The study carried out in our center reflects a challenging recruitment process with 185 enrolled dyads out of 1562 candidates (patient or caregiver) who were contacted. Of the 1562 candidates, 22.6 % were ineligible on initial contact, 56.7 % declined to participate on initial contact, and 8.9 % orally accepted the invitation to participate but did not complete the enrollment. Of those who declined to participate on initial contact and provided reasons for declining, 35.0 % were not interested, 33 % were interested, but wanted to think about it and contact the researcher team later, and 13 % did not feel the need to participate in the study. The study also presents the response rate and evaluations of seven different recruitment approaches, testing both opt-in and opt-out methods. Our data show that 2 of the opt-out techniques used, on-site recruitment and the routine care letters recruitment strategy, yielded a higher number of recruited participants with 79 dyads and 58 dyads respectively, constituting 42.7 % and 31.4 % of the total number of enrolled candidates. The other 5 recruitment approaches provided a total of 49 dyads to the study, and relying on providers at the hospital represented almost half of them.
Treweek and colleagues conclude that opt-out techniques must be considered carefully due to methodological and ethical aspects [7]. They maintain that opt-in techniques benefit patients and are, therefore, more ethically sound, because the patients have explicitly agreed to be contacted by the researcher. However, Hewison and Haines [14] argue that the requirement for patients to agree to be contacted by the researchers may also affect the quality of the primary research. They argue that the opt-in approaches can lead to low response rates, which results in research with limited validity, and wasted resources being used [14]. Trevena and colleagues also suggest that trials conducted with opt-in methods are more likely to include participants who are active health decision-makers, which might affect generalizability [16]. The results of our study are consistent with studies that show higher response rates associated with using an opt-out recruitment strategy compared to opt-in strategies [15, 16]. The low response rate for an opt-in strategy may be due to the responsibility this approach places on the recruitment candidates. The candidates must notice the information, recognize if they are eligible, and remember to contact the researcher. Our data indicate that the opt-in strategy, with advertisements in newspapers, gave a very low yield of enrolled patients for a relatively expensive advertisement fee in addition to required considerable researcher time similar to all the other employed recruitment approaches employed in this study. Advertising in two newspapers yielded 10 enrolled participants, at a cost of approximately USD 1000. However, those who actually contacted the researcher team due to the advertisement showed more interest in participating: 10 out of 13 candidates (77 %) completed the enrollment, representing the highest number of enrollments among the recruitment approaches used in this study. Advertising on the Internet was free of charge and thus much cheaper as it required less researcher time. In addition, this method had a greater potential for information reaching the target individuals. However, the response rate for the Internet approach in the present study was low (11 dyads showed interest, and 6 of them were enrolled into the study). A reason for the low response rate might be the overwhelming amount of information available on the web, making it difficult to distinguish reliable from questionable information. Although we did not need to pay an advertisement fee for some opt-in recruitment strategies, it is worth mentioning that all the recruitment approaches employed in this study required considerable researcher time and were thus expensive.
When using on-site recruitment by health providers, Miller [15] experienced challenges similar to those found in our study. Reported challenges were providers’ lack of time, forgetting to mention the study to participants, recruitment confusion, and not prioritizing recruitment. However, this recruitment technique is still a commonly used approach. Our data indicate a relatively low response rate (28 enrolled dyads). Additionally, this method demanded an effort from the researcher team providing the health personnel with support and information. The same challenges were observed for recruitment at the rehabilitation center.
The two opt-out techniques (the routine care letters recruitment strategy and the on-site recruitment) resulted in a larger proportion of participants, as mentioned earlier. An advantage of using the on-site recruitment approach is the possibility of establishing a personal relationship and a growing trust between the researcher and the potential participant. This may work as a positive factor in recruitment, but it might also result in oral agreement to participate just to be kind to the researcher without real interest in participation, in addition to introducing sampling bias. Using this method, of 1181 potential candidates who were contacted, only 6.7 % were enrolled into the study and it was, therefore, not a cheap approach as it required considerable researcher time. A comparison of the on-site recruitment approach and the routine care letters technique used reveals several positive aspects of the routine care letters technique: i) in contrast to on-site recruitment (face-to-face strategies) in which candidates are encouraged to say “yes” or “no” on the spot, the routine care letters technique provided the individuals with some time to consider whether to participate or not; ii) by routine care letters approach, patients were contacted when they were at home by phone in a more relaxed environment and not before or after their daily treatment; iii) it removed the burden of on-site recruitment from the researcher; iv) it gave the individuals an opportunity to be prepared before the information conversation with the recruitment team; v) it gave the researchers an opportunity to precisely target a specific population and avoid contacting ineligible candidates; vi) the strategy did not rely on the physical presence of participants in a clinical setting, and vii) it was independent of physicians and other providers to oversee recruitment. Steinhauser with colleagues tested a method similar to the routine care letters recruitment strategy, named an “alternative strategy,” and concluded that patients were less overwhelmed and more informed prior to the initial telephone call [6]. The biggest challenge with this approach in our study was the large number of people who did not answer the phone and those who had not read the information prior to the phone call. The latter can be easily solved by sending the information in a separate letter. Additionally, like the other approaches, this approach resulted in a high proportion who did not complete the enrollment even though they had agreed orally to participate. Therefore, the technique needs further improvements, and we support Ewing with colleagues [8] in pointing out that it is generally important to secure additional resources in research budgets to improve recruitment and to take into account unforeseen and extra challenges during the process.
In the current case study, there were some prominent challenges in the recruitment process. First, the recruitment procedures became more complex because they included both patient and caregiver. Study population including patient and caregiver is emphasized to be more challenging because it may require an extended recruitment phase due to the higher number of people to contact and to ask for consent [17]. This was confirmed in the present study, and the requirement for dyads was one of the reasons for drop-out of candidates who had shown interest in participating or who had agreed to participate. In addition, the high response rate in the category “lack of interest” might be because both patient and caregiver needed to be interested. The second prominent challenge was that the case study involved an intervention testing a specific web-based support tool. To become participants, the individuals should, therefore, perceive the tool as useful and have personal motivation and interest to try it. In this study, some declined to participate because they had no interest in using a computer or in using it in connection with their sickness. However, such responses might provide useful information about the market and need for implementing the product in a population. The third prominent challenge was to recruit seriously ill people. Patients might be too sick or too fatigued to participate and might have their focus on going through treatment and recovering. Some responded that they might have been interested in the project and the web-based support system in another period of their sickness, such as when they received the diagnosis. We experienced that cancer patients’ stage of health at the time of approach was important for not causing an extra burden for patients as well as for the recruitment. For example, head and neck cancer patients have much poorer health when they receive radiation treatment compared to prostate cancer patients. For head and neck cancer patients it was best to contact them at the beginning of the radiation treatment when radiation-induced side effects have not yet developed and they are able to talk to us. Therefore, it is essential to investigate the best possible timing of recruitment in the course of the sickness. Additionally, both the patient and the caregiver have the treatment and sickness on their minds. This might be a reason why people orally agreed to participate or showed interest in participating, but did not complete enrollment. Employment of methods to reduce the burden for potential participants is important. However, it is difficult to be sure whether the researcher doing as much as possible to facilitate participation by, for example, having the responsibility for making the contact, sending reminders to facilitate participation, etc, will contribute to reduce the burden or not. Therefore, there is a need for further investigation in this area.
In the process of designing the recruitment techniques, we highly recommend seeking inspiration from mistakes and success factors from previous studies. Learning from previous studies and focusing on a thorough planning phase might prevent the need for adjustments after implementation and enhance the chance of stable recruitment in the same population. That will most certainly pay off when the project is initiated.
A limitation for this study could be the lack of appropriate comparator groups to control for confounding factors and co-interventions, making the direct comparison of opt-out and opt-in strategies challenging. Another limitation for this study is the lack of information about the number of people reached by each recruitment approach since the total number of contacted candidates were not documented by all the health staff involved in the recruitment. The third limitation for this study is lack of quality assurance and documentation as to how the information has been given to the patients by the health staff involved in the recruitment. It is also not possible for the study to describe or include the possible ethical implications; i.e. it is not possible to document what the contacted candidates may have felt having impact on their choice of participation in the study.