Introduction
The rate of childhood emergency admissions to hospital with acute medical diagnoses in England rose by 28% between 1999 and 2010 [
9]. Emergency admissions include those where a child has an unscheduled admission to a hospital ward, and does not include cases seen and discharged home after an emergency department attendance. In the UK, emergency admissions occur when a child is admitted to a hospital inpatient ward after having been referred to medical paediatric services by a doctor or specialist nurse working in either the community (general practice or the primary care out of hours service) or in the hospital emergency department.
The factors driving the rise in paediatric emergency admissions are multifactorial [
9], and include patient (or parent) health seeking behaviour, the capacity and expertise for child health care delivery in the community and management in hospital. The rise in paediatric emergency admissions before 2006, which was seen in England [
9] and also in Scotland [
19], seems likely to be at least in part explained by the 2004 change to the general practice contract leading to near-universal loss of GPs’ responsibility for out-of-hours services for their patients [
6] and a 4-hour cap on waiting time in accident and emergency units. The emergence of short stay paediatric assessment units in the late 1990s may have reduced costs but may also have led to increasing emergency admissions [
4,
14].
What is not clear is whether the rise in emergency admissions is due to a uniform increase across all children and all diagnoses or whether there are some groups of children and certain diagnoses where admission is becoming more prevalent. Understanding this issue is a prerequisite for the design of any intervention to halt or even reverse the rising number of emergency admissions. In Scotland, the proportion of emergency admissions accounted for by ten “primary care sensitive [
16]” diagnoses has remained constant [
19] whilst the absolute number of children with these presentations rose between 1999 and 2011 [
19]. In England, on the other hand, there was an absolute rise of 20–40% between 1999 and 2010 in the number of emergency admissions with infections, including upper respiratory tract infection [
9], lower respiratory tract infection [
9], urinary tract infection [
9], gastroenteritis [
9] and bronchiolitis [
10]. A second study which was published by the Nuffield Trust in 2017 of emergency admissions in England and Wales between 2005/6 and 2015/16 reported that overall emergency admissions for those aged up to 24 years rose by 14%, and emergency admissions with viral infection, bronchiolitis and intestinal infection doubled with asthma admissions falling by 20%. Similar trends are seen outside the UK, for example in Sweden there was a doubling in the number of infants admitted with lower respiratory tract infection between 1987 and 2000 [
3] and in Denmark there was more than a 50% increase in under 5-year olds admitted with infections between 1980 and 2001 [
12]. Previous studies [
9,
19] have analysed publicly available data which are limited due to data protection reasons and this limits analysis of admission data to descriptive trends where, for example, covariates cannot be considered.
We have obtained very detailed administrative data for all paediatric emergency admissions in Scotland. Our individual patient analysis allowed us to carry out a comprehensive analysis of emergency admission data whilst adjusting for covariates. Here we extend the previous work in this area [
9,
19] by describing trends in emergency admissions and testing the hypothesis that characteristics of patients admitted and details of emergency admissions have changed over time.
Discussion
To our knowledge this is the most comprehensive assessment of paediatric emergency admissions to hospital for a whole nation. Our study has described trends in all emergency admissions, zero day emergency admissions, readmissions and also the diagnoses and characteristics of the children admitted. Our work is the first consider covariates in the analyses, meaning that our results are independent of the child’s socioeconomic status, sex and age and also any difference in practice between hospitals. The main finding was that over time, the characteristics of the children with an emergency admission were constant but there were obvious differences in the details of the admissions. There has been considerable change in the number and profile of emergency admissions over a short period of time, and these changes are highly likely to continue and place further burden on the healthcare system in Scotland, across the whole of the UK and other European countries, despite differences in health care infrastructures. For example, there are rising admissions of young children with acute infections in Scandinavia [
3,
12] and in Italy there is concern that a considerable minority of admissions may be avoidable [
2].
Rising emergency admissions in the UK have been attributed to a “systematic failure, both in primary care….and in hospital…..in the assessment of children with acute illness [
9]” however reductions in the relative proportion and absolute number of children admitted with some chronic conditions, e.g. asthma and afebrile convulsions, argues that there is capacity to manage children in the community without the need for emergency admissions. The introduction in Scotland of managed clinical networks for paediatric medical specialities in the mid-2000s may be relevant to the fall in emergency admissions with chronic conditions What is not clear is why the number of emergency admissions with some acute conditions has fallen, e.g. croup and gastroenteritis, whereas those for other conditions have risen, e.g. bronchiolitis, lower respiratory tract infection and tonsillitis.
Our results are consistent with results from England described in one published study [
9] and the 2017 report from the Nuffield Trust which also finds increasing absolute numbers of children with emergency admissions due to lower respiratory tract infections and falling numbers admitted with asthma. A study from Spain also reports falling asthma admissions in children and young adults [
8]. Our study findings are also consistent with a year-on-year rise in bronchiolitis admissions in England [
10]. Rises in the number and proportion of emergency admissions lasting less than one [
9] or two days [
17] have been reported in England up to 2010 and our study demonstrates that this trend continues. Previous work has described a greater increase in emergency admissions among younger relative to older children [
9,
15] and whilst we see that the greatest absolute increase in the number of emergency admissions of infants compared to other age groups, the relative increase in emergency admission prevalence was mostly constant across all ages. One result which is not consistent with previous reports is the fall in gastroenteritis emergency admissions in Scotland which is in contrast with a rise England [
9]. The mostly comparable trends in the number of acute admissions in England and Scotland plus similar trends in diagnoses made indicate that our results are generalisable outside Scotland.
This study was not designed to explain why the number of emergency admission is rising, but some of the results might give a steer as to factors which do not contribute large numbers of emergency admissions. For example there was divergence in the trends in duration of stay (falling) and emergency admissions (rising) and these trends might be linked if children were being discharged home “too early” only to be readmitted, and whilst there was a rise in readmissions before 2006, this proportion remained stable from that point onwards and the numbers involved were too small to explain a substantial proportion of the rising emergency admissions seen. A second insight is that the change in readmissions was parallel to that of zero day emergency admissions but the former was less than 10% of the latter, and therefore readmissions are not a substantial reason for rising zero day emergency admissions. The rise in emergency admissions was entirely due to zero day admissions and this might reflect a combination of factors such as (i) more rapid assessment and effective treatment of acute presentations, (ii) changes in health seeking behaviour of parents, (iii) declining expertise and resources in the community to “watch and wait” and (iv) risk aversion among admitting clinicians in the context of availability of acute assessment wards.
Overall, the diagnoses for zero day admissions were similar to those for all admissions although non-specific abdominal pain and rash were only seen among the ten most common diagnoses for zero day admissions. The trends for rising numbers and proportion of all admissions with acute respiratory infections and falls in asthma and gastroenteritis were also seen within zero day admissions (supplemental Table
5). The characteristics of diagnoses for readmissions was different to zero day and all admissions in at least two aspects: first the ten most common diagnoses for readmission explained a smaller proportion (30%) of all readmissions compared to the corresponding proportion for zero day admissions (50%); second the proportion of readmissions with different diagnoses remained roughly unchanged with the exception of bronchiolitis (which rose). These observations suggest that there are different drivers for zero day admission and readmission.
A 2005 review concluded that short stay assessment units might be effective in reducing paediatric emergency admissions [
15] but a 2012 systematic review did not confirm this finding and concluded that there was no clear solution to the question “how do we reduce emergency admission?” [
7]. An observational study published in 2003 which reported activity in one assessment unit suggested that one third of emergency admissions can be immediately sent home [
13] and only 2% were readmitted [
14]. Many UK centres now have paediatric assessment wards or short stay units yet the number of children being admitted continues to rise and this suggests that short stay units are not effective in reducing emergency admissions (and might be part of the reason for rising admission). Interventions “outside” the hospital should now be explored. There is little understood about what factors in primary care affect emergency admissions, although one systematic review suggested that continuity of care with a single primary care clinician was important [
11]. Other potential interventions could include assessment of the child at the point of referral by specialist via videolink and brief educational interventions delivered to clinical decision-makers focussed on specific conditions.
This study has a number of strengths and limitations. We considered the potential for change in trends of coding to cause an apparent fall or rise in a diagnosis presenting to hospital, and we detected a fall in URTI at the same time as admissions with viral infection rose, and also we used clusters of codes to ensure that, for example, we captured any drift in gastroenteritis coding from R52.9 to A09.9. A limitation is that diagnostic coding is known to be imperfect but is sufficiently robust for research and managerial decision-making [
5], and moreover these errors are likely to reduce and not magnify the trends we have described. A second limitation is that some of the change in diagnoses might be due to changes in the true prevalence and not the threshold for emergency admission and the fall in asthma prevalence since 2004 [
1] might at least partly explain the fall in asthma emergency admissions we report here. A third limitation is that we were able to identify emergency admissions within the same calendar month of discharge but not within one month per se, anecdotally most readmissions occur within a few days of discharge so this is unlikely to affect many readmissions but might underestimate the true readmission rates. A final limitation is that although the trends reported here are valid for the whole country and we considered “health board” as a covariate, some aspects of patient management are likely differ to some degree across the regions within Scotland.
In summary, our whole population study has identified that within the overall rise in the number of emergency admissions in children, the patient demographics have not changed substantially but details of the emergency admissions have changed considerably over a 14-year period. Understanding trends in conditions leading to emergency admissions can inform interventions aimed at safely arresting and perhaps reversing these trends.