Article Text
PDF
Abstract
Objectives To determine the indirect consequences of the COVID-19 pandemic on paediatric healthcare utilisation and severe disease at a national level following lockdown on 23 March 2020.
Design National retrospective cohort study.
Setting Emergency childhood primary and secondary care providers across Scotland; two national paediatric intensive care units (PICUs); statutory death records.
Participants 273 455 unscheduled primary care attendances; 462 437 emergency department attendances; 54 076 emergency hospital admissions; 413 PICU unplanned emergency admissions requiring invasive mechanical ventilation; and 415 deaths during the lockdown study period and equivalent dates in previous years.
Main outcome measures Rates of emergency care consultations, attendances and admissions; clinical severity scores on presentation to PICU; rates and causes of childhood death. For all data sets, rates during the lockdown period were compared with mean or aggregated rates for the equivalent dates in 2016–2019.
Results The rates of emergency presentations to primary and secondary care fell during lockdown in comparison to previous years. Emergency PICU admissions for children requiring invasive mechanical ventilation also fell as a proportion of cases for the entire population, with an OR of 0.52 for likelihood of admission during lockdown (95% CI 0.37 to 0.73), compared with the equivalent period in previous years. Clinical severity scores did not suggest children were presenting with more advanced disease. The greatest reduction in PICU admissions was for diseases of the respiratory system; those for injury, poisoning or other external causes were equivalent to previous years. Mortality during lockdown did not change significantly compared with 2016–2019.
Conclusions National lockdown led to a reduction in paediatric emergency care utilisation, without associated evidence of severe harm.
- mortality
- epidemiology
- COVID-19
Data availability statement
Data are available in a public, open access repository. The source code for all the analyses, alongside the anonymised data sets used in these, has been published on GitLab (https://git.ecdf.ed.ac.uk/twillia2/indirect_effects_covid-19_open_data).
This article is made freely available for use in accordance with BMJ’s website terms and conditions for the duration of the covid-19 pandemic or until otherwise determined by BMJ. You may use, download and print the article for any lawful, non-commercial purpose (including text and data mining) provided that all copyright notices and trade marks are retained.
https://bmj.com/coronavirus/usageStatistics from Altmetric.com
What is already known on this topic?
In different settings, the COVID-19 pandemic has lead reduced paediatric emergency department attendances and paediatric intensive care unit (PICU) admissions for respiratory causes.
The pandemic has also led to concerns among clinicians about delayed care seeking for serious paediatric conditions, and a possible increase in non-accidental injuries.
Overall effects of the pandemic on national emergency healthcare utilisation in primary, secondary and critical care, and its impact on severe paediatric disease, remains unquantified.
What this study adds?
We found a reduction in paediatric emergency care utilisation rates at a national level in Scotland, associated with lockdown.
This reduction is likely to be due to combination of changes in healthcare-seeking behaviour, and a fall in the overall burden of paediatric infectious disease.
These measures did not appear to be associated with severe harm to children in Scotland evidenced by severity scores on PICU presentation or overall mortality.
Introduction
The global impact of the coronavirus (SARS-CoV-2) pandemic has been extensive with over 81 million confirmed cases and 1.8 million deaths globally as of 1 Jan 2021.1 The first case of SARS-CoV-2 in Scotland was identified on 1 March 2020,2 with evidence of sustained community transmission 10 days later.3 With a rising number of cases across the UK, a UK-wide lockdown was implemented on 23 March 2020 (figure 1A). This included a number of non-pharmacological interventions (NPIs), such as widespread closures of schools/nurseries, leisure centres, indoor play spaces and non-essential workplaces, as well as promotion of home working, hand washing and wearing of face coverings.3 4 Phased easing of lockdown measures by the Scottish Government began on 29 May 2020; pupils returned to schools on 12 August 2020 (figure 1B).3
Annotated SARS-CoV-2 epidemic curve for Scotland (A) and key events during the lockdown period (B). (A) New positive SARS-CoV-2 RT-PCR cases by day, annotated with key dates and events of relevance for this study. Data from Scottish Government.27 (B) Date ranges for key social distancing measures in Scotland during the lockdown period.3
Despite a significant direct impact of SARS-CoV-2 on adult mortality and morbidity, the paediatric population appear to have been relatively protected during the first wave of the pandemic, with a low number of reported infections, hospitalisation rates and mortality.5 Public Health Scotland (PHS) and National Records of Scotland (NRS) data show that by the end of August 2020 in the 0–14 age range there had been 345 confirmed SARS-CoV-2-positive cases, 43 hospital admissions6 and no deaths.7
While severe disease directly associated with acute SARS-CoV-2 infection in children is rare,5 the health consequences of NPIs, widespread societal lockdown and possible changes in caregiver healthcare-seeking behaviour on children have not been fully quantified. A reduction in asthma exacerbations8 and paediatric intensive care unit (PICU) admissions for respiratory conditions have been seen,9 with both presentations commonly being triggered by respiratory viruses. Paediatric emergency department (ED) attendance rates in Ireland were reduced during the lockdown period compared with previous years.10 Other case series have shown delays in care-seeking behaviour11 and an increase in non-accidental injuries during lockdown,12 suggesting that there may be risks associated with decreased availability of healthcare and lower levels of surveillance from health, education and social care workers during this period. Our study used complete national data sets to quantify the indirect effects of the COVID-19 pandemic on emergency paediatric healthcare utilisation rates and paediatric morbidity and mortality in Scotland during the lockdown period.
Methods
Data sources and data collection
The overall design was a retrospective, population-based analysis of all emergency paediatric healthcare utilisation. We examined unscheduled primary care and ED attendances, emergency hospital admissions, emergency paediatric intensive care (PICU) admissions requiring invasive mechanical ventilation and paediatric mortality for all children in Scotland. Disclosure controlled, aggregate count data of relevant healthcare utilisation events (apart from PICU admissions) were provided by PHS. The data used are publicly available on request, with no ethical permissions required. Emergency hospital admissions data are derived from the Rapid Preliminary Inpatient Data (RAPID) data set.13
Data from the two PICUs in Scotland were collected, the Royal Hospital for Children (NHS Greater Glasgow and Clyde) and the Royal Hospital for Children and Young People (NHS Lothian), representing the entire PICU service provision for Scotland and complete national data for children requiring critical care. Data on unplanned, emergency admissions requiring invasive mechanical ventilation were collected. Analysis was performed as part of a service evaluation programme and in concordance with both NHS Greater Glasgow and Clyde and NHS Lothian quality improvement policies, so ethical approval is not required. PICU data were collected in an electronic clinical information system MetaVision provided by iMDsoft and in the Paediatric Intensive Care Audit Network (PICANet) database.14 NRS provided detailed childhood mortality data, stratified by age group (0–4 and 5–14 years), and by cause of death classified by International Classification of Diseases 10th Revision (ICD-10) code. A number of deaths for 2019 and 2020 had not yet reached final coding stage and therefore were documented as ‘Not yet classified’. National data on midyear population estimates by age group were referenced from an NRS data set.15
Study design
Calculation of population at risk
The paediatric population was calculated by taking the number of children within each age category from the midyear population estimates from NRS for each year. For the year 2020 to date, the population at risk was estimated by taking the midyear population estimates for years 0–13 for 2019, equating to ages 1–14 in 2020. To estimate the midyear population of infants aged <1 in 2020, the ratio of the number of births in weeks 1–12 in 2020 to those in the equivalent period in 2019 was calculated and applied to the midyear population for age group <1 year in 2019.
Study outcomes
Data for healthcare utilisation and mortality rates were examined for all children aged 0–14 years living in Scotland. Rates of healthcare use, including unscheduled primary care and ED attendance, and emergency hospital admissions were analysed. Healthcare utilisation rates from the period 23 March to 9 August 2020 (epidemiological weeks 13–32, corresponding to the start of lockdown, and the reopening of schools) were compared with the mean of the equivalent period in previous years, calculated as a rate per 100 000 children at risk. Unscheduled primary care (including telephone consultations) and ED attendances were available from 1 January 2016 onwards. Emergency hospital admission rates were obtained from the RAPID data set for years 2018–2020. Attendance and admission rates in all settings were calculated by referencing the same epidemiological weeks for previous years.
Data referencing diagnosis codes for PICU admissions were available until 30 June 2020. The rate of admissions for the period 23 March to 30 June 2020 was compared with the equivalent epidemiological weeks in 2016–2019. Admission diagnoses were classified according to ICD-11 diagnostic codes.16 Illness severity at presentation to PICU was determined by reference to the Paediatric Index of Mortality-2 Recalibrated (PIM2R) score,17 a score calculated for purposes of submission to PICANet using a recalibration dating to 2016 (PIM2R 2016).18 As PICU admissions occur up to the age of 16 years, the population at risk used for calculations was that of children aged 0–15 years. Childhood mortality data, categorised by age groups 0–4 and 5–14 years, were available for epidemiological weeks 13–30 in 2020 (23 March to 31 July) and the equivalent periods in 2016–2019 for comparison. Codes for the cause of death were categorised by ICD-10 code.19 Mortality rates were normalised by the total population at risk as described above.
Statistical analysis
Statistical analysis was conducted using R V.3.4.120 and GraphPad Prism V.6.04 (La Jolla, California). For unscheduled primary care consultations, ED attendances and emergency admissions, geometric means with 95% CIs were used to compare epidemiological weeks 5–32 (2 February to 9 August 2020) with equivalent weeks in 2016–2019. Two-sided Fisher’s exact tests with an alpha value of 0.05 were used to compare events in the lockdown period to previous years (epidemiological weeks 13–32, equating to 23 March to 9 August 2020). PICU admissions were analysed using geometric means with 95% CIs used to compare admissions by month (February to June 2020) to equivalent months in 2016–2019. A Fisher’s exact test was used to compare the number of admissions (23 March to 30 June 2020) with admissions during equivalent weeks in 2016–2019. A Kolmogorov-Smirnov test was used to compare mean severity scores on presentation to PICU. Childhood mortality rates from epidemiological weeks 13–30 (23 March to 26 July 2020) were compared with the equivalent periods in 2016–2019. A z-score was calculated for childhood deaths and PICU admissions by category in 2020, by subtracting the 2016–2019 mean from the value for 2020 and by dividing the difference by the SD. P values were calculated using the pnorm function in R.20 All scripts and data used for the analysis are published on GitLab (https://git.ecdf.ed.ac.uk/twillia2/indirect_effects_covid-19_open_data). This study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology statement.21
Results
Reduction in primary care out-of-hours consultations and ED attendances
The onset of lockdown on 23 March 2020 (occurring 1 hour after the end of epidemiological week 12, figure 1A) until the end of the study period was associated with an almost two-thirds reduction in the number of unscheduled primary care attendances (OR 0.36, 95% CI 0.35 to 0.37, p<0.001) when compared with epidemiological weeks 13–32 in 2016–2019 (figure 2A). For the same period, there was a reduction in ED attendances compared with previous years (OR 0.49, 95% CI 0.48 to 0.49, p<0.001) (figure 2B).
Unscheduled primary care and emergency department (ED) attendances for Scottish children during/after lockdown. (A) Out-of-hours (OOH) unscheduled primary care consultations. Number of unscheduled primary care consultations by epidemiological week compared with mean number of consultations, with 95% CIs, for 2016–2019. Data shown from a total of 273 455 consultations (including telephone consultations). (B) ED attendances. Data shown as for (A) for a total of 462 437 attendances. (C) Emergency medical paediatric admissions. Number of emergency medical paediatric admissions in 2020, compared with admissions in 2018 and 2019. Data shown for a total of 51 596 admissions. (D) Emergency surgical paediatric admissions. Data presented as in (C) for a total of 2480 admissions. All data in (A)–(D) normalised by population at risk per year. (E) Proportion of hospital admissions to ED attendances.
Reduction in emergency hospital admissions for medical but not surgical diagnoses
Comparison of nationwide emergency medical paediatric admissions showed a reduction in these during the lockdown period (figure 2C), with an OR of 0.56 for an admission occurring in epidemiological weeks 13–32, compared with equivalent weeks in 2018–2019 (95% CI 0.55 to 0.57, p<0.001). There was no equivalent reduction in paediatric surgical admissions in the postlockdown period overall, with an OR for admission of 1.03 (95% CI 0.95 to 1.12, p=0.46) (figure 2D).
Proportion of ED attendances to hospital admission during lockdown
As a marker for seriousness of symptoms on presentation to the ED, the proportion of the number of children admitted as an emergency to the number who attended the ED was calculated (for data from 2018 to 2020) (figure 2E). In the early weeks of lockdown, there was a rise in the proportion of children admitted to hospital relative to ED attendances, with a peak in the first week of lockdown (epidemiological week 13; 23–29 March 2020). In this week, there was a 1.74-fold increase in the odds of an emergency admission as a proportion of ED attendances, compared with 2018–2019 (95% CI 1.57 to 1.92, p<0.001). Comparing ED activity and the proportion of hospital admissions to ED attendances, by epidemiological week 22 the activity for ED attendances was still at a lower rate than the 2016–2019 mean (figure 2B); however, the proportion of admissions to attendances returned to baseline levels for years 2016–2019 (figure 2E).
Reduction in emergency admissions to PICU for children requiring invasive mechanical ventilation but no increase in severity scores on presentation
Comparison of the number of events in epidemiological weeks 13–26 (23 March to 30 June 2020; figure 3A) with equivalent periods in 2016–2019 showed that the OR of an emergency admission requiring invasive mechanical ventilation during the lockdown study period was 0.52 (95% CI 0.37 to 0.70, p<0.001). The most marked decrease (by 77%) was seen in admissions for disorders of the respiratory system (ICD-11 Chapter 12) (p<0.001) (figure 3B, table 1). There was also a decrease in admissions for disorders of the neurological system (ICD-11 Chapter 8, p=0.01, table 2). No change was seen when compared with previous years in admissions for injury, poisoning or certain other consequences of external causes (ICD-11 Chapter 22, table 3). The PIM2R score for PICU emergency admissions requiring invasive mechanical ventilation during the lockdown study period did not differ from the equivalent epidemiological weeks in 2016–2019 (p=0.23, Kolmogorov-Smirnov test) (figure 3C). During the lockdown study period there were no emergency admissions requiring invasive mechanical ventilation for bronchiolitis, lower respiratory tract infection or respiratory failure (table 1).
Paediatric intensive care unit (PICU) admissions for Scottish children. (A) The overall number of emergency admissions to PICU requiring invasive mechanical ventilation in March to June 2020 fell when compared with the mean for previous years. Geometric mean and 95% CIs shown for 2016–2019, expressed as number of admissions per 100 000 children at risk. Data shown for a total of 413 admissions during the study period. (B) The reduction in admissions was most marked for diseases of the respiratory system. Geometric mean and 95% CIs shown for five most common International Classification of Diseases 11th Revision (ICD-11) chapters. There was a reduction in overall admissions for respiratory causes (77% reduction), and no change in admissions for injury, poisoning or other external causes. P values calculated from z-scores for 2020 compared with means and SD for 2016–2019. (C) A reduction in admissions was not associated with an increase in severity scores on arrival in the PICU. Median and IQRs shown for log10 Paediatric Index of Mortality-2 Recalibrated (PIM2R) scores on admission for admissions between 23 March and 30 June 2020 compared with the equivalent period in 2016–2019. Distributions compared using a Kolmogorov-Smirnov test, p=0.23.
Emergency admissions for Scottish children requiring invasive mechanical ventilation falling within ICD-11 Chapter 12 (disorders of the respiratory system)
Emergency paediatric intensive care unit (PICU) admissions in Scotland requiring invasive mechanical ventilation, diagnoses within ICD-11 Chapter 18 (disorders of the neurological system)
Emergency paediatric intensive care unit (PICU) admissions in Scotland requiring invasive mechanical ventilation, diagnoses within ICD-11 Chapter 22 (injury, poisoning or certain other consequences of external causes)
Childhood mortality
Childhood mortality in epidemiological weeks 13–30 in 2020 did not differ significantly from mean in equivalent weeks in 2016–2019, for overall deaths, deaths in ages 0–4 years and deaths in ages 5–14 years (figure 4A). Of note, the cause of death for a proportion of cases in years 2019 (10/70) and 2020 (14/78, online supplemental table 1) had not yet been finalised and therefore could not be included (figure 4B, grey bars).
Supplemental material
Childhood mortality during lockdown study period in Scotland. (A) There was no significant change in child mortality in the lockdown study period compared with the equivalent period in previous years. Number of deaths per 100 000 children in Scotland in epidemiological weeks 13–30 for 2020, compared with the 2016–2019. Horizontal lines show means with 95% CIs for years 2016–2019. P values calculated from z-scores for 2020 compared with means and SD for 2016–2020. (B) Childhood deaths by cause in 2020 compared with previous years. Deaths per 100 000 children classified by International Classification of Diseases 10th Revision (ICD-10) chapter. A number of deaths for 2019 and 2020 have not yet been classified by ICD-10 code and fall into the ‘Not yet classified’ group.
Discussion
Overall emergency paediatric healthcare utilisation by children in Scotland reduced significantly during the lockdown period, with no immediate, measurable severe adverse effects on physical child health at a population level observed. Specifically, a reduction in unscheduled primary care and ED attendances, and unplanned hospital admissions did not result in increased presentations or increased disease severity on presentation to PICU, or mortality rates. The number of admissions to PICU with respiratory disease in those requiring invasive mechanical ventilation during the lockdown study period was significantly reduced. This finding supports evidence from recent reports,9 which are consistent with the hypothesis that widespread introduction of NPIs is associated with an overall reduction in non-COVID-19-related respiratory infections. During the same period, no increase was seen in children admitted to PICU within the category of injury, poisoning or other external causes, which may have been expected if lockdown was associated with reduced surveillance by health, education and social care workers. Our findings support those of an English study which examined paediatric ED presentation and reported a very small number of presentations with concerns relating to safeguarding.22 However, in both our study and that by Roland et al,22 a low incidence of attendances/admissions for this category of conditions limits our ability to make inferences about the consequences of reduced surveillance by healthcare professionals.
In our study, no significant changes were seen in paediatric mortality for this period across any age group examined. This was despite the PICU admission rate being decreased. As disorders of the respiratory and nervous systems contribute disproportionately to PICU admission, but are under-represented in overall mortality, one would not necessarily expect a reduction in the burden of this category of disease to equate to a lower overall mortality.
The causes for the observed reduction in rates of paediatric emergency healthcare utilisation during the lockdown study period are likely to be multifactorial. Changes in patterns of interaction between children, shielding measures for at-risk groups and restricted domestic and international travel during the lockdown study period are likely to have reduced the transmission rates of infectious pathogens, and may have reduced exposure to some of the triggers for respiratory diseases (eg, air pollution in asthma).8 Increased parental supervision may have improved adherence with medical therapies and thus reduced exacerbation frequency by improved compliance with prescribed medications, such as anticonvulsants. Reduction in attendance rates to healthcare services might also be associated with caregiver reluctance to attend healthcare settings in the context of a viral pandemic. Reduced attendance rates without notable impact on severity of presentation might suggest that caregivers were able to use a higher threshold for seeking medical attention, therefore decreasing the burden on health systems for children without increasing harms. However, the observed changes in healthcare utilisation outlined in this study appear mainly related to medical issues since surgical presentations remain relatively unchanged (figure 2C,D). This discrepancy suggests that the interruption of pathogen transmission by NPIs was probably the most significant factor in reducing medical presentations, rather than an alteration in caregiver care-seeking behaviour.
Our study uses high-quality, standardised population data from a variety of sources which address the study question from a number of different viewpoints. The use of complete national data sets allowed us to examine the entire paediatric population of Scotland during the lockdown study period and compare this with data collected in previous years. The results from all analyses are concordant with each other and similar analyses,23 highlighting a picture of reduced healthcare seeking without short-term increase in severe paediatric disease or mortality.
Although further subcategorisation of age groups would provide information on how presentation rates differed by age, unfortunately more granular data on age groupings were not available within the data sets used. Minor differences in the coding of PICU admission (ICD-11) and causes of mortality (ICD-10) may make these groups harder to compare. Similarly, mortality data were incomplete for 2019–2020, with some deaths still to be attributed a cause. It is possible that these missing deaths (online supplemental table 1), once coded, could influence our findings. Another limitation of this study is the inability to examine non-severe disease and there is a growing body of evidence that the lockdown period may have had a detrimental effect, both in the short term and the long term,24 particularly on the mental health of children.25 26
Conclusion
Our analysis shows a reduction in paediatric emergency care-seeking utilisation that occurred as a consequence of the widespread adoption of NPIs and societal lockdown measures. This reduction is likely to be due to a combination of changes in healthcare-seeking behaviour, and a fall in the overall burden of infectious causes of childhood disease. These measures do not appear to have been associated with evidence of severe harm to children in Scotland during the lockdown period.
Data availability statement
Data are available in a public, open access repository. The source code for all the analyses, alongside the anonymised data sets used in these, has been published on GitLab (https://git.ecdf.ed.ac.uk/twillia2/indirect_effects_covid-19_open_data).
Ethics statements
Acknowledgments
We would like to thank Julie Ramsay and Marie Kay from National Records of Scotland, and Kathy McGregor and Jaime Villacampa from Public Health Scotland, for their help in securing the data underlying these analyses, and for their comments on the manuscript. We thank Andrew Bretherick for helpful discussions regarding the statistical approaches used, and Rachael Wood for insightful comments into an initial draft of the manuscript.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Twitter @williams_t_c
BG and RJL contributed equally.
Contributors RJL and TCW conceived the study. CMR, OS, SC, PD, NG, SWT, DSU and BG made substantial contributions to the design of the work. CL, RL, CMM, CMR, JMF, IP, AT and MVD contributed towards acquisition of data for the work. TCW, CMR and HH performed the analysis and interpretation of the data. RJL, TCW, CMR and OS drafted the manuscript. SC, PD, NG, SWT, DSU, BG, CL, RJL, CMR, JMF, IP, AT and MVD revised it critically for important intellectual content.
Funding TCW is the recipient of a Wellcome Trust Award (204802/Z/16/Z).
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.