Prolongation of disease-free life: When is the benefit sufficient to warrant the effort of taking a preventive medicine?

The prolongation of disease-free life (PODL) required by people to be willing to accept an offer of a preventive treatment is unknown. Quantifying the required benefits could guide information and discussions about preventive treatment. In this study, we investigated how large the benefit in prolongation of a disease-free life (PODL) should be for individuals aged 50 – 80 years to accept a preventive treatment offer. We used a cross-sectional survey design based on a representative sample of 6847 Danish citizens aged 50 – 80 years. Data were collected in 2019 through a web-based standardized questionnaire administered by Statistics Denmark, and socio-demographic data were added from a national registry. We analyzed the data with chi-square tests and stepwise multinomial logistic regression. The results indicate that the required minimum benefit from the preventive treatment varied widely between individuals (1-week PODL = 14.8%, ≥ 4 years PODL = 39.2%), and that the majority of individuals (51.1%) required a PODL of ≥ 2 years. The multivariable analysis indicate that education and income were independently and negatively associated with requested minimum benefit, while age and smoking were independently and positively associated with requested minimum benefit to accept the preventive treatment. Most individuals aged 50 – 80 years required larger health benefits than most preventive medications on average can offer. The data support the need for educating patients and health care professionals on how to use average benefits when discussing treatment benefits, especially for primary prevention.


Introduction
Patients with chronic diseases and those at risk of such diseases may require lifelong medication in combination with lifestyle changes to prevent or postpone the progression of disease.Decisions on initiation or continuation of such treatment should ideally be based on comprehensible quantitative information on expected benefits, costs and inconveniences to the patient (Bibbins-Domingo, 2016).The benefit may be presented in many ways such as relative or absolute risk reduction or prolongation of life.While longer life is desirable, longer life without disease is even more desirable.Prolongation of disease-free life (PODL) is therefore a relevant metric when expressing health benefit from prevention (Wright and Weinstein, 1998).Previous research indicates that prolongation of life may vary from days to several years depending on disease type, disease state, and available interventions (Davis et al., 2017;Hansen et al., 2019;Kristensen et al., 2015;Lahoud et al., 2012;Wright and Weinstein, 1998).
When survival or disease-free survival is estimated from clinical trials or observational studies, the estimates represent an average prolongation of (disease free) life.In reality, the prolongation is unlikely to be distributed equally across all those taking the intervention (Wright and Weinstein, 1998).Thus, the distribution of benefits within a population may vary from most patients having the average benefit to some patients having small benefits while others have large benefits (de Vries et al., 2019;Hansen et al., 2019).Knowledge of the distribution of benefits is important to better understand and appreciate treatment effectiveness, and to enable good and valuable communication between health care professionals and patients (Stovring et al., 2008).Unfortunately, we are usually unable to predict the distribution, and the average benefit continues to be the usual way to communicate the information on treatment benefits.

Contents lists available at ScienceDirect
Preventive Medicine  Kristensen et al., 2015;Majidi et al., 2020), while a few also reported rather inconclusive results (Todd et al., 2017;Xie et al., 2020).A few studies noted that the average survival gain from preventive medications can be up to several months if treatments continued for longer durations of time (Davis et al., 2017;Hansen et al., 2019;Albarqouni et al., 2017;Kristensen et al., 2015;Lahoud et al., 2012), i.e. accounting for benefits beyond the trial period.However, there might be a discrepancy between what these preventive medications offer in terms of average survival gains, and what patients expect.For instance, previous research indicates that a majority of patients expect prolongations exceeding 8 months (Albarqouni et al., 2017;Dahl et al., 2007;Halvorsen et al., 2007;Halvorsen et al., 2015;Stovring et al., 2008).Misperceptions about disease risks and treatment benefits are common.
Patients as well as physicians may overestimate treatment benefits and may need guidance as to what benefit can be realistically expected and how to interpretate data on expected average gain in lifetime (Halvorsen et al., 2015;Jaspers et al., 2018;Katz et al., 2015).People aged 50 years and older, who are most at risk for noncommunicable diseases, are frequently offered support for lifestyle changes or long-term medications for the potential benefit of PODL or quality of life (Lionis and Midlov, 2017).Currently it is not very clear what benefit people aged 50 years and older expect and require, and how they value treatment-related costs in terms of side effects and outof-pocket payments.Such information is important because quantifying these benefits could guide discussions and decisions related to preventive interventions (Galesic and Garcia-Retamero, 2011).The aim of the present study was to explore how large an expected benefit expressed as PODL, would need to be for patients to accept it.
In this study, we hypothesized a disease and treatment scenario and asked 6847 Danes aged 50-80 years about the required benefit expressed in PODL for them to accept a specified treatment offer.

Sample and procedure
We used a cross-sectional survey design based on a representative (age, gender, geographic residency) group of 15,072 Danish citizens aged 50-80 years.Data were collected in 2019 through a web-based standardized questionnaire (national digital mailbox, e-Boks.dk)administered by The National Bureau for Statistics in Denmark (Statistics Denmark, www.dst.dk), and socio-demographic data were subsequently added by Statistics Denmark.Two reminders were sent through the national digital mail.In total, 6847 individuals (45%) returned a completed questionnaire.

Treatment scenario
All respondents were asked to relate to the following scenario: "Imagine that you have a health condition that means that you may die earlier than the majority at your age.The condition is presently without any symptoms.Now imagine that you are offered a treatment taken as two 2 pills daily for the next 10 years.The treatment can prolong your expected disease-free life.You will have to pay 300 DKK (approximately 40 Euro; €1.00≈7, 5 DKK) per month out of your own pocket."
Smoking habits were assessed with a single item: "Do you smoke?".The variable was coded with three levels: "current smoker", "quit smoking", and "never a smoker."Alcohol consumption was assessed with a single item "How many units (equivalent to one glass of wine) of alcohol do you drink usually in a week?", and answers were categorized into "non-drinker", "1-7 units" "8-14 units" "15-21 units", and ">21 units".
Taking medicine daily was captured by a single item: "Do you take any medicine every day (excluding vitamins, minerals, omega-3 fatty acids, herbal medicine and other equivalent products)?"which was answered with "yes" or "no".
Willingness to take health risks was assessed by the question, "How do you evaluate your willingness to take a risk related to your health situation?"Participants could answer on a scale from 0 (no risk willingness) to 10 (high risk willingness), and for the analyses, the scores were categorized as "low" (0-4), "moderate" (5-6) and "high" (7-10) (Dohmen et al., 2011).
Satisfaction with life was assessed with the single item, "How satisfied are you all in all with your life?" and satisfaction with health was assessed with a single item, "How satisfied are you with your health?"Both questions were presented with response scales from 0 (lowest level of satisfaction) to 10 (highest level of satisfaction), which were subsequently categorized into "low" (0-4), "moderate" (5-7), and "strong" (8-10).
Self-assessed health was based on the question: "How would you rate your current state of health?" which was rated on a 5-point Likert scale: "nearly perfect", "very good", "good", "poor", or "very poor", and was later dichotomized into "good" and"poor" (Bruin et al., 1996).
Health literacy assessment was based on four Likert-scale questions considering health literacy in disease management about (1) finding information about diseases, (2) finding professional help when ill, (3) a good understanding when communicating with physicians and (4) understanding how to use medicine.Cronbach's alpha for the 4-item health literacy scale was 0.83.Cronbach's alpha is a measure of internal consistency that explains how closely related a set of items are as a group, and a value of >0.70 is generally considered good.The sum scores were dichotomized as "adequate" (≤8 points) or "non-adequate" (≥9 points).The cut-off was chosen to reflect the situation in Denmark, where approximately 20% of the population have been rated as having limited health literacy (Bo et al., 2014).
According to the Danish Act on a Biomedical Research Ethics Committee System, the project was not a biomedical research project and did not need the ethic committee's approval.Data include information that could potentially identify individuals, and the project is therefore registered at the University's Research and Innovation Office, and data handling is in accordance with the General Data Protection Regulation (EU 2016/679).

Statistical analyses
The analyses were performed in Stata 16.0 (StataCorp LP, College Station, TX).Chi-square tests were used to examine bivariate associations.Stepwise multinomial logistic regression was used to test whether variables were independently associated with requested treatment benefit.A stepwise regression with a p-value <0.20 for variable inclusion was performed.Significance levels for testing individual variables were set at p-value <0.05.

Sample characteristics
Table 1 shows the individual characteristics of the study sample and the Danish population (Table 1).Of the initial sample of 6847 respondents, 52.5% were females, which corresponds with the equivalent Danish population of the selected three age decades.The mean age of respondents was slightly higher than in the respective Danish population segment because we had more respondents in the age group from 61 to 70 years.The sample population was slightly better educated than the national average.A slightly higher percentage was still active on the labor market and also had higher incomes.Birthplace and place of residence (geographical) in the sample population correspond to the national distribution.

Expected minimum benefit to accept the preventive treatment offer
The majority of the respondents (51.1%) stated that the expected PODL should be ≥2 years to consider the offered prevention attractive (Fig. 1a).About one in every seven respondents (14.8%) stated that 1week benefit was enough for them to warrant treatment.The oldest respondents (71-80 years) had the largest requirement for benefit, with close to half of them requiring ≥4 years prolongation (Fig. 1b).

Factors associated with expected minimum benefit to accept the preventive treatment offer
Education, income, alcohol consumption, willingness to take health risks, satisfaction with life and health, self-assessed health and health literacy were all negatively associated with the required minimum benefit (i.e., the higher the educational attainment, the lower the required benefit in terms of PODL) (Table 2).In contrast, age, smoking, and taking medicines daily were positively associated with required minimum benefit to accept the preventive treatment offer (i.e., the higher the age, the higher the required benefit in PODL, and smokers requesting higher benefits before accepting preventive treatment as compared to non-smokers) (Table 2).
The multivariable analysis indicated that education and income were independently and negatively associated with required minimum benefit to accept the preventive treatment offer (Table 3).Age and smoking habits were independently and positively associated (Table 3).

Discussion
The results of the study indicate that people's required minimum benefit from a preventive interventions varies widely.A minimum PODL of ≥2 years was required by the majority of individuals aged 50 to 80 years.The wide variation in patients' minimum required benefit is consistent with previous studies (Albarqouni et al., 2017;Jaspers et al., 2018;Trewby et al., 2002).A considerable proportion of respondents choosing lowest and highest possible benefit was also observed by Fontana et al. (2014) (Fontana et al., 2014).As contrast to previous studies, however, our study presented a more realistic treatment scenario with a well-defined treatment regime and a fixed price.Also, we targeted a larger number of individuals aged 50-80 years from the general population, which allowed comparisons of required treatment benefit among various population subgroups related to e.g., age, income, education, smoking.
The expected ≥2 years prolongation stated by the majority of the respondents is far from the average benefit for most types of medications, not the least those taken for preventive purposes (e.g.lipid lowering drugs and antihypertensives).For most types of medications, the average life gain is reported to be merely few weeks to few months, and this is perhaps even shorter for the oldest part of the population and people with underlying health conditions (Davis et al., 2017;Hansen et al., 2019;Lahoud et al., 2012;Majidi et al., 2020;Støvring et al., 2013;Todd et al., 2017;Xie et al., 2020).According to Geoffrey Rose's "Prevention Paradox", when a preventive treatment is targeted at the population levelsimilar to traditional screening programsmany people will gain very little because they are less likely to develop the disease, whereas a few who are at high risk of developing the disease might gain a lot (Rose, 1981).As a result, a sizeable segment of the population including those at high risk, may feel uncertain about taking preventive intervention because they perceive their benefits as too small, which is generally associated with non-use or delayed use of such treatment and could potentially lead to an increased burden of chronic diseases and premature deaths (Meid and Haefeli, 2016).
A key question is how to manage a situation where the expectations of the typical patients by far outweigh what a preventive medication can realistically offerat least on average.Thus, a main challenge in "risk communication" is to create better agreement between effectiveness evidence and patients' beliefs about treatment benefit (Ghosh and Ghosh, 2005).A starting point to improve risk communication may be to analyze what characterizes the population groups with the highest benefit requirements.
Those requesting a PODL of ≥4 years constituted approximately 40% of our respondents.Based upon the bivariate associations, this group was dominated by people with lower educational attainment, low income, and smokers.These three socio-demographic variables are often jointly present (Hiscock et al., 2012).Therefore, it was not surprising that associations related to education, income, and smoking could not be seen in the multivariable analysis model without the response category ≥4 years (Model H).
Individuals with low educational attainment, low income and smoking have the largest challenges health-wise (Di Cesare et al., 2013), but also require the largest benefit to accept an intervention offer.These groups may, however, also have the most to gain, i.e. the largest potential for prolonging their disease-free life, if they, through more tailored and targeted initiatives, can be convinced to engage in preventive lifestyle changes and supportive medications.
In general, PODL can be a useful tool for shared decision making.However, the average gains in life expectancy from preventive treatment can be perceived as small because of the effect of averaging across a population (Wright and Weinstein, 1998).It is important that health care professionals can interpret the data of average benefits (or the distribution of benefits across population) and are able to present the information about predicted benefits (e.g., reduction in the level of LDL cholesterol, PODL, postponement of adverse events) in a way that patients can understand (Bibbins-Domingo, 2016;Goodyear-Smith et al., 2011;Halvorsen et al., 2007;Zipkin et al., 2014).As suggested by the findings of the present study as well as previous studies (Bibbins-Domingo, 2016;Simon-Tuval et al., 2014), knowledge or at least perception of potential benefits is what determine the patients' views as well as compliance towards the preventive treatment.
The positive association of age as well as socio-economic factors with required treatment benefit is not consistent with previous studies.The discrepancy could be due to lower statistical power to detect the difference and different patient (risk) groups targeted in the previous studies (Jaspers et al., 2018;McAlister et al., 2000;Trewby et al., 2002).For instance, Jaspers et al. (2018) noted that a marginally positive association of age with desired treatment benefit among patients with cardiac disease (Jaspers et al., 2018).Trewby et al. (2002) found a negative association of age with expected treatment benefit among a group of patients with and without cardiac disease (Trewby et al., 2002), while McAlister et al. (2000) found no significant effect of age on expected treatment benefit among hypertensive patients without overt cardiovascular disease (McAlister et al., 2000).The impact of other individual characteristics, such as gender, alcohol consumption, satisfaction with health and self-assessed health on expected treatment benefit was observed inconsistently in our subgroup analyses and with confidence intervals approaching the inclusion of 1 in the present study.Considering the large number of comparisons, this is not unexpected, and would need confirmation in other and perhaps larger study populations.

Study limitations and strengths
The large and relatively representative national sample with a range of background variables are important strengths of this study.Still, the results should be interpreted within its limitation, not least because of a modest response rate (45%).Another limitation lies in the possibility that the respondents might not have clearly understood the presented treatment scenario or might have had difficulty in answering the question, and basically used one of the two extreme answering categories.To rule this out, an additional subgroup analysis performed by removing both the lowest and the highest answering categories from the analyses resulted in about the same overall findings (see Supplementary file 1).Further to this end, as very few respondents used the answering option 'uncertain', we think that the respondents generally understood the question.
Key issues in prevention are uncertainty with respect to the true prolongation of disease-free survival and whether all individuals achieve the same prolongation.To hint about such uncertainties, we used the term expected (i.e., average) prolongation as attitude to risk and uncertainty may influence preferences for prevention.
We can, however, not rule out the possibility that the desired treatment benefit was influenced by a respondent's underlying disease or past medical history or perceived sensitivity to negative effects (e.g., side effects).We therefore performed an additional sensitivity analysis by comparing groups taking medications daily vs. not-taking medications daily, and observed relatively small differences between the groups, and did not find robust changes in expected treatment benefits that could be associated with regular use of medications.
Additionally, to address the argument that a higher prevalence of disease among the older population group may change their perception of the scenario presented, we did a subgroup analysis by excluding the oldest decade of participants from the analysis.The results of this subgroup analysis did not change the existing significant associations observed in the total group and thus supported our findings for the entire group of three age decades.

Conclusions
In the present study, a majority of the respondents required two or more years PODL to accept a preventive treatment.Age, educational attainment, income, and smoking were predictors of the required  a "Medium education" includes tertiary and bachelor's education, and "high education" includes master and PhD-educations. benefit.

Fig. 1 .
Fig. 1. a/ Bar-diagram showing expected prolongation of a disease-free life to accept the treatment offer.b.Expected prolongation of a disease-free life by age.(*Note: color should be used in print for the Fig. b)

Table 1
Comparison of sample characteristics with the Danish population.

Table 2
Bivariate association of factors associated with expected prolongation of a disease-free life.