Additional information from Professor
Regarding the response to Lord Turnberg's question:
How the HPA would help me with my work?
In my reply I will draw the distinction between
the following terms:
Passive microbiological surveillance;
surveillance of samples sent according to clinical indication.
Enhanced passive clinical surveillance;
using existing sources of routine data (eg NHS direct, GPs' computerised
Systematic microbiological surveillance;
routine microbiological sampling from all patients presenting
with clinical syndromes (and possibly from asymptomatic individuals).
1. The HPA should detect outbreaks (new
symptom patterns, or severe illness with high morbidity/mortality)
using enhanced passive clinical surveillance
The HPA should make better use of using existing
routine data sources is likely to be the most effective way to
spot new outbreaks. NHS direct is a rich source of such information.
Another option could also be better use of clinical data from
GP systems, which could be downloaded anonymously each day.
Wherever the data originates from, once a severe
local problem is identified the data could be fed back quickly
ie within 24-36 hours to the local CCDC who could then initiate
targeted microbiological sampling if appropriate (eg to confirm
that an outbreak of chest infection with high morbidity in a nursing
home was influenza rather than Legionella, or to confirm an outbreak
of food poisoning). The sampling could either use infection control
nurses visiting sites and patients, or if appropriate selected
local GP surgeries (eg asked to send stool sample for anyone presenting
with diarrhoea for a 24 hour period).
A few principles apply:
(a) Preferably use staff to collect samples
that are not involved in routine clinical care (ie with their
own dedicated resources, such as infection control nurses) given
the stretched resources and low morale in primary care.
(b) If staff performing routine clinical
care are involved (eg GPs practice nurses), then setting the threshold
for action appropriately is very important; ie only feed back
to GPs or practice nurses that they need to do something when
a suitably high threshold is reached.
(c) Pilot the system before introducing it
nationally so that lessons can be learnt.
(d) Provide adequate training if usual samples
are required (eg nasopharyngeal swabs).
(e) Explain clearly the rationale for data
collection (otherwise the system will fall into disrepute).
2. The HPA should use the cascade system,
and enhance training, to increase clinical index of suspicion
for exotic infections.
GPs have a low index of suspicion for unusual
infections. Where there is a real danger of a new illness for
which GPs have a low clinical index of suspicion, the Health Protection
Agency will need to liaise with GP tutors and CCDCs to ensure
that training about these illnesses is provided for all GPs and
practice nurses. This should be complemented by the cascade system:
where there is information that a new infection is likely (eg
intelligence about bio-terrorism, or outbreak of new disease in
common travel destination) the cascade system needs to be activated
with a succinct clinical description and advice about management
and sampling (if appropriate).
3. The HPA should monitor antibiotic
resistance use occasional systematic microbiological surveillance
from selected practices (eg spotter practices)
Antibiotic resistance is a major threat and
there clearly needs to be a robust monitoring. The problem for
clinicians is that if routine microbiological surveillance was
increased greatly ie widespread systematic microbiological surveillance
this would (a) incur greatly increased time costs of sampling
in an already over-stretched service, and (b) would also create
the perception to the public that they need to see the doctor
to have sampling performed. This would medicalise acute self limiting
illness, probably drawing more people into the system who would
otherwise have happily managed their own acute illness, and arguably
increase antibiotic prescribing overall. The information from
routine microbiological surveillance may also not be terribly
useful clinically. This is in part because the documentation of
laboratory resistance does not mean there is a clinical problem
(since most illnesses are self limiting). Also, even where there
is an apparent clinical problem (eg slowly resolving infection)
laboratory resistance does not mean that in practice the organism
is resistant due to two factors (a) the accuracy of the test and
(b) in vivo antibiotics may still reach concentrations, that mean
that low/intermediate level of in vitro resistance is not relevant
(eg high concentrations in the urine of trimethoprim may overcome
low levels of resistance).
Thus greatly increased routine systematic microbiological
sampling to monitor antibiotic resistance would result in an expensive
data collection exercise, creating the wrong message to the public,
and would have limited utility for clinical practice. The vital
public health information about trends in antibiotic resistance
could much more efficiently be achieved using intermittent systematic
microbiological sampling in "spotter" practices (eg
in patients with suspected UTI sending all samples of urine for
an MSU for three days twice each year). The only question is how
many spotter practices and how often should routine sampling be
performed. This should be determined by calculations of the precision
of estimates required. If the spotter practice scheme is enhanced
the system will need to be properly funded.
4. The Health Protection Agency needs
to identify a dedicated stream of funding for research into the
targeting of antibiotics
Nobody doubts that antibiotic resistance is
one of the major threats to public health. To tackle the issue
and contain resistance we need to be able to selectively use antibiotics
for those that will benefit, and avoid antibiotics for those that
will not. It is a much greater and urgent priority to tackle this
issue than to spend much more money on surveillance.
Unfortunately, we do not know who benefits for
any of the common acute infections: this evidence base is simply
not there (I can provide more detail on this point if required).
We have shown that currently GPs manage their uncertainty by using
variable and ad hoc clinical and socio-demographic features to
target prescribing. We need several pieces of evidence to rationalise
prescribing and improve care to patients:
(a) Evidence of the validity of clinical
scoring methods and of near patient tests
(b) Evidence that clinical scoring methods
and near patient tests predict benefit from treatment (otherwise
why use them)
(c) Evidence about who is at risk of severe
complications or severe/prolonged symptoms
(d) Evidence about whether those at risk
of severe complications or severe/prolonged symptoms benefit from
Paragraphs (a) and (b) above require validation
studies nested in large randomised trials, (c) and (d) require
case control studies, cohort studies, and trials. This work is
not expensive since it requires large studies with associated
microbiology. Nevertheless, it would easily repay the taxpayer
if it were performed, given the huge costs to the Nation of managing
The problem for gaining funding for this kind
of straightforward applied clinical epidemiology is that in open
competition with other priorities (eg cancer, heart disease, back
pain, mental health etc) and other methodologies (eg laboratory
science) it is difficult to persuade existing funders that this
is a priority (acute infections are common, debilitating, and
incur considerable costs to society, but for most people do not
cause major morbidity or mortality). This is exacerbated by the
fact that NHS regional R+D funding stream has been abolished,
and the MRC is currently suffering cash flow problems.
Thus the Health Protection Agency has an opportunity
to provide a major service to clinical practice, to provide the
key information, which will enable GPs to rationalise prescribing
and contain antibiotic resistance. It will do so by providing
a stream of funding to answer basic questions about the targeting
I hope these points are reasonably clear and
I would be happy to provide clarification or amplification if