Healthy
Conditions
Keywords
Adherence, medication, Pandemic Influenza
Brief summary
During the last pandemic influenza antivirals were prescribed both as prophylaxis and treatment. However, adherence rates were suboptimal. This study assessed the effect of theory-based and evidence-based health messages, which promoted the use of antivirals as prophylaxis for pandemic influenza, on intentions to use antivirals. Using hypothetical scenarios, the investigators tested whether written health communications about pandemic flu and recommended preventative medication (i.e. a prophylactic treatment with antivirals) had an effect on study participants' beliefs about the pandemic flu and the advice received, and their intention to adhere to the recommendation. In particular, the investigators assessed the behavioural impact of health messages presented in four different linguistic formats, defined by a 2×2 (agency assignment × attribute framing) factorial design. The originality of this study relies on the attempt to maximise the behavioural impact of written health messages by combining the agency assignment and attribute framings, which have never been tested together, and by systematically targeting specific predictors of adherence intentions through these messages. The findings of this study may be used to improve the behavioural impact of health communications to the general public in case of a pandemic flu outbreak in the UK.
Detailed description
Background: In 2009, a new strain of influenza virus, called pandemic influenza A(H1N1)pdm09, was detected in humans. During the pandemic in 2009/2010, antiviral medications were prescribed both as prophylaxis and treatment for influenza. However, many people, including those with chronic health conditions, did not used the antivirals as prescribed. This study focused on adherence to antivirals used as prophylaxis, as this is a key preventative stage before any vaccine can be available to the public. To ensure protection of public health during public health emergencies, such as pandemic flu, it is essential that the general public follow the health advice received. Therefore, future risk communications about the use of antivirals for pandemic influenza will need to address the barriers and facilitators of adherence. This study investigated the effect of theory-based and evidence-based health messages, which promoted the use of antivirals as prophylaxis for pandemic influenza, on intentions to use antivirals. In particular, using hypothetical scenarios, the investigators tested whether written health communications about pandemic flu and recommended preventative medication (i.e. a prophylactic treatment with antivirals) had an effect on study participants' beliefs about the pandemic flu and the advice received, and their intention to adhere to the recommendation. The linguistic manipulation used in the present study was based on an adaptation of a successful manipulation carried out by McGlone et al. (2013), who found that using the virus agency assignment in medical factsheets increased students' perception of susceptibility to and worry about the virus A/H1N1, perception of the severity of the consequences of catching the virus, perception of efficacy of the vaccine and reported intention to get the vaccine compared to the human agency assignment. The investigators wanted to test whether these results could be replicated in the context of taking antivirals for pandemic flu, using a more representative sample of the general population and controlling for baseline intentions. For what concerns the framing of the side effects, research shows that the way information on medication side-effects is presented affects people's intentions to accept the recommended medications. The same side-effect likelihood can be expressed in a positively framed message (i.e. 90% of people will not develop side-effects) or in negatively framed message (i.e. 10% of people will develop side-effects). Although a previous meta-analysis (Moxey et al. 2003) found no significant framing effect on either actual immunization or behavioural intention, it found that positively framed information increased positive attitudes toward immunization and perception of the benefits of immunization, decreased expectation of side-effects and led to a more realistic understanding of the side-effects of influenza vaccines. In addition, the health messages used in this study targeted specific variables, which have been in the past associated with higher adherence to antivirals for pandemic influenza. In particular, a recently published systematic review (Smith et al. 2016) investigated the factors affecting adherence to antivirals for influenza, including pandemic flu. They included 26 studies, of which 23 assessed pandemic influenza. Although many of the reviewed studies had serious methodological flaws, the review identified several psychosocial factors associated with adherence to antivirals prescribed as prophylaxis for pandemic flu. In particular, 11 studies investigated actual adherence to prophylactic antivirals. These studies showed that actual adherence was associated with previous adherence to other preventative measures for pandemic flu, beliefs that the recommended prophylactic measures were necessary and having discussed taking oseltamivir, a commonly used antiviral, with someone who had not experienced side effects. Although only one study found that the presence of side-effects was a significant predictor of non-adherence, the occurrence of side-effects was the most commonly self-reported reason for the discontinuation of oseltamivir. The review also included 8 studies investigating psychosocial predictors of intention to adhere: for instance, higher perceived susceptibility to the pandemic flu, higher perceived severity of pandemic flu, knowledge of pandemic flu, higher self-efficacy and response efficacy, and having previously accepted the flu vaccine were all associated with higher adherence intentions. Methods: In May/June 2016 the investigated conducted an online experiment. After reading a hypothetical pandemic flu scenario, 349 adult UK residents were randomly allocated to one of four conditions, defined by a 2x2 (agency assignment × attribute framing of side effects) factorial design. Each condition presented messages describing the pandemic flu using linguistic expressions that assigned agency to either the humans (HA: human agency) or the virus itself (VA: virus agency), whilst describing the antivirals side effects in terms of the chances of either experiencing (NF: negative framing) or not experiencing side effects (PF: positive framing). Although presented with different framings the health messages provided equivalent information that targeted specific predictors of adherence. Participants were blinded to group assignment. Intentions to use the antivirals and potential mediating factors were measured. The aim was to understand the potential behavioural impact of the four health messages developed and to better understand the factors influencing adherence intentions.
Interventions
Linguistic framing used in written health messages. Each version of the messages described the pandemic flu and the efficacy of the antivirals using linguistic expressions that assigned transmission to either humans (HA: 'You can contract the virus when you touch…') or the pandemic flu virus itself (VA: 'It can infect you when you touch…')
Linguistic framing used in written health messages. Each message described the side effects of the antivirals in terms of chances of either experiencing (negative framing: 'Uncommon side effects (10% of people will be affected)') or not experiencing side effects (positive framing: 'Uncommon side effects (90% of people will not be affected)') after using them.
Sponsors
Study design
Masking description
Participants were blinded to group assignment.
Intervention model description
This study was a vignette-based randomised controlled trial, with a 2 (human vs virus agency assignment) x 2 (negative vs positive frame of antivirals side effects) between-participants factorial design. All participants were presented with a hypothetical pandemic flu scenario describing moderate health consequences (i.e. about 1 in 1000 infected people dying). The scenario: 1) in the initial scenario participants were told to imagine that their GP (although not confirming whether it was a pandemic case) had advised them to take antivirals as a precaution; 2) after measuring baseline intentions, participants were introduced to the second part of the scenario where they were asked to imagine being at their local pharmacy where a pharmacist provided them with some health messages and information about the antivirals. Participants were randomised to one of the four health messages described below. After reading the health messages their intentions to use antivirals were re-measured.
Eligibility
Inclusion criteria
* To be eligible to participate, respondents had to be members of the UK general public in the adult age range (18-65 years), and be fluent in English.
Exclusion criteria
* non-UK residents * younger than 18 or older than 65 * not fluent in English
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Change in Intentions to Take Antivirals for Pandemic Flu | Straight after exposure to the health messages | Mean adherence intentions post- exposure to the health information in the 4 groups. Intentions were measured by self-report items: participants were asked to agree with three statements about their intentions to take antivirals as recommended in the hypothetical scenario (on 9-point scale, where 1=strongly disagree to 9=strongly agree). The scores reported below represent a composite variable 'change in intentions', which reflects the average of the three items that compose it. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Worry of the Pandemic Flu Threat | At 20 minutes (i.e. straight after exposure to the health messages) | Measured by self-report items adapted from Witte et al. (2001): participants were asked to agree with two statements about their perceived worry on 9-point scale, from 1=strongly disagree to 9=strongly agree. Higher scores indicate higher levels of reported worry about pandemic flu threat. This is a composite variable reflecting the average of the items that compose it. |
| Perceived Susceptibility to the Pandemic Flu | At 20 minutes (i.e. straight after exposure to the health messages) | Measured by self-report items: participants were asked to state how likely they were to get sick with pandemic flu, had they not taken prophylactic medication o a 9-point scale, where 1=not likely at all, to 9=extremely likely. |
| Perceived Severity of the Pandemic | At 20 minutes (i.e. straight after exposure to the health messages) | Measured by one self-report item (adapted from Witte et al. (2001): participants were asked to agree with a statement about their perception of the severity of the pandemic flu on 9-point scale, from 1=strongly disagree to 9=strongly agree. Higher scores indicate higher levels of perceived severity of the pandemic flu threat. |
| Perceived Self-efficacy | At 20 minutes (i.e. straight after exposure to the health messages) | Measured by one self-report item (adapted from Witte et al. (2001): participants were asked to agree with a statement about their perceived ability to take the antivirals as recommended on 9-point scale, from 1=strongly disagree to 9=strongly agree. Higher scores indicate higher levels of reported self-efficacy. |
| Perceived Efficacy of the Antivirals | At 20 minutes (i.e. straight after exposure to the health messages) | Measured by two self-report items (adapted from Godinho et al. (2016): participants were asked to agree with two statements about their perception of the efficacy of the antivirals against pandemic flu on 9-point scale, from 1=strongly disagree to 9=strongly agree. Higher scores indicate higher levels of perceived efficacy of the antivirals. This is a composite variable reflecting the average of the items that compose it. |
| Perceived Response Costs | At 20 minutes (i.e. straight after exposure to the health messages) | Measured by self-report items adapted from Godinho et al. (2016): participants were asked to agree with four statements about their beliefs about the side effects and negative consequences of using antivirals on 9-point scale, from 1=strongly disagree to 9=strongly agree. Higher scores indicate higher levels of perceived response costs. This is a composite variable reflecting the average of the items that compose it. |
Participant flow
Pre-assignment details
Of 349 submitted questionnaires, 30 were incomplete and 103 were deemed to be invalid, leaving to analyse 216 valid questionnaires.
Participants by arm
| Arm | Count |
|---|---|
| Virus Agency/Negative Attribute Framing Participants in this condition, after reading a hypothetical scenario, received health messages describing the pandemic flu and the efficacy of the antivirals using linguistic expressions that assigned transmission to the pandemic flu virus itself (Virus Agency Assignment framing), whilst describing the side effects of the antivirals in terms of chances of experiencing side effects after using them (Negative Framing).
Agency Assignment framing: Linguistic framing used in written health messages. Each version of the messages described the pandemic flu and the efficacy of the antivirals using linguistic expressions that assigned transmission to either humans (HA: 'You can contract the virus when you touch…') or the pandemic flu virus itself (VA: 'It can infect you when you touch…')
Attribute framing: Each message described the side effects of the antivirals in terms of chances of either experiencing or not side effects | 47 |
| Human Agency/Negative Attribute Framing Participants in this arm, after reading a hypothetical scenario, received a health message describing the pandemic flu and the efficacy of the antivirals using linguistic expressions that assigned transmission to humans (Human Agency Assignment framing), whilst describing the side effects of the antivirals in terms of chances of experiencing side effects after using them (Negative Framing health messages).
Agency Assignment framing: Linguistic framing used in written health messages. Each version of the messages described the pandemic flu and the efficacy of the antivirals using linguistic expressions that assigned transmission to either humans (HA: 'You can contract the virus when you touch…') or the pandemic flu virus itself (VA: 'It can infect you when you touch…')
Attribute framing: Each message described the side effects of the antivirals in terms of chances of either experiencing or not side effects | 60 |
| Human Agency /Positive Attribute Framing Participants in this arm, after reading a hypothetical scenario, received a health message describing the pandemic flu and the efficacy of the antivirals using linguistic expressions that assigned transmission to humans (Human Agency Assignment framing), whilst describing the side effects of the antivirals in terms of chances of not experiencing side effects after using them (Positive Framing).
Agency Assignment framing: Linguistic framing used in written health messages. Each version of the messages described the pandemic flu and the efficacy of the antivirals using linguistic expressions that assigned transmission to either humans (HA: 'You can contract the virus when you touch…') or the pandemic flu virus itself (VA: 'It can infect you when you touch…')
Attribute framing: Each message described the side effects of the antivirals in terms of chances of either experiencing or not side effects | 56 |
| Virus Agency /Positive Attribute Framing Participants in this arm, after reading a hypothetical scenario, received a health message describing the pandemic flu and the efficacy of the antivirals using linguistic expressions that assigned transmission to the pandemic flu virus itself (Virus Agency Assignment framing), whilst describing the side effects of the antivirals in terms of chances of not experiencing side effects after using them (Positive Framing).
Agency Assignment framing: Linguistic framing used in written health messages. Each version of the messages described the pandemic flu and the efficacy of the antivirals using linguistic expressions that assigned transmission to either humans (HA: 'You can contract the virus when you touch…') or the pandemic flu virus itself (VA: 'It can infect you when you touch…')
Attribute framing: Each message described the side effects of the antivirals in terms of chances of either experiencing or not side effects | 53 |
| Total | 216 |
Baseline characteristics
| Characteristic | Virus Agency/Negative Attribute Framing | Human Agency/Negative Attribute Framing | Human Agency /Positive Attribute Framing | Virus Agency /Positive Attribute Framing | Total |
|---|---|---|---|---|---|
| Age, Customized Age Range 18-24 years | 22 Participants | 25 Participants | 21 Participants | 15 Participants | 83 Participants |
| Age, Customized Age Range 25-34 years | 16 Participants | 21 Participants | 24 Participants | 21 Participants | 82 Participants |
| Age, Customized Age Range 35-49 years | 7 Participants | 14 Participants | 7 Participants | 13 Participants | 41 Participants |
| Age, Customized Age Range 50+ years | 2 Participants | 0 Participants | 4 Participants | 4 Participants | 10 Participants |
| Baseline Intentions to take the antivirals as recommended | 8.16 units on a scale STANDARD_DEVIATION 0.21 | 7.93 units on a scale STANDARD_DEVIATION 0.19 | 7.95 units on a scale STANDARD_DEVIATION 0.19 | 8.26 units on a scale STANDARD_DEVIATION 0.2 | 8.08 units on a scale STANDARD_DEVIATION 0.1 |
| Race/Ethnicity, Customized Ethnicity Other | 18 Participants | 9 Participants | 14 Participants | 10 Participants | 51 Participants |
| Race/Ethnicity, Customized Ethnicity White | 29 Participants | 51 Participants | 42 Participants | 43 Participants | 165 Participants |
| Sex: Female, Male Female | 28 Participants | 35 Participants | 33 Participants | 28 Participants | 124 Participants |
| Sex: Female, Male Male | 19 Participants | 25 Participants | 23 Participants | 25 Participants | 92 Participants |
Adverse events
| Event type | EG000 affected / at risk | EG001 affected / at risk | EG002 affected / at risk | EG003 affected / at risk |
|---|---|---|---|---|
| deaths Total, all-cause mortality | — / — | — / — | — / — | — / — |
| other Total, other adverse events | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 |
| serious Total, serious adverse events | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 |
Outcome results
Change in Intentions to Take Antivirals for Pandemic Flu
Mean adherence intentions post- exposure to the health information in the 4 groups. Intentions were measured by self-report items: participants were asked to agree with three statements about their intentions to take antivirals as recommended in the hypothetical scenario (on 9-point scale, where 1=strongly disagree to 9=strongly agree). The scores reported below represent a composite variable 'change in intentions', which reflects the average of the three items that compose it.
Time frame: Straight after exposure to the health messages
Population: ANCOVAs, setting baseline intentions as a covariate, were performed to determine whether post-exposure mean adherence intentions differed between groups.
| Arm | Measure | Value (MEAN) | Dispersion |
|---|---|---|---|
| Virus Agency/Negative Attribute Framing | Change in Intentions to Take Antivirals for Pandemic Flu | 8.29 score on a scale | Standard Deviation 0.12 |
| Human Agency/Negative Attribute Framing | Change in Intentions to Take Antivirals for Pandemic Flu | 8.16 score on a scale | Standard Deviation 0.12 |
| Human Agency /Positive Attribute Framing | Change in Intentions to Take Antivirals for Pandemic Flu | 8.23 score on a scale | Standard Deviation 0.12 |
| Virus Agency /Positive Attribute Framing | Change in Intentions to Take Antivirals for Pandemic Flu | 8.35 score on a scale | Standard Deviation 0.11 |
Perceived Efficacy of the Antivirals
Measured by two self-report items (adapted from Godinho et al. (2016): participants were asked to agree with two statements about their perception of the efficacy of the antivirals against pandemic flu on 9-point scale, from 1=strongly disagree to 9=strongly agree. Higher scores indicate higher levels of perceived efficacy of the antivirals. This is a composite variable reflecting the average of the items that compose it.
Time frame: At 20 minutes (i.e. straight after exposure to the health messages)
Population: Based on the literature, we predicted that Positive framing of the side effects would lead to higher response efficacy compared to the Negative Framing. To test our hypothesis about differences between PF conditions and NF conditions, in this analysis we combined the VAPF and HAPF groups on one hand and the VANF and HANF groups on the other.
| Arm | Measure | Value (MEAN) | Dispersion |
|---|---|---|---|
| Virus Agency/Negative Attribute Framing | Perceived Efficacy of the Antivirals | 6.45 score on a scale | Standard Deviation 0.15 |
| Human Agency/Negative Attribute Framing | Perceived Efficacy of the Antivirals | 6.59 score on a scale | Standard Deviation 0.14 |
Perceived Response Costs
Measured by self-report items adapted from Godinho et al. (2016): participants were asked to agree with four statements about their beliefs about the side effects and negative consequences of using antivirals on 9-point scale, from 1=strongly disagree to 9=strongly agree. Higher scores indicate higher levels of perceived response costs. This is a composite variable reflecting the average of the items that compose it.
Time frame: At 20 minutes (i.e. straight after exposure to the health messages)
Population: Based on the literature, we predicted that Positive framing of the side effects would lead to lower response costs compared to the Negative Framing. To test our hypothesis about differences between PF conditions and NF conditions, in this analysis we combined the VAPF and HAPF groups on one hand and the VANF and HANF groups on the other.
| Arm | Measure | Value (MEAN) | Dispersion |
|---|---|---|---|
| Virus Agency/Negative Attribute Framing | Perceived Response Costs | 3.53 score on a scale | Standard Deviation 0.17 |
| Human Agency/Negative Attribute Framing | Perceived Response Costs | 3.70 score on a scale | Standard Deviation 0.17 |
Perceived Self-efficacy
Measured by one self-report item (adapted from Witte et al. (2001): participants were asked to agree with a statement about their perceived ability to take the antivirals as recommended on 9-point scale, from 1=strongly disagree to 9=strongly agree. Higher scores indicate higher levels of reported self-efficacy.
Time frame: At 20 minutes (i.e. straight after exposure to the health messages)
Population: Based on the literature, we predicted that the Agency Assignment (i.e. VA vs. HA) would not affect self-efficacy measures. To test our hypothesis, in this analysis we combined the VANF and VAPF groups on one hand and the HANF and HAPF groups on the other.
| Arm | Measure | Value (MEAN) | Dispersion |
|---|---|---|---|
| Virus Agency/Negative Attribute Framing | Perceived Self-efficacy | 8.23 score on a scale | Standard Deviation 0.12 |
| Human Agency/Negative Attribute Framing | Perceived Self-efficacy | 8.45 score on a scale | Standard Deviation 0.13 |
Perceived Severity of the Pandemic
Measured by one self-report item (adapted from Witte et al. (2001): participants were asked to agree with a statement about their perception of the severity of the pandemic flu on 9-point scale, from 1=strongly disagree to 9=strongly agree. Higher scores indicate higher levels of perceived severity of the pandemic flu threat.
Time frame: At 20 minutes (i.e. straight after exposure to the health messages)
Population: Based on the literature, it was predicted that the Virus Agency would lead to higher perceptions of severity of the pandemic than the Human Agency assignment. To test our hypothesis about differences between VA conditions and HA conditions, in this analysis we combined the VANF and VAPF groups on one hand and the HANF and HAPF groups on the other.
| Arm | Measure | Value (MEAN) | Dispersion |
|---|---|---|---|
| Virus Agency/Negative Attribute Framing | Perceived Severity of the Pandemic | 5.92 score on a scale | Standard Deviation 0.19 |
| Human Agency/Negative Attribute Framing | Perceived Severity of the Pandemic | 6.59 score on a scale | Standard Deviation 0.21 |
Perceived Susceptibility to the Pandemic Flu
Measured by self-report items: participants were asked to state how likely they were to get sick with pandemic flu, had they not taken prophylactic medication o a 9-point scale, where 1=not likely at all, to 9=extremely likely.
Time frame: At 20 minutes (i.e. straight after exposure to the health messages)
Population: Based on the literature, it was predicted that the Virus Agency would lead to higher perceptions of susceptibility than the Human Agency assignment. To test our hypothesis about differences between VA conditions and HA conditions, in this analysis we combined the VANF and VAPF groups on one hand and the HANF and HAPF groups on the other.
| Arm | Measure | Value (MEAN) | Dispersion |
|---|---|---|---|
| Virus Agency/Negative Attribute Framing | Perceived Susceptibility to the Pandemic Flu | 6.32 score on a scale | Standard Deviation 0.14 |
| Human Agency/Negative Attribute Framing | Perceived Susceptibility to the Pandemic Flu | 6.37 score on a scale | Standard Deviation 0.15 |
Worry of the Pandemic Flu Threat
Measured by self-report items adapted from Witte et al. (2001): participants were asked to agree with two statements about their perceived worry on 9-point scale, from 1=strongly disagree to 9=strongly agree. Higher scores indicate higher levels of reported worry about pandemic flu threat. This is a composite variable reflecting the average of the items that compose it.
Time frame: At 20 minutes (i.e. straight after exposure to the health messages)
Population: Based on the literature, it was predicted that the Virus Agency (main effect) would lead to higher worry than the Human Agency assignment. To test our hypothesis about differences between VA conditions and HA conditions, in this analysis we combined the VANF and VAPF groups on one hand and the HANF and HAPF groups on the other.
| Arm | Measure | Value (MEAN) | Dispersion |
|---|---|---|---|
| Virus Agency/Negative Attribute Framing | Worry of the Pandemic Flu Threat | 6.06 score on a scale | Standard Deviation 0.19 |
| Human Agency/Negative Attribute Framing | Worry of the Pandemic Flu Threat | 6.50 score on a scale | Standard Deviation 0.2 |