Objective   To explore prognostic risk factors and develop a prognostic index (PI) for non-smoking, non-drinking females (NSNDF) with oral cancer in Fujian province, and to provide evidences for prognosis prediction and individualized treatment of NSNDF with oral cancer.

  Methods  We conducted a follow-up study among 657 NSNDF with pathologically confirmed primary oral cancer firstly diagnosed in the First Affiliated Hospital of Fujian Medical University during a period from January 2008 through August 2018. Face-to-face interview with International Cancer of the Head and Neck, Genetics and Environment (InterCHANGE) and follow-ups of every 6-month till the end of 2018 were carried out among the NFNDF. Multivariate Cox proportional hazard regression model was used to analyze prognostic risk factors and the PI model was constructed based on the β value of the established multivariate Cox proportional risk regression model.

  Results  Among all the NSNDF, the overall 1-, 3-, and 5-year cumulative survival rate were 91.81%, 80.02% and 69.34% during the follow-up period of 31 584.60 person-months. Multivariate Cox proportional hazard regression analysis revealed that aged ≥ 60 years, with the IV stage of tumor-node-metastasis (TNM) staging, without surgical treatment, with a higher neutrophil-lymphocyte ratio (NLR), and with a higher platelet-lymphocyte ratio (PLR) were prognostic risk factors for the NSNDF. The 5-year survival rates were 84.66%, 65.18% and 24.34%, respectively, for the NFNDF with low, medium and high PI score. Compared with those with low PI scores, the mortality risk increased for the NFNDF with medium PI scores (hazard risk HR = 2.264, 95% confidence interval 95% CI: 1.507 – 3.402) and high PI scores (HR = 6.564, 95% CI: 4.013 – 10.737). A significant inverse correlation between PI score and mortality risk was observed (χ²_trend = 55.06, P_trend < 0.001). The established PI model is of good performance in prognosis prediction, with the concordance index of 0.677, the area under the receiver operating characteristic curve of 0.624, and the Akaike information criterion of 1 566.398, respectively.

  Conclusion  Age, TNM stage, surgical treatment, NLR and PLR are independent influencing factors for the prognosis of NSNDF with oral cancer in Fujian province. The established PI could be used well to predict the prognosis for the NSNDF with oral cancer.