Paramedicine, like only a few other medical professions, has a nature which is ‘specialist generalist’ (College of Paramedics (CoP), 2013; 2015). As such, paramedics are expected to be able to manage any patient they are presented with, from minor to major illness or injury (Department of Health and Social Care (DHSC), 2005; CoP, 2017). This broad scope of practice may present issues for paramedics when the infrequent nature of some incidents, such as major trauma and cardiac arrest, is considered (DHSC, 2005; National Audit Office (NAO), 2010). In addition to and compounding this, within a changing NHS, there are various historical and contemporary political drivers altering the practice landscape, and therefore incidents attended by paramedics (DHSC, 2005; Association of Ambulance Chief Officers (AACE), 2011; NHS England, 2015).
The 2005 report on a strategic review of ambulance services, Taking Healthcare to the Patient (DHSC, 2005), started one such drive that still exists. It aimed to see patients treated at home where safe to do so, rather than arbitrarily transferring them to hospital, as had traditionally happened. This same publication estimated that 20% of wider NHS patients could be treated in the community. A more recent study by Dixon and Gaisford (2014), carried out in the same trust as this research, established non-specialist paramedic non-conveyance rates to be 27%.
The drive to discharge more patients on scene, as well as the potential infrequency of managing the critically ill/injured patient highlights the requirement for a developing skill set in paramedicine, so that these patients can be safely and effectively managed. Although the specialist paramedic role has been widely discussed in contemporary literature as a way of refining skills in order to maintain competence and provide a quality service to the public (Newton, 2012), the focus of this current work is on the non-specialist paramedic and the incidents they attend.
In addition to the above, there are raised public expectations in relation to the skill set of paramedics and incidents they can manage, largely owing to the ease of access to online medical information (Lowry and Anderson, 2006; Edwards, 2014).
These factors combine to make the traditional, non-specialist paramedic role more complex and challenging than it has ever been before, both in the context of expected skills ability, and types of incidents they are expected to manage (DHSC, 2005; AACE, 2011; CoP, 2013; 2017; NHS England, 2013a; 2013b).
Aims and objectives
The aim of this study was to identify the frequency of incidents attended with a view to highlighting potential issues that may exist for paramedic practice. This aim has been achieved through the following objectives:
Methods
Design
This study is a quantitative retrospective review of paramedic-completed patient clinical records (PCRs) seeking to identify information in relation to types of incident attended by paramedics.
Inclusion/exclusion criteria
All PCRs reviewed were completed by paramedics, practising in a non-specialist role, working in one of two pre-identified areas, within the South West of the UK.
Exclusion criteria during data collection were:
Non-conveyance was excluded as it was felt it would be less likely that this patient group would undergo clinical intervention. A reason was recorded for all excluded incidents during data collection.
PCR sample
The sample size was 600 PCRs; this was designed to identify incidents and skills occurring in 5% or more of ambulance service activity, with a precision of 0.025% (Naing et al, 2006).
The sample was split equally between PCRs completed in rural and urban areas; the decision about the specific areas used was made in consultation with the Research & Audit department of the local ambulance trust, based on the procedures involved with their PCR collection and processing. This collaboration allowed for the most contemporaneous sample possible between May 2013 to April 2014. Within each of the selected geographical areas, an equal number of clinical records were randomly selected from each month (n=25).
To achieve this random sample, chronological lists of all incident numbers for each location and month were created. These were then randomised in terms of order, creating non-chronological lists of incident numbers for each of the months within each location. These lists were then used, working top to bottom, to identify PCRs for data collection. When 25 suitable records had been selected for each month, at each ambulance station, they were redacted to conceal any non-essential, sensitive data.
Data collection
The data were collected and recorded using a custom-designed data-collection tool, informed by the design of the clinical record being reviewed. This is split into two main sections: ‘tick box’ and ‘free text’. The skills chosen were those that were technical and required marking of a tick-box to record completion; it was felt that this would be more reliable in relation to accurate data collection and measurement.
During data collection, 8.3% of clinical records were reviewed by an independent researcher against the data collection tool to verify reliability (n=50); in addition, frequency analysis was employed by the chief investigator to identify errors (Gray, 2009).
There was minimal control over missing data during collection, largely as a result of the chosen study design (retrospective review); this would suggest any omission of data was not deliberate.
Incident categories
The collected data were used to look at frequency of incident types and the factors that may influence variation.
Of the 145 provisional diagnosis codes used by the trust in question, 89 separate codes were identified during data collection. Owing to the size of this data set, the 89 codes were reconsidered under eight broader, more manageable groups (Table 1). These revised groups were informed by the general groups that the provisional diagnosis codes fall into; for example, cardiac arrest and angina would be within the cardiac group. The groups were duplicated from the trust PCR pads.
| Full Name | Abbreviation |
|---|---|
| Cardiac | Cardiac |
| Gastrointestinal | GI |
| Neurological | Neuro |
| Other | Other |
| Other medical | Oth Med |
| Psychiatric/mental health | Psych/MH |
| Respiratory | Resp |
| Trauma (injury) | Trauma |
Most of the categories are self-explanatory; however, ‘other’ and ‘other medical’ are less intuitive. Table 2 offers detail of these groups.
| Other | Other medical |
|---|---|
| Obstetrics/gynaecology | Hypo/hyperglycaemia |
| Poisoning (accidental) | Syncope |
| Environmental | Allergic reaction |
| Social | Epistaxis |
| Other | Urinary tract infection |
| Sepsis | |
| Medical other | |
| Anaphylaxis |
Data analysis
The data were initially analysed using descriptive statistics; this approach provided an overview of incident types, location and time of year. The chi-square test (χ2) was then used to establish relationships between categorical variables such as incident type and time of year.
Results
Frequency of broad incident groups
Table 3 shows the frequencies of the broad incident categories for the rural and urban areas combined. In general, trauma was the largest broad category of incidents encountered by paramedics at 24% (n=144). This is a very diverse group that consists of both minor trauma and major/multi-system trauma, possibly explaining its high frequency.
| Broad incident code | Frequency |
|---|---|
| Trauma | 24% (n=144) |
| Other medical | 18.5% (n=111) |
| Gastrointestinal | 14.5% (n=87) |
| Cardiac | 14% (n=87) |
| Respiratory | 10.3% (n=62) |
| Neurological | 7.5% (n=45) |
| Other | 5.7% (n=34) |
| Psychiatric/mental health | 5.2% (n=31) |
Continued review of the subsequent broad groups revealed ‘other medical’ as the second largest group at 18.5% (n=111). Again, this is a diverse group that represents multiple medical conditions such as diabetic emergencies, sepsis, epistaxis, urinary tract infections, syncope and more. In addition, it perhaps reflects increased calls to the ambulance service for minor illness that would traditionally have been seen in primary care.
Gastrointestinal and cardiac incidents have the next highest frequencies at 14.5% (n=87) and 14.0% (n=84) respectively. Notably, within the category of cardiac incidents, cardiac arrests were only identified in 0.5% (n=3) of the incidents reviewed.
Analysis of broad incident types in relation to the geographical area shows slight variance between urban and rural areas; this can be seen in Table 4. The urban data mimic that of the combined picture; though for the rural area, the frequencies of cardiac and gastrointestinal incidents were reversed. These differences were not however found to hold statistical significance, χ2 =10.15, p=0.18.
| Broad incident group | Rural frequency | Urban frequency |
|---|---|---|
| Trauma | 22.3% (n=67) | 26.3% (n=79) |
| Other Medical | 19.6% (n=59) | 17.3% (n=52) |
| Gastrointestinal | 14.0% (n=42) | 15.0% (n=45) |
| Cardiac | 16.6% (n=50) | 11.3% (n=34) |
| Respiratory | 11.6 (n=35) | 9.0% (n=27) |
| Neurological | 7.6% (n=23) | 7.3% (n=22) |
| Other | 4.0% (n=12) | 7.3% (n=22) |
| Psychiatric/Mental Health | 4.0% (n=12) | 6.3% (n=19) |
When broad incident types and season were analysed, there was an increase in trauma frequency during the summer months; although this was not found to be a significant difference, χ2 (21, n=600)=15.283, p=0.809.
Frequency of provisional diagnosis codes
When frequency of incident was reviewed in terms of the more specific ambulance service provisional diagnosis codes, a different picture emerges. Table 5 shows the most frequently attended codes for urban and rural records combined (1 being the most common). Contrary to the broad groups, there is no trauma code within the more focused provisional diagnosis codes; all are medical.
| Order of frequency | Provisional diagnosis code/condition | Percentage of incidents reviewed (out of 600 records) |
|---|---|---|
| 1 | M21 – Acute abdomen | 8.3% (n=50) |
| 2 | M58 – Medical other | 7.3% (n=44) |
| 3 | M15 – Acute coronary syndrome | 4.8% (n=29) |
| 4 | M05 – Chest infection | 4.2% (n=25) |
| 5 | M62 - Sepsis | 3% (n=18) |
In total, 89 different provisional diagnosis codes were recorded through the 600 PCRs; the top three are more expansive conditions (as identified within the broad groups). The top five provisional diagnosis codes accounted for 27.6% (n=166) of all the incidents reviewed.
When provisional diagnosis codes are explored in the context of geographical location, there seems to be more variance than with the broad groups, albeit still not statistically significant. Table 6 shows the top five occurring provisional diagnosis codes for each geographical location.
| Order of frequency | Urban station (prov diag code, % frequency) | Rural station (prov diag code, % frequency) |
|---|---|---|
| 1 | M58 – Medical Other 9% (n=27) | M21 – Acute abdomen 10.6% (n=32) |
| 2 | M21 – Acute abdomen 6% (n=18) | M15 – Acute coronary syndrome 6% (n=18) |
| 3 | M62 – Sepsis 4% (n=12) | M58 – Medical Other 5.6% (n=17) |
| 4 | M15 – Acute coronary syndrome 3.6% (n=11) |
M05 – Chest infection 5% (n=15) |
| 5 | D02 – Overdose (non opiate) 3.3% (n=10) |
M57 – Syncope (faint) 4% (n=12) |
A greater number of provisional diagnosis codes make up the top five most frequent incidents within the urban area when compared with the rural setting. The urban top five provisional diagnosis codes see the re-emergence of a trauma category (T03A – Head Wound); this was accompanied by the presence of sepsis, gastrointestinal, other and overdose. These did not appear in the rural top five; however, syncope is common in the rural area, but not in the urban setting.
Review of specific provisional diagnosis codes, in the context of the month, do not show significant difference. This is likely owing to the size of the data set and the infrequent occurrence of many codes.
Discussion
When incident types were divided into broad, general categories, ‘Trauma’ was the most commonly occurring at 24% (n=144) of all records reviewed; this figure is not reflected by the AACE Academic review of demand (Edwards, 2014). Using data from five UK ambulance trusts, the review reported an incidence of 3.4% for ‘traumatic injury’ over the 2013–2014 period. In the CoP (2017) Curriculum Guidance document, a slightly higher frequency for trauma incidents is reported by the South East Coast Ambulance Service, at 7.64%. The large variation between the three sources is likely explained by the difference between the incident codes used for reporting.
Despite the seemingly high frequency of this group, at the most serious end of the scale, major trauma accounted for only 0.001% (n=1) of the 600 records reviewed; the same can also be said of the multisystem trauma code.
For many years now, the infrequency of major trauma exposure has been documented. Comment has been made about the proportion of time during paramedic education/training (initial and ongoing) spent preparing for incidents that they will encounter only very occasionally, and the stark infrequency of exposure to this patient group—less than 1% of a paramedic's working time (DHSC, 2005; National Confidential Enquiry into Patient Outcome and Death (NCEPOD), 2007; Royal College of Surgeons (RCS), 2009; National Audit Office (NAO), 2010; CoP, 2013). This generates potential for paramedics to suffer skill decay due to infrequent exposure to the patient group and infrequent use of the skills required to manage them, such as intravenous cannulation, intraosseous access and advanced airway management. The ultimate risk is a reduced standard of trauma care being offered by non-specialist paramedics within the UK, when compared with other countries (NCEPOD, 2007; RCS, 2009; NAO, 2010).
A similar situation arises with the infrequency of attendance at cardiac arrest incidents; only 0.5% (n=3) of the 600 reviewed recorded this provisional diagnosis code. Management of cardiac arrest is a skill often considered a staple of the paramedic, perhaps due to an increasing number of ‘fly on the wall documentaries’ following paramedics, showing attendance at the most dramatic incidents, such as cardiac arrests.
Similarly to the trauma-related publications already mentioned, the DHSC's Taking Healthcare to the Patient 1 and 2 (DHSC, 2005; AACE, 2011) report on a strategic review of ambulance services and its subsequent progress 6 years on, identifying that life-threatening illness/injury accounts for up to 10% of a paramedic's workload. Cardiac arrest obviously sits within this category, alongside many others, and would therefore only make up a small proportion.
The second report, Taking Healthcare to the Patient 2 (AACE, 2011), suggests that nationally, cardiac arrest patients make up approximately 1% of ambulance attendances—a slightly higher occurrence than was reflected in the 600 records reviewed for this study. Again, the infrequency of this group creates potential for skill decay when carrying out certain procedures on the critically ill cardiac patient.
There has long been a drive to prepare paramedics for the workload that they will actually encounter (DHSC, 2005; CoP, 2013; NHS England, 2015), with suggestion that ambulance services need to change from a service providing resuscitation, to that of a mobile health resource responding to the varied, more regular needs of its patients. This sounds very simple; however, paramedics responding to incidents must still demonstrate competence when managing the rarely seen major/multisystem trauma patient or the cardiac arrest patient (Health and Care Professions Council (HCPC), 2014; CoP, 2017).
The infrequency of incident types such as ‘cardiac arrest’ and ‘major trauma’ hold potential issues for the standard of care delivered by non-specialist paramedics. Pusic et al (2012), in a literature review, discuss the concept of forgetting curves; they assert that competence in clinical skills, not practised regularly, will deteriorate at an alarmingly fast rate initially before this deterioration slows down.
This is likely to apply more significantly to the more complex skills and those with a cognitive element. Management of the critically ill/injured or the patient in cardiac arrest fits this, infrequent, complex and cognitive criteria.
Discussion around forgetting curves suggests that degradation stops at around the level of a novice; paramedics should therefore still be able to satisfy the regulatory body threshold standards (HCPC, 2014). However, it is unlikely that either they or their patients would be satisfied with this in the context of increasing public expectation (Edwards, 2014).
The rate of degradation in performance is linked to the initial teaching or ‘encoding’; this makes teaching and learning strategies vitally important (Pusic et al, 2012). Additionally, the opportunity for regular, ‘deliberate’ practice with feedback can be beneficial (Ericsson, 2004). It therefore seems reasonable to suggest that these two aspects in combination, have the potential to influence the trajectory of the forgetting curve.
In the context of the most commonly encountered provisional diagnosis codes and their often, non-time-critical nature, the AACE (2011) highlights some potential issues. It suggests that these incidents can sometimes be more challenging to manage than time-critical situations such as cardiac arrest, where decisions are often relatively binary and follow more intuitive and practical guidelines. It again raises questions about the balance of paramedic education and training, which historically spent a disproportionate amount of time preparing paramedics for life-threatening emergencies (DHSC, 2005; AACE, 2011).
Patients not conveyed may be considered part of the non-time-critical population. Non-conveyance of patients, although not included in this study, may still require the use of clinical skills, potentially within complex situations. Work from the same area as this study reported non-specialist paramedics non-conveyance rates of between 27–32% (Cooper et al, 2007; Dixon and Gaisford, 2014), and other UK studies have reported similar figures (Snooks et al, 2003).
An international systematic review suggested that 52% of ambulance service use was unnecessary/inappropriate; this is different to specific conveyance rates but does have some obvious similarities (Mikolaizak et al, 2013). These figures suggest an additional source of data that may offer further insight into the issue of skills competence and confidence for the future. Discussion around the complexities of managing specific patient groups safely and effectively, intuitively leads to an acknowledgement of specialist paramedic roles such as in critical, primary and urgent care fields of practice. Literature has espoused the benefits to the patient of this (DHSC, 2005; AACE, 2011; Newton, 2012; CoP, 2018); however, non-specialist paramedics practising in the traditional role must still be able to manage all patients in the absence of a specialist. This means that the specialist paramedic role does not solve the issues raised by the frequency of incidents reported in all situations (HCPC, 2014; CoP, 2017).
Urban and rural populations can have a different demographic structure. Edwards (2014) highlights that more densely populated areas have a higher ambulance demand but did not go so far as differentiating between call types in each area. This suggests that it is reasonable to conclude that rural, less populated areas, have a lower ambulance demand, maybe explaining the narrower variety of provisional diagnosis codes seen in the top five.
Anecdotally, paramedics have considered that clinicians working in these two distinct areas are exposed to different workloads. There is no focused research within the UK to demonstrate this. When the data from this study were reviewed, there was no significant difference in the broad incident groups across locations; similarly, when incidents were reviewed in terms of the more specific provisional diagnosis codes across areas, despite some variance, there was still no significant difference, as shown in the results section of this study.
Conclusion
Non-specialist paramedics are attending a wide range of incidents. Whether in terms of broad groups or provisional diagnosis codes, the most frequent incident types tend to be diverse in nature. There was no statistical significance found in the difference between incident types attended and location or time of year.
The study confirms previous literature in relation to the low frequency of incidents such as major/multisystem trauma and cardiac arrest. Furthermore, contemporary literature suggests there may be an issue in relation to skill retention at incidents that are only encountered rarely. This provides cause for careful consideration by paramedics and education providers about how clinicians practising in a non-specialist role will continue to meet the standards of proficiency, as well as maintain confidence and competence in their practice.
Limitations
A larger sample of PCRs would have offered a more detailed picture of the frequently attended provisional diagnosis codes which may have allowed for greater specificity during analysis. In addition, widening the data collection to include qualitative aspects of the PCR would also perhaps have offered better insight.
The study design excluded patients not conveyed to hospital; however, inclusion of this group may have highlighted patterns and trends with skills used in these situations. The inclusion of this group would also have offered a more detailed insight into the wider frequency of incident types as a proportion, without this patient group removed from the sample.