Outcomes Reported for Australian First Nation Populations for the Influenza A(H1N1) 2009 Pandemic and Lessons for Future Infectious Disease Emergencies: a Systematic Review

Aims: Enhanced data collection during infectious disease emergencies, such as the COVID-19 pandemic, must inform the clinical and public health responses appropriate for Australian First Nations populations. To inform the design of such data collection protocols, we systematically reviewed the reported outcomes for the First Nations population related to A(H1N1) 2009 pandemic influenza infection. Methods: We searched PubMed and Google using the search terms: pandemic AND Australia AND 2009 AND (Indigenous OR Aboriginal OR “Torres Strait”). Data extracted included: location; study design; data source(s), number of study participants and the number and percentage that were First Nations; completeness of First Nations status; and reported outcomes (stratified by First Nations status). Each study was also reviewed for documentation of engagement or consultation with First Nation individuals, communities or health services regarding the study design, data collection, analysis, interpretation and reporting. Results: Our search identified 53 citations, with 13 deemed eligible for inclusion. Most studies were case-series (n=6) and used primary data (n=8) and/or secondary data (n=10). The number of First Nations participants ranged from 13 to 3,966. The proportion of First Nations participants per study varied from 1.8% to 100%. Completeness of reporting First Nations status ranged from 62% to 100%. Reported outcomes stratified by First Nations status included notification rate (n=3), comorbidities/risk factors (n=4), severity of disease (hospital admission (n=8), intensive care unit admission (n=8), death (n=5)) and interventions (anti-viral use (n=2) and vaccination (n=4)). There were no studies that described engagement/consultation with First Nations individuals, communities or health services regarding any aspect of the study process. Conclusion: Studies identified in this review mostly used secondary data and reported on outcomes relating to severity, and comorbidities and other risk factors. Studies specifically designed for First Nations populations are required to fully understand the contributing factors for the frequency and severity of disease in an infectious disease emergency and inform appropriate responses. First Nations communities and health services need to be adequately engaged and participate in the design, implementation, analysis and reporting of such enhanced data collection studies.


Introduction
The rapid onset of the COVID-19 pandemic is a case study of the need for rapid data collection, analysis and reporting on clinical and epidemiological characteristics of cases and contacts to inform appropriate clinical and public health responses (1). Surveillance systems for detecting such infectious disease emergencies, which includes influenza pandemics and other outbreaks of novel diseases, have been established in Australia (2). However, once an event is detected, information needs for responding to infectious disease emergencies are increased far beyond routine surveillance systems, particularly early on when many clinical and epidemiological factors are unknown. Further, information requirements change over the course of the event, with a focus first on verification and detection of cases, then risk and severity assessments, and then monitoring the course of the pandemic and detecting any changes in risk and severity (3). Information requirements include, but are not limited to, identification of risk groups, age distribution of cases, clinical features, comorbidities, health care utilisation (e.g. hospital and/or intensive care unit admission), and spread of disease (for example household transmission and geographic distribution) (3).
Early in the influenza A(H1N1) 2009 pandemic, Aboriginal and Torres Strait Islanders, hereafter respectfully referred to as First Nations, were identified as disproportionately affected both in terms of incidence (notifications) and severity (hospital and intensive care unit (ICU) admissions) of infection (4)(5)(6). At the time of the influenza A(H1N1) 2009 pandemic, First Nations peoples were 3% of Australia's population, yet accounted for 11% of cases, 20% of hospital admissions and 13% of deaths (7). This difference in severity has been attributed to the higher prevalence of risk factors and comorbidities within the First Nations population (8). During the pandemic an annex to the Australian Health Management Plan for Pandemic Influenza 2008 was developed to specifically address issues for First Nations population (9). However this annex and the current Australian Health Management Plan for Pandemic Influenza does not include any specific considerations for data collection systems nor special studies to inform the response for the First Nations population (9,10). While First Nations status is routinely collected for notifiable diseases, and was for the influenza A(H1N1) 2009 pandemic, the completeness of this field is suboptimal (11).
First Nations Australians are unique and culturally diverse, and are the oldest continuing cultures in the world (12). Due to the collectivist nature of First Nations Australians cultures, people often live in large family groups and place a greater emphasis on the need for family and social connectedness, which could leave families in a more vulnerable position in relation to infectious disease emergencies (12,13). Collaborative processes for pandemic planning have been identified and participatory approaches to development of responses for the 2009 influenza pandemic for First Nations peoples were documented (13,14). However, sufficient data on outcomes for First Nations peoples is critical for evidence-based planning and response.
There is a need to ensure that enhanced data collection during an infectious disease emergency, such as the COVID-19 pandemic, includes a sufficient number of First Nations people to ensure meaningful analysis and also meets the needs for both communities and responders. Through understanding the clinical and epidemiological characteristics of an infectious disease emergency within the First Nations population, public health responders and affected communities can develop evidence-based responses. In order to inform the design of such protocols for enhanced data collection, we reviewed what outcomes were reported for the First Nations population during the influenza A(H1N1) 2009 pandemic.

Methods
We conducted a systematic literature review on the outcomes for First Nations populations that were reported during the influenza A(H1N1) 2009 pandemic. The systematic review process was guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines as outlined in Figure 1 (15). We searched both PubMed and Google on 27 th March 2019 using the search terms: ((pandemic) AND (2009) AND (Australia)) AND ((Indigenous) OR (Aboriginal) OR (Torres Strait)) in PubMed and pandemic AND Australia AND 2009 AND (Indigenous OR Aboriginal OR "Torres Strait") in Google. The first three Google pages were hand searched for grey literature. Grey literature refers to any research or materials produced by organisations published outside the traditional academic channels. This commonly includes publications relevant to our review such as government reports. Lastly, the references of each included study were screened to identify additional studies.
We included all descriptive and analytical studies using primary and/or secondary data. We excluded books, book chapters, commentaries, literature reviews, editorials, poster abstracts, case reports, published languages other than English, and studies where data were not stratified by First Nations status. Studies that only reported the total number of participants by First Nations status, and no outcomes, were also excluded as these studies did not add any results to the aim of this review. The search was limited to publications from 2009 onwards as 2009 was year of the influenza A(H1N1) 2009 pandemic. AG conducted the searches, and then both AG and EF screened titles and abstracts independently. Once decisions were made, AG and EF discussed discrepancies.
The following data were extracted for each article: lead author, year of publication; study location; study design; number of overall participants included in the study and the number and percentage of those that were First Nations; completeness of First Nations status (percentage of study participants with First Nations status reported); outcomes reported in the study that were stratified by First Nations status; results for each outcome, and data source(s). Each paper was also reviewed for any documentation of processes for any type of engagement or consultation with First Nation individuals or communities or First Nations health services regarding the study design, data collection, analysis, interpretation or reporting.

Results
The database and grey literature search identified 53 citations, of which 17 were deemed relevant for full text review after reviewing the title and abstracts. Upon full-text review, 8 studies were excluded. Four studies were excluded because they either had no analyses of data (16) or the analyses were not stratified by First Nations status (17)(18)(19). A fifth study was excluded due to only reporting the number and percent of First Nations participants (nil outcomes stratified by First Nations status) (20). Two serosurveys were excluded because their design focussed on serum samples collected prior to and after the pandemic to test immunity, and therefore could not be used to inform action during the pandemic (21,22). The eighth study, a government report, was excluded because the same dataset was analysed in more detail in an included study (7). Reference lists of included articles were checked for any other relevant articles, from these six additional studies were identified for full text review. Two were excluded as they only reported the number and/or percent of participants by First Nations status (4,5). In total 13 studies met our inclusion criteria (eight from PubMed, one from Google and four from reference lists) ( Figure  1).

Study sample sizes
Total participant numbers ranged from 57 (31) to 75,154 (30). The latter did not report the total number or proportion of First Nations participants (30). The number of First Nations participants ranged from 13 (31) to 3,966 (26) and the proportion of First Nations participants included in each study varied from 1.8% (34) to 100% (24). Nine studies reported on completeness of First Nations status (8, 23-26, 28, 29, 31, 34) where the completeness ranged from 62% to 100%.

Outcomes reported Demographics
The most common demographic variable stratified by First Nations status was age (8,(24)(25)(26)29). The median age for notification was the same or slightly lower for First Nations compared to non-Indigenous cases ( Table 2) (8,25,29) as was the median age for hospitalisation, ICU admission and death. The national study by Pennington et al (2016) found the median age of First Nations hospitalised cases slightly older than non-Indigenous cases (32 years versus 30 years)(26). Goggin et al (8) found First Nations cases who were hospitalised were comparatively older than non-hospitalised First Nations cases (39 years versus 21 years).
Three small studies reported no difference in sex between First Nations and non-Indigenous hospitalised cases nor between hospitalised and nonhospitalised First Nations cases (8,25,29).The large national study by  reported lower ratio of males to females First Nations compared to non-Indigenous notifications and hospital admissions (26).

Transmission
The age-standardised notification rates for First Nations were 3.5-5.2 times the non-Indigenous notification rates (25,26). No studies reported attack rate or household transmission.
The hospital admission rate for the First Nations population ranged from 63 to 269 per 100,000 population with the relative risk compared to the non-Indigenous population ranging from 3.6 to 12 (23, 25-27, 29, 32). One study reported a standardised morbidity ratio of 7 (26). One study, based in North Queensland, found non-Indigenous cases were more likely to be hospitalised while a study in Western Australia reported a higher proportion of First Nations cases hospitalised compared to non-Indigenous cases (8,23).
The rate of ICU admission for First Nations population ranged from 9 to 15 per 100,000 population with the relative risk compared to the non-Indigenous population ranging from 3.9 to 5.5 (23, 25-27, 29, 32). Two studies reported a standardised morbidity ratio ranging from 4.0 to 7.3 for the First Nations population (26,29). First Nations status was not associated with ICU admission amongst hospitalised cases two studies, one based on univariate analysis and in another study when initial signs and investigations were included in a multivariate analysis (25,28). Observed ICU admissions for First Nations populations exceeded predicted values for 2009 for influenza only, influenza/pneumonia and all respiratory illness (30).
The rate of death for the First Nations population ranged from 3 to 5 per 100,000 population with a relative risk range of 3.2 to 5.6 compared to the non-Indigenous population (23,26,27,29,32). Two studies reported a standardised mortality ratio ranging from 4.5 to 7.6 for the First Nations population (26,29).

Comorbidities and other risk factors
Four of the 13 included studies reported on some form of co-morbidity and/or risk factors (8,23,25,31). Pregnancy, as a risk factor, was reported by First Nations status for three studies (8,25,31). Two studies found no difference in the proportion of hospitalised and non-hospitalised First Nations cases that were pregnant (8,25). One study found a higher risk of ICU admission for First Nations pregnant women compared to non-Indigenous pregnant women (31). Two studies reported on smoking and one reported hazardous alcohol use as risk factors. (8,25) There was no difference in the proportion of First Nations cases that were smokers who were hospitalised or not (8). However, a higher proportion of First Nations hospitalised cases reported smoking compared to non-Indigenous hospitalised cases (25). Similarly, a higher proportion of First Nations hospitalised cases reported hazardous alcohol consumption compared to non-Indigenous hospitalised cases (25).
Three studies reported on co-morbidities for influenza A(H1N1) pandemic cases for First Nations populations, including respiratory conditions, metabolic disorders, cardiac disease, renal disease, neurological disease, chronic liver disease and other medical conditions (8,23,25). Two additional studies provided the number and/or proportion of First Nations participants with a co-morbidity but did not include a comparison group (26,29). Two studies found comorbidities were more prevalent amongst First Nations cases, specifically for diabetes, heart disease, respiratory conditions, obesity, and multiple comorbidities, compared to non-Indigenous cases (8,23). However, one study of cases in the Northern Territory did not find any difference in comorbidities (asthma, COPD, bronchiectasis, obesity, cardiac disease, diabetes, chronic liver disease, neurological disease, immunosuppression or ≥1 comorbidity) in First Nations and non-Indigenous hospitalised cases, with the exception of chronic kidney disease, which was more prevalent amongst First Nations hospitalised cases (25). In one study that conducted a multivariate analysis that accounted for comorbidities, First Nations status was not associated with hospitalisation (8).  • First Nations status was not significantly associated with ICU admission among hospitalised cases (25) • Difference between observed and predicted rates of ICU admission for the First Nations population of 10.9 per 100,000 for Influenza only, 17.2 for influenza/pneumonia and 17.0 for all respiratory (30) Death • First Nations death rate was 3 per 100,000 population compared to 0.6 per 100,000 population for the non-Indigenous population. RR 5.6, standardised mortality ratio 4.5 (29) • First Nations death rate was 4 per 100,000 population compared to 0.8 per 100,000 population for the non-Indigenous population. RR 4.6 standardised mortality ratio 7.6 (26) • First Nations death rate was 3 per 100,000 population, RR 4.7 compared to non-Indigenous population (32) • RR for death for First Nations population was 3.2 compared to the non-Indigenous population (23) • RR of death for First Nations Australians was 5.2 compared to the non-Indigenous population (27) Interventions Antiviral use • A higher proportion of First Nations cases were treated with antivirals versus non-Indigenous cases (73% vs 41%) (8)

Interventions
Two studies reported on the use of antivirals (8,23), with one study specifying Oseltamivir (23). Both studies compared antiviral use between First Nations and non-Indigenous cases and found that First Nations people were more likely to be treated with antivirals (8,23). One study compared antiviral use between hospitalised and non-hospitalised First Nations cases and found no difference (8). Four studies reported on influenza vaccination status (8,24,34,35). Two studies reported on the uptake of influenza A(H1N1) 2009 pandemic vaccine with one national study reporting similar levels of vaccination coverage and a WA based study indicating that the First Nations population had higher uptake compared to non-Indigenous population (34,35). One study reported that there was no difference in vaccination coverage between First Nations hospitalised and non-hospitalised cases (8). One study reported no sociodemographic variables were associated with vaccination uptake amongst First Nations pregnant women (24).

Engagement
None of the included studies documented processes for engagement or consultation with First Nations individuals, communities or health services regarding the study design, data collection, analysis, interpretation and reporting before the pandemic commenced. One study, Rudge and Massey (2010), reported on the collaboration between NSW Health and the Aboriginal Community-Controlled Health Sector, that preceded pandemic and was further strengthened through the urgency of the situation (29). The study highlights the value of the existing collaboration in enabling speedy access to six different First Nations communities to gain insights into what these communities needed on the ground at that time. While this consultation was not done prior to the 2009 pandemic, it provided invaluable recommendations from First Nations communities during the pandemic. None of the other 12 studies that met the inclusion criteria included any further information on community engagement, consultation or cultural governance.

Discussion
We found a small number of studies that reported outcomes for First Nations populations, a key risk group, during the influenza A(H1N1) 2009 pandemic. Largely these studies used secondary data and reported on outcomes relating to severity (hospital and/or ICU admission), and comorbidities and other risk factors (e.g. pregnancy). Most of the studies had a small number of First Nations participants, which may have reduced the ability to statistically identify differences in outcomes, particularly when stratified by sub-groups (e.g. by hospital/ICU admission status). Further, the study with the largest number of First Nations participants (n=3,966) also had the lowest completeness for First Nations status which may have resulted in conservative estimates of the impact for First Nations peoples (26).
The majority of the studies undertook univariable analyses of outcomes by First Nations status. There would be benefits to more complex analyses that sought to understand the contributing factors to the higher frequency and severity of disease observed amongst First Nations populations. While First Nations status was a risk factor in the severity of disease in univariable analysis in the studies in this review, multivariable analysis in two studies identified that being First Nations was not a risk factor for hospitalisation when chronic conditions were considered nor were they at higher risk of ICU admission when initial signs and investigations at hospital admission were considered (8,25). Further, it is difficult to understand the causes for the higher rates of notification among First Nations peoples as testing rates were not included in any of the studies. Such understanding during infectious disease emergencies may help inform more nuanced clinical and public health responses, for example, special considerations for First Nations with chronic diseases. No study reported on transmission outcomes for First Nations populations. There are multiple factors that may contribute to differences in transmission of infectious diseases within First Nations populations, compared to non-Indigenous populations, such as frequent travel and inadequate housing infrastructure, which results in crowded housing (36)(37)(38). Household transmission studies where data are collected on cases and their household contacts can allow description of transmission and severity across the spectrum of clinical presentation (39). Such studies could be considered for future infectious disease events for First Nations populations. Acknowledging the differences in household structures for First Nations peoples, visual tools such as a magnetic boards have been developed to assist with defining the household structures for studies (40).
There was limited reporting on interventions by First Nations status. The clinical interventions reported on were antiviral use and vaccination status. Goggin (2011) reported on both antiviral use and vaccination status in a state-wide study (WA) (8). The other studies on antiviral use and vaccination status were small, sub-state studies. Use of clinical and public health interventions may vary by both First Nations status as well as by location. This is evident with higher uptake of seasonal influenza immunisation amongst First Nations populations compared to non-Indigenous populations, although coverage is still sub-optimal (41,42). No study reported on use of non-pharmaceutical interventions, such as isolation. While participatory action research had been conducted to identify public health interventions to reduce transmission, the implementation and evaluation of these interventions has not been documented (43). Ideally, public health responders around Australia will need to know the uptake and effectiveness of interventions during an infectious disease emergency specifically for the populations which they serve.
Only one study referred to any level of engagement with First Nations populations, specifically Aboriginal Community Controlled Health Service Organisations. Further, this was for the response rather than the study design, analysis and interpretation of the data presented (29). Rudge (2010) cited the need for meaningful and extensive consultation with First Nations communities and community-controlled health services across Australia as not only important, but paramount to ensuring the same over-representation of disease during a pandemic does not happen again (29). Preferably this engagement builds upon existing relationships, rather than study specific relationships, and involves active participation of First Nations communities and health services (43)(44)(45). While engagement may have occurred in other studies, but was not documented, we argue that the description of the engagement with First Nations individuals, health services and communities in studies of infectious disease emergencies is an integral part of the methodology and should be documented to encourage best practice.
There are several limitations with our study. Firstly, we only reviewed publicly available studies. Online situation reports produced by the Commonwealth Department of Health may have provided additional information but were no longer available at the time of this study. Further, additional studies may have been available to responders that was not made publicly available. Secondly, while the focus of the review was on data collected related to influenza A(H1N1) pandemic cases, the publication of some studies occurred well after the pandemic. If the results of these studies were not available to responders at the time of the pandemic, it would indicate that the information for responders was even more limited.

Conclusion
A decade on from the 2009 influenza pandemic, and in the midst of the COVID-19 pandemic, a protocol for enhanced data collection, a FFX study, for the general population has been developed by the World Health Organization and is to be implemented in Australia (prospective study of confirmed COVID-19 cases and their close contacts, aiming to collect real-time data on clinical, virological, and epidemiological characteristics of cases) (46). This protocol may only include a small number of First Nation COVID-19 cases, making analyses of outcomes by First Nations status difficult. Such studies for First Nations peoples are required for participatory development of responses that are evidence-based. As our study has demonstrated, we cannot rely on routine data collection on COVID-19 or ad-hoc research studies to ensure adequate data are collected on all outcomes to inform the response for COVID-19 within First Nations populations. To ensure that appropriate responses to COVID-19, or any other an infectious disease emergency within the First Nations populations in Australia, adequate engagement and participation, as defined by First Nations peoples, in all steps of enhanced surveillance studies (design, implementation, analysis, interpretation and reporting) is required.