Introduction

Severe burn injury is potentially catastrophic for a patient, often requiring prolonged intensive care support and causing significant acute and long-term complications [1]. The ultimate goal of burn care is to restore a patient to a functional level as close to pre-injury status as possible. In the acute phase of severe burn injury, intensive care interventions are focussed on resuscitation and largely short-term based goals. The extent to which these initial interventions impact on long term patient-centred outcomes is unclear.

The quality and consistency of outcome reporting in studies of patients with severe burn injury has been questioned, with numerous calls for a core outcome set (COS) to improve reporting [2,3,4]. Core outcomes are defined as an ‘agreed, standardised collection of outcomes measured and reported in all trials for a specific clinical area’ [5], which facilitates comparison of findings between clinical trials and improves the body of evidence in a particular field. In 1992, the Outcomes Measures in Rheumatoid Arthritis Clinical Trials (OMERACT) group developed a comprehensive framework to establish a set of core outcomes in clinical trials of rheumatology, which has seen significant improvement in outcome reporting in rheumatological trials [6]. The full framework has been well described by Boers et al. [7]. The framework consists of four key domains, from which outcomes relevant to each must be reported. The domains are mortality, life impact (patient-centred outcomes including quality of life, pain, functional status), economic/resource use and pathophysiological manifestations (such as clinical and biochemical outcomes). Part of the success of the framework has been emphasising patient-centred outcomes into COS development, in order to ensure that outcomes relevant to the patients are given importance [8]. Given the broad applicability of these domains to other medical fields, the framework has been implemented into other specialities including cardiothoracic surgery, maternity care, inflammatory bowel disease and paediatric illnesses [9,10,11]. The implementation of core outcome sets into critical care research has been lagging, and there have been a number of critical care research projects working on COS development, many of which have been guided by the OMERACT framework and its broadly relevant domains [11].

The consistency of outcome reporting in the literature pertaining to the intensive care management of severe burns is unclear. Furthermore, the extent to which patient-centred outcomes are reported in this literature is unclear. Therefore, we performed a systematic review to assess the nature of outcome reporting in studies of critically ill patients with severe burn injury.

Our study wished to address whether there is firstly consistency in outcome reporting and secondly whether studies report burns-specific patient-centred outcomes. To answer this, we applied two separate frameworks. Given the uptake of the OMERACT domains for COS development in other critical care fields, we chose to classify outcomes according to the framework as a means of assessing the consistency of outcome reporting. To assess if trials report burns-specific patient outcomes, we applied a framework proposed by Falder et al that assesses crucial long term outcomes in burns survivors [12]. The framework assesses patients’ skin, neuromuscular function, somatosensory perception (pain, itch), psychological function, physical role function, community participation and perceived quality of life.

Methods

The study was conducted according to a pre-specified protocol (see Appendix C), in alignment with the PRISMA guidelines and checklist on systematic review design [13].

Study eligibility

The study included randomised clinical trials (RCTs), pseudo-randomised clinical trials, comparative studies with concurrent controls, and intervention studies without concurrent controls that investigated adult burns patients managed in the ICU. Studies were deemed as pseudo-randomised if patients were assigned to a study arm by alternate allocation rather than true randomisation [14]. Studies were included only if the intervention was deemed a key component of severe burns management (as per our pre-specified protocol). These interventions included fluid resuscitation, transfusion strategy, ventilation strategy, nutrition, analgesia, haemodynamic monitoring or timing of surgery. Studies were included if they were written in English, enrolled human subjects and a primarily adult population. Systematic reviews, meta-analyses and case series were not included.

Data search

We conducted a literature search through PubMed and Medline (via Ovid), using MeSH terms for burns and intensive care, and the domains listed above.

The search strategy for the study was:

((((((((isotonic solution OR crystalloid OR saline OR intravenous fluid)) OR (analgesia OR anaesthesia and analgesia OR pain management)) OR physiologic monitoring) OR (pulmonary ventilation OR invasive ventilation OR non-invasive ventilation)) OR (blood transfusion OR blood product transfusion OR transfusion)) OR enteral nutrition) AND (burns OR thermal injury OR burns injury OR chemical injury OR electrical injury)) AND (intensive care OR critical care OR intensive care unit OR critically ill OR critical illness)

The time frame for the search was limited to studies published between January 1, 1960, and December 31, 2019.

Data collection

Each study was reviewed by two authors to ensure consistency in data collection. We documented the following information about each study: first author, year of publication, type of study (RCT, pseudo-RCT, comparative study with concurrent controls or intervention study without concurrent controls), patient population (degree of burn injury and salient inclusion/exclusion criteria), intervention and control, as well as the primary outcome and longest documented follow-up. If the longest follow-up was not reported, we attempted to derive it by taking the longest reported outcome.

Outcome classification

Primary outcomes were reviewed and classified as either patient centred or surrogate outcomes [15, 16]. Patient-centred outcomes were defined as those deemed relevant to patients in both the short and long term. Examples of these include mortality, measures of quality of life (e.g. psychological function, functionality, independence), pain (acute or chronic), adverse outcomes from therapy, duration of mechanical ventilation and ICU/hospital length of stay (LOS). Surrogate outcomes included biomarkers, vital signs, radiological or histological findings and other markers that were not perceived to correlate with patients’ quality of life. In addition, when classifying outcomes with the OMERACT and Falder frameworks, both primary and secondary outcomes were reviewed.

For the OMERACT outcome classification, we tabulated whether each study reported outcomes relevant to each domain in their results or discussion sections. The data was recorded as whether an outcome relevant to the domain was reported or not. Examples of outcome measures classified into each domain were: Mortality (was death reported as an outcome Yes/No), Life Impact (was a patient-centred outcome, either short or long-term, reported?), Pathophysiological Manifestation (biomarkers, clinical manifestations, vital signs) and Resource Use (direct measurement of costs or surrogate markers of cost including ICU length of stay, hospital length of stay).

For the burns-specific outcomes listed by Falder, we recorded whether each study documented an outcome relevant to any of the seven domains listed in the framework above. The data was reported as a Yes/No whether an appropriate outcome was reported.

Data synthesis

Quantitative and qualitative data from the studies was derived and tabulated with counts and proportions reported. To present both the OMERACT and Falder outcomes, data was broken down into intervention subheadings, and the number of studies reporting each domain presented as absolute numbers and percentages. The total number of papers and percentages for each framework domain were also calculated and presented in the tabulated data.

Results

A total 6154 studies were initially identified, with 98 papers meeting inclusion criteria (see PRISMA flow diagram in Fig. 1). There were 19 studies on analgesia, 26 studies on fluid resuscitation, 4 on haemodynamic monitoring, 31 on nutrition, 5 on surgical timing, 8 on transfusion strategies and 5 on ventilation strategies.

Fig. 1
figure 1

PRISMA flow diagram of study exclusion and inclusion into systematic review

Full size image

Trial characteristics

Trial characteristics and overall findings per trial type are listed in detail in Table 1 (see below—findings expressed as absolute numbers and percentages). Detailed tables with outcomes for each study have been included in the Appendix (Appendix A: Tables 2-10). The search returned 53 RCTs, 13 pseudo-RCTs, 19 clinical studies with concurrent controls and 13 intervention studies without concurrent controls. Across all studies, the median number of patients per trial was 40 (IQR 24–60), with only 13 (13%) enrolling greater than 100. Seven studies (7%) were conducted across more than one centre while the remainder were single centre studies. Median longest follow-up (LFU) was low across all study types. The type of study did not appear to affect the frequency and consistency by which OMERACT or Falder outcomes were reported. Findings for each study domain are reported in greater detail subsequently.

Table 1 Findings broken down by study type
Full size table

Analgesia

The results of the analgesia studies are listed in Table 2 in Appendix A . In all except one study, the primary outcome measured was pain. However, three different pain scales were used with the visual analogue scale (VAS) the most commonly applied (83%). One study (5%) did not report a patient-centred primary outcome; this was the study by Promes et al, which assessed area under the curve for patient temperature as a primary outcome. Median longest follow-up was 2 days (IQR 1 to 7). Only four studies (20%) had a follow-up greater than 14 days and only 1 study assessed pain at 6 months.

Fluid resuscitation

The results of the fluid resuscitation studies are listed in Table 3 in Appendix A. The primary outcome in nine studies (35%) was fluid volume administered. In the remaining 17 studies, there were 13 different primary outcomes. Only fluid balance, urine output, cardiac output and multiple organ dysfunction score (MODS) were common primary outcomes. Only two studies (7%) reported a patient centred-outcome. Median longest follow-up was 3 days (IQR 2 to 28). Longest follow-up was not available in two studies.

Haemodynamic monitoring

The results of the haemodynamic monitoring studies are listed in Table 4 in Appendix A. There was no consistency in the primary outcomes measured in all four studies, three (75%) of which were surrogate outcomes. Median longest follow-up was 3 days (IQR 3-37).

Nutrition

The results of the nutrition studies are listed in Table 5 in Appendix A. There were 26 different primary outcomes across the 31 studies; 25 of these (81%) were surrogate outcomes. Only four of these outcomes were common to more than one study with nitrogen balance the most frequent measure (three studies). Median longest follow-up was 6 days (IQR 3–28). Longest follow-up was not available in three studies.

Surgical timing

The results of the surgical studies are listed in Table 6 in Appendix A. Mortality was the primary outcome in two of the studies. The remaining three studies had differing primary outcomes, all of which were surrogate measures. Median longest follow-up was 42 days (IQR 30–180). One study reported follow-up at 6 months and one study did not report follow -up.

Transfusion strategies

The results of the transfusion studies are listed in Table 7 in Appendix A. Transfusion requirement and haemoglobin concentration pre- and post-operative were the primary outcome for two studies each. The remaining four studies had different outcome measures. Only one study reported a patient-centred primary outcome. Median longest follow-up was 25 days (IQR 11–30). Follow-up was not reported in one study.

Ventilation strategies

The results of the ventilation studies are listed in Table 8 in Appendix A. Of the five studies, two had the same primary outcome (lung injury score). Two primary outcomes were patient-centred (assessing duration of mechanical ventilation). Median longest follow-up was 18 days (IQR 6–33).

OMERACT and Falder outcome classification reporting

Tables 9 and 10 in Appendix A summarise the outcome classification for the papers found in this systematic review according to the OMERACT and Falder frameworks. Regarding the OMERACT classification, in summary, 53% of papers reported mortality as an outcome, 28% reported outcomes related to life impact, 30% reported on resource and economic use and 95% reported on outcomes related to pathophysiological manifestations. Pathophysiological manifestations were consistently reported across all domains; the outcomes were predominantly haemodynamic parameters, measurements of organ function and biochemical values and biomarkers. Economic-related outcomes were mainly surrogate outcomes of cost (ICU or hospital LOS), although one paper (Saffle et al.) measured hospital costs as an outcome. The majority of the life impact outcomes were in the analgesia studies, where all 19 studies reported pain as an outcome.

The burns specific outcome set proposed by Falder et al was poorly adhered to. Across the 98 papers only 29 times out of a possible 686 occasions did the outcome apply to one the suggested domains, equating to a reporting rate of 4.4%. Nearly all of these were reported in the analgesia studies, with all reporting pain, and only two studies reporting psychological function and 1 study reporting quality of life. The surgical timing study by Puri et al. was the only study with intention to assess neuromuscular function however it could not be completed due to poor follow-up.

There did not appear to be a trend in improved patient-centred outcome reporting with more recent studies compared to older studies.

Discussion

This systematic review was undertaken to assess the outcomes reported in the literature on management of severe burns patients from January 1960 to December 2019. We reviewed studies that addressed the seven fields of care that are fundamental to the management of burns patients in the acute care of severe burns.

Our study has highlighted deficiencies in outcome reporting in acute burn care literature. The outcomes measured are highly variable, at inconsistent time frames (usually short) and are rarely meaningful, patient-centred end points. Application of the OMERACT framework has demonstrated that studies of critically ill patients with severe burns patients only consistently report pathophysiological manifestations; however, the specific outcomes are highly variable. Even on review of the RCTs or pseudo-RCTs (which should theoretically be well-designed studies with pre-specified outcomes), there did not appear to be any consistency in outcomes when the OMERACT framework was applied. Reporting of patient-centred outcomes overall was poor. Our study has demonstrated that only the analgesia study subgroup consistently reported these outcomes; however, it is important to note that this was limited almost only to pain with short time end points. Survivors of severe burn injury are at high risk of chronic pain and given the current opioid epidemic, studies of analgesia should ideally look at long-term pain outcomes.

While the initial focus of burn care in the ICU is resuscitation and prevention of complications, the improved rates of burn survival mandate that studies of severe burn injury look at long term outcomes. We have demonstrated a low rate of burns specific outcome reporting (from the Falder framework), and moving forward, it is important to establish how early interventions are impacting on patients’ day-to-day lives when they are discharged from hospital and return to the community.

The absence of consistent, meaningful outcome reporting in the acute burns literature impacts the overall quality of the evidence and limits the ability to use it to guide clinical practice. A recent survey of Australian and New Zealand burns intensive care units demonstrated marked variability in practices [17], which likely reflects the quality status of the burns literature. Standardisation of outcome reporting would allow better comparison between burn units and help to identify areas of variable outcomes. These could then be the focus of research to determine what constitutes best practice and ultimately lead to improved patient outcomes.

Systematic reviews of other domains of intensive care have demonstrated inconsistent outcome selection and timing of outcome reporting between trials, which has hindered the development of guidelines and recommendations [11]. Further this to, with improving survival in intensive care units, there is a push to investigate outcomes beyond mortality and assess long-term patient-centred outcomes including morbidity, functionality and mental health [18]. The establishment of COS in critical care research has become a focus within the last decade. The COMET (Core Outcome Measures in Effectiveness Trials) initiative and the InFACT (International forum for acute care trialists) initiative have been instrumental in developing COS in cardiac arrest [19] and acute respiratory failure [20]. Studies are currently underway to establish COS in other important critical care domains including aneurysmal subarachnoid haemorrhage [21], physical rehabilitation [22] and delirium [23].

Strengths and limitations

This is the first systematic review of the critical care burns literature that has investigated core outcome reporting, spanning six decades of clinical research. By including a variety of trial designs in our search, we have aimed to encompass a large body of the literature relevant to critically ill burns patients. While there are obvious differences between rheumatological diseases and acute burns, we applied the OMERACT framework in our study as it is well validated and has a broad set of domains that have previously guided COS development in other specialities [24, 25]. The outcome set proposed by Falder is specific to burns patients; however, recording that dataset requires a two-hour patient interview so may not meet the standard for feasibility. We excluded studies not written in English and therefore may have omitted studies from developing countries, which may have a higher incidence of burn injury and therefore valuable data. While our search strategy was limited to Medline (Ovid) and PubMed, the content within these two databases should be representative of the vast majority of high quality scientific evidence.

Recommendations

Based on the findings of this review, we recommend that a clinically relevant COS is created which can be applied to future burns research. Development of a COS will require multidisciplinary consensus input from burn care specialists, surgeons, intensivists, anaesthetists, rehabilitation specialists and allied health workers. Support from the COMET and InFACT initiatives would aid this process. Given that severe burns occur with higher frequency in developing countries, it would be important that development of a COS takes into account limitations of resource-poor nations.

Conclusion

This study has demonstrated that outcome reporting in the literature related to severe burns patients in the ICU is highly variable, rarely patient-centred and with a lack of long term follow-up. The development of an accepted and validated core outcome dataset that encompasses outcomes meaningful to our patients would improve the quality and standardisation of outcome reporting. This would lead to improvement in the quality of the burns literature, and eventually improved care and patient outcomes.