Background: Zinc deficiency is easily treated and has been associated with worse outcomes in hospitalized patients. Zinc testing is time-consuming and relatively costly. We identified every zinc level measured at our teaching hospital and quantified how much zinc variation is explained by other hospital factors.

Methods: We linked tables from our hospital data warehouse from 1996-2019 to identify all patients who had at least 1 serum zinc measured during their admission. We determined the status of factors that could influence zinc levels including severity of illness, presence of bleeding or inflammation, and factors influencing zinc absorption.

Results: We identified only 318 adult patients having zinc measurement during their hospitalization. Patients were elderly (median age 71 [IQR 56-78]) and arrived by ambulance 45% of the time. Zinc was measured a median of 5 days into the hospitalization (IQR 3-13) with 154 (51.6%) recording a low level. Almost half of patients were missing at least one covariable laboratory test. Multilinear regression models using complete case analysis returned more extreme parameter estimate values and deemed as significant only two thirds of the factors identified as significant in models using data with missing values imputed. Imputed models found significant associations between lower zinc levels and recent surgery, decreased albumin, creatinine, and sodium, earlier hospitalization day of sampling, and increased patient comorbidity. These models explained 32% of zinc variation.

Conclusions: Zinc testing is rare, low zinc levels are very common, and one third of its variation in hospitalized patients is explained by other covariables.

背景：锌缺乏症易于治疗，并且与住院患者的预后差有关。锌测试既耗时又相对昂贵。我们确定了在教学医院测量的每个锌水平，并量化了其他医院因素可以解释多少锌变异。

方法：我们将1996-2019年间医院数据仓库中的表格进行链接，以识别所有入院时测得的血清锌含量至少为1的患者。我们确定了可能影响锌水平的因素的状况，包括疾病的严重程度，出血或炎症的存在以及影响锌吸收的因素。

结果：我们仅确定了318名在住院期间进行锌测量的成年患者。患者年龄较大（中位年龄为71岁[IQR 56-78]），并有45％的时间通过救护车到达。入院后5天（IQR 3-13）的锌中位数为154（51.6％），表明其水平较低。几乎一半的患者至少缺少一项协变量实验室检查。使用完整案例分析的多线性回归模型返回了更多的极端参数估计值，并且在使用推算缺失值的数据的模型中，只有三分之二的因素被认为是有意义的。推算模型发现较低的锌水平与最近的手术，白蛋白，肌酐和钠的减少，更早的住院日以及患者合并症之间存在显着的关联。

结论：锌测试很少见，低锌水平很常见，住院患者中锌含量变化的三分之一由其他协变量解释。