Effects of an abnormal mini-mental state examination score on postoperative outcomes in geriatric surgical patients: a meta-analysis

Surgical safety and success rates have improved considerably as a result of remarkable medical breakthroughs, which has led to the extension of life expectancy and the rise in the number of aged patients undergoing surgical procedures [1]. Compared to 2017, the number of persons aged 60 and older is expected to more than double by 2050 and more than triple by 2100, thereby rising from 962 million globally in 2017 to 2.1 billion in 2050 and 3.1 billion in 2100 [2], which will necessitate an increasing demand for operations in aged individuals. In England, less than 1.5 million people over 75 years old underwent surgery between 2006 and 2007, contrasted with 2.5 million between 2014 and 2015. Among these 2.5 million patients, up to 30% were above 85 years of age. Similarly, in Australia, female patients aged 85 years and older have tended to be the largest group of the overall emergency surgical population [3]. In America, a research [4] study reported that in 2015, patients 65 years of age and older accounted for 34.1% of all surgeries.

Comorbidities are more common in older patients, the ability for their bodies to compensate decreases, and some health problems may be underdiagnosed; these factors contribute to an inability to tolerate surgery [5]. Surgery in older patients presents medical workers with formidable challenges as they must weigh long-term benefits and risks carefully to make the best surgical decisions. Previous studies have reported that both preexisting and new-onset cognitive impairment (CI) following surgeries, which has been observed in 19–83% of elderly patients varying with age and type of surgery [6, 7], was associated with an increased incidence of postoperative complications such as delirium. Other long-term problems and poor outcomes have also been of concern, such as mortality, impaired functional status, readmission, prolonged hospitalizations and increased expenses [3, 8‐10]. The Mini-Mental State Examination (MMSE) is long established and the most widely used instrument to help clinicians detect cognitive impairment and grade the severity of cognitive change [11]. This tool is frequently applied in the perioperative period of older patients due to its adequate performance in a rule-out capacity [12]. However, several studies have confirmed that the MMSE score could be a predictor of adverse postoperative outcomes, but those findings are limited by the sample sizes or other confounders. Thus, whether the MMSE could be an indicator of the prognosis of geriatric surgical patients remains unclear. To our knowledge, there have been no reviews systematically exploring and quantifying the association between impaired cognition diagnosed by the MMSE and different postoperative outcomes.

We conducted this systematic review to investigate whether the MMSE used as a perioperative assessment tool has prognostic value on adverse clinical outcomes in aged surgical patients and to search for a better understanding and guidance for clinicians to evaluate and make optimal patient care decisions.

We identified a total of 2492 records. After removal of duplicates, we screened 947 titles and abstracts, of which 156 full text articles were selected for eligibility. Twenty-one studies met the inclusion criteria for the systematic review (Fig. 1).

The designs of the included studies were a randomized controlled trial [19](N = 1), retrospective cohort trials [7, 20](N = 2), and prospective cohort trials [21‐38](N = 18). Two sets of data were extracted from the intervention group and control group of the RCT (Table 1). When defining cognitive impairment or cognitive intact, most of the included studies used the MMSE cutoff score of 24, while another 6 studies used a cutoff score of 18 [31], 20 [22], 26 [25], 27 [27, 32], or 28 [36]. Considering the inconsistency of the cutoff point, we grouped our study populations into CI and NCI based on a clear definition by every study instead of selecting a specific cutoff point.

Generally, most of the included studies were judged to be of moderate to high quality. The risk of bias concerns in all the cohort studies were frequently about the comparability of the CI and NCI groups. Other risks of bias included short follow-up durations and the presence of outcomes at the start of the studies. The RCT was at high risk of performance and detection bias and unclear risk of reporting bias (Table 2, Fig. 2).

The principal findings of our meta-analysis are that older patients with perioperative diagnoses of cognitive impairment by the MMSE had higher risk of postoperative delirium, in-hospital mortality and mortality within 1 year. We investigated the timing of assessment with the MMSE and the respective effect on adverse outcomes. Preoperative diagnosis of CI appeared to yield a more significant association with postoperative delirium than postoperatively diagnosed CI, however, according to the sensitivity analysis, the wide variance in the observed effect was due to high heterogeneity, primarily due to the inclusion of the study of Moncada’s; therefore, definitive conclusions cannot be drawn. We suspected that as postoperative delirium is most common on the first and third postoperative days [39], the results of the meta-analysis may have been altered if the onset of delirium was earlier than the timing of postoperative diagnosis of CI; thus, preoperative use of the MMSE may be preferable to predict the incidence of postoperative delirium. Postoperative CI patients did not have an extended stay length in orthopedic wards or rehabilitation wards. However, these results raised questions about insufficient sample sizes and heterogeneity in the eligible studies.

Our systematic review and meta-analysis was novel in providing data showing the predictive value of perioperative assessment by the MMSE on postoperative outcomes in older patients. We also took into consideration the timing of assessment of cognition, as preoperative cognitive impairment revealed a chronic aging-related change while postoperative CI usually developed with acute onset and was confounded by surgery, anesthesia, medication and a stress response [40]. We summarized the data from 21 studies and used both the NOS and the Cochrane risk of bias tool to appraise the quality of selected cohort studies and an RCT, respectively. The methodological quality of studies in our review was fair to high. No significant bias of publication was observed in the report of postoperative delirium and mortality. These factors contributed to more powerful evidence than any single study or previous systematic review that failed to conduct a quantitative data synthesis.

However, there are some potential limitations should be considered. One important limitation of our study was the innate defects and the use of inconsistent cutoff scores for the MMSE across studies, which might have impacted the positive diagnosis rate of cognitive impairment. The MMSE has been validated widely around the world; however, it has been suggested that the MMSE does not perform better as a rule-out tool than a definitive diagnostic tool, which means that for those positive on the MMSE, a more detailed evaluation and inspection are required [12]. The MMSE is not the most efficient tool. Typically, it will take 8 min to complete an assessment in NCI individuals while taking 15 min or longer to evaluate CI patients [41]. Previous studies have indicated that the MMSE would be most suitable in specialist settings compared with community setting and primary care settings due to significant intraobserver differences [42]. The MMSE does not perform well enough in patients with mild cognitive impairment and early dementia [43]. Performance can be disrupted by education, age, language, ethnicity and cultural differences [44], thus optimal cutoff values change in different clinical settings. For instance, threshold values of 21, 23 and 24 are suggested in populations with primary school, high school and university education, respectively [12]. Therefore, we decided to group our study population based on a clear definition of CI by every investigator instead of selecting a specific cutoff score.

Another limitation was that some of our included studies lacked clear discharge criteria and detailed rehabilitation interventions, making the outcome of length of stay less convincing. Selection and measurement biases existed as medical and rehabilitation strength varied from different districts and hospitals. Consequently, the results found in our study need to be interpreted with caution.

Most of our included studies have restricted enrollment or matched the two cognitive groups in terms of age, sex, education levels, and surgery types and so on, even though the variables in individual studies were different. There was an inconsistency distribution of age between the two cognitive groups in some studies (Additional file 2: Material 2), and age did have a negative effect on postoperative outcomes (Additional file 3: Material 3). We could reasonably have concluded that age may act as a confounder. Adjustment for a wide range of potential confounders in individual studies were listed in Additional file 4: Material 4. We carried out a subgroup analysis by whether studies adjusted for age to decrease the effects of age on postoperative delirium. we failed to do Subgroup analysis of other outcomes owing to the lack of adjusted ORs in original studies.

Delirium and dementia are among the most common causes of cognitive impairment in clinical settings, yet their interrelationship remains poorly understood and they are often either unrecognized or mistaken for each other [45]. Our study had limitations that warrant consideration. Cognitive status was measured using the general screening tool, MMSE, and did not specifically identify dementia or delirium., Almost all our included studies chose not to use the more specific concepts of delirium or dementia, as they frequently relied on second-hand observations or precise diagnostic methods instead of a single examination like MMSE, and thus were often under- or misdiagnosed, especially by non-psychiatric staffs. Instead, they investigated the broader concept of cognitive impairment using MMSE. Thus we classified the patients into two cognitive groups and thereby minimize its false positive and negative results. We have not investigated whether acute or chronical cognitive impairment had different surgical outcomes as almost all our included studies did not provide baseline level of MMSE or a clear history of pre-existing dementia prior to admission or earlier. As a matter of fact, it was hard and exhausting for clinicians to get these medical history of patient especially geriatric ones in clinical settings. Previous studies proved that both chronic and acute cognitive impairment were independent risk factors for a poorer outcome after hip fracture [46‐50]. Yet it still remained unclear which one would be involved with worse surgical outcomes. Beyond all question, dementia as well as delirium during an acute and intense stress, like surgical procedures or hip fracture, unveils the preexisting or subclinical frailty of the geriatric individual. It is critically important for Clinicians and researchers to screen for this part of surgical patients and pay attention for the presence of a frailty syndrome, which make sense of our study and the perioperative application of MMSE. Because of its wide acceptance as a general screening instrument for cognitive dysfunction, and because the test can be performed at the bedside within a relatively short period of time.

Some directions for future research should be drawn out from our findings. With the population aging, the demand for surgery in the elderly is growing. There are striking differences in the tolerance, recovery and clinical outcomes between cognitive impaired and cognitive intact older patients following surgery. The American College of Surgeons has provided both preoperative and postoperative rounding checklists for geriatric surgical patients, including a strong recommendation of the assessment and documentation of cognitive dysfunction [51, 52]. Our review further recommends that the MMSE may be used as a reliable preoperative screening tool as well as a postoperative follow-up index in geriatric surgical settings to optimize risk stratification, assess prognosis in this population, and provide indications for early and effective interventions. Moreover, further research is needed to search for more and better assessment instruments to help make clinical decisions for older patients.

We found that older patients with perioperative CI were more likely to suffer from postoperative delirium, in-hospital mortality and mortality within 1 year. The MMSE showed certain value on risk stratification and prognosis evaluation in the geriatric surgical population.