Implementing broad consent for research with routinely collected clinical data and residual biosamples in a cancer hospital: using mixed methods approach to evaluate consent rates and patients’ perspectives

Abstract

Background

Patients are generally willing to contribute to research with routinely collected health data and residual biosamples, but transparency and being able to (to some extent) have control over data are important conditions. A broad consent procedure ensures that patients are informed, without overloading the patients with too many or repeated study-specific consents. In the context of implementation of broad consent, we investigated five aspects: response rates, whether patients felt informed and were able to reach a decision, whether the registered consent was in line with their desired consent, reasons for giving (no) consent or not responding, and suggestions to improve the procedure.

Methods

We analyzed consent decisions of 31,894 patients, recorded between May 2018 and December 2020 in a specialized cancer hospital. We also interviewed 64 patients selected from first-time visiting patients between October and November 2018 (25 with consent, 16 with no consent and 23 with no response).

Results

Consent rates were: 85.2% consented, 3.8% did not consent and 11% did not respond. The majority of the interviewees, who recalled that consent was asked, felt sufficiently informed. Those that needed more information, mostly had not (yet) read the information given to them, due to the hectic and emotional period. For the majority of our interviewees the desired consent decision matched with what was registered in the hospitals’ system. Reasons for giving consent were mostly motivational, e.g., altruism and solidarity. Reasons for not giving consent or not responding yet were mostly contextual, e.g., insufficient headspace and needing more time. Privacy concerns, e.g. mistakes resulting in data being publicly accessible, data linkage and hacking, were mentioned as well. Sometimes the reason to not give consent or not respond was based on misunderstanding, e.g. that consenting would require bureaucratic entanglements.

Conclusions

For high quality research with patient data and samples, broad consent from a large and representative patient population is essential, and patients must feel informed and be able to register their consent decision easily. Our novel consent procedure led to an 85.2% consent rate and desired consent decisions were mostly registered correctly. In addition, patients felt sufficiently informed.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12910-025-01367-2.

Background

Biosamples and patient data (including images such as CT scans) gathered during hospital care are a highly valuable resource for research to improve future care. Patients are generally willing to contribute to research, but transparency and being able to, to a certain extent, have control over data are important conditions that have to be met [1–4].

European and Dutch legislation entail several consent procedures: opt-in, opt-out and broad consent [5–8]. The Dutch Medical Treatment Contracts Act provides the possibility for an informed opt-out procedure under certain conditions, which allows patients to refuse the use of their data and biosamples for research purposes. This leads to less bias in research since few patients opt-out [9]. However, an opt-out procedure may lead to a minority of patients being aware of biosample storage and secondary use [10], and to a majority of patients not remembering having read information about research [11]. A broad consent procedure ensures that patients are informed, without overloading patients with too many or repeated study-specific consents, and is accepted by patients [12–14].

The Antoni van Leeuwenhoek Hospital - Netherlands Cancer Institute (hereafter AVL), an integrated, specialized cancer hospital and research institute in The Netherlands and an OECI accredited Comprehensive Cancer Center, implemented a broad consent (opt-in) procedure for residual biosamples and data use in research in October 2018. The AVL is the first hospital in the Netherlands to change from an opt-out procedure to a hospital-wide opt-in procedure for the further use of biosamples and data. The AVL considers this a GDPR compliant consent, since the scope is limited to cancer research and allows withdrawal of consent at any time. In the context of implementation, we investigated five aspects retrospectively: the response rate, whether patients felt they had received sufficient information to decide, whether the registered consent was in line with their desired consent, reasons for giving (no) consent or not responding, and suggestions to improve the procedure.

These aspects are important for several reasons. First, to ensure sufficient data and sample availability; the number of non-responding patients should be minimized and not biased towards certain groups. Second, patients should feel sufficiently informed. Without information, patients are unable to make an informed, autonomous decision. With too much information, patients might be overwhelmed [15]. We investigated patients’ perspectives on the process through semi-structured telephone interviews. Third, patients’ desired consent decision is reflected in the outcome of the procedure, i.e., the patients’ opinion is accordingly registered in the system. This might sound simple, but sometimes patients don’t give consent even though they do not explicitly want to withhold consent [16]. A high threshold to consent (e.g., having to post a letter) could be a reason. We investigated this by comparing the recalled and desired consent decision with the hospital’s administration. Fourth, we investigated reasons patients provide for their decision, to identify potential improvements in the process. Although motivational aspects of why people provide consent are investigated often [e.g., 17–20], little is known about organizational aspects of a broad consent procedure. To our knowledge, Barazzeti et al. and Lutomski & Manders conducted the only two studies that investigated broad consent for secondary use of data and residual samples in a hospital setting [21, 22]. Barazzeti et al. took into account the contextual and relational factors [21]. They defined contextual factors as “contingent and organisational variables that influence the practice of broad consent […], such as time, space, privacy, organisation of the work, policies or procedures […]”. Relational factors are “participants’ feelings or perceptions that describe or arise from the relationship between two or more people engaged in an ongoing or past interaction”, e.g., trust, the perception that the person who asks consent is not sufficiently trained, or satisfaction with information provided [21]. Precisely those contextual and relational factors are important, since these are the factors that hospitals can influence. Therefore, we also mapped the factors that stem from thematic analysis of the interviews in motivational, relational, and contextual factors.

Methods

Broad consent procedure in the AVL

Implementation of the broad consent procedure started on October 3rd, 2018. The preparation phase started in 2017 and included the installment of a data warehouse that can issue pseudonymized data and beforehand, checks whether the patient consented or not. An Institutional Review Board (IRB) for evaluation of studies including secondary use of data was already in place. The implementation team, consisting of professionals with expertise in research, care, legislation, communication, IT, and a representative of the board of directors, designed the consent procedure, considering feasibility, physicians time, patient preferences, and IT possibilities. They defined the extent of the consent. Consent covers most pseudonymized data and residual biosample use. The IRB of the AVL reviews each protocol and considers, among others, whether principles such as purpose limitation and data minimization are adhered to. This broad consent procedure was discussed multiple times with health care professionals (physicians, nurses, medical students, managers), researchers, the IRB, and the patient advisory board.

A Dutch patient information brochure with general information about medical research was developed and improved together with experts, an organization specialized in easy-to-understand language, and the patient advisory board of the hospital. The English version can be found here: www.avl.nl/consent. On the hospital website several frequently asked questions and examples of completed studies can be found (only available in Dutch: www.avl.nl/toestemming). Approximately two weeks before new patients come to the hospital for a first visit, they receive a brochure package via (e-)mail, including the broad consent brochure.

During their first visit to the hospital, patients receive a brief verbal introduction about the use of their health data and residual biosamples for research and are asked whether they provide broad consent for using their health data and residual biosamples in research. This is part of a 30-minute intake procedure, conducted by a medically trained professional who is not directly involved in the patients’ treatment. This employee conducting the intake registers the decision in the patient’s electronic health record. Patients can change their decision by contacting the hospital through phone, e-mail, by going to the outpatient clinic desk or (since 2025) online in their patient portal. Real-time consent rates were monitored through a dashboard, enabling intervening where necessary.

The broad consent procedure was implemented for first-time visiting patients. Patients who already had their intake before implementation of broad consent were not approached. Data of these patients were considered to be covered by the previous opt-out procedure. Under this previous procedure, patients could object to the use of data and samples in research. Information about this was described in a brochure, but patients were not actively notified. Therefore, with implementing the new procedure, all patients who fell under the previous opt-out procedure but had an appointment scheduled after May 24, 2018 were explicitly reminded about their right to opt-out by letter in May 2019 (n = 31,131). This opt-out letter was also sent to non-responders of the first months of implementation, since the new question was not always asked in the beginning and those non-responders potentially never received the question.

During implementation, some improvements were identified and immediately adjusted. See additional file 1 for an overview.

Study component 1: consent rates

Before implementing the broad consent procedure, the AVL set the goal of achieving 95% consent to ensure sufficient data availability for research and to not introduce an unacceptable amount of bias. We performed descriptive analyses of the final consent decisions of all the patients who were registered up to December 8, 2020 and had an appointment scheduled after May 24, 2018 (n = 59,813 patients) (additional file 2). This database includes the full hospital patient population, half of the database consisted of patients who fall under the new broad consent procedure (n = 28,682) and half fall under the previous opt-out procedure (n = 31,131).

Study component 2: consent rates during intake

Ideally, most patients answer the broad consent question during the intake, to prevent that patients remain non-responders unintentionally. To monitor how many patients provided broad consent during intake, the number of intakes were extracted and linked with the consent decision valid on the intake day. Only patients who did not already had a previous intake scheduled in the last 12 months were included (n = 85,807; additional file 2) since it is likely that some questions (including potentially the broad consent question) were skipped when a patient recently received an intake. Monitoring took place between October 2018 and April 2025.

Study component 3: main interview study

The goal was to include 30 to 90 patients with equal division over three groups: patients who gave consent, patients who gave no consent, and patients who did not respond (additional file 2). Though most patients gave consent, we chose to oversample the other groups since we assumed that most could be improved from the experiences of those patients (stratified purposeful sampling strategy [23]). Between October 3rd and November 9th 2018, each week, the Data desk randomly selected a maximum of five patients from each consent group of the sample of first-time visiting hospital patients who had their intake that week; in total 78 patients were selected (additional file 2). Names and telephone numbers of selected patients were provided to the interviewer, who contacted each individual by phone to request consent prior to conducting the interview. Selection was stopped when saturation (i.e., no new topics emerged during the last couple of interviews) was achieved.

Telephone interviews were conducted by three interviewers (MB, AVO, SR) using a semi structured interview guide (additional file 3). Patients were called, a maximum of three times on different days and times, one to two weeks after they were asked for broad consent. Patients who could not be reached were not replaced with another patient. Interviewers were blinded to the patients’ broad consent decision, in order not to unintentionally influence or steer the patient.

Relevant parts of the interviews were transcribed verbatim and analyzed in R (version 3.6.0) using RQDA (R-package, version 0.3-1). Standard inductive qualitative methods were used: a set of codes was derived from the interview guide and literature, which was refined and expanded using ‘in vivo coding’, meaning that the codes were based on themes that arose from multiple readings of the transcripts [24]. All transcripts were coded independently by MB and SVS; differences were discussed (MB, SVS, SR) to reach consensus and establish the final set of codes. Answers to quantifiable questions were quantified by MB and checked for correctness by SVS.

Study component 4: add-on interview study, zooming in on patients who consented for the research blood biobank but not for residual biosamples and data

In 2021, an add-on interview study was conducted with patients who gave broad consent for drawing blood for the research blood biobank but who did not give broad consent for residual biosamples and data (additional file 2). The objective was to obtain insight into reasons for the different responses and identify improvements for the broad consent procedure. For question and methodology, see additional file 4.

The results of both interview studies were mapped to the framework of Barazzetti et al., structuring the factors into motivational, relational and contextual categories [21]. The initial mapping was done by MB and checked by SR.

Results

Study component 1: consent rates

Consent rates of the broad consent procedure were: 85.2% consented, 3.8% did not consent and 11% did not respond (n = 28,682). Consent rates of the full hospital patient population, including patients who already had their intake before implementation of broad consent and thus were covered by the previous opt-out procedure, were 92.4% consent, 2.3% no consent and 5.4% no response (n = 59,813).

Study component 2: consent rates during intake

Consent rates registered during intake went up gradually during 15 months after the start of the procedure, and again after a drop during COVID-19 (Fig. 1). During the last months of this analysis (between January 1st 2025 and April 9th 2025), of 3784 patients, 88.2% consented, 3.1% did not consent and 8.7% did not respond. The figure does not contain the final consent decisions: patients who give (no) consent after intake are still registered as ‘no response during intake’ in this figure. This includes patients who received a short intake by phone during the first COVID-19 lockdown in March 2020 and therefore received the consent question during follow-up, causing a temporary increase in no response during intake.

Fig. 1.

Trends in consent rates during intake, measured each week between October 2018 and April 2025. The total number of patients was 85,807. T1: Oct - Nov 2018: Main interview study. T2: March 2020: First COVID measures (lock down). T3: May-June 2021: Add-on interview study

Study component 3: interview results

Of the 78 selected patients, 64 agreed to participate in the interview (response rate 82.1%) (additional file 2). Of the 64 interviewees, 25 had initially consented, 16 had not consented, and 23 had not responded, according to hospital registration. Interviews lasted between 5 and 15 min. In two cases, the interview was conducted with (or with help from) a representative, who had joined the patient during intake as well.

Do patients have sufficient information to decide (if not: which additional information is needed)?

Of the 32 patients who remembered the consent question and were asked if they had sufficient information to decide whether they wanted to provide broad consent for research, the majority (21 patients) agreed. Some patients who did not had sufficient information, received the brochure during the intake instead of before. Interviewee 39: “I have indicated that I will first read that brochure and that I first want to know how and what.”. Although patients were able to make their decision later in time, not all patients did that. Interviewee 6: “We took it [the brochure] with us with the idea that I was going to look at it, and then have an opinion about it at the next appointment. And so I didn’t actually do that.”.

Of the 37 patients who were asked if they read the brochure, nine reported that they had not. Patients indicated that this was due to the hectic and emotional period they were in. Interviewee 23: “I just want to read the brochure completely first. But there are many days that we are in the hospital and this is not the priority.”. The amount of information that is given to patients was also mentioned. Interviewee 37: “I have to be honest, I saw a lot of text and it wasn’t that I thought: I’m going to read it right now. […] But it also has to do with me getting it at the same time as the diagnosis, so it’s not a priority for me to read. I get a lot of brochures.”. However, not (remembering) reading the brochure does not necessarily mean that patients felt insufficiently informed. Of the 21 patients who felt sufficiently informed, one patient had not read the brochure and eight did not remember whether they read the brochure or not. Reasons for consenting without (remembering) reading the information included already having knowledge about or experience with health care and research, or being supportive of research.

When patients indicated that they did not have sufficient information, they were asked what information was missing. Most patients answered that there was probably no information missing, but that they just had not read the brochure yet. Interviewee 73: “Well, if I’d read the brochure, I probably would [have sufficient information]. But I didn’t do that.”. Only two patients had specific questions or topics that they wanted to be informed about. Those related to whether it requires extra biosamples, traceability and hacking. These specific questions are in fact addressed in the brochure, indicating that it is likely that these patients would have sufficient information after reading the brochure.

Is the patients’ recalled consent decision registered correctly in the system, and does this match with the desired decision?

We compared decisions patients thought they had made (recalled decisions) with what is registered in the system (registered decisions) (Table 1). Of 64 interviewed patients, 39 recalled their consent decision. The recalled and registered decisions matched for 26 patients (67%). At the end of the interview, we asked each patient: “Imagine, if broad consent was asked today, what would you answer? Would you give consent, no consent, or would you not know which decision to make?”. We considered this answer the desired decision; and for 61% (39 out of 64) of the patients the desired and registered decision matched (Table 1).

Table 1.

Registered decisions versus desired and recalled decisions by consent group

Consent group N = 64		Desired decision			Recalled decision
		Consent	No consent	I don’t know	Consent	No consent	No response	Don’t remember/ was not asked
Registered decision	Consent	23	2	0	14	2	1	8
	No consent	5	10	1	0	8	6	2
	No response	17	0	6	4	0	4	15

The diagonal cells with bold numbers show the number of patients whose desired or recalled decision matched the registered decision. Be aware that the sampling is not representative of the total response but a stratified samples selection for the interviews by consent/response group (see methods)

Seventeen patients who were registered as non-responders in the system and wanted to give consent during the interview. Almost all those patients could not recall having received the brochure or question. For the rest of the patients, the mismatch was not a result of a conscious change of mind, it was often unclear why patients recalled or desired something else than what was registered. Sometimes misunderstandings came to light: patients confused secondary use of data and biosamples for research with research entailing extra work or hospital visits (e.g., trials).

Reasons to give consent, no consent, or not decide

Twenty-three of 25 interviewees with registered consent provided reasons to give consent. Twenty-two interviewees who registered no consent mentioned nonetheless reasons to give consent. The reasons for consent showed altruism (interviewee 78: “If I can help someone with it, great.”), solidarity (interviewee 2: “Science is always a step forward. And you do that with all together.”), and reciprocity (interviewee 4: “I’m not paying for it myself. There is insurance for that. You should be thankful for that, I think. And so you should, if possible, do something in return.”). Patients explicitly mentioned cancer and the importance of research in their arguments. Interviewee 69: “I am all for more research on all cancers in the world.”. Interviewee 38: “Well, I want cancer eliminated from this world, so I’m willing to do anything. If it helps, for others, yes. I know what it’s like to have cancer.”. For some, giving consent went without saying. They considered consenting the norm or a moral duty. Interviewee 1: “I am the curious type, and I also think that humanity has a duty, more or less, to explore the universe, the roof that surrounds us.”. Trust was rarely mentioned explicitly as a reason to provide consent. Interviewee 21: “It is handled in a prudent way, I was told. It’s based on trust that you give consent.”.

Thirteen of 16 interviewees who did not give consent and 11 of 23 that did not respond according to the registration provided reasons for their decision. Reasons included insufficient ‘cognitive bandwidth’ to think about this topic during a hectic period and the timing of asking consent (for seven out of 13 for no consent and three out of 11 for not responding). Interviewee 10: “So, at some point your plate is full. Not now. Then someone else can make himself available. No matter how useful research is, we endorse that.”. Interviewee 51: “My head is not in it.”. Interviewee 72 did not make a decision and explained: “If you come for the first time, […] it disappears among all… That first time makes such an impression that you put it [the brochure] in your bag and you don’t really look at it anymore.”.

Privacy concerns were also mentioned in both groups (six out of 13 for no consent and one out of 11 for not responding). Concerns related to mistakes resulting in data being publicly accessible, data linkage, hacking, and the feeling that there is already so much known about the patient and consent was never asked before. Interviewee 67: “Well, because I’m not convinced that that data isn’t spread.”. International transfer of data was voiced as a reason for concern and therefore a reason not to provide consent. Interviewee 8: “If I’m correct, the decision was ‘no’, because the brochure, one of the comments was that it was all anonymous and stuff, but that it could be used or passed on to foreign companies or institutions, and there, I thought, was also mentioned that there is no privacy control.”. This quote also shows that sometimes, the reason for giving no consent or no response was based on a misunderstanding; this was the case for four more patients.

Another example of misunderstanding was that giving broad consent would require extra work or bureaucratic entanglements. Interviewee 64: “We don’t have to go to the Antoni van Leeuwenhoek hospital [anymore] and then [when you consent] I think you had to go there more often, and we don’t really live nearby.”. Interviewee 32: “I didn’t like the idea of ​​potentially having to deal with bureaucratic entanglements that I couldn’t foresee at the time, or that I didn’t feel like doing.”. Interviewee 73 (no response): “I thought maybe I would have to do all sorts of other things and stuff, and of course I’d rather not. Or well, of course… No, I would rather not, because yes, my cancer is very clear, I thought it was clear what should be applied to it in terms of treatment, so that is not necessary.”. One interviewee mentioned she felt like she was just a research subject, which led her to give no consent.

The most common reason for not making a decision (not responding) yet, mentioned by six interviewees, was related to wanting to think about it more. Interviewee 6: “Well, it was short, so I wanted to think about that before making a decision.”. A couple of interviewees felt they needed more information before making a decision. Interviewee 23: “I just don’t have enough information yet. I just want to read the brochure completely first.”. One interviewee wanted to talk about the subject with her kids before deciding.

Patients’ suggestions to improve the procedure

During interviews, patients were asked for suggestions to improve the procedure. The only suggestion that was given related to the timing of the question. As mentioned above, patients sometimes have insufficient ‘cognitive bandwidth’ to think about this topic during their first day in the hospital. Interviewees were also asked for suggestions to improve the brochure. Although some interviewees could not remember the brochure well, the majority of the interviewees indicated that the brochure was clear. One potential improvement that was mentioned a couple of times was increasing the font size.

Study component 4: results of the add-on interview study, zooming in on patients who consented for the research blood biobank but not for residual biosamples and data

In April 2021, 253 patients had not given consent for residual biosamples and data but had given consent for the blood biobank. Of the selected 50 patients, 25 agreed to participate in the interview (response rate 50%) (additional file 2), and saturation was achieved. Interviews lasted between 6 and 15 min. Two interviews were conducted in English because these interviewees did not speak Dutch. Language issues appeared to be more prominent in the add-on study than in the main study.

There were differences between registered and recalled decision (Table 2). For two of the 25 interviewed patients, the different responses to asking consent for the blood biobank and for the use of residual biosamples and data was a conscious decision. One patient indicated that the word ‘images’ (in Dutch ‘beeldmateriaal’, which applies not only to pictures but also to footage) scared her. Interviewee 21: “It was not clear to me what kind of images were meant. It got a little too personal, a little too intimate, and that scared me, and I said no.”. This patient did not receive or read the brochure (which does give examples of what kind of images could be used) before coming to the hospital. If consent would be asked today, she was still not sure if she would provide this, because: “A tube of blood doesn’t have that much of a thing [value] for me, first of all it is added again, it replenishes itself. […] Several pieces of my cervix have already been removed […]. And I find that quite difficult psychologically, because I literally keep losing pieces of myself.”. The other patient indicated that a couple of tubes of blood more would not matter, and she would hope that this would lead to better treatment and that this was in her own interest as well. Research with samples from her malignant tumor was vaguer to her, she needed more explanation about the importance of this type of research and whether it would benefit herself.

Table 2.

Registered decision versus desired and recalled decision

Consent group N = 25	Desired decision				Recalled decision
	Consent	No consent	I don’t know	Unclear from interview	Consent	No consent	Don’t remember/ was not asked	Unclear from interview
Registered no consent*	19	1	1	4	7	2	15	1

*The add-on study only included patients who, according to the registration, gave consent for the blood biobank but did not give consent for the use of residual biosamples and data

In four interviews, it was unclear whether patients made a conscious decision to consent for one but not for the other, since the patients did not seem to understand the concept of scientific research. Patients confused scientific research with diagnostic research, autopsy, and organ donation, despite multiple explanations of the differences (see interview script in additional file 3). For example, interviewee 28 stated: “If something happens to me, my kids have to decide what happens to my body.”.

In 15 interviews, patients indicated that they did not remember the question for using samples and data for research. All indicated that they would (probably) give consent if it would be asked now. In seven interviews, patients indicated that they thought they had given consent for both.

In Table 3, the factors that stem from thematic analysis of both interview studies are mapped in the framework from Barazzetti et al. [21].

Table 3.

Mapping of factors that stem from thematic analysis of both interview studies

	Motivational factors Those that give sense to the decision of giving or withholding consent	Relational factors Interviewees’ feelings or perceptions that describe or arise from the relationship between two or more people engaged in an ongoing or past interaction	Contextual factors Organizational variables that influence the practice of broad consent
Consent	• Altruism • Solidarity • Reciprocity • Importance of (cancer) research • Norm	• Potentially: trust	• -
No consent	• Privacy concerns • Feeling like just a research subject	• Implicit: potential lack of trust	• Timing • Mental state (insufficient ’cognitive bandwidth’) • Consenting means extra work (misunderstanding) • Implicit: language issues
No response	• Privacy concerns	• Implicit: potential lack of trust	• Timing • Mental state (insufficient ‘cognitive bandwidth’) • Consenting means extra work (misunderstanding)

Discussion

For broad consent to be successful from the research perspective, i.e. avoid bias and secure sufficient data availability for research, a large majority of patients must provide consent. Our novel (opt-in) broad consent procedure led to an 85.2% consent rate. Taking into account the complete hospital patient population, including patients that fall under the previous opt-out procedure who were explicitly informed about this, data of 92.4% of the hospital patient population can be used for research. This did not meet the 95% goal set by the AVL but is above the average weighted consent rate of 84% for opt-in procedures as investigated by De Man et al. [9]. The consent rate could be further improved by contacting non-responders, e.g. by calling them or sending them a letter or email. However, we deem it unlikely that this will raise the consent rate to 95%, because we expect that only a part of the non-responder group would provide consent when asked later in time, under different circumstances. Based on the current study, we cannot conclude whether the 85.2% consent rate would lead to insufficient data or biased research results.

For a successful procedure from patient and ethical perspective, patients should feel sufficiently informed to decide freely. The majority of the interviewees who remember the consent question felt sufficiently informed, despite oversampling of the group that gave no consent and no response. If they needed more information, this was mostly because they had not yet read the brochure given to them.

Importantly, the patients’ desired consent decision should also be reflected in the outcome of the procedure, i.e., the patients’ true intention is registered in the system. For the majority of our interviewees, the desired and registered decision matched. However, one patient reported feeling sufficiently informed despite not having read the brochure, and eight patients could not recall whether they had read it, yet still felt adequately informed. Prior research indicates that failing to read consent materials is common [25–27]. Achieving full comprehension among all patients therefore appears unrealistic, even when information is offered through multiple channels. This raises the ethical question whether consent is valid when patients do not take in information. Kongsholm and Kappel conceptualize this issue as the distinction between trust-based and information-based consent [28]. They argue that consent grounded in trust is not morally inferior to information-based consent, as it can safeguard the four moral values essential for valid consent (autonomy, voluntariness, non-manipulation, and non-exploitation) to a comparable extent. Crucially, however, such trust must be well-placed, and participants should not be exposed to unanticipated risks. In the context of data and sample reuse, these risks predominantly relate to privacy rather than physical harm, and are relatively small. These privacy risks are mitigated as much as possible, e.g., through rigorous review of research involving patient data and technical measures.

When the desired and registered decision did not match, the cause of the mismatch often remained unclear. However, several potential reasons were discovered. Patients confused the different consent procedures (e.g., for the blood biobank or trials), and patients were unaware of the information and question. These are the main reasons for a mismatch in Lutomski and Manders’ study as well [26]. Some patients potentially did not receive the question, due to the very early phase of implementation. Others potentially did receive the information and question, but do not remember, which is in line with previous research indicating that understanding and recall of the consent procedure is generally low [29, 30]. Also, hospitals employees might have registered decisions incorrectly. In addition, patients might have given socially desirable answers during interviews, indicating that they had given consent, even though they had not. To minimize this, we clearly mentioned that we wished to evaluate the new procedure (checking whether the broad consent question is asked in the right way), that there are no right or wrong answers during the interview and that the patient will not be recognizable in the scientific paper. Additionally, we used telephone interviews instead of face-to-face interviews [31].

Our study led to the identification of several factors that affected the procedure. We structure these according to the categories defined by Barazzetti et al. [21]. In addition, we highlight (potential) strategies that the hospital used or could use to improve the procedure.

Motivational factors

Interviewees mentioned mostly motivational factors as reasons for giving consent: altruism, solidarity, reciprocity, seeing the value of research and (to a lesser extent) voicing that it should be the norm to contribute to research. Interviewees explicitly mentioned cancer and emphasized the importance of research for this disease. Previous research shows that cancer patients might be more willing to share (sensitive) data for research than patients with other diseases [32–34]. Since the AVL is a cancer hospital, we suspect that the patient population of this hospital is more willing to give consent.

Patients mentioned privacy concerns related to data protection. It is important to communicate clearly how data are protected. Previous research has also shown that the assurance of individuals’ confidentiality is a crucial condition for trust, leading to more support for research [1, 2, 19, 35–40]. Indicating that participation in research is considered the norm or even a moral duty is not a commonly reported motivation for providing consent in previous studies. A possible explanation for why some participants in this study perceived consent as self-evident is the strong research-oriented identity of the AVL hospital. It is well known that the hospital conducts research. The hospital’s central hall prominently displays multiple messages and signs highlighting its research activities, which may heighten patients’ awareness and prime them to view research participation as routine or expected. Moreover, Dutch citizens rated science as the most trusted social institution and scored the influence of science on society generally positive [41, 42]. This might lead to some patients’ perception of research participation as the norm.

Relational factors

In our study, trust was rarely mentioned explicitly as a reason to give consent. Thus, it is possible that trust is a condition for giving consent, rather than a reason. A lack of trust was never explicitly mentioned during interviews, but some patients voiced privacy concerns, which could stem from a lack of trust that the data could be kept sufficiently safe [36, 43]. The concerns mostly related to not being comfortable with (international) data sharing, which could signal not trusting third parties. Previous research highlights that relational factors such as trust can be addressed through transparent governance mechanisms. One study advocates for oversight committees representing expert but also patient values and expectations, that should carefully manage conflicts of interest and ask public accountability from those who access data and biosamples [38]. The hospital’s IRB includes a patient representative. This is currently not explicitly mentioned in the information, adding this might help reinforce trust. According to a cross-country study, transparent information about who will benefit from data access is the most trust-increasing measure [44]. This information is already in the brochure.

According to international legislation and declarations1, consent should be given freely. The dependent relationship between patient and hospital makes it difficult to completely eliminate all pressure to give consent, even when consent is asked by an independent recruiter that is not linked with care [30]. In the AVL, the consent question is integrated into the regular intake and asked by hospital employees who are linked to care, which could lead to pressure to give consent. This aspect was not mentioned by patients, potentially, because this effect might be unconscious. Additionally, the brochure explicitly mentions that not giving consent would not impact the patients’ care.

Interviewees mentioned the stress on their first day in the hospital. During outpatient clinic visits in the implementation phase, we learned that some of the health care professionals skipped the broad consent question when they deemed the patient too emotional to be able to answer the question. Although adjusting the process to the patients’ needs is important to some extent, this protective behavior might lead to gatekeeping, introducing bias in research [30, 45].

Contextual factors

Interviewees mentioned mostly contextual factors as reasons for giving no consent or no response yet: during intake, some patients have insufficient ‘cognitive bandwidth’ to grasp information that does not directly affect their treatment or health. However, previous research shows that asking later in the process of care is not necessarily a more convenient moment, since the patients’ mind can still be occupied with treatment and health [30]. Additionally, the AVL is a cancer hospital, and some patients’ health may deteriorate quickly, making a decision more difficult.

Although some interviewees wanted to register their decision after instead of during intake, this was sometimes forgotten. Unintentionally, patients remained registered as non-responder. It is important to remind patients of the consent question [9, 22]. This reminder is important not only for non-responders, since patients who have made a decision sometimes forget that they did, for example, because participating in research can be considered a small detail compared to having cancer [46]. Ethically, this is not optimal. Research participants should be able to reassess their consent for the duration of the research [47]. However, it is likely that not all patients would appreciate recurring notifications from the hospital after completing the treatment, for example, because they do not want to be reminded of the negative period in their life. We deem providing clear insight into the decision the patient made, for example by enabling patients to change their decision in the patients’ portal, the most appropriate strategy.

Previous research shows that awareness of the use of health data in research is low [48]. We suspect that more general awareness of health data reuse in society might help. This way, people can think about this subject before they become a patient and the consent question would come less as a surprise.

In the AVL, it took little over a year before non-response during intake dropped below 10%, which could be a sign that the consent question was not always (clearly) asked in the beginning. This is despite the fact that in the AVL, every new patient undergoes a 30-minute intake, which made adding the broad consent question to this intake relatively easy. Implementation requires more work, awareness, and collaboration in hospitals where there is not yet a convenient moment to ask consent. Other researchers investigated a digital procedure that might be more feasible, for example, electronic consent [49], using tablets in the waiting room to provide information and register consent [50] or a video procedure that patients can complete on their own [51].

How to include patients who have difficulties understanding the question?

One important aspect overarches the framework of motivational, relational, and contextual factors. Four interviewees did not seem to understand the concept of scientific research, despite the interviewer trying multiple times to explain. This might be due to having low (health) literacy, not speaking the language well, or cognitive problems due to old age, the disease, treatment side effects, etcetera. Noticeable language issues, i.e., not speaking Dutch and/or English, were more prominent in our add-on study (in which patients were interviewed who did give consent for the blood biobank but did not give consent for using residual samples and data) than in the main study. This could potentially be a reason for the different consent decisions.

Comprehension is a key construct in informed consent [15]. Including patients that have difficulty understanding consent, without their informed consent, reduces their autonomy. However, structural exclusion will lead to biased research, potentially harming these vulnerable groups when research is translated into improvement of care. Diverse datasets are therefore highly important [52]. Cheah et al. investigated a broad consent procedure using an illustration of data sharing for health research among a relatively poor and low-literate population in Thailand, showing that it takes considerable time to explain health research to this group [53]. Currently, in the regular procedure in the AVL, there is not much time to explain broad consent to patients. However, taking more time will not solve this issue, since this too can be burdensome: patients go through an intense period, their ‘cognitive bandwidth’ is limited, and scientific research is not a priority. Previous research shows that animated videos improve the ability to understand information [51], and the AVL implemented an animated video in the waiting room, but this is not sufficient for all patients. There is no consensus about how much information low-literate patients should receive [53].

When the required capacity to understand information and make a well-considered decision is not fully present, asking consent will not reach its full purpose. Relying on the decision can then undermine the persons’ dignity and expression of will. In those cases, recognition of the (partial) incapacity may serve human dignity better [54]. It is very important to obtain more insight into how to meaningfully inform those patients and their representatives to prevent the structural exclusion of these patient groups and investigate how to responsibly include these patients in research when we cannot just rely on informed consent.

In the literature, the following strategies to include patients with difficulties understanding the question, are mentioned. The level of vulnerability should be determined, for example, by using the Kipnis’ taxonomy that not only mentioned cognitive vulnerability, but also juridic, deferential, medical, allocational and infrastructural [55]. If the person is not able to provide meaningful informed consent, consent of a legal representative should be sought. The person should still receive information about research and their assent should be sought to the extent of the person’s capacity [56]. An ethics board or Institutional Review Board should review whether extra measures must be taken before issuing data for research. The risks of the research must be minimized and outweighed by the potential individual benefit [56]. Risks can be minimized by paying extra attention to preventing traceability to an identifiable person with techniques like anonymization, pseudonymization, multiparty homomorphic encryption, etcetera [57]. Additionally, when data of vulnerable groups are used, community engagement is important. Examples are including representatives of the vulnerable group in the ethics board or seeking additional consent at group level through cultural authorities [58].

Strengths and limitations

Opt-in or opt-out consent procedures are subject of ongoing debate, particularly given that the European Health Data Space (EHDS) poses an opt-out (raising the possibility of reverting to an opt-out system) but leaves room for EU Member States to regulate an opt-in for specific types of health data. For the member States who decide to regulate an opt-in, the results of this paper provide insight into several aspects of how to implement such a procedure.

This study is the first mixed method study to investigate a broad consent procedure during implementation in a hospital setting. By quantitative analysis, taking into account the complete hospital patient population, we have an accurate view on the effect on data availability. By interviewing patients, we were able to investigate the patients’ perspective on the procedure and identify improvements to better align the procedure with patients’ wishes and thoughts. The limitation of this method is that we rely on the patients’ memory: we do not know what exactly happened and why differences between what the patient recalls and what is registered in the system of the hospital occurred. This could not have been clarified during the interviews, the interviewer was blinded to the registered consent decision and could therefore not explicitly ask why the patient’s decision was different from what was registered in the hospital system. Second, the interviews were intentionally kept brief to minimize the burden on participants, which limited the opportunity for a more in-depth exploration of patients’ experiences and perspectives. Additionally, we have not systematically collected data from employees who asked for consent.

To monitor how many patients provided broad consent during the first intake, aggregated consent rates were extracted by the hospitals’ data desk. Whether consent was indeed registered at this first intake or later could not in all cases be extracted correctly from the hospital system. In theory, the appointment should be scheduled just once, but in practice it is sometimes scheduled more often. Additionally, when patients who received an opt-out letter in the past and received an intake after (e.g. because of a new tumor), the employee might not have asked the broad consent question since there was already a response registered. For these intakes, the original response (opt-out or no opt-out) is included in the data. These patients could not be excluded from these data. However, the data are sufficiently accurate to provide insight into the trend lines.

Conclusions

Our study investigated whether a broad consent procedure for using routinely collected health data and residual biosamples in research works effectively in a cancer hospital and what lessons could be learned from the implementation process. Our novel consent procedure led to a 85.2% consent rate. Taking into account patients that fall under the old opt-out procedure who were explicitly informed about this, data of 92.4% of the patient population can be used for research, which is close to the 95% goal that the hospital has set for itself. The field would benefit from future research into whether this consent rate is sufficient for research and how much bias it is likely to cause. The majority of the interviewed patients felt sufficiently informed, despite oversampling of the group that gave no consent and no response. If they needed more information, this was mostly because they had not yet read the information given to them. For the majority or our interviewed patients, the desired consent decision and the decision that was registered in the hospitals’ system matched. We conclude that the procedure works effectively and identified lessons learned that could help other hospitals that want to implement a broad consent procedure.

Abbreviations

AVL

Antoni van Leeuwenhoek Hospital

IRB

Institutional Review Board

Authors’ contributions

MKS, SR, AB and HvL conceived the studies. MB, SR, IK and AvO collected the data. MB, SvS and SR analyzed the data. MB, SR and MKS interpreted the data and wrote the first version of the manuscript. All authors played a role in the implementation of the consent procedure, read and commented on the draft manuscript, and approved the final version.

Funding

Source of support: This research was supported by an institutional grant of the Dutch Cancer Society and of the Dutch Ministry of Health, Welfare and Sport.

Data availability

Given the comprehensive nature of the dataset of study component 1, which includes the entire patient population at the Antoni van Leeuwenhoek hospital between October 3, 2018, and December 8, 2020, the data are not publicly available. However, data can be requested through the data request procedure at the Antoni van Leeuwenhoek hospital.The datasets analyzed in study components 2, 3 and 4 are (without direct identifiers) available from the corresponding author on reasonable request. Transcripts will not be shared because of potential identifiability.

Declarations

Ethics approval and consent to participate

Study component 1 included the complete hospital patient population. The study was approved by the IRB of the Antoni van Leeuwenhoek Hospital on October 3rd, 2019 (IRBd19-242) as evaluation of the new consent procedure. To achieve this, all patients were included, irrespective of their consent decision (consent was waived). All data were issued pseudonymized, and no directly identifiable information was used. For study component 2, only aggregated percentages per week were extracted by the Data desk, therefore, no consent was required. Study components 3 and 4 were also reviewed and approved by the IRB (references IRBd18076 and IRBd21-011). Audio recordings of interviews were made with prior verbal consent from participants. All but five interviewees (three in the main study, two in the add-on study) consented to recording. Those who declined allowed detailed note-taking instead in order to analyze their answers, and we did not quote these participants. Research data were stored in a pseudonymized way, and communication data (name and telephone numbers) were stored separately and removed when no longer necessary. Data were processed in accordance with privacy regulations (COREON, 2022; Dutch Implementation Act GDPR, 2018; General Data Protection Regulation, 2016; Medical Treatment Contracts Act, 1994). Clinical trial number: not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.