From the results of our analyses, Mini-Cog© is not independently associated with POD or length of stay in patients undergoing elective major colorectal surgery. Whilst the Mini-Cog© is associated with POD on univariate analysis, this association is lost when other variables which are also plausibly involved in the mediation of POD are added into the logistic regression model. This is important as it highlights the likely multifactorial nature of how POD arises and the necessity to provide a comprehensive, holistic preassessment of the high-risk patient.
The Mini-Cog© was also associated with 90-day mortality in univariate analysis, although we were unable to explore this association with multiple regression due to low mortality rates. Our results are both in keeping with, and build upon, previous literature by highlighting the difficulty in predicting in-hospital delirium (Lindroth et al., 2018) and demonstrating that the link between preoperative cognition and POD may be mediated through other physiological variables (Culley et al., 2017; Dworkin et al., 2016; Robinson et al., 2012). Analysis showed that at baseline, patients with poor Mini-Cog© scores were on average 10 years older (p = 0.001) and had a higher ASA grade (p = 0.012), which may reflect a link between multimorbidity and cognitive function which is controlled for in multivariate analysis.
In our sample, 16% of patients were found to have cognitive impairment as assessed by a Mini-Cog© score < 3, and 6.3% of patients were confused postoperatively. This is at the low end for incidence of POD found in systematic reviews, which ranges from 5 to 52% (Dasgupta & Dumbrell, 2006; Raats et al., 2016; Yang et al., 2016). However, most of these studies have involved orthopaedic (mostly hip-fracture) and major vascular surgery patients. Due to the emergent and traumatic nature of clinical presentation, they would likely have higher levels of haemodynamic and general physiological instability, with poorer perioperative optimisation than the elective patients in our sample, which may explain the higher rates of POD.
We chose a Mini-Cog© cut-off of < 3, which is validated for dementia screening. For broader preoperative confusion assessment that does not reflect cognitive impairment reaching the requirements for dementia, a more generous cut-off of < 4 has been used, with relationships still found between Mini-Cog© score and POD, length of stay and even mortality (Robinson et al., 2009a; Robinson et al., 2012). Dworkin et al. showed that raising the threshold from a score < 3 to a score < 5 only slightly increased NPV from 90 to 94%, indicating minimal benefit to a higher Mini-Cog© score in the diagnosis of preoperative cognitive stability (Dworkin et al., 2016). Indeed, our NPV was 95%, further justifying our choice of 3 as the cut-off value. Our more stringent cut-off is appropriate for investigating the further role of Mini-Cog© in identifying the risk of subsequent POD.
Only two of our patients had a formal dementia diagnosis. This likely reflects underdiagnosis in our surgical population and is in keeping with a previous study that also identified no dementia diagnoses in their elderly patients (Dworkin et al., 2016). Screening for dementia via Mini-Cog©, and using the cut-off of < 3 to do so, may still be an important tool for realising long term benefits in terms of patient cognition and surgical outcomes. Possible dementia identified at a comprehensive preoperative assessment could be passed on to family doctors or specialist memory clinics either preoperatively or postoperatively, to identify those with true dementia and ensure long-term optimisation of patients.
Mini-Cog© score was associated with 90-day mortality in the unadjusted analysis (OR 4.65, 95%CI 1.36 to 15.9, p = 0.02). This finding is in keeping with previous evidence identifying a role for Mini-Cog© in postoperative mortality in high-risk surgical patients (Robinson et al., 2009a; Robinson et al., 2012). This could be due to poor cognition reflecting a lack in other physiological reserve, or maybe due to confused patients being more at risk of other morbidities such as falls. It is possible that poor cognition could affect patients’ ability to engage in postoperative rehabilitation techniques which act to minimise risk of developing chest infections and ileus, whilst ensuring safe mobilisation. It is unsurprising that the other physiological variables (frailty, ASA grade and age) were also associated with this, but it is not clear if Mini-Cog© score acts independently in this regard as our data did not allow us to perform a multivariate logistic regression.
Delirium is thought to be preventable in up to 40% of patients (American Geriatrics Society Expert Panel, 2015). Identification of at-risk patients would allow clinical teams to institute measures to ameliorate the effect of anaesthesia and surgery on cognition. Screening with Mini-Cog© is very quick and cheap, and whilst we have demonstrated that it does not act independently of other variables, its use could increase the awareness of clinical teams to the likelihood of patients developing POD. Increased awareness could then lead to early diagnosis via targeted screening for cognitive change using a validated screening tool such as the Confusion Assessment Method (CAM). Intraoperative measures to reduce the incidence of POD may include the use of processed electroencephalographic (e.g. bispectral index) monitoring by anaesthetists, as a lighter depth of anaesthesia has been shown to reduce rates of POD. Pharmacological management could be targeted, for example by preferentially using regional blocks in high-risk patients to minimise the use of opioid painkillers, polypharmacy and pain-induced delirium. Non-pharmacological approaches such as mobility enhancement, cognitive orientation, sleep enhancement measures and targeted nutrition and fluid therapy have also been demonstrated to be effective (American Geriatrics Society Expert Panel, 2015).