Risk of ovarian cancer in women with symptoms in primary care: population based case-control study
BMJ 2009; 339 doi: https://doi.org/10.1136/bmj.b2998 (Published 26 August 2009) Cite this as: BMJ 2009;339:b2998All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Authors should come clean about their limited findings compared to to their extensive media exposure
Hamilton et al's response to criticism of their paper is welcome, if
belated.
The authors continue to promote certain symptoms as effective
predictors of
ovarian cancer, although their acceptance now that these features are in
fact
of "low risk" of being useful is also welcome. Despite the authors'
enthusiasm,
their study has significant problems of retrospectiveness and lack of
precision
in defining symptoms (for example, when does "abdominal bloating" become
"abdominal distention"?) There are other significant problems described
previously in rapid responses, for example the (typically) six week period
between referral and diagnosis which, if removed from the right side of
the
authors' charts then much of the increase if symptom frequency disappears
too.
The central point they miss about the high community prevalence of
such
symptoms is that in widely publicising their study they have actively
encouraged women experiencing these symptoms to present for
investigation, when the predictive value of these symptoms is far from
certain.
But perhaps the most breathtakingly audacious claim by the authors is
that
the media reports of their findings were apparently none of their
responsibility. However, it is quite clear that they quite deliberately
arranged
to speak to the national non-medical media, who quoted them claiming that
a variety of early symptoms should suggest ovarian cancer when their paper
provided no evidence at all to support this. For example, in The Times
they
were quoted as saying:
“Many of the symptoms described by women ... include fatigue,
abdominal
pain or subtle changes in urination or the bowels. With those, quite
simply
ovarian cancer doesn’t spring to the GP’s mind. Unquestionably, some women
have their cancer missed and have to return, sometimes repeatedly. Ovarian
cancer is not a ‘silent killer’ — it is just not being heard.”
In The Telegraph they are quoted as follows: "quite simply ovarian
cancer
doesn't spring to the GP's mind. Does that make it late diagnosis, missed
diagnosis, or difficult diagnosis? What it does mean is that there is
potential
for improvement. Unquestionably some women have their cancer missed and
have to return - sometimes repeatedly.”
Despite this, they make the remarkable claim that such media reports
were
"unprompted by us". Either the quotes are accurate, in which case they go
far
beyond the limited findings in their paper, or they are not accurate, in
which
case the authors have a responsibility to publicly disown them and take
steps
to correct them. They say that "journalism will always be like this, often
preferring emotive reportage to balanced coverage." Yet in fact it
appeared to
be the authors themselves who provided the emotive and inaccurate
statements associated with wide media reportage, which appeared to be an
end in itself.
I would welcome a further explanation from the authors as to their
role in the
widespread inaccuracies in the reportage.
Competing interests:
None declared
Competing interests: No competing interests
Dear Sir,
We are glad that Wheatley agrees that general practitioners would
appreciate reliable information on the predictive values of symptoms of
ovarian cancer: that is what our study set out to do.1-2 He is concerned
about the non-specific nature of the symptoms we identified. It certainly
would be simpler for everyone (particularly women with cancer) if there
were a symptom with high sensitivity and specificity for ovarian cancer.
Unfortunately, none exists, so we have to deal with low-risk symptoms.
Mourits and de Bock bring a lot of experience of ovarian cancer
screening to bear on this problem.3 They state, without a supporting
reference, that “we can only diagnose ovarian cancers at an advanced
stage.” This is not so. Table 2 in our paper shows quite clearly that
women with early cancers have symptoms – and at a much higher rate than
controls.2 After all, the 53 women in this study with stages I or II
cancer (from 212 in total) were diagnosed symptomatically – there was no
other route to diagnosis at the time. This is not to say clinical
diagnosis is easy: our research makes this clear. We suggested that women
with low risk symptoms should be offered, “..full clinical examination,
followed up by review, and investigation.” It was surprising to see this
approach dismissed as unlikely to be cost-effective – we agree that this
requires rigorous evaluation but it is worth noting that screening has a
yield of cancers many times smaller than even the lowest risk symptom we
identified.4
Both correspondents make an error in conflating two entirely
different phenomena: symptoms experienced in the community and those
presented to primary care. It has been well recognised for many years that
most symptoms are never reported.5 Wheatley’s example of bloating is a
perfect one: we agree community surveys find a prevalence of 16-30%, but
it is important to recognise that few sufferers choose to consult with the
problem. We were encouraged that a response by Tate (unselected for print
publication) found exactly the same percentage of patients (2%) who had
actually reported bloating to their doctors. When deciding upon referral,
it is the prevalence of symptoms in patients in the consulting room that
is relevant, not the community prevalence.
Finally, Wheatley is concerned by the media response to our paper.
Certainly, some of the coverage (unprompted by us) was inappropriate, with
the Sun’s headline, “Docs told me to lose my gut ... it was ovarian
cancer.” arguably the worst of the offenders. We suspect journalism will
always be like this, often preferring emotive reportage to balanced
coverage.
William Hamilton, Tim J. Peters, and Deborah Sharp for the authors (Clare
Bankhead is on maternity leave)
References
1. Wheatley G. Usefulness of abdominal symptoms in early diagnosis. Bmj
2009;339(sep29_3):b3954-.
2. Hamilton W, Peters TJ, Bankhead C, Sharp D. Risk of ovarian cancer in
women with symptoms in primary care: population based case-control study.
Bmj 2009;339(aug25_2):b2998-.
3. Mourits MJ, Bock GHd. Symptoms are not early signs of ovarian cancer.
Bmj 2009;339(sep29_3):b3955-.
4. Menon U, Gentry-Maharaj A, Hallett R, Ryan A, Burnell M, Sharma A, et
al. Sensitivity and specificity of multimodal and ultrasound screening for
ovarian cancer, and stage distribution of detected cancers: results of the
prevalence screen of the UK Collaborative Trial of Ovarian Cancer
Screening (UKCTOCS). Lancet Oncol. 2009;10:327-40.
5. Green LA, Fryer GE, Jr., Yawn BP, Lanier D, Dovey SM. The Ecology of
Medical Care Revisited. N Engl J Med 2001;344:2021-25.
Competing interests:
None declared
Competing interests: No competing interests
Dr Hamilton and colleagues(1) present the results of their excellent
study of symptoms in women with ovarian cancer in terms of “positive
predictive values”. Whilst these have intuitive appeal, some care is
needed in their interpretation when used in the context of symptoms.
The usual formula for positive predictive value (PPV) uses the
prevalence of disease. In this context, one would want the prevalence of
undiagnosed ovarian cancer. The authors use one year’s incidence instead.
Thus, strictly speaking they are estimating the probability that a woman
with particular symptoms will be diagnosed with ovarian cancer over the
next 12 months. This may be a reasonable surrogate for having undiagnosed
ovarian cancer at the time given that the prevalence of ovarian cancer on
first screen tends to be somewhat greater than one year’s incidence(2).
Moreover, it is particularly appropriate that the authors used the same
time period of 12 months for incidence and symptom data in the PPV
calculations.
A bigger concern is the suggestion that the predictive values refer
to women not already under investigation for ovarian cancer. Our own as
yet unpublished data suggest that women in the UK take a median of 1.6
months (interquartile range 1.1, 5.0) to get from first GP referral to
ovarian cancer diagnosis (AWW Lim [PhD Thesis], University of London,
2009, unpublished data). Thus, it seems likely that within a month of
diagnosis, any symptom that appears will be in a woman who is already
being investigated (for an earlier symptom). If one were to exclude
symptoms within 30 days of diagnosis, the data presented in the authors'
Figure 1, indicate that the positive predictive value for abdominal
distension could be between a third and a half of the 2.5% cited in the
paper.
1. Hamilton W, Peters TJ, Bankhead C, Sharp D, Hamilton W, Peters TJ,
et al. Risk of ovarian cancer in women with symptoms in primary care:
population based case-control study.[see comment]. Bmj 2009;339:b2998.
2. Menon U, Gentry-Maharaj A, Hallett R, Ryan A, Burnell M, Sharma A,
et al. Sensitivity and specificity of multimodal and ultrasound screening
for ovarian cancer, and stage distribution of detected cancers: results of
the prevalence screen of the UK Collaborative Trial of Ovarian Cancer
Screening (UKCTOCS). Lancet Oncol 2009.
Competing interests:
None declared
Competing interests: No competing interests
Dear Sir,
Whilst reading both the William Hamilton paper on ovarian cancer and the
accompanying editorial in the BMJ of 12th September I was struck by an
apparently glaring omission. Both papers caught my interest as they make
a strong case that ovarian cancer patients are not silent, but inform
their GP of symptoms in 85% of cases before a diagnosis is made. It is
certainly useful to be aware of the positive predictive values for ovarian
cancer in a female population over 40 years of age, but the symptoms
identified are very common in the wider population. This helps explain why
there is no current workable screening program for ovarian cancer and why
GPs often fail to spot the symptoms of this cancer.
The future for this type of patient identification is not mentioned
in either paper and lies with the ever progressing field of health ICT.
Read Codes used in general practice struggle to capture information
required by the study investigators such as severity or duration of
symptoms. They also have a limited capacity to accurately code symptoms
which can be limited by the clinician’s ability to remember the codes.
Cutting edge technology using the next generation coding system -
SNOMED CT - is used successfully by private healthcare ICT providers such
as the Med⁺DBase system. The result is automatically coded patient
symptoms, diseases and treatments using free-text typed by the doctor at
consultation.
With 730, 000 terms of reference an intelligent ICT system using
SMOMED understand what symptoms the patient has and what symptoms they
don’t, in addition to markers of severity and duration including synonyms,
misspellings and abbreviations for the same disease. This makes it easy
to identify either prospectively or retrospectively female patients over
40 with these positive predictive values (or a combination of them), so
that a reminder for ovarian cancer examination or investigation could
prompt the clinician.
Is it time that mainstream medicine harnessed the power of ICT both
to provide more detail for future researchers and for clinicians to better
identify at-risk groups?
Kind regards,
Competing interests:
None declared
Competing interests: No competing interests
Dear Editor,
With great interest we read a primary care based case-control study
on the value of symptoms in diagnosing ovarian cancer in primary care [1].
Although the title of the article is compelling, we found it confusing and
premature as well and we would like to make a few comments on the
conclusions of this study.
Although the authors detected significant associations between symptoms of
bloating and increased abdominal distension with the diagnosis of ovarian
cancer, these associations do not mean that these symptoms are early signs
of ovarian cancer. Once ovarian cancer causes abdominal distension this is
usually due to a large ovary, omental cake and/or ascites. These are
symptoms of an advanced stage disease and visiting a doctor at that time
will result in the diagnosis, but not improve prognosis. Another comment
we would like to make is that symptoms of bloating and abdominal
distension are very a-specific signs. Almost every woman knows them as a
frequent complaint in the second half of the menstrual cycle. Also women
with irritable bowel syndrome (IBS) experience these symptoms very
frequently if not daily. As most women will not visit their GP due to
these symptoms, this study can not be considered as a population based
study. And last, but surely not least, giving much attention to these a-
specific signs and advising women to see a gynaecologist for
ultrasonography is most likely not cost effective and will cause a lot of
anxiety and cancer worries. Based on the presented results it can be
estimated that a symptom such as abdominal distension will yield
additional 14 referrals per year for each GP and a symptom such as
abdominal pain will give additional 217 referrals per year for each GP.
Whether these referrals will result in earlier diagnosis of ovarian cancer
has not been proven with this study. Only a prospective trial will do.
We have more than 15 years of experience in ovarian cancer screening in a
high risk population. Even in this high risk population with an annual
gynaecological screening and the possibility to come earlier once there
are symptoms, we were only able to diagnose ovarian cancers in an advanced
stage (IIIC) [2]. Therefore, instead of putting great effort, workload and
money in late and a-specific signs and symptoms without proven efficacy in
reducing mortality from ovarian cancer, we would like to recommend putting
more effort in research on biomarkers to detect ovarian cancer in an early
stage.
Prof. MJ Mourits, PhD, MD (1), GH de Bock, PhD (2)
Departments of Gynaecologic Oncology (1) and Epidemiology
(2)University Medical Center Groningen, University of Groningen,
Groningen, The Netherlands
References:
1. Hamilton W, Peters TJ, Bankhead C, Sharp D. Risk of ovarian cancer in
women with symptoms in primary care: population based case-control study.
BMJ 2009; 339:b2998.
2. Van der Velde NM, Mourits MJE, Arts HJG, De Vries J, Leegte LK,
Dijkhuis G, Oosterwijk JC, De Bock GH. Time to stop ovarian cancer
screening in BRCA1 and BRCA2 mutation carriers? Int J Cancer 2009; 124:
919-23.
Competing interests:
None declared
Competing interests: No competing interests
One wonders why some GPs resist the pressure to just make life easier
by referring everything, particularly when they would be regarded by the
patients as 'good doctors' for doing so. I suppose it's because, against
all the odds, the idealism has been retained and they're trying to do the
right thing. Bloating, back pain, TATT- why not just declare: 'You want me
to break the NHS? Then I'll do it by passing the buck and reduce my chance
of complaints at the same time'.
Competing interests:
None declared
Competing interests: No competing interests
Thanks to Rosemary Tate - the difference between the prevalence of
abdominal bloating in community surveys and that recorded in medical
record computer databases is exactly the point made in my response.
Computerised medical records are great for performing searches, but
no-one
would claim that everything in the consultation will be recorded verbatim,
or
even comprehensively - recording is selective. Patients may not reveal the
abdominal bloating they may have, and if they do it may not be recorded if
not thought to be a key element of the consultation by the highly time-
pressured GP.
Several points come through from some of the excellent responses
above.
The authors in the paper effectively claim that the clinical feature
of
abdominal distension is being ignored by GPs and that this leads to
delayed
diagnosis of ovarian cancer. However, what patients and GPs meant by
abdominal distension is never defined, and importantly there is little
sense of
how, when used by patients and GPs, this symptom differs from abdominal
bloating, a much more common symptom and one with little or no predictive
value. Further work is needed to define what the patients and GPs meant by
"distension" as opposed to "bloating", before there can be any credible
claim
of use of this symptom as an effective way of predicting who should go on
to
further investigation.
As pointed out above, one of the problems with the retrospective
design of
this study is that, when those unfortunate enough to develop cancer look
back, their early symptoms of disease may be obvious to them, but in fact
these same symptoms may have been experienced by those without cancer to
more or less a similar degree. Their significance is obvious to the
sufferers
only because, quite understandably, they focus down on to their own
subsequent experience.
Finally, the limited new findings from this paper make the various
claims of
Hamilton to the non-medical media after publication of his paper all the
more surprising, particularly when he refers to clinical features
supposedly
"missed" when he has failed to put forward credible evidence that they
have
any worthwhile predictive value at all.
For example, to The Times ("Women with ovarian cancer ‘dying because
GPs
fail to spot signs’ " 26 Aug), he states: “Many of the symptoms described
by
women ... include fatigue, abdominal pain or subtle changes in urination
or
the bowels. With those, quite simply ovarian cancer doesn’t spring to the
GP’s
mind. Unquestionably, some women have their cancer missed and have to
return, sometimes repeatedly. Ovarian cancer is not a ‘silent killer’ — it
is just
not being heard.”
in The Telegraph ("Ovarian cancer patients face delays 'because
symptoms go
unrecognised' " 25 Aug), he opines that: "quite simply ovarian cancer
doesn't
spring to the GP's mind. Does that make it late diagnosis, missed
diagnosis,
or difficult diagnosis? What it does mean is that there is potential for
improvement. Unquestionably some women have their cancer missed and
have to return - sometimes repeatedly.”
I'm sure Hamilton is aware that of his obligation to, when
"presenting your
research findings to the non-medical press you should make every effort to
ensure that your research findings are reported in a balanced way." On the
face of it, it's difficult to see how Hamilton can square his extensive
media
claims prompted by his paper with the findings of the study itself, and
his
obligation to comment on them responsibly.
Competing interests:
None declared
Competing interests: No competing interests
The rapid response by Graham Wheatley suggests that the annual
prevalence of abdominal bloating and distension is much higher than the
2% reported by Dr Hamilton et al. However, if this is the case, the
symptoms are either not being reported or are not being recorded by GPs. A
recent analysis I carried out using the General Practice Research Database
(GPRD) revealed identical proportions coded as having reported abdominal
distension and bloating to Hamilton et al, i.e 2%.
These data (not yet published) are based on the GPRD records taken
between June 2002 and June 2007 for 8600 (control) women aged between 40
and 80 at the start of data collection.
Competing interests:
None declared
Competing interests: No competing interests
Another important study from Hamilton et al. When implementing study findings such as this, two important factors need highlighting. One is that symptoms associated with obvious causes (such as pregnancy)which will not be detected in the study, need to be excluded. Secondly, the increased predictive value of multiple and persistance of symptoms needs emphasising. Severity is a similar consideration that is difficult to identify in studies.
Competing interests:
Guideline Team leader of New Zealand Guidelines Group "Suspected cancer in primary care: Guidance for referral and reducing disparities guideline."
Competing interests: No competing interests
A polite reply to Dr Wheatley's questions
Dr Wheatley has made a most odd – and rather personal – third
contribution. We wonder why he has such disbelief in our findings, and in
our explanation of the reportage. We are happy to let the science stand
for itself, and will restrict our reply to his concerns about media
coverage.
By agreement, all media communications were made by Dr Hamilton. In
our previous response we described the media coverage as unprompted by us.
This is accurate. The BMJ selected the paper as being newsworthy, and
wrote a press release outlining the main findings. This release was
reviewed and agreed by Dr Hamilton; it essentially reiterated the abstract
of the paper. After the press release, but during the embargo period
(which allows the media to write their stories), several media
organisations contacted Dr Hamilton. These were mostly from radio and
television. At no point, did he – or anyone within the Universities of
Bristol and Oxford – seek out a
single media outlet.
Dr Hamilton gave three radio and two television interviews for
release on the day that the embargo was lifted. He also gave a brief
telephone question and answer session to a press organisation (which was
probably the source of the Times and Telegraph story, though he spoke to
no individual newspaper). All this can be corroborated, as in fact Dr
Hamilton is profoundly deaf and can only use a telephone through an
interpreter. He avoids the phone where at all possible.
So, those are the facts behind the media coverage – we imagine most
authors selected for a press release would have a similar story to tell.
We have already described some of the media coverage as
inappropriate, and made it clear in our earlier response that we were
disappointed by it. What we cannot quite understand is just what Dr
Wheatley finds so wrong in the quotation from the Times (the Telegraph’s
is similar).
“Many of the symptoms described by women ... include fatigue,
abdominal pain or subtle changes in urination or the bowels. With those,
quite simply ovarian cancer doesn’t spring to the GP’s mind.
Unquestionably, some women have their cancer missed and have to return,
sometimes repeatedly. Ovarian cancer is not a ‘silent killer’ — it is just
not being heard.”
Women do often have subtle symptoms with ovarian cancer. Ovarian
cancer does not often spring to a GP’s mind with the subtle symptoms
listed. Some cancers are missed. Ovarian cancer is not silent. Perhaps
saying it is just not being heard was a little unfair to us GPs – but we
cannot afford to be complacent about a cancer with nearly 7,000 new cases
a year and such a dismal prognosis.
Competing interests:
None declared
Competing interests: No competing interests