The aim of this study was to assess whether anaesthesia has an impact on the development of post-operative cognitive dysfunction in hip fracture patients. This is part one of a two-part article.
First published May 2018, updated July 2022
It is estimated that there are 70,000-75,000 fractured neck of femurs sustained in the UK every year. Around 10% of these patients will die within one month and around one third will die within one year.1,2
Alongside an increased risk of mortality and physical morbidity, there are a large number of patients every year that are admitted with a fractured neck of femur that have an intercurrent diagnosis of cognitive impairment or dementia. Indeed, dementia is a recognised risk factor for hip fracture3 and it has been shown that cognitive limitations increases the risk of morbidity and mortality in such patients who fracture their hip.4,5
It would therefore seem logical to try and limit any potential iatrogenic causes that may induce or worsen cognitive impairment. Berggren et al described that post-operative confusion was as prevalent as 44% of elderly patients after surgery for fractured neck of femur,6 with other studies suggesting the incidence of post-operative cognitive impairment in elderly patients for any surgery in general is around 37%.7 However, the incidence may be even higher with some studies suggesting that post-operative cognitive dysfunction (PCOD) is underdiagnosed by nurses and clinicians.7
Despite this association it still remains unclear if anaesthesia plays a role in the onset of cognitive impairment8 in post-operative patients. There is some literature available supporting the notion that general anaesthesia can be a predisposing factor for post-operative cognitive impairment,9 with a metaanalysis by Mason et al also advocating use of regional anaesthesia over general anaesthesia due to this potential increased risk of PCOD, but they identified no obvious link with development of post-operative delirium.10
Anaesthetists may base their choice of anaesthesia for fractured neck of femur surgery, partially, on the patients cognitive state, avoiding general anaesthetic where possible in patients with impairment. However, the literature on the matter appears at times contradictory with multiple studies stating that the use of general anaesthesia has, in fact, no effect on PCOD.11-15 One paper has even stated that, by reducing the ‘surgical stress’ of an operation by undergoing general versus regional anaesthesia, that general anaesthesia may actually provide a neuroprotective effect in the post-operative period.16
A systematic review performed by Davis et al found that three studies they analysed showed a difference in cognitive function between regional anaesthesia, whilst 13 did not.17
There are multiple potential mechanisms by which general anaesthesia has been thought to cause or worsen cognitive impairment, including; oligomerisation and cytotoxicity of amyloid peptides18,19 increasing tau protein phosphorylation20 and altering transmission of neurotransmitters.21,22
In addition, studies have looked at whether anaesthetic type has an impact on clinical outcomes, including length of stay and mortality, with no differences found between differing types of anaesthesia.23,24
The aim of this study was therefore to assess whether anaesthesia, specifically the use of ‘general anaesthesia’ versus ‘no general anaesthesia’ had an impact on cognitive dysfunction given the inconsistent nature of the current literature.
A retrospective cohort study was conducted. Data was analysed for patients, over a five-year period 2011 to 2016, who underwent surgery for a fractured neck of femur. Data was collected from the National Hip Fracture Database for all patients who underwent surgery for a fractured neck of femur at a single hospital.
The patient demographics were recorded (including age and sex) along with; American Society of Anesthesiologists (ASA) grading, operation type and pre and post-operative abbreviated mental test score (AMTS). As it was not possible to obtain information on all patients comorbities, to ensure patients were matched for perioperative risk, ASA was used as a surrogate.
From the pre and post-operative AMTS, a change in AMTS was calculated and it was this that was utilised for a statistical analysis to account for patients having varying baseline AMTS values.
AMTS was used as it is the only objective measurement currently taken both pre and post operatively. Whilst the authors realise it would be beneficial to have adequately validated delirium/ cognitive testing pre and post operatively, this data is currently not available.
Patients were excluded if they had an incomplete AMTS either pre or post-operative or if there was no information on the anaesthetic type that they had received. We also looked to see if the type of operation the patient was likely to receive would have an impact on the anaesthetic they received.
Statistical analysis was performed, utilising a student T-Test for continuous dependant variables and Chi squared test for categorical dependent variables. Where non parametric, ordinal analysis was required the Kruskal-Wallis test was used. Due to the non parametric nature of some of the data and inability to effectively utilise confidence intervals, P values were utilised for determining statistical significance. A P value of <0.05 was considered statistically significant.
The hypothesis’ that we looked to test were; ‘there is no difference in change in AMTS for those having a general anaesthetic vs those not having a general anaesthetic’, and ‘fracture type and operation type results in no difference in choice of anaesthetic type’.
Within the specified data collection period, 1,527 patients underwent surgery for a fractured neck of femur. After excluding those who did not have a pre and post-operative AMTS there were 1,367 patients, of whom three did not have information on anaesthetic type and again were excluded leaving a cohort of 1,364.
The 1,364 patients were grouped based on whether they had received a general anaesthetic or not, with the ‘general anaesthetic group’ including those who had; a general anaesthetic only, general anaesthetic and block, general anaesthetic and spinal, general anaesthetic and spinal and block. The ‘no general anaesthetic group’ included those who had a spinal or those who had a spinal and a block.
In the ‘general anaesthetic’ group there were 870 patients and within the ‘no general anaesthetic’ group there were 494 patients.
Age ranged from 36–104 years in the ‘general anaesthetic’ group and 55–102 years in the ‘no general anaesthetic’ group. There was no significant difference in age between both groups. In the ‘general anaesthetic’ group there were 207 males and 663 females and in the ‘no general anaesthetic’ group there were 143 males and 351 females. Again, there was no statistically significant difference for sex between both groups. ASA information was available for 802 patients in the general anaesthetic group and for 474 (1,276 total) in the no general anaesthetic group. There was no statistically significant difference in ASA grade between groups. p=0.098. (Table 1)
|TABLE 1 – COMPARISON OF GROUPS BY AGE, SEX AND ASA|
|Demographic detail||(General Anaesthetic)||(No General anaesthetic)||P value|
|Age(1)||36 – 104||55 - 102||P=0.282|
|Sex(2)||207 male / 663 female||143 male / 351 female||P=0.362|
|ASA*||1 – 15
2 – 154
3 – 498
4 – 135
5 – 0
|1 – 4
2 – 84
3 – 288
4 – 96
5 – 2
|(1) Analysed using student T test
(2) Analysed using chi squared
*69 patients did not have relevant information on ASA in GA group and 19 did not in SA group (hence 802 GA/474 No GA)
Using the pre and post-operative AMTS data for both sets of patients a ‘change in AMTS was calculated’. This ranged from a deterioration in 9 points to an improvement in 9 points for the ‘general anaesthetic group’ and a deterioration of 10 points to an improvement of 8 in the ‘no general anaesthetic’ group. (Table 2)
|TABLE 2 – CHANGE IN AMTS|
|Association between anaesthetic type and change in AMTS(1)|
|Anaesthetic||Number||Range of change in AMTS (between pre and post-operative)|
|General anaesthetic||870||(deterioration) -9 to 9 (improvement)|
|No general anaethetic||494||(deterioration) -10 to 8 (improvement)|
|(1) Analysed using student T test|
Alongside the impact of anaesthetic on change in AMTS, the authors also looked to ascertain whether the type of operation that the patient was undergoing influenced the choice of anaesthetic.
Operation type was divided into; hemiarthroplasty (uni/bipolar—both cemented and uncemented), total hip replacement (both cemented and uncemented), cannulated screws, DHS/SHS/CHS—categorised as DHS and intramedullary nail (long/short).
- Of the 633 who had a hemiarthroplasty 390 underwent a general anaesthetic and 243 did not undergo a general anaesthetic—expected values from chi squared analysis were 403.87 versus 229.13, hence, patients, were more likely to receive ‘no general anaesthetic’.
- Of the 46 total hip replacements, 22 underwent a general anaesthetic and 24 did not undergo a general anaesthetic—expected values from chi squared analysis were 29.35 versus 16.65—hence, patients were more likely to receive ‘no general anaesthetic’.
- Of the 34 patients who had cannulated hip screws 22 underwent a general anaesthetic and 12 did not undergo a general anaesthetic—expected values from chi squared analysis were 21.69 versus 12.31—there appears to be no bias towards one anaesthetic, over another, for this group.
- Of the 524 DHS/SHS/CHS performed 344 patients underwent a general anaesthetic and 180 did not undergo a general anaesthetic—expected values from chi squared analysis were 334.33 versus 189.67, hence, patients were more likely to receive a ‘general anaesthetic’.
- Out of the 125 patients who underwent an intramedullary nail, 91 patients underwent a general anaesthetic and 34 did not undergo a general anaesthetic—expected values from chi squared analysis were 79.75 versus 45.25, hence, patients were more likely to receive a ‘general anaesthetic.’
Despite similar expected values for some operation subtypes, overall, the impact that operation type had, on choice of anaesthetic, was statistically significant with p = 0.021. (Table 3)
|TABLE 3 – ANAESTHETIC GIVEN FOR EACH OPERATION TYPE|
|General anaesthetic||No General anaesthetic|
|X2 = 11.567, df = 4, X2/df = 2.89, p=0.0209
*1 patient in each group was listed as “other” – operation not known expected values are displayed in italics
individual X2 values are displayed in (parentheses)
Given that there was a statistically significant difference in whether patients received a general anaesthetic or no general anaesthetic based on their operation type, further analysis was performed to aim to determine the reason for this difference. We, further, looked to test the hypothesis that either the patients ASA or the patients cognitive function influenced operation type. As an anaesthetist would not know the change in AMTS (as it requires the post-operative AMTS) to justify a choice of anaesthetic, we postulated that this statistical difference may be based on the patients preoperative AMTS. A Kruskal-Wallis calculation was performed, looking to see if pre-operative AMTS was significantly different between operation groups.
Mean Rank value for hemiarthroplasty was 656.3 and this was the lowest out of the operative groups, suggesting lower pre-operative AMTS and hence more cognitive impairment, in this group than the other operative groups. The mean rank values for the other operative groups were; THR: 896.0, Cannulated screws: 861.3, DHS: 675.1, IM nail 732.1. The difference in these groups was statistically significant with p=0.000021. As patients in the THR group would not have their anaesthetic choice based on the presence of pre-existing cognitive impairment (as it is a pre requisite that this is not present—reinforced by mean of ranks f 896.0) a sub analysis was performed where the THR group was removed. This still showed a statistically significant difference between groups p=0.006.
Looking at the influence ASA grading had on operative choice, there was no statistically significant difference in which operation the patient would receive based on their ASA (p=0.15).
This study would support the hypothesis that type of anaesthesia does not significantly effect PCOD, when considering this with respect to change in AMTS. Despite this, patients appear to be having their anaesthetic type influenced by the presence of pre-existing cognitive impairment.
Mark Sohatee, County Durham and Darlington NHS Foundation Trust, University Hospital of North Durham
Hannah Wilkinson, County Durham and Darlington NHS Foundation Trust, University Hospital of North Durham
Andrew Gower, County Durham and Darlington NHS Foundation Trust, University Hospital of North Durham
Conflict of interest: none declared