背景：目前大量研究表明，钠盐摄入与高血压密切关联，减少钠的摄入可以有效降低血压水平。包括美国、欧洲、中国、世界卫生组织等一些国家地区和国际组织都在相关文件或临床指南中提出了将限盐策略作为控制高血压的重要生活方式干预手段之一，并且给出了针对不同人群膳食钠的推荐摄入量。利用时点尿估算24小时尿钠排出（24-hour Urinary Sodium Excretion，24-h USE）是比较常用的估算人群钠摄入水平的方法，目前最为常用的是Kawasaki公式（K法）、INTERSALT公式（I法）和Tanaka公式（T法）。但这三种公式均是以西方发达国家人群数据建立的，目前尚缺乏针对中国居民，利用时点尿估算24小时尿钠排出量的科学方法。

目的：评价时点尿估算24小时尿钠排出量现有常用方法在中国居民人群中准确性。在中国正常居民中，利用不同时点尿估算24小时尿钠排出量，探索适用于中国居民的估算方法，并评价该方法的准确性。

方法：本研究根据研究目的和设计方案，通过三项独立研究人群探索时点尿估算24小时尿钠排出量方法。其中，在国家心血管病中心西山院区入选100名相对健康成年人群（西山人群），收集完整24小时尿样本，各次时点尿样本均单独采集，并将各次时点尿混匀作为24小时尿样本。同时利用以往的研究数据，即PURE-China尿验证人群（425名）和慢性病防治适宜技术在社区的应用项目部分尿样本数据（德兴人群152名），上述两项研究仅采集清晨时点尿样本和24小时尿样本。对时点尿估算24小时尿钠排出量进行建模探讨。在考虑分析策略时，首先剔除24小时尿肌酐排出量超过正常值范围的研究对象数据，而后采取两种建模思路，分步进行拟合。（1）由西山人群（有效分析77例）作为训练样本建模随机尿新方法I，PURE-China尿验证人群（有效分析331例）和德兴人群（有效分析133例）进行外部数据验证。而后对西山人群数据进行不同时点尿分组比较分析（有效分析60例），并用不同时点尿数据集进行内部数据验证；（2）以三项研究人群合并后重新随机抽样（有效分析541例），按照训练样本与验证样本7:3比例进行随机抽样（训练样本379例，验证样本162例），建模清晨晨起第二次时点尿（Second Morning Urine，SMU）新方法II，而后进行内部数据验证。拟合过程分步进行，即先通过年龄、性别、身高、体重进行24小时尿肌酐排出量的模型拟合，再根据时点尿钠和尿肌酐比值与24小时尿钠和尿肌酐比值的关联，通过线性模型进行24小时尿钠排出量的建模拟合。通过随机尿新方法I和SMU新方法II对人群估计值与实际测量值之间采用相关分析、残差分析、Bland-Altman法进行验证分析，并分析个体估计值与实测值之间偏差分布分析等统计学方法进行评价，同时将新方法与K法、I法和T法进行比较，评价新方法的准确性。根据验证比较结果，利用新方法对PURE-China基线人群进行24小时尿钠排出量估计，并评价盐摄入量，以及在不同人群亚组之间盐摄入量差异的比较。

结果：根据两种建模思路，所构建的时点尿估算24小时尿钠排出量的新方法如下：

（1）随机尿新方法I：

男性：PrCRE_24h=11.87×体重+0.417×身高-11.47×年龄+832.566

女性：PrCRE_24h=9.638×体重-0.085×身高-3.337×年龄+490.743

e-24-h USE=2.06×PrUCr_24h×(Na_RU / Cr_RU)^0.431

（2）SMU新方法II：

男性：PrCRE_24h=9.39×体重-3.567×身高-4.683×年龄+1498.014

女性：PrCRE_24h=8.09×体重+4.417×身高-1.906×年龄-159.528

    e-24-h USE=2.41×PrUCr_24h×(Na_SMU / Cr_SMU)^0.461

注：e-24-h USE为估计的24小时尿钠排出量，单位为mg /天；

Na_RU为RU尿钠排出浓度，单位为mg/dL；

Cr_RU为RU尿肌酐排出浓度，单位为mg/dL；

PrUCr_24h为估计的24小时尿肌酐排出量，单位为mg /天；

体重单位为kg，身高单位为cm。

新方法I估计的24小时尿钠排出量人群均值水平与实测值相比，低估1311 mg /天，偏差（估计值-实测值）较大；有30.6%的个体水平估计值与实测值相对偏差，即（估计值-实测值）/实测值，在±20%以内，36.4%的个体水平估计值绝对偏差（估计值-实测值）在±800 mg /天以内；新方法I的Bland-Altman图呈现横轴相对左侧的个体估计值易高估，右侧的个体估计值易低估的趋势。在利用不同时点尿样本进行新方法I验证时，Bland-Altman图显示SMU样本个体估计值在1.96倍标准差之间散在分布，其余各时点尿样本估计值呈现可能的线性趋势。新方法I与K法、I法和T法比较结果显示，K法准确性较好。

新方法II估计的24小时尿钠排出量人群均值水平与实测值相比，低估273 mg /天，偏差（估计值-实测值）较小；有30.8%的个体水平估计值与实测值相对偏差，即（估计值-实测值）/实测值，在±20%以内，32.1%的个体水平估计值绝对偏差（估计值-实测值）在±800 mg /天以内；新方法II的Bland-Altman图大部分个体估计值散在分布在±1.96倍标准差的区间内，效果相对较好。总体看，两种新方法中，SMU新方法II的准确性相对较好，与K法准确性相当。

新方法II对PURE-China基线人群钠排出量的估计，中国居民人群平均24小时尿钠排出量水平为5142 mg /天，折合成盐约相当于13.06 g /天。我国居民盐摄入水平呈现北方人群高于南方人群，东部高于中西部地区，农村高于城市的趋势；较高教育文化水平的人群盐摄入水平低于低教育文化水平人群。高血压人群盐摄入水平高于非高血压人群，但非高血压人群血压随尿钠排出量的变化幅度，高于高血压人群。

结论：通过随机尿新方法I和SMU新方法II利用时点尿对中国居民24小时尿钠排出量估计的准确性一般，不推荐用于个体尿钠排出量的衡量，但可适用于人群24小时尿钠排出量平均水平的估计，以便用于估算人群钠摄入量。SMU新方法II与Kawasaki方法对于估计24小时尿钠排出量的效果相当。通过新方法II估算的24小时尿钠排出量评价盐摄入量，我国居民盐摄入水平非常高，约2.6倍于WHO <5 g /天的盐每日推荐摄入量。

Background Several studies have confirmed dietary salt intake is positively associated with hypertension. While several guidelines included recommendations regarding dietary sodium intake and take the salt reduction strategy as one of important lifestyle interventions. 24-h urinary sodium excretion (24-h USE) is the preferred method and is considered the gold standard for assessing dietary salt intake, including Kawasaki formula (K), INTERSALT formula (I) and Tanaka formula (T), which all modeled from developed countries.

Objectives To develop and validate of the new estimation methods for 24-h urinary sodium excretion by using spot urine for Chinese residents.

Methods 100 healthy participants were enrolled in this study (Xishan population) and they all collected each spot urine sample and mixed finally as the 24-hour urine samples according to the study design and protocol. Data from previous researches were also used in this study, including Prospective Urban Rural Epidemiology study – China urine validation trial population (PURE-China urine trial population, 425 participants) and Application of prevention and treatment technique in community for noncommunicable diseases (Dexing population, 152 participants), for which these two researches were both only collected the morning urine and 24-hour urine samples. Eliminated the data, whose urinary creatinine concentration was exceeding the normal range for adults, the various model establishment paths was taken. (1) 77 training dataset from Xishan population was used to modeled New method I. PURE-China urine trial population (n=331) and Dexing population (n=133) were applied as the external validation dataset to assess the accuracy of New method I for estimation of 24-h USE in Chinese. (2) The merged population from three independent researches were randomized and re-sampled as training and validation dataset with 7:3 ratio (541 training samples to 162 validation samples). The model established by using linear model step by step. And the validation analysis conducted through correlation, bias distribution, residual analysis and Bland-Altman plot to assess the accuracy and the applicability of New methods (I and II) in Chinese residents. 24-h USE was estimated by using New methods and Kawasaki formulae to evaluated the salt intake in PURE-China baseline population. Subgroup analysis was also performed to explore the salt intake gap in different population characteristics.

Results Two New methods for 24-h USE estimation using spot urine were modeled and finally established.

Random urine New method I:

Male：PrCRE_24h=11.87×weight(kg)+0.417×height(cm)-11.47×age+832.566

Female：PrCRE_24h=9.638×weight(kg)-0.085×height(cm)-3.337×age+490.743

e-24-h USE=2.06×PrUCr_24h×(Na_RU / Cr_RU)^0.431

Second morning urine New method II:

Male：PrCRE_24h=9.39×weight(kg)-3.567×height(cm)-4.683×age +1498.014

Female：PrCRE_24h=8.09×weight(kg)+4.417×height(cm)-1.906×age -159.528

e-24-h USE=2.41×PrUCr_24h×(Na_SMU / Cr_SMU)^0.461

 New method I, with big bias at population level, underestimated the mean value of 24-h USE about 1311 mg/day. The bias distribution both in relative and absolute bias showed a poorer accuracy of 24-h USE at individual level. Compared with K, I, T methods, the accuracy of estimation of K method was better than others. New method II underestimated the mean value of 24-h USE about 273 mg/day, which was better for population level estimation. Furthermore, the accuracy of estimation at individual level also was not improved through bias distribution analysis. With the similar effect of K method, New method II showed a dispersed randomly Bland-Altman plot and indicated a relative accuracy. Although the gap between estimation values and real measured values was the least by using New method II (Estimated – Measured), the bias of individual estimated values were all big no matter which methods used.

 24-h USE means of Chinese urban and rural residents from PURE-China baseline population by using New method II was 5142 mg/day, which amounted to 13.06 g/day after transform the salt equivalent. Residents live in North area, or Eastern region with the higher sodium excretion. The salt intake might higher in rural area than that in urban. The differences between higher education level and lower level was significant statistically. Furthermore, salt intake was higher in hypertension population, and the effects to blood pressure by salt or sodium were much more severe in normotension people than that in hypertension people.

Conclusions The estimation for 24-h USE by using spot urine through New methods I and II did not showed a good accuracy at individual level. So this method might be not recommended to apply. However, New II method, which was similar with Kawasaki formula, could be an alternative for 24-h USE estimation at population level to evaluate the salt intake in given population. The means of salt intake of Chinese residents estimated through spot urine methods indicated an incredible high level, about 2.6 times over the recommendation of salt daily intake by WHO.