Quantifying long-term morbidity following cervical cancer treatment: A retrospective comparison of the Royal College of Radiologist’s (RCR) Morbidity Audit and routine population-based cancer datasets


Year:

Session type:

Theme:

Katie Spencer1,Amy Downing2,Jane Maher3,Louise Whitehouse2,James Thomas4,Eva Morris2
1University of Leeds/Leeds Teaching Hospitals NHS Trust,2University of Leeds,3Macmillan Cancer Support/Mount Vernon Cancer Centre,4NHS England

Abstract

Background

As the number of people surviving cancer increases, understanding the long-term morbidity resulting from cancer treatment becomes increasingly important.  Currently, there is little robust evidence quantifying such cancer-related morbidity but by linking and exploiting population-based cancer datasets it may be possible, for the first time, to generate this information at a population-level.  This study aimed to investigate the feasibility of this approach by quantifying treatment-related morbidity in a population of cervical cancer patients.

Method

RCR audit data for cervical cancer treatment-related morbidity were linked, at patient level, to English cancer registration and Hospital Episode Statistics (HES) data. Patterns in diagnostic reasons for hospital admissions and proportion of survival time in hospital for people with different levels of morbidity (graded using the Franco-Italian-Glossary) recorded in the audit were investigated. Where strong patterns were identified, analyses were undertaken to estimate the proportion of the English cervical cancer population affected by high-levels of treatment-related morbidity.

Results

A total of 682 people in the audit could be identified in the English cervical cancer population of 26,879 people diagnosed between 1998 and 2008. Significant associations between high-levels of morbidity recorded in the audit and patterns of diagnoses and hospital attendance were observed.  For example, individuals with high levels of bowel morbidity have a significantly higher proportion of hospital admission time (37%) related to morbidity compared to those with no/low morbidity (10%/14%). A threshold of >30% of inpatient stays relating to bowel morbidity had a 56.0% sensitivity and 9.4% specificity for identifying high-morbidity patients. Patients with FIGO stage I/II disease, treated with radiotherapy were more likely to have high-bowel-morbidity scores than those treated with surgery alone (11.89% versus 2.26% p <0.001).

Conclusion

This study suggests it is possible to identify treatment-related morbidity using routine population-based data but further work, involving larger populations, is required to fully validate the method.