Examining compliance with ethical standards for animal research: is there a need for refinement? A qualitative study from northern Europe

Abstract

Ethical guidelines for research on animals such as the 3Rs (Replacing, Reducing, Refining) and positive harm-benefit evaluations are anchored in EU Directive 2010/63. In this qualitative study we investigated how ethical guidelines interact and/or compete with other considerations when animal research is planned. Four focus groups consisting mainly of researchers involved in animal use were conducted in four Northern European countries and findings were analysed thematically with the support of NVIVO. Practical issues and the importance of doing good science were dominant topics. Practical issues could not easily be separated from the goal of good science. Participants expressed concerns which accord with the core-values of the 3Rs, but in one group they explicitly referred to the 3Rs as a concept. Conflicts between reductions in animal numbers and the risk of creating unreliable results were addressed. They also criticized the practice of using more animals to improve statistical figures to get results published in highly ranked journals – a finding we believe is new. The main conclusion of this study is that ethical values could not easily be separated from the goal of producing good science. Whereas policy makers seem to expect researchers to explicitly take ethical considerations into account, we found that their ethical thinking is mainly manifested as an implicit part of methodology and design. We don’t see this as a problem as long as the underlying core values are implicitly respected, or promoted, in the relevant experimental practice.

Introduction

Animal research raises ethical issues. In most societies there are conflicting views about under what conditions it is ethically justifiable.^1–7 In Europe, this kind of research is regulated through a common directive implemented in different national legislations that defines minimum standards for animal use in research, teaching and testing and is designed to reflect and enforce widely accepted ethical standards.

Among the criteria defined in EU Directive 2010/63⁸ are the principles that all experiments must implement the 3Rs ⁶ and be based on a positive harm-benefit analysis. ⁵ The 3Rs aim to reduce negative impacts on sentient animals either by Replacing animals, Reducing their number, or by Refining procedures to minimize or eliminate harm to them. ⁶ Directive 2010/63⁸ article 1 explicitly employs the language of the 3Rs:

This Directive establishes measures for the protection of animals used for scientific or educational purposes. To that end, it lays down rules on the following: (a) the replacement and reduction of the use of animals in procedures and the refinement of the breeding, accommodation, care and use of animals in procedures.

The 3Rs – and especially refinement – have had a central place in the evaluation by the competent bodies (IACUC, Ethical committees).^9,10 Although the directive defines the role of advisory bodies on the 3Rs, the researchers are mainly responsible for ensuring that the 3Rs are properly applied. The 3Rs have been investigated in earlier studies. The findings were mixed,^3,11–13 and differences between scientists’ views and policy goals were identified. ¹⁴

Harm–benefit assessment is another important,^5,9,15–19 mandatory ⁸ part of project evaluation.^20–23 The harm–benefit analysis brings the particular project into a context where the harm to animals is justified by expected benefits. ⁵ Harm–benefit evaluations resonate with the public^1,2,24 as well as policymakers,^8,25–27 although they can be difficult to compare and weigh against each other.^15,18

Research activities face time and resource pressures, and researchers must define priorities when they plan experiments. There is a risk that the aspiration to apply high ethical standards will compete with other interests or needs. Guidelines requiring researchers to continue to adhere to ethical standards have been defined both for the planning phase ²⁸ and the reporting of animal research.^29,30 Several bodies have defined codes of conduct describing acceptable research practice.^31–33 Compliance with these codes is crucial if scientific integrity and credibility is to be maintained. ³¹

The aim of this qualitative study, based on interviews with researchers in four Northern European countries, was to deepen our understanding of the ways in which, in practice, ethical guidance, and in particular that given by the 3Rs, interacts and/or competes with other considerations when research involving animal use is being planned.

Methods

Four focus group interviews were performed in four countries (the Netherlands, Denmark, Sweden and Norway) between 2011 and 2013. The interviewees were recruited through contacts who volunteered to organize the events, and no incentives to participate were offered. In total 21 interviewees participated, 11 males and 10 females. The interviewees’ contributions were anonymized. Their age was between 25 and 56 years (average 35, median 34 years) with between 0 and 30 years of experience in animal research (average 8.4, median 5 years). Eight participants identified themselves as PhD students, two as post-docs, five as researchers and three as senior researchers. In addition, one identified him/herself as a research animal veterinarian, one as a technician and one as a bioethical assistant. The contributions of the bioethical assistant did not change the dynamics or the content of the discussion as compared with the other focus groups. Different levels and types of competence were represented in all focus groups. At the start of the interviews the participants introduced themselves, as they did not always know each other, and described their research projects. The interviews lasted for 60–90 minutes and were recorded and transcribed verbatim. Two of them were conducted in English and two in Scandinavian languages.

The participants were first asked what they thought was important, second about their motivation, and finally about difficult issues when planning animal research. The question about important factors was asked because we assumed that the comments participants spontaneously made in response to it would be likely to indicate the matters to which they gave most attention in planning their research. For the question about motivation, it was assumed that these would also receive much attention. The final question about difficulties was asked to explore if there were any ambivalences, or conflicts, in the planning of the animal experiments. As a follow-up question, interviewees were asked how they solved their difficulties. No attempt was made to lead the discussion towards ethical issues, but relevant issues came up spontaneously in all interviews, and all participants contributed by referring to their own experiences or concerns. Additional follow-up questions were put to the informants to encourage them to further elaborate their contributions. All participants contributed to the open discussions, bringing in new elements or commenting on other participants’ contributions. There was no need to interrupt or rein in dominant informants.

We assumed that ethical standards must be a part of the planning of an experiment if they are to be taken into consideration in the performance of subsequent research. Further, we assumed that spontaneously referring to ethical norms reflects moral sensitivity about the use of animals in research ³⁴ which emphasizes the researcher’s commitment to practices that maintain ethical standards.

The transcribed interviews were coded and thematically analysed using NVIVO11 software. A provisional list of codes ³⁵ included ‘3R’, ‘replacement’, ‘reduction’ and ‘refinement’. In applying these we were guided by interpretations of the meaning of the comments, that is we did not require explicit matches with the wording of the comments. The codes were then sub-coded as ‘3R explicit’ or ‘3R values’. In an inductive approach, additional codes were added during the analysis to systematize issues raised during the discussion. ³⁵ The secondary codes also included ‘other ethical concerns’. This sub-code, which covered ethical topics that could not be directly related to the context of the 3Rs, was subdivided into animal related or non-animal related issues. The secondary codes also included ‘operational issues’ and ‘good science’, both of which turned out to be major topics.

Participants signed informed consent forms before the interviews. Demographic data was recorded in other documents and stored in a different folder to assure anonymity. Studies such as the present one does not require ethical approval in Norway, but they must be and were notified to the Norwegian Social Science Data Service.

Results

Solutions to practical issues and the importance of doing good science were dominant topics in all focus groups. The practical issues could not easily be separated from the goal of doing good science in the analysis, because these themes are interrelated.

The practical issues the respondents considered included the acquisition of ethical permissions, the availability of resources, the choice of animal procedures, logistics and information or communication challenges. The discussions of good science addressed the planning and designing of experiments, the choice of models, pilots, statistical considerations revolving around p values, and the reporting of animal experiments. There was agreement that proper design is necessary to obtain valid, reproducible results, and that animal use can be justified only if the design is in order. This may, however, create a tension with the demand for a reduction in the number of animals used.

Which means that if you are very concerned about the ethics, or the large number of animals you are using, trying to downsize it [animal number] in some way the study is unethical, because you are decreasing your chances to actually get the correct answer. (2M3 – interview/sex/informant)

On the other hand, criticism was voiced about the idea that more animals should be used simply to improve statistical power – with the aim of getting results published in a more highly ranked journal.

but then there is more needed to get it published, to get it published in a better journal. People they look at your output, in what kind of journals you publish. (1M1)

So it depends on impact factors, you will increase how many animals you will need. (1M2)

You got a p value different or journal … The question is, of course, is it ethical? (1F2)

Reducing animal numbers by excluding pilots or controls was also discussed. Pilots can provide important information for further studies. Controls are important elements of experimental design. Some respondents wondered whether historical controls should be used more often.

you can use historical data actually as a control. It depends on what type of study you’re [doing]. (2M3)

Better sharing of organs and tissues from animals to reduce animal numbers, was also discussed. Respect for the interest of healthy animals to stay alive was conveyed in another interview, and frustration that research regulations require healthy animals to be killed was also expressed.

That is the basic rule here – all animals come in – they don’t come out … so therefore we have to kill it. That is stupid – a healthy animal! (1M2)

The dilemma between using fewer animals and refining the housing conditions was raised in one of the interviews. The respondents appeared to think that this dilemma has no clear solution, but they demonstrated ethical awareness and the ability to identify and reflect on such dilemmas.

I think sometimes that there is too much focus on the reduction bit and too little on the refinement … To me it makes more sense to look at the refinement, to look at how do we reduce the stress on the animals … (2F1)

Solving practical tasks was a recurrent topic in all of the interviews. The collection of information enabling good practical solutions while taking animal-welfare concerns into consideration, was an important focus. Although such strategies fall into the refinement category, the word ‘refinement’ was not often used. However, the word ‘stress’ was frequently used when negative impacts on animals were being described. The importance of treating animals well to avoid stress of the kind that may bias an experiment was frequently raised.

you ensure that the animals are not stressed, so you’re getting something that is physiologically meaningful …

you want the animal to be happy, and to be as close to its normal [way of life] as possible, to be sure that … I mean, both to reduce the data variability, but also to, I mean, to be sure that the data you are getting actually reflects like a normal physiological regulation, and not a regulation in response to stress … (2M2)

Accurate information about tolerance level for an animal – for example by injection – makes a significant difference to the welfare of that animal. Searches for information on how to improve or refine procedures were important in the planning of experiments.

When we start with new tumors, we look up in the literature – What can we inject in these mice, and how should we inject it? (1F2)

Web-based literature searches were identified as useful methods for collecting information. However, talking to experienced colleagues seemed to be a preferred strategy for collecting information, especially when it came to refinement.

There are a lot of things written down, but other things not, and sometimes you go through the hard way, asking people, asking people – and calling people. And then you find out. (1M2)

I have also talked with people – I mean, a lot of people already have experience, and with this type of model, and what are the pitfalls. (1M3)

Talking with experts also served the purpose of building good collaborative relationships – for example, with the animal care staff.

My experience is that the more contact, the more they look after the animals … (1M1)

Replacing animal research with experimental approaches using alternative methods is a goal of policy makers. ⁸ However, the respondents voiced concerns about the consequences for science if this goal is enforced when no full replacements are available. On the other hand, they expressed disapproval of the situation where animals are still used, and where regulations may prohibit the use of available alternatives.

actually the in-vitro test is so much more sensitive than the rabbit test, eh? So it’s quite astonishing that the FDA has been so reluctant to accept that.

It’s the health authorities that actually have the last word … (2F1)

When the necessity of animal models was discussed, respondents often focused on the idea that there was no need to use animals. Already, in the planning of an animal experiment and the choice of model, the question of alternatives was raised.

Does it [the animal model] tell you what you want to know? Is it really needed? I mean some basic questions can sometimes be done in vitro. (1M2)

Consideration of alternatives was not limited to the planning phase. According to one participant, alternatives should be reconsidered as the research project progresses.

At a certain point you have to say we cannot see this in an animal – maybe we should go back to the lab and see if we can do some in vitro [investigations]. (1F3)

Although the informants raised concerns about alternatives, animal numbers and strategies to refine and improve procedures, only a few occasions they referred explicitly to the 3Rs. When they did so they treated the 3Rs as a tool for continuous improvement and change of practices.

can we do it in a different way? And having that policy put into the system, also with the 3Rs award and so on, I think that’s a good instrument actually to try to making it … eh … make it more natural, you can say – to put your ideas actually into a format that is also put into reality. (2M2)

Discussion

The aim of this study was to explore researchers’ attention to ethical standards and concerns when they are planning animal experiments, with a special focus on the 3Rs.

During the analysis the practical issues anchored in ethical values could not easily be separated from the goal of producing good science because these two elements are interdependent. This created some methodological problems for the analysis. On the other hand, it demonstrates that the identification of practical solutions is a prerequisite of good science, and there is a room for the application of ethical standards in achieving this.

While the present study confirmed that scientists put a strong emphasis on scientific standards, it also demonstrated that researchers address values in accordance with ethical standards for research, including the 3Rs. When the 3R issues were addressed, they were usually connected with practical solutions. On only a few occasions, and in only one of the interviews, were the 3Rs explicitly referred to in their own right. The relevant EU directive, however, requires researchers to prepare an explicit statement explaining how compliance with the 3Rs has been ensured. It requires: A demonstration of the compliance with the requirement of replacement, reduction and refinement (article 43 1.b). ⁸

The spirit of the 3Rs is mentioned in several places in the directive (Article 1.1 a, Article 4, Article 13.2 a-c, Article 27.1.b, ANNEX V.10, ANNEX VI.2). ⁸ In ANNEX VI.2 the 3Rs are explicitly referred to: ‘2. Application of methods to replace, reduce and refine the use of animals in procedures’. ANNEX VI. 3–5 states that the researcher must describe any methodological strategies that have been employed to avoid pain, suffering and to achieve humane endpoints – all of these are examples of refinement. Similarly, ANNEX VI.6 requires researchers to describe strategies they have used to minimize or reduce the number of animals used. The ANNEX VI guidance may indeed encourage researchers to perceive the instruction to demonstrate compliance with the 3Rs as a detached element rather than an integrated part of the methods they select in order to maximize scientific benefits or reduce negative impact on animals.

A recent paper by Kirk ³⁷ stresses that Russell and Burch framed the 3Rs, not as an ethical tool, but as a strategy to maximize high-quality science. The 3Rs are only one pathway among others to high-quality science, and this might explain the informants’ tendency to focus on ‘good science’ without explicitly referring to the 3Rs.

As the 3Rs were launched as a concept for humane experimental technique by Russell and Burch, ⁶ Würbel ¹⁹ launched the 3Vs as a principle to maximize scientific validity. According to Würbel and others, the assessment of scientific quality should be part of the ethical evaluation of animal experiments.^{19,21,23,31,38,39} Some authors regard the scientific quality of experiments as a part of the harm domain.^21,23,39 Others treat it as part of the benefit domain.^40–42 By contrast, Bateson ³⁸ and Würbel ¹⁹ assess quality as a separate domain.

In a previous study of researchers’ attitudes to the 3Rs, only a minority were found to be aware of the 3Rs and able to properly name the principles. ³ This accords with findings of the present study, that is only in one interview were the 3Rs explicitly mentioned. However, it might be asked whether it is important to be able to name 3Rs correctly as long as the core value of limiting harm are implicitly respected, or promoted, in the relevant experimental practice, as they seem to be in this study.

The same study ³ showed that researchers favoured refinement over reduction. The dilemma of reduction versus refinement was spontaneously raised in one of the focus groups. Here, the view that refinement was more important was again expressed. In our study, however, reduction issues were addressed more frequently than refinement issues. Concerns about using too few animals to generate reliable data, were voiced and regarded as an unnecessary and unethical animal use. The inclusion of a sufficient number of animals to assure statistical power was very important, and necessary to justify animal use, according to the researchers we interviewed. This still accords with Russell and Burch’s definition of reduction:

Reduction in the number of animals used to obtain information of a given amount and precision. ⁶

The use of animals for a certain experimental purpose, or in a certain context, brings the benefit domain into play. Some benefits are more legitimate than others. Increasing animal numbers to improve the statistical validity of a study is acceptable; using more animals to strengthen a study statistically in order to get studies published in higher ranked journals as a personal benefit was regarded unacceptable – although, of course, there is a fine line between these two goals. To our knowledge, the issue of increasing animal numbers to improve statistical parameters in order to get results published in higher ranked journals, or journals with higher impact scores, has not been raised before. Such a practice has troubling implications for animals and can also be criticized for bad data management and violating principles of good science.^31–33 Young scientists especially may be under pressure from senior colleagues or journal reviewers asking for more experiments.

The preference for talking with experts rather than conducting a literature search is at odds with an earlier study from Canada, where web-based searches were the preferred method of collecting information. ¹¹ Information collection was one of the priorities mentioned in connection with the planning of experiments, but the participants in this study preferred to talk to colleagues rather than search databases. This accords with van Luijk et al., ¹³ a study showing that 3Rs databases were rarely used and that other forms of knowledge exchange were preferred. ¹³

A web-based study by Fenwick et al. ¹¹ discovered that 3Rs assistance should be ‘constructive’ and ‘neutral’ and provided by an ‘expert’. The same study also suggested that such 3Rs assistance should not be mandatory but be offered when problems are identified. Results from our qualitative study indicated that researchers are very aware of their need to seek advice, they seem to have a good relationship with their 3Rs advisors, and that advice was provided in a collegial, collaborative manner. Studies have shown that the collaborative approach is more efficient, when it comes to compliance with the advice given, than a ‘paternal’ relationship where a superior instruct on the right solution. Hence, the collaborative approach has been described as the best way of building a culture in which the 3Rs are governing values. ³⁶

Lack of reproducibility, and the validity of experiments more generally, is a major concern in preclinical science.^43,44 A strong commitment to principles of good science, as addressed in this study, reflects a serious emphasis on achieving this. However, the attention given to principles of good science expressed in this study seems to be at odds with the existence of a reproducibility crisis.^43,44 One explanation may be that good intentions in the planning phase are not necessarily followed up in practice. This may be because animal experiments are complex, with many pitfalls, so that good intentions drown in practicalities.

Both the finding that researchers preferred to talk with, and collect information from, colleagues and the point about the complexity and pitfalls of animal studies suggest that it is necessary to establish strong, competent, updated support functions, as well as nurturing a collaborative culture around animal research activities to better facilitate both good science and high standards of animal welfare.

A qualitative method was chosen for this study, since the aim was to explore the role of ethical standards in the planning of animal experiments, and to investigate how these standards are addressed and complied with – all with a focus on the 3Rs. A qualitative approach opens up a more nuanced picture than can be obtained by quantitative investigation^11,35,45 or by using questionnaires of the sort adopted in earlier studies.^3,11,46 A qualitative approach provides a better understanding of the way in which concepts like those of replacement, reduction and refinement are perceived, interpreted, phrased and applied. ⁴⁵ It provides insight into the researcher’s perspective that is useful in planning as well as reporting and evaluating studies. Qualitative studies are not suitable for quantifying, or discovering the distribution of certain opinions and their results should not be generalized. The strength of the focus group method is that participants in such groups share, compare and discuss different views, and give feedback on each other’s opinions. ⁴⁵ There is always a danger that some individuals will dominate the discussion and obscure other group-members’ opinions, but this was not seen in the present study. Four northern European counties were chosen, as we assumed that they are rather similar in terms of their levels of legislation and more generally in the attention they give to animal welfare.

The data in this study were collected during the period of transition to full implementation of the new directive. Although the 3Rs as such are referred to for the first time in the 2010 directive, they are also present, albeit without being explicitly set out, in the previous 1986 directive (Articles 7 2, 3 and 4). We suspect that stronger explicit emphasis on the 3Rs in the 2010 directive has affected the way researchers talk about changes and improvement in terms of the 3Rs, even if the changes are part of a process that takes time. We therefore believe that our findings still give valuable insight into the thinking of researchers working with animals.

The results of this study show that researchers do indeed address values and work in accordance with the 3Rs. However, terms such as ‘necessity’ or ‘stress’ are rather used than the official terminology. This has implications for the way we should examine the application of the 3Rs. For example, instead of asking what replacement alternatives have been considered, it may be more productive to ask a researcher: why are animals necessary for this study? Similarly, instead of asking what refinement strategies have been applied, we could ask: what decisions have been made which reduce unnecessary stress in the animals? Talking the same ‘language’ will probably nurture a more fruitful and collaborative collegial relationship, and it may also stimulate reflection on current practices and changes to these.

The respondents in this study were never asked directly what place ethical standards had in the planning of their experiments. Nevertheless, they spontaneously brought ethical issues into the discussion, and in doing this they demonstrated an ethical awareness of the issues raised by the use of animals in research, and empathy for their animals.