Editor
BMC Infectious Diseases
BioMed Central
Floor 6, 236 Gray’s Inn Road
London WC1X 8HB
United Kingdom
14th June 2013
To the Editor,
MS: 1708016998822896
Chronic hepatitis B prevalence among Aboriginal and Torres Strait Islander
Australians since universal vaccination: a systematic review and meta-analysis.
Thank you for reviewing our manuscript.
Please find below point-by-point responses to the reviewer’s comments.
Yours sincerely,
Simon Graham
Aboriginal and Torres Strait Islander Health Program
Kirby Institute
University of New South Wales
Sydney Australia
Reviewer 1
1. It is still not very clear to me whether HBsAg is more prevalent in the
indigenous Australian population because it is endemic or because this
population accumulate risk factors (or both). This should be better clarified
in the introduction to understand which is the contribution of vertical
transmission, sexual transmission, drug use, tattoo etc..to HBsAg
prevalence. This may be particularly helpful also to understand how much
time we may need to observe an effect of the HBV vaccination strategy in
this population. Of course if most indigenous infected individuals have
been vertically infected, more time may be needed to observe an important
reduction in HBsAg prevalence in this population following newborn and
indigenous vaccination in the 2000.
Additional information has been added to the Background (Paragraph 2, lines 511 and paragraph 3, lines 6-9) to describing the main mode of transmission in the
Indigenous population and the risk factors which contribute to endemic levels of
HBsAg prevalence among Indigenous Australians.
Aboriginal and Torres Strait Islander (Indigenous) people represent 2.6% of the Australian
population, however they account for an estimated 16% of those living with chronic HBV.[7,
8] Reported endemic levels of HBsAg among Indigenous people in Australia have historically
occurred through vertical transmission from mother to child. In Australia, due to high levels
of antenatal screening and the implementation of universal infant vaccination including a
birth dose since 2000, transmission of HBV through from mother to child has decreased.[9,
10]
Among Indigenous Australians, a variety of modes of HBV transmission are believed to have
contributed to endemic levels of chronic HBV. However it is likely that a higher proportion of
infections have historically occurred at birth or early in life, resulting in a higher prevalence
of chronic HBV infection due to the increased risk of progression to chronicity during
childhood infections.[9] In Australia, due to high levels of antenatal screening and the
implementation of universal infant vaccination including a birth dose since 2000,
transmission of HBV through from mother to child has decreased.[10, 11]
A greater burden of other diseases such as diabetes, sexually transmitted infections and
renal disease have been reported among Indigenous compared with non-Indigenous people
and may increase the Indigenous population’s vulnerability to HBV infection.[15, 16]
2. The discussion is too focused on the indigenous Australian population.
Though there is some effort to make comparisons with other populations,
the article would benefit from more comparisons with similar populations
(or populations in other countries sharing the same pattern of risk factors)
or in with estimates in other countries where HBV is endemic.
Further information about the hepatitis B epidemiology in New Zealand has been
included in the discussion (paragraph 3, lines 3-11).
In 1990, New Zealand (NZ) introduced a HBV screening program in response to the higher
HBV prevalence in Maori compared with non-Maori New Zealanders.[52] A study by
Robinson and colleagues aimed to report the overall coverage and distribution of HBsAg
among participants in the ‘NZ Hepatitis B Screening Programme’. A total of 177,000 people
were tested for HBV (51% of the program targets and 27% of Census 2001 eligible
population), with highest coverage among women (29%) and people from the Pacific Islands
(34.9%). This study found that HBsAg prevalence was higher among Maori (5.6%) compared
with non-Maori (2.8%) people, people aged 40 year or older (6.5%) compared to those aged
15 years or younger (3.5%) and among males (6.1%) than females (5.4%).[52]
3. The authors should also try to make some hypotheses, using results from
other countries or from modelling concerning a possible reduction of HBVassociated liver diseases following the decline in prevalence rates
observed (or potentially observable) as effect of HBV vaccination.
Further information about the potential for a reduction in liver morbidity, similar to
the experience in New Zealand has been added to the discussion (paragraph 4,
lines 2-4).
4. It is not methodologically correct to use the median of the prevalence of the
studies performed before 2000 to obtain an overall indicator of prevalence
before the introduction of the HBV vaccination strategy. I think it would be
better to build a prevalence indicator as a weighted average of the
prevalence of each study (eventually weighted for the inverse of the
variance) or use other similar weighted approaches – see also methodology
used in meta-analyses.
We agree, and have conducted a meta-analysis on adults/pregnant women – the
only subgroup which had studies before and since 2000, and also had
breakdowns for both Indigenous and non-Indigenous people. We have removed
the median results, and thus restructured the methods and results accordingly.
A meta-analysis was conducted to estimate the HBsAg prevalence by Indigenous status and
time period (before 2000 compared with since 2000). We focused the meta-analysis on subgroups with studies before and since 2000, and each study provided HBsAg prevalence
estimates for both Indigenous and non-Indigenous people. Only two sub-groups met this
criteria, adults and pregnant women. The meta-analysis was conducted using weighted
inverse variance methods, the DerSimonian-Laird method[29] assuming a random-effects
model. We used the I2 test to estimate the approximate proportion of total variability in
point estimates that could be attributed to heterogeneity other than that due to chance.
The results were presented as forest plots.
5. It would be useful, if possible, to focus on the subgroup of adults (as more
estimates are available) and try to have a combined estimate per time
intervals before and after the change in HBV vaccination strategy
(eventually see whether it is possible to statistically prove a significant
change.
See response to comment 4 above.
6. The graphs need some improvements, maybe splitting subgroups (one
graph per subgroups) and appropriately reporting prevalence and
confidence intervals with clear legends. It is not clear what the bar and the
line indicate.
The authors have removed the two graphs and replaced these with two metaanalysis forest plots.
Reviewer 2
1. Abstract:
The first sentence of the conclusion should be more cautiously
rephrased (‘…prevalence has decreased’), because of the limits and
probable biases and the lack of a proper statistical technique to test the
apparent difference observed (see comments below).
The wording has been modified to focus more on the decline in the disparity
between HBsAg prevalence among Indigenous and non-Indigenous people.
2. Background:
2.1 Overall, the general aspects regarding HBV could be shortened; eg. Box
1 seems dispensable to me and might be summarized quickly within the
text, eg.2 details on the benefits provided by vaccination.
The authors have removed Box 1 and summarized these points in the
Background in paragraph one.
2.2 Conversely, a clearer focus on Indigenous people should be made to
help the reader unfamiliar with the local situation. Specifically, what is the
estimated overall population of Aboriginal people in Australia? What are
the public health challenges faced by the Indigenous people (risk factors,
access to care, ….). Number and geographical distribution of the
Territories/States involved [this could be explained within the methods
section as well, and use of maps might be helpful for that matter]?
The authors have provided more information regarding the Indigenous Australian
population in two different areas of the Background. These describe the
Aboriginal Australian population in Australia and across the states and territories,
and other health issues faced by this population (see methods under the subheading Settings, paragraph 1).
In 2011, the Australian Bureau of Statistics (ABS) estimated that there were 548,370
Indigenous people (2.6% of the Australian population).[23] The median age of the
Indigenous population was 21 compared with 37 years in non-Indigenous people.[24] The
number of Indigenous people varied across the states and territories of Australia with the
largest population residing in New South Wales (172,624) followed by Queensland
(155,825). The ABS uses the Australian Standard Geographical Classification (ASGC) to define
whether a geographical area in Australia is urban, regional or remote. This allows the
Australian government to develop different policies suitable to people living in regional or
remote areas as they often have less access to services such as specialist health care.[25] A
higher proportion of Indigenous compared with non-Indigenous people live in regional and
remote areas.[26]
A greater burden of other diseases such as diabetes, sexually transmitted infections and
renal disease have been reported among Indigenous compared with non-Indigenous people
and may increase the Indigenous population’s vulnerability to HBV infection.[15, 16]
2.3 Also, the importance of doing this review is not sufficiently highlighted:
since Indigenous people are already identified as a priority population for
vaccination, what is expected from this study? The authors interestingly
note that the eligibility criteria for free coverage vary across jurisdictions:
what are the potential policy challenges to address?
The authors have included a clear aim of this review in the last paragraph of the
Background which is included below. HBV screening guidelines recommend
screening of Aboriginal people at risk of HBV. The high HBsAg rates in pregnant
women since 2000, and much high rates than non-Indigenous people suggest
opportunistic screening of all Aboriginal people may be warranted similar to the
program in New Zealand. The current screening guidelines have been added to
the Background (paragraph 5) and the recommendation to change to
opportunistic screening in the last paragraph of the discussion.
In 1996, a HBV review provided prevalence estimates of HBsAg in Australia and
aimed to provide evidence for the development of a HBV vaccination policy. This
review estimated HBsAg prevalence among Indigenous pregnant women was
10%.[22] Since 1996, a number of studies have reported HBsAg prevalence among
Indigenous people and in 2000 Australia implemented a universal infant and
adolescent HBV vaccination program. We undertook a systematic review to assess if
the disparity in HBsAg prevalence between Indigenous and non-Indigenous people
has decreased over time, especially since the implementation of universal HBV
vaccination in 2000. This review also provides an up-to-date picture of the burden of
HBsAg among Indigenous people and will help to inform current screening policies.
3. Methods:
3.1 With regard to the study eligibility criteria, the authors do not state any
lower temporal bound. As a result, the two periods of interest are very
unbalanced (before 2000 [which results in including studies from 1972] vs
after 2000). What is the rationale for not using periods such as e.g. (19901999) vs (2000-2010)?
The authors decided to include all articles in the review because if it restricted the
review to a similar time period before 2000 there would only 4 studies included
and as a result we would not be able to conduct a meta-analysis.
3.2 Medians are calculated to provide overall prevalence numbers based on
all available prevalence estimates: why did the authors not use techniques
such as weighted mean prevalence or meta-regressions, including
assessment of the underlying heterogeneity?
See response to Reviewer 1 comment number 4.
3.3 The following sentence is too vague: ‘the authors calculated a p-value’.
What were the tests performed?
P-values are no longer necessary as a meta-anaysis has been conducted.
3.4 Because the universal vaccination program targets infants/adolescents,
the comparison of before/after 2000 is questionable for the prevalence
estimates in adults / pregnant women / prisoners; at least, issues like
delay/latency should be discussed.
We agree, and in the discussion have focused more on the potential impact of
the adolescent catch-up campaign on pregnant women (see paragraph 2, lines 18).
4. Results
4.1 The process behind studies selection is not detailed. It is expected (and
stated in PRISMA statement) that the authors give the ”numbers of studies
screened, assessed for eligibility, and included in the review, with reasons
for
exclusions at each stage, ideally with a flow diagram.”
The authors have included a flow diagram as suggested (see Figure 1).
4.2 Again, the description of the different geographic areas is rather unclear
to
the unfamiliar reader (what are ‘regional areas’?)
The authors have included a description in the methods section under the subheading Setting which describes urban, regional and remote areas of Australia
(Paragraph 1).
The ABS uses the Australian Standard Geographical Classification (ASGC) to define whether a
geographical area in Australia is urban, regional or remote. This allows the Australian
government to develop different policies suitable to people living in regional or remote
areas as they often have less access to services such as specialist health care.[25] A higher
proportion of Indigenous compared with non-Indigenous people live in regional and remote
areas.[26]
4.3 Figures 1 and 2 have no title and - if I am not mistaken - are not referred
to in the body of the text.
The authors have removed these two Figures and replaced them with forest
plots.
5. Discussion
5.1 The authors acknowledge a number of limitations, including the risk of
overestimation when considering clinical audits. However, there is no
comprehensive discussion about important issues such as publication
bias,
selection bias in the reporting for other study designs (response rates,
sampling methods for cross-sectional surveys). Mostly, the conclusions
regarding the comparison before-after 2000 should be more cautious,
because too many potential biases prevent from ascertaining whether that
apparent difference is definitely real or not.
The authors have described the potential limitations of many of the studies in the
results (see paragraphs 5), and also focused the meta-analysis on studies which
provided breakdowns for both Indigenous and non-Indigenous people.
5.2 Overall, it is not clear to me what are the practical implications of the
findings, since Indigenous people are apparently already identified as a
group at risk justifying recommendations for vaccination (see intro): maybe
additional elements could be given with respect to the price/coverage of
vaccination and access to care? The authors might detail more why “there
remains room for improvement in the coverage of infant and adolescent [...]
especially among Indigenous children”.
The authors have stated a clearer aim of why they conducted this review
(Background paragraph 6) Please see response to 2.3. In the discussion the
author also state that we are advocating for opportunistic screening of Indigenous
people in Australia to identify those who need vaccination and those who are
living with chronic HBV and need treatment and care (last paragraph of
discussion). In the discussion the authors also state that opportunistic screening
is best delivered through General Practices and Aboriginal Community Controlled
Health services across Australia (Discussion, paragraph 3).
Minor Essential Revisions
1. Background: The very last sentence is unclear: ‘We will also present
groups in Australia who are a higher risk…’ At this stage, it is unclear
whether the authors consider subgroups within Indigenous people, or other
groups at risk apart from Indigenous.
The authors have removed this sentence from the manuscript.
2. Methods:
2.1 Please specify the initials of the ‘two authors’ who did the review and
information extraction.
The authors have re-worded this sentence to highlight that the lead author
conducted the search. This is stated in the Methods section.
2.2 The term ‘Other adults’ to characterize the group not being ‘pregnant
women’/prisoners may be inadequate, since it refers to the majority of the
community.
The authors have changed the classification of ‘Other adults’ to ‘Adults’ to better
describe this sub-group. This has been changed throughout the manuscript.
3. Box 2: is it referred to somewhere? The current title is not sufficiently
informative: recommendations by the authors after considering their
findings?
The authors have removed Box 2 from the manuscript.
Discretionary Revisions
- Beginning of the Discussion: the authors state “over a 20 year period”,
whereas it is actually an almost 40y period (1972-2008).
The authors have removed this sentence from the discussion.