Assessment of the potential risks of perioperative morbidity and mortality is important in the process of deciding on surgical treatment strategy. Traditionally this has been done by the surgeon and the anaesthesiologist, primarily based on clinical experience. Although the final decision on strategy always will have to be based on this subjective clinical evaluation, more objective models for prediction of risk could come useful in this process. Furthermore, objective preoperative information on potential risks of the treatment is essential as a part of the process of informed consent on the selected procedure.

In the context of performance evaluation of different hospital units, a comparison of crude in-hospital or 30-day mortality can be misleading due to case mix. In order to compensate for the variation in physiological condition of the patient and the severity of surgery, different risk scoring models were developed. Subsequently these multivariable regression models have proved useful in risk prediction. The number of included variables differs between the scoring systems, as does the availability of these variables and thus the possibility for effective utilisation in the clinical setting. Furthermore, the assessment of the models occurs at different points throughout the course of hospitalisation, typically in the pre-, peri- and postoperative phase. The first models introduced were devised to predict the outcome among patients undergoing surgery in general but gradually more diagnosis-related and diagnosis-specific models have been developed.

Initially meant as a tool for anaesthesiologists to improve communication and compare results of anaesthesia, the ASA classification of physical status was introduced already in 1941 [10]. Over the years this classification proved to correlate well with overall surgical mortality and developed to become an estimate of operative risk. Although vague and subjective, with a wide interobserver variability, this classification still is an important predictor of outcome after surgery in many hospitals and thus as a tool for preoperative selection among CRC patients.

At present the most widely accepted risk prediction score in gastrointestinal surgery is the POSSUM score, a Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity, described by Copeland in 1991 [11]. This is a dual scoring system including assessment of both mortality and morbidity and was designed to be used in general surgery, both in the elective and emergency setting. The system consists of a 12-factor, four-grade, physiological score, in combination with a six-factor, four-grade, surgical operative severity score. Further evaluation of the POSSUM score revealed a tendency to over prediction of mortality, especially among low-risk patients, and a modification of the score, the Portsmouth POSSUM, p-POSSUM, was presented in 1998 [12]. The two systems use the same physiological and operative severity scores but different regression equations. The model for predicting morbidity is identical in the two systems. Due to limitations of applying POSSUM scoring in the oldest patients undergoing colorectal surgery, a speciality-specific model based on the POSSUM methodology have been developed, the colorectal POSSUM (CR-POSSUM) [13, 14]. In this system for prediction of mortality, the number of variables included both in the physiological and operative severity score have been reduced [15].

External validation of this model in the UK has proved it to be more accurate than the previous POSSUM models in risk prediction among CRC patients [16]. Evaluation of the applicability of all three POSSUM scores in the US has concluded with overprediction of mortality for colon cancer resections. This indicates the need for calibration when the score is applied on other health-care systems, outside the UK [17].

In general, as the different models are based on cohorts of patients from different hospitals and even from different health-care systems, an evaluation of the predictive value of the model is necessary regarding applicability in a specific hospital. Furthermore, as the quality of treatment improves and the rates of complications and mortality decrease, there has to be a continuous updating of these models in order to maintain an adequate prediction [18].

CRC is a disease mainly affecting the elderly. In the Western world, approximately 60 % of the patients are more than 70 years at diagnosis, and, as a result of an increasing life expectancy, the incidence of CRC in older patients will continue to rise [19].

Thus, the management of elderly CRC patients represents a considerable challenge for the health-care system in the future.

The literature, primarily based on selected cohorts of elderly patients, most often supports that surgical treatment is feasible irrespective of chronological age. In large national observational cohort series, age appears to influence on the rate of resection, both in terms of overall and curative resection rate, and on the choice of surgical procedure [20, 21].

In stage III colon cancer, adjuvant chemotherapy has been demonstrated to reduce the risk of disease recurrence and to improve survival, but the likelihood to receive chemotherapy decreases with age [22]. The same trend is seen in the use of palliative chemotherapy among older CRC patients. Furthermore, radiotherapy seems to be underutilised in the treatment of rectal cancer in the elderly [23]. To avoid a potential substandard treatment of the elderly CRC patient, knowledge regarding results of treatment in the elderly population and individual evaluation of patients in this heterogenic group is imperative.

Several series evaluating the results of elective curative major surgery in elderly CRC patients present rates of local recurrence, metastases and cancer-specific and relative survival comparable to those in younger patient cohorts [20, 22–24]. The rate of postoperative mortality and morbidity increases with age in most of these series, with a mortality of 8 % among the oldest rectal cancer patients undergoing curative major surgery [20]. The rates of postoperative morbidity range between 30 and 60 % in different series [24–26]. This illustrates the challenges of selection in older patients planned for elective CRC surgery.

During the last years, laparoscopic colorectal resection for cancer has emerged as a minimally invasive alternative. Series comparing elderly CRC patients undergoing open or laparoscopic resection have revealed a significant decrease in postoperative morbidity among those undergoing laparoscopic resection [22, 26, 27].

Furthermore, hepatic and pulmonary resections have become safer and 5-year survival rates of up to more than 50 % in selected patients have been reported [28, 29]. Major liver resections can be safely performed in elderly patients, with similar short- and long-term outcome as in their younger counterparts [28]. Although age is a prognostic factor in multivariable analysis on survival after pulmonary resection of CRC metastases, several of the presented series include elderly patients, indicating that high chronological age is not a contraindication for resection [29]. Hepatic and pulmonary resections of colorectal metastasis are feasible in selected older patients.

The rate of patients undergoing emergency surgery for CRC increases with age [21–23, 30]. Furthermore, the rate of curative resections decreases in the emergency setting and the rate of postoperative mortality increases significantly, reaching 35 % in patients aged over 80 years [30]. Colonic stenting has proved to be a safe and effective procedure and should be considered essential as a bridge to surgery or a palliative procedure in emergency treatment of CRC in elderly patients [22, 31].

Elderly rectal cancer patients treated with major rectal resection more often undergo procedures resulting in a permanent stoma than younger patients [20]. This is probably due to an assumption of increased rate of anastomotic leak and poor functional results after an anterior resection. In the literature the rate of anastomotic leak does not seem to increase with age, and although some investigators have found that increasing age has an adverse effect on postoperative functional outcome, most series find no differences between the age groups.

Moreover, the majority of elderly patients are satisfied with their functional outcome after surgery with a primary anastomosis [23, 32]. Thus, age per se does not seem to be a contraindication for primary anastomosis in major rectal resection.

The use of oncological treatment in terms of adjuvant chemotherapy in colon cancer, neoadjuvant chemoradiation in rectal cancer and in palliative treatment of metastatic disease decreases with age [20–22]. Elderly patients are usually underrepresented in clinical trials evaluating oncological treatment, and thus the information regarding the benefits and tolerability of the different treatment options among older CRC patients is limited.

As adjuvant chemotherapy in stage III colonic cancer among older patients, 5-FU-based regimens improve survival to the same extent as in younger patients. Infusional 5-FU seems to have a favourable toxicity profile compared to bolus 5-FU, and infusional 5-FU in combination with leucovorin is recommended as the treatment of choice [22]. Oral capecitabine, although proved to be as effective as 5-FU, has an increased toxicity related to renal function in older patients, while there is limited information regarding the use of the combination of oxaliplatin, 5-FU and LV (FOLFOX) in adjuvant treatment among older patients.