A decline in cognitive status after surgery is common among older surgical patients. This may present in the form of delirium or more subtly in the form of postoperative cognitive dysfunction (POCD).1 Postoperative delirium occurs in up to 10-60% of older patients and is associated with longer hospital stay, increased morbidity and mortality, and higher health care costs.2 Delirium has been described as an acute confusional state featuring disturbances in attention and decreased awareness of the environment.3 The definitive diagnosis of delirium in hospitalized patients is typically made by psychiatrists or neurologists according to the criteria described in The Diagnostic and Statistical Manual of Mental Disorders – fifth edition (DSM-V).4

The development of delirium is thought to be a multifactorial process in which there is a complex interrelationship between baseline patient vulnerability and precipitating factors or insults.5 In surgical patients, predisposing risk factors that increase the vulnerability for delirium include many factors, including age, preexistent cognitive impairment, and pain.2,6 Insults that serve as precipitating factors include events related to surgery, such as type of surgery, blood loss, exposure to medications with central nervous system effects (e.g., opioids), and sleep disruption.2,6-8

In contrast, POCD is broadly defined as a decline from preoperative performance in cognitive domains such as memory, processing speed, or executive functioning after surgery. Postoperative cognitive dysfunction refers to subtle declines in cognitive functioning that can occur in the absence of delirium and are detectable through formal cognitive testing. Unlike delirium, POCD is not recognized in the International Classification of Diseases and is not listed as a diagnosis in the DSM-V. A typical patient with POCD is oriented but exhibits significant declines from his or her own baseline level of performance on one or more cognitive domains.9,10 Assessing patients for POCD requires cognitive testing before (which serves as a baseline) and after surgery. A further distinction should be made between mild cognitive impairment (MCI) and POCD, because MCI is described as a memory impairment beyond that expected for subjects who are not demented (controlling for age and education).11 Individuals with MCI have a greater decline in long-term cognitive functioning compared with those with no MCI,12 whereas POCD that occurs soon after surgery has been described as short-lived.1

Postoperative cognitive dysfunction has been observed in 19-47% of noncardiac surgical patients13 and frequently persists beyond the early postoperative period.13 Postoperative cognitive dysfunction has also been associated with impairments in daily functioning,14 premature departure from the labour market,15 and dependency on government economic assistance after hospital discharge.15 Both POCD and delirium should be distinguished from dementia, which describes a chronic and often insidious decline in cognitive function with significant functional impairment.

Few have considered whether POCD exists among patients whose cognitive symptoms do not meet standardized criteria for delirium. Among studies that measured POCD, most did not concurrently assess for delirium.1,16,17 Furthermore, most studies did not measure POCD in the days soon after surgery, a time when many patients are scheduled for early discharge planning that includes receiving postoperative self-care instructions. Accordingly, this study aimed to determine the incidence of postoperative delirium and POCD in the first two days after noncardiac surgery and whether the presence of delirium or POCD on the first postoperative day persisted to the next day.

Methods

The study was conducted at one of the teaching hospitals at the University of California, San Francisco. This study is part of a larger prospective study evaluating the pathophysiology of postoperative delirium and POCD, including their interrelationship.18,19 The study received approval from the Institutional Review Board (approved April 2001; renewed March 2014), and all patients provided written informed consent.

The inclusion criteria for the cohort were fluency in English, age 65 yr or older, scheduled for major noncardiac surgery requiring anesthesia, and expected to stay in the hospital for at least 48 hr postoperatively. The patients were recruited consecutively if their surgical procedures were conducted on Mondays to Wednesdays to allow for the two days of postoperative follow-up to be completed during a regular work week. Patients were typically recruited, and informed consent was obtained at the preoperative clinic within one week of the planned surgery.

Each patient was interviewed approximately one week before surgery and on the first two days after surgery. Patients’ postoperative visits were conducted daily between 9 am - noon at bedside. At each time point, the presence of delirium was measured using the Confusion Assessment Method (CAM) via a structured interview.18 The CAM was developed as a screening instrument based on operationalization of DSM-III-R criteria for use by non-psychiatric clinicians in high-risk settings. This method has a sensitivity of 94-100% and a specificity of 90-95% for delirium.20,21 All research personnel administering the CAM were trained based on a detailed CAM manual developed by Inouye et al. 20 All cases of delirium were validated by a second investigator (L.P.S.) who reviewed a written summary of each patient’s responses to the structured interview performed by the first investigator and discussed the assessment with that interviewer.

Postoperative cognitive dysfunction was measured by the Digit Symbol Substitution Test,22 the timed Verbal Fluency Test,23 and the Word List Learning Task24 in order to assess the cognitive domains of memory and learning (word list), verbal and language skills (verbal fluency), and attention, concentration, and perception (digit symbol test). These tests target domains that are sensitive to drug effects25 and have been used and validated in a large number of older surgical patients.16 The other important consideration is that, in contrast to previous studies evaluating patients at least one week after the planned surgery, our study focused on evaluating cognitive status in the immediate two days after surgery. Hence, we had to be sensitive to reduce the burden of cognitive testing for patients who have recently undergone surgery. To address the potential confounding effects of learning from repeated cognitive testing, we administered different forms of the cognitive tests using a Latin square design to prevent potential confounding of form by occasion.26 The multiple test forms used were randomized using the Latin square treatment structure where the two blocking factors were measurement occasion (e.g., first and second postoperative days) and form number. Earlier research provides evidence of equivalence of the four forms of each cognitive test.27

For each test, we determined whether the patient experienced a significant decline from preoperative baseline using prediction intervals.25 A decline from preoperative performance of four or more points for the word list or seven or more points for the verbal fluency and the digit symbol tests was considered significant decline, and the patient was classified as having POCD for that day. If decline in performance was observed for at least one postoperative day, we concluded that POCD occurred for that patient.19 The above scores were determined in earlier studies of community living older adults who were similar in demographic and clinical characteristics to the subjects in this study but had not undergone an intervention that could cause cognitive change.28-30 In addition to the above tests, cognitive status was also measured preoperatively using the Telephone Interview for Cognitive Status (TICS) instrument31 which was adapted from the Mini Mental Status Examination. The TICS allowed us to evaluate the baseline cognitive status of patients, but patients were not excluded based on the TICS values.

To distinguish whether patients had developed postoperative delirium and/or POCD (or neither), we used a hierarchical method. If a patient met the criteria for delirium, that patient was classified as having delirium. If a patient did not meet criteria for delirium but met the criteria for POCD, that patient was classified as having POCD. Finally, the patient who did not meet criteria for delirium or POCD was classified as having neither condition.

Other demographic variables measured included age, sex, education level, and alcohol intake. Preoperative symptoms of depression were measured by the 15-item Geriatric Depression Scale.32 Functional status was measured using the Activities of Daily Living (ADL)33 and the Instrumental Activities of Daily Living (IADL).34 If a patient could not perform independently without assistance in one or more of the activities related to either ADL or IADL, that patient would be considered dependent on support to perform the activities listed.

The level of comorbidity was determined using the Charlson comorbidity index.35 Other perioperative data obtained from chart review included the type of surgery and the American Society of Anesthesiologists’ risk classification.36 Additional perioperative variables measured included the type of surgery, type of anesthesia, anesthesia duration, and type of postoperative analgesia.

The Pearson Chi square test of association was used to test if there was a relationship between cognitive status on postoperative day one and that on postoperative day two, and the Cramer’s V statistic was used to measure the strength of the relationship. The Pearson Chi square statistic was also used to assess the relationship between place of discharge and postoperative cognitive status, and Fisher’s exact test was used to assess the relationship between the incidence of preexistent dementia and postoperative cognitive status. One-way analysis of variance (ANOVA) was used to analyze the relationship between preoperative cognitive status (as measured by the Telephone Interview for Cognitive status) and postoperative cognitive status. A two-way ANOVA was used to analyze the relationship between hospital length of stay and the explanatory variables, including cognitive status and surgery type. For this analysis, hospital length of stay was analyzed on the log scale but transferred back to the original scale for interpretation. All statistical analysis was performed with SAS® version 9.2 (Cary, NC, USA) and R 2.15.1.

Results

Four hundred sixty-one patients provided consent to participate in the study during June 2001 to December 2010. Forty patients were excluded from the analysis because they did not meet criteria for delirium and did not complete the cognitive tests on either postoperative day one or two, which left 421 patients for analysis. Specifically, eight of the forty patients were discharged early, 22 refused, one was too sedated, and nine were physically incapable of performing the cognitive tests.

This current study reports on patients with complete delirium and cognitive assessments on the first two postoperative days. The mean (SD) age of the study patients was 73.7 (6) yr; 53.3% were female. The majority of patients underwent either abdominal or spinal surgery, and 62% of patients received patient-controlled analgesia using hydromorphone for postoperative pain control (Table 1).

Table 1 Demographic data by postoperative day one cognitive status
Full size table

No patient had preoperative delirium. Two hundred (48%) of the 421 patients experienced postoperative delirium; delirium occurred in 162 (39%) patients on postoperative day one and 152 (36%) patients on the subsequent day. The relationship between cognitive status on postoperative day one and postoperative day two is shown in Fig. 1. The Cramer’s V statistic of 46.7 indicates a strong association between the cognitive status on day one and that on day two. The presence of delirium on the first postoperative day was strongly associated with the presence of delirium on the subsequent day (χ2 = 183.9; P < 0.0001). For example, 114/162 patients (70%) who had delirium on the first postoperative day also had delirium on the next day.

Fig. 1
figure 1

The study flow chart is shown in Fig. 1. Overall, 461 patients were recruited, but 421 patients were included in this study. The reasons for excluding 40 patients included eight who were discharged early, 22 who refused testing, one was too sedated, and nine were physically incapable of performing the cognitive tests. The patients’ cognitive status for postoperative days one and two is shown here, stratified by the presence of delirium, no delirium but presented signs of POCD, or neither. POCD = postoperative cognitive dysfunction

Full size image

In patients without concurrent delirium, POCD occurred in 166 (39%) patients on postoperative day one and 157 (37%) patients on postoperative day two. The presence of POCD on the first postoperative day was likely to persist to the second postoperative day; 105/166 (63%) patients who had POCD on postoperative day one continued to have POCD on the next day. More than half of the patients (54/93; 58%) with neither postoperative delirium nor POCD on postoperative day one continued to be free of either delirium or POCD on postoperative day two. Descriptive details of the changes in each cognitive test are shown in Table 2.

Table 2 Descriptive summary of cognitive tests
Full size table

Only 22 (14%) patients who had delirium on the first day were free of delirium or POCD on the second day. Similarly, only 36 (22%) patients who had POCD on the first day were free of POCD or delirium on the second day. Altogether, only 54 (13%) patients were free from either delirium or POCD on the first two days after surgery (Fig. 1).

The preoperative cognitive status, as measured by the TICS, was significantly different between the subgroups [delirium: 30.9 (4.6); POCD: 33.8 (3.1); neither: 32.2 (3.8); F = 21.27; P < 0.001]. Specifically, patients with postoperative delirium had lower TICS scores compared with those with only POCD or neither condition. Nevertheless, the incidence of pre-existent dementia, as documented on medical records, was not significantly different between the subgroups [neither: 16 (4%); POCD: 4 (1%); and delirium: 16 (4%); P = 0.123]. Patients with postoperative delirium were more likely to be discharged to a destination other than home after surgery (42% of those with delirium vs 18% with POCD and 17% with neither condition; χ2 = 29.4; P < 0.001).

Results of the two-way ANOVA showed only cognitive status to have an association between hospital length of stay and the explanatory variables, surgery type and cognitive status (cognitive status F = 18.6, P < 0.001; surgery type F = 1.16, P = 0.28; surgery type * cognitive status F = 2.69; P = 0.069) (Fig. 2).

Fig. 2
figure 2

For hospital length of stay, post hoc analysis after Tukey adjustment (alpha = 0.05) showed that the hospital length of stay was significantly longer in patients with delirium than in those with POCD or with neither condition. Nevertheless, the hospital length of stay in patients with POCD was not significantly different from those with neither condition. The red line indicates the mean for each group. Specifically, the mean (SD) length of stay in patients with delirium was 8.2 (7.5) days, 5.9 (4.8) days in those with POCD, and 4.7 (3.6) days in those with neither condition. POCD = postoperative cognitive dysfunction

Full size image

Discussion

In this prospective cohort study of a group of older patients who have undergone major noncardiac surgery, we found that 80% of the patients experienced either postoperative delirium or POCD on the first day after surgery. For patients who did not develop delirium on the first postoperative day, close to 40% had POCD. Furthermore, the majority of patients who had delirium or POCD on the first postoperative day did not recover by the second day after surgery. Approximately 16% of the patients who were delirious on the first postoperative day recovered but met criteria for POCD. Similarly, the same proportion of patients who were not delirious on the first day after surgery became delirious the next day. Our results are novel in that few studies simultaneously measured both delirium and POCD, particularly in the early postoperative period. Furthermore, our study results may be particularly important because of the focus on early cognitive testing which may be more likely to uncover the effects of drugs or anesthetics on cognition.

Assessment of cognitive functioning using a combination of a validated delirium scale and cognitive tests known to be sensitive to drug effects revealed that a substantial number of older patients undergoing noncardiac surgery had either delirium or POCD on the first two days after surgery. Our results suggest there is validity to patient and/or caretaker complaints of memory and thinking problems in the period soon after surgery.37 The types of cognitive dysfunction exhibited by patients in this study may affect patients’ ability to understand and carry out self-care instructions which, in turn, could reduce postoperative functioning, delay discharge, or even risk re-admission. This observation poses a clinical dilemma because detailed cognitive status is not currently evaluated in the clinical setting. Greater clinical attention to the evaluation of cognitive status soon after surgery may help with appropriate discharge planning, including the need for home care or discharge to a destination other than home.13

Our study provides novel information about the incidence of early postoperative delirium and POCD using validated measures. Nearly all studies that included assessments of POCD measured cognitive status one week after surgery, which prevented observation of the peak onset period of cognitive dysfunction and underestimated its incidence. Few studies assessed POCD and delirium in the early postoperative period. One study that measured both delirium and cognitive status for the first three days after surgery likely underestimated the incidence of cognitive dysfunction after surgery due to the measurement tools. The investigators determined delirium by reviewing medical records, which has previously been shown to provide inaccurately low rates of delirium.13 Also, they defined POCD as a two-point decline on the Mini Mental State Exam, a tool that is validated as a screen for chronic cognitive impairment, such as dementia, but not validated for acute cognitive change that can occur in the postoperative setting. Thus, it is not surprising that the incidence of delirium (17%) and POCD (23%) in the first two days after surgery were half that reported in this study. Although methods of measurement and incidence rates differed between the two studies, both studies provide evidence that early postoperative cognitive changes are common occurrences in older surgical patients.

The overall rate of delirium of 48% reported in our study is comparable with previous studies of older surgical patients undergoing major surgery.2,38-40 Among high-risk groups (such as hip fracture), rates of postoperative delirium amongst studies have been reported from 16-62%.41 Nevertheless, studies performed in more homogeneous patient populations, such as those undergoing orthopedic major joint arthroplasty, have reported lower rates of delirium.39,42

The incidence of POCD in our study (52%) was higher than that reported by previous investigators;1,17,43,44 however, previous investigators measured POCD at one week after surgery. For example, the International Study of Postoperative Cognitive Function (ISPOCD1) evaluated 1,218 elderly patients who had undergone major noncardiac surgery and found that 26% of them had POCD one week after surgery.1 In a second report by the ISPOCD group in middle-aged patients (aged 40-60 yr), they reported an incidence of 19.2% of POCD in patients after noncardiac surgery.16 Nevertheless, the incidence of POCD reported by Monk et al. was similar to that reported in our present study. In this study of 1,064 patients aged 18 yr of age or older undergoing noncardiac surgery, these investigators found that POCD was present in 117 (37%) young, 112 (30%) middle-aged, and 138 (41%) elderly patients45 at hospital discharge. Therefore, older age appears to be related to a higher incidence of POCD. Other factors that may account for the difference in results include different cognitive tests being used, different criteria used to define POCD, and timing of cognitive assessments.

Although a variety of scoring methods for the detection of POCD have been used across studies, investigators generally agree that scoring methods should consider baseline performance, practice effects, and change on more than one neuropsychological test.46-49 Prior studies using controls often included control groups that differed from the subjects, e.g., control groups with fewer males,45,50 lower depression levels,45 lower rates of comorbidity,47 and lower attrition rates.50 Therefore, the use of controls to correct for learning effects for repeated cognitive tests remains controversial.

It remains to be determined whether patients who did not meet the criteria for delirium but have POCD should be considered as a group with subsyndromal delirium. The definition of subsyndromal delirium remains ill defined, as investigators have used a variety of definitions, including requiring one or more core CAM symptoms, not meeting the criteria for delirium, or not progressing to delirium.51 Because the definition of POCD is equally heterogeneous among published studies and differs substantially from that for subsyndromal delirium, it remains unclear whether POCD and subsyndromal delirium are similar entities on the same spectrum of cognitive dysfunction. Further research is necessary to determine the relationship between subsyndromal delirium and POCD.

There are a few potential limitations of our study. First, because we performed delirium screening and cognitive testing on only the first two postoperative days, later cases of delirium and cognitive changes might be missed. Second, we measured delirium only once daily, and given the fluctuating nature of delirium, we might have missed cases of delirium that occurred in the afternoons or evenings. Third, we did not include a tool to measure the severity of delirium, which should be considered in future studies of delirium and POCD.

The National Institute for Health and Care Excellence has published guidelines that discuss the importance of the assessment for delirium in older surgical patients and described the cost-effectiveness of interventions to prevent delirium. Nevertheless, no such practice guidelines exist for the assessment of POCD in the early postoperative setting. Given that 52% of older surgical patients were found to have POCD in the first two postoperative days, it is imperative to follow up with future investigations of POCD in the early postoperative setting. Future studies are needed to determine how such cognitive changes affect patients’ post-discharge functioning and ability to perform self care and to assess the cost-effectiveness of treatments to prevent POCD.