Cognitive impairment and pain relief following hip fractures: a case control study

Abstract

Hip fractures represent a significant workload of both emergency and orthopaedic departments within the National Health Service (NHS). Pain relief is key in treating hip fractures as highlighted by both National Institute of Clinical Excellence (NICE) and British Orthopaedic Association Standards for Trauma (BOAST) guidelines. However, the literature shows that patients with cognitive impairment tend to have inconsistent pain management, leading to worse outcomes. We conducted a case–control study looking at 296 patients who presented with hip fractures to a major trauma centre between 1 December 2019 and 30 May 2020. Cognition was assessed using pre-recorded Abbreviated Mental Test Scores (AMTS). There was no significant difference between pain relief provided to patients with or without cognitive impairment in both the pre-hospital (p = 0.208) and Accident & Emergency (A&E) (p = 0.154) setting. A larger proportion of patients in A&E did not receive any pain relief (18.6% versus 42.2%). Pre-hospital, the higher the pain score, the stronger the analgesia given (R = 0.435, p = 0.000). This relationship was present in both the cognitively impaired (R = 0.572, p = 0.000) and cognitively intact groups (R = 0.390 p = 0.000). Strength of analgesia and pain scores did not correlate in A&E (R = 0.014, p = 0.826). Cognition did not impact the time to analgesia both pre-hospital (p = 0.291) and in A&E (p = 0.332); however, patients waited significantly longer to receive pain relief in A&E (29.61 minutes versus 150.28 minutes). Fascia-iliaca blocks were administered to 58.4% of the cohort, with no significant difference noted between cognition status. Overall, cognition does not impact pain management both pre-hospital and in A&E. There is still room for improvement, particularly in the assessment of pain in the cognitively impaired. A possible solution is the utilisation of the Bolton Pain Assessment Tool, a validated pain assessment tool for the cognitively impaired that has been utilised in the trauma setting with good effect.

Background

Hip fractures make up a significant workload of emergency and orthopaedic departments in the United Kingdom. An estimated 65,000 patients with hip fractures present annually to UK emergency departments and this number is set to grow to approximately 101,000 in 2020.^1,2 This represents a significant cost to the National Health Service (NHS); estimated at around £2 billion, encompassing both medical and social care. ³ Hip fractures are found more commonly in the elderly, with 86% of fractures occurring in patients over 65. ⁴ This is due to the shared risk factors between several chronic disease states and hip fractures. Fifty-four per cent of individuals aged 65–69 years have two or more chronic conditions; this rises to 73% in individuals over the age of 85, closely mimicking the rise in incidence of hip fractures with age. ⁵

Of interest to us is the comorbidity of cognitive impairment, either pre-existing or de-novo following a hip fracture. Data show that 25% of the population with a hip fracture are defined as having at least moderate cognitive impairment (Abbreviated Mental Test Score (AMTS) < 7). ⁶ This can manifest itself as a long-term decline in cognition such as Alzheimer’s dementia; alternatively, it can be an acute delirium precipitated by the fracture itself. In other individuals, there is an acute worsening of cognition on a background of long-standing cognitive decline. Cognitive impairment (CI) has been shown to increase mortality when compared with age- and gender-matched individuals without CI. ⁷ Hip fractures and CI have a close relationship as people with long-standing CI such as Alzheimer’s dementia have a 2.7-fold higher risk of sustaining a hip fracture. ⁸ On the contrary, individuals who undergo surgical fixation of their hip fracture have around a 4–53% risk of developing a post-operative delirium. ⁹ This illustrates the complex interplay between the shared risk factors and aetiology of CI and hip fractures, highlighting the need for a multi-disciplinary approach in their management.

A key tenet highlighted in the National Institute of Clinical Excellence (NICE) and British Orthopaedic Association Standards for Trauma (BOAST) guidance is the adequate provision of pain relief for hip fractures. ³ Traditionally, individuals who have a degree of CI have always fared worse when it comes to adequate pain relief. ⁹ The gold standard of pain assessment is self-reported; a medium those with CI may not be able to navigate. The majority of healthcare professionals tend to utilise numerical rating scale (NRS) to assess pain; however, this method does not work for 43% of patients with moderate CI and 100% of patients with severe CI. ¹⁰ Retrospective analysis of hospital records show that patients with CI are significantly more likely to not have a documentation of their pain score. ¹¹ Pre-hospital analgesia is also significantly lacking in patients with CI, with 45% of patients not receiving any analgesia for their hip fracture as compared with 8% of their non-CI counterparts. ¹² Patients with CI wait longer to receive analgesia in Accident & Emergency (A&E) and are significantly more likely to receive weaker analgesia than their cognitive intact counterparts. ¹² These data are replicated in patients with CI who present with a long bone fracture; the results show significant increased waiting times to receive analgesia within A&E. ¹³ The above data make for poor reading, with the consensus being people with CI have significantly undertreated pain secondary to their hip fracture. This exposes them to the sequelae of untreated pain, tachycardia, increased blood pressure, increased myocardial oxygen demand and metabolic disturbances. ¹⁴ The consequences of untreated pain lead to longer length of stay, worse functional outcomes and greater risk of developing delirium.^15,16

Aims

We set out to analyse the prescribing practices in the pre-hospital setting and A&E of a major trauma centre. Our primary aim was to identify whether cognition had any impact on the quality and type of analgesia patients received for their hip fractures. Our secondary aims were to determine whether cognition had any impact on the strength of analgesia received and how long it took to administer analgesia.

Methods

All patients presenting with a hip fracture between 1 December 2019 and 30 May 2020 were eligible for inclusion. Patients were identified retrospectively through the local fragility fracture database. Data were collected and formatted in Excel© from several sources. Data were collected from ambulance documentation, emergency department records and the local hip fracture proforma. This information was collected retrospectively and recorded anonymously. Exclusion criteria were as follows:

• Fracture pattern other than hip fractures;

• Transfers from other hospitals/not presented via ambulance;

• Significant missing data fields (AMTS, A&E prescription chart);

• Taken analgesia prior to ambulance arrival (as this may impact what they received with the paramedic crew).

Parameters recorded were as follows:

• Pain score;

• AMTS;

• Type of analgesia;

• Time to analgesia;

• Fascia-iliaca block (FIB) given or not.

The above data were recorded both in the pre-hospital setting and A&E. An AMTS of < 7 was used as a cutoff for CI. ¹⁷ Analgesia given was divided into distinct categories as per the World Health Organization (WHO) pain ladder ¹² and are presented in Table 1.

Table 1.

Categorical breakdown of analgesia provided as per WHO pain ladder.

No analgesia given	Entonox	WHO Level 1	WHO Level 2	WHO Level 3
		Paracetamol Ibuprofen Aspirin	Dihydrocodeine Codeine phosphate Co-codamol Tramadol	Morphine IV Oral Morphine Fentanyl

WHO: World Health Organization.

Patients were categorised by the strongest analgesia they received. If multiple doses were given of the same type of analgesia, the highest dose was recorded. Statistical analysis was performed using the IBM® SPSS® Statistics software. Means testing between cohorts was done using the Mann–Whitney U test. Chi-square analysis was used to assess significance between categorical variables. Spearman’s rank correlation coefficient was used to quantify if a greater pain score resulted in stronger analgesia, as well as to assess if AMTS had any impact on the time waited to receive analgesia. Kruskall–Wallis test with post hoc analysis (Dunn’s test) was conducted to look for significant correlation between pain score and the strength of analgesia given. Significance was set at p ˂ 0.05.

Results

Our search revealed 376 records of hip fractures in the study period. After the removal of ineligible records, a total of 296 patients were included in the final analysis. There were 85 patients with an AMTS of < 7, with 211 patients with an AMTS ⩾ 7. The demographics and fracture pattern are summarised in Table 2.

Table 2.

Breakdown of demographics and fracture pattern by AMTS.

	AMTS ⩾ 7	AMTS < 7
Age	80.55 (± 10.62)	87.47 (± 7.87)
Male	61	26
Female	150	59
Intracapsular	118	46
Extracapsular	92	39

AMTS: Abbreviated Mental Test Score.

Patients with an AMTS < 7 are significantly older than their cognitively intact counterparts (Mann–Whitney U = 5529, Z = −5.164, p < 0.001). Furthermore, both groups show a greater proportion of women sustaining hip fractures. There is no significant difference in the gender ratio (χ² = 0.0212, df = 1, p = 0.884) or fracture pattern observed between the two groups (χ² = 0.1053, df = 1, p = 0.746).

Pre-hospital analysis

Our analysis of pre-hospital analgesia revealed no significant difference in the pain relief patients received regardless of their cognitive status (p = 0.208) (Figure 1). The most prescribed analgesics were WHO level 3 opioids, with Entonox being the least commonly used. Table 3 highlights the breakdown of analgesia received by AMTS.

Figure 1.

Analgesia given grouped into distinct categories, comparing prescribing practices between the cognitively impaired and the cognitively intact. Chi-square analysis reveals no statistically significant difference between the cognitively impaired or intact (χ² = 5.882, df = 4, p = 0.208).

Table 3.

Number of patients receiving levels of analgesia by AMTS (Pre-hospital).

	None	Entonox	WHO Level 1	WHO Level 2	WHO Level 3
AMTS < 7	20	3	20	5	37
AMTS ⩾ 7	35	19	38	19	100

AMTS: Abbreviated Mental Test Score; WHO: World Health Organization.

We then proceeded to test the hypothesis that the higher your pain score is, the more likely you are to be given stronger analgesia. This was done across the pre-hospital cohort and then individually broken down between the cognitively impaired and the cognitively intact. Figure 2 shows that there is a significant trend between the pain score and level of analgesia given; patients with a higher pain score were significantly more likely to receive stronger analgesia (p = 0.000).

Figure 2.

The graph above shows that the patients who received stronger analgesia tended to have higher mean pain scores. Kruskall–Wallis with post hoc analysis (Dunn’s test) showed a significant difference between the groups (χ² = 52.802, df = 4, p = 0.000). Spearman’s rank correlation also shows a significant trend (R = 0.435, p = 0.000); as pain score increases, so does the level of analgesia given.

Our analysis then proceeded to see if this relationship was consistent across the cognitively impaired and the cognitively intact. Figure 3 shows that cognition did not make a significant difference to the trends noted. Both groups received analgesia in accordance with their pain score.

Figure 3.

When the mean pain score is plotted against analgesia for both impaired versus intact cohorts, we see significant trends emerge. Kruskall–Wallis with post hoc analysis (Dunn’s test) shows significant results for both the cognitively impaired (χ² = 20.570, df = 4, p = 0.000) and the cognitively intact (χ² = 33.979, df = 4, p = 0.000). Spearman’s rank correlation for cognitively intact (R = 0.390, p = 0.000) and cognitively impaired (R = 0.572, p = 0.000) both respectively show significant trends. As the pain score increases, so does the level of analgesia given; cognition has no impact on this.

Accident and emergency analysis

As with pre-hospital analgesia, we found that cognition did not impact the analgesia given to patients who were admitted with a hip fracture (p = 0.154) (Figure 4). The most common group of analgesics prescribed were WHO Level 3 opioids. However, a much larger proportion of patients did not receive any analgesia as compared with pre-hospital numbers. Table 4 highlights the breakdown of analgesia given by AMTS.

Figure 4.

Analgesia given grouped into distinct categories, comparing the prescribing practices between the cognitively impaired and the cognitively intact. Chi-square analysis reveals no statistically significant difference between the cognitively impaired or intact (χ² = 5.257, df = 3, p = 0.154).

Table 4.

Number of patients receiving levels of analgesia by AMTS.

	None	WHO Level 1	WHO Level 2	WHO Level 3
AMTS < 7	38	12	10	25
AMTS ⩾ 7	87	22	14	88

AMTS: Abbreviated Mental Test Score; WHO: World Health Organisation.

We then looked at the correlation between pain score and level of analgesia given, first looking at the whole cohort, then breaking it down by cognitive status. Figure 5 shows an interesting trend line with a significant Kruskall–Wallis result (p = 0.026), however with an insignificant Spearman’s Rank correlation (p = 0.826). Although a statistically significant result, the hypothesis that the higher your pain score, the stronger the analgesia you receive is not true in this situation, as patients who received no analgesia had a higher pain score than those receiving WHO Level 1 analgesia.

Figure 5.

The graph above shows that the patients who received stronger analgesia tended to have higher mean pain scores. Kruskall–Wallis with post hoc analysis (Dunn’s test) showed a significant difference between the groups (χ² = 9.293, df = 4, p = 0.026). Spearman’s rank correlation, however, did not show a significant correlation (R = 0.014, p = 0.826).

This trend was then broken down by cognition to see if that influenced the analgesia these patients received. Figure 6 shows significance when testing the cognitively intact group (p = 0.006), with the cognitively impaired group showing insignificant trends. Post hoc analysis of the cognitively intact group reveals a significant difference in pain scores between the None and WHO Level 1 groups (p = 0.013), with people receiving no pain relief having significantly higher pain scores. This likely explains the significant difference picked up by the Kruskall–Wallis test. Spearman’s rank correlation shows no significant correlation between pain score and level of analgesia given, regardless of cognition.

Figure 6.

When the mean pain score is plotted against analgesia given across cognitive status, no clear significant trend emerges. Kruskall–Wallis test with post hoc analysis (Dunn’s test) shows insignificant results for the cognitively impaired (χ² = 4.303, df = 3, p = 0.231); and significant difference between groups for the cognitively intact (χ² = 12.526, df = 3, p = 0.006). Spearman’s rank correlation for cognitively intact (R = −0.054, p = 0.462) and cognitively impaired (R = 0.223, p = 0.065) both respectively show insignificant trends. The level of analgesia given does not significantly correlate with the pain score of the patient.

Time till analgesia analysis

On average, patients waited 29.61 minutes (± 15.478) to receive analgesia pre-hospital and 150.28 minutes (± 106.524) to receive analgesia in A&E. As shown in Figure 7, cognition did not impact time waited till analgesia either in pre-hospital (p = 0.291) or in the emergency department (p = 0.332).

Figure 7.

(a) Both cohorts show cognition has no impact on the time till analgesia. Mann–Whitney test for pre-hospital cohort (U = 5152.5, Z = −1.057, p = 0.291) and for the emergency department cohort (U = 2676, Z = −0.971, p = 0.332) are both insignificant. However, there is a significantly greater wait to receiving analgesia in the emergency department (U = 1652.0, Z = −15.947, p = 0.000). (b) Scatter plot of time to analgesia and AMTS was plotted to assess for any correlation. Spearman’s rank correlation for pre-hospital cohort (R = −0.101; p = 0.118) and the emergency department cohort (R = −0.111; p = 0.148) did not show any significant correlation.

Further analysis

We also looked at the presence of pain scores in the documentation to see if pain scores were less frequently recorded in the cognitively impaired cohort. Pre-hospital pain scores did not show a significant difference in the presence of valid recorded pain scores between the cognitively impaired and the cognitively intact (χ² = 2.9923, df = 1, p = 0.084). However, in hospital, there was a greater chance of not having a pain score recorded if the patient was cognitively impaired (χ² = 4.3594, df = 1, p = 0.037).

Our department has a policy of offering all hip fractures who meet the inclusion criteria a FIB. Table 5 shows the breakdown of FIBs versus AMTS. Cognition had no impact on whether the patient received a FIB or not (χ² = 0.3297, df = 1, p = 0.566) (patients who refused were not included in this analysis). Of the 125 patients in A&E who did not receive any analgesia, 77 were given a FIB. Cognition was not a significant factor in determining whether these patients went on to have a FIB (χ² = 0.3169, df = 1, p = 0.573). Furthermore, there was no significant difference in the mean AMTS between these two groups (U = 1633, Z = −1.141, p = 0.254).

Table 5.

Number of patients receiving a FIB.

	Received FIB	Did not receive FIB	Refused
Cognitively impaired	47	36	2
Cognitively intact	126	83	2

FIB: fascia-iliaca block.

Discussion

To the authors’ knowledge, this is the largest study conducted looking at the impact of cognition on pain relief in patients with a hip fracture and we achieved all our stated aims. Furthermore, this is the first analysis looking at time to analgesia in the pre-hospital cohort as well as the impact of cognition on receiving a FIB. Poor cognition has been established in the literature as a persistent barrier to patients receiving adequate analgesia; however, our results show that the practices observed in our study overall do not discriminate based on cognition.

Patients with AMTS < 7 tended to be older and women are more commonly found in cohorts of patients who have hip fractures. This is in keeping with the wider literature as rates of cognitive decline significantly increase as age increases; ¹⁸ furthermore, women on the whole are more likely to sustain a hip fracture. ¹⁹ Pre-hospital, we did not find any significant difference in analgesia given to patients based on cognition, with the most common analgesics given falling into WHO Level 3 (opioids). This is in contrast to previous studies which show that patients with cognitive impairment are significantly more likely to not receive any analgesia and were less likely to receive stronger opioids. ¹² Pain score and level of analgesia given correlated well across the whole cohort, and these trends were consistent regardless of the cognitive status of the patient. While there are no strict guidelines on what pain score merits what level of analgesia, this is the first study showing that the greater the patient’s pain, the more likely they were to receive stronger analgesia.

Arriving into A&E, our data are once again at odds with the literature. Cognition did not impact the level of analgesia patients received and both groups received equivocal pain relief. However, a significantly larger proportion of patients did not receive any analgesia in both cohorts (cognitively impaired = 44.7%, cognitively intact = 41.2%). The presence of the FIB pathway may alter staff perceptions about analgesia prior to a FIB. Patients who have been given a FIB in their admission benefit from reduced reliance of opioid analgesics and tend to have improved post-operative outcomes. ²⁰ However, FIBs are only given once the patient has been assessed by a clinician and radiographs obtained; this should not preclude patients from receiving analgesia if required prior to their FIB block. Fifty per cent of patients in both cohorts received FIBs and once again, cognition was not a barrier in the likelihood of a patient receiving a FIB. This is in keeping with departmental policy, whereby a low AMTS is not a contraindication to a patient receiving FIB. Interestingly, in hospital, we did not observe a correlation between the pain score and level of analgesia given. Patients who received no analgesia had comparable pain scores to those who received WHO Level 3 analgesia. Once again, FIBs could be the reason why patients with high pain scores do not receive any analgesia. However, this is still concerning as there is still a delay between patient arrival into A&E and FIB. This represents an area for improvement moving forward.

Patients waited just under 30 minutes on average in the pre-hospital setting prior to receiving analgesia. As far as the authors are aware, this is the first study looking at pre-hospital analgesia times in hip fracture patients. Time till analgesia was not affected by the patient’s cognitive status. However, patients waited significantly longer in A&E to receive analgesia, with average wait times of just over 2.5 hours. This wait time again was not influenced by cognitive status. This data is at odds with the literature, which shows patients with reduced cognition wait longer to receive analgesia.^12,13,21 The longer wait time within A&E is likely because pre-hospital, there are significantly fewer barriers to the prescription and administration of analgesia. The person who assesses an individual’s pain is both the prescriber and the administrator of medication. Within A&E, this is hampered by the need to be assessed by a clinician, prescription of adequate analgesia, controlled drug checks and administration of the analgesia by a different member of staff.

Pre-hospital records showed no significant difference in the presence of a pain score regardless of cognition. However, in A&E, patients with cognitive decline were significantly more likely to not have a pain score recorded. This is reflective of a wider trend that patients with cognitive decline are less likely to be able to communicate their pain on a NRS, reducing the likelihood of having a pain score recorded.^11,21 NRS have shown to put patients with cognitive decline at a disadvantage when it comes to reporting their pain. ¹⁰ Most acute care settings still rely on the use of NRSs due to their ease of use and lack of significant staff training required. Yet these are not fit for task for hip fractures, particularly when a quarter of patients have a degree of cognitive impairment. ⁶ The Bolton Pain Assessment tool is a potential solution; an observational tool built on previous pain scoring tools such as the Abbey Pain scale as well as input on a patient’s behaviour from a carer or partner. ²² This tool has been validated in the trauma setting with good staff response and positive pain management interventions. ²² We hope the use of this novel tool will significantly improve accurate pain assessment in patients with cognitive decline.

Our study shows that overall, cognitive decline does not impact the pain relief that patients receive, both via the ambulance service and A&E. We hope that this is representative of a wider trend as pain recognition in the cognitively impaired improves across the board. This positive change may be driven by the greater emphasis placed by NICE and BOAST guidelines on adequate pain relief in patients who have sustained a hip fracture. Furthermore, our data represent an opportunity for centres to compare outcomes and establish a positive benchmark in pain management in the cognitively impaired. The biggest hurdle to overcome is to reduce the delay in A&E to receiving adequate pain relief. This can be successfully achieved using Patient Group Directions (PGD) to reduce barriers for staff when administering pain relief. ²³ Furthermore, initiating analgesia at triage and further prioritising pain management in A&E training and education have potential to reduce delays and adequately tackle pain earlier in the patient journey. ²³

However, there are some limitations to the conclusions we can draw from this study. It relies on the use of retrospective data which is inherently reliant on adequate completion on clinical paperwork and may not necessarily represent day-to-day factors. Furthermore, we have treated the provision of analgesia as a binary decision, when in reality it takes into account a multitude of factors such as renal function, sensitivities and patient preference. Our study utilises data from one centre and may not be representative of wider practice, therefore further data from other major trauma centres would be required. Despite our study concluding data contrary to the wider literature, we believe it is sufficiently powered to adequately reflect current practice.

Conclusion

Contrary to the broader literature, our study shows that cognition does not impact the provision of analgesia in patients who have a hip fracture. We commend our paramedic colleagues for ensuring patients with hip fractures receive adequate analgesia regardless of their cognition. Our case–control study represents the largest analysis of the impact of cognition on pain relief following hip fractures to date and we are pleased to report our centre bucks the wider trend. While cognition does not impact pain management, there is still room for improvement, particularly in the delays faced in A&E. We hope the potential introduction of the Bolton Pain Assessment in A&E will lead to improved recognition of pain in the cognitively impaired and therefore appropriate pain management.