Individual and contextual deprivation: Association with cancer outcomes and implications for cancer policy

点击翻译按钮获取中文摘要

In the UK, people who reside within more income-deprived areas live a shorter period of time after a diagnosis of cancer compared to people living in less income-deprived areas. At least part of these inequalities in cancer survival are due to inequalities in cancer care, even considering differential patient and tumour factors such as stage at diagnosis. The specific mechanisms by which area-based deprivation levels lead to poorer individual health outcomes within the context of a universal healthcare system, free at the point of use, are not well understood. These analyses will enable, for the first time, the examination of how an individual patient's socio-economic status is associated with poorer cancer survival in England, and will demonstrate how these associations might be modified by the level of deprivation in the small area within which the patient resides.Our aim is to perform an in-depth study of the association between the individual patient's deprivation and cancer survival, considering in particular how this is influenced by their socio-economic context, whether it varies over time since diagnosis and whether it has changed over calendar time. We will focus on three indicators of deprivation: income, education and occupation. We will first examine the correlation between individual and area deprivation, by each of these indicators, and then secondly describe the association between individual deprivation and survival. Third, we will assess whether the association between individual deprivation and patients' survival is modified by area deprivation; that is, whether equally deprived individuals in different areas fare better, or worse, according to the socio-economic context of the area within which they live. Finally we will gain the insights of patients, carers, and healthcare professionals on these data, and communicate these to cancer policy makers.We will use data from the ONS Longitudinal Study, which contains individual deprivation information on 1% of individuals in England present at each national census since 1971. The ONS-LS is linked to Cancer Registry data so that information on all diagnoses of invasive cancer amongst cohort members can be examined. These data will be further linked to the Indices of Multiple Deprivation (IMD) defined at the Lower Super Output Area (LSOA, mean population 1,500) to obtain area-based measures of deprivation. We will fit regression models for the excess mortality hazard (EMH) in order to examine how cancer survival varies with individual deprivation status. The EMH represents the mortality hazard that is observed among cancer patients, taking into account each patient's expected mortality. We will construct life tables stratified by individual deprivation, using the ONS-LS mortality and population data, in order to account as accurately as possible for this background mortality. We will assess whether these associations are modified according to the ecological level of deprivation through the use of interaction terms. We will examine time-dependent associations between individual deprivation and the EMH using flexible functions, and will account for the correlation between observations coming from the same small area through a shared random-effect (defined at LSOA level).The results will allow us to determine the degree of correspondence between individual and area deprivation for cancer patients; to assess how individual deprivation (income, education or occupation) is associated with cancer-specific mortality; and to quantify if and how these associations are modified according to area deprivation, as well as how these associations vary by the time since diagnosis. During the final phases of the project we will seek the insights of patients, carers and healthcare professionals upon our findings. These will then be directly shared with policy makers in a half-day workshop, so that future policies aimed at reducing socio-economic inequalities are more targeted.