Does message framing affect changes in behavioural intentions in people with psoriasis? A randomized exploratory study examining health risk communication

Abstract

Message framing is important in health communication research to encourage behaviour change. Psoriasis, a long-term inflammatory skin condition, has additional comorbidities including high levels of anxiety and cardiovascular disease (CVD), making message framing particularly important. This experimental study aimed to: (1) identify whether health messages about psoriasis presented as either gain- or loss-framed were more effective for prompting changes in behavioural intentions (BI), (2) examine whether BI were driven by a desire to improve psoriasis or reduce CVD risk; (3) examine emotional reactions to message frame; and (4) examine predictors of BI. A two by two experiment examined the effects on BI of message frame (loss vs. gain) and message focus (psoriasis symptom reduction vs. CVD risk reduction). Participants with psoriasis (n = 217) were randomly allocated to one of four evidence-based health messages related to either smoking, alcohol, diet or physical activity, using an online questionnaire. BI was the primary outcome. Analysis of variance tests and hierarchical multiple regression analyses were conducted. A significant frame by focus interaction was found for BI to reduce alcohol intake (p = .023); loss-framed messages were more effective for CVD risk reduction information, whilst gain-framed messages were more effective for psoriasis symptom reduction information. Message framing effects were not found for BI for increased physical activity and improving diet. High CVD risk was a significant predictor  of increased BI for both alcohol reduction (β = .290, p < .01) and increased physical activity (β = −.231, p < .001). Message framing may be an important factor to consider depending on the health benefit emphasised (disease symptom reduction or CVD risk reduction) and patient-stated priorities. Condition-specific health messages in psoriasis populations may increase the likelihood of message effectiveness for alcohol reduction.

Keywords:

• Psoriasis
• risk communication
• health behaviour
• behavioural intentions
• message framing

Introduction

Psoriasis is a complex, long-term, inflammatory skin condition affecting around 2% of the population (Parisi et al., 2013). Severe psoriasis has been associated with cardiovascular disease (CVD) however the independent risk factor for this group is co-morbid inflammatory arthritis (Parisi et al., 2015). Of perhaps greater importance is that people with psoriasis are more likely to engage in unhealthy behaviours such as excessive alcohol intake, smoking, being over-weight and sedentary than those without psoriasis (Hayes & Koo, 2010; Naldi et al., 2005; Parisi et al., 2015; Samarasekera et al., 2013). This contributes to psoriasis onset, severity, and consequently increases the risk of CVD (National Institute for Health & Care Excellence, [NICE], 2010; Yusuf et al., 2004). Despite this, risk factors are often undetected (Rutter et al., 2016) or sub-optimally managed (Nelson et al., 2016). People are unaware of the links between health behaviours and psoriasis exacerbation, and they do not receive the necessary behaviour change support consistent with clinical guidelines (Nelson et al., 2013).

Effective health risk communication methods are crucial for CVD risk reduction (Bonner et al., 2014; Keyworth, Nelson, Chew-Graham et al., 2015). Condition-specific information may improve risk perceptions and promote more informed decision-making about behaviour change (Nelson et al., 2017; Waldron et al., 2011). To date, no studies have examined the effectiveness of psoriasis-specific health information in prompting changes in behavioural intentions.

Theoretical framework

Message framing theory suggests that message effectiveness is influenced by the type of behaviour promoted and the framing of the health message (Rothman et al., 2006). Gain-frame messages emphasise the benefits of behaviour change (‘quitting smoking lowers your risk of lung cancer’), and loss-frame messages emphasise the costs of failure to act (‘by not quitting smoking, you increase your risk of lung cancer’). The theory suggests that: (1) gain-frame messages are more persuasive for prevention behaviours (‘low risk’ behaviours) such as increasing physical activity (Latimer et al., 2008), and (2) loss-frame messages are more persuasive for ‘detection’ behaviours (‘high risk’ behaviours) such as promoting mammography use (Abood et al., 2005).

However systematic reviews identified inconsistent results (Gallagher & Updegraff, 2012; O’Keefe & Jensen, 2007, 2009), and other important mechanisms may influence message effectiveness. Emotional reactions to health messages may act as a barrier to accurate risk perception (Naqvi et al., 2006; Quartz, 2009; Ten Hoor et al., 2012), and self-efficacy, an important motivational factor for behaviour change (Armitage & Conner, 2001) may increase the impact of risk information on health behaviours (Sheeran et al., 2014).

Considering immediate versus future health consequences is recognised as a central characteristic in tailored health communication messages (Morris et al., 2016; O’Connor et al., 2009). Whether an individual is concerned about their immediate health (for example their physical appearance), or reducing future health risks (reducing risk of future health conditions) is an important area for research in appearance-related conditions. Messages tailored around these considerations have been shown to be effective in contexts such as skin cancer (Thomas et al., 2011) and safe-sun practices (Aspden et al., 2015). To-date no studies have examined whether people with psoriasis are driven to protect their immediate health (prevention of psoriasis flares) or their future health (CVD risk).

This exploratory study addressed four specific research questions:

(1)	Are gain- or loss-framed messages more effective for prompting changes in behavioural intentions?
(2)	Are behavioural intentions driven by a desire to improve psoriasis symptoms (i.e. immediate consequences; psoriasis skin flare reduction) or reduce CVD risk (i.e. reduce long-term health risks; CVD prevention)?
(3)	To what extent do differently framed health messages produce emotional responses?
(4)	What are the predictors are behavioural intentions in people with psoriasis?

Methods

Participants

People with psoriasis were invited to take part in an online questionnaire via advertisements placed on a patient organisation website (the Psoriasis Association), a University intranet site, in community locations (such as local supermarkets), and through the use of social networking websites (Twitter and Facebook). There were no geographical restrictions on patient participation. Ethical approval was obtained from the relevant university ethics committee (reference: 13118).

Design

A 2 (message frame: loss vs. gain) by 2 (message focus: psoriasis symptom reduction vs. CVD risk reduction) between-participants design was used.

Materials

Health messages were constructed using clinical guidelines for CVD (NICE, 2010) and psoriasis (NICE, 2012), using information about the effects of health behaviour change on either psoriasis symptoms or CVD risk (presented in Figure 1). Using this methodology allows for tight experimental control, allowing the researcher to specifically identify which messages are effective for prompting behavioural change (de Bruijn et al., 2014; French et al., 2004; Wright et al., 2006).

Figure 1. Health messages randomly assigned to participants.

Measures

Intention to follow the advice offered to reduce risk or promote symptom management was the primary outcome, measured on a 7-point scale (1–7; possible range 3–21) in line with published recommendations (Ajzen, 1991; Francis et al., 2004). Three items were used e.g. ‘I intend to reduce the amount of alcohol I drink in the next month’. Items were related to intentions to change the following health behaviours: smoking, alcohol, diet, and physical activity. Emotional response was the secondary outcome. Items consisted of two items on a 10-point scale (1–10; possible range 2–20), adapted from the literature (French et al., 2004; Klein, 1997) e.g. ‘How concerned are you by reading this information?’. Three items measured self-efficacy (Armitage & Conner, 2001; Wright et al., 2006) measured on a 7-point scale (1–7; possible range 3–21) e.g. ‘How confident are you that you can stop smoking in the next month?’

Procedure

Prior to commencement, detailed information about the association between psoriasis, CVD and health behaviours was presented on the participant information sheet. Participants provided demographic information, current and past severity of psoriasis and were asked about their current health behaviours. Participants were informed about recommended physical activity levels and units of alcohol according to public health guidelines, and were asked to rate whether they met the recommendations (to determine high vs. low risk profile). Participants were randomly allocated to read one of the four health messages after which they answered the post-message items. The questionnaire was piloted-tested (Dillman, 2000) with a small convenience sample (n = 16) of members of an established research user group of people living with psoriasis.

Sample size and power calculation

In order to test the interaction between the two main effects (message frame: loss vs. gain; and message focus: psoriasis symptom reduction vs. CVD risk reduction), a larger than is feasible sample size would be required. The study was, therefore, powered to detect a standardised difference of 0.5 between the two groups (message frame and message focus) defined by a single main effect. As we wanted to test both main effects, an alpha of 2.5% was used for the test of each separate main effect, thus maintaining an overall alpha of 5%. With 80 participants per main effect (40 per group; 160 in total), the study achieved 80% power to detect differences.

Analysis

A series of 2 by 2 analysis of variance tests (ANOVAs) were conducted to investigate the main effects of message frame, message focus and the interaction effects on behavioural intentions. Hierarchical multiple regression analyses were used to determine the predictors of behavioural intentions. Effect sizes were calculated using the means and SDs of each experimental group. SPSS v22 was used to conduct the analyses.

At step 1 demographic variables (age and sex) and self-efficacy were added. At step 2 message frame (gain vs. loss), message focus (psoriasis vs. CVD message) and the frame x focus interaction term were added to test for any effect of the wording of health messages on behavioural intentions. At step 3 risk category (according to current health behaviour profile; high vs. low) was added. The adjusted multiple correlations (R²) measured the proportion of the variation in behavioural intentions predicted by the independent variables. Changes in adjusted R² indicate the amount of variance explained by each step of the model.

Results

The sample (n = 217; Figure 2) was mostly female (58.1%), White British (82%), and reported having mild (38.2%), moderate (29.5%) or moderately severe (19.4%) psoriasis (Table 1). Descriptive statistics are shown in Table 2. Data are presented for primary variables (behavioural intentions: alcohol, diet, and physical activity) and the secondary variable (emotional response to information). Due to the low numbers of smokers (n = 42) included across the four conditions we excluded behavioural intentions related to the smoking in the final analysis.

Figure 2. Study flow diagram.

Table 1. Demographics of participants randomized.

Variable	N (%)
Gender^1
Male	75 (34.6)
Female	126 (58.1)
Did not state	16 (7.4)
Age^2 M (SD), Range	41.23 (14.31), 15–83
0–19	3 (1.4)
20–39	94 (43.3)
40–59	70 (32.3)
60–79	23 (10.6)
80>	1 (.5)
Did not state	20 (9.2)
Ethnicity
White British	178 (82)
White Irish	6 (2.8)
Mixed – White/Asian	2 (.9)
Asian or Asian British –Pakistani	1 (.5)
Asian or Asian British – Indian	5 (2.3)
Melanesian	1 (.5)
Mixed – Hispanic European	1 (.5)
White – Dutch	1 (.5)
Other White	1 (.5)
White European	2 (.9)
Welsh	1 (.5)
Chinese	2 (.9)
Did not state	26 (12)
Age of Pso diagnosis
<16	71 (32.7)
16–22 (early onset)	54 (24.9)
23–39	62 (28.6)
40–57	19 (8.8)
57> (late onset)	5 (2.3)
Self-reported Pso severity
Current
Mild	83 (38.2)
Mild to moderately severe	64 (29.5)
Moderately severe	42 (19.4)
Moderately severe to severe	13 (6)
Severe	9 (4.1)
Ever
Mild	18 (8.3)
Mild to moderately severe	24 (11.1)
Moderately severe	49 (22.6)
Moderately severe to severe	70 (32.3)
Severe	49 (22.6)
Did not state	6 (2.8)
Diagnosis of PsoA	40 (18.4)
Smoker
Ever	113 (52.1)
Current	42 (17)
Did not state	4 (1.8)
Meeting DOH guidelines for physical activity^3
Achieve this every week	86 (39.6)
Almost there, but not quite	54 (24.9)
Around half of recommended	33 (15.2)
A long way off recommended	41 (18.9)
Did not state	3 (1.4)
Typical month, how often do you drink alcohol?
Never	29 (13.4)
Once a month or less	43 (19.8)
2 to 4 times a month	66 (30.4)
2 to 3 times a week	49 (22.6)
4 or more times a week	28 (12.9)
Did not state	2 (.9)
Units of alcohol on a typical day
1–2	49 (22.6)
3–4	66 (30.4)
5–6	32 (14.7)
7–8	20 (9.2)
9+	21 (9.7)
Did not state	29 (13.4)
6 or more units on one occasion
Never	66 (30.4)
Less than monthly	73 (33.6)
Monthly	25 (11.5)
Weekly	38 (17.5)
Daily or almost daily	9 (4.1)
Did not state	6 (2.8)

¹ Gender UK psoriasis population characteristics (males 47.7%) according to Springate et al. (2017).

² Age Psoriasis UK population characteristics (0–19 years [12.1%), 20–39 years [29.6%], 40–59 years [31.4%], 60–79 years [22.8%], 80 > years [4%]) according to Springate et al. (2017).

³ The Department of Health recommends adults are moderately active for 2 and a half hours or vigorously active for 75 min each week. For example 30 min on at least 5 days a week.

Pso = Psoriasis; PsoA = Psoriatic Arthritis; DOH = Department of Health.

Table 2. Descriptive statistics for all primary and secondary outcome variables.

Variable	M (SD)
	Gain-frame			Loss-frame
	Psoriasis symptom reduction (n = 56)	CVD risk reduction (n = 56)	Overall	Psoriasis symptom reduction (n = 52)	CVD risk reduction (n = 48)	Overall
Behavioural intentions (Alcohol)	10.56 (6.30)	9.60 (5.33)	10.00 (5.74)	7.24 (4.56)	10.35 (5.15)	9.15 (5.13)
Behavioural intentions (Diet)	13.21 (5.37)	13.23 (5.55)	13.22 (5.45)	12.63 (6.04)	14.22 (5.78)	13.53 (5.91)
Behavioural intentions (Physical activity)	13.87 (5.27)	13.62 (5.71)	13.73 (5.50)	12.97 (5.19)	14.43 (4.89)	13.80 (5.04)
Emotional response to information	8.13 (4.46)	8.47 (5.34)	8.32 (4.95)	7.00 (5.10)	10.16 (4.91)	8.76 (5.21)

The impact of message frame and message focus on behavioural intentions is presented in Table 3. Results are presented in detail according to each outcome measure.

Table 3. Results of ANOVA examining effects of message framing and message focus on behavioural intentions and emotional responses to health messages.

Effect	df	F	P	Partial n^2
Alcohol intention
Message frame	1	2.081	.151	.013
Message focus	1	1.453	.230	.009
Message frame X message focus interaction	1	5.241	.023*	.032
Total	163
Diet intention
Message frame	1	.062	.804	.000
Message focus	1	.966	.327	.005
Message frame X message focus interaction	1	.903	.343	.005
Total	197
Physical activity intention
Message frame	1	.003	.956	.000
Message focus	1	.619	.432	.003
Message frame X message focus interaction	1	1.232	.268	.006
Total	201
Emotional response
Message frame	1	.145	.704	.001
Message focus	1	5.743	.018*	.029
Message frame X message focus interaction	1	3.752	.054	.019
Total	197

* Significant at the p < .05 level.

Alcohol behavioural intentions

Neither main effects for message frame [F(1, 159) = 2.08, p = .15] nor message focus [F(1, 159) = 1.45, p = .23] reached statistical significance.

There was a significant message frame by message focus interaction effect (p = .023) indicating when presented with different messages, there was a difference in reported behavioural intentions based on the message frame and focus. When presented with information about psoriasis symptom reduction, reported behavioural intentions were significantly higher in the gain-frame condition (mean = 10.56, standard deviation (SD) = 6.30) compared to the loss-frame condition (M = 7.24, SD = 4.56). However, when presented with information about CVD risk reduction, reported behavioural intentions were higher in the loss-frame condition (M = 10.35, SD = 5.15) compared to the gain-frame condition (M = 9.60, SD = 5.33), however the effect size (d = .01) was small.

Diet behavioural intentions

Neither of the main effects were statistically significant [message frame: F(1, 193) = .06, p = .80; message focus: F(1, 193) = .97, p = .33], nor was the interaction effect [F(1, 193) = .90, p = .34].

Physical activity behavioural intentions

Neither of the main effects were statistically significant [message frame: F(1, 193) = .00, p = .96; message focus: F(1, 193) = .62, p = .43], nor was the interaction effect [F(1, 193) = 1.23, p = .27].

Emotional response

There was a significant main effect for message focus [F(1, 193) = 5.74, p = .02]; however, the effect size was small (d = .03). The main effect for message frame was not statistically significant [F(1, 193) = .15, p = .70], neither was the interaction effect [F(1, 193) = 3.75, p = .05].

Multiple regression analyses

Results are shown in Table 4 (intentions to reduce alcohol intake), Table 5 (intentions to improve diet) and Table 6 (intentions to increase physical activity).

Table 4. Regression model for continuous (age and self-efficacy) and categorical variables on alcohol behavioural intentions (n = 165).

Model		R^2	ΔR^2	β	95%CI	p-value
1		.008	−.014
	Age			.049	−.055 to .080	.572
	Sex			.054	−1.395 to 2.593	.536
	Self-efficacy			.051	−.160 to .289	.554
2		.046	.004
	Age			.056	−.055 to .079	.514
	Sex			.035	−1.533 to 2.490	.692
	Self-efficacy			.047	−.162 to .284	.581
	Message frame			−.636	−13.506 to −1.072	.027*
	Message focus			−.441	−11.042 to .202	.085
	Frame X focus interaction			.784	.333 to 7.923	.040*
3		.121	.076
	Age			.098	−.042 to .088	.246
	Sex			−.014	−2.094 to 1.875	.870
	Self-efficacy			.119	−.077 to .369	.164
	Message frame			−.608	−12.874 to −.811	.028*
	Message focus			−.403	−10.478 to .430	.101
	Frame X focus interaction			.722	.079 to 7.448	.049*
	Risk category			.290	1.167 to 5.152	.001*

* Significant at the < .05 level.

Table 5. Regression model for continuous (age and self-efficacy) and categorical variables on diet behavioural intentions (n = 197).

Model		R^2	ΔR^2	β	95% CI	p-value
1		.144	.130
	Age			−.261	−.247 to .040	.157
	Sex			.046	−3.670 to 4.736	.801
	Self-efficacy			.266	−.114 to .808	.142
2		.153	.124
	Age			−.258	−.254 to .049	.185
	Sex			.050	−3.950 to 5.109	.799
	Self-efficacy			.269	−.137 to .839	.160
	Message frame			−.195	−16.303 to 11.831	.751
	Message focus			−.146	−14.369 to 11.061	.795
	Frame X focus interaction			.313	−6.949 to 10.211	.705
3		.228	.035
	Age			−.291	−.268 to .031	.131
	Sex			−.024	−4.918 to 4.261	.902
	Self-efficacy			.341	−.044 to .947	.080
	Message frame			−.303	−17.661 to 10.126	.617
	Message focus			−.263	−15.863 to 9.347	.636
	Frame X focus interaction			.376	−6.297 to 10.514	.641
	Risk category*			.314	−.831 to 5.692	.139

* model based on n = 36 with BMI data.

Table 6. Regression model for continuous (age and self-efficacy) and categorical variables on physical activity behavioural intentions (n = 201).

Model		R^2	ΔR^2	β	95% CI	p-value
1		.397	.387
	Age			−.177	−.113 to −.025	.004*
	Sex			.041	−1.092 to 1.465	.49
	Self-efficacy			.587	.540 to .796	<.001*
2		.402	.381
	Age			−.175	−.113 to −.024	.004*
	Sex			.037	−1.128 to 1.495	.55
	Self-efficacy			.586	.536 to .795	<.001*
	Message frame			−.127	−5.631 to 2.530	.51
	Message focus			−.054	−4.695 to 2.681	.760
	Frame X focus interaction			.170	−1.476 to 3.501	.513
3		.454	.432
	Age			−.188	−.117 to −.032	.001*
	Sex			.055	−.834 to 1.712	.35
	Self-efficacy			.613	.565 to .816	<.001*
	Message frame			−.137	−5.799 to 2.084	.46
	Message focus			−.075	−5.014 to 2.121	.66
	Frame X focus interaction			.195	−1.119 to 3.694	.43
	Risk category			−.231	−3.495 to −1.080	<.001*

* Significant at the < .05 level.

Alcohol

At step 1 neither age, sex nor self-efficacy were independent predictors of intentions to reduce alcohol consumption. At step 2 the addition of message frame (β = −6.36, p = .03) and the frame x focus interaction (β = .784, p = .04) significantly explained an additional .4% of the variance. This indicates that gain-framed messages resulted in higher behavioural intentions, for messages about psoriasis symptom reduction, gain-framed messages resulted in higher behavioural intentions, and for messages about CVD risk reduction loss-framed messages resulted in higher behavioural intentions. Message focus alone did not significantly contribute to the variance. In the final regression model (step 3) the addition of risk category (β = .290, p ≤ .01) significantly explained 7.6% of the total variance, indicating that participants identified as high risk (according to self-reported current alcohol intake) reported higher behavioural intentions (M = 11.29, SD = 5.40) to reduce alcohol compared to those identified as low risk (M = 8.44, SD = 5.32). Message frame (β = −6.08, p = .03) and the frame x focus interaction (β = .722, p = .05) remained significant in the final model.

Diet

At step 1, neither age, sex nor self-efficacy were independent predictors of intentions to improve participants’ diet. At step 2, the addition of message frame, message focus and frame x focus explained no additional variance. At step 3, the addition of risk category (according to self-reported BMI) also explained no additional variance in the model.

Physical activity

At step 1, both age (β = −1.77, p ≤ .01) and self-efficacy (β = .587, p ≤ .001) were independent predictors of intentions to increase physical activity levels and explained 38.7% of the variance, indicating that younger participants reported higher behavioural intentions, and those with higher self-efficacy reported higher behavioural intentions. At step 2 neither the addition of message frame, message focus or frame x focus interaction significantly added to the variance. Both age (β = −1.75, p ≤ .01) and self-efficacy (β = .586, p ≤ .001) remained significant but accounted for no additional variance. In the final regression model (step 3), the addition of risk category (β = −.231, p ≤ .001) explained an additional 5.1% of the variance, indicating that participants self- identified as high risk reported higher intentions (M = 14.22, SD = 5.20) to reduce alcohol compared to those identified as low risk (M = 12.49, SD = 5.34).

Discussion

This is the first study to examine the effects of message framing on reported intentions to modify health behaviours in people with psoriasis. There was a small but significant interaction between message frame and message focus in the context of alcohol reduction, contrary to previous studies. For messages focused on short-term health risk (reducing visible psoriasis symptoms), gain-framed messages were more persuasive than loss-framed messages in increasing behavioural intentions. Messages about long-term health risk (CVD risk reduction) were more persuasive when presented as a loss-framed message rather than a gain-framed message. This finding must be interpreted with caution given the small interaction effect observed, and the nature of this study in the specific context of a long-term condition.

Our study suggests that loss-framed messages may be effective in situations beyond detection behaviours as previously suggested (O’Keefe & Jensen, 2009; Rothman et al., 1999). Using this approach in people with psoriasis and greater CVD risk may encourage behaviour change but needs further testing beyond our initial findings.

There was a significant effect for message focus (psoriasis symptom reduction compared to CVD risk reduction) on emotional responses to information. Messages about CVD produced a higher emotional response compared to messages about psoriasis symptom reduction. This finding has important implications for how sensitive health information is delivered to patients (McBride et al., 2010).

We found no effect of message framing on behavioural intentions related to increasing physical activity or diet. Similar findings have been observed for physical activity and making dietary changes (de Bruijn et al., 2014; van Assema et al., 2001), and health behaviours more generally (Avraham et al., 2016).

Implications for practice

Message framing has been examined in a range of behaviours (Abhyankar et al., 2008; Churchill et al., 2016; Morris et al., 2016; Pavey & Churchill, 2014; Updegraff et al., 2015), and now in a psoriasis population. We demonstrate a clear need for condition-specific health information for people with psoriasis (Keyworth, Nelson, Chew-Graham et al., 2015; Keyworth, Nelson, Griffiths et al., 2015; Nelson et al., 2017).

The precise reason for observing significant effects in changes to behavioural intentions for reducing alcohol consumption must be examined in future research. However one hypothesis may be that in this population, the desire to reduce alcohol consumption outweighed all other health behaviours. One-third of our sample was exceeding recommended guidelines for alcohol consumption, consistent with the psoriasis population generally (McAleer et al., 2011). High alcohol consumption is particularly problematic in this population (Parisi et al., 2017), and our results suggest some wish to make the necessary behavioural changes.

Risk category was a significant predictor of increased intentions for both alcohol reduction and increasing physical activity. People self-identifying as high risk according to their alcohol consumption or by not meeting guidelines for recommended activity reported greater intentions to modify their health behaviours. This illustrates the need for targeted health risk communication strategies. Psoriasis-specific information may increase message salience and facilitate cognitive processing (Kreuter et al., 1999; Latimer et al., 2005).

Based on our findings we recommend psoriasis-specific health messages that consider both the health behaviour being targeted and the message frame used. This will be more effective if done in a way that considers a range of psychological mechanisms involved in behaviour change, as communicating risk information alone may not be sufficient to induce sustained behaviour change (French et al., 2017). Finding new ways of communicating this information in a way that aids understanding and facilitates behaviour change can only serve to improve the clinical management of patients.

Limitations

The low interaction effect observed in the context of alcohol reduction, and a lack of significant effects in further health behaviours, may be due to limitations in the specific measures used in the study. Validated measured for psoriasis populations would benefit from being sensitive to the known psychological impact of living with psoriasis (Kurd et al., 2010). The cross-sectional nature of the study limits our ability to assess the longer-term impact of appropriately framed health messages on behavioural intentions. Including a measure of self-reported behaviour in future studies would strengthen such conclusions. Additionally, the majority of participants in the study were recruited from the UK. Not all demographic information is available for the psoriasis population generally, and whilst we could not match population characteristics for psoriasis identically in terms of age and gender (Springate et al., 2017), alcohol intake closely resembled known levels of consumption (Parisi et al., 2017) with strengthens our hypotheses related to desire to intentions to reduce alcohol intake.

Conclusions

This is the first study to examine health risk communication, in the context of psoriasis, by investigating the role of message framing about future disease risk. For alcohol reduction, loss-framed messages appear to be more effective for CVD risk reduction information, and gain-framed messages are more effective for psoriasis symptom reduction information. Messages about CVD result in higher emotional responses compared to messages about psoriasis symptom reduction. Taken together, our findings should be used to develop important health information for patients with psoriasis.

Funding

This article presents independent research funded by NIHR under its Programme Grants for Applied Research scheme [grant number RP- PG-0608-10163). The views expressed are those of the authors and not necessarily those of the National Health Service, the NIHR or the Department of Health. CEMG is an NIHR Senior Investigator.

Disclosure statement

No potential conflict of interest was reported by the authors.