JAC Advance Access originally published online on February 5, 2007
Journal of Antimicrobial Chemotherapy 2007 59(3):537-543; doi:10.1093/jac/dkl511
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Antimicrobial practice |
Can mass media campaigns change antimicrobial prescribing? A regional evaluation study
1 Department of Health Sciences, University of York, York YO10 5DD, UK 2 Regional Drug and Therapeutics Centre, Wolfson Unit, Claremont Place, Newcastle upon Tyne NE2 4HH, UK
* Corresponding author. Tel: +44-191-4915713; Fax: +44-191-4915727; E-mail: mark.lambert{at}doctors.org.uk
Received 24 August 2006; returned 16 September 2006; revised 19 November 2006; accepted 20 November 2006
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionContributionsTransparency declarationsReferences |
|---|
Background: Antimicrobial drug resistance is a significant cause of avoidable morbidity and mortality. Inappropriate prescribing of antimicrobials is acknowledged as a key determinant of this phenomenon. Many approaches are advocated for reducing this inappropriate prescribing, including regulatory, professional and educational interventions. Mass media campaigns are often suggested as a useful tool in managing public expectations, but the evidence to support this is weak, as no controlled studies of such campaigns exist. Evaluating such campaigns is problematic, and uncontrolled observations are misleading. We report here the first controlled study of such an intervention in the use of antimicrobials.
Methods: Two sequential mass media campaigns, providing information on the appropriate use of antimicrobials, were conducted during early 2004 and 2005 in the North East of England. These messages were articulated in the campaign by the cartoon character ‘Moxy Malone’. The campaigns were supported by printed materials, and in parts of this area, with professional education and prescribing support. A retrospective controlled before–after study was conducted, examining the effects on observed prescribing of antimicrobials for the populations covered by these two cycles of mass media campaigns. These populations were controlled with matched populations in the North of England. The primary outcome examined was prescribing rates (items) for all microbial agents for these populations, corrected for population structure (STAR-PU). A repeated measures analysis of variance (ANOVA) was used to analyse factors that had a possible effect on the prescribing of antibacterial drugs. This was supported by a survey of primary care organizations (PCOs) of all interventions undertaken around antimicrobial use in the intervention and comparison populations.
Results: In this retrospective study, there was incomplete reporting of adjuvant interventions undertaken by the PCOs intervention and comparison areas, so isolating the intervention, and attributing cause and effect is difficult. In this pragmatic evaluation the campaign was found to significantly reduce the volume of antibacterial drugs during the winter months of the intervention years. There were 21.7 fewer items prescribed per 1000 population (P < 0.0005), for the intervention populations over these winter months, equivalent to a 5.8% absolute reduction in prescribing.
Conclusions: Mass media campaigns have a role in changing antimicrobial prescribing practice.
Keywords: prescribing practice , population interventions , seasonal differences , time series , primary health care
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionContributionsTransparency declarationsReferences |
|---|
Antimicrobials have an essential role in combating infection. But resistance to antimicrobials is in itself a significant cause of avoidable morbidity and mortality. It also threatens the delivery of many modern health care programmes, through the emergence of organisms that are difficult to eliminate. A worldwide strategy for minimizing this phenomenon, and its impact includes action to educate patients and prescribers, improvements in management strategies in hospitals, as well as more conservative agricultural and veterinary practices as well as better regulation of the manufacture and use of antimicrobials.1
Resistance to antimicrobials has been recognized since the 1950s, but inappropriate prescribing of antimicrobials in the community remains a significant concern in the UK.2 Although antimicrobial use appears to be declining,3 this trend appears to be explained, at least in part, by declining rates of respiratory infection in UK.4
So, despite the existence of these international strategies and national plans, local policy makers are still faced with difficult decisions about how best to influence prescribing practice. Research into decision-making in the use of antimicrobials in the community is far from conclusive, but does suggest a complex interaction between doctors and patients.5,6
Despite the absence of strong theoretical basis for action, several strategies for influencing prescribing in infective conditions have at least some supportive evidence for their use. These include targeting information about the effective use of these drugs to parents and professionals7–9 and use of delayed prescriptions given in the course of consultations, which can be presented for dispensing at a later date.10
The mass media are often advocated as a vehicle for communicating with potential patients, informing them about the beneficial and adverse effects and managing expectations about prescribing. But to date the evidence for this has been circumstantial, using general reviews of effectiveness of such campaigns.
The Cochrane Review Group examining the effects of health service utilization gives some grounds for supposing that the use of mass media can promote the appropriate use of health care interventions including some examples of safer prescribing practice.11 Applying the results of this review to decisions about whether and how to use the mass media specific clinical scenario remains problematic. Moreover, direct evidence for their use in changing the use of antimicrobials is lacking.
Since this review, one uncontrolled study has indicated positive effect from a media campaign12 on antimicrobial prescribing. However, without an appropriate control, it is impossible to eliminate the potential for confounding by secular trends in prescribing practice.
Despite this relative lack of evidence, a regional intervention was undertaken, based on use of mass media, in conjunction with educational and supportive interventions. Initial reports suggested dramatic changes in antimicrobial prescribing following the intervention. This retrospective study was undertaken to establish the accuracy of these initial reports, and in doing so we describe the first controlled study of this type of intervention in changing use of antimicrobials.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionContributionsTransparency declarationsReferences |
|---|
Intervention
The intervention we have studied is the application of a regional mass media campaign used in two consecutive years on a single geographical population in the North East of England. The campaign was sponsored by the North East PCTs under the auspices of the Tyne and Wear Health Action Zone. It was led by a cartoon character called ‘Moxy Malone’ (stylized as a ‘private eye’), who was developed specifically for this regional campaign. The character, developed in conjunction with a former Viz cartoonist, was given dialogue that included simple messages ‘facts’ and ‘fictions’ about the effects of antimicrobials, and about the availability of community pharmacists to support self-care for the management of usually self-limiting health problems. These messages were drawn together under the title of ‘Antibiotics- tracking down the truth’, which appeared as a short cartoon strip, with supporting leaflets and posters. In addition, the campaign received editorial coverage from local newspapers, TV and radio.
The campaign was run over two successive winters 2004 and 2005 (in the months of January and February). In both years, advertisements were run on local radio (Metro Radio, main audience aged 25–40), with posters on buses and local Metro system. Printed materials (leaflets and posters) were made available to GP surgeries throughout the target area. In the second phase (early 2005), regional television advertising was added, with the same artwork and character. The timing of the campaigns was chosen to coincide with the annual peak in consultations for respiratory infections.
The Moxy Malone campaign was not the only intervention being used to tackle inappropriate antimicrobial prescribing at this time. To assess the significance of alternate interventions, primary care organizations (PCOs) who commissioned the mass media intervention and whose populations were exposed to it (intervention PCOs) were surveyed to establish other supportive interventions which were being actively promoted. A short questionnaire was sent to each PCO requesting details of interventions designed to reduce the inappropriate use of antibiotics. The questionnaire listed ten commonly used strategies: incentive schemes, practice visits, professional education events, public education events, use of printed materials (not related to the Moxy Malone campaign), minor ailment schemes, delayed prescription schemes, guideline promotion, and use of electronic reminders or of opinion leaders. It also asked for details of any other strategies not listed but employed during the time periods in question.
The initial (2004) programme was focused on North Tyneside, Newcastle, Gateshead, Sunderland and South Tyneside populations. This included radio coverage, posters and leaflets in GP surgeries and posters on public transportation systems. The Northumberland population was also covered by the radio campaign, but did not receive any of the supportive elements of poster or leaflet distribution.
In early 2005, when the campaign was run for a second time, the programme included television advertising and the wider populations of Northumberland and County Durham & Tees Valley. A total of sixteen PCOs took up the initiative in this phase, combining it with the supportive elements described above.
The intervention population examined here are those living in Gateshead, South Tyneside, Sunderland, North Tyneside and Northumberland, all of whom received the intervention in two successive years.
The populations of all PCOs in the North of England (the former Directorate of Health and Social Care North) were grouped according to the Office for National Statistics (ONS) clusters.13 All PCOs in the intervention group fell into ‘regional cities A’, ‘prospering smaller towns B’, ‘industrial hinterlands A’ or ‘industrial hinterlands B’. All other PCOs in the North of England which fell into these ONS clusters, and were not exposed to either phase of the intervention, were used as control populations. These clusters were introduced to allow the model to adjust for excess variability caused by differences in socio-economic and demographic characteristics. To avoid contamination of the control population with populations exposed to a partial intervention (i.e. a single cycle), PCO populations from County Durham and Tees Valley, which were exposed to only one cycle of the intervention, are excluded from both the intervention and control arms of the analysis. Control PCOs were also surveyed to provide information on interventions used to reduce inappropriate antimicrobial prescribing.
The primary outcome of the study is the weighted volume of medications prescribed in section 5.1 of the British National Formulary (all antimicrobial drugs). The volumes prescribed were derived from the Prescription Pricing Authority database, which includes the items dispensed from all primary care prescriptions. Prescribing rates were corrected for population structure, and are expressed in prescriptions (items) per 1000 STAR-PU (Specific Therapeutic group Age-sex Related Prescribing Units). This weighting system corrects the list size of an individual practice for the age and sex of patients for whom drugs in specific therapeutic groups are usually prescribed.14 Data were collected for the months July 2002 to August 2005 inclusive.
A repeated measures analysis of variance (ANOVA) model was used to analyse the prescribing frequencies within the intervention and control groups (a ‘group’ effect). The possibility that the trends within these groups may vary differently over time was also considered. The nuisance factor ‘ONS cluster’ was also examined, as was the consideration that these clusters may react differently over time. There is a strong seasonality component inherent in the prescribing of antibiotics. Fitting ‘time’ as a factor (with each month representing one level) allowed the model to adjust for the seasonal variation. However, this method will not allow for the direct comparison of separate campaigns since confounding exists between campaign one (during 2004) and campaign two (during 2005).
The difference in prescribing volume between the intervention and control groups during July 2002 to August 2005 was analysed using a repeated measures analysis of variance model.
|
|
A P value of 0.05 or lower was considered statistically significant.
Repeated measures ANOVA analyses were done with the SPSS software package (version 12.0).
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionContributionsTransparency declarationsReferences |
|---|
In response to the survey of alternate interventions, after two reminders, four PCOs in the intervention group (67%) and eight (47%) in the control group responded. The extent to which other adjuvant interventions (directed at the appropriate use of antimicrobials) were being applied at the same time is described in Tables 1 (intervention populations) and 2 (comparison populations).
|
|
Repeated monthly tests for differences between the control and intervention groups prior to the initial campaign identified four out of the total 18 months where there was a statistically significant difference between them. Specifically there were higher prescribing rates in the intervention group during September 2002 (5.0% higher than control, P = 0.03), May 2003 (5.7% higher than control, P = 0.008), June 2003 (4.0% higher than control, P = 0.02), and lower in November 2003 (6.7% lower than control, P = 0.004). This implies the two groups were not significantly different 78% of the time (14 out of 18 months).
Other than this baseline prescribing position, there are no direct measures to test the comparability of the populations. The sizes of PCT populations were generally larger in the intervention groups (mean 230 000, range 152 000–308 000) than in the comparison populations (mean 135 000, range 93 000–216 000). However mortality was comparable in the intervention populations (mean SMR 113 range 104–122) and in the comparison populations (mean SMR 111 range 91–134).
The final repeated measures ANOVA model included the significant factors ‘ONS cluster’ (P = 0.001), ‘time’ (P < 0.0005), ‘ONS cluster’ by ‘time’ interaction (P < 0.0005) and ‘group’ by ‘time’ interaction (P < 0.0005). The model showed that there is not a direct difference between the control and intervention groups (a ‘main effect’) inferring that, when averaged over the whole period, these two groups have similar prescribing rates. However this effect is implicitly included in the model since the interaction between ‘time’ and ‘group’ is significant (P < 0.0005). Figure 1 illustrates how the campaign may affect the prescribing over time using the estimated means of the ‘time x group’ interaction. Comparable patterns were observed after fitting the model (Figure 1) and in the raw data (Figure 2). However, the model did suggest a clearer difference between intervention and control populations.
|
|
There is significantly reduced prescribing in the intervention group during the winter months. The estimated means and P values for winter months 2004–05 are included in Table 3.
|
Performing repeated tests increases the risk of type I error however the run of significantly reduced prescribing (November 2004–March 2005, excluding December 2004) is an indication that the two campaigns may be having an effect during the winter months.
The model predicts a reduction of 21.7 items per 1000 STAR-PU during the winter months (November–March) in a PCO subjected to two years of the Moxy Malone campaign compared to the control group PCO. This is equivalent to a 5.8% reduction in prescribing (over these winter months only), over and above any background trend (see Table 3).
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionContributionsTransparency declarationsReferences |
|---|
This study has found that antimicrobial prescribing during a mass media campaign directed towards population beliefs about use of antimicrobials was significantly reduced. The observed effect is however, more measured than the preliminary uncontrolled reports of the Moxy Malone programme. It provides a stronger weight of evidence than any other study of this type of intervention for this clinical problem.
Whilst a 5% change in any common clinical behaviour is significant, the change is small on an international scale as there is an almost four fold difference in prescribing rates of antimicrobials across Europe.15 Understanding the effectiveness of policy interventions in changing clinical practice is important, but the nature of the evidence-base is relatively weak for many interventions16 and few high-quality evaluation studies have looked at the effects of mass media interventions on clinical practice.11 The gold standard for evaluating health care interventions is the randomized controlled trial. However, for interventions that rely to a large extent on mass media interventions it is difficult to identify and isolate appropriate units of intervention. The alternative is some form of quasi-experimental design,17 all of which have important limitations and threats to validity.
With most non-randomized designs, ensuring equivalence of the groups is key to determining the validity of any findings.18 This equivalence relates both to the characteristics of the populations, and to other factors (such as alternate interventions) which may differentially impinge on either population during the study period.
Retrospectively identifying appropriate comparison populations, and eliminating the potential for contamination is difficult. There are no reliable measures of incidence of infectious disease for which antimicrobials might be used. We therefore selected populations by clusters that would reflect the wider social and economic circumstances as well as health determinants to select control populations, specifically the clusters used by the Office for National Statistics to group populations for a range of statistical comparisons. In doing so, we have attempted to take into account population differences that could influence patterns of illness behaviour and health care utilization. Our selection processes have therefore attempted to ensure the comparability of the populations at baseline. From the selected indicators included for illustrative purposes, it is possible to assert that the populations are different in some respects; the populations being drawn from different geographical areas, with potentially different consulting behaviours, (the comparison populations being drawn principally from the North West of England, and from North Yorkshire) with many smaller PCOs, but with comparable standardized mortality ratios in the comparison group. However, we believe these will be balanced by the range of indicators used to assemble the ONS clusters (Table 4). Significant baseline differences in antimicrobial prescribing were not however observed between the intervention and comparison populations. We were unable to control, or even comprehensively document the complementary interventions that may otherwise explain any observed differences in antimicrobial prescribing before and after the intervention. It is difficult to establish from our survey whether the PCOs were balanced in the intensity of other interventions undertaken at the same time. Of all the possible reasons for non-response, the one most likely to affect the validity of these findings is the possibility that it reflects a lower priority and lower intensity of intervention during the study periods. Whilst we cannot assume equivalence of intensity of adjuvant or complementary interventions in the intervention and control group, merely counting the number of interventions used is unlikely to be a good predictor of the likely effects.19
|
We did not isolate the effects of the mass media campaign from all other interventions that were taking place alongside it. What is being tested here is, in effect, the effects of a choice to implement a mass media campaign directed towards the general population. It does not throw any light on the mechanistic question of what the most effective combination of factors in any intervention package might be.
In this evaluation, the key factors in choosing an analytic approach include the background declining trends in antimicrobial prescribing,4 which makes uncontrolled intervention studies more likely to give a type I error. In addition, there is both significant background variation in levels of antimicrobial prescribing for populations of this size, and seasonal patterns in antimicrobial prescribing, which may mask any real effect of the intervention.
The outcome measure used is however free from any systematic bias, as these prescribing data are routinely collected, against longstanding and consistent data protocols, which are entirely independent from the intervention. One approach which was considered at an early stage was to use defined daily doses (DDDs) rather than items prescribed. DDDs convert the items prescribed into a numbers of standard daily regimens for individual drugs. However, not all antimicrobials have a recognized DDD, and studies that have used it have introduced other measures of daily dose.15 Items are also preferred as a measure in the case of antimicrobials as there is a further interaction with the use of (frequently) short courses of these drugs.
The mechanism for the observed change in prescribing practices has not been explored here. It is possible that patients consulted less often in intervention populations, or their expectations in the consultation were changed. It is also possible that some who might have benefited from antimicrobials were discouraged from consulting, or from using any prescription they received for antibiotics.
The analysis used here demonstrates the effect on prescribing during the second winter period, following two years of the campaign. We cannot be sure whether the drop in year two is largely due to the continued effect of year one alone (the effect was noticed to start in November), or due to a cumulative effect of the second campaign. Neither does it predict whether there are enduring effects beyond the intervention period. It will be difficult to establish a clinically significant effect beyond this period, given the ‘noise’ from the background variation and downward trend in prescribing patterns. This study has not examined the cost-effectiveness of the campaign. The direct costs of the campaign were £25 000 in each campaign year. The direct benefits of the programme in diminished antimicrobial prescribing, examined here, are important. But a more rounded view of the programme would only consider this as an intermediate outcome, in that the potential effects on diminishing use of primary care consultations and minimizing the additional contribution towards burden of antimicrobial resistance are more appropriate end-points.
Of these, prescribing costs are perhaps the easiest to document, but have the smallest financial contribution to make. The majority of prescribing avoided will be of the generic preparations more commonly used for respiratory tract infections. We have not therefore documented these here. There remain significant limitations to this retrospective study. It will always be difficult to undertake randomized studies of this type of intervention; however, greater certainty could be achieved through routine prospective design of campaigns of this kind, allowing prospective selection and data collection on population and provider characteristics as well as the range of other interventions undertaken. More research is needed to determine the most efficient strategies for informing specific populations, and changing their expectations of antimicrobial use. In addition, more research is needed on optimizing the strategies for complementing the use of mass media with support for professionals. Finally, more detailed economic studies are required to determine more clearly the cost-effectiveness of using mass media in changing the use of antimicrobials.
|
Contributions |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionContributionsTransparency declarationsReferences |
|---|
M. L. had the idea for the study. M. L., S. B. and G. M. designed the study. S. B. was responsible for collecting new data for the study. M. L. and G. M. agreed the analysis plan, which was undertaken by G. M. All authors contributed to the writing of this paper. M. L. takes overall responsibility for the work.
|
Transparency declarations |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionContributionsTransparency declarationsReferences |
|---|
M. L. is also a director of Gateshead NHS Primary Care Trust, one of the NHS organizations that joint-funded the intervention.
|
Acknowledgements |
|---|
We would like to thank Anita Chalmers and Elaine Wilson for their accounts of the interventions, Professor J. N. S. Matthews for advice on statistical analysis and Robbie Foy and Ian Watt for commenting on an earlier draft of this paper, and to the anonymous reviewers whose comments have been used in the revision of this manuscript. No funding was received for this evaluation study.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionContributionsTransparency declarationsReferences |
|---|
1 WHO. (2001) Global Strategy for Containment of Antimicrobial Resistance(World Health Organization, Geneva).
2 Chief Medical Officer. (2003) Winning Ways: Working together to reduce Healthcare Associated Infection in England(Department of Health, London).
3 Unsworth L and Whalley T. (2001) Trends in primary care antibiotic prescribing in England 1994–1998. Pharmacoepidemiology and Drug Safety 10::309–314.[Medline]
4 Fleming DM, Ross AM, Cross KW, et al. (2003) The reducing incidence of respiratory tract infection and its relation to antibiotic prescribing. Br J General Practice 53:778–83.
5
Macfarlane J, Holmes W, Macfarlane R, et al. (1997) Influence of patients' expectations on antibiotic management of acute lower respiratory tract illness in general practice: questionnaire study. BMJ 315:1211–4.
6
Britten N and Ukoumunne O. (1997) The influence of patients' hopes of receiving a prescription on doctors' perceptions and the decision to prescribe: a questionnaire survey. BMJ 315:1506–10.
7
Finkelstein JA, Davis RL, Dowell SF, et al. (2001) Reducing antibiotic use in children: a randomized trial in 12 practices. Pediatrics 108:1–7.
8 Smaabrekke L, Berild D, Giaever A, et al. (2002) Educational intervention for parents healthcare providers leads to reduced antibiotic use in acute otitis media. Scand J Infect Dis 34:6579.
9
Gonzales R, Steiner JF, Lum A, et al. (1999) Decreasing Antibiotic Use in Ambulatory Practice: Impact of a Multidimensional Intervention on the Treatment of Uncomplicated Acute Bronchitis in Adults. JAMA 281:1512–9.
10 Arroll B, Kenealy T, Kerse N. (2003) Do delayed prescriptions reduce antibiotic use in respiratory tract infections? A systematic review. Br J General Practice 53:871–7.
11 Grilli R, Ramsay C, Minozzi S. (2002) Mass media interventions: effects on health services utilisation. Cochrane Database of Syst Rev.
12 Dollman W, Leblanc V, Stevens L, et al. (2005) A community-based intervention to reduce antibiotic use for upper respiratory tract infections in regional South Australia. Med J Austr 82:617–20.
13 Health Area Clusters www.statistics.gov.uk/about/methodology_by_theme/area_classification/datasets.asp. (last accessed 10 October 2006).
14
Lloyd D, Harris C, Roberts D. (1995) Specific therapeutic group age-sex related prescribing units (STAR-PUs): weightings for analysing general practices' prescribing in England. BMJ 311:991–4.
15 Cars O, Molstad S, Melander A. (2001) Variation in antibiotic use in the European Union. Lancet 357:1851–3.[CrossRef][ISI][Medline]
16 Grimshaw J and Eccles M. (2004) Implementing clinical guidelines: current evidence and future implications. J Cont Edu Health Professions 24:S31–37.[CrossRef]
17 Cook T and Campbell D. (1979) Quasi-Experimentation: Design and Analysis Issues for Field Settings.(Houghton Mifflin, Boston).
18 Ramsay C. (2003) Designs in health technology assessment: lessons from two systematic reviews of behavior change strategies. Int J Techn Assess Health Care 19:613–23.
19 Grimshaw J, Thomas R, MacLennan G, et al. (2004) Effectiveness and efficiency of guideline dissemination and implementation strategies. Health Technology Assessment 8:6.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. B. Planta The Role of Poverty in Antimicrobial Resistance J Am Board Fam Med, November 1, 2007; 20(6): 533 - 539. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Coenen, M. Costers, and H. Goossens Comment on: Can mass media campaigns change antimicrobial prescribing? A regional evaluation study J. Antimicrob. Chemother., July 1, 2007; 60(1): 179 - 180. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||