Focus Group

Focus Group in Health Economics and HTA

In health economics, focus groups are mostly used before the numbers are collected. They help analysts find out which outcomes, attributes and words matter to patients, so that a questionnaire, a discrete choice experiment or a submission to an HTA body measures the right things. The method trades depth on each individual for the range of views that emerges when people react to one another. Its findings describe what people think and how they talk about it, not how common a view is in a population.

Group interaction as the source of data

Sim and Waterfield quote a broad definition of the focus group as a type of group discussion about a topic under the guidance of a trained moderator, and trace the method to group interviewing described by Merton and colleagues in 1956. Tausch and Menold note that focus groups served mainly in market research before social scientists adopted them, and that their use in health care research is more recent still.

What sets the method apart from a set of interviews is that the interaction itself is data. Sim and Waterfield recommend, citing Kitzinger among others, that collection and analysis take account of the interaction as well as the dialogue, and they add that analysis should capture how meaning is negotiated and co-produced in the group. The FDA guidance on methods to identify what is important to patients describes a moderator who can explore issues with individuals and also encourage discussion among participants, an approach that elicits a range of experiences. In the most common form, a single group meets in one place, and the moderator steers the conversation with a semi-structured discussion guide.

Composition shapes what the group can produce. Sim and Waterfield report the common advice that members of a group be homogeneous, so that differences in status or power are small and higher-status members do not dominate the discussion.

Where focus groups enter economic evaluation and HTA

Focus groups rarely produce a number that goes into a model. Their value lies upstream, in deciding what a quantitative instrument asks, and in giving committees structured evidence of patient experience.

Attributes and levels for discrete choice experiments

A discrete choice experiment can only value the attributes it includes, so attribute selection fixes the scope of the resulting preference weights and of any willingness to pay estimates derived from them. Vass, Rigby and Payne reviewed 254 healthcare DCEs published from 2001 to June 2012. Of these, 111 (44%) reported no qualitative work, 114 (45%) reported basic information and 29 (11%) reported or cited extensive qualitative work. Among the studies that reported qualitative methods, 63 (44%) used focus groups and 109 (76%) used interviews, mostly to select attributes and levels or to pilot the survey.

Coast and colleagues, drawing on eight studies, describe attribute development as a two-stage process of conceptual development followed by refinement of language so that the attribute conveys the intended meaning. They identify a tension in condensing complex qualitative findings into precise terms, and suggest that the nature of data collection depends on the characteristics of the question, such as its sensitivity, and on the qualitative information already available.

Content validity of outcome and preference instruments

A patient-reported outcome measure is only useful in an evaluation if its items cover what matters to the people being measured, which is the question of content validity. The FDA's 2009 guidance on patient-reported outcome measures lists focus group transcripts among the sources from which items can be generated, and states that, without adequate documentation of patient input, an instrument's content validity is likely to be questioned. For an existing instrument developed without patient involvement, it accepts new qualitative work as documentation of content validity if patient interviews or focus groups use open-ended methods.

The ISPOR task force on content validity divided the work in two. Its first report covers eliciting concepts through qualitative focus groups or interviews, and its second covers testing respondents' understanding of a draft instrument through cognitive interviews.

The same logic reaches the generic measures behind the QALY. The NICE manual (PMG36) says that a case that the EQ-5D is inappropriate for a condition needs qualitative empirical evidence on its lack of content validity, showing that key dimensions of health are missing, supported by evidence of poor performance on tests of construct validity and responsiveness.

Patient experience as evidence in technology appraisal

The NICE manual states that qualitative research can explore values, preferences, acceptability, feasibility and equity implications. It names patients' experience and quality of life with a disease or a treatment, and patients' views on the acceptability of different treatments, among the aspects such evidence is particularly useful to assess. It adds that qualitative data may come from formal qualitative research studies, and when that evidence is extensive it prefers recognised synthesis methods such as framework or thematic synthesis. The manual does not name focus groups, but a focus group study can reasonably be read as formal qualitative research, and the manual says the committee considers any qualitative evidence on the experiences of patients, carers and experts alongside randomised and non-randomised studies. On that reading, focus group findings have a recognised route into health technology assessment.

Designing and running a focus group

The FDA guidance treats most design choices for focus groups as similar to those for interviews, such as sampling and piloting the discussion guide. It adds three specific to groups, namely a trained moderator with the right skills, the number of groups and the size of each group.

The number of groups depends on the complexity and sensitivity of the topic, the diversity of the sample and the number of planned subgroups, and the guidance gives the example of separating the sexes for a topic such as sexual function. On size, it notes that groups of 5 to 10 are common and that too many participants fragment the discussion into simultaneous conversations. Tausch and Menold interviewed moderators of health research groups, who considered four to six participants optimal, against the eight to 20 that they report social science sources recommending, so that each person had time to tell their story.

The FDA guidance says that no universal number of groups exists and that, for some methods, the number may be guided by saturation. Its 2009 guidance defines saturation as the point at which no new relevant or important information emerges, and its 2022 guidance gives a worked grid in which four groups of six patients are tracked concept by concept. Analysis choices are reported less often than they should be. In the Vass review, only 44 studies (31%) of those using qualitative methods reported their analytical approach, most often content or framework analysis.

Ethics differ from those of one-to-one interviews. Sim and Waterfield point out that researchers have limited control over what participants later repeat outside the group, that group discussion can encourage some members to disclose more than they intended, and that consent is harder to make fully informed because the course of the discussion depends on other participants.

Choosing between focus groups, interviews and Delphi panels

The FDA guidance compares the strengths and limitations of each qualitative method. Focus groups let participants react to and build on each other's ideas, collect feedback from several people at once and are relatively inexpensive. One-to-one interviews give more depth on each person, suit sensitive topics, and suit conditions whose symptoms differ widely from patient to patient. The guidance gives a sequence in which focus groups explore a broad topic first and individual interviews then obtain detail.

A Delphi method panel sits at the other end of the spectrum. The FDA guidance describes it as a multistage survey that seeks consensus among experts, who may include patients, and notes that an anonymous process reduces the role of ego and interpersonal issues. A focus group does the opposite, exposing each view to the group, and the guidance describes its aim as eliciting a range of experiences. Where a committee or guideline group needs a judgement agreed, consensus building methods are the relevant family. Cognitive interviews are a third neighbour, used to test whether respondents understand a draft instrument rather than to generate its content.

Worked example with four groups and a choice design

The figures below are illustrative. A team planning a DCE on a community weight-management service runs four focus groups of six adults, 24 people in total, and records how many participants in each group raise each candidate attribute.

Group 1 raises five attributes, Group 2 adds two, Group 3 adds one and Group 4 adds none, so the cumulative count runs 5, 7, 8 and 8. After checking that the four groups covered the intended mix of participants, the team judges that saturation has been reached for this purpose.

The size of a full factorial design shows why the next step, cutting the list, matters.

$$P = L^{K}$$

where $P$ is the number of distinct profiles in a full factorial design, $L$ is the number of levels per attribute, assumed equal across attributes, and $K$ is the number of attributes. With $L = 3$, keeping all eight attributes gives $P = 3^8 = 6561$ profiles, while keeping five gives $P = 3^5 = 243$, which is 27 times fewer. The team then uses experimental design methods to choose a subset of choice tasks, but the attribute list it carries forward is a qualitative judgement, and the tension Coast and colleagues identify in condensing rich findings applies at exactly this point.

The table also invites a misreading. Cost was raised by 10 of 24 participants, about 42%, yet that figure says nothing about the share of eligible adults who care about cost. Participants were recruited purposively and spoke in response to each other, so the weight given to cost has to come from the DCE itself.

What focus group findings cannot show

The FDA guidance gives an example in which focus groups and interviews generate concepts and a survey instrument then measures their prevalence in a larger patient population. The division of labour is the right one, because a focus group samples for range and the counts it produces are not estimates of frequency. Its 2009 guidance makes a related point, that the number of patients matters less than interview quality and the diversity of the sample relative to the intended trial population.

Group effects are the main threat to validity. The FDA lists the limitations that one person may dominate, that participants may feel peer pressure, and that the group setting may inhibit people from sharing sensitive information. Tausch and Menold report that young medical assistants did not feel free to speak in front of their superiors. A finding that every group agreed may reflect the dynamics of the room as much as shared views.

Reporting is the weakest link in economic applications. In the Vass survey, all 50 responding authors said qualitative methods had added value to their DCE, but 22 (44%) reported that journals were uninterested in the qualitative results. An attribute list without a reported qualitative basis leaves readers unable to judge whether the preference weights rest on what patients actually value.

Sources

• Coast J, Al-Janabi H, Sutton EJ, Horrocks SA, Vosper AJ, Swancutt DR, Flynn TN. Using qualitative methods for attribute development for discrete choice experiments: issues and recommendations. Health Economics. 2012;21(6):730-741. https://doi.org/10.1002/hec.1739
• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). London: NICE; 31 January 2022, last updated 13 October 2026. Sections 3.3.18 to 3.3.20, 4.3.10 and 6.2.5. https://www.nice.org.uk/process/pmg36
• Patrick DL, Burke LB, Gwaltney CJ, et al. Content validity, establishing and reporting the evidence in newly developed patient-reported outcomes (PRO) instruments for medical product evaluation: ISPOR PRO good research practices task force report, part 1, eliciting concepts for a new PRO instrument. Value in Health. 2011;14(8):967-977. https://doi.org/10.1016/j.jval.2011.06.014
• Patrick DL, Burke LB, Gwaltney CJ, et al. Content validity, establishing and reporting the evidence in newly developed patient-reported outcomes (PRO) instruments for medical product evaluation: ISPOR PRO good research practices task force report, part 2, assessing respondent understanding. Value in Health. 2011;14(8):978-988. https://doi.org/10.1016/j.jval.2011.06.013
• Sim J, Waterfield J. Focus group methodology: some ethical challenges. Quality & Quantity. 2019;53(6):3003-3022. https://doi.org/10.1007/s11135-019-00914-5
• Tausch AP, Menold N. Methodological aspects of focus groups in health research: results of qualitative interviews with focus group moderators. Global Qualitative Nursing Research. 2016;3:2333393616630466. https://doi.org/10.1177/2333393616630466
• US Food and Drug Administration. Guidance for industry. Patient-reported outcome measures: use in medical product development to support labeling claims. Silver Spring, MD: FDA; December 2009. https://www.fda.gov/media/77832/download
• US Food and Drug Administration. Patient-focused drug development: methods to identify what is important to patients. Guidance for industry, Food and Drug Administration staff, and other stakeholders. Silver Spring, MD: FDA; February 2022. https://www.fda.gov/media/131230/download
• Vass C, Rigby D, Payne K. The role of qualitative research methods in discrete choice experiments: a systematic review and survey of authors. Medical Decision Making. 2017;37(3):298-313. https://doi.org/10.1177/0272989X16683934