Postoperative delirium and health related quality of life after coronary artery bypass grafting

Abstract

Objectives. We wanted to identify determinants for postoperative delirium and its influence on health related quality of life (HRQoL) during 36-month follow-up of coronary artery bypass (CABG) patients. Design. A total of 302 patients were retrospectively analyzed. HRQoL was assessed prospectively by the15D instrument. Delirium was diagnosed clinically. Results. The incidence of delirium was 6.0%. The cumulative survival (all-cause death) in 36 months was 96.1% in patients without delirium and 77.8% in patients with delirium. Age, cerebral disease, chronic heart failure, male gender, postoperative pneumonia and low output syndrome were predictors for delirium. Delirium patients needed more resources i.e. intensive care or total duration of hospitalization and experienced no positive change in HRQoL. Moreover patients with high preoperative 15D score tended to suffer fairly severe but reversible impairment during the first 6 months after the operation. Conclusions. Preoperatively older and sicker patients with complicated postoperative course are at higher risk of developing delirium after CABG. Preoperative status and operative complications together with delirium may exert negative influence on forthcoming HRQoL, which is seen especially in patients with a relatively high preoperative level of HRQoL.

• CABG
• delirium
• Quality of Life
• 15D

Coronary artery bypass grafting (CABG) has reduced mortality and morbidity of coronary artery disease and has also served to relieve angina and to improve health-related quality of life (HRQoL). Despite these advances postoperative adverse neurobehavioral disturbances continue to be a fairly common complication and are associated with increased morbidity, prolonged stay in intensive care, longer hospitalization stay and negative impact on HRQoL 1–5. In the present retrospective study, we wanted to ascertain the incidence and pre- and perioperative determinants for postoperative delirium among CABG patients. We also aimed to examine a possible impact on changes in HRQoL during 36-month follow-up.

Material and methods

Study population

The material consists of 302 patients undergoing elective CABG at Vaasa Central Hospital over a period from October 2000 to January 2003. Patients were considered for prospective evaluation of HRQoL as a part of a larger study of HRQoL for coronary artery-diseased (CAD) patients. After obtaining approval from the Ethical Board and written informed consent from the patients, the research nurse interviewed the patients in order to elicit preoperative HRQoL. The 15D was used to determine HRQoL 6. Re-evaluation of HRQoL was done by a questionnaire mailed to the patients 6 and 18 months after surgery, while the research nurse personally interviewed surviving patients 36 months postoperatively.

Operative and anaesthesiological technique

The study cohort underwent CABG in which cardiopulmonary bypass (CPB) was performed by using membrane oxygenator and an arterial line filter in 93.3% of operations, whereas 6.7% of operations were performed off-pump (OPCAB). During the study period epiaortic ultrasound imaging was not available and the ascending aorta was assessed by the attending surgeon's palpation to determine the placement of the aortic cannula, x-clamp or proximal anastomoses of grafts. All anastomoses were performed with a single-clamp technique with aortic root venting. Myocardial protection was ensured with continuous retrograde blood cardioplegia combined with antegrade infusion into grafts.

On the evening before the operation patients were premedicated with lorazepam (1–2 mg orally) and on the morning of the operation premedication consisted of diatzepam (5–15 mg orally) and morphine (0.1 mg/kg intramuscularly). A standard combined anaesthetic technique consisted of propofol and fentanyl supplemented with isoflurane or sevoflurane in 40–60% oxygen in air. Neuromuscular blockade was established with pancuronium (0.1 mg/kg intravenously). Thoracic epidural anaesthesia was used in 25.7% of operations, in 25.2% of patients without delirium and in 33.3% of patients with delirium (NS). An epidural catheter was introduced on the day before the operation and the duration of epidural anaesthesia was until the 3^rd to 4^th day after operation. Postoperative care in the intensive care unit (ICU) was standard with stabilization of haemodynamics with volume infusion, inotropics (dobutamine) and vasopressors (noradrenalin) when necessary. Patients were extubated when hemodynamically stable and normothermic. After uncomplicated recovery patients left the ICU on the first postoperative day. Standard postoperative pain medication consisted of intravenous (0.05–0.07 mg/kg) or intramuscular (0.1–0.14 mg/kg) oxycodone when necessary during the first days. Pain medication was usually changed to oral paracetamol or paracetamol combined with codeine after 2–4 days after the operation.

Definitions

Data concerning clinical risk factors, findings of coronary catheterization, details of surgery, extra-corporeal circulation (ECC) and details related to the postoperative outcome, including morbidity and mortality were collected according to the Society of Thoracic Surgeons (STS) definitions 7 and according to the EuroSCORE criteria 8, 9. Thus congestive heart failure (CHF) was defined as paroxysmal nocturnal dyspnea or dyspnea on exertion due to heart failure, or chest x-ray showing pulmonary congestion. Patients should also have received diuretics or digoxin. Chronic renal insufficiency was defined as elevated preoperative serum creatinine ≥140 µmol/l or if hemodialysis was in use. Cerebral vascular disease (CVD) was defined as previous transient ischaemic attack (TIA), stroke, previous surgery of carotid arteries or as a detected asymptomatic stenosis >75% in one or both carotid arteries. Postoperative stroke was defined as a new reversible or permanent (>24 hours) neurological deficit. Low output syndrome (LOS) was defined as prolonged need for vasoactive medicine support from the first postoperative morning onwards, or as a need of intra-aortic balloon pump (IABP). A perioperative myocardial infarction (MI) was diagnosed by elevation of creatine isoenzyme mass (CKMBm) >50 IU/l any time after the operation, with or without a new Q-wave in the electrocardiogram (ECG).

In the present study diagnosis of delirium based on fulfilment of criteria, which included acute onset and fluctuating course, inattention, delusions, confusion, and cerebral excitement or altered level of conciousness and having a comparatively short duration 10. A definitive criterion was also that temporary medication, i.e. diazepam or haloperidol, was needed to sedate the delirious patient. A case of delirium developing at any time during the primary hospitalization was included in the analysis. The state was diagnosed clinically and no neurocognitive test was used to identify neurobehavioral disturbances before or after surgery. Cerebral tomography scan (CT) of the brain, for detecting obvious embolization, was performed only in the case of major cerebral complication, i.e. stroke.

Determination of the HRQoL

The 15D is a non-disease-specific, comprehensive, multidimensional, standardized and self-administered measure of HRQoL. It describes the health status along 15 dimensions (Figure 2). Each dimension comprises five answer options. The single index score, 15D score on a 0–1 scale, represents the overall HRQoL. The maximum score is 1 (no problems on any dimensions) and the minimum score 0 (being dead). The 15D scores have proved highly reliable, sensitive and responsive to treatment related change 6.

Statistical analysis

Patient-related and postoperative outcome variables are presented as percentages. Continuous variables are reported as mean±standard deviation (SD) or as median [range]. Cumulative survival was calculated using the Kaplan-Meier method and the log-rank test was used to compare survival between patient groups. Baseline and follow-up variables were compared using Pearson χ² Test or Fisher's Exact Test. Continuous variables with non-normal distributions were compared by Mann-Whitney test. Binary logistic regression (forward stepwise method) was carried out to assess the factors associated with postoperative delirium. The preoperative variables analyzed were age, gender, previous CVD, CHF, chronic renal insufficiency and additive EuroSCORE. Of the postoperative events we analyzed re-operation for bleeding, perioperative MI, LOS, pneumonia, postoperative stroke, renal failure and need for transfusions. The level of statistical significance was set at <0.05. Statistical analyses were performed using the SPSS statistical software programme 14.0 for Windows.

Results

There were three early deaths in the original study cohort. One patient was lost on the table as a consequence of irreversible LOS and another patient died on the first postoperative day when still connected to the respirator and comatose. These two patients were excluded from the study, thus there remained a total of 300 patients for final analysis. 30-day mortality of the primary study cohort was three patients, 1.0%. During the study period of 36 months 15 more patients died giving an overall 3-year survival of 95.0%. Cumulative survival was 96.1% in patients without delirium and 77.8% in patients with delirium (log-rank test: p < 0.001) at 3 years (Figure 1). During the hospitalization in the operating unit 18 patients (6.0%) suffered delirium. All these patients had undergone CABG with cardiopulmonary bypass. The patients with delirium were older and they more often had a history of CVD (p = 0.002) and higher incidence of CHF. Both additive and logistic risk score of mortality was higher in this group (Table I). Moreover, five patients of 18 with postoperative delirium (27.8%) and 15 patients without delirium (5.3%) (p = 0.001) had a history of malignancy, all in remission at the time of the operation and one patient in the delirium groups and two patients in the no-delirium group had problems with alcohol. Postoperatively patients who developed delirium were more frequently subjected to resternotomy for bleeding (p = 0.036) and they also needed more often vasoactive support for hemodynamic stabilization on the first postoperative morning (p = 0.017). They also experienced significantly more often readmission to the ICU (p < 0.001) and reintubation (p < 0.001). Postoperative complications, pneumonia (p = 0.006) and atrial fibrillation (AF) (p = 0.004) were significantly more common in patients with delirium. Stay in the ICU and total length of hospitalization including recovery in other medical care units was also longer in this group of patients (Table II).

Figure 1.  Kaplan-Meier survival curves for CABG patients without or with postoperative delirium.

Figure 2.  Baseline (preoperative) 15D profiles and scores of CABG patients.

Table I.  Univariate analysis of preoperative variables of CABG patients.

Variable	No delirium n = 282 %	Delirium n = 18 %	p-value
Age, years (SD) [range]	65.8 (9.0) [44.1–84.5]	72.0 (7.0) [54.2–80.1]	0.005
Male gender	78.1	94.4	0.136
Chronic pulmonary disease	9.6	16.7	0.405
Extracardiac arteriopathy	4.6	5.6	0.588
Neurological dysfunction	0.4	0	1.000
Previous cardiac surgery	3.2	0	1.000
Renal dysfunction	0	5.6	0.060
Unstable angina	9.9	16.7	0.413
LVEF 30–49%	42.9	50.0	0.627
LVEF <30%	3.2	5.6	0.467
Recent myocardial infarction	9.9	16.7	0.413
Pulmonary hypertension	0.4	5.6	0.117
Emergency	1.1	5.6	0.220
Diabetes	26.6	11.1	0.174
Hypercholesterolemy	70.9	72.2	1.000
Hypertension	55.7	66.7	0.465
Previous cerebral disease	6.7	33.3	0.002
Chronic heart failure	2.8	16.7	0.022
NYHA Class 3–4	61.3	66.7	0.653
ASA 4	42.9	50.0	0.556
EuroSCORE, additive, mean (SD)	3.3 (2.3)	4.9 (2.5)	0.007
EuroSCORE, logistic, mean (SD)	3.3 (3.8)	5.3 (4.5)	0.005

ASA, American Society of Anaesthesiologists, LVEF, left ventricular ejection fraction, NYHA, New York Heart Association.

Table II.  Details of operations and postoperative outcome of CABG patients without or with delirium.

Variable	No delirium n = 282%	Delirium n = 18%	p-value
Aortic occlusion time, min, mean (SD)	87 ( 28)	95 (29)	0.270
Perfusion time, min, mean, (SD)	108 (32)	118 (40)	0.337
Number of distal anastomoses, mean (SD)	4.1 (1.2)	4.4 (1.2)	0.200
Respiratory time, h, mean (SD)	6.8 (6.2)	6.7 (3.4)	0.442
Reintubation	0.4	33.3	<0.001
Reoperation for bleeding	3.5	16.7	0.036
Perioperative MI	9.9	22.2	0.111
Pneumonia	6.0	27.8	0.006
Sternal wound infection	0.7	5.6	0.170
Renal failure	2.1	11.1	0.077
Stroke	2.5	5.6	0.394
Atrial fibrillation	42.6	77.8	0.004
LOS	5.0	22.2	0.017
ICU stay, primary, days, median [range]	1 [1–7]	1 [1–134]	0.182
Readmission to ICU	0.3	33.3	<0.001
Total ICU stay, days, median [range]	1 [1–7]	1 [1–150]	<0.001
Length of stay, days, median [range]	10 [3–46]	13 [7–179]	0.005

ICU, intensive care unit.

Independent predictors of delirium

When adjusted for age, gender, CVD, CHF, chronic renal insufficiency and additive EuroSCORE, age cerebral disease, CHF and male gender had an independent influence on postoperative delirium. On the other hand, among postoperative complications, pneumonia and LOS were predictors for delirium. The result of logistic multivariate analysis is summarized in Table III.

Table III.  Logistic multivariate analysis for predictors of delirium after CABG.

Variable	Odds ratio	95% CI	p-value
Age	1.08	1.01–1.17	0.029
CVD	4.91	1.43–16.85	0.011
CHF	5.28	1.02–27.29	0.047
Male gender	7.60	0.93–62.20	0.059
Postoperative pneumonia	5.34	1.64–17.37	0.005
LOS	4.71	1.29–17.13	0.019

CI, confidence interval.

Development of the HRQoL

Complete data on HRQoL in 36-month follow-up were retrieved in 92.3% of survivors. Preoperative 15D score differed between patients with delirium and patients without delirium (0.774 vs. 0.833, p = 0.024). Patients with postoperative delirium had poorer preoperative score on five out of fifteen dimensions: seeing, hearing, eating, speaking and elimination (Figure 2). In the follow-up the mean 15D score of the patients without delirium improved significantly from the preoperative level until six months and also at 18 months after CABG the 15D score of the group was significantly higher than preoperatively (Figure 3). However, in the patients with postoperative delirium the mean 15D score did not change during the study period compared to base level (Figure 3). Moreover, in patients with high preoperative 15D score (above the median for the study cohort) in conjunction with postoperative delirium (n = 5) the score tended to deteriorate during the first 6-month period but improved considerably later on. In patients with low initial score (under the median for the cohort) and with postoperative delirium no significant change of score was observed in follow-up (Table IV).

Figure 3.  The mean15D index scores (±SD) of patients without and with delirium during the study period. P-values indicate difference in the 15D score of no-delirium patients against preoperative level at the time of control. In delirium patients the difference was not significant (NS) at each time point of follow-up.

Table IV.  15D score change during 6 months after CABG in patients with delirium against to median preoperative 15D score (0.8374) of the study cohort.

			15D score
Patient group	n	Age, mean (range)	Preoperative	6 months	p-value
Delirium, score > median	5	70.8 (67.4–78.5)	0.9225	0.8071	0.080
Delirium, score < median	13	72.4 (54.2–80.1)	0.7164	0.7727	0.222

Discussion

Delirium is a dramatic syndrome having acute onset and varying course. It is characterized by impairment of cognitive functions, incoherent thinking, a reduced level of consciousness, attention abnormalities, increased or decreased psychomotor activity and a disordered sleep-wake rhythm 10, 11. The reported incidence of delirium after CABG varies widely depending on the method used in assessing neurobehavioral changes. If the diagnosis has been made clinically, without specialized tests, the reported incidence varies from 2.1 to 32% 1, 5, 11–13. When validated neuropsychological assessment has been used the incidence of type II neurological outcomes (neurocognitive decline, delirium included) rises up to 30–50% 14–18. Predictors of postoperative delirium include advanced age, preoperative poorer mental status, history of cerebral disease, atherosclerosis of ascending aorta and carotid arteries and peripheral vascular disease 5, 11, 14, 19. During operation with CPB microemboli, consisting of white cell and platelet aggregates, fat or air, may be implicated in more subtle diffuse cerebral dysfunction 1, 21. Among the postoperative variables stroke and infection have an independent impact on delirium 5. Postoperative neurobehavioral and neurocognitive complications are associated with prolonged stay on the ICU and with the length of hospitalization after CABG 5, 13 and may impair the HRQoL of CABG patients 16, 20. On the other hand, beating-heart surgery has been shown to protect an individual against postoperative delirium 11.

In the present study clinically evident delirium occurred in 6.0% of patients. The incidence correlates with those earlier studies in which delirium was diagnosed by clinical estimate of deviant behavior. Certainly a much higher incidence of neurobehavioral complication might well have been found if neuropsychological tests had been used. However, clinical symptoms occurring on the ward form the basis for decisions regarding the treatment of an individual patient.

In our study male gender, history of CVD, CHF and age remained independent predictors for delirium, supporting findings in earlier studies. The same concerns postoperative variables, particularly pulmonary infection and LOS, which were seen to influence independently on delirium. The rate of postoperative complications was high in the delirium group. No doubt a complicated outcome (i.e. reoperation, reintubation) requires a more aggressive plan for pain management and continued intravenous sedation, something which can lead to delirium. Unfortunately it was not possible to ascertain the amount and duration of opioid medication used for individual patients in this retrospective analysis. The role of thoracic epidural anaesthesia remains questionable in connection with delirium, despite of the fact that epidural anaesthesia reduces the need of opioid medication. Also a history of malignancy, if only in remission during the time of operation, would rather have impact on postoperative HRQoL and survival than on development of delirium. In summary, preoperatively older and sicker patients with complicated postoperative course are at higher risk of developing delirium after CABG. The most common adverse event in our study was AF, which appeared even more often in the patients with delirium. However, the role of AF as a cause of delirium may be controversial although the incidence of AF is at highest during first postoperative days after CABG and although it may cause temporary haemodynamic instability, hypoperfusion, intracardiac thrombus formation and embolism.

A significant finding in the present study was difference between no-delirium and delirium patients in the requirement of resources i.e. stay in the ICU or total duration of hospitalization. Actually this means higher total costs although ascertaining costs was not our aim in the present survey. Taking into account the complications which have a major influence on postoperative delirium, the need for resources may merely correlate with the underlying causes of delirium rather than with delirium itself. The survival of delirious patients seemed to be poorer than that of non-delirious patients. In fact, subsequent mortality consisted in both groups almost exclusively of malignancies, some of which had already been identified at the time of the operation.

An end point in our analysis was to evaluate the influence of postoperative delirium on HRQoL after CABG. Patients without delirium recovered well after surgery and HRQoL improved to a significantly higher level after 18 months than before the operation. After 36-month follow-up HRQoL regained its preoperative level. This later progress in HRQoL may be natural over a fairly long observation time, 3 years, which may be long enough to accomplish qualitative change in several dimensions of life in patients who basically were quite old in age. The observed difference in preoperative HRQoL between non-delirious and delirious groups depends partly on the higher mean age of the patients in the latter group. Another interpretative factor is that patients in the delirium group had more co-morbidity, malignancies included. The postoperative course was also more often complicated in the group of patients with later delirium, which may indicate that advancement in HRQoL was not seen after surgery. On the other hand patients with a relatively high preoperative level of HRQo L and with postoperative delirium seemed to have negative but later on reversible development of HRQoL during first 6 months after the operation. One explanation may be that patients with a basically good standard of HRQoL may be more sensitive to even small dimensional changes and thus experience quite radical deterioration in case of as severe an alteration as delirium. However, this finding should be addressed by further studies with a greater amount of patient material and with more sensitive instruments to assess neurobehavioral change after CABG. Elderly patients, who have a poorer development of HRQoL after CABG 22 and who are also at considerable risk for postoperative delirium may have to be subjected to careful preoperative evaluation considering the benefits and disadvantages of CABG. It is important to try to prevent pre-delirium complications, to identify early signs of postoperative delirium and to focus effective care on these patients to prevent subsequent complications and thereby improving later HRQoL. In addition, patients with postoperative delirium may need careful follow-up to detect and treat subsequent disorders in neuropsychological recovery.

Limitations of the study

The main limitations in our study are the retrospective nature of the analysis and the small sample size, affecting particularly the subgroup with postoperative delirium. Therefore any claim about an associative relationship between pre- or perioperative variables and outcomes should be viewed with caution. Aortic atherosclerosis as a possible source of arterial emboli was estimated by palpation which is no longer recommended. Delirium was diagnosed clinically and, moreover, a need for sedative medication was necessitated. This may have led to drop-out of cases with mild confusion or disorientation, in which no medication was used. The study group was also not treated entirely uniformly, as OPCAB patients were included and epidural analgesia was no principal option. However, perioperative care, perfusion technique included, was standardised, and concerning HRQoL the study was prospective with long observation period, which serves to strengthen conclusions on the results.

Conclusion

In the present study we identified higher age, previous cerebral disease and CHF of preoperative factors, respiratory infection (pneumonia) and LOS of postoperative factors to have independent influence on postoperative delirium. Delirium was associated with consumption of more ICU resources and longer totals duration of hospitalization. Patients without postoperative delirium experienced significant improvement in HRQoL. However postoperative HRQoL did not change in patients with delirium, which indicates a complicated postoperative course mainly in the patient group with older and sicker individuals. It is important to prevent postoperative complications i.e. pneumonia, to identify early signs of postoperative delirium and to focus effective care on these patients to avoid subsequent problems and improve HRQoL. Patients with postoperative delirium after CABG should be subjected to careful monitoring and supported in their rehabilitation.

Acknowledgements

The study was supported by the Medical Research Fund of Vaasa Hospital District.