a b s t r a c t

Objective: We aimed to identify predictors of effective management of acute pain in children in the pre-hospital setting.

Methods: A retrospective cross-sectional study using electronic clinical records from one large UK ambulance ser- vice during 01-Oct-2017 to 30-Sep-2018 was performed using multivariable logistic regression. We included all children b18 years suffering acute pain. Children with a Glasgow Coma Scale score of b15, no documented pain or without a second pain score were excluded. The outcome measure was effective pain management (abolition or reduction of pain by >=2 out of 10 using the Numeric pain rating scale, Wong-Baker FACES(R) scale or FLACC [face, legs, activity, crying and consolability] scale).

Results: 2312 patients were included for analysis. Median (IQR) age was 13 (9-16), 54% were male and the cause of pain was trauma in 66% of cases. Predictors of effective pain management include children who were younger (0-5 years) compared to older (12-17 years) (adjusted odds ratio [AOR] 1.53; 95% confidence interval [CI] 1.18-1.97), administerED analgesia (AOR 2.26; CI 1.87-2.73), attended by a paramedic (AOR 1.46; CI 1.19-1.79) or living in an area of low deprivation (index of multiple deprivation [IMD] 8-10) compared to chil- dren in an area of high deprivation (IMD 1-3) (AOR 1.37; CI 1.04-1.80). Child sex, type of pain, transport time, non-pharmacological treatments and clinician experience were not significant.

Conclusion: These predictors highlight disparity in effective pre-hospital management of acute pain in children.

Qualitative research is needed to help explain these findings.

(C) 2019

Introduction

Pain is “an unpleasant sensory and emotional experience associated with actual or potential tissue damage” [1]. Considering access to pain management is a fundamental human right [2] and has recently been identified as a main quality outcome measure for ambulance services in the UK [3], pre-hospital pain management in children is poor [4]. One Australian study found that more than half (55%) of children with severe pain (verbal Numeric rating scale 8-10) did not receive any anal- gesia [5]. A United States study [6] found that from 55,642 pre-hospital

? Presentations: This study was presented at the College of Paramedics National Research Conference on 24th September 2019 in Cardiff, UK.

* Corresponding author at: Community and Health Research Unit, School of Health and Social Care, University of Lincoln, Sarah Swift Building, Brayford Wharf East, Lincoln LN5 7AT, England, United Kingdom.

E-mail address: [email protected] (G.A. Whitley).

patients aged b19 years, 42.1% suffered a traumatic injury or pain, yet only 0.3% received analgesia. A recent UK study found that of injured children who reported pain (n = 7483), 38.8% received no treatment [7].

The management of pain is complex, especially in children, as age, developmental level, cognitive ability, communication skills and associ- ated beliefs must be considered [8]. Without effective pain treatment, children may suffer long-term psychological changes (e.g. altered pain perception) [9,10] and are at risk of developing posttraumatic stress dis- order [11,12]. A greater understanding of pre-hospital pain manage- ment in children is required to improve quality of care and clinical guidance, reducing the number of children at risk.

The UK national ambulance service clinical practice guidelines [13] (Joint Royal Colleges Ambulance Liaison Committee [JRCALC]) state that all children suffering pain should receive pharmacological treat- ment. The Royal College of Emergency Medicine (RCEM) state that ‘rec- ognition and alleviation of pain should be a priority when treating ill

https://doi.org/10.1016/j.ajem.2019.11.043

0735-6757/(C) 2019

and injured children.’ [14] Neither JRCALC nor RCEM guidelines are based on strong evidence; only one reference supports the JRCALC guideline on child pain management and all the RCEM recommenda- tions are based on the lowest quality of evidence (expert opinion). Therefore, there is significant potential for clinical practice guideline improvement.

Three studies have used regression analysis to identify predictors of effective pre-hospital pain management in children [15-17]. These stud- ies showed that predictors of effective pain management included child sex (male), child age (younger), type of pain (traumatic), initial pain score (high) and analgesia administration. We recently performed a systematic mixed studies review which included these three studies and recommended further research [18]. Predictors of effective pre- hospital pain management in children have not previously been identi- fied in the UK.

Therefore, we aimed to identify predictors of effective management

of acute pain in children in the pre-hospital setting.

Methods

Study design and setting

We performed a retrospective cross-sectional study of anonymised electronic clinical records from one UK ambulance service (East Mid- lands Ambulance Service NHS Trust [EMAS]).

EMAS is one of 10 ambulance services in England and is based in the Midlands. It serves a population of 4.8 million over an area of 16,666 km² across six counties covering both urban and rural areas [19]. Approxi- mately 2500 emergency calls are received per day and EMAS employ ap- proximately 2300 ambulance staff.

Participants

Inclusion criteria

• • • Patients aged 0-17 years
    • Attended by the East Midlands Ambulance Service NHS Trust
    • Suffering acute pain (pain b12 weeks duration [20] for example ‘Ab- dominal Pain’ or ‘soft tissue injury‘)

Exclusion criteria

• • • Patients with a Glasgow Coma Scale (GCS) b15 at any time or no doc- umented GCS
    • Patients without two documented pain scores

Patients with a GSC b15 are less likely to provide accurate pain reporting due to the reduced level of consciousness.

Data collection

Anonymised data were extracted from 1st October 2017 to 30th September 2018 using EMAS electronic clinical records.

Initially all electronic patient report forms for children aged b18 years were screened by Clinical impression (pre-hospital diagnosis; selected from a pre-determined list of conditions) to include only children who were likely to be experiencing acute pain. This was performed in dupli- cate with the inter-rater reliability determined by Cohen’s Kappa statistic. Full data extraction included: patient age, gender and ethnicity, index of multiple deprivation (IMD) code (generated from home post- code by East Midlands Ambulance Service NHS Trust), clinical impres- sion(s), clinical observations to include all available GCS scores and pain scores (documented as numeric or visual), times of observations, medications and non-pharmacological treatments administered, times

of medications and non-pharmacological treatments, incident call time, arrival at scene time, left scene time, arrival at hospital time, clini- cian rank (paramedic/technician), clinician experience (years of NHS service), clinician gender and clinician age.

Medications included all drugs administered. Analgesics were counted when administered by anyone (clinicians, patient or parents/relatives) at any time and included paracetamol (tablets, suspension, intravenous), ibuprofen (tablets, suspension), Entonox, morphine sulphate (oral, intra- venous), aspirin, codeine, naproxen, Buscopan, co-codamol, diamorphine, dihydrocodeine, ketamine, pethidine, Solpadeine and tramadol. Non- pharmacological treatments included slings, splints, bandages and dressings.

Index of multiple deprivation (IMD) data was taken from the latest (2015) Ministry of Housing, Communities and Local Government [21]. IMD considers 7 domains; income, employment, education, skills and training, health and disability, crime, barriers to housing and services and the living environment.

Outcome of interest

The primary outcome measure was ‘effective pain management’, de- fined as the abolition or reduction of pain by >=2 out of 10 on the numeric pain rating scale, Wong-Baker FACES(R) scale or FLACC scale [22-27].

The numeric pain rating scale is a self-reporting verbal scale that re- quires the child to state their level of pain on a scale from 0 to 10, vali- dated in children aged 8+ years [28]. The Wong-Baker FACES(R) scale, as seen in Fig. 1 is another self-reporting scale that requires the child to view the tool and choose the face that best illustrates their pain, val- idated for children aged 3+ years [29]. The FLACC scale is an objective scale that requires the clinician to determine the child’s pain by assessing five criteria, the face, legs, activity, crying and consolability and is validated for pre-verbal children and children not able to verbally score their pain [30].

Previous validation studies [22-27] have determined, on average, the minimum clinically significant difference in pain score is 2 points on a 10-point scale. This 2-point reduction represents a child’s ability to recognise a reduction in pain, therefore its significance is recognised and utilised in this study.

Data analysis

Descriptive statistics are displayed with means (standard deviation [SD]) and/or medians (interquartile ranges [IQR]) for continuous data and numbers (n) with percentage (%) for categorical data. Differences between included and excluded patient characteristics were deter- mined using the t-test (means), binomial probability test (proportions) and Wilcoxon rank-sum test (medians). Univariable logistic regression analysis was performed showing odds ratios with 95% confidence inter- vals (CI) and p-values. Multivariable logistic regression analysis was shown with adjusted odds ratios, 95% CI and p-values. The independent variables used in the multivariable logistic regression analysis were pa- tient age, patient gender, type of pain, analgesia administration, non- pharmacological treatment administration, paramedic crew, time to hospital and index of multiple deprivation. These variables were chosen because of their clinical relevance to the dependent variable. Initial pain score was excluded from both regression analyses to avoid mathemati- cal coupling [31].

Continuous data were categorised as follows: patient age (0-5, 6-11, 12-17 years), hospital travel time (b30 min, >=30 min), patient level of deprivation (index of multiple deprivation 1-3, 4-7, 8-10), clinician

age (20-29, 30-39, 40-49, 50-59, 60-69, 70-79) and clinician experi- ence (b5 years and >=5 years).

The variable ‘type of pain’ was dichotomised into ‘traumatic pain’ and ‘medical pain’. ‘Traumatic pain’ included alleged assault, head in- jury, limb injury, soft tissue injury and thermal injury, for example. ‘Medical pain’ included accidental overdose/poisoning, acute abdominal

Fig. 1. Wong-Baker FACES(R) Pain Rating Scale.

problem, back pain, chest pain and headache for example. Patients with traumatic and medical sources of pain were categorised as ‘traumatic pain’ (due to traumatic pain being perceived as more urgent and receiv- ing preferential treatment over medical pain [15-17]); patient’s with only medical sources of pain were categorised as ‘medical pain’ (see Supplementary File).

Where multiple clinicians were on scene, we did not merge age or experience nor did we group clinician sex into ‘male’, ‘female’ or ‘mixed’. Instead we identified the ‘senior clinician’ on scene, as it was likely this clinician would make the decisions regarding pain manage- ment. The ‘senior clinician’ was the highest-ranking clinician on scene; paramedic N technician N other (including trainee technician, emer- gency care assistant and urgent care assistant). In cases of double para- medic or double technician crews, the clinician with the lowest unique PIN number was assigned as the ‘senior clinician’.

We extracted all available pain scores, including numeric and visual. We did not combine pain scores when both numeric and visual data were available, as their comparability is not well established. Instead, we assessed them separately and determined that ‘effective pain man- agement’ was achieved when at least one of the scores met the criteria (abolition or reduction of pain score by >=2 out of 10).

Analyses were considered significant when p b 0.05. Stata Statistical Version 15 (StataCorp, College Station, TX) was used for data analysis.

Ethical considerations

Ethical approval was gained from the National Health Service (NHS) Health Research Authority (HRA) following research ethics committee approval (18/NI/0120). Approval was gained from the Clinical Audit and Research Unit, East Midlands Ambulance Service NHS Trust.

Results

During 1st October 2017 to the 30th September 2018 the East Mid- lands Ambulance Service NHS Trust received 818,340 calls. A vehicle was dispatched and arrived on scene for 662,100 of these calls whilst the remainder were treated over the phone. This resulted in treatment on scene (n = 195,523) or conveyance to further care (n = 466,577). 517,190 electronic patient report forms (ePRFs) were created during this period, giving an ePRF compliance rate of 78%. The remainder were either paper PRFs, which were not included in this study due to the manual data extraction being unfeasible or a PRF was not completed for the incident. 41,494 of these electronic clinical records were of chil- dren aged b18 years. Of these, 8052 had a documented clinical impres- sion suggestive of acute pain, as deemed by two clinical reviewers (GAW and CW), independently. The level of inter-rater agreement was deemed ‘almost perfect’ [32], with a Cohen’s Kappa score of

0.8972. All clinical impressions deemed suggestive of acute pain by both or one reviewer were included (see Supplementary File).

From the 8052 records suggestive of acute pain a number of further exclusions were made (see Fig. 2) resulting in 2312 clinical records in- cluded for analysis.

Characteristics for the 2312 included children can be seen in Table 1, segregated by the dependent variable (effective pain management).

Fig. 2. Patient flow diagram. Calls – all calls including emergency (999) and non- emergency (111/GP referral), PRF – patient report form, GCS – Glasgow Coma Scale.

Table 1

Patient characteristics.

Characteristic Effective pain management (n = 903)

Age y, median (IQR) Age y, mean (SD)	12 (8-15) 11.2 (4.9)	14 (9-16) 12.0 (4.8)	13 (9-16) 11.7 (4.8)
Age, n (%)
0-5 y	140 (15.5)	189 (13.4)	329 (14.2)
6-11 y	263 (29.1)	322 (22.9)	585 (25.3)
12-17 y	500 (55.4)	898 (63.7)	1398 (60.5)
Sex, n (%)
Female	383 (42.4)	671 (47.6)	1054 (45.6)
Male	516 (57.1)	733 (52.0)	1249 (54.0)
Not documented	4 (0.4)	5 (0.4)	9 (0.4)
Type of pain, n (%)
Medical	267 (29.6)	509 (36.1)	776 (33.6)
Trauma	636 (70.4)	900 (63.9)	1536 (66.4)
Pain score, median (IQR) Initial numeric	8 (6-9)	6 (4-7)	7 (4-8)
Final numeric	4 (2-6)	6 (4-8)	5 (3-7)
Difference numeric	3 (2-5)	0 (0-0)	0 (0-2)
Initial visual	6 (4-8)	4 (2-6)	4 (2-6)
Final visual	2 (0-4)	4 (2-6)	2 (2-4)
Difference visual	2 (2-4)	0 (0-0)	0 (0-2)
Hospital travel time (mins), 22 (13-31)		20 (13-28)	20 (13-31)

median (IQR) Analgesia, n (%)

Not effective pain management

(n = 1409)

Total

(n = 2312)

scene and/or referred to another health care professional. Data were missing for senior clinician age (n=52) and experience (n=37).

Non-pharmacological treatments were administered to 137 children and included dressings (n = 57; 42%), splints (n = 39; 28%) and slings (n = 2; 1%) along with various other treatments such as patient posi- tioning, cold compresses and eye irrigations for example (n = 39; 28%). Of those who achieved effective pain management (n = 903), 191 (21%) achieved complete abolition of pain with the remaining 712

(79%) achieving a 2-point reduction.

Results of the univariable and multivariable logistic regression anal- ysis can be seen in Tables 2 and 3, respectively.

Table 3 shows that predictors of effective pain management include children who were younger, administered analgesia, attended by a paramedic and living in an area of medium or low deprivation.

Discussion

For children suffering acute pain in the pre-hospital setting, predic- tors of effective pain management include patients who are younger,

Table 2 Univariable logistic regression analysis assessing the odds of achieving effective pain man- agement (abolition or reduction of pain >=2 out of 10).

Non-pharmacological

Administered	669 (74.1)	794 (56.4)	1463 (63.3)	Predictor	Odds ratio	Significance
Not administered	234 (25.9)	615 (43.7)	849 (36.7)		(95% CI)	(p-value)
treatment, n (%)				Patient age, y

Administered	62 (6.9)	75 (5.3)	137 (5.9)	0-5	1.33 (1.04-1.70)	0.022
Not administered	841 (93.1)	1334 (94.7)	2175 (94.1)	6-11	1.47 (1.21-1.78)	b0.000

Index of multiple deprivation, median (IQR)

Paramedic crew, n (%)

5 (3-7) 4 (2-7) 4 (2-7)

12-17 (reference) 1

Patient sex

Male 1.23 (1.04-1.46) 0.015

Female (reference) 1

Paramedic 669 (74.1) 934 (66.3) 1603 (69.3)

Non-paramedic Senior clinician age (y),

median (IQR)

Senior clinician experience (y), median (IQR)

Senior clinician sex, n (%)

Female 307 (34.0) 475 (33.7) 782 (33.8)

Male	525 (58.1)	824 (58.5)	1349 (58.4)	Administered	1.31 (0.93-1.86)	0.126
Not documented	71 (7.9)	110 (7.8)	181 (7.8)	Not administered (reference)	1

Type of pain

Non-pharmacological treatment administration

234 (25.9) 44 (34-50)	475 (33.7) 44 (34-50)	709 (30.7) 44 (34-50)	Trauma Medical (reference)	1.35 (1.13-1.61) 1	0.001
			Analgesia administration

11 (3-16)	10 (3-16)	10 (3-16)	Administered	2.21 (1.85-2.66)	b0.001
			Not administered (reference)	1

y – year, IQR – interquartile range, SD – standard deviation, numeric – numeric pain rating scale, visual – Wong-Baker FACES(R) scale, non-pharmacological treatment – slings, splints, bandages and dressings, senior clinician – highest rank clinician N lowest PIN number, ex- perience – total NHS (National Health Service, UK) employment, index of multiple depri- vation (IMD) (2015) – calculated from home postcode (IMD is based on income, employment, education skills and training, health and disability, crime, barriers to housing and services and living environment [1 = highest deprivation, 10 = lowest deprivation]).

A large group of children were excluded for no initial pain score or no second pain score (n = 3872) (see Fig. 2). This excluded group were sig- nificantly younger (median (IQR) 8 years (2-14) [p b 0.0001]), closer to hospital (median travel time minutes (IQR) 17 (11-24) [p b 0.0001]), suffered more traumatic pain (n = 2801 (72%) [p b 0.0001]), were attended by a paramedic more often (n = 2815 (73%) [p = 0.0046]), were from more deprived areas (median index of multiple deprivation (IQR) 4 (2-7) [p = 0.0002]) and received less analgesia (n = 1552 (40%) [p b 0.0001]) than those included. See Supplementary File for the table of comparison.

Index of multiple deprivation scores were available for 1585 (69%) children, with 670 (29%) having no home postcode documented and 57 (2%) home postcodes being unmatched / erroneous. Lack of home postcode documentation was likely due to the incident occurring in a public place or at school. The low percentage of unmatched/erroneous postcode data suggests a low risk of bias. hospital transport time was available for 1894 (82%) children as the remainder were discharged at

Index of multiple deprivation

Highest deprivation (reference) 1

Medium deprivation 1.41 (1.12-1.78) 0.003

Lowest deprivation 1.35 (1.03-1.76) 0.027

Missing data 1.18 (0.94-1.46) 0.148

Hospital travel time

b30 min (reference) 1

>=30 min 1.03 (0.84-1.28) 0.756

Missing data 0.92 (0.74-1.15) 0.480

Paramedic crew

Paramedic 1.45 (1.21-1.75) b0.001

Non-paramedic (reference) 1

Senior clinician age, y

20-29 (reference) 1

30-39 1.09 (0.81-1.45) 0.569

40-49 1.15 (0.88-1.49) 0.304

50-59 1.06 (0.80-1.41) 0.680

60-69 0.50 (0.28-0.91) 0.022

70-79 0.55 (0.06-5.37) 0.610

Missing data 0.95 (0.52-1.75) 0.882

Senior clinician sex

Male 0.99 (0.82-1.18) 0.877

Female (reference) 1

Senior clinician experience, y

>=5	1.06 (0.89-1.27)	0.515
b5 (reference)	1
Missing data	0.99 (0.50-1.95)	0.969

y – years, non-pharmacological treatment administration – slings, splints, bandages and dressings, index of multiple deprivation (2015) – highest deprivation (1-3), medium dep- rivation (4-7), lowest deprivation (8-10), senior clinician – highest rank clinician N lowest PIN number, experience – total NHS (National Health Service, UK) employment.

Table 3

Multivariable logistic regression analysis assessing the odds of achieving effective pain management (abolition or reduction of pain >=2 out of 10).

hypothesis that non-pharmacological treatments are associated with ef- fective pain management in Younger children.

A number of findings did not confirm previously published data.

Predictor Adjusted^a odds ratio

(95% CI)

Patient age, y

0-5 6-11 12-17 (reference)	1.53 (1.18-1.97) 1.49 (1.21-1.82) 1	0.001 b0.001
Patient sex Male	1.17 (0.98-1.39)	0.090
Female (reference)	1
Type of pain Trauma	1.18 (0.97-1.43)	0.091
Medical (reference)	1
Senior clinician experience b5 years (reference)	1
>=5 years	0.97 (0.80-1.18)	0.744
Missing data	1.42 (0.70-2.91)	0.334
Analgesia administration Administered	2.26 (1.87-2.73)	b0.001
Not administered (reference)	1
Non-pharmacological treatment administration
Administered	1.08 (0.75-1.55)	0.695
Not administered (reference)	1
Paramedic crew Paramedic	1.46 (1.19-1.79)	b0.001
Non-paramedic (reference)	1
Hospital travel time b30 min (reference)	1
>=30 min	1.00 (0.80-1.25)	0.981
Missing data	1.00 (0.78-1.27)	0.986
Index of multiple deprivation
Highest deprivation (reference)	1 (ref)
Medium deprivation	1.41 (1.11-1.79)	0.005
Lowest deprivation	1.37 (1.04-1.80)	0.027
Missing data	1.18 (0.94-1.48)	0.158

Number of observations: 2303.

Significance (p-value)

Jennings et al. [15], Lord et al. [16] and Bendall et al. [17] identified child sex (male) as a predictor of effective pain management, although Bendall et al. [17] defined ‘effective pain management’ as a pain score reduction >=30% of initial pain score. We found no significant association (p = 0.090) between child sex (male) and effective pain management. Our estimate of effect is perhaps more conservative given the number of included independent variables (n = 9). There is currently no evidence to explain why children of male sex may be more likely to achieve effec- tive pain management than children of female sex.

We found no significant difference (p = 0.981) between children who face a shorter (b30 min) journey to hospital versus a longer journey (>=30 min). This conflicts with data previously reported by Bendall et al.

[17] suggesting children with a shorter care time (b30 min) are less likely to achieve effective pain management than those with a longer care time (30-59 min). We acknowledge that the exclusion of clinical records of children without an initial pain score or second pain score (n = 3872) may have influenced the “time to hospital” predictor, as the excluded group faced a significantly shorter median hospital trans- port time of 17 min (IQR 11-24) compared to 20 min (IQR 13-31) (p b 0.0001) for the included group.

Analgesia administration was not selected as the primary outcome measure as this was deemed a proxy for effective pain management. The results of this study reinforce this decision as 849 children reported pain and received no analgesia, yet 234 (28%) of these still achieved ef- fective pain reduction. This highlights the multivariable nature of pain management in this Paediatric population and context.

The floor and ceiling effect along with regression to the mean was also considered. We categorised those who achieved a numeric pain score reduction from 1 to 0 as ‘effective pain management’ using the ‘ab- olition of pain’ criteria. The initial pain score could not be included as an

y – years, non-pharmacological treatment administration – slings, splints, bandages and dressings, index of multiple deprivation (2015) – highest deprivation (1-3), medium dep- rivation (4-7), lowest deprivation (8-10), senior clinician – highest rank clinician N lowest PIN number, experience – total NHS (National Health Service, UK) employment.

^a Adjusted for patient age, patient sex, type of pain, senior clinician experience, anal- gesia administration, non-pharmacological treatment administration, paramedic crew, hospital travel time and index of multiple deprivation.

administered analgesia, attended by a paramedic or living in an area of medium or low deprivation.

A number of our results confirm previous findings. Jennings et al.

[15] found that analgesia administration was a predictor of effective pain management in children, with an adjusted odds ratio (95% confi- dence interval) of 6.6 (5.9-7.3) when compared to those who did not receive analgesia.

Bendall et al. [17], Jennings et al. [15] and Lord et al. [16] found that younger children were more likely to achieve effective pain manage- ment than older children. This conflicts with qualitative evidence that found younger children are more difficult to assess, cannulate and ad- minister inhaled analgesia [33-35]. Therefore we conclude that pain management strategies other than analgesics were associated with a higher chance of effective pain relief for younger compared with older children. In an attempt to confirm this finding, we restricted our multi- variable logistic regression to include only children aged 0-5 years and found the adjusted odds ratio (95% CI) for achieving effective pain man- agement for those receiving analgesia and non-pharmacological treat- ments was 1.19 (0.75-1.89) and 0.94 (0.39-2.29), respectively. This subgroup analysis showed that for 0-5 year old children, analgesia ad- ministration was not associated with effective pain management. It also showed that non-pharmacological treatments were not associated with effective pain management. This may be due to the difficulty in ac- counting for all non-pharmacological techniques due to lack of docu- mentation. Future prospective studies may be needed to support the

independent variable in either logistic regression analysis to avoid mathematical coupling [31]. A percentage reduction, as used in Bendall et al. [17] was considered for the outcome measure, however much of the validation work around the minimum clinically significant differ- ence in pain has been performed using a point reduction [22-27].

We were able to account for some non-pharmacological treatments such as slings, splints, bandages and dressings within this study; al- though their administration was not significantly associated with effec- tive pain management. This data adds strength to the evidence base as previous studies were unable to account for these techniques [5,36]. Pa- tient deprivation level was also assessed as a predictor of effective pain management. To our knowledge this is a novel predictor not previously reported in this population and context.

Limitations

The electronic patient report form (ePRF) compliance rate for the study period was 78% therefore we have excluded a number of paper PRFs due to the manual data extraction of these forms being unfeasible given the time restraints of this study.

The retrospective nature of this study meant that data could only be collected when clinicians documented their assessments and treat- ments appropriately. Due to time constraints we were unable to screen the ‘free text’ section of clinical records, therefore it is possible that some observations or treatments have been missed.

EMAS clinicians could not report behavioural pain scores on the electronic PRF during the period of data extraction, except in the ‘free text’ section. EMAS follow the Joint Royal Colleges Ambulance Liaison Committee clinical practice guidelines [13] which advocate the use of FLACC (face, legs, activity, crying and consolability) as the choice behav- ioural pain assessment scale for pre-verbal children, therefore we as- sume clinicians have used this scale where appropriate and reported it as a numeric pain score during documentation.

The validity of pain scoring tools for children is a concern, as none have been validated in the pre-hospital setting [13]. We assume the nu- meric pain rating, Wong-Baker FACES(R) and FLACC scale are valid and are used appropriately.

Defining ‘effective pain management’ as an objective measure (abo- lition or reduction of pain by >=2 out of 10) may not be reflected in the patient’s perceived experience. Although validation work had been un- dertaken [22-27] it might be useful to consider subjective outcome measures in addition to objective measures in both clinical practice and future prospective research.

We excluded children with no initial pain score or no second pain score (n = 3872). Their characteristics were significantly different to those included. These exclusions were necessary to satisfy the depen- dent variable, however we acknowledge this has introduced a degree of selection bias. We expected to see a difference in these two patient groups as the lack of pain assessment documentation indicated perhaps a more challenging group of children in terms of pain assessment and management.

We were not able to measure the influence of non-pharmacological pain management techniques such as distraction and staying close to relatives, due to the difficulty of quantifying these approaches retro- spectively. However, we were able to analyse the effects of other non- pharmacological treatments such as slings, splints, bandages and dressings.

Due to the lack of documentation we were unable to determine the impact of patient ethnicity on the dependent variable. Data for clinician ethnicity was not available for extraction, therefore no ethnicity data was presented. We acknowledge the need to address ethno-diversity of data in future research.

We felt that clinicians who were parents versus clinicians who were not may have influenced the rates of effective pain management; how- ever, we were unable to assess this due to this data not being available. Future prospective research should consider this.

internal validity was deemed high due to the large number of poten- tial confounders considered and the minimisation of selection bias by screening clinical records independently and in duplicate.

External validity was deemed high as many of our results match pre- viously reported evidence [15-17]. Our study population was diverse, encompassing a wide age range from urban and rural areas within a modern EMS system.

Implications for future research

We aim to explain these identified predictors utilising a qualita- tive approach within a mixed methods study. This integration of data will enrich the quantitative findings from this study by provid- ing context from experience, culture and social norms. This will allow clinicians, policy makers and stakeholders to more compre- hensively understand the reasons for the disparity in effective pain management in children.

Pain scoring tools for children should be validated within the pre- hospital setting and the appropriateness of their use should also be ex- plored. Future research into predictive factors of effective pain manage- ment in children should consider patient and clinician ethnicity along with the clinicians’ status as a parent.

Conclusion

These predictors highlight disparity in effective pre-hospital man- agement of acute pain in children. Qualitative research is needed to help explain these findings.

Acknowledgements

We acknowledge the East Midlands Ambulance Service NHS Trust for sharing their data, in particular Deborah Shaw and Robert Spaight

from the Clinical Audit and Research Unit and Darren Coxon from the Performance Management Information Team.

Funding

This study is funded by the National Institute for Health Research (NIHR) Applied Research Collaboration East Midlands (ARC EM) United Kingdom. The views expressed are those of the author(s) and not neces- sarily those of the NIHR or the Department of Health and Social Care.

The funders had no role in the design and conduct of the study; col- lection, management, analysis and interpretation of the data; prepara- tion, review or approval of the manuscript or decision to submit the manuscript for publication.

Declaration of competing interest

None.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi. org/10.1016/j.ajem.2019.11.043.

References

1. International Association for the Study of Pain. IASP taxonomy. [Available from https://www.iasp-pain.org/Taxonomy?navItemNumber=576; 1994.
2. Brennan F, Lohman D, Gwyther L. Access to pain management as a human right. Am J Public Health 2019;109(1):61-5. https://doi.org/10.2105/AJPH.2018.304743.
3. Turner J, Siriwardena AN, Coster J, Jacques R, Irving A, Crum A, et al. Developing new ways of measuring the quality and impact of ambulance service care: the PhOEBE mixed-methods research programme. 2019;7(3). https://doi.org/10.3310/pgfar07030.
4. Samuel N, Steiner IP, Shavit I. prehospital pain management of injured children: a systematic review of current evidence. Am J Emerg Med 2015;33(3):451-4. https://doi.org/10.1016/j.ajem.2014.12.012.
5. Lord B, Jennings PA, Smith K. The epidemiology of pain in children treated by para- medics. Emergency Medicine Australasia: EMA 2016;28(3):319-24. https://doi.org/ 10.1111/1742-6723.12586.
6. Lerner EB, Dayan PS, Brown K, Fuchs S, Leonard J, Borgialli D, et al. Characteristics of the pediatric patients treated by the Pediatric Emergency Care Applied Research Network‘s affiliated EMS agencies. prehospital emergency care: Official Journal Of The National Association Of EMS Physicians And The National Association Of State EMS Directors 2014;18(1):52-9. https://doi.org/10.3109/10903127.2013.836262.
7. Whitley GA, Bath-Hextall F. Does current pre-hospital analgesia effectively reduce pain in children caused by trauma, within a UK ambulance service? A service evaluation. British Paramedic Journal 2017;1(4):21-8. https://doi.org/10.29045/14784726.2017.1. 4.21.
8. Srouji R, Ratnapalan S, Schneeweiss S. Pain in children: assessment and nonPharmacological management. Int J Pediatr 2010;2010. https://doi.org/10.1155/ 2010/474838.
9. Taddio A, Katz J, Ilersich AL, Koren G. Effect of neonatal circumcision on pain re- sponse during subsequent routine vaccination. Lancet (London, England) 1997; 349(9052):599-603. https://doi.org/10.1016/s0140-6736(96)10316-0.
10. Weisman SJ, Bernstein B, Schechter NL. Consequences of inadequate analgesia dur- ing painful procedures in children. Arch Pediatr Adolesc Med 1998;152(2):147-9.
11. Sheridan RL, Stoddard FJ, Kazis LE, Lee A, Li NC, Kagan RJ, et al. Long-term posttrau- matic stress symptoms vary inversely with early opiate dosing in children recover- ing from serious burns: effects durable at 4 years. J Trauma Acute Care Surg 2014;76 (3):828-32. https://doi.org/10.1097/TA.0b013e3182ab111c.
12. Saxe G, Stoddard F, Courtney D, Cunningham K, Chawla N, Sheridan R, et al. Rela- tionship between acute morphine and the course of PTSD in children with burns. J Am Acad Child Adolesc Psychiatry 2001;40(8):915-21. https://doi.org/10.1097/ 00004583-200108000-00013.
13. Joint Royal Colleges Ambulance Liaison Committee, Association of Ambulance Chief Executives. UK Ambulance Services Clinical Practice Guidelines 2016. Bridgwater: Class Profesional Publishing; 2016.
14. Royal College of Emergency Medicine. Management of pain in children. 30-Mar- 2017. Available from https://www.rcem.ac.uk/docs/RCEM%20Guidance/RCEM% 20Pain%20in%20Children%20-%20Best%20Practice%20Guidance%20(REV%20Jul% 202017).pdf; 2017.
15. Jennings PA, Lord B, Smith K. Clinically meaningful reduction in Pain severity in chil- dren treated by paramedics: a retrospective cohort study. Am J Emerg Med 2015;33 (11):1587-90. https://doi.org/10.1016/j.ajem.2015.06.026.
16. Lord B, Jennings PA, Smith K. Effects of the introduction of Intranasal fentanyl on re- duction of pain severity score in children: an interrupted time-series analysis. Pediatr Emerg Care 2017. https://doi.org/10.1097/PEC.0000000000001376.
17. Bendall JC, Simpson PM, Middleton PM. Effectiveness of prehospital morphine, fen- tanyl, and methoxyflurane in pediatric patients. Prehosp Emerg Care 2011;15(2): 158-65. https://doi.org/10.3109/10903127.2010.541980.
18. Whitley GA, Hemingway P, Law GR, Jones AW, Curtis F, Siriwardena AN. The predic- tors, barriers and facilitators to effective management of acute pain in children by ambulance services: a systematic mixed studies review (poster). College of para- medics 3rd national research conference; 2019 Cardiff.
19. East Midlands Ambulance Service NHS Trust. EMAS annual report 2016-2017. [Available from https://www.emas.nhs.uk/about-us/trust-documents/; 2017.
20. British Pain Society. British pain society: Useful definitions and glossary. [Available from https://www.britishpainsociety.org/people-with-pain/useful-definitions-and- glossary/#pain; 2019.
21. Ministry of Housing Communities and Local Government. English indices of depriva- tion 2015. [Available from http://imd-by-postcode.opendatacommunities.org/; 2015.
22. Powell CV, Kelly AM, Williams A. Determining the minimum clinically significant difference in visual analog pain score for children. Ann Emerg Med 2001;37(1): 28-31. https://doi.org/10.1067/mem.2001.111517.
23. Bulloch B, Tenenbein M. Assessment of Clinically significant changes in acute pain in children. Acad Emerg Med 2002;9(3):199-202.
24. Bailey B, Daoust R, Doyon-Trottier E, Dauphin-Pierre S, Gravel J. Validation and prop- erties of the verbal numeric scale in children with acute pain. Pain 2010;149(2): 216-21. https://doi.org/10.1016/j.pain.2009.12.008.
25. Voepel-Lewis T, Burke CN, Jeffreys N, Malviya S, Tait AR. Do 0-10 numeric rating scores translate into clinically meaningful pain measures for children? Anesth Analg 2011;112(2):415-21. https://doi.org/10.1213/ANE.0b013e318203f495.
26. Myrvik MP, Brandow AM, Drendel AL, Yan K, Hoffmann RG, Panepinto JA. Clinically meaningful measurement of pain in children with Sickle cell disease. Pediatr Blood Cancer 2013;60(10):1689-95. https://doi.org/10.1002/pbc.24624.
27. Tsze DS, Hirschfeld G, von Baeyer CL, Bulloch B, Dayan PS. Clinically significant differ- ences in acute pain measured on self-report pain scales in children. Academic

    Emergency Medicine: Official Journal of the Society for Academic Emergency Medi- cine 2015;22(4):415-22. https://doi.org/10.1111/acem.12620.

    von Baeyer CL, Spagrud LJ, McCormick JC, Choo E, Neville K, Connelly MA. Three new datasets supporting use of the numerical rating scale (NRS-11) for children’s self- reports of pain intensity. Pain 2009;143(3):223-7. https://doi.org/10.1016/j.pain. 2009.03.002.

28. Wong-Baker FACES(R) Foundation. Wong-Baker FACES(R) Pain Rating Scale. Retrieved [19-Jul-2019] with permission 2019 [Available from http://www.WongBakerFACES. org.
29. Merkel SI, Voepel-Lewis T, Shayevitz JR, Malviya S. The FLACC: a behavioral scale for scoring postoperative pain in young children. Pediatr Nurs 1997;23(3):293-7.
30. Archie Jr JP. Mathematic coupling of data: a common source of error. Ann Surg 1981;

    193(3):296-303.

    Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33(1):159-74. https://doi.org/10.2307/2529310.

31. Murphy A, Barrett M, Cronin J, McCoy S, Larkin P, Brenner M, et al. A qualitative study of the barriers to prehospital management of acute pain in children. Emergency Medicine Journal: EMJ 2014;31(6):493-8. https://doi.org/10.1136/emermed-2012- 202166.
32. Holmstrom MR, Junehag L, Velander S, Lundberg S, Ek B, Haggstrom M. Nurses’ ex- periences of prehospital care encounters with children in pain. Int Emerg Nurs 2019; 43:23-8. https://doi.org/10.1016/j.ienj.2018.07.004.
33. Gunnvall K, Augustsson D, Lindstrom V, Vicente V. Specialist nurses’ experiences when caring for preverbal children in pain in the prehospital context in Sweden. Int Emerg Nurs 2018;36:39-45. https://doi.org/10.1016/j.ienj.2017.09.006.
34. Murphy AP, Hughes M, McCoy S, Crispino G, Wakai A, O’Sullivan R. Intranasal fenta- nyl for the prehospital management of acute pain in children. Eur J Emerg Med 2017;24(6):450-4. https://doi.org/10.1097/MEJ.0000000000000389.