Patient demographics and database
Newly diagnosed oral cancer patients treated with resection of the primary oral tumor and neck dissection with or without adjuvant therapy were identified from our hospital Cancer Registry Database within the years 2009 to 2018. Patients without complete treatment, electric medical records, regular follow up, and distant metastasis on diagnosis were excluded in our study. The type of neck dissection included elective neck dissection (END) for clinical N0 disease and radical neck dissection (RND) for clinical N1-3 disease. Patients with bilateral neck dissection and those with retrieved LN less than 10 were also excluded. The variables collected from the Cancer Registry Database included patient demographic data, such as age, gender, tumor status such as clinico-pathological TNM stage, and pathological risk features (such as margin status,  tumor differentiation, perineural invasion, lympho-vascular permeation, number of retrieved LNs, number of positive LNs, extranodal extension, adjuvant treatment modality, radiation dose, and chemotherapy regimen). All staging was done according to the American Joint Committee on Cancer (AJCC) cancer staging (7th edition). As shown in figure 1, the study periods were categorized into two stages: period 1 (before 2015 Oct), and period 2 (after 2015 Oct). In the period 1, the operative procedure of primary tumor resection and neck dissection was performed as per the guidelines or literatures in target group and control group.3,14Since period 2, soft but non-mandatory intervention, lymph node promotion activiy (LNPA) for cervical lymph node improvement was implemented in target group alone. The LNPA used in the target group including: (1) promotion of the importance of survival rates and LN yield in head and neck cancer in weekly meetings, and resident journal club conferences; (2) surgeons were encouraged to identify important landmarks, like cervical ansa, internal jugular vein, carotid artery, spinal accessory nerve, splenius capitus, levator scapula, anterior scalene muscle, and transverse cervical artery during neck dissection ; (3) the average number of LN yield was announced per month, but individual data related to the patients or surgeons were not disclosed (Figure 1).
The hypothesis was that target group surgeons’ behavior response to the LNPA could improve LN yield in the period 2. The effect of LNPA on lymph node yield and regional recurrence was evaluated by difference-in-differences analysis. For regional recurrence, we used a concept of regional recurrence density which meant the proportion of regional recurrence among those treated with neck dissection during a specific period. This method could provide us a comparison of regional recurrence between the target group and control groups in different periods.