AbstractBackground:Lower treatment rates in older patients with cancer exist but may reflect contraindications of advanced disease, comorbidities and frailty. This thesis aimed to firstly review the evidence for under-treatment (UT), secondly to assess the potential for UT using patient-level cancer registry data and secondary population-based statistics, finally to compile a dataset and develop a method to quantify UT in the older breast cancer (BC), lung cancer (LC) and colorectal cancer (CRC) patients.
Methods: The potential for UT was assessed using EUROCARE-5 – a large population-based survival estimates database, and data from Northern Ireland Cancer Registry (NICR). To quantify UT, information for patients diagnosed with BC, LC and CRC between 2011-2015 was collated from hospital episodes, multidisciplinary team meetings, the NICR and self-reported health status (SRHS) from the 2011 NI Census data. Age was categorised into younger (<75 year of age) and older (≥75), and primary treatment defined as surgery, chemotherapy, radiotherapy or no treatment. A causal inference mediation approach, g-computation, was used to estimate 1) the total causal effect (TCE) of age on 1-year observed survival, which was partitioned into 2) the natural indirect effect (NIE) mediated through treatment and 3) the natural direct effect (NDE) not mediated through treatment but through stage, comorbidities, and performance status.
Findings: Potential for UT was indicated by, (a) lower conditional relative survival (conditional on surviving one year) in the older, compared to the younger, BC and LC patients (p<0.05), (b) a drop in stage-specific surgery rates in the older (-3%) compared to the increase (+2.5%) in the younger, and lower increase in 2-year net survival in the older (5.9%) compared to the 7.5% increase in the younger CRC patients, over two time periods; 2000-2007 and 2008-2015, (c) lower treatment rates and lower survival in the older BC, LC and CRC patients in a systematic review conducted as part of this thesis. In the mediation analyses quantifying UT, (a) compared to the younger patients, older patients were less likely to receive surgery, radiotherapy, chemotherapy, but were more likely to receive hormone therapy (for BC only), (b) the TCE of older age on survival for LC, CRC and BC patients was (-14.1% 95% CI, -17.5% — -10.7%), (-23.6% 95% CI -28.7% — -18.4%) and (-10.1% 95% CI -14.0% — -6.1%), respectively. UT (i.e., NIE) was quantified for LC, CRC and BC as (-4.5%; 95% CI -6.8% —-2.2%), (-9.6%; 95% CI -13.4% — -5.8%) and (-0.7%; 95% CI -5.1% — 3.6), respectively. Therefore, of every 100 older patients, an extra 4 to 5 LC and 10 CRC patients would be alive at one year had they received the clinically specific treatment rates of the younger patients. There was no substantial evidence to suggest UT in the older BC patients.
Interpretation: About 32% (for LC) and 40.7% (for CRC) of the reduced survival in older patients were attributed to UT. This thesis highlights the potential of offering more clinically-specific treatments to older LC and CRC patients and the need for comprehensive geriatric assessment in these patients with complex health needs.
|Date of Award||Jul 2020|
|Sponsors||Northern Ireland Department for the Economy|
|Supervisor||Finian Bannon (Supervisor), Frank Kee (Supervisor) & Gerard Hanna (Supervisor)|
- breast cancer
- lung cancer
- colorectal cancer
- hormone therapy
- conditional relative survival
- net survival
- observed survival
- self-reported health status
- g-computation formula