Perspectives

“The girls don’t drink much; ’bout the same as the fellas.” I am an Aboriginal woman, with traditional connections to the Bidjara people from central western Queensland and extensive experience in working with Aboriginal women who consume alcohol during pregnancy. During several decades of working in the health field, I have asked myself and others why Aboriginal people drink alcohol at such dangerous levels. As I have reported ...

Lorian G Hayes MAppEpid, BAppHlthSc(Hons)

A systematic search at the source of the smoking problemWhile there has been a redoubling of efforts to reduce smoking rates among Indigenous Australians, Indigenous health researchers have paid almost no attention to the tobacco industry — the “vector” of smoking-related diseases. The exceptions are two reports of the use of Aboriginal images to advertise Winfield cigarettes in Europe.1,2 We decided to examine industry documents ...

David P Thomas MB BS, PhD, FAFPHM · Laura Bond BHSc, MPH

Key performance indicators for medical research do not always indicate the long-term success of a treatmentTranslation into reduced morbidity and mortality is the ultimate measure of success for medical research. However, this may take decades and often researchers are dead before the full impact of their research is realised. Gregor Mendel’s research is an excellent example, with the most important publication in genetics remaining uncited ...

David A Mackey MD, FRANZCO

Is now the time to send a signal that poor-quality care should not be rewarded in activity-based funding?Public hospitals in Australia are in for a shake-up over the next few years, with boards being reintroduced in many states and activity-based funding (ABF) being rolled out nationally.1 ABF will replace global or historic budgets for hospitals in most states. National casemix classifications will be agreed, and ...

Stephen J Duckett PhD, DSc, FASSA

Rachel Nowak explains why improving communication standards will be essential in the drive to speed research translationPssst ... Want your research findings or health guidelines to change how doctors treat patients? Then take a tip from an ex-journo: Use plain English. Keep it short. Tell stories. Journalists have, for centuries, used these tools first to hook ...

Rachel E Nowak BSc, MA, PhD

The moral panic surrounding human embryo research has fuelled unrealistic public expectations for the imminent success of induced pluripotent stem cell therapyTraditionally, treatment for chronic degenerative conditions revolved around reducing symptoms and improving physiological functioning in the hope of gaining some (often limited) increase in life expectancy. However, in the past decade, the advent of ...

Ronald K F Fung BSc(Adv) · Ian H Kerridge MPhil, FRACP, FRCPA · Loane L C Skene LLB(Hons), LLM, LLD · Megan J Munsie BAppSc, MRepSc, PhD

Ray Moynihan hears a new professional–consumer alliance ringing the alarm on overused interventionsSometimes truth is stranger than fiction. In the United States a few weeks ago, nine leading doctors’ groups published lists of the top five overused tests and treatments within their specialty areas.1 What’s more, they did it with enormous fanfare and in partnership ...

Ray N Moynihan BA

Youth mental health expert Alison Yung says safe treatments should be offeredSchizophrenia carries a great personal and economic burden. It begins with a non-psychotic prodromal phase before the onset of characteristic positive symptoms such as delusions and hallucinations. This prodromal phase is therefore a potential target for early intervention. Criteria have been developed to identify the ...

Alison R Yung MD, MPM, FRANZCP

Schizophrenia expert David Castle suggests caution, especially regarding medication useIt has been suggested that one can “predict” schizophrenia at an individual level. Features posited as being part of these so-called ultra-high risk (UHR) criteria1 include risk factors (eg, a family history of schizophrenia), symptoms (eg, attenuated positive psychotic symptoms) and longitudinal trajectory (decline in functioning). ...

David J Castle MD, FRANZCP, FRCPsych

It is now less clear how parents should be advisedIn Australia, stillbirth is defined as the loss of a fetus of at least 20 weeks’ gestation, or 400 g birth weight if gestational age is unknown. Every year in Australia there are 1750 stillbirths.1 Several studies have reported that stillbirth, regardless of how it is managed, can have ...

Kelly A Cunningham BSc, PhD

John Cartmill and David Butt discuss how surgery might change with robot-like enhancements to surgeons’ capabilitiesThe advent of the surgical “robot”, exemplified by the da Vinci Surgical System (Intuitive Surgical, Sunnyvale, Calif, USA), brings fresh approaches, possibilities and questions to the field of surgery. Because of their capacity to augment dexterity, strength, coordination and vision, ...

John A Cartmill · David Butt

Data sharing would speed progress in improving the Australian health systemThe benefits of data sharing have been demonstrated in genomics, where it has accelerated progress and provided an understanding of complex issues sooner.1 Hoping to ...

Michael D Coory MB BS, FAFPHM, PhD · Craig L Fry BSc(Hons), DrPH

Reconsidering the place of disease monitoring after treatmentWomen who have completed primary chemotherapy for ovarian cancer commonly have serial assessment of the serum tumour marker cancer antigen 125 (CA-125).1 This practice has been based on the ...

Paul Harnett MB BS(Hons), FRACP, PhD · Ian H Kerridge BMed(Hons), FRCPA, FRACP · Christopher F C Jordens BA(Hons), MPH, PhD · Kim Hobbs BSocStud(Hons) · Catherine Mason MB BS, MPH, FRANZCP · Bronwen M Morrell BA(Hons)

Opioid use is increasing, and not just for chronic conditions or cancer Global concern about the prescribing of opioid analgesic medications (particularly oxycodone) for chronic pain is growing. In this Journal in September 2011, Roxburgh and colleagues examined the rise of oxycodone and morphine prescribing and the harm associated with opioid use.1 In the accompanying editorial, Hall and Farrell argued that this suggests that most prescribing of opioids was for chronic non-malignant pain.2 We tested this assertion using general practitioner prescribing data linked to the problem under management, and examined the characteristics of the GPs who prescribed opioids and the patients who received these prescriptions. We analysed data from the Bettering the Evaluation and Care of Health (BEACH) program (detailed methods are described elsewhere3). For the period April 2010 to March 2011, we examined encounters at which opioids were either prescribed or supplied directly to patients. We then measured changes in opioid and oxycodone prescribing rates over the past decade and extrapolated these results to all Medicare GP service items claimed nationally.4 While Roxburgh and colleagues1 reported dispensing rates as measured by the Pharmaceutical Benefits Scheme (PBS), BEACH data measure GP prescription and supply actions. Using the international Anatomical Therapeutic Chemical classification,5 we defined opioids as “N02A” and oxycodone as “N02AA05”. We determined the statistical significance of differences by non-overlapping 95% CIs. In the 2010–11 BEACH data year, at least one opioid was prescribed or supplied at 4666 of the 95 839 encounters (4.9%) recorded from 892 GPs (93.1% of the 958 GPs who participated in that year). There were a total of 5350 opioids prescribed (5237; 97.9%) or supplied (113; 2.1%) by the GP at the 4666 encounters. From this point, we shall refer to them all as being prescribed. Compared with encounters at which no opioids were recorded, those at which an opioid was prescribed were four times more likely to be claimable through workers compensation (6.8%; 95% CI, 5.7%–7.8% v 1.7%; 95% CI, 1.5%–1.9%), and more likely to be long or prolonged consultations (10.0%; 95% CI, 8.8%–11.2% v 7.8%, 95% CI, 7.2%–8.4%). Opioids were prescribed at a rate of 3.7 per 100 problems managed, accounting for 5.8% of all medications prescribed or supplied by participating GPs. There was no significant difference in opioid prescribing rates for male and female patients. Opioid prescribing was rare for young patients, rose significantly for patients aged 25–44 years, peaked among 45–64-year-old patients and stayed almost as high in older age groups. Opioid prescribing was significantly less for patients new to the practice, possibly because of their younger age (results not shown) or GP reluctance to prescribe opioids at an initial consultation (Box 1). Commonwealth Health Care Card holders had a significantly higher rate of opioid prescribing than other patients; this rate was even higher than for patients aged 65 years and older, suggesting that age was not the only contributing factor. While the prescribing rate for opioids did not differ with the age group of the GP, male GPs prescribed significantly more opioids per 100 encounters than did females (Box 2), adding to previous research evidence showing that female GPs prescribe fewer medications than males.6 Malignant neoplasms accounted for only 3.5% of opioids prescribed, and chronic non-cancer conditions for a further 43.9%, so a significant proportion were prescribed for conditions classed as non-chronic.7 Almost 60% of prescriptions were for musculoskeletal problems. The higher prescribing rate by male GPs may reflect their higher management rate of musculoskeletal problems when compared with female GPs.6 Back problems accounted for over a quarter of all opioids prescribed and was the most common individual problem for which opioids were provided. About one in 10 opioids prescribed were for osteoarthritis, provided at almost one in five occasions of osteoarthritis management. Less than 7% of opioids prescribed were for generalised multisite pain, but 72.7% of generalised multisite pain managements had an opioid prescribed (Box 2). The most common opioids prescribed were combinations of paracetamol with at least 30 mg codeine (the minimum amount for classification as an opioid). Extrapolation suggests that there were over two million prescriptions for the year, nationally. The next most common opioid was oxycodone, followed by tramadol (Box 2). Over the past decade, the total prescribing rate for total analgesics remained steady4 but the prescribing rate for opioids rose from 3.8 per 100 encounters in 2000–01 to 5.6 per 100 in 2010–11; this was particularly the case for oxycodone (up from 0.3 prescriptions per 100 encounters to 1.5 per 100 over the period). Extrapolation suggests GP prescriptions for opioids almost doubled (from 3.83 million to 6.65 million), while the number of oxycodone prescriptions increased almost sevenfold (from 0.26 million to 1.73 million). These increases reflect both the increased prescribing rate and the increase in the number of GP visits claimed through Medicare in recent years. Our results show that Hall and Farrell were correct in their assertion that chronic non-malignant pain accounted for a far greater proportion of opioid prescriptions than did malignant neoplasms.2 However, we have also shown that a significant proportion of opioid prescriptions are for problems classified as non-chronic. The classification of a problem as non-chronic does not preclude it being chronic in some patients (although it is not usually chronic), nor does it preclude involvement of severe pain over an extended period. 1 Opioid prescribing at Australian general practice encounters by patient characteristics (based on weighted data) Patient characteristics Patients at opioid encounters Patient characteristic opioid prescribing rate per 100 encounters (95% CI) Total encounters 4666 5.6 (5.3–5.9) Sex (missing data, 44) Male 2572 (44.4%) 5.8 (5.4–6.3) Female 2050 (55.6%) 5.4 (5.0–5.7) Age group (missing data, 30) 0–14 years 6 (0.1%) 0.1 (0.0–0.1) 15–24 years 152 (3.3%) 2.0 (1.6–2.3) 25–44 years 1154 (24.9%) 6.1 (5.5–6.6) 45–64 years 1729 (37.3%) 7.7 (7.1–8.3) 65–74 years 734 (15.8%) 6.7 (6.1–7.3) ≥ 75 years 860 (18.6%) 6.5 (5.9–7.0) New patient to practice (missing data, 56) 213 (4.6%) 3.4 (2.8–4.0) Commonwealth concession card (missing data, 304) 2753 (63.1%) 8.2 (7.6–8.7) 2 Opioid prescriptions or supplies (n = 5350) at 4666 Australian general practice encounters (based on weighted data) A: Characteristics of general practitioners prescribing opioids GP characteristics Opioid prescriptions GP age- and sex-specific opioid prescribing rate per 100 encounters (95% CI) Sex (missing data, 0) Male 4150 (77.6%) 6.1 (5.7–6.6) Female 1200 (22.4%) 4.2 (3.8–4.7) Age group (missing data, 17) < 35 years 275 (5.2%) 6.1 (4.6–7.7) 35–44 years 889 (16.7%) 5.1 (4.4–5.7) 45–54 years 1686 (31.6%) 5.4 (4.9–5.9) ≥ 55 years 2482 (46.6%) 5.9 (5.4–6.4) B: Health problems managed with opioids Health problem Opioid prescriptions % of this problem treated with an opioid Back problem* 1449 (27.1%) 37.9% Osteoarthritis† 517 (9.7%) 18.4% Generalised multisite pain 351 (6.6%) 72.7% Fracture† 205 (3.8%) 20.3% Prescription—all† 198 (3.7%) 6.9% Malignant neoplasms—all‡ 187 (3.5%) 10.7% Sprain/strain† 147 (2.7%) 10.5% Migraine 108 (2.0%) 18.1% All musculoskeletal 3114 (58.2%) 17.3% Chronic non-cancer conditions§ 2350 (43.9%) 4.1% C: Main opioids prescribed or supplied Opioids Opioid prescriptions Estimated number nationally in 2010–11 Paracetamol/codeine (30 mg)¶ 1730 (32.3%) 2 130 000 Oxycodone 1406 (26.3%) 1 730 000 Tramadol 864 (16.2%) 1 060 000 Buprenorphine 427 (8.0%) 530 000 Morphine sulphate 371 (6.9%) 460 000 Fentanyl 252 (4.7%) 310 000 Dextropropoxyphene/paracetamol 180 (3.4%) 220 000 * Includes back symptom/complaint, low back symptom/complaint, back syndrome without radiating pain and back syndrome with radiating pain. † Includes multiple International Classification of Primary Care – Version 2 (ICPC-2) or ICPC-2 PLUS codes (see appendix 4 of Britt et al3). ‡ Includes all malignant neoplasms except of skin, for which only melanomas are included. § Chronic conditions are defined as in O’Halloran et al.7 ¶ Paracetamol and codeine combinations with less than 30 mg codeine are not classified as opioids.

Christopher M Harrison BPsych(Hons), MSocHlth · Janice Charles BA, MSc(Med) · Joan Henderson BAppSc(HIM)(Hons), PhD(Med) · Helena Britt BA, PhD

Knowing if a patient is Indigenous can be key to improving their care An article published in 2010 in the New England Journal of Medicine advocated the collection of data on the race and ethnic groups of patients by medical practices in the United States. This was part of an initiative to computerise medical records and broaden the collection of demographic data. The authors suggested that such data could be used to detect health disparities, optimise the effectiveness of quality improvement interventions, and generate more reliable data on quality of care and outreach to patients. According to the authors, the “most common and strongest objection” was that doctors (and other practice staff) believed that “knowing a patient’s race and ethnic group is, or should be, clinically irrelevant”.1 Doctors are motivated by quality of care,1 but in this US initiative, there is an incentive for the data collection in the form of additional reimbursement. In Australia, incentives have been introduced for doctors to improve the management of infectious and chronic disease among Indigenous Australians (eg, payment for additional immunisations for children, access to health assessments at an earlier age and programs to improve access to medicines).2,3 However, these incentives have generally had limited uptake.2,3 This has been attributed to a failure to collect information on whether or not patients are Indigenous. In 2009, using a combination of an analysis of the available literature and de-novo qualitative research, we examined strategies for improving the identification of Indigenous people in general practice in order to facilitate the uptake of new measures associated with the Closing the Gap objective.4 Three issues emerged from discussions with general practitioners that have important implications for improving the identification of Indigenous patients in general practice. First, there was a clear consensus that the poorer health status of Indigenous patients indicated a need to improve their medical care. Second, in order to convince clinicians and overcome resistance to identifying Indigenous patients, the relevance of identification to quality of care needs to be demonstrated. Third, the experienced clinicians interviewed for our report could not provide any examples demonstrating how identifying Indigenous patients had improved medical care. In part, this last issue seemed to arise from the way clinicians conceptualise quality of care. Campbell and colleagues proposed a model of quality of care with two dimensions — access and effectiveness.5 Access means availability, accessibility, affordability and acceptability while effectiveness is operationalised as the processes of interpersonal and clinical care. Interpersonal care relates to the interaction between health care providers and their patients. Clinical care relates to the technical aspects of care.5 This distinction is also reflected in guidelines developed in the US to address disparities in health.6 Our research indicated that one of the barriers preventing clinicians from making the link between identifying their Indigenous patients and and quality of care is that they tend to focus on effectiveness of care rather than access to care. For example, in focus groups, none of the clinicians talked about access to additional or modified health services as a reason for identifying Indigenous patients, even though this is a cornerstone in reducing health disparities.4 However, at case study sites, we found that clinicians who did associate identification with accessing specific services or applying different sets of clinical guidelines were more confident and willing to ask patients about their ethnicity. Focusing on awareness of health services specifically available to Indigenous people, the rationale underlying these services and the associated incentives resulted in significant increases in completed health assessment for Indigenous people.4 Making the link between the collection of identifying data with quality of care is also important from the perspective of the community. Consultation with the community has revealed that, while many community members were prepared to identify as Indigenous, they did not see this information as relevant to the quality of care they received.7 Understanding that they might directly benefit from information provision was seen as further incentive for patients to identify as Indigenous. Not surprisingly, displaying promotional material in practices (including Indigenous flags etc.) significantly improved Indigenous identification. Knowing a patient’s ethnicity can help orientate clinicians to a range of issues and health service options that they might not otherwise have considered. Ensuring that identification results in improved care requires that this information (like any other aspect of a patient’s social history) is considered a starting point for enquiry that reflects the clinical and interpersonal imperatives of any interaction. Increasing the willingness of doctors to enquire about ethnicity will require a doctor more easily making the link between knowing a patient’s ethnicity and his or her quality of care. While this will involve improving doctor and patient awareness, it will also require having systems in place that enable information about ethnicity to contribute explicitly to improved quality of care.

Margaret Kelaher BScPsych, PhD · Amy Parry BA, DipPH · Susan Day BA, PhD · Yin Paradies MMedStats, MPH, PhD · Ian Anderson MBBS, PhD

Social networking and search engine data confer real-time advantages The population-level pattern-based nature of epidemiological research makes it well suited for computational work in general, and for machine learning in particular. The social nature of disease spread makes recent trends in social media computing specifically amenable to epidemiological research, but can computational techniques be reliable indicators and predictors of communicable disease? Previous approaches to epidemiology have had to rely on self-report mechanisms (eg, online health surveillance at www.flutracking.net), or on reports from health care services, such as the United Kingdom Health Protection Agency (HPA), the United States Centers for Disease Control (CDC) and the European Centre for Disease Prevention and Control. The recent growth in the use of social media means that people now volunteer large amounts of information on a real-time and location-specific basis. Along with this, advances in computational intelligence in the form of machine-learning and data-mining techniques have proven useful in knowledge discovery and predictions in other domains. So, how best can we make use of and interpret that vast quantity of information? The information is of two main kinds: that provided by users (status updates and microblogging, such as on Facebook and Twitter); and requests for information using search engines such as Google and Yahoo. Patterns found in this information can be interpreted by applying machine-learning and data-mining techniques. Roughly speaking, machine learning and data mining are used to make predictions based on patterns learned from data and discovering patterns in data. In the words of Tom Mitchell, “A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E”.1 Researchers from the University of Bristol in the UK have been using Twitter to investigate the possibility of tracking influenza spread. They collected about 160 000 tweets per day over 24 weeks from the 54 most populated areas in the UK, in which they sought, for example, reports of sore throat, fever or headache. Reports from the HPA (based on general practitioner consultations per 100 000 citizens resulting in influenza diagnoses) were used as the “gold standard” basis for disease activity when learning to predict influenza rates. Learning and discovering patterns resulted in the ability to predict HPA flu rates with about 90% accuracy.2 The University of Iowa used a similar approach, whereby public sentiment about pandemic (H1N1) 2009 influenza and actual disease activity was tracked using Twitter posts. Using about a million influenza-related tweets and the CDC’s reported data, machine learning was used to construct a predictive model. Even though the model does not predict disease activity, it was able to estimate activity in real-time, reflecting the data in the official CDC reports. The advantage is that “real-time” is typically 1–2 weeks ahead of CDC reports.3 People do not use use Twitter to discuss their health — many use search engines to research symptoms, treatments, spread of diseases, and so on. Google conducted an experiment on monitoring search queries related to influenza. An analysis of 5 years’ worth of log-files in combination with available CDC information was used to construct a regression model for influenza surveillance, which contains the top 45 search terms. This model was used to predict the spread of influenza in the 2007–2008 influenza season. As with the University of Iowa example, the system was able to consistently estimate the percentage of the population with influenza 1–2 weeks ahead of official CDC reports.4 I have highlighted just some of the emerging work on applying clever algorithms and large amounts of computational power to vast amounts of information generated by users of search engines and social media. Despite the impressive results, these approaches are not a panacea for epidemic surveillance. There are challenges, some shared with existing techniques and some that are unique. People discussing their health on Twitter are not representative of the general population, and Twitter use is not uniform across time and geography. Demographic data provided by traditional surveillance cannot (yet) be supplied by search queries. Additionally, it is difficult to separate discussion about epidemics from actual cases. Regardless of current shortcomings, these approaches will prove to be important parts of modern medicine.

Anders Kofod-Petersen DrScient

Veterinary pharmacologist Stephen Page believes that antibiotic use in animals benefits human and animal health While antibiotic use in livestock has the potential to cause harm to human health, the likelihood of harm arising from the use of antibiotics under current Australian practices is highly unlikely. There is ample independent evidence supporting the long history of appropriate use of antibiotics in Australian livestock. For example, the past six annual reports of the National Residue Survey contain the results of tests for the presence of up to 38 antibacterial agents in the edible tissues of cattle, pigs, poultry and sheep.1 Almost 400 000 assays were performed and compliance with Australian standards was at or above 99.99% for each of the four species tested. Given the high morbidity and mortality associated with many of the diseases of livestock (eg, necrotic enteritis and coccidiosis in poultry, bovine respiratory disease and mastitis in cattle, and post-weaning diarrhoea and porcine respiratory disease in pigs), prevention of infectious diseases is of fundamental importance. Judicious use of antibiotics in integrated animal health plans has an important role to play in protecting the health and welfare of animals and in contributing to the safety and quality of the food that is produced. Only 39 antibacterial agents are approved and available for use in Australian livestock, six of which are from classes with no human medical use counterpart. As an example of contemporary patterns of use, a recent survey of the use of antimicrobial agents in the Australian pig industry found most use was for control of infections with Lawsonia intracellularis, Mycoplasma hyopneumoniae and Escherichia coli and relied on sulfonamides, tetracyclines and penicillins2 — all drugs considered by the Expert Advisory Group on Antimicrobial Resistance to be of low importance to public health.3 What led the Australian Department of Health and Ageing to conclude that “Australia’s food supply is one of the safest and cleanest in the world”4? In 2008, Food Science Australia was commissioned to survey the presence of bacteria with antimicrobial resistance in samples of raw whole poultry, beef mince and pork chops collected from shops in Melbourne, Sydney, Brisbane and Perth. Among the isolates of Campylobacter jejuni, Salmonella and E. coli that were collected, no fluoroquinolone or ceftiofur resistance was identified. No Enterococcus faecium was isolated and all Enterococcus faecalis isolates were susceptible to ampicillin, penicillin and vancomycin; only one of 300 E. faecalis isolates was resistant to gentamicin. In contrast, Denmark, despite being recognised as having one of the most regulated and conservative approaches to the use of antibiotics in livestock, noted significant antibiotic resistance in their most recent antimicrobial resistance monitoring report: fluoroquinolone-resistant C. jejuni in 17% of broiler meat samples; high-level gentamicin resistance in E. faecalis from pork samples; vancomycin-resistant E. faecium in 47% of isolates from broiler chickens; extended-spectrum β-lactamase producing E. coli (ceftriaxone resistant) in 27% of broiler cloacal swabs and 9% of broiler meat samples; fluoroquinolone-resistant Salmonella enterica subspecies enterica serovar Enteritidis in 8% of sporadic cases of human salmonellosis; multiresistant Salmonella enterica subspecies enterica serovar Typhimurium definitive type 104 (DT104) in 7% of isolates from pigs; and ceftiofur-resistant E. coli in 1% of domestic pork and broiler meat samples.5 OzFoodNet (http://www.ozfoodnet.gov.au) was established by the Australian Government in 2000 to enhance surveillance of foodborne disease. It has published eight annual reports, which do not identify any cases of domestically acquired foodborne illness caused by infections with Salmonella, Campylobacter or Shiga toxin-producing E. coli complicated by antibiotic resistance of livestock origin. By contrast, a number of human cases of fluoroquinolone-resistant Campylobacter infection and S. Typhimurium DT104 infection have been described, but each originated from overseas (especially South-East Asia) or from imported food. Since 2001, the Australian Salmonella Reference Centre has tested more than 18 000 Salmonella isolates from cattle, pigs, sheep, chickens (broilers and layers), raw meat and eggs. No fluoroquinolone-resistant isolates were found and only nine isolates (most of which were submitted before 2004) were phenotypically resistant to cefotaxime at 1 g/mL. Australians should be reassured that food safety is the highest priority in livestock production. The Australian Veterinary Association is currently updating a series of guidelines for responsible use of antibiotics, the poultry industry is supporting leading edge research to develop a vaccine to prevent necrotic enteritis, and the pig industry has spent over $7 million on projects to refine and replace the use antibiotics and has committed to halving antibiotic use by 2018. Antibiotics used prudently and selectively can clearly benefit animal health and welfare, leading to overall human health benefits, not risks.

Stephen W Page BVSc(Hons), MVetClinStud, MAppSci(EnvTox)