You are here

Article Text

Article menu

Download PDFPDF

Original research
Is socioeconomic position associated with bronchiolitis seasonality? A cohort study
  1. Kate Lewis,
  2. Bianca De Stavola,
  3. Pia Hardelid
  1. Population, Policy and Practice Research and Teaching Department, UCL Great Ormond Street Institute of Child Health, London, UK
  1. Correspondence to Kate Lewis, Department of Population, Policy and Practice, UCL Great Ormond Street Institute of Child Health, 30 Guilford Street, London WC1N 1EH, UK; kate.lewis.14{at}ucl.ac.uk

Abstract

Background Understanding differences in the seasonality of bronchiolitis can help to plan the timing of interventions. We quantified the extent to which seasonality in hospital admissions for bronchiolitis is modified by socioeconomic position.

Methods Using Hospital Episode Statistics, we followed 3 717 329 infants born in English National Health Service hospitals between 2011 and 2016 for 1 year. We calculated the proportion of all infant admissions due to bronchiolitis and the incidence rate of bronchiolitis admissions per 1000 infant-years, according to year, month, age, socioeconomic position and region. We used harmonic Poisson regression analysis to assess whether socioeconomic position modified bronchiolitis seasonality.

Results The admission rate for bronchiolitis in England increased from 47.4 (95% CI 46.8 to 47.9) to 58.9 per 1000 infant-years (95% CI 58.3 to 59.5) between 2012 and 2016. We identified some variation in the seasonality of admissions by socioeconomic position: increased deprivation was associated with less seasonal variation and a slightly delayed epidemic peak. At week 50, the risk of admission was 38% greater (incidence rate ratios 1.38; 95% CI 1.35 to 1.41) for infants in the most deprived socioeconomic group compared with the least deprived group.

Conclusion These results do not support the need for differential timing of prophylaxis or vaccination by socioeconomic group but suggest that infants born into socioeconomic deprivation should be considered a priority group for future interventions. Further research is needed to establish if the viral aetiology of bronchiolitis varies by season and socioeconomic group, and to quantify risk factors mediating socioeconomic deprivation and bronchiolitis rates.

  • Infection
  • inequalities
  • respiratory DI
  • social epidemiology
  • child health

http://dx.doi.org/10.1136/jech-2019-213056

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    INTRODUCTION

    Bronchiolitis is an acute lower respiratory tract infection that commonly affects children under 1 year.1 Mild cases of bronchiolitis can be managed at home, but a significant number of infants require supportive hospital care to aid feeding and respiration.2 In 2011, 3.8% of infants aged less than 1 year required admission to hospital for bronchiolitis in England.3 4 Respiratory syncytial virus (RSV)—a virus for which there is currently no vaccine available—causes an estimated 80% of bronchiolitis cases that require hospitalisation.1 5 A monoclonal antibody called palivizumab (Synagis) can reduce the risk of severe infection during periods of RSV circulation; however, it is costly and therefore only recommended for infants at high risk of serious complications.6 7 Notably, severe symptoms of RSV infection are associated with an increased risk of wheeze and asthma in later childhood.

    In temperate climates in the Northern Hemisphere, such as the UK, RSV circulation peaks between December and February,8 leading to an increased burden on both primary and secondary care services during winter months.3 9 Previous research shows that children from lower socioeconomic groups are at particular risk of admission to hospital and paediatric intensive care for bronchiolitis,3 10 but it is not known whether differences also exist in the seasonality of bronchiolitis admissions. RSV transmission may differ because of socially patterned risk factors, such as residential overcrowding and family size, thereby leading to different seasonal patterns of admissions.7 11 Understanding these patterns could help to inform preventive interventions to reduce bronchiolitis admissions, including optimal timing of palivizumab prescribing in different population groups, or targeting of maternal RSV vaccination, when this becomes available in the future.12

    The aim of this study was to determine how socioeconomic position is associated with the seasonality of hospital admissions for bronchiolitis among infants in England. Our objectives were twofold: to describe seasonality of infant bronchiolitis hospital admissions in England from 2012 to 2016; and quantify the extent to which this seasonality is modified by socioeconomic position.

    METHODS

    Data source

    We used Hospital Episodes Statistics Admitted Patient Care (HES APC) data to identify bronchiolitis admissions and the infant population at risk. HES APC is a database of all hospital inpatient admissions funded by the English National Health Service (NHS) and captures approximately 97% of all births in England.13 14 NHS Digital, the body that houses and supplies access to HES, links patient records over time and provides each database extract with a pseudo-anonymised identifier (HESID) unique to each individual, enabling researchers to create longitudinal patient cohorts. HES APC was linked to the Office for National Statistics mortality records.15

    Cohort population and follow-up

    A birth cohort of all live singleton births recorded in HES APC between 1 January 2011 and 31 December 2016 was created from HES APC. To enhance data completeness, we linked infant birth records to mothers’ delivery records following methods described in Harron et al. 16 Using these methods, 43.5% of birth and delivery records were deterministically linked, 51.5% were probabilistically linked and 5.0% were unlinked. Using the infant’s HESID, data on admissions to hospital within the first year of life were added to birth data; figures from other sources show that 90% of bronchiolitis admissions occur in the first year of life and two-thirds of admissions are in children less than 6 months old.5 17

    Infants were followed from birth or 1 January 2012, whichever occurred last, until their first birthday, date of out-migration, 31 December 2016 or date of death, whichever occurred first. This study period ensured that we had follow up for children aged up to 12 months in all study years, including the first. Infants were excluded from cohort analyses if they had missing information, were from a multiple birth or were stillborn.18 Infants with a non-English address recorded at birth were excluded to prevent potential loss to follow-up. Where a non-English address was recorded in a subsequent infant admission, a censoring date was placed halfway between the date of the infant’s last known resident admission and the non-resident admission.

    Outcome

    We identified hospital admissions with a diagnosis of bronchiolitis in HES APC using the International Classification of Diseases version 10 code J21 for acute bronchiolitis. All J21 subcategories were included in our definition (J21.0 acute bronchiolitis due to RSV, J21.1 acute bronchiolitis due to human metapneumovirus, J21.8 acute bronchiolitis due to other specified organisms and J21.9 acute bronchiolitis, unspecified) because of the low sensitivity of RSV-specific ICD-10 codes.19 To calculate the total burden of infant admissions attributable to bronchiolitis, we extracted all hospital admissions (excluding birth admissions) between 1 January 2012 and 31 December 2016 where the patient was ≤365 days old. All admissions with bronchiolitis recorded as either the primary or secondary diagnosis for an infant during their first year of life were included in the analyses. To calculate admission rates, any admission for RSV bronchiolitis within 14 days of discharge from another RSV bronchiolitis admission was assumed to be associated with the same infection, and therefore only the first of these admissions was included in the analyses.

    We used the Index of Multiple Deprivation (IMD) 2010 to capture socioeconomic position for each cohort member. IMD is constructed by assigning to each lower super output area (approximately 650 households and 1500 residents) across England a composite score summed from seven domains of deprivation: income, employment, health and disability, education skills and training, barriers to housing and services, living environment and crime.20 IMD was derived from mother’s postcode at delivery or the earliest mention of postcode in infant’s hospital admission. IMD was split into fifths for analyses; highest ranked IMD means the most deprived category.

    Infant sex (male or female), month and year of birth were extracted from infant’s birth record. Infant’s age at admission for bronchiolitis was calculated by subtracting the bronchiolitis admission date from the infant’s admission date in their birth episode. Age was split into three groups: <3 months, 3 to <6 months and 6 to <12 months. Government office region of residence was used to indicate the area of England within which the infant lived at time of birth.21 HES APC includes information on additional risk factors such as gestational age, mode of delivery and congenital anomalies; however, as these variables are likely on the causal pathway between socioeconomic position and admission for bronchiolitis,22–24 they were not included in our analyses. Inclusion of mediators in associational analyses introduces overadjustment bias into the model.25

    Statistical analysis

    To calculate the proportion of all infant admissions attributed to bronchiolitis, we divided the number of bronchiolitis admissions (including those within 14 days of one another) by the total number of hospital admissions among infants. Bronchiolitis admission rates were calculated by dividing the number of new admissions for bronchiolitis by person-time at risk for all infants at risk in the birth cohort, and are expressed as annual admission-based rates per 1000 infant-years. We also calculated person-based rates where, for infants with multiple admissions, only the first admission was counted. Stata 15.026 was used for data analysis and Microsoft Excel 2013 to create graphs.

    Event rates were modelled using a Poisson regression model with robust SEs to account for multiple admissions for the same child. Negative binomial models were also considered, but there was no evidence of overdispersion in event rates. The impact of seasonality on rates of admission was modelled using a harmonic function of time in weeks (t) expressed as β 1(sin(2πt/T))+β 2(cos(2πt/T)), where T is the number of periods within one cycle (ie, 1 year=52.14 weeks).27 28 The model included interaction terms of IMD group with the sine and cosine regression coefficients (denoted δ 1j and δ 2j) to assess evidence of effect modification of IMD group by seasonality, formally tested using a Wald χ 2 test. Year of admission, region, sex, month of birth, age group and interaction terms of age group with the sine and cosine regression coefficients were selected a priori as covariates in the model to increase precision given their known associations with the outcome as identified in the literature.3 7 29 See online supplementary Box S1 for full model parameterisation.

    Using the estimated model coefficients, we calculated the following quantities for the epidemic curve of each IMD group j=1, …, 5 at reference values of the other covariates (where betas, gammas and deltas are replaced by their estimates)30: the amplitude (log)=√((β 1+δ 1j)2+(β 2+δ 2j)2); the phase (in radians)=arctan(β 1+δ1j )/(β 2+δ 2j); the peak week =52.14*(phase j /2π)+1; and incidence rate ratios (IRR) of admissions at the average peak week of the top four IMD groups relative to the lowest IMD group. The delta method, as implemented by the Stata command lincom, was used to calculate 95% CIs for each parameter. Epidemic duration for each IMD group was calculated as the time in weeks from the first of 3 consecutive weeks with increasing predicted rates to the first of 3 consecutive weeks with decreasing predicted rates.

    RESULTS

    Our cohort comprised 3 717 329 singleton infants, of which 48.7% were female and 19.0% were born to mothers residing in London (table 1). Missing information about IMD, region or sex was present for 67 422 (1.8%) infants, who were excluded from further analyses (see online supplemental figure S1). The average annual proportion of all infant hospital admissions in England attributed to bronchiolitis was 15.0% over the five study years. The burden of bronchiolitis was concentrated in the winter months. In December 2012–2016 combined, 39.8% of the total admissions for infants included a diagnosis of bronchiolitis compared with 2.3% of August admissions over the same period (online supplemental table S1).

    View this table:
    Table 1

    Distribution of births and bronchiolitis admissions in the cohort, with bronchiolitis admission rates per 1000 infant-years

    Supplemental material

    There were 155 479 admissions for bronchiolitis by cohort members over the study period—an average annual admission-based rate of 50.3 admissions per 1000 infant-years (95% CI 50.0 to 50.5). The average follow-up time per infant was 304.5 days (indicating that most infants were followed up until their first birthday) and 12 279 (7.9%) of events were readmissions. The overall infant-based admission rate was 46.3 per 1000 infant-years (95% CI 46.1 to 46.6). Rates of admissions were highest in the most socioeconomically deprived groups, males, younger infants and those residing in the North West or North East of England. The average admission rate was 79.3 per 1000 infant-years (95% CI 78.3 to 80.4) for infants born in October compared with 33.7 (95% CI 33.0 to 34.4) for March births. Bronchiolitis admission rates increased from 47.4 per 1000 infant-years (95% CI 46.8 to 47.9) in 2012 to 58.9 per 1000 (95% CI 58.3 to 59.5) in 2016.

    Figure 1 displays weekly crude rates of admissions by IMD, illustrating differential rates of admissions by the level of socioeconomic position across the year and a clear annual peak in December. The multivariable model that included interaction terms between IMD and the two harmonic functions fitted the data better than the model without interactions, Wald test, χ 2(8)=238.07, p<0.001 (table 2). Fitted values and seasonal estimates from the model are presented in figure 2 and table 3. The amplitude of the epidemic curve ranged from 3.33 (95% CI 3.25 to 3.41) in the most deprived to 4.05 (95% CI 3.89 to 4.21) in the least deprived group, pointing to less seasonal variation in the rates of admissions in the more disadvantaged groups. The phase shift is greatest in the most deprived, corresponding to the slightly delayed peak timing. The average peak timing of the annual epidemic varied marginally across the groups, from week 49.5 (95% CI 49.3 to 49.6) in the least deprived to week 50.2 (95% CI 50.1 to 50.4) in the most deprived group. The estimated epidemic duration was 26 weeks across all groups. After adjustment for covariates, infants in the most deprived group had a 1.38 (95% CI 1.35 to 1.41) greater risk of admission to hospital for bronchiolitis at week 50 compared with infants in the least deprived group.

    View this table:
    Table 2

    IRRs from multivariable harmonic Poisson regression model with 95% CIs

    View this table:
    Table 3

    Derived average annual seasonal estimates following Poisson regression*, by IMD group

    Figure 1
    Figure 1

    Observed weekly rates of hospital admission for bronchiolitis, by IMD: England 2012–2016. IMD, Index of Multiple Deprivation.

    Figure 2
    Figure 2

    Predicted weekly rates of hospital admission for bronchiolitis, by IMD: England 2012–2016. IMD, Index of Multiple Deprivation.

    DISCUSSION

    This study presents a continuation of previous findings that show ever-increasing rates of bronchiolitis admissions to hospitals in England,3 with admission rates reaching 58.9 per 1000 infant-years in 2016. The data present a single annual peak in bronchiolitis admissions across England in mid-December, when 40% of all infant admissions included a bronchiolitis diagnosis. There is a clear socioeconomic gradient to these admissions, and at week 50 (peak admission week), infants born into the highest level of socioeconomic deprivation had a risk of admission 38% greater than the lowest group. Our results suggest that the association between seasonality and bronchiolitis admission rates is marginally moderated by the level of socioeconomic position, with increasing socioeconomic deprivation associated with less seasonal variation and a slightly delayed epidemic peak.

    Our study data were derived from a national hospital administrative data set, which enabled us to create a representative birth cohort with minimum selection bias and apply linkage methodology, which was crucial for acquiring socioeconomic data as this information was missing for infant records before 2014 (but available in maternal records). However, the probabilistic method used is likely to have included erroneous links, introducing some bias into the data set.31 Using IMD as a proxy for individual socioeconomic position may have led to a weakened association between socioeconomic position on rates of bronchiolitis admissions.32 Defining RSV bronchiolitis without laboratory testing meant that we could not attribute the bronchiolitis admissions to particular pathogens, although previous work from England shows that 80% of bronchiolitis admissions in infants (defined using ICD-10 code J21) are due to RSV.3 5 More broadly, this study is a measure of hospital use and therefore does not measure community, or primary care burden of bronchiolitis.

    We added harmonic functions to our regression model to account for sinusoidal patterns in a highly seasonal infection33; however, our model underestimates the size of the amplitude across socioeconomic groups. Improved estimation could have been achieved with the addition of more harmonic pairs or by using splines; however, these methods would have added undue complexity to the estimation of seasonal parameters required by our research aim.29 30 The likeness in the relative difference between amplitudes by IMD across observed and fitted rates (figures 1 and 2) gives us confidence in our results.

    This is the first study investigating seasonality and socioeconomic position in relation to bronchiolitis admissions in the English population. Our work presents a small, but not consequential, difference in epidemic timing—the two most extreme IMD groups had a 0.7 week difference in predicted peak—and no difference in the relative duration of the epidemic. Less seasonal variation in lower socioeconomic groups may reflect continued (but low level) admissions for bronchiolitis during warmer months of the year among these infants. Understanding precise viral aetiology of bronchiolitis by time of the year and socioeconomic group may help to delineate seasonal differences further. Future work would benefit from linkage to surveillance data sets, such as Public Health England’s second-generation surveillance or Respiratory DataMart Systems, to assess these factors.34 35

    Research from Western Australia found no difference in the seasonality of RSV-positive specimens among aboriginal children (a population that experience high levels of socioeconomic deprivation) compared with non-aboriginal children.29 US-based research, on the other hand, found a negative association between the proportion of the population from a black ethnic group (correlated with socioeconomic position of the population) and seasonal peak timing at the ZIP-code level.36 We have previously investigated the relationship between socioeconomic deprivation and seasonality by geographical area, finding a 3-week difference in peak timing across Clinical Commissioning Groups (CCGs) in England.37 In this study, CCGs with both low and high IMD scores were associated with earlier peak timing of bronchiolitis, after adjustment for population density. However, these two factors explained less than 40% of the variation in the timing of epidemic peak, highlighting the potential role of other unmeasured factors such as living in areas with major transport connections.

    In the context of current UK guidelines for administering palivizumab, our results present little evidence to support differential timing of interventions for RSV bronchiolitis among socioeconomic groups in England at the current time. Our work does, however, present substantial difference in admission rates across population subgroups irrespective of season. This study highlights that, at the very least, infants from poorer backgrounds should be considered a priority group for future interventions. Based on previous research, we hypothesise that a combination of individual risk factors, such as prematurity and presence of congenital anomalies, and environmental factors, such as exposure to tobacco smoke and housing conditions, contribute to inequities.3 11 38 39 Further work using formal methods to establish pathways through which these factors affect the risk of bronchiolitis is needed to guide the most appropriate interventions. We also demonstrate that the previously observed increase in bronchiolitis admissions in England is continuing through to 2016,3 highlighting the need to stem the flow of bronchiolitis admissions and the urgent need for a vaccine against RSV. Further, other research has shown that hospital and healthcare policies such as changes in admission thresholds, accessibility of primary care services and hospital bed availability are an important driver of this increase, rather than changes in the transmissibility or severity of RSV.3 38 Evaluations of alternative models of care for acutely ill infants which reduce the need for hospital admission could therefore also contribute to the reduction in bronchiolitis admissions.40

    CONCLUSION

    Our study used a population-based cohort, created from administrative data, to investigate the association between seasonality and socioeconomic position on bronchiolitis admissions. We have updated current knowledge on the burden of this infection and have presented results using harmonic analysis, which allowed us to model seasonality—an essential component of RSV bronchiolitis. Our results suggest that differential timing of interventions may not be necessary for the groups studied but highlight the continued gap in admission rates between socioeconomic groups. Moving forward, investigating the precise viral aetiology of bronchiolitis by season and socioeconomic group, as well as the risk factors mediating the link between socioeconomic deprivation and bronchiolitis rates, will aid understanding further of this complex condition.

    What is already known on this subject

    • Bronchiolitis causes a substantial burden on the English health system during winter.

    • In addition to seasonality, socioeconomic position is associated with the risk of hospital admission for bronchiolitis.

    What this study adds

    • Almost 4 in every 10 infant admissions to hospitals in England during December include a diagnosis of bronchiolitis.

    • The seasonality of admissions varied marginally by socioeconomic group; the predicted peak epidemic week was 49.5 among the least deprived and 50.2 among the most deprived infants.

    • Socioeconomic disparities in admissions to hospital for bronchiolitis persist after taking seasonality into account.

    Acknowledgments

    Research at UCL Great Ormond Street Institute of Child Health is supported by the NIHR Great Ormond Street Hospital Biomedical Research Centre. This work uses data provided by patients and collected by the NHS as part of their care and support. This research benefits from and contributes to the NIHR Children and Families Policy Research Unit but was not commissioned by the NIHR Policy Research Programme. The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research, the Department of Health and Social Care or its arm’s length bodies, and other government departments.

    REFERENCES

    1. ↵
      2. Smyth RL ,
      3. Openshaw PJM
      . Bronchiolitis. Lancet 2006;368:312–22. doi: 10.1016/S0140-6736(06)69077-6
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      2. Ricci V ,
      3. Delgado Nunes V ,
      4. Murphy MS , et al.
      Bronchiolitis in children: summary of NICE guidance. BMJ 2015;350:h2305–h2305. doi: 10.1136/bmj.h2305
      OpenUrlFREE Full Text
    3. ↵
      2. Green CA ,
      3. Yeates D ,
      4. Goldacre A , et al.
      Admission to hospital for bronchiolitis in England: trends over five decades, geographical variation and association with perinatal characteristics and subsequent asthma. Arch Dis Child 2016;101:140–6. doi: 10.1136/archdischild-2015-308723
      OpenUrlAbstract/FREE Full Text
    4. ↵
      2. Moore HC ,
      3. Hall GL ,
      4. de Klerk N
      . Infant respiratory infections and later respiratory hospitalisation in childhood. European Respiratory Journal 2015;46:1334–41 doi: 10.1183/13993003.00587-2015.
      OpenUrlAbstract/FREE Full Text
    5. ↵
      2. Reeves RM ,
      3. Hardelid P ,
      4. Gilbert R , et al.
      Estimating the burden of respiratory syncytial virus (RSV) on respiratory hospital admissions in children less than five years of age in England, 2007–2012. Influenza Other Respi Viruses 2017;11:122–9. doi: 10.1111/irv.12443
      OpenUrl
    6. ↵
      1. NICE
      . Respiratory syncytial virus: management in children. 2019. Available https://bnf.nice.org.uk/treatment-summary/respiratory-syncytial-virus.html.
    7. ↵
      2. Hardelid P ,
      3. Verfuerden M ,
      4. McMenamin J , et al.
      The contribution of child, family and health service factors to respiratory syncytial virus (RSV) hospital admissions in the first 3 years of life: birth cohort study in Scotland, 2009 to 2015. Eurosurveillance 2019;24. doi: 10.2807/1560-7917.ES.2019.24.1.1800046
    8. ↵
      2. Broberg EK ,
      3. Waris M ,
      4. Johansen K , et al.
      Seasonality and geographical spread of respiratory syncytial virus epidemics in 15 European countries, 2010 to 2016. 2018;23:17–00284. doi: 10.2807/1560-7917.ES.2018.23.5.17-00284
    9. ↵
      2. Taylor S ,
      3. Taylor RJ ,
      4. Lustig RL , et al.
      Modelling estimates of the burden of respiratory syncytial virus infection in children in the UK. BMJ Open 2016;6:6. doi: 10.1136/bmjopen-2015-009337
      OpenUrl
    10. ↵
      2. O’Donnell DR ,
      3. Parslow RC ,
      4. Draper ES
      . Deprivation, ethnicity and prematurity in infant respiratory failure in PICU in the UK. Acta Pædiatrica 2010;99:1186–91. doi: 10.1111/j.1651-2227.2010.01803.x
      OpenUrl
    11. ↵
      2. Colosia AD ,
      3. Masaquel A ,
      4. Hall CB , et al.
      Residential crowding and severe respiratory syncytial virus disease among infants and young children: a systematic literature review. BMC Infect Dis 2012;12:95–95. doi: 10.1186/1471-2334-12-95
      OpenUrlCrossRefPubMed
    12. ↵
      2. Anderson AJ ,
      3. Snelling TL ,
      4. Moore HC , et al.
      Advances in vaccines to prevent viral respiratory illnesses in children. Pediatric Drugs 2017;19:523–31. doi: 10.1007/s40272-017-0257-x
      OpenUrl
    13. ↵
      2. Wijlaars LP ,
      3. Hardelid P ,
      4. Woodman J , et al.
      Contribution of recurrent admissions in children and young people to emergency hospital admissions: retrospective cohort analysis of hospital episode statistics. Arch Dis Child 2015;100:845–9. doi: 10.1136/archdischild-2014-307771
      OpenUrlAbstract/FREE Full Text
    14. ↵
      2. Harron K ,
      3. Gilbert R ,
      4. Cromwell D , et al.
      International comparison of emergency hospital use for infants: data linkage cohort study in Canada and England. BMJ Qual Saf 2017. doi: 10.1136/bmjqs-2016-006253
    15. ↵
      2. Herbert A ,
      3. Wijlaars L ,
      4. Zylbersztejn A , et al.
      Data resource profile: hospital episode statistics admitted patient care (HES APC). Int J Epidemiol 2017;46:1093–93i. doi: 10.1093/ije/dyx015
      OpenUrlCrossRefPubMed
    16. ↵
      2. Harron K ,
      3. Gilbert R ,
      4. Cromwell D , et al.
      Linking data for mothers and babies in de-identified electronic health data. PLoS One 2016;11:e0164667. doi: 10.1371/journal.pone.0164667
    17. ↵
      1. NHS Digital
      . Hospital admitted patient care activity, 2016–17: diagnosis. 2017.
    18. ↵
      2. Zylbersztejn A ,
      3. Gilbert R ,
      4. Hjern A , et al.
      Child mortality in England compared with Sweden: a birth cohort study. The Lancet 2018;391:2008–18. doi: 10.1016/S0140-6736(18)30670-6
      OpenUrl
    19. ↵
      2. Cai W ,
      3. Tolksdorf K ,
      4. Hirve S , et al.
      Evaluation of using ICD-10 code data for respiratory syncytial virus surveillance. 2019. doi: 10.1111/irv.12665
    20. ↵
      1. DCLG
      . The English indices of deprivation 2010 . Government CaL, 2011.
    21. ↵
    22. ↵
      2. Lewis KM ,
      3. De Stavola B ,
      4. Hardelid P
      . Could socioeconomic inequity in bronchiolitis admissions be reduced through intervening on gestational age? A principled analysis of cohort data. Lancet 2018;392:S52. doi: 10.1016/S0140-6736(18)32076-2
    23. ↵
      2. Moore HC ,
      3. de Klerk N ,
      4. Holt P , et al.
      Hospitalisation for bronchiolitis in infants is more common after elective caesarean delivery. Arch Dis Child 2012;97:410–4. doi: 10.1136/archdischild-2011-300607
      OpenUrlAbstract/FREE Full Text
    24. ↵
      2. Koehoorn M ,
      3. Karr CJ ,
      4. Demers PA , et al.
      Descriptive epidemiological features of bronchiolitis in a population-based Cohort. Pediatrics 2008;122:1196–203. doi: 10.1542/peds.2007-2231
      OpenUrlAbstract/FREE Full Text
    25. ↵
      2. Schisterman EF ,
      3. Cole SR ,
      4. Platt RW
      . Overadjustment bias and unnecessary adjustment in epidemiologic studies. Epidemiology 2009;20:488–95. doi: 10.1097/EDE.0b013e3181a819a1
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. StataCorp
      . Stata statistical software: release 15 . College Station: TX: StataCorp LLC, 2017.
    27. ↵
      2. Lofgren ET ,
      3. Wenger JB ,
      4. Fefferman NH , et al.
      Disproportional effects in populations of concern for pandemic influenza: insights from seasonal epidemics in Wisconsin, 1967–2004. Influenza Other Respi Viruses 2010;4:205–12. doi: 10.1111/j.1750-2659.2010.00137.x
      OpenUrlPubMed
    28. ↵
      2. Naumova EN
      , MacNeill IB Seasonality assessment for biosurveillance systems. In: Auget J-L , Balakrishnan N , Mesbah M et al., eds. Advances in statistical methods for the health sciences: applications to cancer and AIDS studies, genome sequence analysis, and survival analysis . Boston, MA: Birkhäuser Boston, 2007:437–50.
    29. ↵
      2. Moore HC ,
      3. de Klerk N ,
      4. Richmond P , et al.
      Seasonality of respiratory viral identification varies with age and aboriginality in metropolitan Western Australia. Pediatr Infect Dis J 2009;28:598–603. doi: 10.1097/INF.0b013e318199cefd
      OpenUrlCrossRefPubMed
    30. ↵
      2. Barnett AG ,
      3. Dobson AJ
      . Analysing seasonal health data . Berlin: Springer-Verlag, 2010.
    31. ↵
      2. Zhu Y ,
      3. Matsuyama Y ,
      4. Ohashi Y , et al.
      When to conduct probabilistic linkage vs. deterministic linkage? A simulation study. J Biomed Inform 2015;56:80–6. doi: 10.1016/j.jbi.2015.05.012
      OpenUrlCrossRefPubMed
    32. ↵
      2. Galobardes B ,
      3. Lynch J ,
      4. Smith GD
      . Measuring socioeconomic position in health research. Br Med Bull 2007;81–82:21–37. doi: 10.1093/bmb/ldm001
      OpenUrlPubMed
    33. ↵
      2. Stolwijk AM ,
      3. Straatman H ,
      4. Zielhuis GA
      . Studying seasonality by using sine and cosine functions in regression analysis. J Epidemiol Community Health 1999;53:235–8. doi: 10.1136/jech.53.4.235
      OpenUrlAbstract
    34. ↵
      2. Zhao H ,
      3. Green H ,
      4. Lackenby A , et al.
      A new laboratory-based surveillance system (Respiratory DataMart System) for influenza and other respiratory viruses in England: results and experience from 2009 to 2012. Eurosurveillance 2014;19:20680. doi: 10.2807/1560-7917.ES2014.19.3.20680
      OpenUrl
    35. ↵
      1. Public Health England. Laboratory reporting to Public Health England
      . A guide for diagnostic laboratories . London: Public Health England, 2016.
    36. ↵
      2. Noveroske DB ,
      3. Warren JL ,
      4. Pitzer VE , et al.
      Local variations in the timing of RSV epidemics. BMC Infect Dis 2016;16:674. doi: 10.1186/s12879-016-2004-2
      OpenUrl
    37. ↵
      2. Lewis KM ,
      3. De Stavola B ,
      4. Hardelid P
      . Geospatial and seasonal variation of bronchiolitis in England: a cohort study using hospital episode statistics. Thorax 2020;75:262–8. doi: 10.1136/thoraxjnl-2019-213764
      OpenUrlAbstract/FREE Full Text
    38. ↵
      2. Murray J ,
      3. Bottle A ,
      4. Sharland M , et al.
      Risk factors for hospital admission with RSV bronchiolitis in England: a population-based birth cohort study. PLoS One 2014;9:e89186. doi: 10.1371/journal.pone.0089186
    39. ↵
      2. Jones LL ,
      3. Hashim A ,
      4. McKeever T , et al.
      Parental and household smoking and the increased risk of bronchitis, bronchiolitis and other lower respiratory infections in infancy: systematic review and meta-analysis. Respir Res 2011;12:5. doi: 10.1186/1465-9921-12-5
      OpenUrlCrossRefPubMed
    40. ↵
      2. Cheung CR ,
      3. Semple MG
      . Stemming the tide of hospital admissions for bronchiolitis. Arch Dis Child 2016;101:118–19. doi: 10.1136/archdischild-2015-309452
      OpenUrlFREE Full Text

    Footnotes

    • Contributors All authors designed the study and revised the paper. KL cleaned and analysed the data, and drafted the paper. PH supervised the study.

    • Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors. KL is funded by a UK Medical Research Council doctoral training studentship (MR/N013867/1).

    • Competing interests None declared.

    • Patient consent for publication Not required.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Date availability statement Data may be obtained from a third party and are not publicly available. This study uses NHS hospital episode statistics data and was provided within the terms of a data sharing agreement (NIC-393510-D6H1D-v3.2) to the researchers by the Health and Social Care Information Centre (‘NHS Digital’). The data do not belong to the authors and may not be shared by the authors, except in aggregate form for publication. Data can be obtained by submitting a data request through the NHS Digital Data Access Request Service.

    • 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.