Factors influencing general practitioners decisions to refer Paediatric patients to the emergency department: a systematic review and narrative synthesis

Background Clinical guidelines are integral to a general practitioner’s decision to refer a paediatric patient to emergency care. The influence of non-clinical factors must also be considered. This review explores the non-clinical factors that may influence general practitioners (GPs) when deciding whether or not to refer a paediatric patient to the Emergency Department (ED). Methods A systematic review of peer-reviewed literature published from August 1980 to July 2019 was conducted to explore the non-clinical factors that influence GPs’ decision-making in referring paediatric patients to the emergency department. The results were synthesised using a narrative approach. Results Seven studies met the inclusion criteria. Non-clinical factors relating to patients, GPs and health systems influence GPs decision to refer children to the ED. GPs reported parents/ caregivers influence, including their perception of severity of child’s illness, parent’s request for onward referral and GPs’ appraisal of parents’ ability to cope. Socio-economic status, GPs’ aversion to risk and system level factors such as access to diagnostics and specialist services also influenced referral decisions. Conclusions A myriad of non-clinical factors influence GP referrals of children to the ED. Further research on the impact of non-clinical factors on clinical decision-making can help to elucidate patterns and trends of paediatric healthcare and identify areas for intervention to utilise resources efficiently and improve healthcare delivery.


Background
General practitioners (GPs) are often the first point of contact to healthcare, acting as a gateway to emergency, specialist, or other secondary care services in many countries [1,2]. Rising utilization of emergency departments (ED) has been recorded internationally, leading to increased attention to reducing demand on emergency services [3,4]. This debate is commonly framed in terms of the appropriateness of visits, and children have been identified as high users of the ED with conditions treatable in primary care [3,5]. GPs contribute substantially to ED attendance rates through referrals. In the UK, GP referrals account for 21% of emergency admissions annually [6], while in Australia 8% of all presentations to the ED are referred by a GP [7]. In Ireland, referrals account for approximately 37-38% of paediatric ED attendances in 2015 [8]. However, little is known about the decisionmaking process behind GP referrals to the ED, particularly when it comes to children [9].
Significant variation in GP referral patterns to secondary care has been recorded, however the reasons are manifold and not fully understood [6,10,11]. Referrals are highly complex and present challenges to GPs, particularly in relation to children, with decisions made in a time pressured manner [6,7]. While clinical aspects of the presenting condition are fundamental to the decision to refer, a multitude of other factors influence GP decision-making, including a complex interplay of clinical and non-clinical factors relating to the GP, the patient and health system considerations [10,12].
Non-clinical factors have been identified in previous literature, although this is predominantly in relation to specialist and other secondary care services and not focused exclusively on paediatric patients [11,[13][14][15]. These include characteristics of the GP such as level of training received, or clinical experience [13]. This may be particularly pertinent with paediatric patients where lack of exposure to paediatric training may lead to a loss of confidence in treating children [7,14]. Risk aversion, tolerance of uncertainty and the interaction between doctor and patient is also significant to understanding referral rates [15][16][17]. Patients' expectations and pressure to refer may have some bearing, with parental requests reported as influencing referral decisions in a number of studies [2,18,19]. Finally, structural considerations such as the organisation of the health system, accessibility of specialist care, and waiting lists may also affect the GP's referral rates [10,20]. Insight into the trends and patterns of GP referrals is critical to informing health system policy and management, particularly in the context of rising pressure on EDs [7].
There is a paucity of research regarding the influence of non-clinical factors on GP decision-making regarding referrals. This review aims to address this by conducting a systematic review exploring the non-clinical factors that may impact a GP's decision to refer a paediatric patient to the ED.

Methods
A systematic review was conducted to establish the nonclinical factors that influence the decision-making of GPs when referring paediatric patients to the ED. The review was conducted following the PRISMA framework [21]. The protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO, registration no. CRD42020145233) [22].

Search strategy
Search terms were developed following a limited search of the databases Medline (Pubmed), CINAHL, Web of Science, Embase and PsycINFO. The search terms utilized are displayed in Additional File 1 (see Additional file 1). Five databases were used: Medline (Pubmed), CINAHL, Web of Science, Embase and PsycINFO. A modified search term strategy was employed for a secondary search of Google Scholar, of which the first 10 pages were selected and reviewed for relevance, and Lenus (an Irish database). References of included articles were also screened. At the outset of the search, the time span covered articles published in English from August 2010 until July 2019. However, as the initial screening produced a small number of studies and to ensure a broader literature was included in the review, the date parameter was extended to capture literature from 1980.

Inclusion and exclusion criteria and screening
The search strategy and screening process is documented in the PRISMA diagram [21] in Fig. 1. Both qualitative and quantitative primary studies published in English that aimed to analyse non-clinical factors that influence GP decision-making when referring paediatric patients to emergency services were included. Studies were excluded if they focused on adult populations only and were expert opinions or editorials. Title and abstracts of articles obtained from the searches were screened independently by two researchers (EN & CC) using the online review management software Covi-dence™ [23]. This was then followed by independent full text review by two researchers (EN & CC) and any conflicts were discussed and resolved.

Data extraction and quality assessment
Data was extracted by the primary researcher (CC) and a second researcher independently extracted data from three included articles (EN). Characteristics of the included studies are displayed in Tables 1 and 2. All included studies had 100% complete data sets.
Each study was quality assessed independently by two researchers (EN & CC) using the Mixed Methods Appraisal Tool (MMAT) [24]. The MMAT provides a framework for appraising quantitative, qualitative, and mixed method studies for methodological quality and rigor, addressing sources of data, analytical process, appropriateness of measurements, selection or researcher bias. Included studies scored moderate to high quality (≥50%). No studies were excluded based on their MMAT score. Detailed scoring is displayed in Additional file 2 (see Additional file 2).

Results
Seven published studies were included in the systematic review [25][26][27][28][29][30][31]. The countries of origin were: U. S (n = 3), U. K (n = 2), Nigeria (n = 1) and Israel (n = 1). Studies conducted in the U. S and U. K make up the majority of the included studies therefore a hybrid private/public and universal healthcare systems are mostly represented. One study conducted in Israel outlines that ED visits are free when referred by a GP [25]. The study carried out in Nigeria does not report details on the health system [27]. Four studies [25,26,29,31] utilized a qualitative methodological approach and the remaining three were quantitative [27,28,30]. Population sizes varied significantly; four studies had fewer than 50 participants [25,26,29,31], one had 364 [30], and the remaining two just below 1000 participants [27,28]. Further information is displayed in Table 1.

Factors that influence GPs decision-making when making referrals
Factors influencing GPs' decision-making and preferences elicited are represented in Table 3.

Factors relating to patients Parental/ caregivers influence
Parents and/or caregivers influence featured in four studies [25][26][27]29]. GPs reported parental anxiety as an influencing factor, with higher levels of perceived anxiety prompting GPs to make the decision to refer [25,29]. Perceived level of parents' health literacy and capability to recognise worsening signs of their children's condition and ability to provide care were reported [25]. GPs considered the parents' perception of severity of illness, and in one study stated they trusted parents' instincts about the deterioration of a child's health and accepted their judgement that it warranted an ED visit [26]. Parental pressure to refer was reported in two studies [25,27]. In the study conducted in Nigeria, parental request accounted for 15.8% of referrals [27]. GPs in another study stated this generated a "moral conflict" between pleasing and/or reassuring parents, and adherence to best medical practices [25].

Socio-economic status
Patients of lower socio-economic status were more likely to be referred to the ED in two studies [25,29]. In one study this was attributed to providing financial help to parents as, in the Israeli context, attendance at the ED without a GP referral is paid out of pocket [25]. In one U. S study, patients with public health insurance were referred to the ED to access specialty care [31].

Age & Previous History
While not statistically significant, patients who were two years old or younger were more likely to be approved  for a referral for a non-urgent ED visit and GPs considered the child's previous admission history [30].

Factors relating to GPs
Risk aversion "Erring on the side of caution" was reported by GPs who felt it was preferable to refer to the ED rather than risk patient's health [26,29]. GPs stated they like to be completely comfortable in sending a child home [29]. GPs cited relying on their "gut instinct" and "rule of thumb" protocols when considering referral [25,29]. They reported practicing defensive medicine by considering legal implications, such as the risk of incurring lawsuits [25].
On the other hand, GPs working in the ED setting were less likely to utilize specialist services such as radiography or microbiology investigations, prescribe medications or refer to outpatient services compared to ED staff [28].

Preference for referral destination
A study with both paediatricians and GPs found that paediatricians held a preference for paediatric urgent care centres, while GPs were happy to refer to any urgent care centre [26].

System level factors Time of day & Distance from ED
Findings in three studies indicated time of day was a factor when referring [25,29,30]. Children were more likely to be referred on the weekends in one study [25], and in another GPs approved a significantly higher proportion (58%; P < 0.01) of non-urgent ED visits due to "full office schedule" after 3.30 pm [30]. Before 3.30 pm, the most common reason was medical urgency [30]. Furthermore, GPs were more likely to deny rather than approve a non-urgent visit before noon [30]. GPs reported considering accessibility to primary services for their patients, and were more likely to refer those living in isolated areas [25].

Resource need
The lack of certain resources within primary services such as tests, treatments, expertise and funds was also reported [26,27]. GPs immediately referred patients for perceived need of sutures, cauterizations and access to laboratories [26]. In the study carried out in Nigeria, a lack of funds to continue treatment (17.1%), lack of facilities (14.5%) and lack of expertise (10.4%) were reported as reasons for referrals [27]. Finally, GPs utilized the ED as a pathway to access outpatient specialty care for children who had public health insurance in the face of long waiting lists for specialists [31].

Discussion
This systematic review makes a unique contribution due to its explicit focus on the non-clinical factors that impact GP decision-making about referrals of children to the ED. The literature synthesised suggests that along with clinical factors, non-clinical considerations relating to GPs, patients and health systems play a role in the decision-making process of GPs. While it may not be fully possible to disentangle the non-clinical issues from the clinical, it is worth isolating and examining them to understand how they influence referral decision-making. Variation of referral patterns and rates has drawn attention for some time now, although reasons for this are not fully understood, indicating they are varied, idiosyncratic and integral to the context. One of the notable findings is the small number of included studies, despite the inclusion of a broad date range, suggesting the impact of non-clinical factors on the decision-making  Qualitative, indepth interviews, Grounded Theory Purposive sample of parents triaged as non-urgent in hospital emergency department on preselected days and PCP's of same children •PCP's stated parent's perception of severity of illness justified a visit to ED •PCP's reported that they prefer to trust the instinct of parents and accept the judgements of parents and refer to the ED rather than take a risk on the child's health. If they were unable to see a child they would refer to the ED if the parent described the child as being in severe pain, "inconsolable" or "very irritable, regardless of whether or not medication was tried before the ED visit. "I would rather be wrong 100 times than make a mistake"  process of GPs when referring children to the ED is an under-researched area. As highlighted by this review, a myriad of complex factors beyond purely clinical considerations impact the GPs decision to refer. While some factors are applicable to both adult and children populations, certain factors may be more pertinent in the case of children. Parental influence featured prominently, reflecting previous research on parental or patient pressure to refer [2,12,15,18,19,32]. When seeking unscheduled healthcare, parental anxiety may be considerably heightened leading them to seek reassurance [18,25,29]. This can stem from an obligation of responsibility for their children's wellbeing and unwillingness to take risks, particularly among those who have had previous traumatic health experiences, dealing with symptoms they are unfamiliar with, or with younger children who cannot verbalize the source of discomfort or pain [33] .
Attending a primary care service before presenting to the ED is not compulsory, however in certain contexts, such as Ireland, a GP referral removes the cost of attending an ED. Parents may feel it is expedient to go beyond primary care services to emergency departments, due to their perception of urgency and an assumption that ED offers higher quality of care [34][35][36]. This review offers some insight regarding GPs reaction to parental request; while in one study GPs reported trust in parents' instincts regarding their children's health status, in another it generated a "moral conflict" for GPs [25,26]. Consideration of a parent's request for a referral shows respect for parent's wishes regarding their children's care [2]. Participation in medical decision-making improves quality of care and health outcomes [37], and has been shown to improve parents' satisfaction and linked to reducing unnecessary antibiotic use for children [38]. On the other hand, GPs must balance this with the necessity of the referral and may feel uncomfortable in their gatekeeper role, highlighting the complexity of shared decision-making, particularly when it comes to referrals [6,37,39].
GPs reported experiencing professional uncertainty leading them to refer to the ED, echoing previous research [16,17]. Erring on the side of caution may be pronounced when dealing with small children whose condition can deteriorate quickly. Research has shown GPs may have less confidence with paediatric patients due to a lack of paediatric training or experience in treating children [2,7,14].
Previous studies have highlighted ED attendance is more frequent among lower socio-economic groups and those with lower levels of educational attainment [11]. This review supports this finding, demonstrating that across contexts GPs reported being more likely to refer those of lower socio-economic status [25,29]. One study showed how U. S public funded insurance recipients, who may not be in urgent need for medical attention, were referred to ensure timely access to specialist services [31]. This indicates ED utilization by those from lower socio-economic backgrounds is influenced by system factors and clinical decision-making, and not just patient health-seeking behaviour.
Health system factors have been attributed to nonurgent use of the ED by healthcare professionals [37]. This review shows the lack of diagnostic equipment and treatments available in primary care may contribute to GPs referrals to the ED [26,27]. Additionally, one study highlighted that lack of capacity in primary care resulted in approval of non-urgent visits to the ED [30]. Nonurgent use of ED during normal business hours has been documented previously, demonstrating the need for enhanced access to primary care in order to redirect non-critical care from emergency to primary care [38]. This is especially relevant to paediatric patients as children are high users of EDs, many with conditions that could be treated in primary care [3]. As health systems differ vastly across different countries and contexts with divergence of access across a range of public and private health systems, these findings suggest the need for an international study of GPs

PATIENT-LEVEL FACTORS
Parental Anxiety and perception of illness •Parental anxiety had an influence on decision to refer [25,29].
•Parental perceptions of the severity of illness may influence the decision [26,29].
•GP's trusted parents' instincts and judgements of severity rather than take a risk on the child's health, especially if they were unable to see the child. Referrals were more likely if parents stated the child was inconsolable or in severe pain, even when parents had not tried medication [26].
Parental pressure to refer •GPs reported parent pressure for referral in order to avoid incurring cost, leading to what doctors reported as a "moral conflict" between offering good service to patients and adherence to best medical guidelines [25].
•15.8% of referrals in one study were due to a parent request [27].

Parental Capability
•GPs considered the parent's health literacy, in order to judge their ability to understand, follow instructions and capability to adequately provide care and recognise a worsening condition. This was especially important when the patient was unfamiliar to them [25].
Child's age & medical history •GPs were more likely to approve a referral to the ED for younger patients especially those under two years old and for presentations of fever or trauma [30]. •GPs considered the child's previous admission history when making a decision [26].

Socio-Economic Status
•GPs referred those of lower socio-economic status more frequently [25,29].
•In one study GPs reported their perception of socio-economic status was based on place of residence, language and distance from medical centre [25].

Risk aversion
•GPs reported they often err on the side of caution by referring rather than risk a child's health [26,29].
•GPs stated they referred to their gut instinct, intuition and rule of thumb protocols when diagnosing patients [25,29].
•A small number of GPs stated a lack of paediatric training would impact on their confidence [29].
•The legal implications and fear of lawsuits was reported as an influence on their decision to refer or not [25].
•On the other hand, GPs temporarily employed in the ED who carried out primary care consultations were less likely to utilize investigative and specialist resources, including radiography, prescriptions of antibiotics and referrals to specialist and outpatient clinics [28].
Preference for Referral destination •In one study featuring both paediatricians and general family practitioners, paediatricians were more likely to refer to paediatric urgent care centres, while general family practitioners were happy to refer to any urgent care services [26].

Time of Day & Distance from ED
•GPs in one study were more likely to refer later in the day than in the morning [29] while in another GPs gave approval for ED visits for different reasons before and after 3.30 pm [30]. Visits were most commonly approved for being clinically urgent before 3.30 pm and after 3.30 pm a full office schedule was the most common reason cited [30]. Denial of a visit to the ED was more likely earlier in the day; 24/40 denials occurred before noon while 56/151 of approvals occurred before noon [30].
•Doctor's reported referring more often on weekends (Friday-Saturday in Israel) especially Fridays [25].
•Patients accessibility and opening hours of medical centres were a consideration for GPs, and those living in isolated areas were referred to the ED more frequently [25].
Access to resources and diagnostics unavailable in primary care settings •Lack of funds to continue treatment (17.1%), lack of facilities (14.5%) and lack of expertise (10.4%) were all reported as reasons for referrals [27].
•Patients were referred in order to access resources such as tests and treatments unavailable in primary care offices. Immediate referral was given for perceived need of sutures, laboratories, and nosebleed cauterization [26].
•The ED was used as a "middle man" in order to ensure publicly insured patients get access to outpatient specialty care and GPs refer them to the ED to facilitate their access to specialists [31].
across health systems to understand the influence of the structure and financing of health systems on GP decision-making processes, and therefore on the pattern of paediatric presentations to the ED [7].

Implications for policy and practice
Internationally, health systems are struggling to meet the demand on emergency departments, with healthcare planners and managers endeavouring to reduce the strain by eliminating non-urgent utilization [3,4]. Strengthening primary care capacity and capabilities, through strategies such as increased supply and extended opening hours, can contribute to treating nonurgent cases in the community [3,40]. This could address the pattern of referrals at certain times of the day and week, highlighted in this review. Enhanced paediatric training for GPs who experience professional uncertainty when treating children may lower referrals to the ED due to an aversion of risk. An alternative strategy could be the provision of remote consultations for GPs to seek advice from paediatricians. Stronger recognition of non-clinical factors and their impact on clinical decision-making is also essential during GP training. Greater awareness of various influences on clinical decision-making is vital to ensuring appropriate and excellent care for patients [41].

Limitations
The findings of this review are limited primarily by the small number of included studies and the variation in study sample size. Literature on GP referral patterns is mostly based on referral to specialist services, and not to the emergency department, and often does not focus specifically on paediatric patients. The focus of this review was empirical studies and therefore policy documents have not been included, though it is accepted that this may shed further light on referral pathways.

Conclusion
The decision to refer a child to the ED is imbued with a complex interplay of parent, GP and structural factors integral to the context upon which that decision is made. Enhanced awareness of non-clinical factors on referral decision-making is crucial to understanding patterns of paediatric unscheduled healthcare and to planning services that respond to parent's and children's needs, whilst allowing GPs to make decisions in the best interest of the child. Literature examining referral variation is rather dated, suggesting up to date research is required to account for system changes in recent years. We have identified scope for further research, such as qualitative research with GPs, which can contribute to understanding the inter-play between primary and emergency services, pertinent in the context of rising paediatric presentations to the ED.