Understanding the concept of medical risk reduction: A comparison between the UK and Germany

Abstract

Objective: To explore the views of German general practitioners, healthcare assistants, and laypeople about the minimum absolute risk reduction needed to justify drug treatment to prevent heart attacks, and to compare these views with those found in the UK. Method: Qualitative content analysis study using the same clinical risk scenario and semi-structured interview schedule concerning a “pill” reducing cardiovascular risk as a recent UK study. The similarly recruited participants included six general practitioners (GPs), four healthcare assistants, and 12 laypeople, interviewed in 10 GP surgeries, two community settings, and five private homes. Results: In both countries, most participants, health professionals as well as laypeople, used risk numbers inconsistently in preventive treatment decisions. In Germany, some people explicitly rejected the probabilistic risk concept as a basis for such decisions. In the UK, people were generally more aware of cost for society than in Germany. Other factors were similar in both countries. Conclusion: In both countries, preventive risk information is not well understood. Our results suggest that this is not only a technical communication problem.

• comprehension of medical risk concept
• professionals
• lay people
• national comparison

Introduction

The concept of medical risk reduction is based on a probabilistic way of seeing the world. It uses information known about a population of broadly similar people, and applies that information to individual patients [1], [2]. This probabilistic concept is widely accepted as a key concept in prevention, for example in the lowering of blood pressure or lipid levels to reduce the probability of an ischaemic vascular event [3–5]. To make adequate treatment decisions based on this concept and to incorporate patients’ preferences and values, we have to understand and communicate risk and risk reduction adequately [6–8].

A recent qualitative study from Great Britain (led by co-author D.K.L.) explored the views of general practitioners, practice nurses, and laypeople about the minimum absolute risk reduction they thought would justify drug treatment to prevent heart attacks and the reasons for their choices [9]. The researchers found a number of factors influencing this decision, including possible side effects, “guaranteed effectiveness”, costs, attitudes towards pill taking, lifestyle changes, and age. Concerning the risk numbers used, “most participants, including the health professionals, found the concepts difficult to grasp at first (for example, stating they were confused) and gave different numerical answers through the interview, some of which were self contradictory” [9].

The way we understand and apply risk information is part of medical culture and may vary from place to place [10], [11]; we wanted to know if results in Germany would be different from the results found in the UK. British medical culture has a tradition of rational thinking, including “chances” and “risks”, often viewed numerically. Clinical epidemiology and evidence-based medicine originated in the UK. Most British people are familiar with betting shops and the concept of “odds”—a tradition Germany does not have—whereas German medical culture has always stressed “personal experience” and the “unpredictable individual case”. With this in mind, we expected to find even more confusion concerning the handling of probabilities in Germany than in the British study. The aim of this study was therefore to explore the views of German general practitioners, healthcare assistants, and laypeople concerning the minimum absolute risk reduction needed to justify drug treatment to prevent heart attacks, and to compare these views with those found in the UK.

Methods

We recruited a similar [12] sample of general practitioners, healthcare assistants (Arzthelferinnen), and laypeople, and used the same semi-structured interview schedule as in the British study [9]. Also, we used the same kind of questions when confronting participants with probabilities and calculations. The English interview schedule as well as the case descriptions were translated into German, and were then back-translated by a bilingual speaker to guarantee as much as possible identical sets.

Table I compares the background characteristics of German and British participants.

Table I.  Summary of participants’ characteristics.

	German study			British study (9)
Characteristic	GPs (n=6)	Healthcare assistants (n=4)	Laypeople (n=12)	GPs (n=4)	Pract. nurses (n=4)	Laypeople (n=18)
Male/female, n	3/3	0/4	6/6	2/2	0/4	10/8
Age range, years	34–63	23–65	35–78	31–52	28–52	45–89
Aged >65 years	0	0	6	0	0	13
Coronary heart disease, n	0	1	4	0	0	7
Close person had heart attack*	1	0	8	1	3	7
Attended further education	6	1	7	4	4	8

* Participants were asked, “Has anyone close to you ever had a heart attack?”

Health professionals

We used a list of all general practitioners (GPs) in the region of North-Rhine and picked out by random 30 GPs to recruit the health professionals. We contacted the practices by telephone. If the person asked to take part in the study declined, we approached the next practice on the random list. No participants were known to the researchers, and all worked at different practices within the North-Rhine region. We interviewed participants in their workplace.

Laypeople

As in the British study, the laypeople were recruited from different settings outside a healthcare environment. Different organizations were approached via a key contact who was asked to look for volunteers in their organization. We interviewed five laypeople at a meeting of the German Heart Foundation (Deutsche Herzstiftung) to match the five people of the British Cardiac Support Group; two people from a university department; and five people at the social charity club of a church parish, who were interviewed at home.

Interview

Participants were given the opportunity to read an information sheet about the study. They gave consent to record a confidential face-to-face interview. The interviewer (G.F.) introduced himself as a GP and researcher from a university department.

The interviews were semi-structured using exactly the same clinical scenario and prompts as the British study (Box 1). At the end of the interviews, after “titrating” for the lowest acceptable benefit of taking a pill (as asked in the scenario), we additionally showed participants a visualization tool and asked if this would alter their answer or their understanding of the scenario. The visualization tool depicted 100 faces, which could be pictured as three groups: those being saved by the drug, those having a heart attack despite the drug, and those who would not have had a heart attack anyway.

Box 1. Scenario presented to participants and prompts used during interview [9]. Scenario "I would like you to think about a drug, a tablet, which can prevent heart attacks. It is not perfect, in that it does not prevent them completely. What that means is that if, say, 100 or 1000 people take the tablets every day for 5 years, then some will be saved from a heart attack—they would have had a heart attack without the tablet, but the tablet has prevented it. Some will have a heart attack anyway, even though they take the tablet. And the rest will not have a heart attack, but would never have had one anyway even without the tablets. Therefore, those who are saved from a heart attack will have been helped by the drug. The rest would have been just the same if they had not taken the tablets at all. The problem is that no one can say exactly who will be better and who will not, so they all had to take the tablets.” Clarification: “What I want to know is of, say, 100 similar people, all taking the drug for 5 years, how many have to be saved from a heart attack for it to be worthwhile all of them taking the drug every day?” Prompts used in all interviews “Why did you choose that number and not, say, one a bit higher (or lower)?” “How about if everybody has to take the tablet for 10 years instead of 5? Would that make any difference?” “Do you think it makes any difference how old the people taking the tablets are? How would that change your answer?” “Is there anything else about the people taking the tablets which might alter your answer?” “Let's say that the tablets are mostly safe, but that you have to have a blood test at least every year and that the long-term side effects are not known.” At the end of interview “Thank you. Would you still choose the same number now as you did at the beginning?” If someone said that one person benefiting out of 100 would be worthwhile: “OK. Let's say that 1000 people have to take it every day for just one to be better. Would that be worth it?” (Increasing the number until they said that it would not be worth it)

Interviews lasted 7–29 min and were recorded on minidisk. G.F. and H.H.A. independently analysed the recordings according to the principles of content analysis, for the themes found in the British interviews, and for new themes. A consensus was reached by discussion and repeated examination of the recordings. Relevant passages were transcribed verbatim. D.K.L. re-checked the original data of the British study to validate any differences found [12], [13].

Results

Understanding the scenario (Box 2)

Almost all participants—doctors, healthcare assistants, and laypeople—seemed a bit confused at first. Most participants, including four doctors, showed inconsistencies in relating biological risk factors or expected effects of existing drugs to the actual numbers in the scenario. Three laypeople did not understand at all what these numbers could mean. Two doctors, one healthcare assistant, and three laypeople thought the question was somehow related to a drug trial, perhaps implying that this kind of discussion is unusual in routine practice.

Box 2. Understanding the scenario. Researcher: “[ … ] how many have to be saved [ … ] for it to be worthwhile all of them taking the drug every day?” “I would have taken the tablet!” Researcher: “I see. Could you put that into numbers?” “Well … Fifty-fifty, I'd say … That's what people say, don't they?” Lay interview 4 Researcher: “[ … ] how many have to be saved [ … ] for it to be worthwhile all of them taking the drug every day?” “I don't know. I don't understand what you mean.” Lay interview 5 General practitioner 2 (interrupting the researcher presenting the scenario, which did not include any risk reduction numbers at all): “... oh yes, that means 5 out of 100 would have had a benefit!” At the end of the interview: “If there is a benefit for 10 out of 100, I would include them in this study.” General practitioner 2 Researcher: “[ … ] how many have to be saved [ … ] for it to be worthwhile all of them taking the drug every day?” ”If a patient had three or four risk factors, I would give him the tablet [ … ].” Researcher: “I see. Could you put that into numbers?” “I first need to know what this tablet does. How does it work? On blood pressure, on cholesterol, on what? It must diminish some risk factor, doesn't it? [ … ] I need proof. Black on white! [ … ] Something I can measure.” Researcher: “Alright, let's think about a drug where we don't know how it works biologically, but where we know for sure that it works: it can prevent heart attacks. [ … ] Let's say 10 out of 100 people would be saved. Would that be worth it, or is that not enough?” “Oh sure, 10 would be enough.” Researcher: “So, from what number on would it be worth it?” “[Laughing] … hmm, let's say: three out of 100. [ … ] But there is still something I have to know, which must have been established in tests, or so: what are the side effects?” At the end of the interview: “Three per cent looks so few. In pharmaceutical studies, they always have 20 or 30 per cent [ … ]. Anyway, I'll stick to that.” General practitioner 5

Minimum benefits chosen and reasons given for these choices (Table II)

The minimum acceptable benefits chosen varied widely from “one out of 100” to “60 out of 100” immediately after the presentation of the clinical risk scenario, and from “one out of the whole world population” to “80 out of 100”, at the end of the interview. A common viewpoint, mainly with lay participants, was that “every life counts!” leading to very low minimum benefits chosen. Higher numbers were chosen either because of possible (or “inevitable”) side effects, or because of a general dislike of taking tablets (the reason given for the highest minimum benefits chosen).

Table II.  Summary of minimum benefits chosen.

Doctors chose numbers between one and 30 out of 100; they seemed more confident in their choices, and their opinions changed less during the interviews than the other participants’ opinions (Table II). One doctor chose not to give any number and never changed this choice.

Benefits over time (prompt 2)

Most participants stated that it made no difference to them if everybody had to take the tablet for 10 years instead of five. This is inconsistent, because absolute benefit would increase with time if a tablet had a constant effect. Interestingly, three participants (two doctors and one layperson) expected a disproportional increase in effect over time, which may match reality, as risk (and thus absolute benefit) increases with age.

Visualization

Many participants stated that they better understood the concept with the aid of the visualization tool (“Okay … now I understand what you've been talking about all this time,” lay interview 12). Others were astonished because the visualization looked so different from what they had previously thought. Two doctors and one layperson realized that they had been mixing up absolute and relative risk reduction during the interview.

Showing the visualization tool at the end of the interview rarely altered the number chosen, but most participants thought it was a good means to communicate the information needed and a good starting point for discussion.

Rejection of the concept (Box 3)

Some participants explicitly rejected the probabilistic concept of risk reduction as a basis of decision-making for individuals. Three main reasons were given explicitly. One reason was that biological processes were more important than statistical figures; another reason was that this concept was “too abstract”, and that dealing with these numbers gives an illusion of rationality but does not help to make good decisions (general practitioner 4); the third reason explicitly given was that the “destiny” of an individual could not be deduced from risk information, which necessarily deals with groups of people, and thus risk information could not be a rational basis for individual decision-making (healthcare assistant 4).

Box 3. Rejection of the concept and cost of preventive treatment. Rejection of the concept Researcher: “[  …  ] how many have to be saved [ … ] for it to be worthwhile all of them taking the drug every day?” “I won't give you a number. This is too abstract! [ … ]” At the end of the interview: “I still won't give you a number. I refuse to! It's pseudo-rational. It does no good.” General practitioner 4 “I wouldn't care about this number. Not at all! [ … ] Numbers about risk say nothing about my personal destiny.” Healthcare assistant 4 “I would take them, if my GP told me. This counts for me, not'solid numbers' (...) and by the way: destiny is destiny. You can only try to die healthy.” Lay interview 3 “What counts for me is what this drug does on a biological level. This is much more important than those statistical numbers.” General practitioner 3 Cost of preventive treatment “The more expensive a treatment is, the more we have to know whether it really helps the ones who get it.” Lay interview 1 “I would only prescribe it if it doesn't count on my medication budget!” General practitioner 4 Researcher: “What about cost?” “Oh, I don't care much about cost. I haven't been sued yet [for exceeding the budget] [laughing].” General practitioner 5 Researcher: “What about cost?” “Does the health insurance pay it or not? The costs are THEIR problem! If they pay, I don't care.” Lay interview 9 Researcher: “What about cost?” “I would give my last penny for my health.” Researcher: “And if it was cost for society?” ”In that situation, I'd rather be selfish." Lay interview 4 Researcher: “What about cost?” “Either you give it to everyone or to no one. Not only to the rich!” Lay interview 8 Researcher: “[ … ] how many have to be saved [ … ] for it to be worthwhile all of them taking the drug every day?” “I don't know from what point on it is profitable.” Researcher: “You're thinking of cost?” “Yes, sure. If it is affordable, we should put it into the water supply, then we'd all live to 170 and work till we're 140!” General practitioner 6

Cost of preventive treatment (Box 3)

The participants rarely mentioned cost spontaneously. If the interviewer mentioned it, the participants mainly considered their personal costs as a doctor or as a patient. There were also participants, however, who thought that “the question of cost-effectiveness is very important” (general practitioner 2) and who reflected on the influence of cost on treatment decisions.

Other factors involved

Age (prompt 3) was seen as a factor that should be considered by about half the participants; the others said that it would not matter. Side effects (prompt 5) were mentioned by all participants as important. For major side effects, the minimum benefit required generally increased considerably, or the tablet was refused totally. For most participants, though, unknown long-term side effects in an otherwise mostly safe tablet were not a major problem. An annual blood test was seen as acceptable by all participants.

All the general practitioners except one wanted to incorporate patients’ preferences and values into the preventive treatment decision, which would be a part of shared decision-making. Some older laypeople stated that they would like to follow their doctor's recommendation, asking only for very broad information, e.g., “if it is necessary” (lay interview 2). The others wanted to make their own decision, based on more detailed information provided by their doctor.

Many participants, especially laypeople, preferred changes in lifestyle to taking tablets. In some of the interviews, the researchers got the impression that they massively overestimated the possible preventive effects of lifestyle changes, as they opposed “imperfect” drug treatment, suggesting lifestyle changes as a “perfect” solution to the threat of a heart attack. Many participants stated that they disliked taking pills, some without giving reasons, some in relation to possible side effects, and some because the act of taking tablets in itself made them feel ill by labelling them “in need of medicine”.

Differences between the UK and Germany

The scenario presented to the participants and the underlying probabilistic concept of medical risk reduction was as difficult to understand for the German as for the British participants. In both countries, doctors showed considerable inconsistencies in their answers; and in both countries, most participants did not see that a longer treatment period should increase the expected absolute benefit.

As expected, in Germany, several participants (health professionals and laypeople) explicitly rejected the probabilistic concept of risk reduction as the frame of reference for preventive treatment decisions, whereas in the UK no participant questioned the concept fundamentally.

In the UK, participants were generally more aware of cost for society than in Germany. In the German interviews, unlike the British, participants rarely mentioned cost spontaneously; when prompted, they mainly considered only personal costs. We did not find any other important differences between the British and the German interviews.

Discussion

Summary of main findings

We found for both countries that the concept of probabilistic thinking in preventive medicine is far from being understood and—especially in Germany—far from being accepted or even becoming a part of “everyday routine”. Additionally, in Germany, several participants explicitly rejected the concept of using population-based statistics to make individual decisions. In contrast to the UK, German health professionals and laypeople do not seem to be aware of the cost to society of treatments; starting or not starting a preventive treatment seems to be seen as an individual act only.

Strengths and limitations of this study

The strengths of our qualitative study include its methodological rigour. For the German study, we recruited a similar sample of general practitioners, healthcare assistants, and laypeople from similar social contexts and used the same clinical scenario and semi-structured interview schedule as used for the British study. As different results appeared in the German interviews, the British researchers were asked to re-check their original data concerning these aspects in order to validate the differences found. The limitations of this study include the limitations of all qualitative research: it can provide hints and allow for hypotheses, and it can make certain phenomena visible, but it cannot tell us to what degree or extent they exist.

Comparison with existing literature

Most participants used risk numbers inconsistently. Our results suggest that simple representations and visualization tools could help communication. This has been shown before [1], [20]. It has also been shown that treatment decisions of patients and doctors based on risk information largely depend on the way this information is presented and framed: relative benefits usually sound greater than their equivalent absolute benefits, and so appeal more, although absolute numbers are more informative [1], [21–24].

We still know little about the reasons why—after several decades of using statistics in medicine—neither doctors nor laypeople seem to have adopted probabilistic thinking as a genuinely “reliable tool” to help them make good decisions. This question is undoubtedly very important. Numeracy seems to be an important issue here, as well as the evidence that decision-making under uncertainty is strongly influenced by heuristics and biases of different kinds [1], [25]. There might also be some hints in our data. In Germany, several participants explicitly rejected the concept of using population-based statistics to make individual decisions. The reasons given for this rejection partly reflected difficulties in understanding the concept, but also highlighted a real limitation: risk numbers are about groups of people, making it impossible to deduce what will happen to an individual [14]. This reflects the fact that a decision for an individual based on risk information can never automatically “emerge” from the data. The patient always needs to decide, because there is no such thing as a single “right choice” [15], [16]. Some patients, however, wanted their doctor to give them clear directives, i.e., to make the decision for them. This request implies the belief that the doctor somehow “knows better”. It might also be understood as a “regressive” wish for a “secure base” in an insecure world [17]. We know that—but still not really why—some patients follow their doctors’ preferences and proposed therapies, even though they might not be more logical or consistent then their own preferences [18], [19]. On the other hand, patients simply may not always have well-informed preferences with respect to presumably important issues regarding medical treatments [26].

Conclusion

Our data suggest that risk numbers are difficult to use in shared decision-making. If we want to make probabilistic thinking a better tool in medical communication, we have to understand why it is not used adequately. Our data suggest that the reasons are not only a technical communication problem. They seem to be more deeply rooted: the ability to visualize risk cannot be trained and most risk reductions by medical interventions today appear too small to be believed even when verbally understood. The results of the study suggest that healthcare professionals should be made aware of the difficulties in understanding risk reductions, even when using the right “framing” in a consultation with a patient or in discussion with another healthcare professional. It might sometimes be helpful to use different framings for the same data in order to allow healthcare professionals to better appreciate a patient or colleague's level of understanding. However, as is often the case, the amount of time available for such communication is paramount. Discussing risks and risk reduction is not banal but worth longer explanation using examples and illustrations of what they are intended to show. For GPs, emphasis should be placed on understanding the numbers better and genuinely understanding the individual patient. In Germany, this could imply that even when “using the right data” and presenting them in different framings, it is still not understood, i.e., accepted by the patient, who sometimes “just wants advice on what to do”. Giving the right advice, however, also requires greater understanding of numbers and patients by GPs.

Acknowledgements

We thank the participants, who agreed to be interviewed and gave up their time for this study.

Contributors: D.K.L. had the idea for the British study and designed the clinical scenario and interview questions together with Ewan Wilkinson (Department of Public Health, Central Liverpool Primary Care Trust). G.F. and H.H.A. had the idea for the German-British comparison study and the visualization part of the German interviews. G.F. translated the clinical scenario and the interview questions from English to German and the quotes of the participants from German to English. H.H.A. reviewed and validated the translations. G.F. recruited and interviewed participants and recorded the interviews. G.F. and H.H.A. independently analysed the recordings for the themes found by D.K.L. in the British interviews, and for other themes, and made transcripts of the relevant passages. They discussed and validated their findings. D.K.L. re-checked the original data of the British study to validate the differences found by H.H.A. and G.F. G.F. wrote the paper, and all authors revised it. G.F. is guarantor for the study.

Source of funding: Department of General Practice, Universitaetsklinikum Düsseldorf, GermanyCompeting interests: None declared

Ethical approval: Not required according to the ethical committee of Universitaetsklinikum Duesseldorf “because no personal only fictive data are used in the interview”.