Journal of Alzheimer’s Disease 22 (2010) S67–S79 S67

DOI 10.3233/JAD-2010-101086
IOS Press

The Impact of General and Regional

Anesthesia on the Incidence of Post-
Operative Cognitive Dysfunction and
Post-Operative Delirium: A Systematic
Review with Meta-Analysis
Sam Ewan Masona, Anna Noel-Storrb and Craig William Ritchiea,∗
a
Centre for Mental Health Claybrook Centre, Imperial College London, Hammersmith, London, UK
b
Cochrane Collaboration Centre, John Radcliffe Hospital, Oxford, UK

Accepted 10 August 2010

Abstract. Post-operative cognitive complications such as delirium have been consistently associated with poor short and long
term outcomes, and the role of anesthesia, particularly the role of general versus regional anesthesia, remains unclear. The
objective of this systematic review with meta-analysis was to compare the influence of general, regional, or a combination of
anesthesia on the development of Post-Operative Cognitive Dysfunction (POCD) and Post-Operative Delirium (POD). Standard
bibliographic databases were searched and complimented by hand searching of original and review article references. Included
studies were randomized controlled trials comparing general to regional (spinal, epidural, or intravenous block) or a combination
of these in a cohort who were pre-operatively cognitively normal and had an average age exceeding fifty. Where POD was the
principle outcome, studies must have employed the DSM or ICD criteria. Where POCD was the principal outcome, this was
defined as any objective cognitive impairment. Twenty one studies were considered suitable for inclusion. There was no effect
of anesthesia type on the odds ratio of developing POD (0.88, 0.51–1.51 with 95% confidence) however general anesthesia was
marginally non-significantly associated with POCD (odds ratio of 1.34, 0.93–1.95 with 95% confidence). There was no evidence
of publication bias. In conclusion, it appears that general anesthesia, compared to others, may increase the risk of developing
POCD; however this has not been shown for POD. Possible reasons for this finding have been explored. This data would advocate
for the use of regional anesthesia wherever possible especially in people otherwise vulnerable to developing cognitive symptoms.

Keywords: Confusion, delirium, epidural anesthesia, general anesthesia, post-operative period, spinal anesthesia

INTRODUCTION is remarkably common, especially in the elderly, to

experience a varying degree of cognitive dysfunction
post-operatively, which can vary from mild and short-
The role of anesthesia in the development of post-
lived to severe and permanent. Such manifestations
operative cognitive complications remains unclear. It
are described as post-operative cognitive dysfunction
(POCD) which describes a range of abnormalities, one
∗ Correspondence
of which is post-operative delirium (POD). Delirium is
to: Dr. Craig Ritchie, Centre for Mental Health
Claybrook Centre, Imperial College London, 37 Claybrook Road,
a condition with a wide range of etiologies and whether
Hammersmith, London W6 8LN, UK. Tel.: +44 (0) 20 7386 1233; it is seen in those having undergone surgery or not, has
Fax: +44 (0) 20 7386 1216; E-mail: [email protected]. consistently been associated with a poor short and long

ISSN 1387-2877/10/$27.50  2010 – IOS Press and the authors. All rights reserved
S68 S.E. Mason et al. / POCD After General and Regional Anesthesia

term outcome, including increased morbidity, mortali- may be an end point of many up-stream mechanisms.
ty, length of hospital stay, increased associated health- A functional cholinergic impairment can occur as a re-
care costs, long-term cognitive impairment, and further sult of energy failure intra- and post-operatively due to
decline beyond 12 months [1–4]. hypoxia or ischemia of several causes, and this is likely
The diagnostic criteria for post-operative delirium augmented by the direct action of anti-cholinergic med-
are remarkably similar between the Diagnostic and Sta- ications. Later, peripheral inflammation, as a result of
tistical Manual of Mental Disorders (DSM) IV and In- surgery and the consequent release of pro-inflammatory
ternational Classification of Diseases (ICD) 10, which cytokines such as interleukin 1β and tumor necrosis
include fluctuating consciousness, inattention, memo- factor α (TNFα), may activate CNS microglia. The
ry impairment, and perceptual abnormalities. A key latter then further release pro-inflammatory cytokines
point to note is that post-operative delirium is not at- and cause neuronal dysfunction [14,15]. Cognitive re-
tributed to anesthetic agents, in which case it would serves to compensate during neuronal dysfunction post-
be referred to as emergence delirium, a subtype of operatively may also be diminished by extant but clin-
substance-induced delirium. Emergence delirium is ically occult Alzheimer’s disease, Lewy Body, or oth-
considered more of a concern in pediatric patients and er central neurodegenerative pathology, resulting in a
this review is interested in interval delirium (referred cognitive dysfunction such as delirium. Finally, symp-
to as post-operative delirium) in which patients emerge tom manifestation may be mediated by alteration in
from anesthesia normally. Clinical diagnostic batter- the balance of the reciprocal dopamine and cholinergic
ies include the Delirium Symptom Interview and Con- systems, as cytokine release has been shown to increase
fusion Assessment Method (CAM), with further tests dopamine levels centrally [16].
such as the Delirium Rating Scale (DRS) available to The method of anesthetic administration though is
assess the severity, a prognostic indicator [5]. The natu- a potentially modifiable risk to avoid POCD. Beyond
ral history of POD often involves a lucid post-anesthetic their central pharmacological effect, general anesthet-
phase which lasts 1–3 days, where after the fluctuation ics (GA) influence neuronal processes such as gene
of cognitive abilities becomes apparent and objective transcription, receptor efficacy, synaptic vesicle cy-
recognition of the condition is possible within this first cling, and intracellular calcium homeostasis [17–20].
post operative week. Studies have demonstrated a wide GA’s also appear to specifically influence pathways
variance in the incidence of POD, rates of between 5– currently linked to POCD through anticholinergic ef-
15% have been reported [6] and we demonstrated that fects [21,22]. These effects are not shared by regional
rates of POD were highly variable in a post operative anesthetic (RA). There have been numerous trials com-
hip surgery population (3.6–28.3% of elective and 4– paring anesthetic routes viz a viz the development of
53.3% of trauma patients) [7]. Such a disparity in re- POCD (Table 1), though there remains an absence of
sults may be attributed to the varying methodologies consensus with regards to the association between route
used in previous studies. of anesthetic administration and POCD. This may be
In comparison to the validated criteria for POD, the mediated partially through methodological problems;
definition of POCD is much looser. For instance, there challenges in randomizing patients into different arms
are no diagnostic criteria in the DSM-IV or ICD-10 of a study, maintenance of blinding, the combination of
and in previous studies the neuropsychological test bat- anesthetics used in clinical practice, poor collection of
teries employed and the thresholds considered signifi- baseline confounders, small study size and the effect of
cant have varied considerably [8,9]. One of the largest post-operative interventions such as narcotic analgesia
studies, ISPOCD 1 [10] described POCD in 25.8% of with known cognitive effects all play their part in the
patients using a variety of well recognized tests. Such genesis of this lack of conclusion.
test batteries often assess a variety of cognitive dimen-
sions such as memory, attention and executive func-
tion; many of which may be insensitive to POCD or OBJECTIVES
vary widely within this group [11]. There are a large
number of risk factors for POCD and POD that overlap, The aim of this study was to undertake a systematic
which suggests a shared pathogenesis with POD being review of the literature with meta-analysis to determine
a more severe manifestation of the same process. the influence of method of anesthetic administration on
Many authors have suggested that POD is a mani- the development of POCD and POD comparing GA,
festation of a central cholinergic deficit [12,13], which RA, and a combination of these methods.
 S.E. Mason et al. / POCD After General and Regional Anesthesia S69

Table 1
Summary of reviews assessing the role of anesthetics on the post-operative cognitive outcome
Author and year Design of included Anesthesia studies Methodology of included Findings
studies assessed studies
Bryson 2006 [23] RCT GA, RA, RBB Surgeries: non-cardiac No difference between
Outcome: delirium (including any GA and RA
confusion), any POCD
Follow-up: Unclear
Newman 2007 [8] RCT, OBS GA, RA, LA Surgeries: all No difference between
Outcome: any POCD GA and RA
Follow-up: 7–548 days
Parker 2004 [24] RCT GA, RA Surgeries: Orthopedic A risk ratio of 0.5
Outcome: any confusional state (0.26–0.95 95% CI)
Follow-up: 3–365 days favoring RA over GA
Wu 2004 [25] RCT, OBS GA, RA, LA Surgeries: all No difference between
Outcome: any POCD GA and RA
Follow-up: 0–365 days
RCT = Randomized Controlled Trial; OBS = Observational; GA = General Anesthetic; RA = Regional Anesthetic; RBB =
Retro-Bulbar Block; LA = Local Anesthetic; POCD = Post-Operative Cognitive Dysfunction; CI = Confidence Interval.

METHODS In the meta-analysis a fixed effect assumption was

made, pooling the odds ratios between studies with
Both electronic and hand searching techniques were similar interventions being compared. In POCD stud-
used to identify appropriate literature. MEDLINE, ies, where an odds ratio was not presented (or could not
EMBASE, PsycINFO, and ISI Web of Knowledge, all be calculated), group mean change for a test were de-
via OvidSP were used. Relevant references from the scribed and contextualised with the meta-analysis find-
studies identified from the electronic search were lo- ings. Quality of the included studies was assessed us-
cated. A study was excluded at this stage if the ab- ing the Jadad score which has been adapted to increase
stracts identified an inappropriate intervention or in- its use in this review (Appendix 1).
correct outcome. The manuscripts for all remaining It was anticipated that authors would compare gen-
abstracts were sought and assessed to pre-determined eral anesthesia to epidural, spinal, regional (spinal or
criteria. For foreign language studies a translator was epidural), and intravenous block distal to a cuff and
sought. To be included, the study had to be a RCT combination anesthesia (both regional and general).
presenting data on a cohort of patients with a mean General anesthesia will also be compared to all oth-
age exceeding 50 who received either a general, re- er anesthesia methods (regional and combined), in a
gional (epidural, spinal or iv block), or combination
group entitled ‘non-general’. The heterogeneity of the
of anesthesias. Studies were excluded if participants
sample is defined using a X2 test.
had pre-operative cognitive dysfunction (as defined by
The apparent influence of covariates such as the num-
the reporting author), where there was neurosurgery
ber and type of neuropsychological tests used and fre-
or had received an intervention designed to decrease
quency of assessment will be described. All analysis
the risk of post-operative cognitive dysfunction. If the
was conducted using Stata 11.0.
principle outcome was POD, the study must have em-
ployed either the DSM or ICD criteria. However for
any post-operative cognitive dysfunction, it was antic-
ipated that a large number of neuropsychological tests RESULTS
and thresholds will be employed. In order for a test
to be considered acceptable, the reporting author must The original search yielded 6968 abstracts, the vast
state that it is able to measure the presence or absence majority of which could be excluded due to an inappro-
of cognitive dysfunction and this must be achieved in priate study design, population, intervention, or out-
an objective manner. All scales for POD and POCD come measure. Seventy-five remained and the stud-
had to have been applied within three to seven days ies were reviewed yielding 21 studies which satisfied
post-operatively to avoid studies capturing the etiolog- the final inclusion criteria. Table 2 describes the char-
ically distinct emergent delirium described previously. acteristics of some of the studies that were excluded
Studies included by this systematic review process are and Table 3 summarizes those that have been includ-
described. ed. Studies included are listed in Table 4 and those not
S70 S.E. Mason et al. / POCD After General and Regional Anesthesia

Table 2
Selection of excluded studies with a brief description of their methodologies and reason for exclusion
Study Characteristics of the study and reason for exclusion
Benoit 2004 [26] This cohort study assesses post-operative delirium in 102 patients receiving combined general and epidural
anesthesia. As there is no group randomized to general anesthesia for comparison, it has been excluded.
Bracco 2007 [27] This study follows a 1293 patient cohort with general versus general with thoracic epidural anesthesia in cardiac
surgery. The participants were controlled however not randomized and therefore this study was excluded.
Campbell 1993 [28] This study explores cognitive function in 169 patients undergoing cataract after being randomized to either
general or local anesthesia. This study is designed to assess the impact of anesthetic agents on neuronal function
centrally and largely local anesthetics are not considered to be able to enter the central nervous system to influence
cognitive function.
Crul 1992 [29] This study describes subjective physical well-being and mental function in elderly patients after being randomized
to general or spinal anesthesia for urological surgery. However patients who were unsuitable for randomization
due to a heavy co-morbidity load have been added to the spinal group for analysis, without presenting data only
for those who were randomized.
Haan 1991 [30] This study explores general and spinal anesthesias for urological surgery in elderly men. Only those that did not
express a preference were randomized and the incidence of post-operative cognitive dysfunction was only stated
including those that had not been randomized.
Handley 1997 [31] This study is a randomized controlled trial of general versus general and epidural anesthesia in patients aged
18–74 for abdominal surgery. Not only was the study population too young but the psychological testing to
determine post-operative cognitive dysfunction was on the first post-operative day only and as such an incidence
was not reported within the required 3–7 post-operative day window.
Jones 1990 [32] This randomized study compares general and regional anesthesia in 146 patients undergoing elective knee or hip
replacement. However the psychological outcome measures were employed at 3 months post-operatively only
and not within the required 3–7 day window.
Marcantonio 1998 [33] This study describes the incidence of post-operative delirium in 1341 patients who have undergone non-cardiac
surgery, by looking at their medical records. Therefore this paper is non-randomized, non-controlled and
retrospective.
Nielson 1990 [34] This randomized study evaluates cognition of 98 patients after randomization to general or regional anesthesia
for elective knee arthroplasty. The outcome measures were used at 3 months post-operatively only and therefore
were not within the 3–7 day timeframe as required.
Ryhanen 1978 [35] This study assesses the effects of halothane, methoxyflurane, combined analgesic-relaxant anesthesia and epidural
anesthesia for varicose vein stripping in women. Although the patients were controlled they were unrandomized.
Yoshida 2008 [36] This randomized controlled trial assessed the influence of general and spinal anesthesias in eighty men undergoing
urological surgery. However no objective assessment of cognitive outcome was made.

Table 3
The number and nature of the included studies, comparing the outcome measures of post-operative delirium and post-operative cognitive
dysfunction
Outcome Studies Studies Mean Study Type of surgeries Volume- Volume- Mean
which size (in order of weighted weighted adapted
report (Range) frequency) mean age mean % jadad score
incidences (Range) male (Range) (Range)
All Outcomes 21 14 103 (30–408) ORTH, URO, VASC, 67 (52–84) 51 (6.7–100) 3.1 (0–5)
ABDO, CARD
Post-Operative 5 5 109 (30–262) ORTH, ABDO 72 (69–78) 28 (10–64) 4 (3–5)
Delirium
Post-Operative Cogni- 16 9 101 (30–408) ORTH, URO, VASC, 66 (52–84) 59 (6.7–100) 2.9 (0–5)
tive Dysfunction ABDO, CARD
ORTH – orthopedic; URO – urological; VASC – vascular; ABDO – abdominal; CARD – cardiac.

measuring the odds of POD or POCD are listed in Ta- compared to epidural in two cases [44,56], regional in
ble 5. Studies tended to present risk ratios rather than one case [54] and a combination of either general with
odds ratios and for consistency these are presented in spinal [51] or general with epidural [52]. The pooled
the tables. estimate shows no difference between the groups (0.88,
95% CI = 0.51–1.51) (Fig. 1).
Comparison of GA with all non-GA for POD
Comparison of GA with spinal anesthesia for POCD
In the five studies whose outcome was POD, GA was
Of the five studies which explored general versus sp-
 S.E. Mason et al. / POCD After General and Regional Anesthesia S71

Table 4
Included papers which present incidences of post-operative delirium or cognitive dysfunction
Paper Surgery Anesthesia Outcome
Number Gender Age Assessment Adapted Odds ratio
(%M) timing (days) jadad score (95% CI)
Berggren 1987 [44] ORTH GA, EPI Delirium 57 19 78 1, 7 3 0.70 (0.23–2.07)
Bigler 1985 [45] ORTH GA, SPI POCD 40 18 79 7, 90 4 1 (0.06–17.18)
Casati 2003 [46] ORTH GA, SPI POCD 30 7 84 1, 7 5 3.5 (0.32–38.23)
Chung 1987 [47] URO GA, SPI Confusion 44 50 72 1, 3, 5, 30 2 2.86 (0.27–29.80)
Chung 1989 [48] URO GA, SPI POCD 44 100 72 1, 3, 5, 30 4 1.46 (0.44–4.88)
Cook 1986 [49] VASC GA, SPI Confusion 101 70 67 Until discharge 1 0.61 (0.20–1.85)
Forster 1990 [50] VASC GA, RA POCD 64 33 73 1, 7 4 0.22 (0.02–2.07)
Kudoh 2004 [51] ORTH GA, COM Delirium 150 10 76 0–7 5 1 (0.14–7.29)
Nishikawa 2007 [52] ABDO GA, COM Delirium 30 57 71 0–3 4 1.63 (0.23–11.46)
Pan 2006 [53] ABDO GA, COM POCD 92 53 72 7 4 0.83 (0.37–1.90)
Papaioannou 2005 [54] ABDO GA, RA Delirium 47 64 60+ 0–3 4 1.45 (0.32–6.71)
Rasmussen 2003 [9] ORTH, URO GA, RA POCD 314 41 71 7, 90 3 1.72 (0.97–3.04)
Scott 2001 [55] CARD GA, COM Confusion 408 86 59 0–5 0 3.90 (1.07–14.18)
Williams-Russo 1995 [56] ORTH GA, EPI Delirium 262 30 69 7, 180 4 0.76 (0.35–1.68)
ORTH – orthopedic; URO – urological; VASC – vascular; ABDO – abdominal; CARD – cardiac; GA – general anesthetic; EPI – epidural; SPI –
spinal; RA – regional anesthesia; COM – combined; POCD – post-operative cognitive dysfunction; CI – confidence intervals.

Fig. 1. Forest plot of the five studies which compare GA with RA in the development of POD. GA – General Anesthesia; RA – Regional
Anesthesia.
S72 S.E. Mason et al. / POCD After General and Regional Anesthesia

Table 5
Papers which present only mean group changes in a neuropsychological test battery, with a summary of their findings
Paper Surgery Anesthesia
Outcome Number Gender Age Assessment Adapted Summary of Findings
(%M) Timing (days) Jadad Score
Anwer 2006 [37] ORTH, URO GA, RA POCD 60 60 62 1, 3 3 Significantly less de-
cline in RA compared
to GA
Asbjorn 1989 [38] URO GA, EPI POCD 40 100 69 4, 21 3 Equal risk from GA
and EPI
Dahn 1999 [39] ORTH GA, SPI POCD 30 37 70 0, 1, 3 3 Favors SPI compared
to GA, significantly in
1 of 10 tests
Dahn 2003 [40] ORTH GA, SPI POCD 40 43 65 0, 1, 3 3 Equal risk from GA
and SPI
Ghoneim 1988 [41] ORTH, URO GA, RA POCD 105 67 61 1–7 3 Equal risk from GA
and RA
Riis 1983 [42] ORTH GA, EPI, COM POCD 30 NR 70 2, 4, 7 3 Equal risk from GA,
COM and EPI
Somprakit 2002 [43] ORTH, URO GA, RA POCD 120 38 52 1, 3 3 Equal risk from GA
and COM
ORTH – orthopaedic; URO – urological; GA – general anesthetic; RA – regional anesthesia; EPI – epidural; SPI – spinal; COM – combined;
POCD – post-operative cognitive dysfunction; NR – not reported

Fig. 2. Forest plot of the five studies which compare GA and spinal anesthesia in the development of POCD. GA – General Anesthesia; SPI –
spinal.

inal anesthesia [45–49] there was no significant differ- Comparison of GA with epidural anesthesia for POCD
ence between interventions with the pooled estimate
of the odds ratio being 1.15 (95% CI = 0.56–2.34) There were no studies presenting incidences compar-
(Fig. 2). Two papers [39,40] presented mean group ing general and epidural anesthesia, however there were
changes, one [39] significantly favoring spinal anesthe- two papers which presented mean group changes [38,
sia in one of ten tests and the other finding no significant 42], which both showed an equal risk for POCD be-
differences between the groups [40]. tween anesthesia methods.
 S.E. Mason et al. / POCD After General and Regional Anesthesia S73

Fig. 3. Forest plot of the two studies which compare GA and RA in the development of POCD. GA – General Anesthesia; RA – Regional
Anesthesia.

Fig. 4. Forest plot of the two studies which compare GA and combination anesthesia in the development of POCD. GA – General Anesthesia.
S74 S.E. Mason et al. / POCD After General and Regional Anesthesia

Fig. 5. Forest plot of the nine studies which compare GA and non-GA in the development of POCD. GA – General Anesthesia.

Comparison of GA with RA for POCD pooled estimate of the odds ratio is 1.34 (95% CI =
0.93–1.95), marginally non-significantly favoring non-
Two papers presented incidences for GA and RA [9, general anesthesia (Fig. 5). In total, 7 studies showed
50], with a pooled estimate of the odds ratio of mean group changes, with a significant finding report-
1.51 (95% CI = 0.87–2.64), non-significantly favor- ed in two [37,39]. The pattern of results from these 7
ing RA (Fig. 3). Three studies presented mean group papers does not seem to differ substantially from the
changes [37,41,43], with one [37] significantly favor- pattern observed in the studies which contributed to the
ing RA and two showing no difference between types meta-analysis.
of anesthesia. The rating scales used to assess cognitive dysfunc-
tion across all studies have been categorized in Ta-
Comparison of GA with combination anesthesia for ble 6. The 35 tests tended to be either global measures
(the most commonly used was the mini-mental state
POCD
examination), or recognized tests of specific cognitive
domains which are normally found as components of
Two papers presented an incidence of POCD after
comprehensive test batteries. The studies that did not
surgery under GA or combination (general and epidu-
present criteria derived definitions of POCD tended to
ral) anesthesia [53,55], with a pooled odds ratio of 1.3 employ many individual tests to cover domains of cog-
(95% CI = 0.65–2.60), non-significantly favoring com- nition including memory, executive function and cal-
bination anesthesia (Fig. 4). One paper presented mean culation.
group changes [42] which did not favor either method The adapted Jadad score giving an indication of
of anesthesia. methodological rigor varied from 0–5 across all studies
and on average where delirium was the principal out-
Comparison of GA with non-GA for POCD come, quality appeared higher in comparison to POCD
studies (p = 0.11). However there appeared no clear
Combining the 9 studies which presented inci- relationship between the adapted Jadad score of studies
dences of POCD after GA compared to non-GA, the and their findings for each outcome. Finally, Figs 6
 Table 6
The neuropsychological test batteries employed by the various authors assessing post-operative cognitive dysfunction, categorized by the domains of cognition that they test
Paper Global Memory Higher Other
function Verbal Verbal Visual Visual cognition Attention
recall recognition recall recognition executive
function
Anwer 2006 [37] WAIS
Asbjorn 1989 [38] Digit span, paired Paired Author’s
associates, free associates
recall, story recall
Bigler 1985 [45] AMT
Casati 2003 [46] MMSE
Chung 1987 [47] MMSE
Chung 1989 [48] MMSE
Cook 1986 [49] Author’s
Dahn 1999 [39] Digit span, Stroop Backwards spelling,
free recall calculation, controlled word
association
Dahn 2003 [40] Digit span, Stroop Backwards spelling,
free recall calculation, controlled word
association
Forster 1990 [50] MMSE
Ghoneim 1988 [41] Digit span, free Paired Card sorting, Digit symbol Reaction time.
recall, paired associates, Stroop substitution, symbol Motor: tapping test.
associates word cancellation
recognition
Pan 2006 [53] Paired associates, Paired Visual Digit symbol Pegboard for the favored
digit span forward associates retention substitution, trial and unflavored hand, mental
and backwards making A conorol
Rasmussen 2003 [9] Verbal learning Visual verbal Part C of concept Letter digit
S.E. Mason et al. / POCD After General and Regional Anesthesia

test learning test shifting substitution

Riis 1983 [42] Digit span, paired Paired Visual Picture Sorting test, Trail making,
associates, story associates gestalts, recognition block design symbol digit
recall, selective picture substitution
reminding recall
Scott 2001 [55] Author’s
Somprakit 2002 [43] MMSE
WAIS – Weschler Adult Intelligence Scale; AMT – Abbreviated Mental Test; MMSE – Mini-Mental State Examination; Author’s – authors own test or adaptation of a recognized test.
S75
S76 S.E. Mason et al. / POCD After General and Regional Anesthesia

Fig. 6. Funnel plot of sample size against the logarithm of the odds ratio for those studies who reported the incidence of post-operative delirium
in their patients. GA – General Anesthetic; PE – Pooled Estimate.

Fig. 7. Funnel plot of sample size against the logarithm of the odds ratio for those studies who reported the incidence of post-operative cognitive
dysfunction in their patients. GA – General Anesthetic; PE – Pooled Estimate.

and 7; though with small numbers of studies show little It was an objective to compare individual types of
evidence of publication bias. anesthesia in the development of POD, but as only five
studies were suitable for inclusion, it was required that
they were grouped into GA versus non-GA. The pooled
estimate of the odds ratio showed no impact of anesthe-
DISCUSSION
sia route on the incidence of POD. From review of indi-
vidual studies, it would appear that orthopaedic surgery
This study aimed to determine the impact of the route was more likely to favor general anesthesia, with the
of anesthetic delivery on the development of either POD opposite the case for abdominal surgery. The small
or the more restricted concept of POCD. number of studies in this part of the analysis though
 S.E. Mason et al. / POCD After General and Regional Anesthesia S77

limits any opportunity to draw inference. What was not ing robust outcome measures applied in the critical win-
observed though is important, namely that in terms of dow for POD and POCD namely between 3 and 7 days
delirium, the route of anesthetic would appear to have post operatively. These specific requirements along
little bearing. A larger sample of studies was available with the sensitive search strategy and hand-searching of
for review where POCD was the principal outcome. original and review article bibliographies should have
However, differences in how this was defined limited ensured that this review only incorporates the most rele-
the ability to draw any firm conclusions. Despite this, vant, high quality evidence. However, by strictly defin-
the meta-analysis on this occasion pointed towards re- ing the criteria for inclusion in this study, it has been
gional anesthesia being less likely to cause POCD than impossible to assess specific covariants due to a lack of
GA, however this was non-significant. Comparing this studies to use for metaregression analysis. The study
finding to the one for POD suggests either that the sub- could have been improved further if experts in the field
tle effects of anesthetic on cognition are measurable were contacted to locate relevant grey literature.
and apparent but insufficiently potent to cause delirium
or that the genesis of delirium is multi-factorial and in
RCT’s not manifest. This is not true for POCD work, CONCLUSION
where the impact of anesthetic route is more specific to
the genesis of POCD. This meta-analysis has demonstrated that rates of
The influence of other variables such as surgery type, delirium are unlikely to be influenced by the route of
age of sample, or duration of post operative observa- anesthesia. This is reassuring in some regards because
tion could not be measured statistically due to the high of the work linking delirium to increased morbidity and
degree of variability in these factors in a small sam- mortality post operatively as well as with regards to
ple of studies. Therefore, the ability to draw any con- the longer term risks of neurodegeneration. Perhaps
clusions regarding mapping anesthetic type onto sub- though this optimistic observation should be balanced
populations of patients is limited. against the likely risk our work demonstrated of gen-
Standardization of outcomes in this type of work erating POCD with GA when compared to anesthesia
would help the field progress, indeed with a burgeon- that has no central effect. The short, mid, and long term
ing interest in intervention studies to reduce the risk of consequences of POCD need qualifying and quantify-
POD and POCD, the use of a standardized cognition ing. However, there is an urgent need for a consensus
battery for POCD would be helpful. The validity and on the definition of POCD to aid both observational
reliability of the rating scales used for POD have helped work as well as measuring the impact of interventions
in this regard and similar scales for POCD are over- in this condition.
due. We found that 35 different tests were employed
for measuring cognitive decline, often with different
thresholds for impairment. Global measures such as DISCLOSURE STATEMENT
the MMSE were most popular probably as they are
fast and easy to perform, but they may be insensitive Authors’ disclosures available online (http://www.j-
to the dysfunction seen post-operatively. It has been alz.com/disclosures/view.php?id=576).
described that particularly tests of verbal memory are
sensitive for post-operative cognitive dysfunction [11],
but in a global assessment, the lack of impairment in Appendix 1
other domains may hide the verbal memory dysfunc-
tion. Similarly, studies which employ many individual The adapted Jadad score is as follows.
tests may have increased sensitivity, depending upon Randomization:
how one sets a threshold. In one study [9], the authors +1 if the study uses the word randomized, random
derived an incidence based on several tests and they in- allocation or similar
terestingly reported a higher rate of POCD as well as a +1 if the method of randomization is adequate e.g.,
greater different between the groups than other studies random number generation
in our sample. -1 if the method of randomization is inadequate e.g.,
The strengths and weaknesses of this study must be based on date of operation
assessed in order to understand the implications of the Blinding:
findings. We had a clear aim to only include RCTs us- +1 outcome assessor blinded
S78 S.E. Mason et al. / POCD After General and Regional Anesthesia

-1 outcome assessor clearly unblinded [12] Baxter MG, Chiba AA (1999) Cognitive functions of the basal
Outcome measure validity: forebrain. Curr Opin Neurobiol 9, 178-183.
[13] Koponen HJ (1999) Neurochemistry and delirium. Dement
+1 objective measures of cognitive performance are Geriatric Cogn Disord 10, 339-341.
clearly stated [14] Perry VH (2004) The influence of systemic inflammation on
-1 measures are not adequately stated such that repli- inflammation in the brain: implications for chronic neurode-
cation is difficult generative disease. Brain Behav Immunity 18, 407-413.
[15] Garden GA, Moller T (2006) Microglia biology in health and
Were both groups treated equally: disease. J Neuroimmune Pharmacol 1, 127-137.
+1 patient characteristics well documented and in- [16] Mukadam N, Ritchie CW, Sampson EL (2008) Cholinesterase
different between groups inhibitors for delirium: what is the evidence? Int Psychogeri-
-1 patient characteristics in each group are not re- atr 20, 209-218.
[17] Li QF, Zhu YS, Jiang H (2008) Isoflurane preconditioning
ported
activates HIF-1alpha, iNOS and Erk1/2 and protects against
oxygen-glucose deprivation neuronal injury. Brain Res 1245,
26-35.
REFERENCES [18] Barann M, Linden I, Witten S, Urban BW (2008) Molecular
actions of propofol on human 5-HT3A receptors: enhance-
[1] Cole MG, Primeau FJ (1993) Prognosis of delirium in elderly ment as well as inhibition by closely related phenol derivatives.
hospital patients. CMAJ 149, 41-46. Anesthesia Analgesia 106, 846-857, table of contents.
[2] Marcantonio ER, Goldman L, Mangione CM, Ludwig LE, [19] Hemmings HC, Jr., Yan W, Westphalen RI, Ryan TA (2005)
Muraca B, Haslauer CM, Donaldson MC, Whittemore AD, The general anesthetic isoflurane depresses synaptic vesicle
Sugarbaker DJ, Poss R (1994) A clinical prediction rule for exocytosis. Mol Pharmacol 67, 1591-1599.
delirium after elective noncardiac surgery. JAMA 271, 134- [20] Gomez RS, Guatimosim C (2003) Mechanism of action of
139. volatile anesthetics: involvement of intracellular calcium sig-
[3] Abildstrom H, Rasmussen LS, Rentowl P, Hanning CD, Ras- naling. Curr Drug Targets CNS Neurol Disord 2, 123-129.
mussen H, Kristensen PA, Moller JT (2000) Cognitive dys- [21] Fodale V, Santamaria LB (2003) The inhibition of central
function 1-2 years after non-cardiac surgery in the elderly. IS- nicotinic nAch receptors is the possible cause of prolonged
POCD group. International Study of Post-Operative Cognitive cognitive impairment after anesthesia. Anesthesia Analgesia
Dysfunction. Acta Anaesthesiol Scand 44, 1246-1251. 97, 1207; author reply 1207.
[4] Bickel H, Gradinger R, Kochs E, Forstl H (2008) High risk of [22] Pratico C, Quattrone D, Lucanto T, Amato A, Penna O, Rosci-
cognitive and functional decline after postoperative delirium. tano C, Fodale V (2005) Drugs of anesthesia acting on central
A three-year prospective study. Dement Geriatr Cogn Disord cholinergic system may cause post-operative cognitive dys-
26, 26-31. function and delirium. Med Hypotheses 65, 972-982.
[5] Yang FM, Marcantonio ER, Inouye SK, Kiely DK, Rudolph [23] Bryson GL, Wyand A (2006) Evidence-based clinical update:
JL, Fearing MA, Jones RN (2009) Phenomenological sub- general anesthesia and the risk of delirium and postoperative
types of delirium in older persons: patterns, prevalence, and cognitive dysfunction. Can J Anesthesia 53, 669-677.
prognosis. Psychosomatics 50, 248-254. [24] Parker MJ, Handoll HH, Griffiths R (2001) Anesthesia for hip
[6] Bekker AY, Weeks EJ (2003) Cognitive function after anesthe- fracture surgery in adults. Cochrane Database Syst Rev (4),
sia in the elderly. Best practice & research.Clin Anaesthesiol CD000521.
17, 259-272. [25] Wu CL, Hsu W, Richman JM, Raja SN (2004) Postoperative
[7] Bruce AJ, Ritchie CW, Blizard R, Lai R, Raven P (2007) The cognitive function as an outcome of regional anesthesia and
incidence of delirium associated with orthopedic surgery: a analgesia. Reg Anesth Pain Med 29, 257-268.
meta-analytic review. Int Psychogeriatr 19, 197-214. [26] Benoit AG, Campbell BI, Tanner JR, Staley JD, Wallbridge
[8] Newman S, Stygall J, Hirani S, Shaefi S, Maze M (2007) HR, Biehl DR, Bradley BD, Louridas G, Guzman RP, Fromm
Postoperative cognitive dysfunction after noncardiac surgery: RA (2005) Risk factors and prevalence of perioperative cogni-
a systematic review. Anesthesiology 106, 572-590. tive dysfunction in abdominal aneurysm patients. J Vasc Surg
[9] Rasmussen LS, Johnson T, Kuipers HM, Kristensen D, Siers- 42, 884-890.
ma VD, Vila P, Jolles J, Papaioannou A, Abildstrom H, Silver- [27] Bracco D, Noiseux N, Dubois MJ, Prieto I, Basile F, Olivier JF,
stein JH, Bonal JA, Raeder J, Nielsen IK, Korttila K, Munoz Hemmerling T, PeriCarg Group M (2007) Epidural anesthesia
L, Dodds C, Hanning CD, Moller JT, Investigators I (2003) improves outcome and resource use in cardiac surgery: a
Does anesthesia cause postoperative cognitive dysfunction? single-center study of a 1293-patient cohort. Heart Surg Forum
A randomized study of regional versus general anesthesia in 10, E449-458.
438 elderly patients. Acta Anaesthesiol Scand 47, 260-266. [28] Campbell DN, Lim M, Muir MK, O’Sullivan G, Falcon M,
[10] Moller JT, Cluitmans P, Rasmussen LS, Houx P, Rasmussen H, Fison P, Woods R (1993) A prospective randomized study of
Canet J, Rabbitt P, Jolles J, Larsen K, Hanning CD, Langeron local versus general anesthesia for cataract surgery. Anesthesia
O, Johnson T, Lauven PM, Kristensen PA, Biedler A, van 48, 422-428.
Beem H, Fraidakis O, Silverstein JH, Beneken JE, Gravenstein [29] Crul BJ, Hulstijn W, Burger IC (1992) Influence of the type
JS (1998) Long-term postoperative cognitive dysfunction in of anesthesia on post-operative subjective physical well-being
the elderly ISPOCD1 study. ISPOCD investigators. Interna- and mental function in elderly patients. Acta Anaesthesiol
tional Study of Post-Operative Cognitive Dysfunction. Lancet Scand 36, 615-620.
351, 857-861. [30] Haan J, van Kleef JW, Bloem BR, Zwartendijk J, Lanser JB,
[11] Deiner S, Silverstein JH (2009) Postoperative delirium and Brand R, van der Does IG, Krul EJ, Elshove HM, Moll AC
cognitive dysfunction. Brit J Anesthesia 103 Suppl 1, i41-46. (1991) Cognitive function after spinal or general anesthesia
 S.E. Mason et al. / POCD After General and Regional Anesthesia S79

for transurethral prostatectomy in elderly men. J Am Geriatr Assoc Thailand 85 Suppl 3, S875-883.
Soc 39, 596-600. [44] Berggren D, Gustafson Y, Eriksson B, Bucht G, Hansson LI,
[31] Handley GH, Silbert BS, Mooney PH, Schweitzer SA, Allen Reiz S, Winblad B (1987) Postoperative confusion after anes-
NB (1997) Combined general and epidural anesthesia versus thesia in elderly patients with femoral neck fractures. Anes-
general anesthesia for major abdominal surgery: postanesthe- thesia Analgesia 66, 497-504.
sia recovery characteristics. Reg Anesthesia 22, 435-441. [45] Bigler D, Adelhoj B, Petring OU, Pederson NO, Busch P,
[32] Jones MJ, Piggott SE, Vaughan RS, Bayer AJ, Newcombe Kalhke P (1985) Mental function and morbidity after acute
RG, Twining TC, Pathy J, Rosen M (1990) Cognitive and hip surgery during spinal and general anesthesia. Anesthesia
functional competence after anesthesia in patients aged over 40, 672-676.
60: controlled trial of general and regional anesthesia for [46] Casati A, Aldegheri G, Vinciguerra E, Marsan A, Fraschini G,
elective hip or knee replacement. BMJ 300, 1683-1687. Torri G (2003) Randomized comparison between sevoflurane
[33] Marcantonio ER, Goldman L, Orav EJ, Cook EF, Lee TH anesthesia and unilateral spinal anesthesia in elderly patients
(1998) The association of intraoperative factors with the de- undergoing orthopaedic surgery. Eur J Anaesthesiol 20, 640-
velopment of postoperative delirium. Am J Med 105, 380-384. 646.
[34] Nielson WR, Gelb AW, Casey JE, Penny FJ, Merchant RN, [47] Chung F, Meier R, Lautenschlager E, Carmichael FJ, Chung
Manninen PH (1990) Long-term cognitive and social sequelae A (1987) General or spinal anesthesia: which is better in the
of general versus regional anesthesia during arthroplasty in the elderly? Anesthesiology 67, 422-427.
elderly. Anesthesiology 73, 1103-1109. [48] Chung FF, Chung A, Meier RH, Lautenschlaeger E, Seyone
[35] Ryhanen P, Helkala EL, Ihalainen O, Hollmen A, Rantakyla S, C (1989) Comparison of perioperative mental function after
Merila M, Tuohino V, Pietarila M, Horttonen L (1978) Effects general anesthesia and spinal anesthesia with intravenous se-
of anesthesia on the psychological function of patients. Ann dation. Can J Anesthesia 36, 382-387.
Clin Res 10, 318-322. [49] Cook PT, Davies MJ, Cronin KD, Moran P (1986) A prospec-
[36] Yoshida S, Nishikawa K, Shimodate Y, Igarashi M, Namiki A tive randomized trial comparing spinal anesthesia using hy-
(2008) Efficacy of spinal anesthesia with small-dose lidocaine perbaric cinchocaine with general anesthesia for lower limb
in ambulatory prostate biopsy procedures for elderly patients vascular surgery. Anesthesia Intensive Care 14, 373-380.
: comparison with propofol anesthesia. Masui 57, 605-609. [50] Forster A, Altenburger H, Gamulin Z (1990) Effects of anes-
[37] Anwer HM, Swelem SE, el-Sheshai A, Moustafa AA (2006) thesia on higher brain functions in the elderly. Presse Med 19,
Postoperative cognitive dysfunction in adult and elderly 1577-1581.
patients–general anesthesia vs subarachnoid or epidural anal- [51] Kudoh A, Takase H, Takazawa T (2004) A comparison of
gesia. Middle East J Anesthesiol18, 1123-1138. anesthetic quality in propofol-spinal anesthesia and propofol-
[38] Asbjorn J, Jakobsen BW, Pilegaard HK, Blom L, Ostergaard fentanyl anesthesia for total knee arthroplasty in elderly pa-
A, Brandt MR (1989) Mental function in elderly men after tients. J Clin Anesthesia 16, 405-410.
surgery during epidural analgesia. Acta Anaesthesiol Scand [52] Nishikawa K, Kimura S, Shimodate Y, Igarashi M, Namiki
33, 369-373. A (2007) A comparison of intravenous-based and epidural-
[39] Dahn J, Oster M, Moltner A, Wohrle C, Ratzer-Frey A, van based techniques for anesthesia and postoperative analgesia in
Ackern K, Holzl R, Segiet W (1999) Anesthesia in geriatric elderly patients undergoing laparoscopic cholecystectomy. J
patients. The determination of physiological variables for cog- Anesthesia 21, 1-6.
nitive function in geriatric patients after regional or general [53] Pan LF, Wang DX, Li J (2006) Effects of different methods
anesthesia. Der Anaesthesist 48, 379-386. of anesthesia and analgesia on early postoperative cognitive
[40] Dahn J, Eckert S, Oster M, Suselbeck T, Ellinger K, van Ack- dysfunction after non-cardiac surgery in the elderly. Beijing
ern K, Daffertshofer M, Segiet W (2003) Cognitive functions Da Xue Xue Bao38, 510-514.
and cerebral oxygenation of older patients after general and [54] Papaioannou A, Fraidakis O, Michaloudis D, Balalis C, Ask-
regional anesthesia. Der Anaesthesist 52, 596-605. itopoulou H (2005) The impact of the type of anesthesia on
[41] Ghoneim MM, Hinrichs JV, O’Hara MW, Mehta MP, Pathak cognitive status and delirium during the first postoperative
D, Kumar V, Clark CR (1988) Comparison of psychologic days in elderly patients. Eur J Anaesthesiol 22, 492-499.
and cognitive functions after general or regional anesthesia. [55] Scott NB, Turfrey DJ, Ray DA, Nzewi O, Sutcliffe NP, Lal
Anesthesiology 69, 507-515. AB, Norrie J, Nagels WJ, Ramayya GP (2001) A prospective
[42] Riis J, Lomholt B, Haxholdt O, Kehlet H, Valentin N, randomized study of the potential benefits of thoracic epidu-
Danielsen U, Dyrberg V (1983) Immediate and long-term ral anesthesia and analgesia in patients undergoing coronary
mental recovery from general versus epidural anesthesia in artery bypass grafting. Anesthesia Analgesia 93, 528-535.
elderly patients. Acta Anaesthesiol Scand 27, 44-49. [56] Williams-Russo P, Sharrock NE, Mattis S, Szatrowski TP,
[43] Somprakit P, Lertakyamanee J, Satraratanamai C, Wanick- Charlson ME (1995) Cognitive effects after epidural vs gen-
samban S, Silapadech A, Chainchop P, Muangsuay V (2002) eral anesthesia in older adults. A randomized trial. JAMA 274,
Mental state change after general and regional anesthesia in 44-50
adults and elderly patients, a randomized clinical trial. J Med