Article Text

Interventions to improve the management of pain in emergency departments: systematic review and narrative synthesis
  1. F C Sampson1,
  2. S W Goodacre2,
  3. A O'Cathain3
  1. 1Health Services Research, ScHARR, University of Sheffield, Sheffield, UK
  2. 2ScHARR, University of Sheffield, Sheffield, UK
  3. 3Medical Care Research Unit, ScHARR, University of Sheffield, Sheffield, UK
  1. Correspondence to Fiona C Sampson, Health Services Research, ScHARR, University of Sheffield, 30 Regent Street, Sheffield S1 4DA, UK; f.c.sampson{at}sheffield.ac.uk

Abstract

Introduction Pain management in emergency departments (ED) is often inadequate despite the availability of effective analgesia, with many patients receiving insufficient and untimely analgesia. We conducted a systematic literature review to identify interventions that could improve pain management in the ED.

Methods We systematically searched seven databases for studies reporting pain management outcomes after intervention to change professional practice to improve pain management in the ED, compared with pain management before or without intervention. Data was synthesised using principles of narrative synthesis.

Results We identified 43 relevant studies, including 40 uncontrolled before-and-after studies. Interventions included implementation of guidelines and protocols, educational interventions, pain scoring tools and changes in nursing roles, with many multifaceted interventions incorporating two or more of these elements. Interventions aimed to improve assessment and documentation of pain, knowledge and awareness of pain management and reduce time to analgesia. Due to the high probability of bias in study design and significant variation between studies, it was not possible to estimate the overall effectiveness of interventions, or identify which had the greatest impact. Intervention to improve pain management was reported to have some positive impact in most studies, but these findings may be explained by limitations in study design.

Conclusions Many interventions reported improvements in pain management, but current evidence is insufficient to recommend any for widespread adoption. In order to improve pain management we need to understand more about the theory underlying interventions, the context in which interventions work, and develop interventions based on this stronger theoretical understanding.

Statistics from Altmetric.com

Introduction

The inadequate treatment of pain within emergency departments (ED) is a well documented problem worldwide.1 ,2 Suggested reasons for the undertreatment and untimely treatment of pain include lack of awareness of pain management, difficulties in assessing and reassessing pain and structural problems within the ED contributing to delays.2 ,3 Various effective pharmacological and non-pharmacological treatments to reduce pain are available within EDs, but despite the existence of comprehensive guidelines to assist the management of pain within EDs4–6 underprescribing and inappropriate prescribing of analgesia, and delays to analgesia for patients with painful conditions, remains a significant problem. Interventions to change professional behaviour within the ED may help to improve the management of pain within the ED.

A number of interventions to change professional behaviour have been evaluated in other settings. Change in practice is more likely to be effected by use of active methods and by multifaceted strategies that incorporate a range of methods to change practice.7–9 Similarly, interventions are more likely to work if they have some ‘ownership’ or organisational support and are feasible in practice.8 ,10 Interventions to improve the use of analgesia and processes for providing analgesia within EDs potentially include the introduction of protocols and guidelines, incorporating and mandating pain-scoring tools within the triage process and the use of educational interventions to improve awareness and knowledge of pain management within the ED. We are not aware of any attempt to systematically evaluate the various potential methods and draw conclusions about which should be recommended for general adoption.

This systematic review of the literature aims to identify interventions that could improve the management of pain in the ED and synthesise the existing literature to identify which interventions work. Specifically, the review sought to identify any intervention seeking to improve the delivery of pain management and change pain management behaviour within an ED, rather than identify optimal treatments or test the efficacy of individual treatment modalities.

Methods

Search strategy

We searched the following databases in December 2012: Medline (via Ovid), Embase (via Ovid), Cinahl (EBSCO), Web of Science, Cochrane central register of controlled trials. We also searched Opengrey (previously SIGLE) and Health Management Information Consortium for grey literature. No limits were placed on year of publication or language. We also searched reference lists of general reviews of pain management in EDs and reference lists of all included studies. Hand-searching of journals was not undertaken as the pain management interventions used in EDs were reported in a wide range of journals and the search criteria were felt to be broad enough to incorporate any relevant articles (see online supplementary appendix 1 for search terms used). Prospero registration number Prospero 2013:CRD42013002542.

Study selection and inclusion criteria

Studies were selected using PICOS criteria (population, interventions, comparator, outcomes, study design). The population included patients presenting to the ED with any condition and of any age. The intervention must have aimed to alter the management of pain for any population of patients attending the ED by changing clinical behaviour around the management of pain. The intervention must have sought to act at an organisational level rather than patient level and needed to include all patients prior to pain assessment being undertaken. Studies reporting efficacy of a drug or method of delivery of analgesia alone were excluded. Studies must have included some form of comparison group who have not received the intervention. Studies reporting the following changes in outcomes related to pain management were included: proportion of patients receiving analgesia, time to analgesia, change in pain score, proportion of patients receiving adequate analgesia, documentation of pain score, reassessment of pain, repeat dosing of analgesia, patient satisfaction. Any study design was included, provided there was some form of comparison group.

As a broad search strategy was used to maximise sensitivity, screening was performed on a two-stage basis; initial screening to identify articles relating to any interventions targeting pain management in the ED were identified by one reviewer (FCS) and these were then reviewed by two reviewers (FCS and SWG) to identify which articles met the above criteria. Any discrepancies were resolved by discussion.

Data extraction and assessment of risk of bias

Data extraction and validity assessment was undertaken by a single reviewer. Double extraction was planned in the event of quantitative synthesis being undertaken, but was not required.

Assessment of the risk of bias within studies was undertaken using study-specific quality assessment criteria designed to address a range of potential sources of bias. This was felt to be more appropriate to the review than existing checklists11 ,12 and was adapted from criteria used in previous reviews that included non-randomised study designs.13 ,14

Data synthesis

We planned to undertake meta-analysis if appropriate data existed, but were ultimately unable to due to the high level of potential bias within the included studies and the level of heterogeneity between studies. Data was synthesised using narrative synthesis, which describes the scope of existing research and summarises data using structured narratives and summary tables. Narrative synthesis was undertaken following the four principles proposed by Popay et al15: (development of theory of how the intervention works, why and for whom; development of a preliminary synthesis of findings of included studies; exploration of relationships in the data and assessment of the robustness of the synthesis). Additionally, the introduction and discussion sections of included articles were reviewed to elicit the aims of the intervention and any lessons around feasibility and acceptability of interventions in the ED.

Studies were categorised according to typology of interventions, developed from theories around the aim of the intervention. Results were also briefly summarised by outcome, although due to the high risk of bias within results, the results focussed upon the types of interventions reported.

Results

A total of 8046 articles were identified and titles and abstracts reviewed. Seventy-five articles were then identified for review by both reviewers and 71 articles included. A further four were excluded at the data extraction process as they were subsequently found unable to meet the inclusion criteria. A total of 42 studies were included in this review. The κ score for inter-rater agreement on articles to include was 0.81.

Characteristics of included studies

There was significant variation between studies in terms of important variables, including design of the intervention, outcomes reported, length of follow-up, patient group and country (see table 1).

Table 1

Characteristics of included studies

Studies were predominantly before and after studies in a single site (n=38), with different lengths of follow-up period. There were two cohort studies of patients with sickle cell disease attending ED for vaso-occlusive crisis pain and one randomised controlled trial of different methods of displaying pain scores within ED charts. One study reported a stepped-wedge design of 55 Australian EDs involved in a national pain initiative project.

Study populations consisted of all patients attending the ED (n=5), patients with a range of painful conditions (n=17) and specific conditions (n=19), including fracture (n=5), renal colic (n=2), sickle cell disease (n=4) and others (n=8). One study did not specify their inclusion criteria.

Results from the assessment of risk of bias are shown in table 2. The level of risk of bias was high, notably due to the uncontrolled before and after design, as well as lack of blinding unmatched data collection periods, and differences in collection of preintervention and postintervention data.

Table 2

Assessment of risk of bias

Data synthesis

Stage 1: development of theory of how the intervention works, why and for whom

There are many different theories about why pain management is poor in the ED, but little empirical evidence supporting any individual theory. As a consequence, the type of intervention used to improve pain management depends upon the prevailing theory of why pain management is poor. Very few studies explicitly reported the rationale or theory behind the development of an intervention. Because of this, we identified the distinct rationales and types of intervention based on reading the articles. This was used as a preliminary theoretical framework for synthesising results (see table 3).

Table 3

Theoretical framework developed by the research team

Results of included studies could also be categorised according to outcome, country of origin or population studied, as there is a clear rationale for not combining results for each of these characteristics. However, as the focus was not on effectiveness due to the design of studies included, the categorisation by ‘type’ of intervention was undertaken to allow lessons about feasibility and acceptability to be included.

Stage 2: development of a preliminary synthesis of findings of included studies

Full details of the interventions and study findings are included in online supplementary appendix 2. The types of intervention, outcomes reported and any significant results are summarised in table 4 and discussed in stage 3 below.

Table 4

Components of interventions and outcomes reported

The most commonly reported outcomes were proportion of patients given analgesia (n=26) and time to analgesia (n=27). For both measures, 10 reported a significant improvement, and the remainder reported no significant difference (n=7, n=8 respectively) or did not report significance levels (n=9, n=8). One study reported a significant increase in time to analgesia. There were 14 studies that reported the proportion of patients who were given appropriate or adequate analgesia as an outcome (though the definition of ‘appropriate’ differed between studies), seven of which reported a significant improvement. Fifteen studies reported documentation of pain score as an outcome, of which 11 reported a significant improvement. Only seven studies reported reduction of pain score as an outcome, of which two saw a significant reduction in score.

The different elements of interventions are discussed in table 5 below. Studies attempted to improve implementation of the intervention by offering training in the use of the intervention (n=8), audit and feedback (n=10) and making use of reminders (n=6). Nearly half the interventions (n=20) were developed in-house, using local staff and knowledge.

Table 5

Key messages from studies grouped by rationale for intervention

Stage 3: exploration of relationships in the data

Key messages emerging from analysis of the studies are summarised in table 5. There was some overlap within the ‘types’ of intervention and some studies were included within more than one category.

Further exploration of outcomes

There did not appear to be any particular type of intervention that may correlate with either improved rates of analgesia or reduction in time to analgesia. Of the seven studies reporting significant improvement in rates of appropriate or adequate analgesia, six included the use of a protocol or guideline. This result, though interpreted cautiously, is encouraging, as many of the protocols included information about the correct route and dosage of analgesia in order to ensure the analgesia is administered appropriately.

Ten of the 11 studies that reported a significant improvement in documentation of pain included pain scoring within their intervention, either alone or within a multifaceted intervention, suggesting that the inclusion of pain scoring may improve documentation. The number of studies reporting reduction in pain score was low, which may be due to the difficulty in recording this as an outcome, as full recording of pain score at the beginning and end of the ED visit is required.

Stage 4: assessment of the robustness of the synthesis

Any attempt to synthesise data across different groups must be interpreted cautiously. There are a number of different factors within studies of pain management in EDs that influence the effectiveness of any interventions attempted. The populations studied varied widely in terms of ages and conditions. Assessing the success of interventions is more difficult in paediatric populations due to communication of pain levels. Pain relief is harder to achieve in certain conditions,21 and pain is more likely to be treated when known to be due to a painful condition (eg, fracture)17 ,42 and less likely when diagnostic workup is required.43

Differences in settings, particularly country, will influence effectiveness of interventions due to different expectations of pain relief and baseline levels of pain management. The implementation of pain protocols may have less impact in countries such as the USA and Australia where there are already strong national guidelines, and national bodies already recommend the mandating of pain scoring.4 ,60

Differences in length and timing of follow-up can affect outcomes, and is a source of significant bias in ‘before’ and ‘after’ studies. Several studies reported follow-up at less than 1 month postintervention, when the ‘honeymoon’ effect would likely still be strong. Outcomes from studies with significantly longer follow-up risk contamination due to secular trends.61 The time periods used to assess preintervention and postintervention outcomes were often not comparable in terms of length of time and seasonality, despite ED attendances being highly seasonal59 and correlation between quality indicators and ‘busyness’ of a department.62 There was considerable variation within the ‘types’ of interventions reported, and there is little value to comparing, for example, a department-specific protocol reinforced by interactive educational sessions, audit and reminders with a more simple protocol reinforced by a single didactic education session.

Discussion

Despite a very broad search and wide inclusion strategy, this evidence synthesis revealed a lack of good quality evidence of effectiveness of interventions to improve pain management within EDs. Over 70 studies were identified and 42 included, yet all but four used an uncontrolled before and after study design, with just one randomised controlled trial (RCT) looking at methods of displaying pain scores. This RCT compared methods of presenting pain scores with a ‘control’ of the visual analogue scale (VAS) recorded at presentation and at 2 h, which will not represent current practice in many EDs and, therefore, limits the usefulness of the study's conclusions.53 We aimed to identify generalisable methods to improve the provision of analgesia within the ED, which requires studies that compare interventions to control groups, preferably using multicentre evaluation. However, a lack of such studies precluded any meta-analysis of results to identify any single method that is most effective at improving pain management. Also, there was significant variation in the design of interventions, populations studied, length of follow-up and outcome measures used. However, the use of narrative synthesis allows a comprehensive synthesis of the literature pertaining to pain management interventions within the ED and offers some lessons about the feasibility of implementing interventions that may be useful in improving local practice.

The primary aim of this systematic review was to identify any interventions that could be adopted to improve pain management within the ED as part of evidence-based practice. The review did not identify any particular intervention that could be recommended for implementation, due to in the inferior quality of evidence. It may also be the case that even with good quality evidence, there is no ‘magic bullet’ that can be recommended as a ‘solution’ for all. Due to the large degree of variation within multifaceted interventions, it was difficult to attribute any level of ‘success’ to an individual element of the intervention.25 ,55 As in other areas, the value of the intervention will depend upon the context, and an individual intervention may only work within the setting for which it was designed.10 The level of ‘success’ of an intervention will depend upon the baseline performance of a department, and the degree to which the intervention has been tailored towards a specific department's needs.25

Many of the studies included within this review were based upon local audits undertaken by nursing and clinical staff with little or no external support or funding. Studies often reported their intervention to be successful in terms of pain management even where most of their prespecified outcomes had not shown significant change. It may be that the implementation of an intervention did have positive effects for that department, although there are too many potential sources of bias for the results to have any external validity. The process of developing an intervention, and in particular feeding back the results of preintervention audits, may have been sufficient to raise the profile of pain management within EDs regardless of the type of intervention used. The use of audit as an intervention in itself has been shown to have a moderate impact upon changing clinical behaviour in other settings.8 Some studies within this review reported that a change in practice had been observed following feedback of the preintervention audit, and prior to an intervention being implemented, as some EDs needed the audit feedback to understand how they were performing.55 ,63

Implications for future research

Future research into interventions for pain management should consider carefully which outcomes to report. While studies may report a change in processes used, this does not always translate into patient-oriented outcomes, such as reduction in pain score, or reduction in time to analgesia. Patient-centred outcomes, such as reduction in pain score or patient satisfaction should be used within future evaluation of interventions to improve pain management.

Although future studies of interventions to improve pain management in EDs would benefit from a stronger research design (eg, cluster RCT), it is unlikely that the evaluation of any individual intervention will provide valid recommendations for adoption that could be generalised to other EDs without a stronger theoretical underpinning for the interventions. It is probable that a ‘magic bullet’ intervention does not exist, and future research needs to focus on factors associated with improved pain management in order for EDs to develop interventions specific to their needs. A stronger theoretical framework for interventions, combined with more robust evaluation designs such as RCTs, will enable EDs to understand how and why an intervention works, and under what conditions it may succeed.

Conclusions

There is currently insufficient evidence to recommend any interventions to improve pain management within EDs for widespread adoption, and it is likely that interventions need to be tailored to individual settings in order to address barriers that exist within that department. Interventions to improve pain management should be formed upon a stronger theoretical understanding of how and why interventions may work. They should be developed following diagnostic analysis of an individual department’s needs, include adequate pain assessment and reassessment and attempt to identify and address structural and attitudinal barriers to pain management. Evaluations of interventions should ensure that patient-oriented outcomes are reported and use robust evaluative designs.

[AQ: As per journal style, a ‘key messages’ box is not required for this article category. This has been deleted.]

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • Contributors FCS undertook the systematic review, analysed the data and drafted the paper. SWG contributed to the design of the systematic review, reviewing papers for inclusion and writing the paper. AO'C contributed to the design of the review, narrative synthesis of results and writing the paper.

  • Funding This study is independent research arising from Doctoral Research supported by the National Institute for Health Research. The views expressed in this publication are those of the authors and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health Research.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.