Prognostic model of the absence of larynx motility recovery in unilateral lesions of the recurrent laryngeal nerve in thyroid surgery

Abstract

One of the most serious complications in thyroid surgery is damage to the recurrent laryngeal nerve, which can significantly worsen the patient’s quality of life, so predicting the course of the disease is very important, as it contributes to timely and targeted aid to patients and the determination of optimal treatment tactics.
 Objective — construction of the prognostic model of the absence of larynx motility recovery in unilateral lesions of the recurrent laryngeal nerve in thyroid surgery
 Materials and methods. In the years 2018 — 2021, a prospective single‑center study was conducted at the Ukrainian Scientific and Practical Center of Endocrine Surgery, Transplantation of Endocrine Organs and Tissues of the Ministry of Health of Ukraine. Investigation involved 164 patients after thyroidectomy with unilateral motility disorder of the larynx. The investigated patients are divided into two groups: the main group (n = 33) included patients with unilateral paralysis of larynx (the mean age 49.7 ± 11.1 years), and control group consisted of 131 patients with unilateral paresis of larynx (the mean age 43.2 ± 14.6 years).
 Results. Prognostication of disease was conducted by means of three methodologies. The method of statistical analysis of Wald was conducted taking into account the most meaningful informing laryngoscopy signs with their gradations (9 laryngoscopy signs and 1 sign that is related to age), that were distinguished by us in previous prognostic research. Diagnostic coefficients and information measures of Kullback J of selected features were determined and a special template was developed using the MS Excel program to automate diagnostic analysis. Method of binary logistic regression allows the expectation of dependence of binary initial variable (present/absent paralysis) after influence of factor variables. Three factor signs that remained after recalculation have been investigated. To ensure quick and convenient execution of calculations, a corresponding template has also been created in the MS Excel program. The prognostic express test, which enabled us to build an easy‑to‑use test based on the most significant two predictors and their gradations, which were converted into ranks. A comparison of the occurrence of paresis/paralysis of the larynx was made for different sums of ranks.
 Conclusions. The diagnostic method based on the sequential Wald analysis, despite good performance in the case of a definite result, gave about 25 % indeterminate results and therefore needs further improvement based on a more representative sample. The diagnostic model built using binary logistic regression led to good results, but working with it requires a personal computer and an appropriate template for calculations in the Excel program. The prognostic express test, which is based on the consideration of only two predictors, is convenient, does not require complex calculations and provides an acceptable level of sensitivity and specificity, which can be widely used in the clinical practice of first‑line otorhinolaryngologists for a timely decision on further conservative or surgical treatment.