Epidemiology of tuberculosis and levels of vitamin D in Australia: person, place and time

To the Editor: We recently reported the impact of latitude on seasonality of tuberculosis in Australia, with greater cyclic variation in southern parts of the continent.1 We hypothesise that this seasonality is partly determined by differences in ultraviolet radiation exposure and subsequent vitamin D synthesis.1–3

Vitamin D deficiency (serum
25-hydroxyvitamin D levels below 50 nmol/L)3 has become a significant public health concern in Australia. Australians with darker skin, including some migrants and Aboriginal and Torres Strait Islander people, are at particular risk of both vitamin D deficiency3,4 and tuberculosis.1,5

Similar to our findings regarding tuberculosis incidence,1 a recent Australian study found that vitamin D deficiency was most prevalent in spring and that risk was highest for residents in major cities, people from socioeconomically disadvantaged areas, and those aged 20–39 and
≥ 80 years.2 These factors also apply to tuberculosis, including the age distribution (Box).

The correlations between seasonal variations in, and risk factors for, vitamin D deficiency and tuberculosis in Australia reinforce the ecological association between these conditions. Such associations cannot determine causality; but their consistency argues that guidelines should consider the potential impact of vitamin D deficiency on people at greatest risk
of tuberculosis.1 The increased risk of tuberculosis conferred by vitamin D receptor polymorphisms supports a causal role for vitamin D deficiency in active tuberculosis.6

Despite increasing observational data regarding vitamin D deficiency and risk of tuberculosis, evidence supporting vitamin D supplementa-tion to reduce this risk is lacking.6 However, given the broader potential health impacts of vitamin D deficiency in high-risk populations, we support recent calls to increase vitamin D testing in these groups and to promote supplementation for those at greatest risk of both tuberculosis and vitamin D deficiency, including migrants with darker skin.3

Serum 25-hydroxyvitamin D levels of at least 50 nmol/L at the end of winter have been recommended for optimal bone and muscle function,3 with supplementation continued into spring to avert depletion.2 This seasonal focus on ensuring adequate vitamin D levels in people at risk of both vitamin D deficiency and tuberculosis could also reduce the seasonal peak of disease each spring, particularly in the southern states of Australia.1

Age distribution of tuberculosis (TB) notifications* and vitamin D levels† in Australia

25-OHD = 25-hydroxyvitamin D. * Source: National Notifiable Diseases Surveillance System. † Source: Boyages and Bilinski.2

Millions unaware of ticking hypertension time bomb

Almost half of patients with hypertension are unaware they have the potentially life-threatening condition, an Australian-led international study has found.

The research, conducted in 17 countries and involving more than 153,000 people, showed that two thirds of those diagnosed with hypertension did not receive adequate care for their condition.

According to the study, published in the Journal of the American Medical Association (JAMA), a majority of those who knew they had high blood pressure received drug treatment for hypertension. However, only a third of those had their hypertension under control.

There is a significant need for better diagnosis and treatment.

Senior author and Professor of Medicine at McMaster University, Dr Salim Yusuf, said that drug treatments that work to control hypertension were well known.

Associate Professor Clara Chow, from The George Institute of Global Health, and the University of Sydney, agreed, and said treatments were generally inexpensive – but only about one third of patients were achieving their target blood pressure.

“The results are alarming,” Associate Professor Chow said.

Associate Professor Chow and her colleagues organised the study in three high-income countries: Canada, Sweden, and United Arab Emirates; 10 middle-income countries: Argentina, Brazil, Chile, Poland, Turkey, Malaysia, South Africa, China, Colombia, and Iran; and four low-income countries: Bangladesh, India, Pakistan, and Zimbabwe.

“Although Australia is not included in the study, it faces similar problems,” she said.

Another major issue with patients who are unaware of hypertension, is lack of symptoms. Patients tend to resist taking daily doses of medication if they do not see or feel any major symptoms.

“Australians should realise from this that hypertension is exceedingly common and nearly half of us are walking around with it, not knowing that our blood pressure is high.

“Australians should have regular annual checks,” Associate Professor Chow said.

Sanja Novakovic

Better information to drive health investment decisions

Strengthening the use of data will prevent health decisionmakers from “flying blind”

Federal and state governments spend a large and increasing proportion of their budgets on health care. Although rising costs are partly driven by an ageing population, increases in the volume of services per patient are also a major cause, while advancing science, community expectations and defensive medicine contribute to “doing more”. The major challenge facing the Australian health care system in the next electoral term will therefore be to rein in rising costs while maintaining high standards of care and meeting new demands.

Better information about the delivery of health programs and services in Australia must drive policy responses to this challenge. Such information will inform community dialogue about our health care spending priorities and underpin decisions about when new programs, tests or treatments should be offered, continued or ended.

The past decade has seen progress towards high-quality, timely and locally relevant information required by health decisionmakers. However, if Australia is to meet the challenge of rising health care costs, the information and its effective use must be considerably strengthened — otherwise, health decisionmakers are “flying blind”. For example, a better understanding of the causes of unwarranted variations in practice can provide powerful insights into how to reduce costs while maintaining the quality of care and meeting growing demands. In the United States, a twofold difference has been found between geographic regions in per capita Medicare spending; but for some conditions, higher spending does not result in better patient outcomes.1 Considerable variations in practice also exist in Australia.

Improved data systems and more sophisticated approaches to understanding the causes and meaning of variation will be required to make good use of this information in health decisions. A recent article in the Journal illustrates the importance of testing the methods and data, if the right conclusions are to be drawn from performance monitoring and analysis.2

Another opportunity to inform health decisions is to embed rigorous evaluations into the rollout of new policies, programs and service delivery models. It is sometimes possible to structure the rollout of a program to provide for rigorous testing of policy alternatives. For example, in the United Kingdom, a randomised trial was established within the implementation of the national breast cancer screening program to test whether inviting women aged in their forties reduced deaths from the disease.3 More often, it would be possible to establish multifaceted evaluations that shed light on different components of the program, such as that established by the Scottish Government to assess the impact of a smoking ban in public places. Seven separate evaluations examined the impacts of the program, including its effect on hospital admissions for acute coronary syndrome, reduction in second-hand smoke exposure in 11-year-olds, reduction in second-hand smoke at public bars and public support for the policy.4 The evaluation of both of these initiatives influenced subsequent policy decisions.

Although evaluating the operation of programs and services in routine practice is critical to investment or disinvestment decisions, rigorous evaluation is not commonly practised in Australia. Where it does occur, evaluation is often insufficient to draw meaningful conclusions about costs and benefits. The failure to routinely incorporate rigorous evaluation means “we have wasted huge opportunities to learn”.5 The political will to encourage or mandate routine high-quality evaluation of new policies, programs or services would do much to inform health decisions. In the UK, for example, the Treasury places evaluation at the heart of programs and planning.6

Finally, a strong information base for health decisions will require closer partnerships between researchers and health decisionmakers. Health decisionmakers often report that research does not address the questions of most interest to them such as program cost and scalability, and the most likely beneficiaries. Engaging health decisionmakers as partners in the design and implementation of research will result in more robust information for health decisions. Several agencies overseas have pioneered new models of research funding, including the Canadian Health Services Research Foundation and the UK National Institute for Health Research. In Australia, the National Health and Medical Research Council has initiated Partnership Projects and the new Partnership Centres for better health.

The recent McKeon Review emphasised the benefits of continued support for research intended to inform health decisions.7 Supporting this type of research will, over time, help improve care while keeping costs under better control.

Doctors ahead in slow early going

Doctors and other professionals have emerged as one of the early winners from the 2013 Federal election campaign after Labor decided to defer its highly unpopular cap on tax deductions for self-education expenses.

While the AMA and more than 60 other professional organisations, business groups and education providers remain committed to having the measure scrapped, the Federal Government hopes to have drawn much of the short-term electoral sting out of the issue by holding the introduction of the cap over until mid-2015.

But, aside from this move – made on the eve of the campaign proper – the Australian Greens have made most of the early running.

Even before the election was called, they had flagged they would reinstate indexation of Medicare rebates.

Barely a week later, the Greens again grabbed the initiative by promising to establish – as per AMA policy – an independent panel of medical experts to monitor and report on health care for asylum seekers being held in detention.

They followed this up with a commitment to “stop the bombardment” of junk food advertisements on television directed at children.

The announcements overshadowed a commitment by Labor to provide $20 million to prevent, research and manage foetal alcohol spectrum disorder, which has blighted the lives of many children, particularly from disadvantaged communities.

Even less visible has been the Coalition, which has so far largely kept its own counsel on health policy.

Adrian Rollins

Promises, promises….

Who has promised what on health policy so far…

Announcement	Date	Cost	Comment/reaction
Labor
$2000 cap on tax deductions for self-education expenses deferred 12 months to mid-2015	2 August	$250 million over four years	AMA and Scrap the Cap Alliance vow to continue fighting measure; Coalition yet to comment
Medicare rebate indexation frozen to mid-2014	14 May	$664 million	Greens promise to reinstate indexation
Foetal alcohol spectrum disorder strategy	6 August	$20 million
Coalition
Medicare Locals to be reviewed	24 May	n.a.
Medibank Private to be sold off		tbc	Greens don’t rule out supporting sell off
Private health insurance means test to be scrapped	24 May	tbc
Government policy to cap indexation of private health insurance rebate backed	27 May	$700 million saving
National Health and Medical Research Council funding guaranteed	24 June	$3.7 billion
Greens
Medicare rebate indexation reinstated	27 July	$600 million
Independent Health Advisory Panel to oversee asylum seeker health care	6 August	$2 million
Universal dental care	27 July	tbc
Ban on junk food ads to children	8 August	n.a.

n.a. – not applicable

tbc – to be confirmed

The SNAPSHOT ACS study: getting the big picture on acute coronary syndrome

How snapshot methodology identifies factors limiting translation of evidence to practice

Translating evidence-based treatments into clinical practice is fundamental to modern health care delivery. Yet numerous studies demonstrate limited uptake of guideline-endorsed treatment recommendations.1,2 Why is this so? There are many possible explanations. Patient characteristics such as age, comorbidity, socioeconomic status, cultural background and frailty are likely to be important. Most trials of novel drugs or devices are funded by industry. Trials are very expensive, and a trial sponsor is understandably keen to ensure that their product is administered to those patients most likely to benefit. Consequently, patients entered into clinical trials are typically younger and have less comorbidity than the broader population of patients with a particular condition. This was found to be the case in a recent Canadian registry report on patients admitted with acute coronary syndrome (ACS), which compared the baseline characteristics of those who were included in clinical trials with the much larger cohort of patients who were not.3 Other variables also limit the translation of evidence-based treatments. For ACS, which requires acute hospital care, the type of hospital (eg, peripheral versus major teaching hospital) and its location (eg, urban versus rural or remote setting) can determine the level of medical expertise and complexity of treatment offered to a patient.

How best then to identify all the factors that limit the translation of evidence-based medicine into clinical practice? One approach is to take a “snapshot” of all patients with an acute condition admitted to every hospital nationwide and to compare patient characteristics, patterns of investigation, treatment and patient outcomes. This would allow hospitals, states and even countries to be compared, if the snapshot is broad enough. Undertaking such a study is logistically daunting, but in a landmark study published in this issue of the Journal, Chew and colleagues have attempted exactly that.4 The investigators approached 525 hospitals across Australia and New Zealand and obtained ethics approval from 478 to perform a prospective audit of patients admitted with suspected or confirmed ACS during 2 weeks in May 2012. They enrolled 4398 patients in 286 hospitals. Hospitals that did not enrol patients were smaller and did not admit patients with ACS during the snapshot window.4

Three design features of the SNAPSHOT ACS study are worth highlighting. First, use of an opt-out consent process (a consent waiver in New Zealand) no doubt contributed to the high recruitment rate.5 Second, standardised data definitions and real-time data collection ensured consistency between participating hospitals, in contrast to other methods of administrative data extraction where data validity is often questioned.6 Third, including a survey to document the range of cardiac services at the participating hospitals facilitated benchmarking between institutions. Facilities for percutaneous coronary intervention and coronary artery bypass surgery were available in less than 30% and 20% of hospitals, respectively. This expected finding probably explains the high rate of interhospital transfer, involving 26% of enrolled patients. However, surprisingly, only two-thirds of the 286 hospitals in the study contained an acute coronary care ward.

The SNAPSHOT ACS study is in every sense a contemporary “real world” audit of the diagnostic evaluation and management of ACS across Australasia. The authors report some noteworthy findings. Although virtually all patients presented to a hospital that was capable of providing acute reperfusion therapy for ST-segment-elevation myocardial infarction, this was administered to only two-thirds of patients with this diagnosis. In addition, patients with higher Global Registry of Acute Coronary Events (GRACE) scores (and higher risk of hospital mortality) were less likely to receive an invasive management strategy. Previous publications from large prospective registries have reported similar findings.7,8 This may be explained by factors such as advanced age and renal failure, which contribute to the GRACE score but also reduce the likelihood of acute intervention.3 Other major comorbidities are also likely to contribute to underuse of evidence-based therapies in these high-risk patients.7 These and other findings deserve further published analysis of this unique dataset.

Most disturbing are the substantial variations found in implementation of evidence-based management and in patient outcomes, not only between hospitals but also between jurisdictions. Differences in the use of invasive diagnostic and therapeutic measures (coronary angiography and percutaneous or surgical revascularisation) are to be expected, as most enrolled hospitals lacked these facilities.4 On the other hand, the significant differences between hospitals in the use of guideline-recommended medications, cardiac rehabilitation, and diet and exercise advice are harder to explain. An important feature of the SNAPSHOT ACS study is the benchmarking of (de-identified) hospital data to enable individual clinicians and institutions to compare their performance against other institutions providing comparable services. But benchmarking needs to be followed by change-management strategies to improve the uptake of evidence-based treatments in underperforming hospitals.

Jurisdictional differences in use of evidence-based therapies for ACS and patient outcomes indicate differences in health service policy, resourcing and delivery. Exploring these differences will be important to identify jurisdictional barriers to implementing evidence-based treatments and will require bringing together national and state government agencies, professional bodies, individual clinicians and researchers across all hospitals. This approach has been applied successfully (albeit on a limited scale) in the management of ACS, acute heart failure and stroke.9,10 However, a clear message from the SNAPSHOT ACS study is that much broader engagement of these stakeholders will be required to improve the outcomes of patients with ACS across the entire health sector.

The impact of a genomic assay (Oncotype DX) on adjuvant treatment recommendations in early breast cancer

Breast cancer is the most common malignancy in Australian women and most of these tumours are hormone receptor-positive (HR+).1–3 Adjuvant therapy for invasive breast cancer has been shown to improve survival.4 For early-stage, HR+, human epidermal growth factor receptor 2-negative (HER2−) cancer, this therapy may consist of hormonal therapy (HT), or a combination of chemotherapy and hormonal therapy (CHT). It is accepted that the vast majority of patients with HR+, HER2− tumours should receive HT; however, the threshold for use of CHT is difficult to define because this group includes patients with a spectrum of recurrence risks.4,5 A large number of patients in whom the disease would not recur still receive CHT because accurate tools to help define the benefit of adjuvant chemotherapy have not been available.6,7

The 21-gene Oncotype DX breast cancer assay (Genomic Health) was developed on mRNA extracted from archived tumour samples of 447 patients from three studies. The expression patterns of 21 genes were used to develop an algorithm that yields a recurrence score. The score has been validated to quantify the risk of distant recurrence in patients treated with tamoxifen and who had both HR+, lymph node-negative or lymph node-positive disease, as well as to predict the magnitude of benefit from the addition of chemotherapy to tamoxifen treatment.8–11 The assay has been widely available in the United States since 2004, and is available in Australia for $4000. There have been a number of studies examining the effect of the Oncotype DX assay on clinical decision making.12–17 The results have been fairly consistent, showing a shift in the treatment decision in about 30% of patients.12–15,17

These studies have largely focused on decisions made by individual practitioners. In Australia, the multidisciplinary meeting (MDM) has become the standard forum for determining treatment recommendations. A national goal is for all patients to have their treatment decisions discussed in an MDM before definitive treatment recommendations are made.18 In this study, we have assessed the effect of recurrence score information on treatment recommendations in the MDM.

Methods

Study design

The study was conducted between November 2010 and September 2011 at three institutions in Melbourne, Australia: the Royal Melbourne Hospital, the Peter MacCallum Cancer Centre and Austin Health. Approval was obtained from the human research ethics committees of Melbourne Health, the Peter MacCallum Cancer Centre and Austin Health. All medical oncologists and surgeons who participated in the MDMs were eligible to participate. Eligible patients included women with HR+ and HER2− early breast cancer and 0–3 positive nodes. Sequential eligible patients were offered enrollment.

After surgery and a pathology evaluation performed in the pathology department of the respective institutions, each patient was reviewed in the MDM and a treatment recommendation was made. Oestrogen receptor (ER), progesterone receptor (PR) and HER2 status was routinely assessed in all participating institutions, but testing for antigen Ki-67 levels was not a routine part of pre-assay assessments. The MDMs were typically attended by surgeons, medical oncologists, radiation oncologists, breast care nurses, radiologists and pathologists. Attendance varied little throughout the course of the study. Consenting patients discussed their treatment recommendation with the treating physician at the first postoperative visit. The Oncotype DX assay was ordered and a tumour sample was sent to the Genomic Health laboratory for analysis. After the assay result was received, the case was discussed in the MDM and a second treatment recommendation was made. The final treatment recommendation was then discussed with the patient.

The primary aim of the study was to characterise the impact of the Oncotype DX assay on the MDM treatment recommendation, as measured by a change in the recommendation. This was assessed separately in the node-negative and the node-positive group. Other objectives included determining the actual treatment decision of the patient.

Statistical analyses

The study was designed to enrol at least 80 patients with node-negative tumours and at least 50 with node-positive tumours. The proportions of treatment recommendations that changed from before to after the assay were calculated. The Fisher exact test was used to compare the difference in the proportions of patients whose recommendation moved from CHT to HT versus those who moved from HT to CHT. The McNemar test was used to assess the change in the proportion of patients who received a recommendation for CHT after the Oncotype DX assay. Exact P values are presented. The statistical software used was SAS version 9.2 (SAS Institute).

Results

Patient and tumour characteristics

One hundred and sixty-one patients were enrolled between 1 November 2010 and 30 September 2011. Of these, 10 were determined to be ineligible: three had HER2−positive tumours, two had multifocal cancer, one had HR−negative disease, and four patients’ samples were unsuitable for Oncotype DX testing. Therefore, 151 patients were eligible: 101 (67%) had node-negative tumours and 50 (33%) had node-positive tumours. Patient and tumour details including recurrence scores are shown in Box 1.

Recurrence score impact on MDM recommendations

For the 101 patients in the node-negative group, the initial recommendations were for HT alone in 71 patients and for CHT in 30 patients (Box 2). Following receipt of the recurrence score information, there was an overall change rate of 24% (24/101), consisting of changes from CHT to HT in 12 patients and from HT to CHT in another 12 patients. Therefore, the overall proportion of patients with node-negative tumours who received a final recommendation for CHT was unchanged from pre-assay to post-assay (P = 1, McNemar test). However, within the pre-assay recommendation groups, 12 of the 30 CHT patients (40%) had their recommendation changed to HT, while only 12 of the 71 HT patients (17%) had their recommendations changed to CHT. The difference in these rates of change (23%) is statistically significant (P = 0.02, Fisher exact test).

A greater proportion of the 50 patients with node-positive tumours received a pre-assay recommendation for CHT (37 patients) versus HT (13 patients) (Box 2). Thirteen patients in the node-positive group (26%) had their treatment recommendation changed after receiving a recurrence score: from CHT to HT in 12 patients (24%) and from HT to CHT in one patient (2%). There was a significant decrease in the proportion of patients with node-positive tumours who received a recommendation for CHT from before the assay (37/50; 74%) to after the assay (26/50; 52%) (P = 0.003, McNemar test).

Details of patients with a changed recommendation are shown in Box 3 and Box 4. The change in treatment recommendation after the assay was consistent with the recurrence score category, and most of those whose recommendations were changed were in the low or high recurrence score category or close to the cut-off between categories.

Of the 12 patients with node-negative tumours whose recommendations changed from CHT to HT, eight had a low recurrence score and four an intermediate result. No patients with a high recurrence score were switched from CHT to HT. Of the 12 patients with node-negative tumours and whose recommendations were changed from HT to CHT, six had an intermediate recurrence score and six had a high score.

Similar results were seen in the node-positive group. The patient whose recommendation changed from HT to CHT had a high recurrence score (37). Of the 12 patients with node-positive tumours whose recommendations were changed from CHT to HT, nine had a low score and three had an intermediate score.

Patient decisions

Fifteen patients chose not to follow the recommendation of the MDM. In the node-negative group, 10/101 patients decided against the post-assay treatment recommendation: four opted for CHT instead of HT, four selected HT instead of CHT, and two refused all therapy. In the node-positive group, 5/50 opted for a different therapy than the post-assay recommendation: two selected CHT instead of HT, and three selected HT instead of CHT.

Discussion

This first decision-impact study of the Oncotype DX breast cancer assay in Australia provides evidence that recurrence score information influenced MDM treatment recommendations for both the node-negative and node-positive groups.

In patients with node-negative tumours, those initially receiving a CHT recommendation were more likely to change to HT than the reverse (change rates of 40% v 17%; P = 0.02), as reported by others.12–15 However, unlike in other studies, a high proportion (more than two-thirds) of patients initially received a recommendation of HT alone and hence there was no overall change in chemotherapy use (12 patients changed from HT to CHT and 12 from CHT to HT). The high proportion of patients with node-negative tumours and with an initial recommendation of HT in this study appears to reflect an approach among this group of clinicians of basing treatments primarily on the degree of hormone sensitivity of the tumour, with the understanding that these patients would gain little extra benefit from chemotherapy. A number of these patients had unexpectedly high recurrence score results, which suggests that such an approach may sometimes lead to undertreatment.

In patients with node-positive tumours, those initially receiving a CHT recommendation were also more likely to change to HT than the reverse (change rates of 32% v 8%), although the difference in rates was not significant (P = 0.14). In contrast to the node-negative group, almost three-quarters of the patients with node-positive tumours received an initial recommendation for CHT, and there was a statistically significant reduction in recommendations for CHT post-assay (from 74% to 52%, P = 0.003). The reduction in CHT recommendations in the node-positive group is consistent with results seen in other studies of the impact of Oncotype DX assay on adjuvant decision making in patients with node-positive tumours.13,16 This suggests that in terms of reducing possible overtreatment, the Oncotype DX assay might have its greatest impact in this group.

While confirmation of the impact of the recurrence score on chemotherapy benefit in RCTs is pending (eg, TAILORx [NCT00310180], RxPONDER [NCT01272037]), our data suggest that use of the assay can spare patients potentially unnecessary treatment as well as identify patients for whom potentially lifesaving therapy might otherwise be omitted. In both patients with node-negative and those with node-positive tumours, changes in recommendations were made in directions consistent with the recurrence score categories, suggesting that assay results had a strong impact on final decisions. As the recurrence score has been validated to predict both distant recurrence risks and responsiveness to chemotherapy in both the node-negative and node-positive groups, these shifts in treatment recommendations arguably allow physicians to direct individual patients toward therapies that ultimately may be associated with better health outcomes. The result is more precise individualisation of therapy from both the MDM and the treating physician.

In both the node-negative and node-positive groups, about 10% of patients elected not to follow the post-assay treatment recommendation. Patient treatment decisions went against the MDM recommendations, both toward CHT and toward HT. Reasons for not following treatment recommendations might include patient preference, patient anxiety about one treatment regimen or the other, and varying levels of patient trust in the Oncotype DX recurrence score assay.

The lack of routine assessment of Ki-67 levels may be seen as a limitation of this study. A report in 2011 suggested that an assessment of Ki-67 levels, along with ER, PR and HER2 status, could provide similar prognostic information to the Oncotype DX assay.19 However, a subsequent international working group concluded that large variations in analytic methods have limited the routine reliance on immunohistochemical assessment of Ki-67 in clinical practice.20

Another limitation of our study is that MDMs at only three academic breast cancer units in Melbourne participated. Therefore, our results (which reflect the collective opinions of participating physicians) may not reflect nationwide practices. Attendance at the MDMs may have varied, and other factors such as the passage of time could have affected the second MDM recommendation, although the time between assessments was generally about 3 weeks. There may have been other biases.

In conclusion, this study provides evidence that the Oncotype DX assay influences treatment recommendations for HR+ early-stage invasive breast cancer in the Australian MDM setting. Most of the recommended changes were toward lower intensity treatment regimens. The health economic impact of this assay in Australia remains to be investigated, but the impact of this shift on quality of life should not be underestimated.

1 Patient and tumour characteristics of 151 women with early breast cancer

		No. of positive nodes (% of patients)*
Characteristics	Patients (%)	0	1–3

All	151 (100%)	101/151 (67%†)	50/151 (33%†)
Mean age (years)	56.2	56.6	55.3
Tumour size
< 2 cm	103 (68%)	71 (70%)	32 (64%)
≥ 2 cm	48 (32%)	30 (30%)	18 (36%)
Tumour grade
1	29 (19%)	22 (22%)	7 (14%)
2	84 (56%)	51 (50%)	33 (66%)
3	38 (25%)	28 (28%)	10 (20%)
Hormone receptor status
ER-positive, PR-positive	136 (90%)	90 (89%)	46 (92%)
ER-positive, PR-negative	14 (9%)	11 (11%)	3 (6%)
ER-negative, PR-positive	1 (1%)	0	1 (2%)
Recurrence score
Low (< 18)	72 (48%)	47 (47%)	25 (50%)
Intermediate (18–30)	59 (39%)	40 (40%)	19 (38%)
High (≥ 31)	20 (13%)	14 (14%)	6 (12%)

ER = oestrogen receptor. PR = progesterone receptor. * Percentage of patients in each column except where otherwise indicated. † Percentage of total patients.

2 Treatment recommendations before and after Oncotype DX testing in 151 women with early
breast cancer

	Before Oncotype DX testing		After Oncotype DX testing		Change from pre-Oncotype DX recommendation*
Nodal status	Therapy recommendation	Patients	Patients recommended HT	Patients recommended CHT	Proportion (%) changed	Difference between proportions changed	P†

Node-negative	HT	71	59	12	12/71 (17%)	23%	0.020
	CHT	30	12	18	12/30 (40%)
	Total	101	71	30	24/101 (24%)
Node-positive	HT	13	12	1	1/13 (8%)	25%	0.141
	CHT	37	12	25	12/37 (32%)
	Total	50	24	26	13/50 (26%)

CHT = combination of chemotherapy and hormonal therapy. HT = hormonal therapy. * The proportion of patients with node-negative tumours who were recommended CHT was unchanged from before (30%) to after Oncotype DX testing (30%) (P = 1.00, McNemar test). The proportion of patients with node-positive tumours who were recommended CHT decreased from before (74%) to after testing (52%) (P = 0.003, McNemar test). † P from Fisher exact test for the difference in change rates between patients originally recommended CHT versus those originally recommended HT.

3 Patients with treatment recommendations changed from chemotherapy plus hormone therapy to hormone therapy alone

Node-negative				Node-positive
Patient age	Tumour size (mm)	Tumour grade	Recurrence score	Patient age	Tumour size (mm)	Tumour grade	Recurrence score	No. of positive nodes

37	47	2	2	56	25	2	10	1*
62	20	3	3	50	20	2	12	1
51	45	2	6	50	24	2	12	1*
56	25	2	11	49	21	2	13	1*
54	25	2	11	75	18	2	14	2
77	53	2	12	45	57	2	14	1*
40	15	3	17	47	24	2	14	1*
38	33	2	17	58	20	2	16	1
76	20	3	18	52	18	3	17	1*
62	15	2	20	63	15	2	19	1
59	20	3	20	46	19	1	22	1
57	25	2	21	54	41	2	23	1

* Micrometastasis.

4 Patients with treatment recommendations changed from hormone therapy to chemotherapy plus hormone therapy

Patient age	Tumour size (mm)	Tumour grade	Recurrence score
Node-negative

37	18	2	22
27	25	2	25
72	12	3	26
63	5	1	29
58	15	1	29
65	8	3	29
66	20	3	33
60	15	2	34
66	12	3	34
66	9	3	34
38	30	2	35
61	13	3	48
Node-positive (1 node)
68	20	2	37

Rethinking psychotropics in nursing homes

To the Editor: We thank Hilmer and Gnjidic for highlighting the excessive use of psychotropic medications in Australian nursing homes.1 We recently completed a national psychotropic audit of over 9000 nursing home medication reviews conducted during the 2011–12 financial year. Our results were alarming: even after excluding residents with schizophrenia or bipolar disorder, more than a quarter of residents reviewed (27%) were taking antipsychotics. Concerningly, 41% of all residents reviewed were prescribed benzodiazepines.

Such findings are not new. The problem of excessive psychotropic use in Australian nursing homes has been reported for nearly 20 years and has been the subject of both federal and state inquiries.

As possible solutions, Hilmer and Gnjidic suggest economic evaluations and investment in staff and in research to develop better management strategies.1 Yet, qualitative research has determined that a major reason for high prescribing lies in an overestimation of the efficacy of psychotropic medications and limited awareness of their adverse effects.2 Consequently, awareness-raising and education of health practitioners and residents’ relatives is key.

We would also like to emphasise that programs to reduce psychotropic use in Australian nursing homes have proven successful in controlled trials. Resident and staff education reduced benzodiazepine use by half in a South Australian nursing home.3 Likewise, through local audit, benchmarking and nurse education in 15 Tasmanian nursing homes, antipsychotic and benzodiazepine use was reduced significantly over a 6-month period.4 It is time to move beyond describing psychotropic usage patterns and evaluating solutions to implementing these successful intervention strategies on a wider scale.

Rethinking psychotropics in nursing homes

To the Editor: The article by Hilmer and Gnjidic1 raises issues about the management of behavioural and psychological symptoms of dementia (BPSD)2 in residential aged care facilities (RACFs).

It is important to remember the context in which RACF residents’ care is provided. The federally managed aged care system is politically and organisationally perceived as supported accommodation, which is geared toward supporting or substituting for residents’ performance of basic and instrumental activities of daily living. This system is outside the state-based health care system. Health care, effectively an optional extra in this context, is provided by general practitioners, who may have no prior knowledge of the resident before admission.

RACF residents in high-level care are too sick and disabled, many suffering from moderate or severe dementia and in the palliative phase of their illness, to be supported by the state health system in the community. BPSD are an everyday fact of life that cause distress to those suffering them and their fellow residents and can make it impossible for carers to provide essential care safely.

Yes, BPSD can be managed by non-pharmacological methods.1 This requires skilled assessment, which goes significantly beyond diagnosis, and skilled intervention, which is much more than just prescription, by carers of all categories. Geriatrician or psychogeriatrician involvement is an obvious and necessary starting point. So why does it not happen when geriatric and psychogeriatric services exist in most regions of the country?

Unlike people living in their own homes, RACF residents are dependent on medical services coming to them, and my experience suggests that GPs are only too happy to refer patients for consultation. Are RACF residents invisible because they can’t come to us, and we won’t go to them?

Rethinking psychotropics in nursing homes

In reply: We thank Westbury and Peterson, and Mykyta for highlighting some of the issues raised in our article.1

While we acknowledge the important contribution of providing education programs to existing staff, staff education alone is insufficient to address the complex issues affecting psychotropic drug use in nursing homes.1 Better access to appropriately trained staff in nursing homes, including general practitioners, geriatricians, psychogeriatricians, psychologists, pharmacists, nurses or nursing aids, is required to provide safer alternatives to psychotropics. As we discussed, provision of these services may require changes to our health care and funding models.1

Inappropriate use of psychotropics in nursing homes is a hazard that requires comprehensive risk management strategies. The Australian Work Health and Safety Regulations require risks to be managed according to a hierarchy of risk control, in which education is among the lowest controls and providing safer alternatives is among the highest.2 The hierarchy of risk control may be applied to psychotropic use in nursing homes as follows:

Eliminate the hazard altogether: this is not feasible for psychotropics, which have a limited therapeutic role for nursing home residents.
Substitute the hazard with a safer alternative: invest in more skilled staff to administer non-pharmacological therapies, and in research to develop newer, safer therapies.
Isolate the hazard from anyone who could be harmed: make policy changes that limit prescribing of psychotropics by indication, dose and duration, and promote withdrawal3 to reduce the risk–benefit ratio.
Reduce the risk through engineering controls: use electronic prescribing to reduce inappropriate psychotropic use.
Reduce the risk through administrative controls: educate and train existing staff.
Use of personal protective equipment: residents who are administered psychotropics should wear hip protectors.

A multifaceted approach to psychotropic use is required to optimise the safety of the vulnerable people living in nursing homes.