This approach is rooted in anthropology and is concerned with the description and interpretation of cultural patterns of groups and the understanding of the cultural meanings people use to organize and interpret their experiences. In such studies the researcher studies the meaning of behaviour, language and interactions of the ‘culture-sharing’ group, for example a hospital, a health unit, or the different ways in which pain is expressed by men and women.

The purpose of phenomenology is to uncover the meaning of how humans experience phenomena through the description of those experiences as they are lived by individuals. For example, what is the meaning of healing to women with cancer, or, what is the meaning of fear of dementia in ageing people? This approach is different from ethnography because it focuses on experiences from the perspective of the individual rather than understanding a group or cultural pattern.

Grounded theory is a method that is primarily used to develop a theory or conceptual model. Usually the researcher begins with a broad query in a particular topic area and then collects relevant information about the topic. As data collection continues, each new piece of information is reviewed, compared and contrasted with earlier collected information. From this constant comparison process commonalities and dissimilarities among categories of information become clear and ultimately a theory that explains the observations is inductively developed. For example, this approach could be used to develop a theory of how people diagnosed with cancer decide whether or not to use CAM or to explain why some people refuse conventional treatment options.

Case studies include either indepth analyses of single cases or comparison across multiple cases (Yin 2003). Usually data collection spans a variety of sources including documents, interviews and observation. Data from different sources are triangulated (i.e. information collected using different strategies is used to examine the same phenomenon from different perspectives) to provide an indepth view of each case. When multiple cases are available, analysis is first completed on a case-by-case basis and then cross-case comparisons are made (Miles and Huberman, 1994 and Stake, 2006). For example, this approach may facilitate identification of success factors for integrative clinics where each clinic is analysed as a separate case.

A biographical study is the investigation of an individual and his or her experiences as told to the researcher or found in documents and archival material (Creswell 1998). Life history inquiry is a form of a biographical study, and this focuses on the individual life course. This type of research is concerned with eliciting life experiences and with how individuals interpret and attribute meanings to these experiences (DePoy & Gitlin 1998).

Interviews are the main data collection method employed by qualitative researchers. Interviewing techniques vary from semistructured interview schedules that ensure all participants are asked to discuss the same topics to completely unstructured interviews or narratives where the participant primarily determines what is discussed (Berg, 1995 and Morse and Field, 1995). Unstructured interviews are generally used when the researcher knows very little about the topic and is unable to prepare a series of questions because s/he doesn’t know what to ask. In this case it is best simply to let the participants tell their stories. Semistructured interviews are used for more focused inquiries where the researcher knows many of the key topics of interest. Questions are worded so that they are open-ended and the interviewer should allow the participant to guide the order in which the topics are discussed. For example, if the participant begins talking about question 5 when answering question 2, the interviewer should follow the participant’s lead by probing about question 5. The questions that have been skipped can be addressed later in the interview. To enhance semistructured interviews it is important to maintain a conversational flow with the participants. Interviews are usually audio-recorded. Face-to-face interviews are best to establish rapport with the participant; however, telephone interviews can facilitate data collection across greater physical distances. An advantage of individual interviews is the ability to elicit information about a specific process or changes over time. Interviews allow the investigation of indepth personal experiences of phenomena and can be particularly useful when discussing sensitive (personal) topics. The two main disadvantages of individual interviews are the fact that they can be time-consuming and expensive. For example, usually no more than two interviews can be scheduled in a single day and analysis of individual interview transcripts is often very time-intensive. Costs related to travel, interviewer time and transcription can make individual interviews too expensive for large studies.

One way to minimize costs and decrease data collection time is to interview participants in groups, usually called focus groups. Typically, focus groups consist of six to eight individuals who share some characteristics (e.g. disease diagnoses; experience of a therapy) and a trained moderator leads the group through a discussion of predetermined topics (Morgan 1998). Focus groups are very useful for gathering a wide range of opinions quickly or to develop consensus guidelines in a group process. Another advantage of focus groups is the ability to capture and analyse interactions between individuals attending the group, for example, one participant trying to explain his/her point of view to another participant, or disagreements that arise between group members. This also emphasizes the need for a skilled moderator who can ensure everyone has an opportunity to express his/her views and that one participant does not unduly influence the opinions of others. Since it is necessary to schedule multiple individuals to attend each group, focus groups do not work well with individuals who are difficult to schedule, such as busy physicians.

Observation is another way to explore the interaction of participants with each other (Lincoln & Guba 1985). The main purpose of observation is to collect data about participants in a natural setting. Observation is often used to provide breadth and context that cannot be achieved by interviews alone. However, it can be time-consuming and expensive, so it needs to be justified based on the research question and the value it can add needs to be clearly specified (Morse and Field, 1995 and Creswell, 1998). Consent is an important issue to consider when designing a study that includes an observational component. Generally ethics boards permit observational data collection in public spaces without obtaining the informed consent of those being observed. However, observation of patients in health care settings, or clinicians interacting with each other in integrated clinics, requires the consent of those being observed. Observational strategies are generally categorized by the degree of researcher involvement in the setting and range from complete participation of the researcher (e.g. a naturopathic doctor who is also a PhD student observing how fellow practitioners interact with patients and other clinicians in an integrative clinic) to complete observer (e.g. a sociologist observing patients receiving acupuncture treatments). The former is challenging due to issues related to informed consent (ensuring consent of those being observed or justifying why the intent to observe cannot be disclosed to those being observed) and the difficulty the participant observer is likely to have with objectivity (it is difficult to maintain the work role while taking a researcher perspective). The latter has the disadvantage of not permitting the researcher to ask questions or clarify what is happening, which can lead to misinterpretation of what is being observed. In practice, normally some compromise between these two extremes is found to be the best way to collect data using observation techniques (Lincoln and Guba, 1985 and Morse and Field, 1995).

Finally, any written document can be used for qualitative inquiries. Diaries, letters, personal notes and official documents can all inform the investigation. Although qualitative analysis is most easily performed on written texts, increasingly photographs, works of art, videos and performance art are also considered as sources of data for qualitative inquiry. Ultimately, the source of data and the data collection technique should be driven by the research question.

Many CAM interventions emphasize patients’ developing understanding of health and illness. Stibich & Wissow (2006) examined how perceptions of health and illness changed in the context of wellness acupuncture. To this end they asked graduating acupuncture students to select three patients and ask them to write a letter about the benefits of acupuncture. A total of 367 letters were received. The letters were analysed using a grounded-theory approach to identify meaning shifts. Five shifts were identified: (1) from a goal of fixing the problem to a goal of increasing health; (2) from symptoms as problems to symptoms as teachers; (3) from healing as passive to healing as active; (4) from being dominated by illness to moving beyond illness; and (5) from regarding the practitioner as a technician to regarding the practitioner as a healer/friend. Numerous quotes are used in the paper to support these changes. In an attempt to assess the transferability of these themes we conducted a secondary analysis of transcripts from qualitative interviews with cancer patients receiving integrative care and found that these shifts could also be identified in this population (Verhoef & Leis 2008). The suggested link between meaning and health led Stibich & Wissow (2006) to look for models that explain how meaning shifts can be created. As such this work has great potential for informing and enhancing clinical practice.

A common concern in CAM efficacy/effectiveness research is that the current arsenal of outcome measures may not be sufficient to assess outcomes experienced by people receiving CAM interventions or integrated care. Bell et al. (2003) identified that changes homeopathic practitioners witnessed in their patients extend beyond health-related domains to a much broader view of the individual’s entire scope and flow of life experience, such as personal perception, creativity, sense of freedom and ‘feeling less stuck’. Qualitative analysis of interviews that explored patients’ experiences of homeopathy (Bell et al. 2004) and of integrative care (reported in Berg 1989) confirmed that several of these outcomes, including ‘unsticking’ and transformation, occur in patients who receive CAM or integrative health care interventions (Koithan et al. 2007). This information is currently being used by a research team headed by Dr Ritenbaugh (University of Arizona, USA) to develop a quantitative instrument to measure where patients are in their transformation process. As long as CAM interventions are being evaluated using conventional physical, psychological and quality of life measures, we will not be able to identify benefits that are specific to whole-system CAM interventions. In fact, using outcome measures that do not target individualized and holistic outcomes may result in bias and compromise clinical relevance. Qualitative analysis is a useful tool to identify and assess the outcomes patients describe as being most relevant.

Given the rise of CAM use it is important to know how patients make the decision to use CAM. Balneaves et al. (2007) have developed a model that explains how women with breast cancer make the decision to use CAM. Participants were recruited from clinical oncology settings, an integrative care clinic and community support groups in Vancouver, Canada. Investigators used grounded-theory methodology and conducted semistructured interviews. The resulting conceptual model is called ‘bridging the gap’. All participants aimed at making treatment decisions that would help them survive cancer and improve their well-being. In the context of their knowledge and beliefs about conventional medicine and CAM as well as their social networks, women contemplated which therapies would help them maximize the benefits and minimize the chances of harm. Once they identified CAM therapies they often moved into a complex and confusing process of gathering and filtering information about these therapies and trying to figure out which ones were credible and which ones were not. They were often confronted by contradictory information and recommendations about the safety and the value of these therapies and started to feel there was a gap between the two worlds of conventional medicine and CAM. Faced with these paradigmatic differences, but wanting to make a treatment choice that would be best for their health and be congruent with their knowledge and beliefs, women engaged in three different decision-making processes of ‘bridging the gap’. These included:

Again the results of this study provide direction for health care practitioners regarding the need for education and decision support strategies to ensure the safety and well-being of women with breast cancer.

Evaluating complex CAM interventions is challenging for an RCT design as it means standardizing individualized treatment; controlling for non-specific and placebo effects, such as the impact of patient preferences, hopes and expectations; the impossibility of blinding; and a focus on group rather than individual results. As qualitative inquiry is an approach that is very suitable to address the impact of several of these issues, it is sometimes added to an RCT design. For example, Brazier et al. (2006) added qualitative interviews to an RCT assessing the effect of a 2-week residential yogic breathing, movement and meditation programme aimed at improving mental health, health status and stress reduction among people living with human immunodeficiency virus (HIV)/acquired immunodeficiency syndrome (AIDS). They found only minor differences between the experimental and control groups; however, qualitative results showed that participants experienced personal growth processes following the intervention. They described how ‘living’ began to feel more meaningful and conscious, and they began to feel everything, pleasant and unpleasant, with greater intensity – but this greater self-awareness also included greater awareness of changes, stress, pain and discomfort. These results illuminate the trial results and aid the development of more sensitive evaluation approaches for future research.

In another study Stalpers et al. (2005) conducted an RCT to determine whether hypnotherapy reduced anxiety and improved quality of life in cancer patients. No statistical differences in anxiety and quality of life were found; however, qualitative interviews, including an open-ended questionnaire, showed that the majority of patients felt their mental and overall well-being had improved and all patients in the hypnotherapy group would recommend such therapy to other patients. The authors suggest that the increased attention of doing the interviews might have been responsible for the changes. However, the qualitative inquiry reveals a number of potential explanations that could be further explored and tested. Studies such as the above identify how qualitative research can explain issues related to the meaning of the intervention, outcomes, process and context of the intervention.

While early-phase research or phase I trials are not uncommon designs, Aickin, 2007 and Fønnebø, 2007 have pointed out the importance of this type of study with respect to CAM trials. Many CAM approaches are un- or underresearched and it is not feasible to investigate them all using large studies that will present definitive results. Early-phase studies are intended to get answers to many design questions, such as: What patients should be targeted for the intervention? Who is willing to accept the intervention? Is placebo really placebo or part of the intervention? Can CAM and conventional practitioners deliver the treatment under circumstances that differ from their normal practice? What are appropriate outcome measures? Are there different versions of the intervention? Do practitioners agree on the nature of the intervention? In order to address such questions, Aickin suggests that the sample size should be small, as in this phase, in which much is unknown, we will find that some research approaches will not work. He also suggests that measurement should be intensive to learn as much as possible about those data that are relevant and those that are not. Most importantly, such studies will only be successful if both quantitative and qualitative inquiry is used. In fact, qualitative research will serve as a ‘corrective check on researcher bias’ via the use of techniques that solicit unstructured responses from participants.