Gait speed and grip strength are simple markers of physical capability, which are associated with adverse outcomes in the elderly. However, there are few studies on the prediction of adverse outcomes in this population by the combination of the two markers.

To investigate the associations of gait speed and grip strength with adverse outcomes in geriatric inpatients.

A cohort design was used in this study. From August 2015 to December 2018, eligible geriatric inpatients aged≥65 years were recruited from Department of Geriatrics, Fuxing Hospital, Capital Medical University. We measured the gait speed and grip strength with 6-meter walking test and dynamometer, respectively. By the gait speed, the patients were divided into tertiles (T1 group: ≤ 0.6 m/s, T2 group: >0.6-0.8 m/s, T3 group: >0.8 m/s). By the grip strength, they were divided into L1, L2 and L3 tertiles (L1 group: ≤ 21.6 kg for males, ≤ 14.6 kg for females; L2 group: > 21.6 kg but ≤ 28.2 kg for males, >14.6 kg but ≤ 19.4 kg for females; L3 group: >28.2 kg for males, >19.4 kg for females). Follow-up was conducted by telephone once every three months within one year after discharge and once half a year after this until December 31, 2019. All-cause mortality and falls were recorded. Survival curves were constructed by the Kaplan-Meier method. Cox regression analysis was used to investigate the association of gait speed, grip strength, or their combination with all-cause mortality and falls. ROC curves for comparing the ability of the two makers or their combination on predicting all-cause mortality and falls.

Among the 685 patients, 29 (4.2%) were lost to follow-up, and the other 656 cases who finished the follow-up with complete data were included for analysis. During the follow-up period, 130 patients (19.8%) died from all causes and 147 patients (22.4%) experienced falls. There were 222, 225 and 209 patients in the low, moderate and high tertiles of gait speed (T1, T2 and T3 groups), and 215, 229 and 212 patients in the low, moderate and high tertiles of grip strength (L1, L2 and L3 groups), respectively. Log-rank test showed that the cumulative survival curves of all-cause mortality and falls differed significantly among T1, T2 and T3 groups (P<0.01). The same results were obtained in L1, L2 and L3 groups (P≤0.01). Cox regression analysis with adjustment for potential confounders showed that compared to patients in high tertiles of both gait speed and grip strength, the risk of all-cause mortality significantly increased in those both in low gait speed and low or moderate tertiles grip strength〔HR=3.29, 95%CI (1.13, 9.55) ; HR=3.09, 95%CI (1.08, 8.85) ; P<0.05〕, and the risk of fall significantly increased in those in low tertiles of both gait speed and grip strength 〔HR=1.92, 95%CI (1.13, 4.27), P<0.05〕. The prediction probability of the joint diagnostic model of gait speed and grip strength was estimated by Logistic regression analysis, and the AUC of the combination of them for predicting all-cause mortality and falls was 0.756 〔95%CI (0.710, 0.801) 〕, and 0.700〔95%CI (0.659, 0.741) 〕, respectively.

In geriatric inpatients, the combination of gait speed and grip strength had higher predictive value for all-cause mortality and falls, which is helpful to optimize the health management.

步速和握力是反映身体机能的简单测量指标，均与老年人不良结局相关。目前将步速和握力联合预测住院老年患者不良结局的研究较少。

探讨步速和握力与住院老年患者不良结局的关系。

采用队列研究方法，选取2015年8月至2018年12月在首都医科大学附属复兴医院老年科住院的年龄≥65岁患者，采用6米步行试验计算步速，电子读表式握力计测量握力。依据步速三分位数将患者分为T1、T2、T3组，其中T1组步速≤0.6 m/s，T2组步速>0.6~0.8 m/s，T3组步速>0.8 m/s。依据握力三分位数将患者分为L1、L2、L3组，其中L1组握力：男性≤21.6 kg、女性≤14.6 kg，L2组握力：21.6 kg<男性≤28.2 kg、14.6 kg<女性≤19.4 kg，L3组：男性>28.2 kg、女性>19.4 kg。以电话方式进行随访，患者出院后每3个月开展1次，1年后每半年开展1次。随访截至2019-12-31。记录患者全因死亡及跌倒的发生情况。采用Kaplan-Meier法绘制生存曲线，Cox回归分析探究步速与握力对患者全因死亡及跌倒的影响，受试者工作特征（ROC）曲线比较步速和握力单独及联合对老年患者全因死亡及跌倒的预测效果。

685例患者中失访29例（4.2%），最终收集完整资料共656例。随访期间共130例（19.8%）患者发生全因死亡，147例（22.4%）患者发生跌倒。T1、T2、T3组分别有222、225、209例患者，L1、L2、L3组分别有215、229、212例患者。Log-rank检验结果显示，T1、T2、T3组全因死亡、跌倒累积发生率的生存曲线比较，差异有统计学意义（P<0.01）；L1、L2、L3组全因死亡、跌倒累积发生率的生存曲线比较，差异有统计学意义（P≤0.01）。多因素Cox回归分析结果显示，在调整混杂因素后，与T3组、L3组组合患者相比，T1组、L1组组合与T1组、L2组组合患者全因死亡发生风险增加〔HR=3.29，95%CI（1.13，9.55）；HR=3.09，95%CI（1.08，8.85）；P均<0.05〕，T1组、L1组组合患者跌倒发生风险增加〔HR=1.92，95%CI（1.13，4.27），P<0.05〕。通过Logistic回归分析计算步速、握力联合诊断模型预测概率，并绘制ROC曲线，步速联合握力预测住院老年患者全因死亡、跌倒的ROC曲线下面积（AUC）分别为0.756〔95%CI（0.710，0.801）〕、0.700〔95%CI（0.659，0.741）〕。

在住院老年患者中，步速和握力联合较步速和握力单独能够更好地预测全因死亡及跌倒发生风险，有助于优化老年人的健康管理。