| Summary: | This thesis investigates prognostic factors influencing outcome and survival in patients managed by the South East Wales Upper GI cancer network. The hypotheses tested were: Socio-economic deprivation and health deprivation adversely influence outcome in patients undergoing surgery for oesophageal and gastric cancer; Patient delay accounts for the majority of the total delay encountered in the diagnosis and treatment of oesophago-gastric cancer and deprivation is an important factor in this regard; Body composition and sarcopenia as measured by Bioelectrical Impedance Analysis (BIA) are important prognostic indicators; Centralisation of oesophago-gastric cancer services significantly improves outcome and survival. In a consecutive cohort of 1185 patients survival was associated with multiple deprivation (P < 0.0001) and health deprivation (P < 0.0001). Total delay consisted of the following components: patient delay (76%); practitioner delay (1%) and hospital delay (23%). Factors influencing patient delay were deprivation (p=0.005) and gender (p=0.030). Survival was significantly related to overall delay (p=0.010). In 125 patients who underwent BIA testing open and close laparotomy was significantly associated with FFM% (p=0.027), and BF% (p=0.030). Post-operative morbidity (Clavien-Dindo ≥ 3) was associated with intracellular fluid volume (ICV) (p=0.018), total body water content (p=0.019), and sarcopenia (p=0.045). Critical care length of stay was associated with ICV (p=0.009), lean muscle mass (p=0.006), the phase angle (p=0.025) and sarcopenia (p=0.011). Treatment with curative intent increased from 21.6% to 29.6% of patients before and after centralisation respectively (p=0.002). Serious post-operative morbidity (Clavien-Dindo ≥ 3) decreased after centralisation (p=0.194), and there were significant reductions in critical care length of stay (p < 0.0001), with overall length of hospital stay reduced by 2.5 days (p=0.008). On univariate analysis of factors influencing two-year survival centralisation was statistically significant (p=0.001).
|
|---|
