Impact of adverse news media on prescriptions for osteoporosis: effect on fractures and mortality

The media play an increasingly important role in disseminating medical information to the general public, predominantly via television and newspapers. While it is important to disseminate knowledge to the public, the media often seem to be more intent on increasing their audience and ratings by issuing alarmist reports than informing the public of new scientific data. For example, in Canada, a public health advisory statement about the use of antidepressants in pregnancy, with subsequent coverage in the media, was widely misunderstood and led to pregnant women making decisions that may not have been in the best interests of their health and their child’s health.1 Similarly, media coverage of the findings of the Women’s Health Initiative2 have had a major effect on patient perceptions about the use of hormone replacement therapy3 and dietary modification4 that were not necessarily evidence-based.

In December 2007, the public broadcaster, the Australian Broadcasting Corporation (ABC), screened an item on its news and current affairs program, The 7.30 Report, about the association between osteonecrosis of the jaw and bisphosphonate use. The program received wide coverage in other media, including newspapers and the internet. A number of factual errors in the program were the subject of a complaint to the ABC, and these errors were subsequently acknowledged in writing by the broadcaster.5 However, the immediate effect of the media coverage was likely to have been considerable misunderstanding among the public, as the program failed to clearly identify the different uses of bisphosphonates in the oncology and osteoporosis settings or the different absolute risks of osteonecrosis in the individual patient.

After the television program aired, there was a reduction in the number of prescriptions filled for bisphosphonates, according to statistics provided by the Drug Utilisation Sub-Committee (DUSC) of the Pharmaceutical Benefits Advisory Committee of the Department of Health and Ageing (Box 1). The graph shows that the increase in antiresorptive drug use over the period 2001–2008 was not linear. The number of Pharmaceutical Benefits Scheme (PBS) and Repatriation Pharmaceutical Benefits Scheme (RPBS) scripts usually declines in the first quarter of each year (due to PBS Safety Net provisions, which subsidise all prescription costs after patients’ costs exceed a specific threshold in a calendar year) and then increases steadily over the remaining three quarters, but the pattern was different in 2008. Before 2008, the number of RPBS scripts in each quarter was always larger than the number in the corresponding quarter in the previous year. It can also be seen that the decline in the number of scripts in the first quarter of 2008 was greater than in the first quarter of previous years and that the growth after the first quarter of 2008 was slower than that in previous years.

We modelled the effect of this decrease in bisphosphonate prescriptions to estimate the impact on fractures and associated mortality in the Australian population.

The DUSC provided quarterly numbers of RPBS scripts filled for antiresorptive drugs for the period 2001–2008, as well as information on bone densitometry services (item numbers 12306, 12309, 12321 and 12323) for the years 2006–2008. These statistics included only claims processed by Medicare Australia, not services provided to public patients in hospital.

We developed two polynomial regression models of the quarterly number of RPBS scripts. Both included continuous covariates for year (Year 1 to Year 7 or 8) and quarter (Quarter 1 to 4), and we tested squared terms in year and quarter, as well as all two-way interaction terms.

For Model 1, we used the data for 2001–2007 and then used this model to predict the number of scripts in the first three quarters of 2008. The average difference between the predicted numbers based on the prior 7 years and the recorded numbers for the first three quarters of 2008 was considered as the effect size of the media coverage.

For Model 2, we used data for the whole period of 2001–2008 and included an extra covariate (0 before the broadcast and 1 after the broadcast) for the effect of the media coverage on the use of antiresorptive drugs.

To be conservative, we selected the model that gave the smaller estimate of the effect of the media coverage. Because each patient usually has three scripts per quarter, we estimated the number of people affected by the media coverage by dividing the selected value by three.

The age and sex distribution of those not obtaining bisphosphonates after the media coverage was assumed to be the same as that for patients taking antiresorptive drugs between April 2007 and March 2008. We estimated fracture incidence rates for osteoporotic fractures and hip fractures by age and sex using the fracture risk calculator developed by the Garvan Institute in Sydney.6 The number of hip fractures and total fractures expected in each group was estimated by multiplying the incidence rate by the number of people in the group. To calculate the number of fractures that may have resulted from the adverse media coverage, we assumed, based on previous studies,7-9 that bisphosphonates produce a 40% reduction in hip fractures and a 25% reduction in other fractures. In order to estimate the additional number of deaths, we further assumed that 20% of people who had a hip fracture would die within the first 6 months after the fracture.10

The results from the two regression models are presented in Box 2. The estimated reduction in the number of prescriptions using Model 1 was 29 633, compared with 31 697 using Model 2. Based on the value of 29 633, the number of men and women affected by the media coverage was estimated to be 9878, representing 3.1% of the average number of people (316 304) receiving antiresorptive treatment in 2007. It is worth noting that this effect persisted to the end of the study (ie, the September quarter of 2008) (Box 1).

The number of hip fractures and total fractures expected in each age–sex group and the estimated number of fractures and deaths that were not prevented as a result of the media coverage are shown in Box 3. In total, 415 osteoporotic fractures, including 174 hip fractures, were expected to occur each year among these 9878 men and women. We estimated that 70 hip fractures, 60 other fractures and 14 deaths would have been prevented if the 9878 people affected by the media coverage had taken antiresorptive drug treatment (Box 3).

Our findings suggest that a television program and ensuing media coverage highlighting the association between bisphosphonate use and an uncommon adverse event, osteonecrosis of the jaw, potentially resulted in 130 fractures and 14 deaths that may otherwise have been prevented.

Although it is important for patients to be informed of the risks of medication, unbalanced media coverage has the potential to do more harm than good.1,4 Indeed, it has recently been reported that the predicted rising trend in hip fractures in Australian women has been reversed, coincident with increased prescribing of bisphosphonates.11 When antiresorptive drugs are used to treat osteoporosis, the very small risk of jaw osteonecrosis must be kept in perspective and balanced against their demonstrated benefit in reducing fractures and death.

The incidence of osteonecrosis of the jaw after oral bisphosphonate treatment for osteoporosis has been estimated to be between 1 in 10 000 and 1 in 100 000 patient treatment-years,12 although a retrospective postal survey in Australia provided higher estimates of risk.13 In cancer patients, for whom bisphosphonates are primarily used as high-dose adjunctive therapy to reduce the number of skeletal-related events or to treat hypercalcaemia, the reported incidence may be as high as 1 in 1000, increasing to 1 in 10 after tooth extraction.12

Other factors may have affected the use of antiresorptive drugs in 2008. For example, in some patients, bisphosphonate therapy may be stopped after 5 years of the treatment if bone density readings are stable. However, this is not a likely explanation for the usage pattern observed, as the number of people undergoing bone density testing in 2008 was in fact lower than in previous years (according to data provided by the DUSC). Some physicians may also have followed guidelines suggesting that selected patients be referred to a dentist before starting bisphosphonate treatment. However, these guidelines should have affected the use of bisphosphonates throughout the period since their introduction in 2006.14 Thus, any effects should have been part of the growth trend from 2001 to 2007, which would have been included in the regression models. We used an average but conservative effect size for the reduction in fractures with different bisphosphonates. Although we modelled prescription usage after the television story aired, other sources of medical information, such as the internet, may also have contributed to patient perceptions about bisphosphonate use.

Responsible reporting of health data to the public is a challenge. Great skill is needed to strike the right balance between alerting but not alarming. Reporters and the general public alike have poor understanding of epidemiological principles and application of the risk to individuals. It behoves the medical community to conduct appropriate studies to gain unbiased assessment of risks, and it behoves media personnel to be more responsible in their reporting so that such alerts first do no harm.

1 Predicted number of prescriptions compared with recorded number of prescriptions for bisphosphonates in Australia, 2001 to 2008

The predicted model fits observed scripts closely until the first three quarters of 2008 (these three quarters are also shown in expanded view).