Adults with SMI are twice as likely as adults without SMI to die with cancer before their 75th birthday. (1). It has been suggested that this inequality is contributed to by reduced uptake of cancer screening services, delayed cancer diagnosis, treatment choices and adherence to treatment plans. (2–7)
Clinical Practice Research Datalink (CPRD) Aurum data will be used to explore associations between cancer screening, SMI diagnoses and a range of person, place and behaviour related variables in England. It will also be used test the null hypotheses that people in England with different SMIs are as likely as people without SMI to participate in cancer screening services, once other non-SMI factors are adjusted for. The analysis will be conducted on a cross section of data from September 2020.
The hypothesis tests will be completed using three multiple logistic regression models. The outcome variables will be participation in bowel, breast and cervical screening. The primary exposure variable will be a previous diagnosis of SMI (split into three categories of SMI, and a reference category of no SMI). Models will be constructed separately for bowel, breast and cervical screening. Three categories of confounding variables will be added to each model: person based (age, gender and ethnicity), place based (deprivation and region) and healthy behaviour based (smoking status and body mass index (BMI)). Alcohol and drug consumption may also be considered. Null hypotheses will be rejected where the confidence intervals around the SMI diagnosis coefficients do not overlap with 1.
This research may contribute to improved cancer screening services and health outcomes for adults with SMI. If the null hypotheses are rejected, it offers adequate evidence that merely targeting non-SMI characteristics is unlikely to fully address the SMI based inequalities. It would provide an evidence-based justification for SMI targeting of the screening services.
The study will include both descriptive and analytical components, and each will measure slightly different outcomes.
• The proportion (with confidence intervals) of adults with and without SMI (separately) who had participated in bowel, breast and cervical screening (separately) in the period up to September 2020, split by:
o Index of multiple deprivation (IMD) quintile
o Patient ethnicity
o Region of England of the GP practice where the patient was registered
o Smoking status (current smoker or not)
o BMI status (greater or less than 30)
o Age group (for cervical screening)
o Gender (for bowel screening)
o Alcohol and drug consumption may also be considered
• The proportion (with confidence intervals) of adults who had participated in bowel, breast and cervical screening (separately) in the period up to September 2020 split by category of SMI (bipolar disorder, psychosis, schizophrenia or no SMI)
• The proportion of females aged between 60 and 64 who had participated in 0, 1, 2 or all three of the screening programmes
• The number of females age between 50 and 64 who had participated in 0, 1 or both of breast and cervical screening
• The number of females aged between 60 and 70 who had participated in 0, 1 or both of bowel and breast screening
Multiple logistic regression models will be used to measure:
• Participation in bowel screening
• Participation in breast screening
• Participation in cervical screening
The outcomes will be summarised and presented by their odds. Odds ratios and their confidence intervals will be estimated for participation given a diagnosis of bipolar disorder, psychosis and schizophrenia respectively against a reference category of no SMI.
Odds ratios and confidence intervals will also be estimated for each of the additional confounding variables.
Amendments - 10/11/21
Review Comments: Please clarify your definition of psychosis (for example, will drug-induced psychosis be included?).
Research team response: Non-affective psychosis diagnostic codes will be included. Therefore we excluded conditions such as organic psychoses or drug- induced psychoses.
Optionally: consider truncating at March 2020, instead of September 2020, to avoid confounding by the covid epidemic.
Research team response: Thank you for this comment. The analysis includes 6 months period in addition to the standard recall period for each cancer screening programme which will allow for some delay due to pandemic restrictions. We expect that the impact of pandemic on access to cancer screening will apply to the SMI and non-SMI. We will compare our cancer screening participation rates with what would be expected and caveat our results accordingly.
Alex Jones - Chief Investigator - Office for Health Improvement and Disparities
Gabriele Price - Corresponding Applicant - Office for Health Improvement and Disparities
Cam Lugton - Collaborator - Office for Health Improvement and Disparities
Danny Yip - Collaborator - Office for Health Improvement and Disparities
Jianhe Peng - Collaborator - Office for Health Improvement and Disparities
Robert Kerrison - Collaborator - University of Surrey
Patient Level Index of Multiple Deprivation;Practice Level Index of Multiple Deprivation