Between-hospital variations of patient-related risk factors and treatment choices in colorectal cancer. A Danish population-based study

There have been consistent reports of between-hospital variations in treatment choices for colorectal cancer (CRC) over the last five years in Denmark. The variations are surprising as national treatment recommendations have been present for the last twenty years.

The thesis is based on three studies that elucidate patterns of patient-related risk factors and essential treatment choices in CRC, with particular attention to variation between hospitals. The treatment choices were the overall aim of treatment, permanent stoma or anastomosis, and adjuvant chemotherapy in CRC. All three studies are register-based cohort studies on the same cohort of almost all adult CRC patients diagnosed with first-time CRC from 2009-2018 in Denmark. Data from several Danish nation-wide registers were combined.

Study I examined variations in patient-related risk factors of short-term postoperative outcomes between hospitals treating CRC in Denmark. The study found noticeable variations in patients aged 75 years or older (31-46%), stage I (12- 21%), stage IV (23-35%), and American Association of Anesthesiologists (ASA) score ≥III (18-40%). Accumulated patient-related risk factors were calculated based on logistic regression analyses. The predicted risk of postoperative complications and 90-day postoperative mortality varied from 17-23% and 3.2-5.5% between hospitals. Overall, the study showed marked variation in the case-mix of CRC patients between hospitals. Differences in case-mix did not explain variations in short-term postoperative complications between hospitals.

Study II elucidated predictors in refraining from curatively intended surgery and adjuvant chemotherapy in potentially curable CRC, including between-hospital variations. Predictors for the non-curative treatment aim were old age, ASA ≥III, World Health Organization performance status (PS) ≥1, distant metastases, and underweight. The study also examined predictors for no adjuvant chemotherapy treatment in stage II or stage III CRC. The predictors for this outcome were old age, ASA ≥III, PS ≥2, kidney disease, postoperative complications, and living alone. The study found marked variations in the overall treatment aim and refraining from adjuvant chemotherapy between hospitals. In conclusion, the betweenhospital variations were not explained by case-mix or other included possible confounders.

Study III examined whether socioeconomic factors or the treating hospital could predict permanent stoma when adjusting for case-mix and various other possible confounders. Multivariable analysis could not be done in the right colon due to less than one percent having a permanent stoma. For the upper two-thirds of the rectum, the lowest quartile of household income was an independent predictor. We found marked variations between hospitals in the sigmoid colon and upper two-thirds of the rectum.

The present project contributes to mapping patterns of between-hospital variations in treatment choices in colorectal cancer in Denmark. The studies found an unequal distribution of patient-related risk factors between hospitals treating CRC in Denmark. High age and poor performance status were the main predictors of the examined treatment choices, while socioeconomic factors partially influenced the choices.

Further studies are warranted on the unexplained variations in treatment choices between hospitals and the consequences of these variations in mortality, morbidity and quality of life.