类，传统菜肴和自制馅饼。沙拉包括所有蔬菜，生的或煮的。

膳食钠评估：方法比较

https://doi.org/10.1017/S1368980021004390 Published online by Cambridge University Press

人体测量参数的评估

参与者的体重在不穿鞋或厚衣服的情况下测量到最接近的 0·1 公斤（Tanita 身体成分分析仪，BC-418）。身高测量不穿鞋，参与者站立时肩膀放松，手臂自由垂下，头部在法兰克福水平面上（SECA 213）。BMI计算为体重/（身高）（kg/m）。

CVD危险因素的评估和定义

高血压被定义为使用抗高血压药物和/或办公室血压测量> 139/89 mmHg（休息至少 10 分钟后仰卧位间隔 1 分钟的三次连续读数的平均值;Microlife WatchBP办公室，Microlife AG，Widnau，瑞士）。血脂异常被定义为使用降脂药物和/或低密度脂蛋白胆固醇水平> 160 mg/dl。糖尿病被定义为空腹血糖高于 126 mg/dl 或 HbA1c ≥6·5% 和/或降糖治疗。吸烟或吸电子烟的定义是每周每天至少使用一支香烟或使用电子烟。

统计分析

所有分析均使用 SPSS 版本 25（IBM Corp. 2017 年发布，IBM Corp.）进行。连续变量表示为平均±，分类变量表示为绝对频率和百分比 （%）。显著性水平设定为P值<0·05。使用 Kolmogorov-Smirnov 检验和直方图检验变量的分布正态性。方法之间的差异（平均值偏差）计算为 24UNa 减去其他 DM 和 UM 的 Na 测量值。在适当的情况下，使用配对样本 t 检验和 Wilcoxon 检验来确定 Na 平均值差异的显着性。为了评估 24UCol 与其他 Na 估计方法之间的相关性，应用了 Pearson 相关系数（对于正态分布变量）和 Spearman 相关系数（对于非正态分布的变量）。还用类内相关系数（ICC）评估了不同Na估计方法之间的一致性。人们普遍认为，对ICC值没有绝对的解释。然而，在本研究中，我们使用了 Koo 和 Li 的建议;因此，ICC 值 <0·5 表示可靠性差，ICC 介于 0·5 和 0·75 之间表示可靠性中等，ICC 介于 0·75 和 0·9 之间表示可靠性良好，ICC 值大于 0·90 表示可靠性极好。

使用Bland-Altman图评估Na估计方法与24UCol之间的差异，并评估它们之间的一致性。通过平均差±1·96×的差值计算两种不同Na估计值之间的一致性上限和下限

线性回归分析用于评估差值和均值（24UCol与每种Na估计方法之间的）的关联。在排除所有 24UCol（敏感性分析）不完整的受试者后，重复了关于 24UCol 与每种 Na 估计方法和 ICC 之间相关性的分析，并在补充中提出。不完整的24UCol的排除标准是根据国际参考书目确定的，并在补充材料中列出（见在线补充表2）。

结果

使用122名具有24UCol可用数据的受试者进行分析（56.0±12.6岁;55.7%男性）（表2）。UM 和 DM 的可用样本量为 Spot UM， n 71;24DR = 119;FFQ，第 87 页;NaFFQ，n 60（表2）。研究人群的描述性特征见表2。7·4%的受试者不完整（表2）。

表 3 显示了所有可用的 UM 和 DM 的平均 Na 摄入量或排泄量，以及 24UNa 与其他每种 Na 估计方法之间差异的显着性。平均24UNa为2810·4±1303·9 mg/d。关于点尿方法，所有方法都高估了24UNa（平均偏差范围：-1780·9至-492·0 mg），其中INTERSALT无点K方程的偏差最小（-492·0±1223·2 mg）（表4）。关于现有的DM，他们都低估了24UNa（平均偏倚范围：876·6至1211·6 mg）。从改进的DM来看，24DR þ 15 % 和 24DR þ SQ 低估了 24UNa（分别为 876·6 ± 1342·6 和 923·3 ± 1345·8 mg，P < 0·001），但 NaFFQ 略微高估了 24UNa，显示出所有 DM 和 UM 的最小偏倚 （-290·2 ± 1336·2 mg）（表 3）。

表 4 显示了 Pearson 和 Spearman 的相关性检验以及 24UCol 与 UM 和 DM 之间的 ICC。 关于点尿方法，Mage 方程与 24UCol 的相关性最强 （r = 0·596，P < 0·001），所有其他方程的可靠性均为中等（ICC 范围：0·59–0·74）。从现有的DM来看，它们都与24UCol呈弱相关（r = 0·232–0·263，P < 0·05）。关于改进的DM，24DR þ 15 %和24DR þ SQ与24UCol的相关性较弱（r = 0·263-0·296，P ≤ 0·01），可靠性较差（ICC范围：0·42–0·44），但NaFFQ与24UCol的相关性最强（r = 0·497，P ≤ 0·01）且可靠性中等（ICC 0·66（95%CI 0·43， 0·80)).在亚组分析中（数据见补充表3）：（a）五个亚组中有四个同意Mage方程与24UCol表现出最强的相关性（r = 0·625–0·700，P < 0·001）和（b）五个亚组中有三个同意川崎方程是唯一具有良好可靠性的方法（ICC范围： 0·76–0·80）和所有

C Tsirimiagkou 等人。

org/10 https://doi.org/10.1017/S1368980021004390 Published online by Cambridge University Press

表2

总样本的研究人群的描述性特征和每种 Na 估计方法

泌尿方法

饮食方法

24UCol

第 122 页

尿斑

第 71 页

第 119 页

第 87 页

NaFFQ系列

年龄、岁月

重量，kg

高度，厘米

169·9

170·7

169·9

170·3

170·8

BMI，kg/m

能量，kcal/d

1998·8

668·3

2238·7

75337

2210·7

695·5

现有 24DR

þ 15 %

Na derived from food, mg/d  

1633·8  

763·6  

1633·8  

763·6  

1633·8  

763·6  

1704·3  

800·0  

1793·5  

873·5 

Na derived from table salt, mg/d  

288·3  

134·8*  

1197·4  

1047·2* 

Na derived from cooking salt, mg/da  

276·4  

261·6  

Males  

Smoking 

Current (cigarette/e-cigarette)  

Ex smoking  

Never  

DMS drugs  

Hypertension  

Hypertension drugs  

Dyslipidaemia  

Dyslipidaemia drugs  

Autoimmune/inflammatory disease  

Infectious disease  

Incomplete 24 h UCol  

24UCol, 24-h urine collection; 24DR, 24-h dietary recalls (three 24DR were performed); 24 DR  

15 %, 24-h dietary recalls Na plus 15 % (discretionary Na); 24 DR

SQ, 24-h dietary recalls Na plus discretionary salt questions; T1DM, type 1 

diabetes mellitus; T2DM, type 2 diabetes mellitus.*Na derived from table and cooking salt. 

Dietary sodium assessment: comparison of methods  

https://doi.org/10.1017/S1368980021004390 Published online by Cambridge University Press

Table 3 Na intake/excretion for each dietary and urinary Na estimation method, bias of mean values and comparisons with the 24-h urine collection 

Na intake or excretion 

Bias (24UNa minus each Na estimation 

method) mean  

Urinary methods 

24UCol, mg/d  

2810·4  

1303·9  

Kawasaki, mg/d  

4523·0  

1331·0  

−1780·9  

1235·2 <0·001 

Tanaka, mg/d  

4862·1  

10 633·2  

−894·8  

1154·1 <0·001 

INTERSALT with spot K, mg/d 

3209·4  

869·0  

−599·0  

1140·0 <0·001 

INTERSALT without spot K, mg/d 

3207·8  

843·1  

−492·0  

1223·2 0·001 

Mage, mg/d  

3438·8  

2494·8  

−722·6  

2050·6 0·016 

Toft, mg/d  

3852·8  

955·7  

−1136·6  

1165·6 <0·001 

Dietary methods Existing dietary methods 24 DR, mg/d 119 1633·8 763·6  

1211·6  

1298·8 <0·001 

FFQ, mg/d  

1704·3  

800·0  

1058·7  

1335·7 <0·001 

Improved dietary methods 24 DR þ 15 %, mg/d 119 1922·2 898·3  

923·3  

1345·8 <0·001 

24 DR þ SQ, mg/d 119 1968·9 917·0  

876·6  

1342·6 <0·001 

NaFFQ, mg/d  

2990·9  

1397·5  

−290·2  

1336·2 0·098 

24UCol, 24-h urine collection; 24UNa, 24-h urine Na; 24 DRNa, 24-h dietary recalls Na; 24 DR þ 15 %, 24-h dietary recalls Na plus 15 % (discretionary Na); 24 DR þ SQ, 24-h dietary recalls Na plus discretionary salt questions. 

Table 4 Pearson’s and spearman correlations & intraclass correlation coefficients between 24-h urine collection and the other Na estimation methods 

Na estimation methods  

Total sample  

95 % CI 

Spot urine methods  

Kawasaki  

r 0·583** 

ICC 0·74 0·58, 0·84 n 71 Tanaka r 0·542** 

ICC 0·66 0·46, 0·79 n 71 INTERSALT with spot K r 0·492** 

ICC 0·63 0·40, 0·77 n 67 INTERSALT without spot K r 0·469** 

ICC 0·59 0·35, 0·75 n 71 Mage r 0·596** 

ICC 0·65 0·44, 0·78 n 71 Toft r 0·570** 

ICC 0·68 0·48, 0·80 n 71 Dietary methods Existing dietary methods 24DR r 0·263** 

ICC 0·40 0·14, 0·58 n 119 FFQ r 0·232* 

ICC 0·39 0·06, 0·60 n 87 Improved dietary methods 24DR þ SQ r 0·296** 

ICC 0·44 0·20, 0·61 n 119 24DR þ 15 % r 0·263** 

ICC 0·42 0·17, 0·60 n 119 NaFFQ r 0·497** 

ICC 0·66 0·43, 0·80 n 60 

ICC, intraclass correlation coefficient; 24UNa, 24-h urine Na; 24 DRNa, 24-h dietary recalls Na; 24DRNa þ 15 %, 24-h dietary recalls Na plus 15 % (discretionary Na); 24-h DRNa þ SQ, 24-h dietary recalls Na plus discretionary salt questions. *P < 0·05. 

**P ≤ 0·01.

C Tsirimiagkou et al. 

org/10 https://doi.org/10.1017/S1368980021004390 Published online by Cambridge University Press

the five subgroups agreed that regarding the existing and the improved DM, NaFFQ was the only method presenting moderate reliability (ICCs range: 0·44–0·51), while all the other DM presented poor reliability (ICC range: 0·20–0·32). 

Bland–Altman plots for all the spot urine methods, existing DM, and improved DM are presented in Figs 1, 2, and 3, respectively. Regarding spot urine methods, the use of equations of Toft, INTERSALT with spot K and INTERSALT without spot K resulted in underestimation at lower levels and overestimation at higher levels of Na excretion in Bland–Altman plots (Fig. 1). On the contrary, Mage equation was the only method providing the opposite finding, presenting overestimation at low levels of Na excretion and underestimation at higher levels. Finally 

the Kawasaki equation exhibited a homogeneous variation as Na excretion levels increase (Fig. 1). All methods presented wide ranges of agreement (Kawasaki: −4201·8 to 640·0; Mage: −4741·7 to 3296·6; Toft: −3421·2 to 1148·0; INTERSALT without spot K: −2889·4 to 1905·4; INTERSALT with spot K: −2833·3 to 1635·4; Tanaka: −3156·9 to 1367·3) (Fig. 1). Linear regression analysis revealed statistically significant associations between the difference and the mean of 24UCol and all the spot urine methods, except from the Kawasaki equation (β = 0·028, P = 0·818) (Fig. 1). Regarding the existing DM (Fig. 2), both of them presented consistent bias in Bland–Altman plots, underestimating the 24UNa in low levels of Na intake and overestimating in high levels of Na intake, while presenting wide ranges of agreement in Bland–Altman plots (24DR: −1334·1 to 3757·4; FFQ: −1559·2 to 3676·7) 

2000·00

–2000·00

–4000·00

(24UNa - Kawasaki)

(24UNa - INTERSALT without spot K)

(24UNa - INTERSALT with spot K)

(24UNa - Mage)

(24UNa - Toft)

(24UNa - Tanaka)

–6000·00

–8000·00

–4000·00

–2000·00

2000·00

2000·00

–2000·00

–4000·00

1000·00 2000·00 3000·00 4000·00

Mean (24UNa - Kawasaki)

Mean (24UNa - INTERSALT without spot K)

Mean (24UNa - INTERSALT with spot K)

Mean (24UNa - Tanaka)

Mean (24UNa - Mage)

Mean (24UNa - Toft)

5000·00 6000·00

1000·00

2000·00

3000·00

4000·00

5000·00 6000·00

1000·00 2000·00

3000·00

4000·00

5000·00

6000·00

1000·00 2000·00 3000·00 4000·00 5000·00 6000·00 7000·00

7000·00

Lower limit = –4201·8

Upper limit = –640·0

Upper limit = 3296·6

bias = –1780·9

b = 0·028

b = 0·632

p = 0·818

p < 0·001

Lower limit = –4741·7

Lower limit = –2833·3

Lower limit = –3421·2

Lower limit = –3156·9

Bias = –894·8

Upper limit = 1148·0

Upper limit = 1367·3

Upper limit = 1635·4

Upper limit = 1905·4

Lower limit = –2889·4

Bias = –492·0

Bias = –722·6

Bias = –599·0

Bias = –1136·6

b = – 0·719

b = 0·511

p < 0·001 b = 0·460

p < 0·001

b = 0·560

p < 0·001

p < 0·001

2000·00 4000·00 6000·00 8000·00

12000·00

1000·00 2000·00

3000·00

4000·00

5000·00

6000·00

10000·00

2500·00

–2500·00

–5000·00

2000·00

–2000·00

–4000·00

2000·00

–2000·00

–4000·00

–7500·00

–10000·00

Fig. 1 (colour online) Bland–Altman plots comparing 24-h urinary Na excretion with Na estimated by spot urine equations. Solid line is the mean difference between methods and dashed lines are the 95 % CI of the difference between methods. Limits of agreement of the two Na assessment methods, defined as mean difference ± 1·96 × of differences. 24UNa, Na estimated by 24-h urine collection 

4000·00

2000·00

–2000·00

4000·00

2000·00

–2000·00

(24UNa - 24DRNa)

∆ (24UNa - FFQ)

1000·00

2000·00

3000·00

4000·00

5000·00

1000·00

2000·00

3000·00

4000·00

5000·00

Mean (24hUNa - 24hDRNa)

Mean (24hUNa - FFQ)

Lower limit = –1334·1

Lower limit = –1559·2

Upper limit = 3757·4

Upper limit = 3676·7

Bias = 1211·6

Bias = 1058·7

b = 0·780

b = 0·749

p < 0·001

p < 0·001

Fig. 2 (colour online) Bland–Altman plots comparing 24-h urinary Na excretion with Na estimated by existing DM. Solid line is the mean difference between methods, and dashed lines are the 95 % CI of the difference between methods. Limits of agreement of the two Na assessment methods, defined as mean difference ± 1·96 × of differences. 24UNa, Na estimated by 24-h urine collection; 24-h DRNa, Na estimated by 24-h dietary recalls 

Dietary sodium assessment: comparison of methods  

org/10 https://doi.org/10.1017/S1368980021004390 Published online by Cambridge University Press

(Fig. 2). Linear regression analysis revealed statistically significant association between the difference and the mean of 24UCol and all the DM (Fig. 2). Regarding the improved DM, the NaFFQ was the only one showing: (a) a homogeneous variation as the mean Na intake increases in Bland–Altman plots, however, presenting wide ranges of agreement (−2909·2 to 2328·8) and (b) not statistically significant association between the difference and the mean of 24UCol and improved DM in linear regression analysis (β = 0·142, P = 0·354) (Fig. 3). The other two improved DM (24DR þ 15 % & 24DR þ SQ) underestimated the 24UNa at low levels of Na intake and overestimated at high levels of Na intake, presenting wide ranges of agreement (24DR þ SQ: −1334·1 to 3508·1; 24DR þ 15 %: −1714·5 to 3561·2) (Fig. 3). 

All the analyses were repeated using 1 dash of salt instead of 2 in the question 2 of the improved dietary recalls (24DR þ SQ), and similar findings were observed (data not shown). 

Discussion 

This study aimed to assess and compare the most commonly used in population studies UM and DM for mean Na intake and develop a new accurate and easy to use clinical tool for Na estimation in high CVD risk 

populations. The main findings of this study are (i) the existing DM tend to underestimate and spot urine methods tend to overestimate the true Na intake; (ii) all the existing DM are weakly correlated and present poor agreement with the 24UCol, and all the spot urine methods are moderately correlated and present moderate agreement with the 24UCol and (iii) the new NaFFQ is the only method that performed better in the analysis, having simultaneously the smallest bias in mean differences, the strongest correlation with the 24UCol regarding DM and a homogeneous variation as the mean Na intake increases in Bland–Altman plots, but still wide limits of agreement. 

Spot urine collection is an easily applicable alternative in estimating dietary Na intake. Increasing studies aim to reveal the most accurate formula for converting spot Na to 24UNa, comparing not only those commonly usedbut also those newly designed against the gold-standard 24UCol. The mostly studied formulas are the INTERSALT equation, the Tanaka equation and the Kawasaki equation. Despite some controversies , a large number of studies support that among the existing equations, the INTERSALT performs better in estimating the 24UNa showing the least bias (44,45,49,51,52). In our findings, the INTERSALT equation presented the lowest bias among all the other equations; however, it was moderately correlated with 24UNa and also presented consistent bias in 

6000·00

2000·00

(24UNa - NaFFQ)

(24UNa - 24DRNa+15 % )

(24UNa - 24DRNa+SQ)

–2000·00

–4000·00

1000·00 2000·00

3000·00

4000·00

Mean (24UNa - 24DRNa+15%)

Mean (24UNa - 24DRNa+SQ)

Mean (24UNa - NaFFQ)

5000·00

6000·00

1000·00

2000·00

3000·00

4000·00

5000·00

6000·00

Lower limit = –1334·1

Lower limit = –1714·5

Upper limit = 3561·2

Upper limit = 3508·1

Upper limit = 2328·8

Bias = 876·6

Bias = 923·3

b = 0·142

b = 0·559

b = 0·524

p = 0·354

p < 0·001

p < 0·001

4000·00

2000·00

–2000·00

2000·00

4000·00

6000·00

–4000·00

4000·00

6000·00

2000·00

–2000·00

–4000·00

4000·00

Lower limit = –2909·2

Bias = –290∙2

Fig. 3 (colour online) Bland–Altman plots comparing 24-h urinary Na excretion with Na estimated by improved DM. Solid line is the mean difference between methods and dashed lines are the 95 % CI of the difference between methods. Limits of agreement of the two Na assessment methods, defined as mean difference ± 1·96 × of differences. 24UNa, sodium estimated by 24-h urine collection; 24-h DRNa þ SQ, Na estimated by 24-h dietary recalls plus salt-related questions 

C Tsirimiagkou et al. 

https://doi.org/10.1017/S1368980021004390 Published online by Cambridge University Press

Bland–Altman plots by underestimating Na intake at low levels of Na excretion and overestimating at high levels of Na excretion. However, it is important to note that the studies supporting the use of INTERSALT equation as the best alternative of 24UCol for Na estimation have all been conducted in general populations (44,45,49,51,52) , which is in contrast to our high CVD risk population. Indeed, the evidence is not supportive of the use of the INTERSALT equation in high-risk patients, having chronic diseases such as chronic kidney disease or hypertension . Dougher et al. compared commonly used equations for Na estimation in 129 chronic kidney disease patients . According to their findings, the authors conclude that spot urine equations do not estimate accurately dietary Na intake in this group of people. Similarly, when Ma et al. assessed Na intake by the INTERSALT, the Tanaka and the Kawasaki equations in 365 high-risk stroke patients, they found poor correlations (r = 0·35–0·38), poor reliability (ICCs = 0·31–0·38) and significant biases among all the three methods compared with the 24UCol . These findings are in agreement not only with our study but also with a significant number of studies, which do not recommend the use of spot equations for dietary Na estimation (17,47,48,54,55). It is important to note that Na excretion presents a circadian variability, which potentially could influence the estimations derived from spot urine collections. A systematic review of studies comparing the 24UCol and spot urine collections for estimating salt intake, conducted by Ji et al., included twenty studies and 1·380·130 participants, concluded that although it is of great interest to replace the 24UCol as a method for Na intake estimation, the best alternative UM remains uncertain as a wide range of correlations (r = 0·17–0·94) between 24UCol and the other methods presented in their work . Also, a systematic review and meta-analysis in 10·414 participants from thirty-four countries showed that ‘estimates based upon spot urine samples have excellent sensitivity (97 %) and specificity (100 %) at classifying mean population salt intake above or below the World Health Organization maximum target of 5 g/d but underestimate intake at high levels of consumption and overestimate at lower levels of consumption’. Even more interestingly, in a recent analysis of TOHP (Trials of Hypertension Prevention) study follow-up data, conducted by He et al., estimated values of Na excretion (using the Kawasaki, INTERSALT with spot K and Tanaka equations) – examining the same population sample – altered the linear association between 24UNa and mortality to J- or U-shaped. The authors concluded that these urinary Na estimation methods ‘were systematically biased with overestimation at lower levels and underestimation at higher levels’, indicating that estimation of Na through spot urine specimens is inaccurate. All these findings are consistent to a WHO/PAHO statement in the protocol for population level Na determination in 24-h urine sample, declaring that ‘the use of spot-urine is discouraged as a 

method to determine Na, potassium or iodine intake because of the limitations and uncertainty inherent in the method’. 

As regard to the existing DM for Na estimation, although it is useful and efficient to highlight food items rich in Na, several methodological disadvantages have been raised. The most commonly discussed include the difficulty or even inability to assess and quantify discretionary Na; deviated estimations of Na due to high variability in Na content in recipes of homemade and manufactured food; the absence of Na derived from medicines and dietary supplements and participant-related issues (underreporting and difficulty to recall all the food and beverages consumed; socially desired answers and dietary behaviour modification) . A small number of studies suggest that DM, such as food diaries or multiple 24DRs, can be used for Na estimation, having the ability to predict over 90 % of 24UNa . However, the majority of the available studies have reported that Na estimation based on DM tends to underestimate 24UNa (levels of underreporting 15–40 %) and correlates weakly or moderately with 24UNa (r≈0·15–0·50) . This is in line with our findings, showing weak correlations, poor reliability and high levels of bias, suggesting that the existing DM for Na estimation are inaccurate. In a recent meta-analysis including twenty-eight studies, McLean et al. compared 24DR with 24UCol . 24DR underestimated mean Na intake by 607 mg/d, but high quality 24DR improved accuracy. The authors concluded that 24UCol remains the most accurate method to assess population Na intake; however, highquality 24DR (use of multiple pass methods, accurate food composition databases and quantification of discretionary salt) could be used if 24UCol is not feasible . 

To our knowledge, studies comparing different DM and UM simultaneously for Na intake estimation are scarce. A recent study compared the spot urine collection (using the INTERSALT equation) v. the 24DR (without quantifying the discretionary use of table salt) in a large sample of adults in New Zealand, consisting of 3321 participants . The authors observed poor agreement between estimated Na intake from spot urine collection and those from 24DR . In another study, a plethora of different DM and UM were compared with a PABA-validated 24UCol . The assessment of Na intake included an FFQ, a modified 24DR and three equations to convert the spot Na to 24UNa (INTERSALT, Tanaka and Kawasaki). In this study neither DM nor UM provided accurate estimations at individual level, but for group means, the DM and some of the UM may be useful for Na estimation . However, the method for the quantification of Na intake has not been clearly described . 

FFQ are commonly used in dietary Na assessment in population-based studies, having the ability to bypass problems related to day-to-day variability of Na intake and cover larger time periods of intake. The last four decades several FFQ have been designed for the estimation 

Dietary sodium assessment: comparison of methods  

https://doi.org/10.1017/S1368980021004390 Published online by Cambridge University Press

of Na (or salt) intake . However, most of them present weak correlations with the 24UCol, ranging from 0·19 to 0·35 . Furthermore, the available FFQ for Na assessment have been designed for particular ethnic groups . To our knowledge, only two of them have been developed for hypertensive subjects but until now, there was no FFQ for Na estimation in other high CV risk groups, such as patients with dyslipidaemia, diabetes mellitus, infectious or autoimmune diseases. Recently, McLean et al. published a systematic review of the literature, regarding the assessment of dietary Na intake using FFQ and 24UCol . This work revealed a poor agreement between estimates of Na from FFQ and 24UCol , indicating that the Na FFQ until now are inadequate to estimate the true intake. 

The novel NaFFQ was created to accurately estimate Na intake in high CVD risk populations, calculating not only Na derived from food content, but table and cooking Na as well. Our aim was to cover the need of an easily applicable in epidemiological studies and reliable tool for group means of Na intake, which could lead to a better management of high CVD-risk populations. According to our findings, this tool presented the best correlation with – and the lowest bias from – the 24UCol compared with all the existing DM, even when attempts to further improve the accuracy of 24DR were applied. However, despite these promising findings regarding NaFFQ, it provided very wide limits of agreement in Bland– Altman plots, reaching ∼3000 mg/d, indicating that future improvements have to be addressed. A limitation of our study is the use of a single 24UCol. Due to the day-today variability in Na intake and excretion, multiple 24UCol are recommended either for assessing accurately usual individual Na intake or for a more reliable record of dietary Na in studies investigating its relationship with health or disease . In our study, our aim was to estimate Na intake in group means and not in individual level, so the use of single 24UCol, which is very common in epidemiological studies, was reasonable. Indeed, the use of a single 24UCol v. three to seven 24UCol have been reported to provide similar mean levels of Na excretion at the population level . Second, an important limitation to be mentioned is the method used for the quantification of discretionary salt. In our study, the use of dashes of salt, as well as the cut-offs that were designed for processed food, may lead to several concerns and systematic bias. However, until today, the estimation of discretionary salt in studies remains a challenge for the investigators, and there is no generally accepted protocol to be applied in dietary surveys . Moreover, the NaFFQ is population specific and has not been externally validated in other populations. Nevertheless, the methodology used here could be used to adapt other FFQ, designed for other population groups, in order to more accurately estimate Na intake. 

In conclusion, the available DM and spot urine methods present poor accuracy compared with the gold-standard 24UCol. The new FFQ – specifically designed for Na estimation – is a promising method to detect a mean Na intake at population level in high CVD-risk people. Future validation of this tool in larger populations would verify its accuracy and/or provide evidence for further amelioration, making it a reliable and easy to use clinical tool for Na quantification, in population-based studies. Similar approaches might be useful for other populations. 

Acknowledgements 

Acknowledgements: We gratefully thank all the partners for their contribution in this work. Financial support: The first author of this research work (C.T.) was supported by the Hellenic Foundation for Research and Innovation (HFRI) and the General Secretariat for Research and Technology (GSRT), under the HFRI PhD Fellowship grant (GA. no. 186619/I2). Conflicts of interest: The authors declare no conflict of interest. Authorship: C.T.: acquisition, analysis and interpretation of data, article drafting and final approval of the version to be submitted; K.K.: conception and design of the study, revising article critically for important intellectual content and final approval of the version to be submitted; E.D.B.: acquisition of data, article drafting and final approval of the version to be submitted; A.A.A.: analysis and interpretation of data, article drafting and final approval of the version to be submitted; T.G.P.: analysis and interpretation of data, revising article and final approval of the version to be submitted; M.Y.: conception and design of the study, revising article critically for important intellectual content and final approval of the version to be submitted; A.D.P.: conception and design of the study, revising article critically for important intellectual content and final approval of the version to be submitted. Ethics of human subject participation: This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the Medical School of the National and Kapodistrian University of Athens. Written informed consent was obtained from all subjects. 

Supplementary material 

For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980021004390 

References 

Blaustein MP, Leenen FH, Chen L et al. (2012) How NaCl raises blood pressure: a new paradigm for the pathogenesis 

C Tsirimiagkou et al. 

https://doi.org/10.1017/S1368980021004390 Published online by Cambridge University Press

of salt-dependent hypertension. Am J Physiol Heart Circ Physiol 302, H1031–H1049. 

Mozaffarian D, Singh GM & Powles J (2014) Sodium and cardiovascular disease. N Engl J Med 371, 2138–2139. 

Strazzullo P, D’Elia L, Kandala NB et al. (2009) Salt intake, stroke, and cardiovascular disease: meta-analysis of prospective studies. BMJ 339, b4567. 

WHO (2012) Guideline: Sodium Intake for Adults and Children. Geneva: World Health Organization. 

Powles J, Fahimi S, Micha R et al. (2013) Global, regional and national sodium intakes in 1990 and 2010: a systematic analysis of 24 h urinary sodium excretion and dietary surveys worldwide. BMJ Open 3, e003733. 

McLean RM (2014) Measuring population sodium intake: a review of methods. Nutrients 6, 4651–4662. 

Colin-Ramirez E, Arcand J & Ezekowitz JA (2015) Estimates of dietary sodium consumption in patients with chronic heart failure. J Card Fail 21, 981–988. 

Schachter J, Harper PH, Radin ME et al. (1980) Comparison of sodium and potassium intake with excretion. Hypertension 2, 695–699. 

Bingham SA, Williams R, Cole TJ et al. (1988) Reference values for analytes of 24-h urine collections known to be complete. Ann Clin Biochem 25, 610–619. 

Joossens JV & Geboers J (1984) Monitoring salt intake of the population: methodological considerations. In Surveillance of the Dietary Habits of the Population with Regard to Cardiovascular Diseases, EURONUT Report 2, pp. 61–73 [GG De Backer, HT Pedoe and P Ducimetiere, editors]. Wageningen: Department of Human Nutrition, Agricultural University. 

Kawasaki T, Itoh K, Uezono K et al. (1993) A simple method for estimating 24 h urinary sodium and potassium excretion from second morning voiding urine specimen in adults. Clin Exp Pharmacol Physiol 20, 7–14. 

Mage DT, Allen RH & Kodali A (2008) Creatinine corrections for estimating children’s and adult’s pesticide intake doses in equilibrium with urinary pesticide and creatinine concentrations. J Expo Sci Environ Epidemiol 18, 360–368. 

Tanaka T, Okamura T, Miura K et al. (2002) A simple method to estimate populational 24-h urinary sodium and potassium excretion using a casual urine specimen. J Hum Hypertens 16, 97–103. 

Toft U, Cerqueira C, Andreasen AH et al. (2014) Estimating salt intake in a Caucasian population: can spot urine substitute 24-h urine samples? Eur J Prev Cardiol 21, 1300–1307. 

WHO/PAHO Regional Expert Group for Cardiovascular Disease Prevention through Population-Wide Dietary Salt Reduction Sub-group for Research and Surveillance (2010) Protocol for Population Level Sodium Determination in 24-h Urine Samples. Washington DC: PAHO. 

Brown IJ, Dyer AR, Chan Q et al. (2013) Estimating 24-h urinary sodium excretion from casual urinary sodium concentrations in western populations: the INTERSALT study. Am J Epidemiol 177, 1180–1192. 

Peng Y, Li W, Wang Y et al. (2016) Validation and assessment of three methods to estimate 24-h urinary sodium excretion from spot urine samples in Chinese adults. PLoS One 11, e0149655. 

Ji C, Miller MA, Venezia A et al. (2014) Comparisons of spot v. 24-h urine samples for estimating population salt intake: validation study in two independent samples of adults in Britain and Italy. Nutr Metab Cardiovasc Dis 24, 140–147. 

Kelly C, Geaney F, Fitzgerald AP et al. (2015) Validation of diet and urinary excretion derived estimates of sodium excretion against 24-h urine excretion in a worksite sample. Nutr Metab Cardiovasc Dis 25, 771–779. 

Elliott P & Brown I (2007) Sodium Intakes around the World. 

Geneva: World Health Organization; available at http:// www.who.int/iris/handle/10665/43738. 

James WP, Ralph A & Sanchez-Castillo CP (1987) The dominance of salt in manufactured food in the sodium intake of affluent societies. Lancet 1, 426–429. 

Mattes RD & Donnelly D (1991) Relative contributions of dietary sodium sources. J Am Coll Nutr 10, 383–393. 

Ferreira-Sae MC, Gallani MC, Nadruz W et al. (2009) Reliability and validity of a semi-quantitative FFQ for sodium intake in low-income and low-literacy Brazilian hypertensive subjects. Public Health Nutr 12, 2168–2173. 

Charlton KE, Steyn K, Levitt NS et al. (2008) Development and validation of a short questionnaire to assess sodium intake. Public Health Nutr 11, 83–94. 

D’Elia L, Manfredi M, Strazzullo P et al. (2018) Validation of an easy questionnaire on the assessment of salt habit: the MINISAL-SIIA study program. Eur J Clin Nutr 73, 793–800. 

Shepherd R, Farleigh CA & Land DG (1985) Estimation of salt intake by questionnaire. Appetite 6, 219–233. 

Faul F, Erdfelder E, Lang AG et al. (2007) G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 39, 175–191. 

Faul F, Erdfelder E, Buchner A et al. (2009) Statistical power analyses using G*Power 3.1: tests for correlation and regression analyses. Behav Res Methods 41, 1149–1160. 

Reinivuo H, Valsta LM, Laatikainen T et al. (2006) Sodium in the Finnish diet: II trends in dietary sodium intake and comparison between intake and 24-h excretion of sodium. 

Eur J Clin Nutr 60, 1160–1167. 

Malekshah AF, Kimiagar M, Saadatian-Elahi M et al. (2006) Validity and reliability of a new food frequency questionnaire compared to 24 h recalls and biochemical measurements: pilot phase of Golestan cohort study of esophageal cancer. 

Eur J Clin Nutr 60, 971–977. 

WHO Regional Office for Europe (1984) Estimation of Sodium Intake and Output: Review of Methods and Recommendations for Epidemiological Studies. Report on a WHO Meeting by the WHO Collaborating Center for Research and Training in Cardiovascular Diseases. Geneva: World Health Organization. 

Bountziouka V, Bathrellou E, Giotopoulou A et al. (2012) Development, repeatability and validity regarding energy and macronutrient intake of a semi-quantitative food frequency questionnaire: methodological considerations. 

Nutr Metab Cardiovasc Dis 22, 659–667. 

Institute of Preventive Medicine Environmental and Occupational Health (2014) National Dietary Guidelines for Adults: Scientific Documentation. Athens. ISBN: 978-960503-559-4. http://www.diatrofikoiodigoi.gr/?Page=entypoyliko-enilikes. 

Bountziouka V, Bathrellou E, Constantinidis TC et al. (2010) Repeatability of dietary patterns derived using a-priori and a-posterior methods. J Appl Biobehav Res 15, 31–60. 

U.S. Department of Agriculture & Agricultural Research Service (2019) USDA National Nutrient Database for Standard Reference, Release 22. http://www.ars.usda.gov/ ba/bhnrc/ndl. 

Trichopoulou A (1992) Composition Tables of Simple and Composite Foods. Athens: Parisianos. 

Kafatos A & Chasapidou M (2001) Composition Tables of Greek Foods. Greece: University of Crete. 

Hellenic Food Authority (2011) Nutrition Policy and Research Directorate. Information for Consumers – Salt Consumption. http://www.efet.gr/portal/page/portal/ efetnew/news/view_new?par_newID=471. 

Williams B, Mancia G, Spiering W et al. (2018) 2018 ESC/ESH guidelines for the management of arterial hypertension: the task force for the management of arterial hypertension of the European society of cardiology and the European society of hypertension: the task force for the management of arterial hypertension of the European society of cardiology and 

Dietary sodium assessment: comparison of methods  

org/10 https://doi.org/10.1017/S1368980021004390 Published online by Cambridge University Press

the European society of hypertension. J Hypertens 36, 1953–2041. 

Bartko JJ (1966) The intraclass correlation coefficient as a measure of reliability. Psychol Rep 19, 3–11. 

Koo TK & Li MY (2016) A guideline of selecting and reporting intraclass correlation coefficients for reliability research. 

J Chiropr Med 15, 155–163. 

Bland JM & Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1, 307–310. 

Bland JM & Altman DG (1995) Comparing methods of measurement: why plotting difference against standard method is misleading. Lancet 346, 1085–1087. 

Cogswell ME, Wang CY, Chen TC et al. (2013) Validity of predictive equations for 24-h urinary sodium excretion in adults aged 18–39 years. Am J Clin Nutr 98, 1502–1513. 

Swanepoel B, Schutte AE, Cockeran M et al. (2018) Monitoring the South African population’s salt intake: spot urine v. 24 h urine. Public Health Nutr 21, 480–488. 

Ma W, Yin X, Zhang R et al. (2017) Validation and assessment of three methods to estimate 24-h urinary sodium excretion from spot urine samples in high-risk elder patients of stroke from the rural areas of Shaanxi province. Int J Environ Res Public Health 14, 1211. 

Charlton K, Ware LJ, Chidumwa G et al. (2020) Prediction of 24-h sodium excretion from spot urine samples in South African adults: a comparison of four equations. J Hum Hypertens 34, 24–33. 

Polonia J, Lobo MF, Martins L et al. (2017) Estimation of populational 24-h urinary sodium and potassium excretion from spot urine samples: evaluation of four formulas in a large national representative population. J Hypertens 35, 477–486. 

Whitton C, Gay GM, Lim RB et al. (2016) Evaluation of equations for predicting 24-h urinary sodium excretion from casual urine samples in Asian adults. J Nutr 146, 1609–1615. 

Mohammadifard N, Marateb H, Mansourian M et al. (2020) Can methods based on spot urine samples be used to estimate average population 24 h sodium excretion? Results from the Isfahan salt study. Public Health Nutr 23, 202–213. 

Meyer HE, Johansson L, Eggen AE et al. (2019) Sodium and potassium intake assessed by spot and 24-h urine in the population-based Tromso study 2015–2016. Nutrients 11, 1619. 

McLean R, Williams S & Mann J (2014) Monitoring population sodium intake using spot urine samples: validation in a New Zealand population. J Hum Hypertens 28, 657–662. 

Dougher CE, Rifkin DE, Anderson CA et al. (2016) Spot urine sodium measurements do not accurately estimate dietary sodium intake in chronic kidney disease. Am J Clin Nutr 104, 298–305. 

Zhou L, Tian Y, Fu JJ et al. (2017) Validation of spot urine in predicting 24-h sodium excretion at the individual level. Am J Clin Nutr 105, 1291–1296. 

Zhang Y, Peng Y, Li K et al. (2019) Assessing whether a spot urine specimen can predict 24-h urinary sodium excretion accurately: a validation study. J Hypertens 37, 99–108. 

Ji C, Sykes L, Paul C et al. (2012) Systematic review of studies comparing 24-h and spot urine collections for estimating population salt intake. Rev Panam Salud Publica 32, 307–315. 

Huang L, Crino M, Wu JH et al. (2016) Mean population salt intake estimated from 24-h urine samples and spot urine samples: a systematic review and meta-analysis. Int J Epidemiol 45, 239–250. 

He FJ, Ma Y, Campbell NRC et al. (2019) Formulas to estimate dietary sodium intake from spot urine alter sodium-mortality relationship. Hypertension 74, 572–580. 

Pietinen P (1982) Estimating sodium intake from food consumption data. Ann Nutr Metab 26, 90–99. 

Colin-Ramirez E, Arcand J & Ezekowitz JA (2015) Estimates of dietary sodium consumption in patients with chronic heart failure. J Card Fail 21, 981–988. 

Caggiula AW, Wing RR, Nowalk MP et al. (1985) The measurement of sodium and potassium intake. Am J Clin Nutr 42, 391–398. 

Espeland MA, Kumanyika S, Wilson AC et al. (2001) Statistical issues in analyzing 24-h dietary recall and 24-h urine collection data for sodium and potassium intakes. 

Am J Epidemiol 153, 996–1006. 

Khaw KT, Bingham S, Welch A et al. (2004) Blood pressure and urinary sodium in men and women: the Norfolk cohort of the European prospective investigation into cancer (EPIC-Norfolk). Am J Clin Nutr 80, 1397–1403. 

Freedman LS, Commins JM, Moler JE et al. (2015) Pooled results from 5 validation studies of dietary self-report instruments using recovery biomarkers for potassium and sodium intake. Am J Epidemiol 181, 473–487.