From a purely semantic perspective, we can define frequent users as individuals that attend a certain place excessively, more often than it is considered normal. The problem with this definition arises when we attempt to apply this term to health care settings, trying to identify a set of patients that visit paediatric services (hospital emergency departments, primary care, specialty clinics) more than normal. At times, normal is understood as what is necessary, so that the term frequent users ends up having pejorative connotations.1

The World Health Organization defines a health emergency as the onset of a health problem of varying severity that leads to the perception of need for immediate attention in the affected individual or his or her family.2 Based on this definition, all visits to emergency departments (EDs) would be necessary and frequent users would be patients that perceive this need often. Frequent use would thus depend on various factors that may influence this perception, including factors related to the hospital EDs themselves (accessibility, opening hours, awareness of its presence in the population), the health care system (promotion and availability of hospitals), the patient (underlying diseases, followup in outpatient care) and socioeconomic and cultural status.3–5

There is a generalised belief that a small set of patients exists that consume a large amount of resources and place a considerable burden on heath services,6 which may have a negative impact both on the EDs and on the patients (assessment by different health providers, repetition of diagnostic tests, lower quality of followup…). There is a considerable interest in reducing the impact of frequent use, and this has led to performance of several studies in the adult population.7,8

For the purpose of analysing frequent use, authors in the published literature have attempted to establish cut-off points based on the number of visits made in a given time interval (between 3 and 12 visits per year),8 and there is no consensus on this point. Thus, while some authors propose a fixed number of visits that vary between different studies, others determine the cut-off points based on statistical constructs such as exceeding a certain percentile.9

The magnitude of frequent and super-frequent use in the paediatric emergency setting has not been well defined. We also do not know the frequency of use of paediatric emergency departments (PEDs) in our area. The first step towards optimising the management of paediatric frequent users is to become acquainted with their characteristics and needs to then develop protocols that will result in increased efficiency, understood in terms of health, for these patients.

The aim of this study was to determine the frequency of use and the prevalence of super-frequent users in PEDs of the autonomous community of Madrid and to analyse the demographic and clinical characteristics of paediatric frequent users. A secondary objective was to assess the differences between isolated visits and second or return visits.

We conducted a descriptive and inferential multicentre, observational, retrospective cohort study including all patients that sought services in the PEDs of five hospitals in the Autonomous Community of Madrid between January 1 and December 31, 2013.

After applying the exclusion criteria, we found a total of 173445 visits corresponding to 92253 patients, with a mean of 1.88 visits per patient. Figures 1 and 2 show the proportions of patients and of the total visits that corresponded to the different frequency-of-use categories. The mean number of visits per patient and the percentages of frequency-of-use categories were similar in all participating hospitals (Table 1).

Figure 1.

Distribution of patients that visited paediatric emergency departments in 2013. Percentage of patients that made one visit, two to nine visits, and ten or more visits to the participating hospitals. H, hospital.

(0.08MB).

Figure 2.

Distribution of the total visits to paediatric departments made in 2013. Percentage of visits made by patients that visited participating hospitals once, two to nine times, and ten or more times. H, hospital.

(0.09MB).

Analysis of frequent use

Our study included 558 super-frequent users that made a total of 6813 visits to EDs (Figure 3), which corresponded to a prevalence of 0.60% (95% confidence interval [CI], 0.56–0.64%) in the total users of emergency services in our area. This group made 3.93% of the total number of visits (95% CI, 3.47–4.39%). Super-frequent users made a median of 11 visits per patient (10–13), with no differences between hospitals (P=.827).

Figure 3.

Flowchart showing the selection of patients that were super-frequent users for the analysis.

(0.18MB).

The mean age was 2.06 years (SD, 2.55) and the median was 1.5years (0.5–2.5), also with no difference between hospitals (P=.919).

The highest volume of health care use in this group corresponded to children aged less than 2 years (66.6% of the children, who made 60.9% of the visits); especially infants aged less than 1 year (48.7% of children and 34.5% of visits), followed by children aged 1 to 2years (17.7% and 26.4%). However, it was the last group that used the ED most frequently, with 18 visits per patient (Table 2).

Table 2.

Number of identified super-frequent users and visits made by age.

Age (years)	Patients		Visits		Visits/patient
	n	%	n	%
0	272	48.7	2351	34.5	8.6
1	99	17.7	1798	26.4	18.2
2	66	11.8	921	13.5	14.0
3	57	10.2	860	12.6	15.1
4	15	2.7	265	3.9	17.7
5	13	2.3	166	2.4	12.8
6	5	0.9	81	1.2	16.2
7	3	0.5	40	0.6	13.3
8	4	0.7	29	0.4	7.3
9	4	0.7	79	1.2	19.8
10	7	1.3	53	0.8	7.6
11	4	0.7	45	0.7	11.3
12	5	0.9	63	0.9	12.6
13	3	0.5	51	0.7	17.0
14	0	0	0	0	0
15	1	0.2	11	0.2	11.0
Total	558	100	6813	100

There was a predominance of male compared to female patients (60% versus 40%, P<.001), and we found no difference in the mean number of visits between them (mean±SD, 12.40±3.68 per year vs 12±2.72 per year; P=.468).

Super-frequent users made between none and sixty-seven visits to their paediatrician at the primary health care centre, with a median of 10visits per patient (5–17). We did not find a substantial predominance of super-frequent users in primary care compared to patients that made fewer visits (230 versus 204 children).

The visits of most of the children were distributed across four trimesters (62.3%) or three trimesters (32.5%), while the visits of 4.8% were concentrated in two trimesters and those of 0.4% in a single trimester (P<.001). The total volume of visits was similar in all months of the year save for a decline in July and August (482 and 368 emergency visits, respectively) and a surge in the last three months (631, 647 and 750 emergency visits) (P<.001). Patients visited more often during the afternoon shift (42.5% of visits), while the volume of visits was similar in the morning shift (29.6%) and the night shift (27.9%) (P<.001).

As for visits to hospitals that did not correspond to the geographical area where the patient resided, they mostly took place in high-level-care rather than low-level-care hospitals (79.8 vs 20.2%; P<.001). Consistent with this, patients that visited low level of care hospitals tended to live in the same geographical area (69.3%; P<.001), while patients that visited high level of care hospitals came from different areas, although the difference was not marked (53.3 vs 45.6%; P<.001).

As for their previous medical history, 13.6% of patients were in followup for a health condition related to the presenting complaint, 30.5% were in followup for reasons unrelated to the presenting complaint, and 56% were not in any type of followup at the hospital (P<.001).

Most of the visits were classified during triage as less urgent or non-urgent (level ≥4, 3186 visits, and level ≤3, 1812 visits; P<.001), although in 27%, the priority level was unknown at the expense of low-level-care hospitals (93.2% of missing data).

The most common reasons for seeking care were respiratory conditions (34.6%), followed by infectious conditions (24.3%), and the most frequent diagnoses were upper respiratory tract infection (22.9%) followed by bronchiolitis/bronchospasm (17.5%) (Table 3).

Table 3.

Presenting complaint and final diagnosis of the visits made by super-frequent users.

Reason for visit	n	%	Discharge diagnosis	n	%
Respiratory	2354	34.6	Upper respiratory tract infection	1558	22.9
Infectious/fever	1656	24.3	Bronchiolitis/bronchospasm	1193	17.5
Gastrointestinal	981	14.4	Vomiting/AGE	752	11.0
ENT	400	5.9	AOM/pharyngotonsillitis	707	10.4
Surgery/trauma	396	5.8	Febrile syndrome	411	6.0
Dermatologic	390	5.7	Skin lesions	371	5.4
Neurologic	265	3.9	Concussion/fracture/wound	367	5.4
Genitourinary	141	2.1	Laryngitis	333	4.9
Other	208	3.1	Non-urgent condition	124	1.8
Unknown	22	0.3	Pneumonia	117	1.7
			Other	853	12.5

AGE, acute gastroenteritis; AOM, acute otitis media; ENT, ear-throat-nose.

In more than half of the visits, no interventions consisting of diagnostic tests, observation or treatment were performed at the ED (54.5%). The most frequently requested diagnostic tests were radiology tests (8.2% of visits), blood tests (7.1%) and urine tests (5.7%). Patients were most frequently discharged to their homes (Table 4).

Table 4.

Characteristics of the visits made by super-frequent users.

	n	%
Hospital visited
Assigned hospital	3620	53.1
Other	3101	45.5
Unknown	92	1.4
Previous care
No	5433	86.9
Yes	822	13.1
Point of origin
Home	6620	97.2
Primary care	149	2.2
Assigned hospital	17	0.2
Another health facility	5	0.1
Out-of-hospital urgent care	16	0.2
Unknown	6	0.1
Reason for visit
Common disease	6422	94.3
Random accident	346	5.1
Traffic accident	9	0.1
Sports-related injury	11	0.2
Other	13	0.2
Unknown	12	0.2
Priority level
1	12	0.2
2	187	2.7
3	1613	23.7
4	3137	46.0
5	49	0.7
Unknown	1815	26.6
Intervention in emergency department
No	3714	54.5
Yes	3099	45.5
Diagnostic tests
No	5410	79.4
Yes	1403	20.6
Observation/treatment
No	4563	67.0
Yes	2250	33.0
Final patient destination
Discharge	3614	53.0
Admission	361	5.3
Return to ED	187	2.7
Followup in primary care	2346	34.4
Referral to hospital clinic	267	3.9

Return visits amounted to 27% of the visits, with the reason for the visit being persistence of symptoms in 56.3% and worsening in 32.3%; while in 11.1% the patient was referred back from primary care or from the ED itself or a hospital clinic. Also, 13.8% had been preceded by a visit to the primary care health centre while in 79.5% the patient had only previously visited the same ED. First visits accounted for the remaining 73% of the total visits. Sixty-seven percent of first visits were due to isolated acute conditions, while 26.6% were due to a recurrent acute condition and 6.4% to a chronic condition.

We did not find differences in the mean age of patients that made return visits compared to patients that did not (P=.063).

Table 5 presents the differences in the characteristics of first visits and return visits.

Table 5.

Analysis of first visits and return visits made by super-frequent users.

	First visit		Return visit
	n	%	n	%
Medical history
In followup for condition unrelated to complaint	1557	31.3	519	28.3
In followup for related condition	644	12.9	281	15.3	P=.008
Not in followup	2775	55.8	1037	56.5
Hospital visited
Assigned hospital	2687	54.0	933	50.8	P=.022
Other	2217	44.6	884	48.1
Previous care
No	3973	88.1	1460	83.7	P<.001
Yes	538	11.9	284	16.3
Type of complaint
Common disease	4630	93.0	1792	97.6
Accident	326	6.6	40	2.2	P<.001
Other	8	0.2	5	0.3
Priority level (more or less urgent)
≥4	2382	47.9	804	43.8	P<.001
≤3	1213	24.4	599	32.6
Intervention in emergency department
Yes	2153	43.3	946	51.5	P<.001
No	2823	56.7	891	48.5
Diagnostic tests ordered
No	4014	80.7	1396	76.0	P<.001
Yes	962	19.3	441	24.0
Observation/treatment
No	3430	68.9	1133	61.7	P<.001
Yes	1546	31.1	704	38.3
Reason for visit
Surgery/trauma	342	6.9	54	2.9
Dermatologic	313	6.3	77	4.2
Gastrointestinal	690	13.9	291	15.8	P<.001
Infectious/fever	1202	24.2	454	24.7
Genitourinary	111	2.2	30	1.6
Neurologic	219	4.4	46	2.5
ENT	325	6.5	75	4.1
Respiratory	1607	32.3	747	40.7
Other	148	3.0	60	3.3
Discharge diagnosis
Bronchiolitis/bronchospasm	768	15.4	425	23.1
Concussion/fracture/wound	318	6.4	49	2.7
Upper respiratory tract infection	1191	23.9	367	20.0	P<.001
Laryngitis	234	4.7	99	5.4
Skin lesions	304	6.1	67	3.6
Pneumonia	34	0.7	83	4.5
AOM/Pharyngotonsillitis	511	10.3	196	10.7
Febrile syndrome	318	6.4	93	5.1
Vomiting/AGE	530	10.7	222	12.1
Non-urgent condition	112	2.3	12	0.7
Other	634	12.7	219	11.9
Final patient destination
Discharge	2749	55.2	865	47.1
Admission	188	3.8	173	9.4
Return to ED	85	1.7	102	5.6	P<.001
Followup in primary care	1729	34.7	617	33.6
Referral to hospital clinic	194	3.9	73	4.0

In bold face: significantly higher group of visits.

AGE, acute gastroenteritis; AOM, acute otitis media; ENT, ear-throat-nose.

This is the first multicentre study conducted in the paediatric population during a prolonged period that includes return visits.

The study was conducted in the Autonomous Community of Madrid, where there is a public health system with universal and free coverage both at the hospital and the primary care levels. Previous studies have found a higher frequency of use in public health users compared to private insurance users or the uninsured,16,17 so we would expect the prevalence of super-frequent users in the population of our study to be higher than that reported in the literature. We cannot compare our results with those of other studies in the paediatric population, as there are very few of them4,16–18 and the cut-off point that defines super-frequent users is not consistent between studies. We can say that the percentage of super-frequent users found in our study was slightly higher than the one reported in the adult population (0.4% of the population; 2.6% of visits).10–12

Despite the variability in the definition of frequent use, we decided to use the definition applied in the literature to super-frequent users12 as it is the group that makes the most visits and because it is also the subset analysed by the working group on high-frequency users in the adult population of our autonomous community.10,11

The follow-up period varies between published studies. Most research groups have established a predetermined time period6,19,20; while others propose a follow-up period relative to the first visit made.16

Our study was consisted with the results of other studies in the paediatric age group in that the highest proportion of super-frequent users corresponded to infants aged less than 1 year,16,17 although we were unable to compare the number of visits made by each age band, since these results have not been reported in the existing literature. The fact that the highest proportion of super-frequent users is found in the youngest children may be explained by the inexperience of parents, the frequent recurrence of infectious diseases at these ages, and the anxiety produced by disease in the parents of such young children.21

We found different frequencies of use of primary care in our sample, so we could not establish the presence of specific user profiles in this setting. Our results also did not provide grounds to state, as has been proposed, that adequate primary care is one of the factors that would contribute to reduce ED overuse.4,8

The distribution of visits throughout the year supports the description of these patients as loyal.11 This loyalty, combined with the lack of an association with the frequency of use of primary care, suggests that the factors that contribute to super-frequent ED use are not only the physical proximity and convenient hours of hospital settings,22 but also the immediacy of care, the available resources and the good reputation of EDs in the population.1,5,18,23

Although the published literature has shown an association between chronic disease and increased frequency of ED use,16 the proportion of children in our case series that were in followup for diseases related to their presenting complaint was relatively low.

We did not find a clear predominance of the need versus lack thereof for diagnostic tests or treatment, although as described in the literature, we found that the majority of visits involved less urgent conditions, at least in high-level-care hospitals, that did not require intervention of any kind.14,16 However, we did find a higher number of interventions in children that made second or successive visits. The level of priority assigned in triage was not documented routinely in low-level-care hospitals, so our results cannot be extrapolated to the entire population.

The diagnoses at discharge were consistent with those reported in the literature,16,17,24 with diseases that are known to have more protracted courses, such as bronchiolitis and bronchospasm, diagnosed most frequently in patients that made return visits. Super-frequent users did not require admission to the hospital more frequently that the general paediatric population in our area,25 although admissions were more frequent in the subset that made return visits.

The proportion of return visits in super-frequent users was nearly double than that of the proportion in the general paediatric population.14 However, and despite most of these visits being due to persisting symptoms, the percentage of patients that visited primary care before returning to the ED was much lower than those described in other studies.26

Considering all of the above, while most super-frequent users present with complaints that are not urgent and do not require intervention, in many cases the course of disease cannot be predicted and they may require urgent care at some point during the process. We may need to promote health education, encourage visiting primary care facilities, give detailed information on the possible course of the diseases that patients experience, and ensure an adequate understanding of the warning signs and the situations that require urgent evaluation. Providing adequate information and increasing the availability of primary care paediatricians during afternoon shifts could reduce the anxiety experienced by parents in relation to disease, and consequently lead to a reduction in the demand for urgent care.

Despite the difficulty of identifying frequent and super-frequent users prospectively, further research is required to explore not only clinical, but also sociocultural parameters in order to gain a broader understanding of the underlying causes of frequent use patterns.

The characteristic super-frequent user profile in our study was a male patient of young age (less than 2 years) with visits distributed throughout the year, that visits the hospital nearest home in the afternoon shift, or otherwise a high-level-care hospital due to an acute condition for which care has not been previously sought at a primary care facility that does not usually require diagnostic tests or urgent treatment, and is discharged to the home or to primary care followup.

Return visits account for a considerable number of emergency department visits among super-frequent users, and their characteristics differ slightly from those of first visits. We would highlight that most are motivated by the persistence of symptoms, that they often involve conditions requiring urgent care, require more interventions—diagnostic tests, observation or treatment—and lead to hospital admission more frequently.

There is no financial gain for this study. None of the main or collaborating researchers will receive remuneration for actively participating in developing the study or data collection and processing.

The authors have no conflict of interests to declare.