Identify factors which have an impact on the length of postoperative hospital stay in patients undergoing major surgical resection for head and neck cancer.Design:
Retrospective study using hospital case notes.Setting:
One hundred and eighty-four patients who underwent 191 major head and neck resections between 2006 and 2010 were identified from the head and neck oncology database. An additional cohort of 52 patients from a second centre was used to externally validate the model.Methods:
Data collected on 13 variables including age, preoperative haemoglobin level, body mass index, timing of surgery (primary or recurrent tumour), category of surgical procedure (clean, clean-contaminated, contaminated, dirty), tracheostomy (yes or no) and duration of the operation were analysed using a multiple linear regression.Results:
In the 13 parameter model, the four variables that were found to significantly (P < 0.05) prolong length of stay (LOS) were surgical wound type (clean-contaminated, P = 2.19 × 10−6versus clean), tracheostomy (P = 0.0034), operation time (P = 0.011) and American Society of Anaesthesiologists (ASA) grades 3 and 4 (P = 0.0067) versus 1 and 2. The statistically significant variables were used to generate a preoperative submodel without the operative time and a postoperative submodel with all four variables. Overall, the best model based on adjusted R2 was the postoperative model. When fitted on the external data set, there was no significant difference in the residuals, indicating that the models generalise across centres.Conclusion:
Clean-contaminated wound was the most significant factor affecting the LOS, with others being ASA grades 3 and 4, longer duration of surgery and presence of a tracheostomy. Using these parameters, it is possible to predict the LOS in patients undergoing major surgical resection for head and neck cancer.