Heath Kelly

Australian National University, Canberra, Australian Capital Territory, Australia

Are you Heath Kelly?

Claim your profile

Publications (178)645.7 Total impact

  • Heath Kelly, Benjamin J Cowling
    The Lancet 01/2015; DOI:10.1016/S0140-6736(15)60074-5 · 39.21 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The influenza virus undergoes frequent antigenic drift, necessitating annual review of the composition of the influenza vaccine. Vaccination is an important strategy for reducing the impact and burden of influenza, and estimating vaccine effectiveness (VE) each year informs surveillance and preventative measures. We aimed to describe the influenza season and to estimate the effectiveness of the influenza vaccine in Victoria, Australia, in 2013.Methods Routine laboratory notifications, general practitioner sentinel surveillance (including a medical deputising service) data, and sentinel hospital admission surveillance data for the influenza season (29 April to 27 October 2013) were collated in Victoria, Australia, to describe influenza-like illness or confirmed influenza during the season. General practitioner sentinel surveillance data were used to estimate VE against medically-attended laboratory confirmed influenza. VE was estimated using the case test negative design as 1 − adjusted odds ratio (odds of vaccination in cases compared with controls) × 100%. Cases tested positive for influenza while non-cases (controls) tested negative. Estimates were adjusted for age group, week of onset, time to swabbing and co-morbidities.ResultsThe 2013 influenza season was characterised by relatively low activity with a late peak. Influenza B circulation preceded that of influenza A(H1)pdm09, with very little influenza A(H3) circulation. Adjusted VE for all influenza was 55% (95%CI: −11, 82), for influenza A(H1)pdm09 was 43% (95%CI: −132, 86), and for influenza B was 56% (95%CI: −51, 87) Imputation of missing data raised the influenza VE point estimate to 64% (95%CI: 13, 85).Conclusions Clinicians can continue to promote a positive approach to influenza vaccination, understanding that inactivated influenza vaccines prevent at least 50% of laboratory-confirmed outcomes in hospitals and the community.
    Vaccine 11/2014; DOI:10.1016/j.vaccine.2014.11.019 · 3.49 Impact Factor
  • Heath A Kelly
  • Source
    Allen C. Cheng, Heath Kelly
    Australian and New Zealand Journal of Public Health 10/2014; 38(5). DOI:10.1111/1753-6405.12303 · 1.64 Impact Factor
  • Eurosurveillance: bulletin europeen sur les maladies transmissibles = European communicable disease bulletin 08/2014; 19(34). DOI:10.2807/1560-7917.ES2014.19.34.20884 · 4.66 Impact Factor
  • H Kelly, Bj Cowling
    Eurosurveillance: bulletin europeen sur les maladies transmissibles = European communicable disease bulletin 07/2014; 19(27). DOI:10.2807/1560-7917.ES2014.19.27.20850 · 4.66 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Vaccine effectiveness may wane with increasing time since vaccination. This analysis used the Victorian sentinel general practitioner (GP) network to estimate vaccine effectiveness for trivalent inactivated vaccines in the 2012 season. A test-negative design was used where patients presenting to GPs with influenza-like illness who tested positive for influenza were cases and noncases were those who tested negative. Vaccination status was recorded by GPs. Vaccine effectiveness was calculated as (1-odds ratio) × 100%. Estimates were compared early versus late in the season and by time since vaccination. Virus isolates were assessed antigenically by hemagglutination inhibition assay in a selection of positive samples and viruses from healthy adults who experienced a vaccine breakthrough were analyzed genetically. The adjusted vaccine effectiveness estimate for any type of influenza was 45% (95% CI: 8,66) and for influenza A(H3) was 35% (95% CI: -11,62). A non-significant effect of waning effectiveness by time since vaccination was observed for A(H3). For those vaccinated <93 days of presentation vaccine effectiveness was 37% (95% CI: -29,69), while for those vaccinated ≥93 days before presentation it was 18% (95% CI: -83,63). Comparison of early versus late in the season estimates was very sensitive to the cut off week chosen for analysis. Antigenic data suggested that low vaccine effectiveness was not associated with poor vaccine match among the A(H3) viruses. However, genetic analysis suggested nucleotide substitutions in antigenic sites. In 2012, the trivalent influenza vaccine provided moderate protection against influenza and showed limited evidence for waning effectiveness. Antigenic and genetic data can provide additional insight into understanding these estimates. J. Med. Virol. © 2013 Wiley Periodicals, Inc.
    Journal of Medical Virology 06/2014; 86(6). DOI:10.1002/jmv.23847 · 2.22 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Parental attitudes towards vaccination significantly influence vaccine uptake. The A(H1N1)pdm09 influenza pandemic was followed in 2010 by an unprecedented increase in febrile reactions in children receiving trivalent inactivated influenza vaccine manufactured by bioCSL. Uptake of TIV in children <5 years in Western Australia (WA) decreased in 2010 and has remained low. The impact of pandemic A(H1N1)pdm09 and adverse-events on parental attitudes towards vaccination is uncertain.
    Vaccine 05/2014; DOI:10.1016/j.vaccine.2014.05.055 · 3.49 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Few studies report the effectiveness of trivalent inactivated influenza vaccine (TIV) in preventing hospitalisation for influenza-confirmed respiratory infections. Using a prospective surveillance platform, this study reports the first such estimate from a well-defined ethnically diverse population in New Zealand (NZ). A case test-negative design was used to estimate propensity adjusted vaccine effectiveness. Patients with a severe acute respiratory infection (SARI), defined as a patient of any age requiring hospitalisation with a history of a fever or a measured temperature ≥38°C and cough and onset within the past 7 days, admitted to public hospitals in South and Central Auckland were eligible for inclusion in the study. Cases were SARI patients who tested positive for influenza, while non-cases (controls) were SARI patients who tested negative. Results were adjusted for the propensity to be vaccinated and the timing of the influenza season. The propensity and season adjusted vaccine effectiveness (VE) was estimated as 39% (95% CI 16;56). The VE point estimate against influenza A (H1N1) was lower than for influenza B or influenza A (H3N2) but confidence intervals were wide and overlapping. Estimated VE was 59% (95% CI 26;77) in patients aged 45-64 years but only 8% (-78;53) in those aged 65 years and above. Prospective surveillance for SARI has been successfully established in NZ. This study for the first year, the 2012 influenza season, has shown low to moderate protection by TIV against influenza positive hospitalisation.
    Vaccine 04/2014; DOI:10.1016/j.vaccine.2014.04.013 · 3.49 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: There are few studies evaluating the effectiveness of trivalent influenza vaccination (TIV) in young children, particularly in children <2 years. The Western Australian Influenza Vaccine Effectiveness Study commenced in 2008 to evaluate a program providing TIV to children aged 6 to 59 months. An observational study enrolling children with influenza-like illness presenting to a tertiary pediatric hospital was conducted (2008-2012). Vaccination status was determined by parental questionnaire and confirmed via the national immunization register and/or vaccine providers. Respiratory virus polymerase chain reaction and culture were performed on nasopharyngeal samples. The test-negative design was used to estimate vaccine effectiveness (VE) by using 2 control groups: all influenza test-negative subjects and other-virus-detected (OVD) subjects. Adjusted odds ratios were estimated from models with season, month of disease onset, age, gender, indigenous status, prematurity, and comorbidities as covariates. Subjects enrolled in 2009 were excluded from VE calculations. Of 2001 children enrolled, influenza was identified in 389 (20.4%) children. Another respiratory virus was identified in 1134 (59.6%) children. Overall, 295 of 1903 (15.5%) children were fully vaccinated and 161 of 1903 (8.4%) children were partially vaccinated. Vaccine uptake was significantly lower in 2010-2012 after increased febrile adverse events observed in 2010. Using test-negative controls, VE was 64.7% (95% confidence interval [CI]: 33.7%-81.2%). No difference in VE was observed with OVD controls (65.8%; 95% CI: 32.1%-82.8%). The VE for children <2 years was 85.8% (95% CI: 37.9%-96.7%). This study reveals the effectiveness of TIV in young children over 4 seasons by using test-negative and OVD controls. TIV was effective in children aged <2 years. Despite demonstrated vaccine effectiveness, uptake of TIV remains suboptimal.
    PEDIATRICS 04/2014; 133(5). DOI:10.1542/peds.2013-3707 · 5.30 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Duration of viral shedding following infection is an important determinant of disease transmission, informing both control policies and disease modelling. We undertook a systematic literature review of the duration of influenza A(H1N1)pdm09 virus shedding to examine the effects of age, severity of illness and receipt of antiviral treatment. Studies were identified by searching the PubMed database using the keywords 'H1N1', 'pandemic', 'pandemics', 'shed' and 'shedding'. Any study of humans with an outcome measure of viral shedding was eligible for inclusion in the review. Comparisons by age, degree of severity and antiviral treatment were made with forest plots. The search returned 214 articles of which 22 were eligible for the review. Significant statistical heterogeneity between studies precluded meta-analysis. The mean duration of viral shedding generally increased with severity of clinical presentation, but we found no evidence of longer shedding duration of influenza A(H1N1)pdm09 among children compared with adults. Shorter viral shedding duration was observed when oseltamivir treatment was administered within 48 hours of illness onset. Considerable differences in the design and analysis of viral shedding studies limit their comparison and highlight the need for a standardised approach. These insights have implications not only for pandemic planning, but also for informing responses and study of seasonal influenza now that the A(H1N1)pdm09 virus has become established as the seasonal H1N1 influenza virus.
    Influenza and Other Respiratory Viruses 12/2013; DOI:10.1111/irv.12216 · 1.47 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To assess the magnitude and severity of the 2012 influenza season in Victoria, Australia using surveillance data from five sources. Data from influenza notifications, sentinel general practices, a sentinel hospital network, a sentinel locum service and strain typing databases for 2012 were descriptively analysed. Influenza and influenza-like illness activity was moderate compared to previous years, although a considerable increase in notified laboratory-confirmed influenza was observed. Type A influenza comprised between 83% and 87% of cases from the general practitioners, hospitals and notifiable surveillance data. Influenza A/H3 was dominant in July and August, and most tested isolates were antigenically similar to the A/Perth/16/2009 virus used in the vaccine. There was a smaller peak of influenza type B in September. No tested viruses were resistant to any neuraminidase inhibitor antivirals. Higher proportions of type A/H3, hospitalized cases and those with a comorbid condition indicated for influenza vaccination were aged 65 years or older. Influenza vaccination coverage among influenza-like illness patients was 24% in sentinel general practices and 50% in hospitals. The 2012 influenza season in Victoria was average compared to previous years, with an increased dominance of A/H3 accompanied by increases in older and hospitalized cases. Differences in magnitude and the epidemiological profile of cases detected by the different data sources demonstrate the importance of using a range of surveillance data to assess the relative severity of influenza seasons.
    11/2013; 4(3):42-50. DOI:10.5365/WPSAR.2013.4.2.007
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Health authorities find thresholds useful to gauge the start and severity of influenza seasons. We explored a method for deriving thresholds proposed in an influenza surveillance manual published by the World Health Organization (WHO). For 2002-2011, we analysed two routine influenza-like-illness (ILI) datasets, general practice sentinel surveillance and a locum medical service sentinel surveillance, plus laboratory data and hospital admissions for influenza. For each sentinel dataset, we created two composite variables from the product of weekly ILI data and the relevant laboratory data, indicating the proportion of tested specimens that were positive. For all datasets, including the composite datasets, we aligned data on the median week of peak influenza or ILI activity and assigned three threshold levels: seasonal threshold, determined by inspection; and two intensity thresholds termed average and alert thresholds, determined by calculations of means, medians, confidence intervals (CI) and percentiles. From the thresholds, we compared the seasonal onset, end and intensity across all datasets from 2002-2011. Correlation between datasets was assessed using the mean correlation coefficient. The median week of peak activity was week 34 for all datasets, except hospital data (week 35). Means and medians were comparable and the 90% upper CIs were similar to the 95(th) percentiles. Comparison of thresholds revealed variations in defining the start of a season but good agreement in describing the end and intensity of influenza seasons, except in hospital admissions data after the pandemic year of 2009. The composite variables improved the agreements between the ILI and other datasets. Datasets were well correlated, with mean correlation coefficients of >0.75 for a range of combinations. Thresholds for influenza surveillance are easily derived from historical surveillance and laboratory data using the approach proposed by WHO. Use of composite variables is helpful for describing influenza season characteristics.
    PLoS ONE 10/2013; 8(10):e77244. DOI:10.1371/journal.pone.0077244 · 3.53 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Background: There are few studies evaluating the effectiveness of the Southern Hemisphere trivalent influenza vaccination (TIV) in children or TIV effectiveness in children <2 years. The Western Australian Influenza Vaccine Effectiveness study commenced in 2008 to evaluate a state-based program providing TIV to children aged 6-59 months. Methods: An observational study enrolling children presenting to a tertiary paediatric hospital with influenza-like-illness was conducted from 2008-12. Vaccination status was determined by parental questionnaire and confirmed using the national immunisation register and/or vaccine providers. Fully vaccinated was defined as i) two doses of TIV ≥21 days apart and ≥14 days prior to presentation or ii) one dose ≥14 days prior to presentation and two or more doses in a previous year. Nasopharyngeal sampling for respiratory virus PCR and culture was performed. The test-negative design was used to estimate vaccine effectiveness (VE). Fully vaccinated and unvaccinated children were compared using two control groups: influenza test negative subjects (test negative controls) and subjects in whom another virus was detected (other virus detected controls). Adjusted odds ratios were estimated from models with season, week of disease onset, age, sex, Indigeneity, prematurity and comorbidities as covariates. Subjects enrolled in 2009 were excluded from VE calculations. VE was estimated for the influenza season. Results: Of 2408 children enrolled (median: 1.9y), 21.0% required hospital admission. Only 15.3% of children had significant comorbidities. Influenza was identified in 469 children (fluA, 13.7%; fluB: 5.8%). Another respiratory virus was identified in 1493 children (62.0%). Overall, 16.6% children were fully vaccinated and 8.4% partially vaccinated. Using test negative controls, VE for children recruited through the emergency department (2008, 2010-2012) was 63.8% (95%CI: 27.4-81.9%). No significant difference in VE was observed with other virus detected controls (68.2%; 95%CI: 26.5-86.2%). The VE for children < 2 years was 87.7% (95%CI: 39.5-97.5%). Conclusion: This study demonstrates the effectiveness of the Southern Hemisphere TIV in young children over multiple seasons using both test negative controls and other virus detected controls. TIV was effective in children aged <2 years.
    IDWeek 2013 Meeting of the Infectious Diseases Society of America; 10/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: There are limited summary data published on the risk of fever and febrile seizures in children following influenza vaccination. We performed a review of the risk of fever and febrile seizures following receipt of trivalent inactivated influenza vaccine (TIV) in children aged ≥6 months to <36 months, searching PubMED and Google Scholar for English language articles from 2000 onwards, and initiated or ongoing unpublished studies since September 2007 using clinicaltrials.gov. Exclusions included other vaccine co-administration, missing ages or participant numbers, or unmeasured fever. We reviewed articles and collated results using a standard data extraction template. We identified a total of 909 published papers and unpublished trials from a search conducted on 23 January 2013, 669 from Google Scholar, 114 from PubMed and 126 from the Clinicaltrials.gov online database. After excluding 890 published papers or unpublished trials, 5 published papers and 14 unpublished trials were included in this review. Extracted data on number of events, children at risk and time of follow-up were converted to the risk of fever, which was averaged per week of follow-up (referred to as 'averaged weekly risk'). Following one dose of TIV, the median averaged weekly risk of any fever (≥37.5°C) was 26.0% (range 10.3-70.0%) in unpublished trials compared to 8.2% (range 5.3-28.3%) in published papers (p=0.04). The median averaged weekly risk of severe fever (≥39.0°C) was 3.2% (range 0-10.0%) and 2.0% (range 0.6-17.0%), respectively (p=0.91). Variation in the reporting of fever by participant age groups, time since vaccination and the definition or measurement of fever resulted in a wide range of risk estimates. Reporting of febrile reactions should be standardised to allow comparison between manufacturers and influenza seasons.
    Vaccine 09/2013; DOI:10.1016/j.vaccine.2013.09.005 · 3.49 Impact Factor
  • Paul V Effler, Heath Kelly
    Vaccine 09/2013; 32(1). DOI:10.1016/j.vaccine.2013.08.095 · 3.49 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: During the 2009 influenza pandemic, uncertainty surrounding the seriousness of human infections with the H1N1pdm09 virus hindered appropriate public health response. One measure of seriousness is the case fatality risk, defined as the probability of mortality among people classified as cases. We conducted a systematic review to summarize published estimates of the case fatality risk of the pandemic influenza H1N1pdm09 virus. Only studies that reported population-based estimates were included. We included 77 estimates of the case fatality risk from 50 published studies, about one-third of which were published within the first 9 months of the pandemic. We identified very substantial heterogeneity in published estimates, ranging from less than 1 to more than 10,000 deaths per 100,000 cases or infections. The choice of case definition in the denominator accounted for substantial heterogeneity, with the higher estimates based on laboratory-confirmed cases (point estimates = 0-13,500 per 100,000 cases) compared with symptomatic cases (point estimates = 0-1,200 per 100,000 cases) or infections (point estimates = 1-10 per 100,000 infections). Risk based on symptomatic cases increased substantially with age. Our review highlights the difficulty in estimating the seriousness of infection with a novel influenza virus using the case fatality risk. In addition, substantial variability in age-specific estimates complicates the interpretation of the overall case fatality risk and comparisons among populations. A consensus is needed on how to define and measure the seriousness of infection before the next pandemic.
    Epidemiology (Cambridge, Mass.) 09/2013; 24(6). DOI:10.1097/EDE.0b013e3182a67448 · 6.18 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Vaccine effectiveness (VE) studies are often made for a "season" which may refer to different analysis periods in different systems. We examined whether the use of four different definitions of season would materially affect estimates of influenza VE using data from the Victorian general practice (GP) sentinel surveillance network for 2007-2012. In general, the choice of analysis period had little effect on VE estimates (≤five percentage points) when there was a statistically significant protective effect of vaccination (2007, 2010 and 2012). In contrast, for years when the analysis period varied widely depending on the method used and when VE estimates were imprecise, the change in VE estimate was as much as 43 percentage points (2008). Studies of influenza VE should clearly define the analysis period used and, where possible, provide sensitivity analyses to align this definition with other VE studies.
    Vaccine 07/2013; 31(40). DOI:10.1016/j.vaccine.2013.06.103 · 3.49 Impact Factor
  • Heath Kelly, Benjamin J Cowling
    Epidemiology (Cambridge, Mass.) 07/2013; 24(4):622-3. DOI:10.1097/EDE.0b013e318296c2b6 · 6.18 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To better understand the role that diagnostic test-ordering behaviour of general practitioners has on current pertussis epidemiology in Australia. Analysis of Australian general practice encounter data (from the Bettering the Evaluation and Care of Health [BEACH] program) on 13 "pertussis-related problem" (PRP) codes that were most likely to result in a pertussis laboratory test request and Australian pertussis notifications data (from the National Notifiable Diseases Surveillance System [NNDSS]) for the period April 2000 to March 2011. The change in the proportion of PRP general practice encounters with a pertussis test request between 2000 and 2011, and the change in national pertussis notifications over the same period. The proportion of PRP encounters resulting in a pertussis test request increased from 0.25% between April 2000 and March 2004 to 1.71% between April 2010 and March 2011 (odds ratio, 7.0; 95% CI, 5.5-8.8). The BEACH data on pertussis testing and NNDSS data on pertussis notifications were highly correlated (r = 0.99), and the notification data mirrored the likelihood of a pertussis test request in general practice. The proportion of NNDSS pertussis notifications with a polymerase chain reaction (PCR)-confirmed diagnosis increased from 16.3% between April 2000 and March 2004 to 65.3% between April 2010 and March 2011. An increase in pertussis testing following recognition of early epidemic cases may have led to identification of previously undetected infections, resulting in a further increase in notified disease and awareness among GPs. The changing likelihood of being tested may also be due to expanding availability and use of PCR testing in Australia.
    The Medical journal of Australia 06/2013; 198(11):624-8. DOI:10.5694/mja13.10044 · 2.85 Impact Factor

Publication Stats

2k Citations
645.70 Total Impact Points


  • 2014–2015
    • Australian National University
      • National Centre for Epidemiology & Population Health Research
      Canberra, Australian Capital Territory, Australia
  • 1999–2014
    • Victorian Infectious Diseases Reference Laboratory
      Melbourne, Victoria, Australia
  • 2013
    • PathWest Laboratory Medicine
      Perth City, Western Australia, Australia
  • 2010
    • Royal Children's Hospital Brisbane
      Brisbane, Queensland, Australia
  • 2009–2010
    • University of Western Australia
      • School of Computer Science and Software Engineering
      Perth, Western Australia, Australia
    • University of Queensland 
      • School of Population Health
      Brisbane, Queensland, Australia
  • 2002–2009
    • University of Melbourne
      • • Population Mental Health Group
      • • Department of Paediatrics
      Melbourne, Victoria, Australia
  • 2008
    • Victoria University Melbourne
      Melbourne, Victoria, Australia
  • 2003–2008
    • Melbourne Health
      Melbourne, Victoria, Australia
  • 2007
    • Royal Melbourne Hospital
      Melbourne, Victoria, Australia
  • 2005
    • Murdoch Childrens Research Institute
      Melbourne, Victoria, Australia