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

Learn More →

Consent for use of personal information for health research: Do people with potentially stigmatizing health conditions and the general public differ in their opinions?

Consent for use of personal information for health research: Do people with potentially... Background: Stigma refers to a distinguishing personal trait that is perceived as or actually is physically, socially, or psychologically disadvantageous. Little is known about the opinion of those who have more or less stigmatizing health conditions regarding the need for consent for use of their personal information for health research. Methods: We surveyed the opinions of people 18 years and older with seven health conditions. Participants were drawn from: physicians' offices and clinics in southern Ontario; and from a cross-Canada marketing panel of individuals with the target health conditions. For each of five research scenarios presented, respondents chose one of five consent choices: (1) no need for me to know; (2) notice with opt-out; (3) broad opt-in; (4) project- specific permission; and (5) this information should not be used. Consent choices were regressed onto: demographics; health condition; and attitude measures of privacy, disclosure concern, and the benefits of health research. We conducted focus groups to discuss possible reasons for observed consent choices. Results: We observed substantial variation in the control that people wish to have over use of their personal information for research. However, consent choice profiles were similar across health conditions, possibly due to sampling bias. Research involving profit or requiring linkage of health information with income, education, or occupation were associated with more restrictive consent choices. People were more willing to link their health information with biological samples than with information about their income, occupation, or education. Conclusions: The heterogeneity in consent choices suggests individuals should be offered some choice in use of their information for different types of health research, even if limited to selectively opting-out. Some of the implementation challenges could be designed into the interoperable electronic health record. However, many questions remain, including how best to capture the opinions of those who are more privacy sensitive. Page 1 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 - people's view of the level of consent required for use of Background The term "stigma" generally refers to a distinguishing per- their information would vary directly with disclosure con- sonal trait that is perceived as or actually is physically, cern and inversely with perceptions of the benefits of socially, or psychologically disadvantageous. [1] Because health care and the potential for health research to of the presence of that trait, an individual may be discrim- improve the lifespan and quality of life of people with inated against – e.g. in employment or in social circles. their health condition. Health conditions will vary in the extent to which they are perceived by those individuals having the condition and This paper reports on the testing of these hypotheses. by others as being stigmatizing. Individuals with a poten- tially stigmatizing health condition may be more inclined Methods Choice of health conditions than members of the public without such conditions to experience concerns over disclosure of their personal In this study, we included seven health conditions with health information out of concern that this could result in varying susceptibility to being labelled as stigmatizing. discrimination against them. For example, a person with Four conditions – hypertension, diabetes, chronic depres- a prior history of cancer may be concerned over denial of sion and alcoholism – were used in a previous public certain employment opportunities, a mortgage, or life opinion survey of members of the public, in which insurance. A person with HIV/AIDS may be concerned respondents were asked to imagine they had one of these about social isolation because of others' concerns that health conditions. [4] In the previous study, hypertension their presence puts others at increased risk of contracting and diabetes were found to be lower-stigma health condi- the condition. tions. Chronic depression and alcoholism were found to be higher-stigma conditions. In this study, we had the There is now an emerging body of literature examining opportunity to obtain the views of people with these con- the opinion of the public regarding consent for use of ditions. To these four conditions we added: HIV, to create one's health information for research. [2-9] However, the an upper extreme category for potential stigma; breast opinion of those who have health conditions that may be cancer; and lung cancer. We anticipated that responses stigmatizing to a greater or lesser degree has been much may differ between breast and lung cancer because of a less studied. [10] There are several reasons for considering greater general public support for breast cancer sufferers – or even giving priority to – the values and expectations and a perception that lung cancer is self-inflicted through of these individuals. People who are unwell have the most smoking. at stake, both because they stand to benefit from research Survey into their health condition and because a breach of pri- vacy may potentially expose them to discrimination in The study proceeded in two phases. In Phase 1 (Novem- obtaining loans, mortgages, insurance, or employment. ber 2006 to July 2007), we surveyed individuals with the Moreover, they are under-represented in surveys targeted target health conditions. In Phase 2 (July to September to the general public and, to the extent that some health 2007), we held focus groups with a sub-sample of partici- conditions are stigmatizing, the perspective of individuals pants from Phase 1 – one group for each health condition. with these conditions may be discounted by the general The chief purpose of Phase 2 was to help inform our anal- public. Further, if we listen only to the voice of the general ysis of the findings derived from Phase 1, by providing public without attention to the concerns of this vulnera- examples of the types of concerns some people took into ble minority we run the risk of committing a form of "tyr- account when making decisions around consent. anny of the majority". [11] Setting and Participants Objective Survey participants were drawn from two sources: (1) a The purpose of this study was to examine the attitudes of pre-existing cross-Canada panel of individuals with iden- people with a range of potentially stigmatizing health tified health conditions, maintained by Harris Interactive, conditions concerning the need for consent for the use of a professional polling firm; and (2) patients recruited by their personal information for different types of observa- the investigators through family physicians' offices and tional health research, and to compare their attitudes with specialty clinics in the vicinity of Hamilton, Ontario, Can- those of the general public. We hypothesized that: ada. The reference group consisted of people recruited through Harris Interactive who had none of the target - responses would differ across health conditions; health conditions and no other serious health conditions. This group was used to approximate the response of the - some patient groups would be more permissive and oth- general public. All survey participants were 18 years or ers more restrictive than the general public; and older. Page 2 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 Participants recruited directly by the investigators were the 7 health conditions according to level of disclosure either sent a letter from the physician's office in the mail concern. This algorithm placed HIV/AIDS first, followed explaining the study or handed an information brochure by alcoholism, lung cancer, breast cancer, depression, dia- in the clinic by staff. In each case, information was pro- betes and hypertension. vided for patients to contact the investigators if interested. In total, 892 brochures were mailed to patients' homes We presented five different scenarios involving use or and 888 brochures provided to physicians and clinics for linkage of personal information for health research: (1) directly handing out to patients. use of health data for quality improvement; (2) use of the same data for marketing; (3) linkage of work/education/ All participants recruited through Harris Interactive com- income information with health information; (4) linkage pleted the survey over the internet. Participants recruited of biosamples with health information (a) assuming no through local family physicians and clinics were given the profit and (b) assuming a profit element. (See additional choice to complete the survey via internet or by telephone. file 1 for a more detailed description.) These scenarios This was done to minimize refusal due to lack of access to were identical to those used two years earlier in a series of or familiarity with the internet, particularly among older seven cross-Canada public dialogues. [13] Participants of patients. Those opting to do the internet survey used the the current study were advised that, in each scenario, same system as the Harris participants. Those opting to names, addresses and any other information that could complete the survey by telephone arranged a scheduled directly identify them were removed. Following each sce- call with telephone surveyors from Harris Interactive. nario, participants were asked which statement best They were mailed a hard copy of the core questions of the described their view (words in italics not included in sur- survey beforehand, so as to minimize differences in vey responses): response due to method of survey administration. (1) There is no need for me to know. Just use it. Sample size was calculated on the basis of the primary outcome variable – consent choice in the use of personal (2) My permission is not needed, but I want to know information for health research. This was expressed on a this is being done and a chance to say "no." [i.e. notice 5-point ordinal scale. (See Survey Data and Key Variables with opt-out] below.) This sample was determined with the primary goal of building a multivariable regression model to com- (3) My general permission is needed. This could be for pare the overall attitudes among the seven groups and the several different research studies. I could withdraw my general public controlling for several demographic and permission in future. [i.e. broad opt-in] other confounding variables. Heuristics based on simula- tion studies indicate that at least five respondents per (4) My permission is needed each time. [i.e. project- degree of freedom for each predictor variable are required specific consent] for the stability of the model. [12] We have 7 predictor variables with a total of 21 degrees of freedom, which (5) My information should not be used for this pur- would require at least 105 participants. We aimed to pose. recruit 1400 and obtained responses from 1137 (734 from Harris Interactive and 403 from physician offices The five response options above, hereinafter referred to as and clinics) with the sampling stratified by health condi- "consent choices", served as the outcome variable in a tion. We inflated our minimum sample size by a factor of multivariable regression analysis using the following as over 10 to account for potential clustering of responses predictor variables: within a patient. Therefore, the sample size was adequate to ensure the stability of the model. - demographics (age, sex, education, marital status, employment, and income) Survey Data and Key Variables We collected information on participant demographics, - health condition (one of the 7 target conditions) attitudes about privacy, disclosure concern, and the bene- fits of health care and health research at improving lon- - self-reported health (6-point scale, varying from gevity and quality of life, and the participant's health "very poor" to "excellent") conditions. Where the participant had more than one tar- get health condition, they were asked to answer the survey - scenario, and questions with only one health condition in mind. In ear- lier pilot work, we established an algorithm for determin- - attitudinal variables (disclosure concern and medical ing which health condition would take priority, ranking benefits score. These are described below. Questions Page 3 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 used to compile these scores and the scoring scheme mial logistic regression which is the more correct analysis are found in additional file 2.) but more difficult to interpret. [4] While there are many dimensions to stigma, for the pur- Initially, individual predictor variables were regressed poses of this study, we chose to focus on individuals' dis- onto consent choice using univarite analysis. Variables closure concern – i.e. concern that others may find out that met the criterion of alpha = 0.20 were then included about their health condition. For this, we asked: How con- in the multivariable regression model. The chief variables cerned would you be if: (a) your employer found out of interest – health condition, disclosure concern, and about any health condition(s) you have; (b) your health medical benefit scores – were forced into the model. We insurer found out about any health condition(s) you tested for interaction effects of disclosure concern and have; or (c) a friend other than those you told found out medical benefits scores with research scenarios and health about any health conditions you have. For each of these condition, No significant interactions were found. questions, respondents replied either: "not at all con- cerned"; "somewhat concerned"; or "very concerned". Focus Groups At the end of the survey, those participants recruited For the medical benefits scale, people replied on a 5-point directly by the investigators in the Hamilton area were scale ("strongly agree, somewhat agree", "neither agree asked if they would be willing to participate in a focus nor disagree", "somewhat disagree", or "strongly disa- group to discuss the reasoning that participants may have gree") to the following statements: (a) Medical treatments used to make their consent choices. Those who agreed can improve the quality of my life; (b) medical treatments were contacted by the study coordinator. One focus group can extend my life; (c) disease prevention programs have was convened for each health condition except for alco- shown me how to live a healthier life; and (d) medical holism, as the survey sample size was too small for this research can improve my life. group. We sampled from the pool of survey participants to provide a representative sample in each focus group with We re-scaled the disclosure concern and medical benefits regard to age, gender, and self-reported health status. scores to a 0–1 scale to facilitate interpretation of relative When making selections, investigators were blind as to attitudes toward disclosure and medical benefit across volunteers' responses to survey questions. health conditions. Six focus groups, ranging in size from 6 to 10 individuals, Dealing with potential sampling bias were conducted between July and September 2007. Focus We checked for sampling bias chiefly through two meth- groups were 90 minutes in length. Participants received an ods. Harris Interactive sampled questions from our survey honorarium of $75 following the session. These meetings in an omnibus random-digit dialled telephone survey and were moderated by two of the researchers. Upon arrival, compared the responses from the telephone survey with participants completed a mini-survey comprised of the those from their internet sample. In addition, we com- key questions from the Phase 1 survey. During the focus pared the consent choices of the reference group from this group, a structured interview protocol was used to ask par- study regarding the five scenarios with the consent choices ticipants to reflect on the survey findings regarding con- of the people who participated in the public dialogues in sent choice for the five different research scenarios. our previous study. [13] Participants were then asked to consider how the survey responses compared with their own responses and to con- Statistical methods sider the reasons for their responses and any possible var- Consent choices were analysed graphically across scenar- iance from the response pattern from the Phase 1 survey. ios and across health conditions. We also plotted disclo- Focus group participants were also asked to comment on sure concern and medical benefit scores across health the relative importance of individual control over use of conditions using a radar graph. their information and safeguards and the relation between them. This question was addressed both in the To test our chief hypotheses, we used regression analysis abstract and by giving them the opportunity to rate spe- controlling the correlation across scenarios using the cific controls and safeguards. Immediately following the method of generalized estimating equations (GEE) meeting, participants again completed the mini-survey, so assuming an exchangeable correlation structure. [14] The a comparison of the responses of focus group participants results are reported as estimates of model coefficients with those of the larger sample could be made. Discus- (with corresponding 95% confidence interval) and associ- sions were audio recorded and transcribed for analysis. ated p-value. Statistical computations used SAS, version 9.1.3 (Cary, NC.). In earlier work, we found the results of Verbatim transcripts from all focus groups were read inde- linear regression to yield equivalent results to multino- pendently by at least two members of the research team Page 4 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 and quotations were selected to exemplify the types of rea- pants through the investigators' sources is summarized in soning that focus group participants used to make their Figure 1. consent choices. These quotations are used below to help illustrate the broader survey analysis, and to provide a Demographic data window on the kinds of issues some people may address Additional file 3 summarizes the demographic character- when making consent choices. istics of study participants by sampling source. Those recruited directly by the investigators are presented by Ethics Review method of survey completion. Overall, the reference The research was reviewed and approved by the research group was younger (mean age 46 years) than participants ethics boards of St. Joseph's Healthcare, Hamilton, with target health conditions (mean 55 years), regardless Ontario, McMaster University Health Sciences REB, Ham- of sampling source. Their self-reported health was better ilton, Ontario and the Social Sciences and Humanities than those with the target health conditions. Those with REB, University of Ottawa, Ottawa, Ontario. the target health conditions completing the survey over the internet were comparable between the two sampling Results sources on most demographic variables. However, the Participants Harris sample was less wealthy and reported poorer Four hundred three survey participants were recruited health than the sample assembled by the investigators. directly by the investigators and 734 were recruited Survey participants recruited by the investigators who through Harris Interactive. Recruitment of survey partici- completed the survey by telephone were older (mean age Summary of recruitment process, Figure 1 using the investigators' sample Summary of recruitment process, using the investigators' sample. Page 5 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 62) with a greater percentage of women (68% vs. 60% in was uniformly high across health conditions. On the general public and 53% and 57% in the Harris and inves- other hand, there was substantial variation across the tigator internet samples). Telephone survey participants health conditions regarding the level of concern partici- had a greater percentage with high school education or pants had about employers, insurers, or friends finding less (56%) than those who completed the survey over the out about their health condition. Disclosure concern was internet (28%). In addition, telephone respondents were lowest for those with hypertension, diabetes, and lung more likely to be separated, widowed, or divorced (30% cancer. It was highest among those with chronic depres- vs. 14–24%). Fewer were employed (21% vs. 49–72%) sion and HIV/AIDS. We note also that disclosure concern and a greater percent had an income less than $40,000 per in the reference group was almost as high as the HIV/AIDS year (42% vs. 20–35%). group yet, they had no serious or chronic health condi- tion. Attitudes A summary score of participants' perceptions about the Consent choice benefits to be accrued from health care and health Across scenarios, consent choice profiles were very similar research (medical benefits score) and disclosure concern for all health conditions. They were also very similar to is provided in Figure 2. Each axis on this radar plot repre- the profile of the reference group. For ease of presentation, sents one health condition or the reference group. The we have combined the responses across health conditions scale on the axis for both attitudes was standardized to in Figure 3 and compared these with the reference group. (0,1) to facilitate comparison. The medical benefits score !&! +,! &"*#! )*! &!'&(   ! %# " # ! $# A Figure 2 ttitudes of survey participants across health conditions Attitudes of survey participants across health conditions. Page 6 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 Cons target health conditions Figure 3 ent choices across research scenarios – A comparison of the general population with a pooled sample of those with the Consent choices across research scenarios – A comparison of the general population with a pooled sample of those with the target health conditions. Effect of intended use and commercialization on consent choice marketing scenario (Scenario 2), and with no particular Scenario 1 involved the use of prescribing information consent choice being preferred. from the medical record for quality improvement and, for this paper, serves as the comparator scenario. Figure 3 In our focus groups, we probed for explanations behind shows a relatively permissive consent profile for this sce- the desire for greater control when commercialization was nario, with the most common response being "Just use it" involved. The responses were wide ranging. A common (35–40%). By contrast Scenario 2, which used the same sentiment was that people felt they were being taken prescribing information for marketing purposes, elicited advantage of: essentially the opposite profile, with over 40% of respondents indicating that this information should not "It was a matter of control. It's the whole idea of profit, that be used (at all) for this purpose. word 'profit'. Once I see that, I just have a sense of being taken advantage of. ... but on the other hand I wouldn't not A similar, though less marked, shift in the pattern of con- do it because it is helpful. I would just want to know." (Par- sent choice was observed when comparing Scenarios 4a ticipant 7, Diabetes group) and 4b. In Scenario 4a, individuals' health information was linked with biological samples in the absence of any "First thing I thought of was 'Well, if they're selling it for a commercialization of any discovery. In this case, the con- profit what do I get out this?' I just don't see your volunteer- sent profile was quite similar to that for Scenario 1. When ing something, if somebody else is making a profit out of it. a potential profit element was introduced through the I don't see that." (Participant 3, Lung cancer group) development of a commercial lab test (Scenario 4b), the consent profile displayed a desire for greater control over use of that information, though not as strong as for the Page 7 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 Different consent profiles for linkage with biological samples vs. you're not sure what types of people are analyzing that. income, education, and occupation, in the absence of profit It's not a white lab coat, microscope it's more of a psy- The consent profile for linkage of health information with chological marketing." (Participant 7, Depression) biological samples in the absence of profit (Scenario 4a) was similar to that for the quality improvement scenario "I trust giving out the information to most of the medi- (Scenario 1). Thirty to 40% of respondents felt it was cal profession but it's not something I would give will- acceptable to link this information without notification. ingly to people that I don't really know well." By contrast, linkage of health information with income, (Participant 3, Depression) education, or occupation (Scenario 3) was associated with a consent profile that reflected a desire for greater control "I have found when I've been asked to [do] this that my over use of the information. For this scenario, only 11– family Doctor, where I go – and I have trust in him and 15% felt it was acceptable to link this information without in his staff – and he sent a letter saying 'This is what notification and 30–43% of respondents felt their consent they are asking you to do and I believe in it.' And so I should be sought for each use before the information feel that gave me the confidence that this wasn't just could be linked. some little thing that showed up at my door. My family doctor... I have trust in him." (Participant 8, Diabe- Because this finding surprised us, we probed this in our tes) focus groups. Similar sentiments were expressed across groups, around 3 categories of explanations: These (and other) references to high trust were directed more toward doctors and hospitals or, even more person- 1. Participants felt that information about income, ally, to their doctor. While there was also a relatively high education or occupation says a lot about "who we are" trust toward university researchers (Figure 4), that trust whereas a biosample only provides information about was more tentative than for doctors. "what we are". Regression analyses "I think the simple answer is that physical tissue sample Additional file 4 compares three regression models. The full model examines the independent contributions of is just a piece of what you are, what you might be...where the rest of the information [education, health condition and attitudes toward disclosure concern income, and employment] is more of who you are. Peo- and medical benefits, controlling for scenario, survey ple are more afraid of the revelation of who you are method, and sex. Reduced model 1 removes disclosure than what you are." (Participant 8, HIV group) concern and medical benefit from the model while reduced model 2 removes health condition from the 2. Participants believed that access to information model. about education, income, and occupation can lead to the identification of the individual, but biosamples In the full model, consent choices among survey partici- cannot. pants did not differ significantly across health conditions, compared with the reference group. Increased disclosure "The work, education and your income. If somebody concern was associated with a more restrictive consent looks at a biological sample they can't look up and say choice while greater perception of the benefits of medical 'That's you or you.' You can identify somebody by all care and medical research was associated with more per- that, those other things." (Participant 6, Depression missive consent choice. When the attitude variables (dis- group) closure concern and medical benefits scores) were removed from the model (Reduced Model 1), the coeffi- "The only thing you can get from tissue or bodily fluids cients for several of the health conditions moved in the is DNA and what the disease is. They can't get infor- direction of being more permissive compared with the ref- mation about you per say. Right?" (Participant 3, erence group. In two cases (breast cancer and hyperten- Lung cancer group) sion) these became statistically significant. On the other hand, when health conditions were removed from the 3. Participants trusted that their doctor and the medi- model (Reduced Model 2), estimates for the attitude vari- cal system will ensure that the information will be ables were s . This suggests that the stronger predictor of used appropriately. consent choice was individual attitude toward medical benefit and disclosure concern than was the person's "I think with tissue you can see a real concrete connec- health condition. tion. You can imagine a scientist in a lab studying that tissue, looking for an answer. With income and work Page 8 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 1* ." " ' /2 0 / " "* / *+ #', -,..) ! "  #$ %&' () Tr Figure 4 ust in organizations Trust in organizations. Regression results also confirm our perceptions that that differed widely in potential for stigma. When exam- research involving an element of profit (Scenarios 2 and ining the confidence intervals around the parameter esti- 4b) or requiring linkage of health information with mates for the health conditions, they did not appear to be income, education, or occupation (Scenario 3) were asso- of a magnitude that would be policy relevant. These find- ciated with more restrictive consent choices. Female ings surprised us. It may be that, across health conditions, respondents were also generally more restrictive in their those who were more privacy sensitive were less inclined consent choices. Those completing the survey by tele- to participate in the study and those were more permissive phone were more permissive in their consent choices than about use of their information were more inclined to par- were those completing the survey over the internet. ticipate. This is discussed further under "Limitations" below. Discussion Chief findings Disclosure concern differed across health conditions and We had hypothesized that consent choice would differ was associated with more restrictive consent choices. By across the selected health conditions and also with indi- contrast, there was a uniformly high rating of medical vidual attitudes regarding disclosure concern and percep- benefits across health conditions. However, at the individ- tions of the benefits of medical care and research. While ual level, a lower perception of medical benefits was asso- we recognized beforehand that hypertension and diabetes ciated with more restrictive consent choices. Thus, were relatively low in stigma and that HIV and alcoholism individual attitudes – and disclosure concern in particular were likely higher in stigma, we did not assume strict ordi- – were more predictive of consent choice than was one's nality across health conditions. We found that, collec- health condition. This suggests that privacy attitudes may tively, participants with target health conditions were be formulated relatively early on and may be robust to slightly more permissive in their consent choices com- one's health condition, which may develop later in life. pared against the reference group (Figure 3). However, when we controlled for survey method, sex, and attitudes, We also observed significantly more restrictive consent we found no statistically significant difference in consent choices in research scenarios involving profit or when choice across health conditions, compared with our refer- linking health information with income, education, or ence group – this despite having chosen health conditions occupation. This is consistent with findings from our ear- Page 9 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 lier public dialogues. [13] We also note that people were frame; and (b) they consisted of individuals who did not relatively trusting with regard to linking their health have any of the target health conditions and who had no records to their biological samples – even more trusting other major health conditions. A true random sample of than with linking to income, education, and occupation. the public would have some proportion of respondents This is an interesting finding worthy of further study, with the target health conditions. given that there is a much higher likelihood of commer- cial application and intellectual property protection in Given all this, we believe our study may have under-rep- research involving biological samples and the great poten- resented those in our society who are the most privacy tial for stigmatization and discrimination. Our focus sensitive. Moreover, depending on the severity of the groups suggest that, at least in part, their lesser concern selection bias, it may be that our failure to observe a dif- was attributable to perceptions that specialized knowl- ference in consent choice across health conditions may be edge was required to interpret one's DNA and those with a result of the absence of more privacy-sensitive respond- that specialized knowledge were trusted to keep the infor- ents. These non-participants may represent varying pro- mation confidential. However, once one's genetic risk portions across health conditions. We cannot assess what profile is recorded in the health record, this information is proportion of the selection bias is due to self-selection or equally subject to misadventure as one's income, should selective approaching by their physicians. that information fall into the wrong hands. With regard to trust, we note that most of the focus group discussion of Finally, in the consumer literature, stated privacy prefer- trust was in reference to their doctors. Perhaps trust in ences are often much more stringent than those revealed one's doctor then confers benefits of trust to the research- in actual behaviour. [15] Thus, it is possible that the stated ers in the process. consent choices for use of one's personal information reported here may be different than what they are pre- Limitations pared to accept in the health care "marketplace". Based on a comparison with a subset of questions in an omnibus survey conducted by Harris Interactive, we deter- Policy implications mined that people recruited into the study were some- No one consent option was preferred by even a simple majority of survey participants in any of the scenarios what less privacy concerned and more research-friendly than the general public. Within our study sample, those (Figure 3). This wide variation in opinion is consistent who completed the survey by telephone were particularly with our earlier work with the general public and with a less privacy concerned. This observation is reinforced by recent American survey. [3] This high heterogeneity in comparison of our current findings with those of our ear- consent choices makes it difficult to put forward any one lier study involving a cross-Canada sample of the general model for consent for research use of one's personal public using random-digit dialling. In the earlier study, health information. One possible response would be to the profiling of consent choices was somewhat less per- offer individuals a menu of choices with a mechanism to missive than observed here – particularly around the link- track their choices re: secondary uses of data. While this is age with genetic information. [13] Intuitively, one would now technically possible, there are multiple challenges expect that individuals willing to participate in ongoing from a systems perspective. For example, who would bro- internet consumer panels may be less privacy concerned ker the consent process and under what conditions? Anec- and more open to participating in research. Our conven- dotal evidence suggests that physicians do not have the tional sampling through clinics, though, was also subject time for this. On the other hand, special "clinics" could be to a similar selection bias. Current ethics rules require that set up through hospitals and centres where applications a researcher not approach potential research participants are made for renewal of health cards. Information could directly, based on prior knowledge about that individual's be available through brochures and DVDs, and knowl- health condition. Individuals must first be asked by some- edgeable individuals could be available either in person one who may be reasonably expected to have access to or over the telephone. All these approaches would require this information if they would be willing to have their a substantial investment in funds and an ongoing infra- name released so they could be called by the researcher, or structure for managing these consent choices. This could if they would be willing to take the initiative themselves also be designed into the planned pan-Canadian interop- to contact the researcher. This two-stage process probably erable electronic health record system, through secure resulted in lack of access to more privacy-sensitive web-based patient portals into their health record. [16,17] research participants. These portals are web-based interactive systems that allow individuals to view their health record and communicate The reference group in this study is not quite representa- in a secure fashion with their health care provider's office. tive of the "general public", to the extent that: (a) they, Another consideration is whether consent choices should too, were drawn from the Harris internet polling sampling be totally unconstrained. There is growing evidence that Page 10 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 opt-in consent processes can result in selection biases that appreciate the rationale for restricting researchers from may affect conclusions as to various causal associations. directly recruiting patients on the basis of prior knowledge [18] Are there certain types of research (e.g. public health, about their health condition, it is ironic that, to meet eth- quality of care) for which the default assumption would ical requirements respecting individuals' privacy, this be that the information may be used for research and peo- study has likely under-represented the interests of those ple's only option would be to selectively opt out? These very people whose voices need to be heard. When address- are but two examples of issues that need to be addressed ing questions like the one we have posed here, there need through further research. to be ways to better reach those who have the most at stake. Our survey and focus group participants placed high trust in medical researchers – higher than that found recently Competing interests by Westin in the United States. [3] However, as in Westin's The authors declare that they have no competing interests. study, much of this trust was qualified. Approximately two-thirds of those expressing trust in university research- Authors' contributions DW conceived and designed the study, led the focus ers only "somewhat trusted" them. Therefore, much of this trust in researchers is vulnerable to erosion in the groups and the analyses and was the primary author of the event of a high-profile breach of confidence. Survey par- manuscript; VS, CC, LS, JR and LT all contributed to the ticipants also valued highly the ability to monitor how development of the protocol. JR led in the analyses of the their information was being used and the ability to say text. LT provided statistical guidance. JC conducted the "No" to particular uses. This "trust but verify" attitude statistical analyses. All co-authors reviewed and revised could also be accommodated through the development of drafts of the manuscript. All authors read and approved a patient portal into one's electronic health record. the final manuscript. We observed a shift toward more restrictive consent Additional material choices when research involved a commercial element. Given the increasing importance of personal information Additional file 1 to health research, and the pervasiveness of the private Box 1 – Description of Research Scenarios. Provides information about sector and commercialization in the research enterprise, the scenarios on which consent choices were made. continued public engagement would be worthwhile. This Click here for file would help researchers to better appreciate the underlying [http://www.biomedcentral.com/content/supplementary/1472- 6939-10-10-S1.doc] concerns of the public around research and profit and to determine whether these views change or are reinforced Additional file 2 with a better understanding of the role of both personal Appendix 1. Survey questions that formed the elements for (a) the dis- information and the private sector in research. closure concern scale; (b) the medical benefits scale; and (c) the con- sent choices outcome variable. This document provides the core questions Conclusions that were used in developing two key predictor variables (a measure of The public is generally supportive of research use of per- individuals' concern over disclosure of information about their health con- dition and a measure of individuals' perceptions of the benefits of health sonal health information, but do not wish to entirely care and health research for their health condition) and the key outcome relinquish control. Despite having a relatively research- variable (consent choice). friendly sample, we still observed substantial individual Click here for file variation in opinion as to the degree of control that peo- [http://www.biomedcentral.com/content/supplementary/1472- ple wish to have over use of their personal information for 6939-10-10-S2.doc] research. Opportunities for individual choice over use of one's personal health information could be designed into Additional file 3 the planned pan-Canadian inter-operable electronic Table 1. Summary of Participant Demographics. This describes basic demographic features of survey participants broken down by survey health record system. method and sample source. We divided investigator sample source into those who responded by telephone and internet to determine whether People's desire for control over use of their personal demographic characteristics varied more by sample source or self-selected health information increases when there is a commercial method of completion of the survey. element to the research. Additional public engagement is Click here for file [http://www.biomedcentral.com/content/supplementary/1472- required to better understand this. 6939-10-10-S3.doc] Finally, this study attempted to address the question of how much control patients would like over use of their personal health information for research. While we fully Page 11 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 portal linked to an ambulatory care electronic health record: patient gateway for diabetes collaborative care. Diabetes Tech- Additional file 4 nology & Therapeutics 2006, 8:576-586. Table 2. Regression analysis of consent choice predictors. This table 17. Hess R, Bryce CL, Paone S, Fischer G, McTigue KM, Olshansky E, presents the results of regression modelling of predictors of consent choice Zickmund S, Fitzgerald K, Siminerio L: Exploring challenges and potentials of personal health records in diabetes self-man- – scenario, survey method, health condition, sex, disclosure concern score, agement: implementation and initial assessment. Telemedi- and medical benefits score. The latter two variables were a composite of cine Journal & E-Health 2007, 13:509-517. several questions in the survey. (See Additional File 2.) The comparison 18. Kho ME, Duffett M, Willison DJ, Cook DJ, Brouwers MC: Written of reduced models 1 and 2 allows one to compare the relative amount of informed consent and selection bias in observational studies explained variance in consent choices that would be attributable to (a) using medical records: systematic review. BMJ 2009, 338:b866. health condition and (b) the combination of perceptions of medical benefit and disclosure concern. Pre-publication history Click here for file The pre-publication history for this paper can be accessed [http://www.biomedcentral.com/content/supplementary/1472- here: 6939-10-10-S4.doc] http://www.biomedcentral.com/1472-6939/10/10/pre pub Acknowledgements This study was supported by a grant from the Canadian Institutes of Health Research (Grant Number MOP-79315). References 1. Anonymous: Dorland's Illustrated Medical Dictionary 31st edition. Phila- delphia: Saunders Elesvier; 2007. 2. Singleton P, Wadsworth M: Consent for the use of personal medical data in research. BMJ 2006, 333:255-258. 3. Westin AF: How the public views privacy and health research. Results of a national survey commissioned by the Institute of Medicine Com- mittee on "Health research and the privacy of health information: The HIPPA Privacy Rule" 2007 [http://www.iom.edu/Object.File/Master/48/ 528/%20Westin%20IOM%20Srvy%20Rept%2011-1107.pdf]. 4. Willison DJ, Schwartz L, Abelson J, Charles C, Swinton M, Northrup D, Thabane L: Alternatives to project-specific consent for access to personal information for health research: What is the opinion of the Canadian public? Journal of the American Med- ical Informatics Association 2007, 14:706-712. 5. Kass NE, Natowicz MR, Hull SC, Faden RR, Plantinga L, Gostin LO, Slutsman J: The use of medical records in research: what do patients want? J Law Med Ethics 2003, 31:429-433. 6. Damschroder LJ, Pritts JL, Neblo MA, Kalarickal RJ, Creswell JW, Hayward RA: Patients, privacy and trust: patients' willingness to allow researchers to access their medical records. Soc Sci Med 2007, 64:223-235. 7. Robling MR, Hood K, Houston H, Pill R, Fay J, Evans HM: Public atti- tudes towards the use of primary care patient record data in medical research without consent: a qualitative study. J Med Ethics 2004, 30:104-109. 8. Willison DJ, Keshavjee K, Nair K, Goldsmith C, Holbrook AM, for the COMPETE investigators: Patients' consent preferences for research uses of information in electronic medical records: Interview and survey data. BMJ 2003, 326:373-376. 9. Whiddett R, Hunter I, Engelbrecht J, Handy J: Patients' attitudes towards sharing their health information. Int J Med Inform 2006, 75:530-541. 10. Page SA, Mitchell I: Patients' opinions on privacy, consent and the disclosure of health information for medical research. Publish with Bio Med Central and every Chronic Dis Can 2006, 27:60-67. scientist can read your work free of charge 11. Tocqueville A: Democracy in America New York, NY: Schocken Books; 1961. "BioMed Central will be the most significant development for 12. Hair JF Jr, Anderson RE, Tatham RL, Black WC: Multivariate data anal- disseminating the results of biomedical researc h in our lifetime." ysis 5th edition. Upper Saddle River, New Jersey: Prentice Hall; 1998. Sir Paul Nurse, Cancer Research UK 13. Willison DJ, Swinton M, Schwartz L, Abelson J, Charles C, Northrup D, et al.: Alternatives to project-specific consent for access to Your research papers will be: personal information for health research: insights from a available free of charge to the entire biomedical community public dialogue. BMC Med Ethics 2008, 9(18):. 14. Diggle PJ, Heagerty P, Liang K, Zeger SL: Analysis of Longitudinal Data peer reviewed and published immediately upon acceptance 2nd edition. Oxford: Oxford University Press; 2002. cited in PubMed and archived on PubMed Central 15. Norberg PA, Horne DR: Privacy attitudes and privacy-related behavior. Psychology & Marketing 2007, 24:829-847. yours — you keep the copyright 16. Grant RW, Wald JS, Poon EG, Schnipper JL, Gandhi TK, Volk LA, Mid- BioMedcentral Submit your manuscript here: dleton B: Design and implementation of a web-based patient http://www.biomedcentral.com/info/publishing_adv.asp Page 12 of 12 (page number not for citation purposes) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png BMC Medical Ethics Springer Journals

Consent for use of personal information for health research: Do people with potentially stigmatizing health conditions and the general public differ in their opinions?

Loading next page...
 
/lp/springer-journals/consent-for-use-of-personal-information-for-health-research-do-people-6ULDF1Iepn

References (22)

Publisher
Springer Journals
Copyright
Copyright © 2009 by Willison et al; licensee BioMed Central Ltd.
Subject
Philosophy; Ethics; Philosophy of Medicine; Theory of Medicine/Bioethics
eISSN
1472-6939
DOI
10.1186/1472-6939-10-10
pmid
19630941
Publisher site
See Article on Publisher Site

Abstract

Background: Stigma refers to a distinguishing personal trait that is perceived as or actually is physically, socially, or psychologically disadvantageous. Little is known about the opinion of those who have more or less stigmatizing health conditions regarding the need for consent for use of their personal information for health research. Methods: We surveyed the opinions of people 18 years and older with seven health conditions. Participants were drawn from: physicians' offices and clinics in southern Ontario; and from a cross-Canada marketing panel of individuals with the target health conditions. For each of five research scenarios presented, respondents chose one of five consent choices: (1) no need for me to know; (2) notice with opt-out; (3) broad opt-in; (4) project- specific permission; and (5) this information should not be used. Consent choices were regressed onto: demographics; health condition; and attitude measures of privacy, disclosure concern, and the benefits of health research. We conducted focus groups to discuss possible reasons for observed consent choices. Results: We observed substantial variation in the control that people wish to have over use of their personal information for research. However, consent choice profiles were similar across health conditions, possibly due to sampling bias. Research involving profit or requiring linkage of health information with income, education, or occupation were associated with more restrictive consent choices. People were more willing to link their health information with biological samples than with information about their income, occupation, or education. Conclusions: The heterogeneity in consent choices suggests individuals should be offered some choice in use of their information for different types of health research, even if limited to selectively opting-out. Some of the implementation challenges could be designed into the interoperable electronic health record. However, many questions remain, including how best to capture the opinions of those who are more privacy sensitive. Page 1 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 - people's view of the level of consent required for use of Background The term "stigma" generally refers to a distinguishing per- their information would vary directly with disclosure con- sonal trait that is perceived as or actually is physically, cern and inversely with perceptions of the benefits of socially, or psychologically disadvantageous. [1] Because health care and the potential for health research to of the presence of that trait, an individual may be discrim- improve the lifespan and quality of life of people with inated against – e.g. in employment or in social circles. their health condition. Health conditions will vary in the extent to which they are perceived by those individuals having the condition and This paper reports on the testing of these hypotheses. by others as being stigmatizing. Individuals with a poten- tially stigmatizing health condition may be more inclined Methods Choice of health conditions than members of the public without such conditions to experience concerns over disclosure of their personal In this study, we included seven health conditions with health information out of concern that this could result in varying susceptibility to being labelled as stigmatizing. discrimination against them. For example, a person with Four conditions – hypertension, diabetes, chronic depres- a prior history of cancer may be concerned over denial of sion and alcoholism – were used in a previous public certain employment opportunities, a mortgage, or life opinion survey of members of the public, in which insurance. A person with HIV/AIDS may be concerned respondents were asked to imagine they had one of these about social isolation because of others' concerns that health conditions. [4] In the previous study, hypertension their presence puts others at increased risk of contracting and diabetes were found to be lower-stigma health condi- the condition. tions. Chronic depression and alcoholism were found to be higher-stigma conditions. In this study, we had the There is now an emerging body of literature examining opportunity to obtain the views of people with these con- the opinion of the public regarding consent for use of ditions. To these four conditions we added: HIV, to create one's health information for research. [2-9] However, the an upper extreme category for potential stigma; breast opinion of those who have health conditions that may be cancer; and lung cancer. We anticipated that responses stigmatizing to a greater or lesser degree has been much may differ between breast and lung cancer because of a less studied. [10] There are several reasons for considering greater general public support for breast cancer sufferers – or even giving priority to – the values and expectations and a perception that lung cancer is self-inflicted through of these individuals. People who are unwell have the most smoking. at stake, both because they stand to benefit from research Survey into their health condition and because a breach of pri- vacy may potentially expose them to discrimination in The study proceeded in two phases. In Phase 1 (Novem- obtaining loans, mortgages, insurance, or employment. ber 2006 to July 2007), we surveyed individuals with the Moreover, they are under-represented in surveys targeted target health conditions. In Phase 2 (July to September to the general public and, to the extent that some health 2007), we held focus groups with a sub-sample of partici- conditions are stigmatizing, the perspective of individuals pants from Phase 1 – one group for each health condition. with these conditions may be discounted by the general The chief purpose of Phase 2 was to help inform our anal- public. Further, if we listen only to the voice of the general ysis of the findings derived from Phase 1, by providing public without attention to the concerns of this vulnera- examples of the types of concerns some people took into ble minority we run the risk of committing a form of "tyr- account when making decisions around consent. anny of the majority". [11] Setting and Participants Objective Survey participants were drawn from two sources: (1) a The purpose of this study was to examine the attitudes of pre-existing cross-Canada panel of individuals with iden- people with a range of potentially stigmatizing health tified health conditions, maintained by Harris Interactive, conditions concerning the need for consent for the use of a professional polling firm; and (2) patients recruited by their personal information for different types of observa- the investigators through family physicians' offices and tional health research, and to compare their attitudes with specialty clinics in the vicinity of Hamilton, Ontario, Can- those of the general public. We hypothesized that: ada. The reference group consisted of people recruited through Harris Interactive who had none of the target - responses would differ across health conditions; health conditions and no other serious health conditions. This group was used to approximate the response of the - some patient groups would be more permissive and oth- general public. All survey participants were 18 years or ers more restrictive than the general public; and older. Page 2 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 Participants recruited directly by the investigators were the 7 health conditions according to level of disclosure either sent a letter from the physician's office in the mail concern. This algorithm placed HIV/AIDS first, followed explaining the study or handed an information brochure by alcoholism, lung cancer, breast cancer, depression, dia- in the clinic by staff. In each case, information was pro- betes and hypertension. vided for patients to contact the investigators if interested. In total, 892 brochures were mailed to patients' homes We presented five different scenarios involving use or and 888 brochures provided to physicians and clinics for linkage of personal information for health research: (1) directly handing out to patients. use of health data for quality improvement; (2) use of the same data for marketing; (3) linkage of work/education/ All participants recruited through Harris Interactive com- income information with health information; (4) linkage pleted the survey over the internet. Participants recruited of biosamples with health information (a) assuming no through local family physicians and clinics were given the profit and (b) assuming a profit element. (See additional choice to complete the survey via internet or by telephone. file 1 for a more detailed description.) These scenarios This was done to minimize refusal due to lack of access to were identical to those used two years earlier in a series of or familiarity with the internet, particularly among older seven cross-Canada public dialogues. [13] Participants of patients. Those opting to do the internet survey used the the current study were advised that, in each scenario, same system as the Harris participants. Those opting to names, addresses and any other information that could complete the survey by telephone arranged a scheduled directly identify them were removed. Following each sce- call with telephone surveyors from Harris Interactive. nario, participants were asked which statement best They were mailed a hard copy of the core questions of the described their view (words in italics not included in sur- survey beforehand, so as to minimize differences in vey responses): response due to method of survey administration. (1) There is no need for me to know. Just use it. Sample size was calculated on the basis of the primary outcome variable – consent choice in the use of personal (2) My permission is not needed, but I want to know information for health research. This was expressed on a this is being done and a chance to say "no." [i.e. notice 5-point ordinal scale. (See Survey Data and Key Variables with opt-out] below.) This sample was determined with the primary goal of building a multivariable regression model to com- (3) My general permission is needed. This could be for pare the overall attitudes among the seven groups and the several different research studies. I could withdraw my general public controlling for several demographic and permission in future. [i.e. broad opt-in] other confounding variables. Heuristics based on simula- tion studies indicate that at least five respondents per (4) My permission is needed each time. [i.e. project- degree of freedom for each predictor variable are required specific consent] for the stability of the model. [12] We have 7 predictor variables with a total of 21 degrees of freedom, which (5) My information should not be used for this pur- would require at least 105 participants. We aimed to pose. recruit 1400 and obtained responses from 1137 (734 from Harris Interactive and 403 from physician offices The five response options above, hereinafter referred to as and clinics) with the sampling stratified by health condi- "consent choices", served as the outcome variable in a tion. We inflated our minimum sample size by a factor of multivariable regression analysis using the following as over 10 to account for potential clustering of responses predictor variables: within a patient. Therefore, the sample size was adequate to ensure the stability of the model. - demographics (age, sex, education, marital status, employment, and income) Survey Data and Key Variables We collected information on participant demographics, - health condition (one of the 7 target conditions) attitudes about privacy, disclosure concern, and the bene- fits of health care and health research at improving lon- - self-reported health (6-point scale, varying from gevity and quality of life, and the participant's health "very poor" to "excellent") conditions. Where the participant had more than one tar- get health condition, they were asked to answer the survey - scenario, and questions with only one health condition in mind. In ear- lier pilot work, we established an algorithm for determin- - attitudinal variables (disclosure concern and medical ing which health condition would take priority, ranking benefits score. These are described below. Questions Page 3 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 used to compile these scores and the scoring scheme mial logistic regression which is the more correct analysis are found in additional file 2.) but more difficult to interpret. [4] While there are many dimensions to stigma, for the pur- Initially, individual predictor variables were regressed poses of this study, we chose to focus on individuals' dis- onto consent choice using univarite analysis. Variables closure concern – i.e. concern that others may find out that met the criterion of alpha = 0.20 were then included about their health condition. For this, we asked: How con- in the multivariable regression model. The chief variables cerned would you be if: (a) your employer found out of interest – health condition, disclosure concern, and about any health condition(s) you have; (b) your health medical benefit scores – were forced into the model. We insurer found out about any health condition(s) you tested for interaction effects of disclosure concern and have; or (c) a friend other than those you told found out medical benefits scores with research scenarios and health about any health conditions you have. For each of these condition, No significant interactions were found. questions, respondents replied either: "not at all con- cerned"; "somewhat concerned"; or "very concerned". Focus Groups At the end of the survey, those participants recruited For the medical benefits scale, people replied on a 5-point directly by the investigators in the Hamilton area were scale ("strongly agree, somewhat agree", "neither agree asked if they would be willing to participate in a focus nor disagree", "somewhat disagree", or "strongly disa- group to discuss the reasoning that participants may have gree") to the following statements: (a) Medical treatments used to make their consent choices. Those who agreed can improve the quality of my life; (b) medical treatments were contacted by the study coordinator. One focus group can extend my life; (c) disease prevention programs have was convened for each health condition except for alco- shown me how to live a healthier life; and (d) medical holism, as the survey sample size was too small for this research can improve my life. group. We sampled from the pool of survey participants to provide a representative sample in each focus group with We re-scaled the disclosure concern and medical benefits regard to age, gender, and self-reported health status. scores to a 0–1 scale to facilitate interpretation of relative When making selections, investigators were blind as to attitudes toward disclosure and medical benefit across volunteers' responses to survey questions. health conditions. Six focus groups, ranging in size from 6 to 10 individuals, Dealing with potential sampling bias were conducted between July and September 2007. Focus We checked for sampling bias chiefly through two meth- groups were 90 minutes in length. Participants received an ods. Harris Interactive sampled questions from our survey honorarium of $75 following the session. These meetings in an omnibus random-digit dialled telephone survey and were moderated by two of the researchers. Upon arrival, compared the responses from the telephone survey with participants completed a mini-survey comprised of the those from their internet sample. In addition, we com- key questions from the Phase 1 survey. During the focus pared the consent choices of the reference group from this group, a structured interview protocol was used to ask par- study regarding the five scenarios with the consent choices ticipants to reflect on the survey findings regarding con- of the people who participated in the public dialogues in sent choice for the five different research scenarios. our previous study. [13] Participants were then asked to consider how the survey responses compared with their own responses and to con- Statistical methods sider the reasons for their responses and any possible var- Consent choices were analysed graphically across scenar- iance from the response pattern from the Phase 1 survey. ios and across health conditions. We also plotted disclo- Focus group participants were also asked to comment on sure concern and medical benefit scores across health the relative importance of individual control over use of conditions using a radar graph. their information and safeguards and the relation between them. This question was addressed both in the To test our chief hypotheses, we used regression analysis abstract and by giving them the opportunity to rate spe- controlling the correlation across scenarios using the cific controls and safeguards. Immediately following the method of generalized estimating equations (GEE) meeting, participants again completed the mini-survey, so assuming an exchangeable correlation structure. [14] The a comparison of the responses of focus group participants results are reported as estimates of model coefficients with those of the larger sample could be made. Discus- (with corresponding 95% confidence interval) and associ- sions were audio recorded and transcribed for analysis. ated p-value. Statistical computations used SAS, version 9.1.3 (Cary, NC.). In earlier work, we found the results of Verbatim transcripts from all focus groups were read inde- linear regression to yield equivalent results to multino- pendently by at least two members of the research team Page 4 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 and quotations were selected to exemplify the types of rea- pants through the investigators' sources is summarized in soning that focus group participants used to make their Figure 1. consent choices. These quotations are used below to help illustrate the broader survey analysis, and to provide a Demographic data window on the kinds of issues some people may address Additional file 3 summarizes the demographic character- when making consent choices. istics of study participants by sampling source. Those recruited directly by the investigators are presented by Ethics Review method of survey completion. Overall, the reference The research was reviewed and approved by the research group was younger (mean age 46 years) than participants ethics boards of St. Joseph's Healthcare, Hamilton, with target health conditions (mean 55 years), regardless Ontario, McMaster University Health Sciences REB, Ham- of sampling source. Their self-reported health was better ilton, Ontario and the Social Sciences and Humanities than those with the target health conditions. Those with REB, University of Ottawa, Ottawa, Ontario. the target health conditions completing the survey over the internet were comparable between the two sampling Results sources on most demographic variables. However, the Participants Harris sample was less wealthy and reported poorer Four hundred three survey participants were recruited health than the sample assembled by the investigators. directly by the investigators and 734 were recruited Survey participants recruited by the investigators who through Harris Interactive. Recruitment of survey partici- completed the survey by telephone were older (mean age Summary of recruitment process, Figure 1 using the investigators' sample Summary of recruitment process, using the investigators' sample. Page 5 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 62) with a greater percentage of women (68% vs. 60% in was uniformly high across health conditions. On the general public and 53% and 57% in the Harris and inves- other hand, there was substantial variation across the tigator internet samples). Telephone survey participants health conditions regarding the level of concern partici- had a greater percentage with high school education or pants had about employers, insurers, or friends finding less (56%) than those who completed the survey over the out about their health condition. Disclosure concern was internet (28%). In addition, telephone respondents were lowest for those with hypertension, diabetes, and lung more likely to be separated, widowed, or divorced (30% cancer. It was highest among those with chronic depres- vs. 14–24%). Fewer were employed (21% vs. 49–72%) sion and HIV/AIDS. We note also that disclosure concern and a greater percent had an income less than $40,000 per in the reference group was almost as high as the HIV/AIDS year (42% vs. 20–35%). group yet, they had no serious or chronic health condi- tion. Attitudes A summary score of participants' perceptions about the Consent choice benefits to be accrued from health care and health Across scenarios, consent choice profiles were very similar research (medical benefits score) and disclosure concern for all health conditions. They were also very similar to is provided in Figure 2. Each axis on this radar plot repre- the profile of the reference group. For ease of presentation, sents one health condition or the reference group. The we have combined the responses across health conditions scale on the axis for both attitudes was standardized to in Figure 3 and compared these with the reference group. (0,1) to facilitate comparison. The medical benefits score !&! +,! &"*#! )*! &!'&(   ! %# " # ! $# A Figure 2 ttitudes of survey participants across health conditions Attitudes of survey participants across health conditions. Page 6 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 Cons target health conditions Figure 3 ent choices across research scenarios – A comparison of the general population with a pooled sample of those with the Consent choices across research scenarios – A comparison of the general population with a pooled sample of those with the target health conditions. Effect of intended use and commercialization on consent choice marketing scenario (Scenario 2), and with no particular Scenario 1 involved the use of prescribing information consent choice being preferred. from the medical record for quality improvement and, for this paper, serves as the comparator scenario. Figure 3 In our focus groups, we probed for explanations behind shows a relatively permissive consent profile for this sce- the desire for greater control when commercialization was nario, with the most common response being "Just use it" involved. The responses were wide ranging. A common (35–40%). By contrast Scenario 2, which used the same sentiment was that people felt they were being taken prescribing information for marketing purposes, elicited advantage of: essentially the opposite profile, with over 40% of respondents indicating that this information should not "It was a matter of control. It's the whole idea of profit, that be used (at all) for this purpose. word 'profit'. Once I see that, I just have a sense of being taken advantage of. ... but on the other hand I wouldn't not A similar, though less marked, shift in the pattern of con- do it because it is helpful. I would just want to know." (Par- sent choice was observed when comparing Scenarios 4a ticipant 7, Diabetes group) and 4b. In Scenario 4a, individuals' health information was linked with biological samples in the absence of any "First thing I thought of was 'Well, if they're selling it for a commercialization of any discovery. In this case, the con- profit what do I get out this?' I just don't see your volunteer- sent profile was quite similar to that for Scenario 1. When ing something, if somebody else is making a profit out of it. a potential profit element was introduced through the I don't see that." (Participant 3, Lung cancer group) development of a commercial lab test (Scenario 4b), the consent profile displayed a desire for greater control over use of that information, though not as strong as for the Page 7 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 Different consent profiles for linkage with biological samples vs. you're not sure what types of people are analyzing that. income, education, and occupation, in the absence of profit It's not a white lab coat, microscope it's more of a psy- The consent profile for linkage of health information with chological marketing." (Participant 7, Depression) biological samples in the absence of profit (Scenario 4a) was similar to that for the quality improvement scenario "I trust giving out the information to most of the medi- (Scenario 1). Thirty to 40% of respondents felt it was cal profession but it's not something I would give will- acceptable to link this information without notification. ingly to people that I don't really know well." By contrast, linkage of health information with income, (Participant 3, Depression) education, or occupation (Scenario 3) was associated with a consent profile that reflected a desire for greater control "I have found when I've been asked to [do] this that my over use of the information. For this scenario, only 11– family Doctor, where I go – and I have trust in him and 15% felt it was acceptable to link this information without in his staff – and he sent a letter saying 'This is what notification and 30–43% of respondents felt their consent they are asking you to do and I believe in it.' And so I should be sought for each use before the information feel that gave me the confidence that this wasn't just could be linked. some little thing that showed up at my door. My family doctor... I have trust in him." (Participant 8, Diabe- Because this finding surprised us, we probed this in our tes) focus groups. Similar sentiments were expressed across groups, around 3 categories of explanations: These (and other) references to high trust were directed more toward doctors and hospitals or, even more person- 1. Participants felt that information about income, ally, to their doctor. While there was also a relatively high education or occupation says a lot about "who we are" trust toward university researchers (Figure 4), that trust whereas a biosample only provides information about was more tentative than for doctors. "what we are". Regression analyses "I think the simple answer is that physical tissue sample Additional file 4 compares three regression models. The full model examines the independent contributions of is just a piece of what you are, what you might be...where the rest of the information [education, health condition and attitudes toward disclosure concern income, and employment] is more of who you are. Peo- and medical benefits, controlling for scenario, survey ple are more afraid of the revelation of who you are method, and sex. Reduced model 1 removes disclosure than what you are." (Participant 8, HIV group) concern and medical benefit from the model while reduced model 2 removes health condition from the 2. Participants believed that access to information model. about education, income, and occupation can lead to the identification of the individual, but biosamples In the full model, consent choices among survey partici- cannot. pants did not differ significantly across health conditions, compared with the reference group. Increased disclosure "The work, education and your income. If somebody concern was associated with a more restrictive consent looks at a biological sample they can't look up and say choice while greater perception of the benefits of medical 'That's you or you.' You can identify somebody by all care and medical research was associated with more per- that, those other things." (Participant 6, Depression missive consent choice. When the attitude variables (dis- group) closure concern and medical benefits scores) were removed from the model (Reduced Model 1), the coeffi- "The only thing you can get from tissue or bodily fluids cients for several of the health conditions moved in the is DNA and what the disease is. They can't get infor- direction of being more permissive compared with the ref- mation about you per say. Right?" (Participant 3, erence group. In two cases (breast cancer and hyperten- Lung cancer group) sion) these became statistically significant. On the other hand, when health conditions were removed from the 3. Participants trusted that their doctor and the medi- model (Reduced Model 2), estimates for the attitude vari- cal system will ensure that the information will be ables were s . This suggests that the stronger predictor of used appropriately. consent choice was individual attitude toward medical benefit and disclosure concern than was the person's "I think with tissue you can see a real concrete connec- health condition. tion. You can imagine a scientist in a lab studying that tissue, looking for an answer. With income and work Page 8 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 1* ." " ' /2 0 / " "* / *+ #', -,..) ! "  #$ %&' () Tr Figure 4 ust in organizations Trust in organizations. Regression results also confirm our perceptions that that differed widely in potential for stigma. When exam- research involving an element of profit (Scenarios 2 and ining the confidence intervals around the parameter esti- 4b) or requiring linkage of health information with mates for the health conditions, they did not appear to be income, education, or occupation (Scenario 3) were asso- of a magnitude that would be policy relevant. These find- ciated with more restrictive consent choices. Female ings surprised us. It may be that, across health conditions, respondents were also generally more restrictive in their those who were more privacy sensitive were less inclined consent choices. Those completing the survey by tele- to participate in the study and those were more permissive phone were more permissive in their consent choices than about use of their information were more inclined to par- were those completing the survey over the internet. ticipate. This is discussed further under "Limitations" below. Discussion Chief findings Disclosure concern differed across health conditions and We had hypothesized that consent choice would differ was associated with more restrictive consent choices. By across the selected health conditions and also with indi- contrast, there was a uniformly high rating of medical vidual attitudes regarding disclosure concern and percep- benefits across health conditions. However, at the individ- tions of the benefits of medical care and research. While ual level, a lower perception of medical benefits was asso- we recognized beforehand that hypertension and diabetes ciated with more restrictive consent choices. Thus, were relatively low in stigma and that HIV and alcoholism individual attitudes – and disclosure concern in particular were likely higher in stigma, we did not assume strict ordi- – were more predictive of consent choice than was one's nality across health conditions. We found that, collec- health condition. This suggests that privacy attitudes may tively, participants with target health conditions were be formulated relatively early on and may be robust to slightly more permissive in their consent choices com- one's health condition, which may develop later in life. pared against the reference group (Figure 3). However, when we controlled for survey method, sex, and attitudes, We also observed significantly more restrictive consent we found no statistically significant difference in consent choices in research scenarios involving profit or when choice across health conditions, compared with our refer- linking health information with income, education, or ence group – this despite having chosen health conditions occupation. This is consistent with findings from our ear- Page 9 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 lier public dialogues. [13] We also note that people were frame; and (b) they consisted of individuals who did not relatively trusting with regard to linking their health have any of the target health conditions and who had no records to their biological samples – even more trusting other major health conditions. A true random sample of than with linking to income, education, and occupation. the public would have some proportion of respondents This is an interesting finding worthy of further study, with the target health conditions. given that there is a much higher likelihood of commer- cial application and intellectual property protection in Given all this, we believe our study may have under-rep- research involving biological samples and the great poten- resented those in our society who are the most privacy tial for stigmatization and discrimination. Our focus sensitive. Moreover, depending on the severity of the groups suggest that, at least in part, their lesser concern selection bias, it may be that our failure to observe a dif- was attributable to perceptions that specialized knowl- ference in consent choice across health conditions may be edge was required to interpret one's DNA and those with a result of the absence of more privacy-sensitive respond- that specialized knowledge were trusted to keep the infor- ents. These non-participants may represent varying pro- mation confidential. However, once one's genetic risk portions across health conditions. We cannot assess what profile is recorded in the health record, this information is proportion of the selection bias is due to self-selection or equally subject to misadventure as one's income, should selective approaching by their physicians. that information fall into the wrong hands. With regard to trust, we note that most of the focus group discussion of Finally, in the consumer literature, stated privacy prefer- trust was in reference to their doctors. Perhaps trust in ences are often much more stringent than those revealed one's doctor then confers benefits of trust to the research- in actual behaviour. [15] Thus, it is possible that the stated ers in the process. consent choices for use of one's personal information reported here may be different than what they are pre- Limitations pared to accept in the health care "marketplace". Based on a comparison with a subset of questions in an omnibus survey conducted by Harris Interactive, we deter- Policy implications mined that people recruited into the study were some- No one consent option was preferred by even a simple majority of survey participants in any of the scenarios what less privacy concerned and more research-friendly than the general public. Within our study sample, those (Figure 3). This wide variation in opinion is consistent who completed the survey by telephone were particularly with our earlier work with the general public and with a less privacy concerned. This observation is reinforced by recent American survey. [3] This high heterogeneity in comparison of our current findings with those of our ear- consent choices makes it difficult to put forward any one lier study involving a cross-Canada sample of the general model for consent for research use of one's personal public using random-digit dialling. In the earlier study, health information. One possible response would be to the profiling of consent choices was somewhat less per- offer individuals a menu of choices with a mechanism to missive than observed here – particularly around the link- track their choices re: secondary uses of data. While this is age with genetic information. [13] Intuitively, one would now technically possible, there are multiple challenges expect that individuals willing to participate in ongoing from a systems perspective. For example, who would bro- internet consumer panels may be less privacy concerned ker the consent process and under what conditions? Anec- and more open to participating in research. Our conven- dotal evidence suggests that physicians do not have the tional sampling through clinics, though, was also subject time for this. On the other hand, special "clinics" could be to a similar selection bias. Current ethics rules require that set up through hospitals and centres where applications a researcher not approach potential research participants are made for renewal of health cards. Information could directly, based on prior knowledge about that individual's be available through brochures and DVDs, and knowl- health condition. Individuals must first be asked by some- edgeable individuals could be available either in person one who may be reasonably expected to have access to or over the telephone. All these approaches would require this information if they would be willing to have their a substantial investment in funds and an ongoing infra- name released so they could be called by the researcher, or structure for managing these consent choices. This could if they would be willing to take the initiative themselves also be designed into the planned pan-Canadian interop- to contact the researcher. This two-stage process probably erable electronic health record system, through secure resulted in lack of access to more privacy-sensitive web-based patient portals into their health record. [16,17] research participants. These portals are web-based interactive systems that allow individuals to view their health record and communicate The reference group in this study is not quite representa- in a secure fashion with their health care provider's office. tive of the "general public", to the extent that: (a) they, Another consideration is whether consent choices should too, were drawn from the Harris internet polling sampling be totally unconstrained. There is growing evidence that Page 10 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 opt-in consent processes can result in selection biases that appreciate the rationale for restricting researchers from may affect conclusions as to various causal associations. directly recruiting patients on the basis of prior knowledge [18] Are there certain types of research (e.g. public health, about their health condition, it is ironic that, to meet eth- quality of care) for which the default assumption would ical requirements respecting individuals' privacy, this be that the information may be used for research and peo- study has likely under-represented the interests of those ple's only option would be to selectively opt out? These very people whose voices need to be heard. When address- are but two examples of issues that need to be addressed ing questions like the one we have posed here, there need through further research. to be ways to better reach those who have the most at stake. Our survey and focus group participants placed high trust in medical researchers – higher than that found recently Competing interests by Westin in the United States. [3] However, as in Westin's The authors declare that they have no competing interests. study, much of this trust was qualified. Approximately two-thirds of those expressing trust in university research- Authors' contributions DW conceived and designed the study, led the focus ers only "somewhat trusted" them. Therefore, much of this trust in researchers is vulnerable to erosion in the groups and the analyses and was the primary author of the event of a high-profile breach of confidence. Survey par- manuscript; VS, CC, LS, JR and LT all contributed to the ticipants also valued highly the ability to monitor how development of the protocol. JR led in the analyses of the their information was being used and the ability to say text. LT provided statistical guidance. JC conducted the "No" to particular uses. This "trust but verify" attitude statistical analyses. All co-authors reviewed and revised could also be accommodated through the development of drafts of the manuscript. All authors read and approved a patient portal into one's electronic health record. the final manuscript. We observed a shift toward more restrictive consent Additional material choices when research involved a commercial element. Given the increasing importance of personal information Additional file 1 to health research, and the pervasiveness of the private Box 1 – Description of Research Scenarios. Provides information about sector and commercialization in the research enterprise, the scenarios on which consent choices were made. continued public engagement would be worthwhile. This Click here for file would help researchers to better appreciate the underlying [http://www.biomedcentral.com/content/supplementary/1472- 6939-10-10-S1.doc] concerns of the public around research and profit and to determine whether these views change or are reinforced Additional file 2 with a better understanding of the role of both personal Appendix 1. Survey questions that formed the elements for (a) the dis- information and the private sector in research. closure concern scale; (b) the medical benefits scale; and (c) the con- sent choices outcome variable. This document provides the core questions Conclusions that were used in developing two key predictor variables (a measure of The public is generally supportive of research use of per- individuals' concern over disclosure of information about their health con- dition and a measure of individuals' perceptions of the benefits of health sonal health information, but do not wish to entirely care and health research for their health condition) and the key outcome relinquish control. Despite having a relatively research- variable (consent choice). friendly sample, we still observed substantial individual Click here for file variation in opinion as to the degree of control that peo- [http://www.biomedcentral.com/content/supplementary/1472- ple wish to have over use of their personal information for 6939-10-10-S2.doc] research. Opportunities for individual choice over use of one's personal health information could be designed into Additional file 3 the planned pan-Canadian inter-operable electronic Table 1. Summary of Participant Demographics. This describes basic demographic features of survey participants broken down by survey health record system. method and sample source. We divided investigator sample source into those who responded by telephone and internet to determine whether People's desire for control over use of their personal demographic characteristics varied more by sample source or self-selected health information increases when there is a commercial method of completion of the survey. element to the research. Additional public engagement is Click here for file [http://www.biomedcentral.com/content/supplementary/1472- required to better understand this. 6939-10-10-S3.doc] Finally, this study attempted to address the question of how much control patients would like over use of their personal health information for research. While we fully Page 11 of 12 (page number not for citation purposes) BMC Medical Ethics 2009, 10:10 http://www.biomedcentral.com/1472-6939/10/10 portal linked to an ambulatory care electronic health record: patient gateway for diabetes collaborative care. Diabetes Tech- Additional file 4 nology & Therapeutics 2006, 8:576-586. Table 2. Regression analysis of consent choice predictors. This table 17. Hess R, Bryce CL, Paone S, Fischer G, McTigue KM, Olshansky E, presents the results of regression modelling of predictors of consent choice Zickmund S, Fitzgerald K, Siminerio L: Exploring challenges and potentials of personal health records in diabetes self-man- – scenario, survey method, health condition, sex, disclosure concern score, agement: implementation and initial assessment. Telemedi- and medical benefits score. The latter two variables were a composite of cine Journal & E-Health 2007, 13:509-517. several questions in the survey. (See Additional File 2.) The comparison 18. Kho ME, Duffett M, Willison DJ, Cook DJ, Brouwers MC: Written of reduced models 1 and 2 allows one to compare the relative amount of informed consent and selection bias in observational studies explained variance in consent choices that would be attributable to (a) using medical records: systematic review. BMJ 2009, 338:b866. health condition and (b) the combination of perceptions of medical benefit and disclosure concern. Pre-publication history Click here for file The pre-publication history for this paper can be accessed [http://www.biomedcentral.com/content/supplementary/1472- here: 6939-10-10-S4.doc] http://www.biomedcentral.com/1472-6939/10/10/pre pub Acknowledgements This study was supported by a grant from the Canadian Institutes of Health Research (Grant Number MOP-79315). References 1. Anonymous: Dorland's Illustrated Medical Dictionary 31st edition. Phila- delphia: Saunders Elesvier; 2007. 2. Singleton P, Wadsworth M: Consent for the use of personal medical data in research. BMJ 2006, 333:255-258. 3. Westin AF: How the public views privacy and health research. Results of a national survey commissioned by the Institute of Medicine Com- mittee on "Health research and the privacy of health information: The HIPPA Privacy Rule" 2007 [http://www.iom.edu/Object.File/Master/48/ 528/%20Westin%20IOM%20Srvy%20Rept%2011-1107.pdf]. 4. Willison DJ, Schwartz L, Abelson J, Charles C, Swinton M, Northrup D, Thabane L: Alternatives to project-specific consent for access to personal information for health research: What is the opinion of the Canadian public? Journal of the American Med- ical Informatics Association 2007, 14:706-712. 5. Kass NE, Natowicz MR, Hull SC, Faden RR, Plantinga L, Gostin LO, Slutsman J: The use of medical records in research: what do patients want? J Law Med Ethics 2003, 31:429-433. 6. Damschroder LJ, Pritts JL, Neblo MA, Kalarickal RJ, Creswell JW, Hayward RA: Patients, privacy and trust: patients' willingness to allow researchers to access their medical records. Soc Sci Med 2007, 64:223-235. 7. Robling MR, Hood K, Houston H, Pill R, Fay J, Evans HM: Public atti- tudes towards the use of primary care patient record data in medical research without consent: a qualitative study. J Med Ethics 2004, 30:104-109. 8. Willison DJ, Keshavjee K, Nair K, Goldsmith C, Holbrook AM, for the COMPETE investigators: Patients' consent preferences for research uses of information in electronic medical records: Interview and survey data. BMJ 2003, 326:373-376. 9. Whiddett R, Hunter I, Engelbrecht J, Handy J: Patients' attitudes towards sharing their health information. Int J Med Inform 2006, 75:530-541. 10. Page SA, Mitchell I: Patients' opinions on privacy, consent and the disclosure of health information for medical research. Publish with Bio Med Central and every Chronic Dis Can 2006, 27:60-67. scientist can read your work free of charge 11. Tocqueville A: Democracy in America New York, NY: Schocken Books; 1961. "BioMed Central will be the most significant development for 12. Hair JF Jr, Anderson RE, Tatham RL, Black WC: Multivariate data anal- disseminating the results of biomedical researc h in our lifetime." ysis 5th edition. Upper Saddle River, New Jersey: Prentice Hall; 1998. Sir Paul Nurse, Cancer Research UK 13. Willison DJ, Swinton M, Schwartz L, Abelson J, Charles C, Northrup D, et al.: Alternatives to project-specific consent for access to Your research papers will be: personal information for health research: insights from a available free of charge to the entire biomedical community public dialogue. BMC Med Ethics 2008, 9(18):. 14. Diggle PJ, Heagerty P, Liang K, Zeger SL: Analysis of Longitudinal Data peer reviewed and published immediately upon acceptance 2nd edition. Oxford: Oxford University Press; 2002. cited in PubMed and archived on PubMed Central 15. Norberg PA, Horne DR: Privacy attitudes and privacy-related behavior. Psychology & Marketing 2007, 24:829-847. yours — you keep the copyright 16. Grant RW, Wald JS, Poon EG, Schnipper JL, Gandhi TK, Volk LA, Mid- BioMedcentral Submit your manuscript here: dleton B: Design and implementation of a web-based patient http://www.biomedcentral.com/info/publishing_adv.asp Page 12 of 12 (page number not for citation purposes)

Journal

BMC Medical EthicsSpringer Journals

Published: Jul 24, 2009

There are no references for this article.