治疗初期听力提升水平对非低频下降型突发性聋的预后预测价值

Predictive Value of Early Hearing Improvement for Prognosis in Non-Low-Frequency-Descending Sudden Sensorineural Hearing Loss

摘要:

目的 探讨平坦下降型、全聋型及高频下降型突发性聋患者治疗初期听力提升水平与预后的相关性，并评价初始听力与初期听力提升水平联合预测模型对预后的预测价值。方法 选取2017年1月-2025年6月于我科住院的162例单耳突发性耳聋患者进行回顾性分析，按最终随访（出院后≥1月末次纯音测听）患耳听力是否痊愈为标准，将患者分为95例完全恢复组和67例未完全恢复组，使用Mann-Whitney U检验和X2检验对两组间的年龄、发病至就医时间、性别、患耳侧别、眩晕、糖尿病、高血压、治疗初期听力提升水平、患耳受损频率初始听阈资料进行分析，使用logistic回归模型确定听力完全恢复的独立影响因素，通过ROC曲线和约登指数确定连续变量最佳截断值，并以此截断值为分组依据行Fisher确切概率检验。将初始听阈与初期提升水平同时纳入logistic回归建立联合预测模型，以Bootstrap内部验证、净重分类改善指数（NRI）、综合判别改善指数（IDI）及决策曲线分析（DCA）评价联合模型的预测增益。结果 162例单耳突发性耳聋患者总体痊愈率58.64%。多因素logistic回归分析显示发病时初始听力（OR=0.922，p<0.001）、治疗初期纯音听阈提升水平（OR=1.114，p<0.001）是听力完全恢复的独立影响因素。单变量ROC分析中，发病时初始听阈与治疗初期听阈提升水平的最佳截断值分别为77 dB HL（灵敏度80.0%、特异度73.1%）和15 dB HL（灵敏度85.3%、特异度61.2%）。联合预测模型AUC为0.939（95%CI 0.902–0.975），Bootstrap内部验证乐观校正后为0.934，显著高于初始听阈（0.832）与初期提升水平（0.799）单一指标；联合模型较单一指标的连续NRI、IDI均为正（较初始听阈：NRI=1.175、IDI=0.268；较初期提升水平：NRI=1.191、IDI=0.341）；决策曲线分析显示联合模型在多数阈值概率下净获益优于“全部治疗”策略。发病时初始听阈≥77 dB HL、治疗初期纯音平均听阈提升<15 dB HL者预后更差（均P<0.001）。结论 发病时初始听力、治疗初期听力提升水平是非低频下降型突发性聋预后的独立预测因素，两者联合可实现对预后的早期预测。本研究结论仅适用于平坦下降型、全聋型及高频下降型突发性聋，不宜推广至低频下降型及全部突发性聋患者。

Abstract:

Abstract: Objective To investigate the correlation between early hearing improvement and prognosis in patients with flat-type, total-deafness-type, and high-frequency-descending-type sudden sensorineural hearing loss (SSNHL), and to evaluate the predictive value of a combined prediction model incorporating initial hearing threshold and early hearing improvement for prognosis.Methods A retrospective analysis was conducted on 162 patients with unilateral SSNHL hospitalized in our department from January 2017 to June 2025. Based on whether the affected ear achieved complete recovery at the final follow-up (last pure-tone audiometry at ≥1 month after discharge), patients were divided into a complete recovery group (n=95) and an incomplete recovery group (n=67). The Mann–Whitney U test and chi-square test were used to compare age, time from onset to treatment, sex, affected ear side, vertigo, diabetes mellitus, hypertension, early hearing improvement, and initial hearing threshold at impaired frequencies between the two groups. Logistic regression was used to identify independent factors for complete hearing recovery. The optimal cutoff values of continuous variables were determined by receiver operating characteristic (ROC) curves and the Youden index, and the patients were regrouped accordingly for Fisher’s exact test. A combined prediction model was established by entering initial hearing threshold and early improvement simultaneously into logistic regression, and its incremental predictive value was evaluated by bootstrap internal validation, net reclassification improvement (NRI), integrated discrimination improvement (IDI), and decision curve analysis (DCA).Results The overall complete recovery rate of the 162 patients with unilateral SSNHL was 58.64%. Multivariable logistic regression showed that initial hearing at onset (OR=0.922, P<0.001) and early pure-tone threshold improvement during treatment (OR=1.114, P<0.001) were independent factors for complete hearing recovery. In univariate ROC analysis, the optimal cutoffs of initial hearing threshold at onset and early threshold improvement were 77 dB HL (sensitivity 80.0%, specificity 73.1%) and 15 dB HL (sensitivity 85.3%, specificity 61.2%), respectively. The combined model achieved an AUC of 0.939 (95% CI 0.902–0.975), with an optimism-corrected AUC of 0.934 after bootstrap internal validation, significantly higher than that of initial hearing threshold alone (0.832) and early improvement alone (0.799). Compared with the single indicators, the combined model yielded positive continuous NRI and IDI values (vs. initial hearing threshold: NRI=1.175, IDI=0.268; vs. early improvement: NRI=1.191, IDI=0.341). Decision curve analysis showed that the combined model provided a greater net benefit than the “treat-all” strategy across most threshold probabilities. Patients with an initial hearing threshold ≥77 dB HL at onset or an early pure-tone average improvement <15 dB HL had a poorer prognosis (both P<0.001). Conclusion Initial hearing at onset and early hearing improvement are independent predictors of prognosis in non-low-frequency-descending SSNHL, and their combination enables early prediction of prognosis. The conclusions of this study apply only to the flat-type, total-deafness-type, and high-frequency-descending-type SSNHL and should not be generalized to the low-frequency-descending type or to all SSNHL patients.