The designated study population consisted of student nurses who were at least 16 years of age and attended one of three nursing schools in Amsterdam, the Netherlands. Before recruitment, the school institutional review boards agreed with the study protocol. In total, four different recruitment techniques were used. First, by e-mail, we invited 154 students who studied in the Amsterdam area and participated in an on-going national cohort study (Visser et al., unpublished data). In this national cohort of approximately 700 student nurses, genetic susceptibility towards HE is studied. Secondly, we gave 2-min introductions in classes to invite students to participate. Thirdly, we placed posters on school message boards and school cafeteria tables. Lastly, by means of convenience sampling, we approached student nurses at the schools directly. We made sure that the proportions of participants recruited with these four techniques were comparable in the focus groups, interviews and questionnaires. All recruitment methods included a brief explanation of the study and a reward for participation. When desired, participants were refunded their travel costs.
The execution and analysis of the three qualitative research methods were based on core literature (Bryman 2001; Denzin and Lincoln 2000; Kitzinger 1995; Kvale 1996). To create a topic list for guiding the involvement methods and the analysis of results, we first performed a literature search on factors (items) that could influence nurses’ decisions, beliefs or attitudes towards the use of a genetic test that estimates the personal risk for HE. The following search strategy was applied in MEDLINE via PubMed: (“Dermatitis, Irritant” [Mesh] OR “Dermatitis, Occupational” [Mesh]) AND (“Nurses” [Mesh]) AND (“Genetic Predisposition to Disease” [Mesh] OR “Genetic Testing” [Mesh]). Because this search did not reveal any relevant studies, we broadened the search with the following strategy: (“Genetic Predisposition to Disease” [Mesh] OR “Genetic Testing” [Mesh]) AND (“Attitude” [Mesh] OR “Public Opinion” [Mesh] OR beliefs [tw] OR facilitator [tw] OR barrier [tw]). This search was limited to information published between September first 1999 and September first 2009, to human studies and to papers published in the English language. This search revealed 1,502 possibly relevant studies. MR and MV independently scanned all retrieved citations based on title and abstracts. Subsequently, the full texts of articles of relevant abstracts were retrieved. Ten relevant studies were selected for the purpose of this investigation (Cameron et al. 2009; Cameron and Muller 2009; Condit 2001; Harel et al. 2003; Henneman et al. 2004, 2006; Sanderson et al. 2004; Sussner et al. 2009; Tercyak et al. 2006; Toiviainen et al. 2003). From these studies and from our personal experience, we formulated 22 items that could influence the use of a genetic test. The items were clustered in 10 domains and processed in a topic list (“Appendix 1”). The 10 domains were: (1) expected use of genetic test (results); (2) test content; (3) feelings and emotions; (4) involvement with HE; (5) principles/beliefs; (6) expected effects of HE; (7) relative risk of developing HE; (8) accessibility, safety and privacy; (9) practical considerations and (10) social influence and media.
All three involvement methods comprised two parts and started with an introduction on the purpose of the study, the time schedule and confidentiality. During the first part, following the introduction, a hypothetical “case” was presented in which a genetic test for susceptibility to HE was introduced (Fig. 1). This presentation was concluded by two questions: (1) Would you use this test? (yes, no or doubt) and (2) What are your motives for using or not using this test? (open question). In the focus groups and interviews, answers were first noted by the participants and were subsequently discussed. During the second part, we introduced and discussed a topic list with items extracted from the literature. Participants were asked if (yes or no) and how (open question) the different items of this topic list would influence their choice to use this test. The items that had already been discussed during the first part were not reviewed. After this discussion, participants were invited to mention supplemental items.
Before application, the focus group protocol, interview protocol and questionnaire were all piloted. Additionally, a draft version of the electronic questionnaire was tested on comprehensibility among four workers from the Academic Medical Center in Amsterdam, the Netherlands. By convenience sampling, we recruited one worker from the catering service, one from the transport service and two student nurses.
The focus groups were held between October and December 2009 and were moderated by MR. MV participated as the case presenter and observer. Both researchers had been trained in qualitative methods. Focus group sessions lasted for about 2 h and were audio-recorded. Five to eight student nurses participated in each group, numbers depending on availability for the scheduled time. Participants received a gift coupon with a value of €20,–. The “case” was presented using PowerPoint and ended with the two discussion questions. We stimulated discussion by asking open-ended, non-guiding questions and encouraged all participants to contribute. To facilitate the discussion of the topic list in the second part of the session, we presented each domain (if not mentioned before) on flip-over sheets. We stopped the data collection at the point of data saturation, i.e. when two subsequent focus groups did not reveal any new items that could influence using a genetic test for HE.
Semi-structured interviews were executed between February and April 2010 by MR, MV and MMV. The interviews lasted for about 45 min, were audio-recorded and took place in a quiet room. Participants received a gift coupon with a value of €10,–. The “case” and the questions were provided in text and read out loud to the participants (Fig. 1). After reading the case, the interviewer left the room for a short period while the participants noted down their answers. Subsequently, the answers were discussed. To facilitate the discussion of the topic list in the second part of the interview, we presented all clustered literature items to the participants (if not mentioned before) on small cards. The interview data collection process was ended at the point of data saturation, i.e. when three subsequent interviews did not reveal any new items.
The electronic questionnaire, with combined closed and open-ended questions, was emailed to 51 participants in May 2010. We sent out one email reminder. Respondents were rewarded with a small gift (value €5,–). Participants received an introductory email with a hyperlink to the electronic questionnaire, which included 56 questions and took about 20 min to complete. The questionnaire mainly followed the protocols of the focus groups and interviews, which involved starting with the “case” and the two discussion questions on the use of the test and related motives. Subsequently, we introduced the domains one by one on separate pages. For each of the items within these domains, participants were asked if (yes or no) and how (open question) the item would influence their choice to use this test. Before proceeding to the next domain, participants were invited to provide supplemental items. Respondents were not able to go back to a previous page. The questionnaire data collection was ended at the point of data saturation, i.e. when five subsequent questionnaires did not reveal any new items.
All three methods were concluded by the participants’ completion of a short questionnaire on personal and professional characteristics and general knowledge of and experience with genetics and genetic testing (“Appendix 2”). Because we believe that the stakeholders’ perceived satisfaction with their involvement and contribution during the involvement method can directly influence their involvement output, we added the question: “How satisfied are you with your contribution during the focus group/interview/survey?” To prevent bias due to socially desirable answering, this questionnaire was completed anonymously.
Data analysis and coding
MR and MV performed a thematic content analysis with the data from all involvement methods. The audio-taped data from the first part of the focus groups and interviews was transcribed and analysed using MAXQDA software (VERBI Software, Marburg, Germany, 2006) that facilitates with organising and presenting large quantities of qualitative data. Each relevant unit of text remark was coded according to the taxonomy of 10 domains and 22 items as extracted from the literature. Remarks that could not be coded according to our taxonomy were iteratively discussed by MR and MV, and if necessary, new items or domains were created. From this point on, “literature items” refer to items spontaneously mentioned during the first part of the involvement methods that corresponded with one of the 22 items extracted from literature. “New items” refer to items spontaneously mentioned that were additional to the literature. We also noted whether the items hindered or facilitated the use of a genetic test for hand eczema susceptibility.
The output per participant of an involvement method was calculated by the total number of items (literature + new) or the total number of relevant remarks (literature + new) obtained per method, divided by the number of participants in that method, i.e. the mean number of items or relevant remarks per participant. The total number of items revealed per method could not be compared statistically as the total number of items is related to the combined group and not to individuals. For interviews and questionnaires, the number of remarks per participant was compared using Wilcoxon’s rank-sum test. The number of remarks per participant in the focus groups could not be compared statistically with that of the interviews and questionnaires because the number of remarks was only available per focus group and not per individual.
To establish (i.e. rule out) possible differences in participant characteristics between the methods, we applied the chi-squared test for dichotomous variables, the Yates and Cochran test for ordinal variables and one-way ANOVA for continuous variables. For this purpose, we used α = 0.1.