Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Designing privacy in personalized health: An empirical analysis

Designing privacy in personalized health: An empirical analysis A crucial challenge for personalized health is the handling of individuals’ data and specifically the protection of their priv- acy. Secure storage of personal health data is of paramount importance to convince citizens to collect personal health data. In this survey experiment, we test individuals’ willingness to produce and store personal health data, based on dif- ferent storage options and whether this data is presented as common good or private good. In this paper, we focus on the nonmedical context with two means to self-produce data: connected devices that record physical activity and genetic tests that appraise risks of diseases. We use data from a survey experiment fielded in Switzerland in March 2020 and perform regression analyses on a representative sample of Swiss citizens in the French- and German-speaking cantons. Our analysis shows that respondents are more likely to use both apps and tests when their data is framed as a private good to be stored by individuals themselves. Our results demonstrate that concerns regarding the privacy of personal heath data storage trumps any other variable when it comes to the willingness to use personalized health technologies. Individuals prefer a data storage format where they retain control over the data. Ultimately, this study presents results susceptible to inform decision-makers in designing privacy in personalized health initiatives. Keywords personalized health, health apps, genetic tests, data storage (Ries et al., 2010). In this study, we look beyond precision Introduction medicine focused on sick patients and zoom-in on the Personalized health is based on the massive integration of prevention-oriented dimension of personalized health biomedical and social data into research, to determine how (Khoury et al., 2016). individuals’ physical and social environments, genetic We ask: what are individuals’ preferences regarding how endowments and behaviors influence their health (Barazzetti they want to store the personal health data they opt to et al., 2021). This data helps to customize preventive and produce? We analyze two voluntary methods of data gener- therapeutic interventions to the individual genetic and clinical ation which cover the non-medical dimensions of persona- characteristics of each patient (Minvielle et al., 2014). lized health: self-performed genetic tests to send to a Personal health data can cover a wide range of forms of laboratory (Phillips et al., 2018) that can be used to create data, from diagnoses recorded by physician in a patient’s personalized plans for health prevention and connected record to simple metrics about the individual, for instance technologies that monitor physical activity (Allen and their weight. A large amount of personal health data is pro- Christie, 2016) such as health trackers and apps. A central duced in the medical context and often involves that issue regarding any new technology is acceptability medical institutions have access to this data, which raises (McCartney et al., 2011; Leonard et al., 2017). To privacy concerns (Baumann et al., 2018; De Pietro and Francetic, 2018; Platt and Kardia, 2015). However, a host of health data may be created and stored outside of the Department of Actuarial Science, University of Lausanne, Switzerland medical context: individuals can self-produce data via Institute of Political Studies, University of Lausanne, Switzerland apps that record, for instance physical activity (Seifert Corresponding author: et al., 2018; Seifert and Van-delanotte, 2021) or elect to Thibaud Deruelle, Political Science and International Relations Department, undergo blood or genetic tests to appraise their risks to University of Geneva, Boulevard du Pont-d’Arve 40 1205 Genève. develop a cancer (Nakagomi et al., 2016) or other diseases Email: thibaud.deruelle@unige.ch Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https:// creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access page (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 Big Data & Society successfully apply new technologies and get citizens to use a health care system, which puts the responsibility for them, we need to know under which conditions such usage health care on individuals, for example regarding becomes more likely. In the context of personalized health, co-payments for treatments (DePietro et al., 2015; De the usage of such technologies is also important for the Pietro and Francetic, 2018). In other words, Switzerland common good, because the availability of individuals’ is a “consumerdriven” healthcare system (Okma and health data might not only improve their personal health, Crivelli, 2013) and as such, citizens’ willingness to use per- but can in addition provide the basis for research aiming sonalized health technologies is highly relevant. at providing new diagnostic and treatments. The data in this paper consists of a sample (N= 1000) of A crucial challenge for the future of personalized health the Swiss population, which is representative according to is thus the handling of individuals’ data and specifically the the following categories: men and women are distributed protection of their privacy (Ostherr et al., 2017; Vayena equally and the participants, aged between 25 and 65 et al., 2018; Blasimme et al., 2019). For instance, privacy years, are evenly distributed into four age groups. Two protection is important because personalized health bears thirds (67%) of the sample is comprised of Swiss the risk of discrimination among individuals based on Germans and the remaining 33% are from Switzerland’s genetic profiles (Feldman, 2012; Lee, 2015; Phillips et al., French-speaking region. The data does not take into 2014), as well as because powerful economic interests are account the Italian- and the Romansh-speaking regions, likely to take advantage of citizens’ cognitive biases and representing 8% and less than 1% of the population respect- weak data protection legislation to access to personal ively (Deruelle et al., 2022). health data (Boyd and Hargittai, 2010; Brown, 2016). The main goal of the survey was to find out about how Previous research (Whiddett et al., 2006; Laurie, 2011; individuals are willing to share their health data. Caenazzo et al., 2015; Patil et al., 2016; Persaud and Therefore, we used a simple survey experiment. In the Bonham, 2018; Buhler et al., 2019; Trein and Wagner, survey, after briefly explaining what is meant by health 2021) has shown that individuals are not willing to have apps and genetic tests, we asked respondents how likely their personal information shared beyond the purpose of would it be for them to use health related apps (question clinical care and prefer to be consulted before their informa- A1) or to take a genetic test (question A2). After that we tion is released. Yet, recent studies demonstrate broad randomly framed individuals into two equal-sized groups public support for the use of health data for the purpose while controlling for the gender, age and language region of research (Garrison et al., 2016; Stockdale et al., 2018; distributions. One group (common good framing – CG) Braunack-Mayer et al., 2021). Furthermore, information received the following framing: “Some consider that per- about who benefits from data and the option to withdraw sonal health data is a common good and should be used access to data increases the likelihood of data sharing to improve public health.” Then we asked them how (Milne et al., 2021). likely would it be for them to use apps (question B1 )or In this study, we use a survey experiment to analyze take genetic tests (question B1 ) if their data were stored Swiss residents’ willingness to produce and store personal by public authorities, such as a biobank. The other group health data as the dependent variable. We probe the expect- (private good framing – PG) was provided with a different ation that privacy concerns are crucial to explain why indi- framing: “Some consider that personal data are private and viduals report that they are willing to take a genetic test or should be used exclusively to improve the health of the use connected technologies. We randomly expose indivi- individuals to whom they belong.” Again, this group was duals to different storage conditions which differ in the asked how likely it is they would use an app (question degree according to which respondents retain control over B2 ) or test (question B2 ) if they were to store their .1 .2 their health data. In other words, we framed the question health data privately, for example in a “datasafe” or on a so to indicate that genetic data is stored as public good secure server. for one group and private good for the other. Our main We illustrate the survey setup in Figure 1. research expectations are that private storage increases indi- Our analysis proceeds in two steps. For the first step, we viduals’ willingness, so does framing genetic data as private analyze the role of storage conditions and framing data as good, which our study confirms. common good or private good and whether citizens are more likely to use health apps and conduct a genetic test if they could store the data themselves or if it is stored by public authority. The experimental aspect of this research Method: Cross-sectional, survey allows us to present respondents with storage options that experiment are coherent with a normative view that articulates the The data set for our analysis was generated from an online nature of the data and what forms of data sharing are accept- survey that was fielded in March 2020 in Switzerland as able solutions. As such framing is coherent with storage part of a larger project on health policy and genetic data. options: there is no cross-comparison of framing vs. Switzerland provides a relevant context as the country has storage option possible. Ultimately the survey experiment Deruelle et al. 3 Figure 1. Operationalization of data sharing in the survey. is concerned with comparing response before and after issues than others (education) and finally whether there framing. This part of the analysis uses descriptive statistics are cultural disparities between the German- and French- of the different variables (see Figure 2). speaking respondents in Switzerland. For the second step, we conduct multivariate regression To operationalize political background variables, we ask analysis to examine whether the results of the framing participants whether it is the role of the state to provide remains statistically significant when controlling for alterna- social security, to reduce economic inequalities, to store tive explanations: variables based on the social background and use data as well as to regulate the storage and sharing of respondents as well as their political background and of data. Respondents’ views on the role of the state in their health history. This empirical strategy allows to test health and welfare indicate whether individuals are support- for the robustness of our findings. Our socioeconomic vari- ive of state intervention in these areas. The first two back- ables include the gender, age, citizenship, education level, ground variables are respondents’ views on whether it is professional status, subjective wealth, marital status, lan- the role of the state to provide social security or to reduce guage region, and subjective health. The goal is to investi- economic inequalities. These two variables are useful to gate whether there are disparities between those who paint big strokes on the role of the state: if individuals are anticipate potential health problems in the near future and largely in favour of state involvement in these aspects but those who do not (age variable), those who can afford to adverse to have their data stored by the state, this would invest in alternative storage solutions that match their pre- indicate that this is more likely due to privacy concerns ferences and those who do not (wealth variable), those rather than political views on the role of the state. If indivi- who have a better understanding of privacy and storage duals are adverse to both, it would mitigate our findings and 4 Big Data & Society Figure 2. Levels of willingness to use apps and make tests along the framing. indicate that political views are significant. We also analyzed do genetic tests (49.7%) than to use apps (70.2%). two other background variables: the respondents’ views on Further, we find that those who received the information whether it is the role of the state to store and use data and that data might be stored with the state (CG framing) are their views on whether it is the role of the state to regulate less likely to use apps (45.6%) and even less likely to storage and sharing data. In contrast to the first two variables, conduct genetic tests (39.2%) compared to those who received if respondents say that it is indeed the role of the state and are the information that data should be stored with the individual willing to have their data stored by the state, this would indi- (PG framing). In the latter case, the willingness to use is cate that political views and personal choice are coherent in slightly lower for apps (66.8%) and slightly higher for tests explaining individuals’ willingness to use personalized (51.4%) when compared to the situation before framing. health technology. In Table 1, we illustrate the respondents’ willingness to Finally, focusing on health specifically we look into indi- use health-related apps or to participate in genetic testing along viduals’ relationships with their own health. We survey the different control variables over the different framing setups. their self-assessed health conditions and whether respon- Again, and across the variables, we observe that individuals are dents have family antecedents of cancer. The role of those much less likely to do genetic tests than to use apps. For variables is to identify whether health-conscientious indivi- example, both men and women are similarly likely to use duals are more likely to want to monitor their health using apps (70.6 respectively 69.8%) but for both genders, the apps or genetic tests. willingness to use tests is lower (51.2 respectively 48.2%). Concerning political factors, we transformed the vari- ables that measure the individual’s level of usage in the Results four statements into binary measures (yes/no). In the ori- ginal survey, the variables measure the statements on a five- For each question laid out in Figure 1, our statistics report point Likert scale, from “do not agree” to “fully agree”, the share (in percent) of those who responded “likely” or allowing for an “undecided” category. To simplify the pres- “very likely.” Our first results are depicted in Figure 2 entation of the results and our econometric analysis, we and illustrate the impact of the framing. The three panels have coded both disagreement categories as well as the of the graph show the share of respondents that are “undecided” category as a “no” since we aim to distinguish willing to use apps and make tests, initially (before actual willingness. Indeed, we observe that the political framing), and in both the common and private good fram- factors seem to make a difference regarding the willingness ings (using a subsample of observations). In the second to use apps and conduct tests. Those who feel that it is the and third panels, we report the willingness to use apps role of the state to store and use data are more likely to use and tests for the subsample before framing (results are in apps and tests regardless of the framing. Finally, let us note line with those of the whole sample given in the first that we have also coded the variable regarding subjective panel) before indicating the results using the framing. First, we observe that individuals are much less likely to health, combining the categories “bad” and “very bad” as Deruelle et al. 5 Table 1. Levels of willingness to use apps and make tests. Before framing CG framing PG framing Before framing CG framing PG framing N (share) A A B1.1 B1.2 B2.1 B2.2 N (share) A A B1.1 B1.2 B2.1 B2.2 1 2 1 2 Socioeconomic factors Language region Gender French 330 (33.0) 72.4 53.9 48.5 41.8 67.3 55.8 Male 500 (50.0) 70.6 51.2 44.8 40.4 69.2 56.4 German 670 (67.0) 69.1 47.6 44.2 37.9 66.6 49.3 Female 500 (50.0) 69.8 48.2 46.4 38.0 64.4 46.4 Subjective health Age Bad 75 (7.5) 60.0 52.0 47.1 44.1 56.1 48.8 25–34 250 (25.0) 78.4 51.2 41.1 35.5 69.8 52.4 Fair 337 (33.7) 71.5 48.4 46.6 38.0 64.4 49.4 35–44 250 (25.0) 72.4 54.8 46.0 40.5 74.2 60.5 Good 588 (58.8) 70.7 50.2 44.9 39.3 69.8 53.0 45–54 250 (25.0) 67.2 47.6 52.4 43.7 66.1 49.2 Cancer in family history 55–65 250 (25.0) 62.8 45.2 42.7 37.1 57.1 43.7 No 540 (54.0) 69.4 47.4 47.4 40.4 66.7 49.6 Swiss nationality Yes 460 (46.0) 71.1 52.4 43.5 37.8 67.0 53.5 No 255 (25.5) 75.3 58.4 51.3 47.9 73.9 60.9 Political factors Yes 745 (74.5) 68.5 46.7 43.9 36.6 64.1 47.8 Role of the state Higher education Provide social security No 613 (61.3) 66.2 47.3 45.5 37.8 63.8 52.8 No 386 (38.6) 68.7 50.5 42.2 37.2 69.5 52.9 Yes 387 (38.7) 76.5 53.5 45.7 41.5 71.4 49.2 Yes 614 (61.4) 71.2 49.2 47.8 40.5 65.2 50.5 Professional status Reduce economic inequalities Full-time 516 (51.6) 72.9 51.4 49.0 41.7 68.5 53.3 No 386 (38.6) 70.2 48.4 37.8 31.9 66.7 51.7 Part-time 278 (27.8) 69.1 42.1 43.7 34.1 61.5 37.8 Yes 614 (61.4) 70.2 50.5 50.2 43.5 66.9 51.2 Other 206 (20.6) 65.0 55.8 39.6 39.6 70.0 66.0 Store and use data Subjective wealth No 783 (78.3) 67.2 45.7 41.8 33.8 62.8 46.3 Below avg. 589 (58.9) 68.3 51.8 43.4 38.5 61.4 50.2 Yes 217 (21.7) 81.1 64.1 58.9 58.0 81.9 70.5 Above avg. 411 (41.1) 73.0 46.7 48.6 40.2 75.1 53.3 Regulate storage and sharing of data Marital status No 533 (53.3) 67.2 47.8 40.2 34.6 64.5 49.8 Married 478 (47.8) 71.5 48.3 48.5 36.9 62.9 47.3 Yes 467 (46.7) 73.7 51.8 51.2 43.9 69.7 53.4 Other 522 (52.2) 69.0 51.0 42.9 41.3 70.3 55.1 The abbreviations “CG” and “PG” stand for common good respectively private good. The column “N” denotes the number of respondents. The share of respondents and the level of agreement (share of answers “likely” and “very likely”) in each question are expressed in percent. Results for the CG and PG framing are based on a total N= 500 observations each. 6 Big Data & Society Table 2. Regression results for the apps and tests usage. Apps and wearable devices Blood and genetic tests (1) Question A (2) Question B (3) Question A (4) Question B 1 x.1 2 x.2 Coeff. Marg. eff. Sig. Coeff. Marg. eff. Sig. Coeff. Marg. eff. Sig. Coeff. Marg. eff. Sig. Constant 0.353 −0.483 0.401 −0.134 Gender Female 0.081 +1.99% 0.022 +0.57% −0.068 −1.66% −0.196 −4.88% Age (baseline: 35 – 44) 25 – 34 0.309 +7.27% −0.151 −3.44% −0.101 −2.44% −0.157 −3.93% 45 – 54 −0.244 −5.97% 0.038 +0.96% −0.295 −7.24% −0.105 −2.65% 55 – 65 −0.365 −9.01% −0.345 −7.68% −0.332 −8.17% −0.349 −8.55% Swiss nationality ∗ ∗ Yes −0.180 −4.38% −0.257 −5.80% −0.370 −9.14% −0.394 −9.60% Higher education ∗ ∗ Yes 0.386 + 8.99% 0.009 + 0.26% 0.308 + 7.11% −0.070 −1.78% Professional status (baseline: Full time) Part time −0.166 −4.04% −0.225 −5.10% −0.379 −9.35% ∗−0.422 −10.26% ∗ Other −0.286 −7.03% −0.086 −1.96% 0.146 +3.45% 0.251 +6.22% Subjective wealth above average Yes 0.139 +3.37% 0.424 +10.43% ∗∗ −0.303 −7.44% ∗ 0.111 +2.72% Married or partnership Yes −0.041 −0.97% −0.016 −0.33% 0.154 +3.63% 0.220 +5.45% German-speaking region Yes −0.114 −2.76% 0.003 +0.12% −0.095 −2.30% −0.050 −1.30% Subjective health (baseline: Bad) Fair 0.472 +10.83% 0.209 +5.09% −0.042 −1.03% 0.046 +1.10% Good 0.379 +8.83% 0.197 +4.81% 0.068 +1.61% 0.121 +2.98% Cancer in family history Yes 0.147 +3.54% −0.065 −1.46% 0.285 +6.60% ∗ 0.073 +1.77% Role of the state (baseline: No) Provide social security 0.160 +3.86% −0.112 −2.55% −0.172 −4.20% −0.142 −3.56% Reduce economic −0.132 −3.20% 0.270 +6.60% 0.069 +1.63% 0.221 +5.48% inequalities Store and use data 0.747 +16.32% ∗∗∗ 0.768 +19.00% ∗∗∗ 0.710 +15.34% ∗∗∗ 0.973 +23.12% ∗∗∗ Regulate storage and 0.033 +0.83% 0.094 +2.30% −0.073 −1.78% −0.036 −0.95% sharing of data Private good framing 0.963 +23.67% ∗∗∗ 0.558 +13.74% ∗∗∗ ∗ ∗∗ ∗∗∗ All models are based on N= 1 000 observations. Significance levels: p< 0.05; p< 0.01; p< 0.001. well as “good” and “very good” into “bad” and “good” robustness tests, we also estimated variables, which does respectively. Further, we have changed the education not yield different results. These findings are available level variable into a binary variable classifying individuals upon request from the authors. into those who at least have a high school diploma and We estimate two models regarding the likelihood to use those who have not. apps and two models concerning the usage of tests. In both To understand deeper how the usage of apps and tests is cases, we estimate one reference model with the control linked to the different explanatory variables, we estimated variables only. Model 1 and 3 use as dependent variables different regression models. Specifically, we built four the questions A1 respectively A2 from Figure 1 (Table 2). regression models using the binary answer variable (see Models 2 and 4 use the questions questions B1 /B1 .1 .2 above) for the likelihood to use an app or to conduct a respectively B2 /B2 from Figure 1 as dependent vari- .1 .2 test. We use logit models because they yield the lowest ables (Table 2). In addition, we add to Models 2 and 4 an Akaike Information Criterion (AIC) values compared to additional binary variable measuring PG (1) vs. CG models with a probit or a cloglog link function. As framing (0). In addition to the regression coefficients, Deruelle et al. 7 Table 2 also shows the marginal effects of variable, since (something that we did not measure) and therefore are the regression coefficients do not allow us to determine willing to share their data from health tests with the the size of the effects. authorities. The main result of the analysis shows that those who Other political factors play a less important role: notably, received the private good framing are more likely to use those who support statements such as “it is the role of the health apps and do genetic tests compared to those who state to provide social security” and “it is the role of the were exposed to the common good framing. In the second state to reduce economic inequalities” are not more or set of regression models (bottom line in models 2 and 4 less likely to use health apps and provide genetic tests. in Table 2), a binary variable indicates if the respondent Although the regression coefficients have an either positive received the private good framing instead of the baseline or negative direction, their statistical significance is weak, common good framing (cf. Figure 1). The results show and they point in different directions for two similar vari- that such a framing significantly increases the likelihood ables. This finding indicates that using health technologies to use apps and wearable devices by almost 24% and is not very strongly related to political ideologies, such as makes it almost 14% more likely that individuals do the left and right political positions (Jensen, 2011). blood and genetic tests. Overall, our findings underline the importance of data The analysis shows also that very few control variables privacy when it comes to implementing practices of perso- are statistically significant (see Table 2). For example, the nalized health. Maintaining privacy over personal health items measuring the respondents’ medical history do not data storage important when it comes to using health-related seem to co-vary with the willingness to do tests, especially apps and tests: more so than beliefs about subjective health when we control for the private good framing variable. and a family history of cancer. Among the socio-economic variables yielding signifi- cant results, the following points are relevant. Regarding Conclusions the use of apps, those reporting a subjective wealth above average are more likely to use apps once the framing is As per research expectations, this study shows that indivi- applied. However, as shown in Table 1, this is due mostly duals are more likely to use personalized health technolo- to the relationship between subjective wealth and app use gies if their data are kept privately, rather than a common in the private group framing. Indeed, while before storage. This shows that privacy concerns are still import- framing 73% of the respondents with a wealth above ant, which is in line with findings from the literature average would use apps, they are only 48.6% once the (Ostherr et al., 2017; Vayena et al., 2018; Jacobs and public good framing is applied. This however rises up to Popma, 2019). Other explanatory variables have a less 75% for the private good group. This shows a meaningful clear effect, but this finding is not statistically significant. relationship between wealth and connected device usage, For instance, respondents with a degree in higher education but only when individuals are presented with the private are more likely to use those technologies. However, this good framing – thus confirming the central result of this increased likelihood disappears once we apply the framing. study. Moreover, those reporting to have a higher level of Nevertheless, ideological views should not be completely education are almost 9% more likely to use apps, but statis- eschewed: indeed, those who believe that it is the role of tically significance is lower. Regarding the use of genetic the state to collect and store data are more likely to use per- tests, the reference model indicates that respondents with sonalized health technologies. at least a high-school diploma and a family history of There are important differences between health apps and cancer are more likely to engage in tests compared to genetic tests. The likelihood for respondents to self-perform those who have a lower level of education and no cancer a direct-to-customer test is, regardless of the framing, lower history in their family. Furthermore, Swiss citizens and than for health apps. It is also lower throughout all back- those who only work part-time are less likely to take such ground variables. Previous findings on health data in tests. Switzerland did not differentiate between means to produce Among political factors, we see that respondents who health data (Pletscher et al., 2022). Genetic data is particularly feel that it is the role of the state to store and use data are sensitive a form of health data and individuals’ willingness to more than 16% more likely to report app usage and 15% even elect to do a test may be mitigated by their knowledge more likely to conduct genetic tests. These results increase and their worldview on genetic information and its use even further if we control for the effect of framing. This (Bearth and Siegrist, 2020; Claytonetal.,2018).Yet,attitudes finding is not surprising because it implies that those who towards genetic testing may not only be rooted in ethical con- are anyways more willing to give away their data to siderations. Middleton et al. (2020) for instance show that in public authorities are more likely to conduct tests, inde- Switzerland most people are unfamiliar with genetic knowl- pendently whether they received a CG or PG framing. edge and do not differentiate those from other medical infor- One possible interpretation of this result is that these mation. Willingness may also depend on costs (Ries et al., respondents have higher levels of trust in the government 2010). 8 Big Data & Society While respondents are more likely to produce and store for PH data outside of the medical context. And thus, it data in the private good group, our survey experiment would be relevant for future research to further analyze shows that willingness to produce and store data is lower how privacy and storage may affect whether individuals once the framing is applied, regardless of the framing that accept to share their data. is applied. This is important for practitioners, as it shows somewhat of a paradox: as soon as respondents are contem- Declaration of conflicting interests plating that the data can be shared, they are less likely to The author(s) declared no potential conflicts of interest with want to produce it, yet, producing this data would respect to the research, authorship, and/or publication of this improve the public good. Ultimately our results show the article. necessity to develop literacy on health data, digital platform and privacy. Potentially, this could include improving sci- Funding entific literacy and educating on the personal and collective benefits of producing this data; however, our results show The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: that respondents from the common good framing group This work was supported by the Schweizerischer Nationalfonds were the least likely to produce and store data. More gener- zur Förderung der Wissenschaftlichen Forschung, (grant number ally, our findings show also that personal health data is no CRSII5 180350). different than other forms of personal data: mitigating data privacy concerns is of the utmost importance for a correct implementation of personalized health objectives. ORCID iDs Therefore, in order to promote the production of persona- Thibaud Deruelle https://orcid.org/0000-0002-3722-0467 lized health, decision makers should create storage solu- Philipp Trein https://orcid.org/0000-0001-6217-6675 tions that could emulate, for instance, the Swiss Joël Wagner https://orcid.org/0000-0002-3712-5494 Electronic Health Record (Baumann et al., 2018; De Pietro and Francetic, 2018; Platt and Kardia, 2015) that References allow citizens to give consent for the use of their data for Allen LN and Christie GP (2016) The emergence of personalized the purpose of research. health technology. Journal of Medical Internet Research 18(5): Like any other research, our study comes with limita- e99. tions. We lay out three limitations. The first one being the Barazzetti G, Bühler N, Audétat M, et al. (2021) Making persona- generalization of our findings. Switzerland is arguably at lized medicine ethical: A critical examination of the new pro- the forefront of the personalized health “turn,” and thus mises of ‘personalized health’ in Switzerland. Science and the debate regarding data privacy is particularly relevant, Public Policy 48(6): 818–828. unlike in other national contexts in Europe or worldwide. Baumann LA, Baker J and Elshaug AG (2018) The impact of elec- Second, the scope of our empirical analysis is limited as tronic health record systems on clinical documentation times: A systematic review. Health Policy 122(8): 827–836. we could not include the Italian-speaking population of Bearth A and Siegrist M (2020) Psychological factors that deter- Switzerland. Finally, our study is limited to personalized mine people’s willingness to share genetic data for research. health technologies (smartphone apps, wearable devices, Clinical Genetics 97(3): 483–491. do-it-yourself tests) where data generation is depending Blasimme A, Vayena E and Van Hoyweghen I (2019) Big data, on the patients’ willingness to use them. Our results are precision medicine and private insurance: a delicate balan- therefore not relevant to personal health data generated as cing act. Big Data & Society 6(1): 1–6. DOI: 10.1177/ a result of a medical intervention. This last limitation may inform the direction for future Boyd D and Hargittai E (2010) Facebook privacy settings: Who research. Further analysis of individuals’ preferences can cares? First Monday 15(8). Epub ahead of print. DOI: 10.5210/ be done by comparing the medical and the non-medical fm.v15i8.3086 dimension of PH data generation. In the same vein, more Braunack-Mayer A, Fabrianesi B, Street J, et al. (2021) Sharing government health data with the private sector: Community storage and privacy options can be explored Considering attitudes survey. Journal of Medical Internet Research that respondents’ willingness to use health technologies is 23(10): e24200. DOI: 10.2196/24200. significantly higher when the state is not involved. Future Brown I (2016) The economics of privacy, data protection and sur- research could explore which private third-party citizens veillance. In: Bauer JM and Latzer M (eds) Handbook on the trust to handle their data, whereas private insurance, Economics of the Internet. Cheltenham: Edward Elgar digital technology firms or cooperative biobanks. On the Publishing, pp.247–261. specificity of genetic testing, future research should Bühler G, Hermann M and Lambertus M (2019) Observatoire explore whether ethical and cost-related considerations “société numérique et solidarité” opinion et comportement matter in the Swiss case or whether privacy concerns de la population suisse. Zürich: Sotomo/Fondation Sanitas. trump it all as one could expect from our findings. Caenazzo L, Tozzo P and Borovecki A (2015) Ethical governance Finally, our results can inform how to design privacy in biobanks linked to electronic health records. European Deruelle et al. 9 Review for Medical and Pharmacological Sciences 19(21): Nakagomi H, Sakamoto I, Hirotsu Y, et al. (2016) Willingness of 4182–4186. Japanese patients with breast cancer to have genetic testing of Clayton EW, Halverson CM, Sathe NA, et al. (2018) A systematic BRCA without burden of expenses. Breast Cancer (Tokyo, literature review of individuals’ perspectives on privacy and Japan) 23(4): 649–653. genetic information in the United States. PloS one 13(10): Okma KGH and Crivelli L (2013) Swiss And Dutch “consumer- e0204417. DOI: 10.1371/journal.pone.0204417. driven health care”: Ideal model or reality? Health Policy DePietro C, Camenzind P, Sturny I, et al. (2015) Health systems in 109(2): 105–112. transition, Switzerland. Health System Rev 17(4): 1–323. Ostherr K, Borodina S, Bracken RC, et al. (2017) Trust and De Pietro C and Francetic I (2018) E-health in Switzerland: The privacy in the context of user-generated health data. Big laborious adoption of the federal law on electronic health Data & Society 4(1): 1–11. DOI: 10.1177/2053951717704673. records (EHR) and health information exchange (HIE) net- Patil SH, Lu H, Saunders CL, et al. (2016) Public preferences for works. Health Policy 122(2): 69–74. electronic health data storage, access, and sharing — evidence Deruelle T, Kalouguina V, Trein P, et al. (2022) Is there a “pan- from a pan-European survey. Journal of the American Medical demic effect” on individuals’ willingness to take genetic Informatics Association 23(6): 1096–1106. tests? European Journal of Human Genetics. Epub ahead of Persaud A and Bonham VL (2018) The role of the health care pro- print. DOI: 10.1038/s41431-022-01223-6. vider in building trust between patients and precision medicine Feldman EA (2012) The genetic information nondiscrimination research programs. The American Journal of Bioethics 18(4): act (gina): Public policy and medical practice in the age of per- 26–28. sonalized medicine. Journal of General Internal Medicine Phillips KA, Deverka PA, Hooker GW, et al. (2018) Genetic test 27(6): 743–746. availability and spending: Where are we now? Where are we Garrison NA, Sathe NA, Antommaria AHM, et al. (2016) A sys- going? Health Affairs 37(5): 710–716. tematic literature review of individuals’ perspectives on broad Phillips KA, Trosman JR, Kelley RK, et al. (2014) Genomic consent and data sharing in the United States. Genetics in sequencing: Assessing the health care system, policy, and Medicine 18(7): 663–671. big-data implications. Health Affairs 33(7): 1246–1253. Jacobs B and Popma J (2019) Medical research, big data and the Platt J and Kardia S (2015) Public trust in health information need for privacy by design. Big Data & Society 6(1): sharing: Implications for biobanking and electronic health 205395171882435. record systems. Journal of Personalized Medicine 5(1): 3–21. Jensen C (2011) Marketization via compensation: Health care and Pletscher F, Mändli Lerch K and Glinz D (2022) Willingness to share the politics of the right in advanced industrialized nations. anonymised routinely collected clinical health data in Switzerland: British Journal of Political Science 41(4): 907–926. a cross-sectional survey. Swiss Medical Weekly 152(23): w30182. Khoury MJ, Iademarco MF and Riley WT (2016) Precision public DOI: 10.4414/smw.2022.w30182. health for the era of precision medicine. American Journal of Ries NM, Hyde-Lay R and Caulfield T (2010) Willingness to pay Preventive Medicine 50(3): 398–401. for genetic testing: A study of attitudes in a Canadian popula- Laurie G (2011) Reflexive governance in biobanking: On the tion. Public Health Genomics 13(5): 292–300. value of policy led approaches and the need to recognise the Seifert A, Christen M and Martin M (2018) Willingness of older limits of law. Human Genetics 130(3): 347–356. adults to share mobile health data with researchers. GeroPsych: Lee SSJ (2015) The biobank as political artifact: The struggle over race The Journal of Gerontopsychology and Geriatric Psychiatry in categorizing genetic difference. The ANNALS of the American 31(1): 41–49. Academy of Political and Social Science 661(1): 143–159. Seifert A and Vandelanotte C (2021) The use of wearables and Leonard NR, Silverman M, Sherpa DP, et al. (2017) Mobile health health apps and the willingness to share self-collected data technology using a wearable sensorband for female college students among older adults. Aging and Health Research 1(3): with problem drinking: an acceptability and feasibility study. JMIR 100032. DOI: 10.1016/j.ahr.2021.100032. mHealth and uHealth 5(7): e90. DOI: 10.2196/mhealth.7399. Stockdale J, Cassell J and Ford E (2018) “giving something back”:A McCartney CI, TJ W and Williams R (2011) Transnational exchange systematic review and ethical enquiry into public views on the use of forensic dna: Viability, legitimacy, and acceptability. European of patient data for research in the United Kingdom and the repub- Journal on Criminal Policy and Research 17(4): 305–322. lic of Ireland. Wellcome Open Research 3(6). Epub ahead of print. Middleton A, Milne R, Almarri MA, et al. (2020) Global public DOI: 10.12688/wellcomeopenres.13531.2. perceptions of genomic data sharing: What shapes the willing- Trein P and Wagner J (2021) Governing personalized health: A ness to donate dna and health data? The American Journal of scoping review. Frontiers in Genetics 12: 650504. DOI: 10. Human Genetics 107(4): 743–752. 3389/fgene.2021.650504. Milne R, Morley KI, Almarri MA, et al. (2021) Demonstrating trust- Vayena E, Haeusermann T, Adjekum A, et al. (2018) Digital worthiness when collecting and sharing genomic data: Public health: Meeting the ethical and policy challenges. Swiss views across 22 countries. Genome Medicine 13(1): 1–12. Medical Weekly 148: w14571. Minvielle E, Waelli M, Sicotte C, et al. (2014) Managing custom- Whiddett R, Hunter I, Engelbrecht J, et al. (2006) Patients’ atti- ization in health care: A framework derived from the services tudes towards sharing their health information. International sector literature. Health Policy 117(2): 216–227. Journal of Medical Informatics 75(7): 530–541. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Big Data & Society SAGE

Designing privacy in personalized health: An empirical analysis

Loading next page...
 
/lp/sage/designing-privacy-in-personalized-health-an-empirical-analysis-3088cLChX5

References (49)

Publisher
SAGE
Copyright
© The Author(s) 2023
ISSN
2053-9517
eISSN
2053-9517
DOI
10.1177/20539517231158636
Publisher site
See Article on Publisher Site

Abstract

A crucial challenge for personalized health is the handling of individuals’ data and specifically the protection of their priv- acy. Secure storage of personal health data is of paramount importance to convince citizens to collect personal health data. In this survey experiment, we test individuals’ willingness to produce and store personal health data, based on dif- ferent storage options and whether this data is presented as common good or private good. In this paper, we focus on the nonmedical context with two means to self-produce data: connected devices that record physical activity and genetic tests that appraise risks of diseases. We use data from a survey experiment fielded in Switzerland in March 2020 and perform regression analyses on a representative sample of Swiss citizens in the French- and German-speaking cantons. Our analysis shows that respondents are more likely to use both apps and tests when their data is framed as a private good to be stored by individuals themselves. Our results demonstrate that concerns regarding the privacy of personal heath data storage trumps any other variable when it comes to the willingness to use personalized health technologies. Individuals prefer a data storage format where they retain control over the data. Ultimately, this study presents results susceptible to inform decision-makers in designing privacy in personalized health initiatives. Keywords personalized health, health apps, genetic tests, data storage (Ries et al., 2010). In this study, we look beyond precision Introduction medicine focused on sick patients and zoom-in on the Personalized health is based on the massive integration of prevention-oriented dimension of personalized health biomedical and social data into research, to determine how (Khoury et al., 2016). individuals’ physical and social environments, genetic We ask: what are individuals’ preferences regarding how endowments and behaviors influence their health (Barazzetti they want to store the personal health data they opt to et al., 2021). This data helps to customize preventive and produce? We analyze two voluntary methods of data gener- therapeutic interventions to the individual genetic and clinical ation which cover the non-medical dimensions of persona- characteristics of each patient (Minvielle et al., 2014). lized health: self-performed genetic tests to send to a Personal health data can cover a wide range of forms of laboratory (Phillips et al., 2018) that can be used to create data, from diagnoses recorded by physician in a patient’s personalized plans for health prevention and connected record to simple metrics about the individual, for instance technologies that monitor physical activity (Allen and their weight. A large amount of personal health data is pro- Christie, 2016) such as health trackers and apps. A central duced in the medical context and often involves that issue regarding any new technology is acceptability medical institutions have access to this data, which raises (McCartney et al., 2011; Leonard et al., 2017). To privacy concerns (Baumann et al., 2018; De Pietro and Francetic, 2018; Platt and Kardia, 2015). However, a host of health data may be created and stored outside of the Department of Actuarial Science, University of Lausanne, Switzerland medical context: individuals can self-produce data via Institute of Political Studies, University of Lausanne, Switzerland apps that record, for instance physical activity (Seifert Corresponding author: et al., 2018; Seifert and Van-delanotte, 2021) or elect to Thibaud Deruelle, Political Science and International Relations Department, undergo blood or genetic tests to appraise their risks to University of Geneva, Boulevard du Pont-d’Arve 40 1205 Genève. develop a cancer (Nakagomi et al., 2016) or other diseases Email: thibaud.deruelle@unige.ch Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License (https:// creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access page (https://us.sagepub.com/en-us/nam/open-access-at-sage). 2 Big Data & Society successfully apply new technologies and get citizens to use a health care system, which puts the responsibility for them, we need to know under which conditions such usage health care on individuals, for example regarding becomes more likely. In the context of personalized health, co-payments for treatments (DePietro et al., 2015; De the usage of such technologies is also important for the Pietro and Francetic, 2018). In other words, Switzerland common good, because the availability of individuals’ is a “consumerdriven” healthcare system (Okma and health data might not only improve their personal health, Crivelli, 2013) and as such, citizens’ willingness to use per- but can in addition provide the basis for research aiming sonalized health technologies is highly relevant. at providing new diagnostic and treatments. The data in this paper consists of a sample (N= 1000) of A crucial challenge for the future of personalized health the Swiss population, which is representative according to is thus the handling of individuals’ data and specifically the the following categories: men and women are distributed protection of their privacy (Ostherr et al., 2017; Vayena equally and the participants, aged between 25 and 65 et al., 2018; Blasimme et al., 2019). For instance, privacy years, are evenly distributed into four age groups. Two protection is important because personalized health bears thirds (67%) of the sample is comprised of Swiss the risk of discrimination among individuals based on Germans and the remaining 33% are from Switzerland’s genetic profiles (Feldman, 2012; Lee, 2015; Phillips et al., French-speaking region. The data does not take into 2014), as well as because powerful economic interests are account the Italian- and the Romansh-speaking regions, likely to take advantage of citizens’ cognitive biases and representing 8% and less than 1% of the population respect- weak data protection legislation to access to personal ively (Deruelle et al., 2022). health data (Boyd and Hargittai, 2010; Brown, 2016). The main goal of the survey was to find out about how Previous research (Whiddett et al., 2006; Laurie, 2011; individuals are willing to share their health data. Caenazzo et al., 2015; Patil et al., 2016; Persaud and Therefore, we used a simple survey experiment. In the Bonham, 2018; Buhler et al., 2019; Trein and Wagner, survey, after briefly explaining what is meant by health 2021) has shown that individuals are not willing to have apps and genetic tests, we asked respondents how likely their personal information shared beyond the purpose of would it be for them to use health related apps (question clinical care and prefer to be consulted before their informa- A1) or to take a genetic test (question A2). After that we tion is released. Yet, recent studies demonstrate broad randomly framed individuals into two equal-sized groups public support for the use of health data for the purpose while controlling for the gender, age and language region of research (Garrison et al., 2016; Stockdale et al., 2018; distributions. One group (common good framing – CG) Braunack-Mayer et al., 2021). Furthermore, information received the following framing: “Some consider that per- about who benefits from data and the option to withdraw sonal health data is a common good and should be used access to data increases the likelihood of data sharing to improve public health.” Then we asked them how (Milne et al., 2021). likely would it be for them to use apps (question B1 )or In this study, we use a survey experiment to analyze take genetic tests (question B1 ) if their data were stored Swiss residents’ willingness to produce and store personal by public authorities, such as a biobank. The other group health data as the dependent variable. We probe the expect- (private good framing – PG) was provided with a different ation that privacy concerns are crucial to explain why indi- framing: “Some consider that personal data are private and viduals report that they are willing to take a genetic test or should be used exclusively to improve the health of the use connected technologies. We randomly expose indivi- individuals to whom they belong.” Again, this group was duals to different storage conditions which differ in the asked how likely it is they would use an app (question degree according to which respondents retain control over B2 ) or test (question B2 ) if they were to store their .1 .2 their health data. In other words, we framed the question health data privately, for example in a “datasafe” or on a so to indicate that genetic data is stored as public good secure server. for one group and private good for the other. Our main We illustrate the survey setup in Figure 1. research expectations are that private storage increases indi- Our analysis proceeds in two steps. For the first step, we viduals’ willingness, so does framing genetic data as private analyze the role of storage conditions and framing data as good, which our study confirms. common good or private good and whether citizens are more likely to use health apps and conduct a genetic test if they could store the data themselves or if it is stored by public authority. The experimental aspect of this research Method: Cross-sectional, survey allows us to present respondents with storage options that experiment are coherent with a normative view that articulates the The data set for our analysis was generated from an online nature of the data and what forms of data sharing are accept- survey that was fielded in March 2020 in Switzerland as able solutions. As such framing is coherent with storage part of a larger project on health policy and genetic data. options: there is no cross-comparison of framing vs. Switzerland provides a relevant context as the country has storage option possible. Ultimately the survey experiment Deruelle et al. 3 Figure 1. Operationalization of data sharing in the survey. is concerned with comparing response before and after issues than others (education) and finally whether there framing. This part of the analysis uses descriptive statistics are cultural disparities between the German- and French- of the different variables (see Figure 2). speaking respondents in Switzerland. For the second step, we conduct multivariate regression To operationalize political background variables, we ask analysis to examine whether the results of the framing participants whether it is the role of the state to provide remains statistically significant when controlling for alterna- social security, to reduce economic inequalities, to store tive explanations: variables based on the social background and use data as well as to regulate the storage and sharing of respondents as well as their political background and of data. Respondents’ views on the role of the state in their health history. This empirical strategy allows to test health and welfare indicate whether individuals are support- for the robustness of our findings. Our socioeconomic vari- ive of state intervention in these areas. The first two back- ables include the gender, age, citizenship, education level, ground variables are respondents’ views on whether it is professional status, subjective wealth, marital status, lan- the role of the state to provide social security or to reduce guage region, and subjective health. The goal is to investi- economic inequalities. These two variables are useful to gate whether there are disparities between those who paint big strokes on the role of the state: if individuals are anticipate potential health problems in the near future and largely in favour of state involvement in these aspects but those who do not (age variable), those who can afford to adverse to have their data stored by the state, this would invest in alternative storage solutions that match their pre- indicate that this is more likely due to privacy concerns ferences and those who do not (wealth variable), those rather than political views on the role of the state. If indivi- who have a better understanding of privacy and storage duals are adverse to both, it would mitigate our findings and 4 Big Data & Society Figure 2. Levels of willingness to use apps and make tests along the framing. indicate that political views are significant. We also analyzed do genetic tests (49.7%) than to use apps (70.2%). two other background variables: the respondents’ views on Further, we find that those who received the information whether it is the role of the state to store and use data and that data might be stored with the state (CG framing) are their views on whether it is the role of the state to regulate less likely to use apps (45.6%) and even less likely to storage and sharing data. In contrast to the first two variables, conduct genetic tests (39.2%) compared to those who received if respondents say that it is indeed the role of the state and are the information that data should be stored with the individual willing to have their data stored by the state, this would indi- (PG framing). In the latter case, the willingness to use is cate that political views and personal choice are coherent in slightly lower for apps (66.8%) and slightly higher for tests explaining individuals’ willingness to use personalized (51.4%) when compared to the situation before framing. health technology. In Table 1, we illustrate the respondents’ willingness to Finally, focusing on health specifically we look into indi- use health-related apps or to participate in genetic testing along viduals’ relationships with their own health. We survey the different control variables over the different framing setups. their self-assessed health conditions and whether respon- Again, and across the variables, we observe that individuals are dents have family antecedents of cancer. The role of those much less likely to do genetic tests than to use apps. For variables is to identify whether health-conscientious indivi- example, both men and women are similarly likely to use duals are more likely to want to monitor their health using apps (70.6 respectively 69.8%) but for both genders, the apps or genetic tests. willingness to use tests is lower (51.2 respectively 48.2%). Concerning political factors, we transformed the vari- ables that measure the individual’s level of usage in the Results four statements into binary measures (yes/no). In the ori- ginal survey, the variables measure the statements on a five- For each question laid out in Figure 1, our statistics report point Likert scale, from “do not agree” to “fully agree”, the share (in percent) of those who responded “likely” or allowing for an “undecided” category. To simplify the pres- “very likely.” Our first results are depicted in Figure 2 entation of the results and our econometric analysis, we and illustrate the impact of the framing. The three panels have coded both disagreement categories as well as the of the graph show the share of respondents that are “undecided” category as a “no” since we aim to distinguish willing to use apps and make tests, initially (before actual willingness. Indeed, we observe that the political framing), and in both the common and private good fram- factors seem to make a difference regarding the willingness ings (using a subsample of observations). In the second to use apps and conduct tests. Those who feel that it is the and third panels, we report the willingness to use apps role of the state to store and use data are more likely to use and tests for the subsample before framing (results are in apps and tests regardless of the framing. Finally, let us note line with those of the whole sample given in the first that we have also coded the variable regarding subjective panel) before indicating the results using the framing. First, we observe that individuals are much less likely to health, combining the categories “bad” and “very bad” as Deruelle et al. 5 Table 1. Levels of willingness to use apps and make tests. Before framing CG framing PG framing Before framing CG framing PG framing N (share) A A B1.1 B1.2 B2.1 B2.2 N (share) A A B1.1 B1.2 B2.1 B2.2 1 2 1 2 Socioeconomic factors Language region Gender French 330 (33.0) 72.4 53.9 48.5 41.8 67.3 55.8 Male 500 (50.0) 70.6 51.2 44.8 40.4 69.2 56.4 German 670 (67.0) 69.1 47.6 44.2 37.9 66.6 49.3 Female 500 (50.0) 69.8 48.2 46.4 38.0 64.4 46.4 Subjective health Age Bad 75 (7.5) 60.0 52.0 47.1 44.1 56.1 48.8 25–34 250 (25.0) 78.4 51.2 41.1 35.5 69.8 52.4 Fair 337 (33.7) 71.5 48.4 46.6 38.0 64.4 49.4 35–44 250 (25.0) 72.4 54.8 46.0 40.5 74.2 60.5 Good 588 (58.8) 70.7 50.2 44.9 39.3 69.8 53.0 45–54 250 (25.0) 67.2 47.6 52.4 43.7 66.1 49.2 Cancer in family history 55–65 250 (25.0) 62.8 45.2 42.7 37.1 57.1 43.7 No 540 (54.0) 69.4 47.4 47.4 40.4 66.7 49.6 Swiss nationality Yes 460 (46.0) 71.1 52.4 43.5 37.8 67.0 53.5 No 255 (25.5) 75.3 58.4 51.3 47.9 73.9 60.9 Political factors Yes 745 (74.5) 68.5 46.7 43.9 36.6 64.1 47.8 Role of the state Higher education Provide social security No 613 (61.3) 66.2 47.3 45.5 37.8 63.8 52.8 No 386 (38.6) 68.7 50.5 42.2 37.2 69.5 52.9 Yes 387 (38.7) 76.5 53.5 45.7 41.5 71.4 49.2 Yes 614 (61.4) 71.2 49.2 47.8 40.5 65.2 50.5 Professional status Reduce economic inequalities Full-time 516 (51.6) 72.9 51.4 49.0 41.7 68.5 53.3 No 386 (38.6) 70.2 48.4 37.8 31.9 66.7 51.7 Part-time 278 (27.8) 69.1 42.1 43.7 34.1 61.5 37.8 Yes 614 (61.4) 70.2 50.5 50.2 43.5 66.9 51.2 Other 206 (20.6) 65.0 55.8 39.6 39.6 70.0 66.0 Store and use data Subjective wealth No 783 (78.3) 67.2 45.7 41.8 33.8 62.8 46.3 Below avg. 589 (58.9) 68.3 51.8 43.4 38.5 61.4 50.2 Yes 217 (21.7) 81.1 64.1 58.9 58.0 81.9 70.5 Above avg. 411 (41.1) 73.0 46.7 48.6 40.2 75.1 53.3 Regulate storage and sharing of data Marital status No 533 (53.3) 67.2 47.8 40.2 34.6 64.5 49.8 Married 478 (47.8) 71.5 48.3 48.5 36.9 62.9 47.3 Yes 467 (46.7) 73.7 51.8 51.2 43.9 69.7 53.4 Other 522 (52.2) 69.0 51.0 42.9 41.3 70.3 55.1 The abbreviations “CG” and “PG” stand for common good respectively private good. The column “N” denotes the number of respondents. The share of respondents and the level of agreement (share of answers “likely” and “very likely”) in each question are expressed in percent. Results for the CG and PG framing are based on a total N= 500 observations each. 6 Big Data & Society Table 2. Regression results for the apps and tests usage. Apps and wearable devices Blood and genetic tests (1) Question A (2) Question B (3) Question A (4) Question B 1 x.1 2 x.2 Coeff. Marg. eff. Sig. Coeff. Marg. eff. Sig. Coeff. Marg. eff. Sig. Coeff. Marg. eff. Sig. Constant 0.353 −0.483 0.401 −0.134 Gender Female 0.081 +1.99% 0.022 +0.57% −0.068 −1.66% −0.196 −4.88% Age (baseline: 35 – 44) 25 – 34 0.309 +7.27% −0.151 −3.44% −0.101 −2.44% −0.157 −3.93% 45 – 54 −0.244 −5.97% 0.038 +0.96% −0.295 −7.24% −0.105 −2.65% 55 – 65 −0.365 −9.01% −0.345 −7.68% −0.332 −8.17% −0.349 −8.55% Swiss nationality ∗ ∗ Yes −0.180 −4.38% −0.257 −5.80% −0.370 −9.14% −0.394 −9.60% Higher education ∗ ∗ Yes 0.386 + 8.99% 0.009 + 0.26% 0.308 + 7.11% −0.070 −1.78% Professional status (baseline: Full time) Part time −0.166 −4.04% −0.225 −5.10% −0.379 −9.35% ∗−0.422 −10.26% ∗ Other −0.286 −7.03% −0.086 −1.96% 0.146 +3.45% 0.251 +6.22% Subjective wealth above average Yes 0.139 +3.37% 0.424 +10.43% ∗∗ −0.303 −7.44% ∗ 0.111 +2.72% Married or partnership Yes −0.041 −0.97% −0.016 −0.33% 0.154 +3.63% 0.220 +5.45% German-speaking region Yes −0.114 −2.76% 0.003 +0.12% −0.095 −2.30% −0.050 −1.30% Subjective health (baseline: Bad) Fair 0.472 +10.83% 0.209 +5.09% −0.042 −1.03% 0.046 +1.10% Good 0.379 +8.83% 0.197 +4.81% 0.068 +1.61% 0.121 +2.98% Cancer in family history Yes 0.147 +3.54% −0.065 −1.46% 0.285 +6.60% ∗ 0.073 +1.77% Role of the state (baseline: No) Provide social security 0.160 +3.86% −0.112 −2.55% −0.172 −4.20% −0.142 −3.56% Reduce economic −0.132 −3.20% 0.270 +6.60% 0.069 +1.63% 0.221 +5.48% inequalities Store and use data 0.747 +16.32% ∗∗∗ 0.768 +19.00% ∗∗∗ 0.710 +15.34% ∗∗∗ 0.973 +23.12% ∗∗∗ Regulate storage and 0.033 +0.83% 0.094 +2.30% −0.073 −1.78% −0.036 −0.95% sharing of data Private good framing 0.963 +23.67% ∗∗∗ 0.558 +13.74% ∗∗∗ ∗ ∗∗ ∗∗∗ All models are based on N= 1 000 observations. Significance levels: p< 0.05; p< 0.01; p< 0.001. well as “good” and “very good” into “bad” and “good” robustness tests, we also estimated variables, which does respectively. Further, we have changed the education not yield different results. These findings are available level variable into a binary variable classifying individuals upon request from the authors. into those who at least have a high school diploma and We estimate two models regarding the likelihood to use those who have not. apps and two models concerning the usage of tests. In both To understand deeper how the usage of apps and tests is cases, we estimate one reference model with the control linked to the different explanatory variables, we estimated variables only. Model 1 and 3 use as dependent variables different regression models. Specifically, we built four the questions A1 respectively A2 from Figure 1 (Table 2). regression models using the binary answer variable (see Models 2 and 4 use the questions questions B1 /B1 .1 .2 above) for the likelihood to use an app or to conduct a respectively B2 /B2 from Figure 1 as dependent vari- .1 .2 test. We use logit models because they yield the lowest ables (Table 2). In addition, we add to Models 2 and 4 an Akaike Information Criterion (AIC) values compared to additional binary variable measuring PG (1) vs. CG models with a probit or a cloglog link function. As framing (0). In addition to the regression coefficients, Deruelle et al. 7 Table 2 also shows the marginal effects of variable, since (something that we did not measure) and therefore are the regression coefficients do not allow us to determine willing to share their data from health tests with the the size of the effects. authorities. The main result of the analysis shows that those who Other political factors play a less important role: notably, received the private good framing are more likely to use those who support statements such as “it is the role of the health apps and do genetic tests compared to those who state to provide social security” and “it is the role of the were exposed to the common good framing. In the second state to reduce economic inequalities” are not more or set of regression models (bottom line in models 2 and 4 less likely to use health apps and provide genetic tests. in Table 2), a binary variable indicates if the respondent Although the regression coefficients have an either positive received the private good framing instead of the baseline or negative direction, their statistical significance is weak, common good framing (cf. Figure 1). The results show and they point in different directions for two similar vari- that such a framing significantly increases the likelihood ables. This finding indicates that using health technologies to use apps and wearable devices by almost 24% and is not very strongly related to political ideologies, such as makes it almost 14% more likely that individuals do the left and right political positions (Jensen, 2011). blood and genetic tests. Overall, our findings underline the importance of data The analysis shows also that very few control variables privacy when it comes to implementing practices of perso- are statistically significant (see Table 2). For example, the nalized health. Maintaining privacy over personal health items measuring the respondents’ medical history do not data storage important when it comes to using health-related seem to co-vary with the willingness to do tests, especially apps and tests: more so than beliefs about subjective health when we control for the private good framing variable. and a family history of cancer. Among the socio-economic variables yielding signifi- cant results, the following points are relevant. Regarding Conclusions the use of apps, those reporting a subjective wealth above average are more likely to use apps once the framing is As per research expectations, this study shows that indivi- applied. However, as shown in Table 1, this is due mostly duals are more likely to use personalized health technolo- to the relationship between subjective wealth and app use gies if their data are kept privately, rather than a common in the private group framing. Indeed, while before storage. This shows that privacy concerns are still import- framing 73% of the respondents with a wealth above ant, which is in line with findings from the literature average would use apps, they are only 48.6% once the (Ostherr et al., 2017; Vayena et al., 2018; Jacobs and public good framing is applied. This however rises up to Popma, 2019). Other explanatory variables have a less 75% for the private good group. This shows a meaningful clear effect, but this finding is not statistically significant. relationship between wealth and connected device usage, For instance, respondents with a degree in higher education but only when individuals are presented with the private are more likely to use those technologies. However, this good framing – thus confirming the central result of this increased likelihood disappears once we apply the framing. study. Moreover, those reporting to have a higher level of Nevertheless, ideological views should not be completely education are almost 9% more likely to use apps, but statis- eschewed: indeed, those who believe that it is the role of tically significance is lower. Regarding the use of genetic the state to collect and store data are more likely to use per- tests, the reference model indicates that respondents with sonalized health technologies. at least a high-school diploma and a family history of There are important differences between health apps and cancer are more likely to engage in tests compared to genetic tests. The likelihood for respondents to self-perform those who have a lower level of education and no cancer a direct-to-customer test is, regardless of the framing, lower history in their family. Furthermore, Swiss citizens and than for health apps. It is also lower throughout all back- those who only work part-time are less likely to take such ground variables. Previous findings on health data in tests. Switzerland did not differentiate between means to produce Among political factors, we see that respondents who health data (Pletscher et al., 2022). Genetic data is particularly feel that it is the role of the state to store and use data are sensitive a form of health data and individuals’ willingness to more than 16% more likely to report app usage and 15% even elect to do a test may be mitigated by their knowledge more likely to conduct genetic tests. These results increase and their worldview on genetic information and its use even further if we control for the effect of framing. This (Bearth and Siegrist, 2020; Claytonetal.,2018).Yet,attitudes finding is not surprising because it implies that those who towards genetic testing may not only be rooted in ethical con- are anyways more willing to give away their data to siderations. Middleton et al. (2020) for instance show that in public authorities are more likely to conduct tests, inde- Switzerland most people are unfamiliar with genetic knowl- pendently whether they received a CG or PG framing. edge and do not differentiate those from other medical infor- One possible interpretation of this result is that these mation. Willingness may also depend on costs (Ries et al., respondents have higher levels of trust in the government 2010). 8 Big Data & Society While respondents are more likely to produce and store for PH data outside of the medical context. And thus, it data in the private good group, our survey experiment would be relevant for future research to further analyze shows that willingness to produce and store data is lower how privacy and storage may affect whether individuals once the framing is applied, regardless of the framing that accept to share their data. is applied. This is important for practitioners, as it shows somewhat of a paradox: as soon as respondents are contem- Declaration of conflicting interests plating that the data can be shared, they are less likely to The author(s) declared no potential conflicts of interest with want to produce it, yet, producing this data would respect to the research, authorship, and/or publication of this improve the public good. Ultimately our results show the article. necessity to develop literacy on health data, digital platform and privacy. Potentially, this could include improving sci- Funding entific literacy and educating on the personal and collective benefits of producing this data; however, our results show The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: that respondents from the common good framing group This work was supported by the Schweizerischer Nationalfonds were the least likely to produce and store data. More gener- zur Förderung der Wissenschaftlichen Forschung, (grant number ally, our findings show also that personal health data is no CRSII5 180350). different than other forms of personal data: mitigating data privacy concerns is of the utmost importance for a correct implementation of personalized health objectives. ORCID iDs Therefore, in order to promote the production of persona- Thibaud Deruelle https://orcid.org/0000-0002-3722-0467 lized health, decision makers should create storage solu- Philipp Trein https://orcid.org/0000-0001-6217-6675 tions that could emulate, for instance, the Swiss Joël Wagner https://orcid.org/0000-0002-3712-5494 Electronic Health Record (Baumann et al., 2018; De Pietro and Francetic, 2018; Platt and Kardia, 2015) that References allow citizens to give consent for the use of their data for Allen LN and Christie GP (2016) The emergence of personalized the purpose of research. health technology. Journal of Medical Internet Research 18(5): Like any other research, our study comes with limita- e99. tions. We lay out three limitations. The first one being the Barazzetti G, Bühler N, Audétat M, et al. (2021) Making persona- generalization of our findings. Switzerland is arguably at lized medicine ethical: A critical examination of the new pro- the forefront of the personalized health “turn,” and thus mises of ‘personalized health’ in Switzerland. Science and the debate regarding data privacy is particularly relevant, Public Policy 48(6): 818–828. unlike in other national contexts in Europe or worldwide. Baumann LA, Baker J and Elshaug AG (2018) The impact of elec- Second, the scope of our empirical analysis is limited as tronic health record systems on clinical documentation times: A systematic review. Health Policy 122(8): 827–836. we could not include the Italian-speaking population of Bearth A and Siegrist M (2020) Psychological factors that deter- Switzerland. Finally, our study is limited to personalized mine people’s willingness to share genetic data for research. health technologies (smartphone apps, wearable devices, Clinical Genetics 97(3): 483–491. do-it-yourself tests) where data generation is depending Blasimme A, Vayena E and Van Hoyweghen I (2019) Big data, on the patients’ willingness to use them. Our results are precision medicine and private insurance: a delicate balan- therefore not relevant to personal health data generated as cing act. Big Data & Society 6(1): 1–6. DOI: 10.1177/ a result of a medical intervention. This last limitation may inform the direction for future Boyd D and Hargittai E (2010) Facebook privacy settings: Who research. Further analysis of individuals’ preferences can cares? First Monday 15(8). Epub ahead of print. DOI: 10.5210/ be done by comparing the medical and the non-medical fm.v15i8.3086 dimension of PH data generation. In the same vein, more Braunack-Mayer A, Fabrianesi B, Street J, et al. (2021) Sharing government health data with the private sector: Community storage and privacy options can be explored Considering attitudes survey. Journal of Medical Internet Research that respondents’ willingness to use health technologies is 23(10): e24200. DOI: 10.2196/24200. significantly higher when the state is not involved. Future Brown I (2016) The economics of privacy, data protection and sur- research could explore which private third-party citizens veillance. In: Bauer JM and Latzer M (eds) Handbook on the trust to handle their data, whereas private insurance, Economics of the Internet. Cheltenham: Edward Elgar digital technology firms or cooperative biobanks. On the Publishing, pp.247–261. specificity of genetic testing, future research should Bühler G, Hermann M and Lambertus M (2019) Observatoire explore whether ethical and cost-related considerations “société numérique et solidarité” opinion et comportement matter in the Swiss case or whether privacy concerns de la population suisse. Zürich: Sotomo/Fondation Sanitas. trump it all as one could expect from our findings. Caenazzo L, Tozzo P and Borovecki A (2015) Ethical governance Finally, our results can inform how to design privacy in biobanks linked to electronic health records. European Deruelle et al. 9 Review for Medical and Pharmacological Sciences 19(21): Nakagomi H, Sakamoto I, Hirotsu Y, et al. (2016) Willingness of 4182–4186. Japanese patients with breast cancer to have genetic testing of Clayton EW, Halverson CM, Sathe NA, et al. (2018) A systematic BRCA without burden of expenses. Breast Cancer (Tokyo, literature review of individuals’ perspectives on privacy and Japan) 23(4): 649–653. genetic information in the United States. PloS one 13(10): Okma KGH and Crivelli L (2013) Swiss And Dutch “consumer- e0204417. DOI: 10.1371/journal.pone.0204417. driven health care”: Ideal model or reality? Health Policy DePietro C, Camenzind P, Sturny I, et al. (2015) Health systems in 109(2): 105–112. transition, Switzerland. Health System Rev 17(4): 1–323. Ostherr K, Borodina S, Bracken RC, et al. (2017) Trust and De Pietro C and Francetic I (2018) E-health in Switzerland: The privacy in the context of user-generated health data. Big laborious adoption of the federal law on electronic health Data & Society 4(1): 1–11. DOI: 10.1177/2053951717704673. records (EHR) and health information exchange (HIE) net- Patil SH, Lu H, Saunders CL, et al. (2016) Public preferences for works. Health Policy 122(2): 69–74. electronic health data storage, access, and sharing — evidence Deruelle T, Kalouguina V, Trein P, et al. (2022) Is there a “pan- from a pan-European survey. Journal of the American Medical demic effect” on individuals’ willingness to take genetic Informatics Association 23(6): 1096–1106. tests? European Journal of Human Genetics. Epub ahead of Persaud A and Bonham VL (2018) The role of the health care pro- print. DOI: 10.1038/s41431-022-01223-6. vider in building trust between patients and precision medicine Feldman EA (2012) The genetic information nondiscrimination research programs. The American Journal of Bioethics 18(4): act (gina): Public policy and medical practice in the age of per- 26–28. sonalized medicine. Journal of General Internal Medicine Phillips KA, Deverka PA, Hooker GW, et al. (2018) Genetic test 27(6): 743–746. availability and spending: Where are we now? Where are we Garrison NA, Sathe NA, Antommaria AHM, et al. (2016) A sys- going? Health Affairs 37(5): 710–716. tematic literature review of individuals’ perspectives on broad Phillips KA, Trosman JR, Kelley RK, et al. (2014) Genomic consent and data sharing in the United States. Genetics in sequencing: Assessing the health care system, policy, and Medicine 18(7): 663–671. big-data implications. Health Affairs 33(7): 1246–1253. Jacobs B and Popma J (2019) Medical research, big data and the Platt J and Kardia S (2015) Public trust in health information need for privacy by design. Big Data & Society 6(1): sharing: Implications for biobanking and electronic health 205395171882435. record systems. Journal of Personalized Medicine 5(1): 3–21. Jensen C (2011) Marketization via compensation: Health care and Pletscher F, Mändli Lerch K and Glinz D (2022) Willingness to share the politics of the right in advanced industrialized nations. anonymised routinely collected clinical health data in Switzerland: British Journal of Political Science 41(4): 907–926. a cross-sectional survey. Swiss Medical Weekly 152(23): w30182. Khoury MJ, Iademarco MF and Riley WT (2016) Precision public DOI: 10.4414/smw.2022.w30182. health for the era of precision medicine. American Journal of Ries NM, Hyde-Lay R and Caulfield T (2010) Willingness to pay Preventive Medicine 50(3): 398–401. for genetic testing: A study of attitudes in a Canadian popula- Laurie G (2011) Reflexive governance in biobanking: On the tion. Public Health Genomics 13(5): 292–300. value of policy led approaches and the need to recognise the Seifert A, Christen M and Martin M (2018) Willingness of older limits of law. Human Genetics 130(3): 347–356. adults to share mobile health data with researchers. GeroPsych: Lee SSJ (2015) The biobank as political artifact: The struggle over race The Journal of Gerontopsychology and Geriatric Psychiatry in categorizing genetic difference. The ANNALS of the American 31(1): 41–49. Academy of Political and Social Science 661(1): 143–159. Seifert A and Vandelanotte C (2021) The use of wearables and Leonard NR, Silverman M, Sherpa DP, et al. (2017) Mobile health health apps and the willingness to share self-collected data technology using a wearable sensorband for female college students among older adults. Aging and Health Research 1(3): with problem drinking: an acceptability and feasibility study. JMIR 100032. DOI: 10.1016/j.ahr.2021.100032. mHealth and uHealth 5(7): e90. DOI: 10.2196/mhealth.7399. Stockdale J, Cassell J and Ford E (2018) “giving something back”:A McCartney CI, TJ W and Williams R (2011) Transnational exchange systematic review and ethical enquiry into public views on the use of forensic dna: Viability, legitimacy, and acceptability. European of patient data for research in the United Kingdom and the repub- Journal on Criminal Policy and Research 17(4): 305–322. lic of Ireland. Wellcome Open Research 3(6). Epub ahead of print. Middleton A, Milne R, Almarri MA, et al. (2020) Global public DOI: 10.12688/wellcomeopenres.13531.2. perceptions of genomic data sharing: What shapes the willing- Trein P and Wagner J (2021) Governing personalized health: A ness to donate dna and health data? The American Journal of scoping review. Frontiers in Genetics 12: 650504. DOI: 10. Human Genetics 107(4): 743–752. 3389/fgene.2021.650504. Milne R, Morley KI, Almarri MA, et al. (2021) Demonstrating trust- Vayena E, Haeusermann T, Adjekum A, et al. (2018) Digital worthiness when collecting and sharing genomic data: Public health: Meeting the ethical and policy challenges. Swiss views across 22 countries. Genome Medicine 13(1): 1–12. Medical Weekly 148: w14571. Minvielle E, Waelli M, Sicotte C, et al. (2014) Managing custom- Whiddett R, Hunter I, Engelbrecht J, et al. (2006) Patients’ atti- ization in health care: A framework derived from the services tudes towards sharing their health information. International sector literature. Health Policy 117(2): 216–227. Journal of Medical Informatics 75(7): 530–541.

Journal

Big Data & SocietySAGE

Published: Jan 1, 2023

Keywords: personalized health; health apps; genetic tests; data storage

There are no references for this article.