Participation of older newly-diagnosed cancer patients in an observational prospective pilot study: an example of recruitment and retention

Cancer is a common health problem for older persons. Nevertheless, the manifestations and course of cancer in this older population have not been well investigated and remain poorly understood [1‐6]. Moreover, with the aging of the population both the absolute and the relative number of older patients affected by cancer can be expected to increase in the future [7, 8]. There is increasing interest in conducting research on health and functional status in older cancer patients [9, 10]. It has been suggested that frailty might be a useful concept to identify vulnerable older patients at risk of adverse outcomes of treatment who then might benefit from comprehensive geriatric assessment [11, 12]. Frailty represents a state of reduced homeostasis and resistance to stress, due in part to aging, which leads, in turn, to increased vulnerability and risk of adverse outcomes although there is no consensus on the definition. Fried et al [13] have developed a now commonly used approach to measure frailty. The five characteristics of their frailty phenotype are: weakness, poor endurance, reduced physical activity, slow walking speed, and unintentional weight loss during the preceding year. Cognitive impairment and depressed mood have recently been added [14, 15]. Unlike their non-frail contemporaries, frail older individuals appear unable to withstand stressors, such as environmental stress, injury, and acute illness. These stressors may provoke a downward spiral, whereby the frail older individual is unable to recover and return to the baseline state [13, 14, 16]. In older patients, cancer treatments, especially chemotherapy, are considered strong stressors that will reveal which patients have sufficient functional reserve to regain stable homeostasis [3, 17]. To our knowledge, there has been no prospective observational study investigating the usefulness of the concept of frailty in the field of geriatric oncology.

There have been many studies that have reported that older patients are underrepresented in cancer clinical trials due to a variety of reasons [6, 18‐27]. Furthermore, investigators have reported that recruiting older patients to any clinical trial is difficult due to co morbidity, cognitive problems, sensory deficits, transportation issues, study burden and influence of significant others [28‐35]. As observational studies do not bring personal advantage or direct benefits for patients as do clinical trials, the reasons for participation and refusal might be different than for clinical trials. Newly-diagnosed cancer patients may also be asked to participate in a cancer clinical trial at the same time as being asked to participate in an observational study which may increase the burden on the older patients and may affect response rates and retention rates in both studies. Few observational studies have reported information on the number needed to screen to recruit the study sample, or the number and reasons for refusal and drop-out [36, 37]. Furthermore, most of the studies have been cross-sectional with small sample sizes and/or with only one type of cancer diagnosis included [38‐42]. The studies with older cancer patients have defined their older patients with no cut-off based on age [36, 37], 65 years and over [39, 41] or 70 years and over [38, 40, 42]. More information on recruitment and retention strategies may facilitate or increase participation of older newly-diagnosed cancer patients in studies on health and functional status.

In the context of an observational prospective pilot-study conducted to describe the health and vulnerability of older newly-diagnosed cancer patients and the adverse outcomes of treatment, we evaluated and report our recruitment and retention strategies to optimize their use for future larger studies. As we were interested whether the concept of frailty is useful to identify older patients at risk of adverse outcomes of cancer treatment, it was crucial to conduct the baseline interview prior to cancer treatment. The window of opportunity for recruitment of older newly-diagnosed cancer patients was thus narrow, making recruitment more challenging. The aim of this study was to examine the feasibility of recruitment older newly-diagnosed cancer patients prior to treatment into an observational prospective pilot study and to retain them during a six-month follow-up period.

From March 1^st 2007 until May 1^st, 2008, 3060 medical charts of patients aged 65 and over were screened in the Segal Cancer Centre; see Figure 2. Of those 3060 patients, 2838 were excluded (ineligible) because they were treated for cancer in the past 5 years, they were being seen for follow-up appointments after cancer treatment or the patient was investigated for cancer but did not have cancer. There were 222 newly-diagnosed patients to review for further eligibility. Of those 222, 66 were excluded (see Figure 2 for the reasons) leaving 156 eligible patients. Of these, 116 patients agreed to participate. 114 signed informed consent and two persons had to be excluded because they received treatment before the planned interview took place. Of the 114 patients who signed informed consent, one patient died before the baseline interview. A second patient completed the baseline interview but following a second opinion the cancer diagnosis was changed to a non-malignant diagnosis and the patient was excluded. Finally, 112 participants completed the baseline interview (72% of all eligible patients).

Of the 40 patients (25.6%) who refused to participate, 16 reported that they felt overwhelmed/sad by the diagnosis, were too anxious in relation to the start of their cancer treatment, 5 patients did not feel well enough to participate, 5 patients had no time (due to care giving responsibilities, moving or work, or start of treatment next day), 2 patients felt too healthy, 10 persons were not interested, 1 patient refused due to study burden as the patient was also enrolled in a clinical trial and the last person refused due to denial of the cancer diagnosis. However, half of them indicated that they would be willing to participate after the start of treatment (i.e. knowing then what the treatment would be like).

The baseline characteristics of those who participated, those who refused to participate and those who were excluded are presented in Table 2. The only statistically significant difference between those who participated and those who refused was that participants were less often married/living common-law (p = 0.006).

The baseline characteristics of participants who completed the study, died during the study or refused further follow-up interviews are presented in Table 3. Due to the small number of patients who refused, statistical analyses were not conducted comparing these groups. During the study, 14 participants died (12.5%) before the last interview and one patient died following completion of the first half of the interview. As shown in Table 3, those who died seemed to be in worse health and functional status at the baseline interview compared to those who completed the study or refused further follow-up.

Seven participants (6.3%) refused follow-up after the baseline interview. There was an additional patient who refused the third and fourth interview due to a lack of time but participated in the final interview. The reasons for refusal were not longer interested (3), no time (1), too overwhelmed by diagnosis and treatment (2), sick husband (1).

Sixty four (57.1%) of all baseline interviews were conducted at home and the remainder were conducted in the hospital, see Table 4. During the course of the study, 61 interviews were postponed by 44 participants. The median time needed to complete the baseline interview (n = 87) was 45 minutes (Interquartile range (IQR) 40–55), 10 minutes for the second interview (N = 77) (IQR 8–15), 45 minutes for the third interview (N = 70) (IQR 35–50), 10 minutes for the fourth interview (N = 70) (IQR 10–15) and 40 minutes for the fifth interview (N = 71) (IQR 35–55). The results of the feasibility of the study according to patients' characteristics showed only some minor differences were observed with not one group for whom participation was less feasible [see Additional file 2].

During the study, 2 participants became very ill and proxy interviews were conducted (both participants had lung cancer). For two other participants with cognitive impairment a proxy respondent completed the telephone interview while the face-to-face interviews were conducted with the patient and the proxy together. Only two participants used the transportation offered to come to the hospital for an interview. Nobody used the offer to be compensated for transportation costs to the hospital.

In addition, patients were asked their opinion on participating in this study and 84 participants provided feedback (see Table 5). Most did not identify specific questions that they felt were unpleasant or too difficult but did mention that nothing was unpleasant or too difficult. The majority of patients indicated that in their opinion there was no item or topic missing. A third of patients indicated that they had enjoyed participating in this study.

Our results show that it was feasible to recruit newly diagnosed cancer patients prior to treatment although this did require considerable time and effort. Although no information was recorded systematically on the number of hours necessary to recruit potential participants by members of the research team, we estimate that on average it took 3 hours to recruit the potential eligible subjects. This estimate does not include the time needed to assess eligibility. One member of the research team worked fulltime on the recruitment and conduct of interviews. We found it necessary to modify our strategies depending on cancer type. It was more difficult to identify and recruit eligible patients with colorectal cancer compared to the other cancer diagnoses for which the research team had access to the clinic lists. For colorectal cancer the treating surgeon had to alert us when they had a new patient. Regardless of strategies implemented to remind physicians, we found that it was important to be present in the clinics when physicians see their patients to reinforce the reminder. We found that at the time of recruitment it was important to meet the patient in the physician's office so that the patient would know who was going to call them to explain the study. This finding is in keeping with the work of others [28]. We also observed that patients wanted to know the treatment plan before they could decide whether or not to participate in this pilot study. During the informed consent procedure a number of questions were asked about the possibility of withdrawing from the study. It was stressed many times that they could withdraw at any time without providing a reason. Despite these initial concerns, during the study this was not a major issue as only two participants dropped out of the study because of the burden of treatment. We also noticed that some patients and family members had different views about whether or not they should participate in the study. In some cases, family members were concerned about the study burden in combination with the cancer treatment even when the patients wanted to participate. When this arose, this concern was successfully addressed by also giving family members all the study information to help them understand the study. Interestingly, in a few cases, patients refused to participate and subsequently their family members asked if they could participate as a proxy, a strategy not included in the study design.

It was important to actively recruit patients after they had agreed to be contacted for the study as none of the patients contacted the research team for information about the study. This is in agreement with Townsley et al. [27] who conducted a survey among elderly patients concerning their willingness to participate in cancer clinical trials and found that most elderly patients were willing to consider participation but did not appear to actively seek information about trials. Our overall recruitment was 72% of eligible patients in spite of the limited time window available to recruit patients. This rate is about the same as reported in other studies with older cancer patients. It is higher than another study with newly-diagnosed cancer patients with breast, colon, lung or colorectal cancer who were recruited within 6 weeks of initial surgery or within 2 weeks of initiating chemotherapy or radiation therapy who reported a response rate of 51% [36‐38]. However, a high number of patients had to be screened to recruit those patients. A review on attrition in large longitudinal aging studies (>1000 participants) around the world showed initial response rates of 63 to 93% [57]. Our response rate compares very favourably considering that this is a population of newly-diagnosed cancer patients recruited before the start of treatment.

We compared the participants on several characteristics to identify potential differences in the feasibility of recruitment and retention in this pilot study. There were only minor differences observed. It seemed that those with a poorer functional status needed more time to complete the interviews, the study overall and required more meetings to complete an interview. Similar findings were reported by Pijls-Johannesma et al[58] in a study comparing different quality of life instruments. They found that patients with lung cancer who all had a poor performance status, needed assistance more often to complete the assessment tool than patients with breast cancer. In addition, we found that patients born outside Canada, required more effort to complete an interview and more often had an incomplete interview. Patients were included in the study if they could understand and speak English or French and the study material was provided in English and French. However, some of the foreign-born participants found the questionnaires were too difficult and this may have led to the higher number of incomplete interviews. The cognition tests and the HADS were most often missing for foreign-born patients. In the 2006 census http://​www.​statcan.​ca, 19.5% of Canada's population was born outside Canada. For future studies it is important to pay attention to the feasibility of study design and measurement tools of foreign-born older adults.

We noted that once enrolled in the study, the retention rate was high (81%) despite the fact that the majority of participants were undergoing active cancer treatment. Most patients did not feel that the interviews were too long. From the feedback received from patients, most had enjoyed participating in the study. In the few prospective studies published with older cancer patients, the attrition rate ranged from 2% to 37%, in part due to study differences in baseline age and disease status, treatment regimes and length of follow-up time (ranging between 6 months to 1 year follow-up) [38, 40, 41, 59‐61]. The review on attrition in large longitudinal aging studies [57] showed that attrition was associated with increasing age and those who are in worse health. In our study, attrition due to refusal was highest among patients with colorectal cancer which might be due to the fact circumstances and different method of recruitment and attrition due to death was highest amongst patients with lung cancer. These findings are in agreement with the study of Neumark et al. who found that dropout was highest among patients with colorectal and lung cancer [37]. At a University teaching hospital newly-diagnosed cancer patients may also be asked to participate in a cancer clinical trial at the same time as being asked to participate in an observational study which may increase the burden on the older patients and may affect response rates and retention rates in both studies. Unfortunately we did not have the information on how often the patients who refused to participate were asked to participate in a clinical trial. However, one patient who refused to participate told us that she was participating in a clinical trial and thought the burden of both studies would be too much. In addition, one physician asked that we not contact an eligible patient because he wanted to enrol the patient in a clinical trial and thought both studies would be too much. Nevertheless, none of our dropouts during the study were participating in a clinical trial. Future studies should take into account the double study burden of participating in both an observational study and clinical trial.

The planning of interviews with participants undergoing cancer treatment needed to be flexible as a significant number of participants had to postpone and reschedule interview appointments due to a variety of reasons. This is in agreement with the study of Shipman et al [62] where patients with advanced lung or colorectal cancer (n = 20) were interviewed to determine their views about participating in serial questionnaires studies. These authors also reported that the interview appointments were often postponed and patients reported that an interview duration of 30–45 minutes in length would be preferred with a preference for face-to-face interviews which is in agreement with our findings [62]. It was also important for participants, as far as possible, to have the interviews done by the same person and to offer home visits and transportation. For each face-to-face data collection about half of the interviews were conducted at home. Furthermore, participants in our study who were born outside Canada and who were not native English or French speakers preferred to have a family member present during the interviews in case they might not understand the question. In most of these cases, interviews were done in the oncology centre while participants were with a family member waiting to see their oncologist and this is partly the reason why some interviews were postponed until the next appointment.

To our knowledge this is one of the first studies with older newly-diagnosed cancer patients examining the recruitment and retention strategies for an observational study that begins prior to cancer treatment. A limitation of our study is that we only recruited participants at one cancer centre and the recruitment and retention process in other hospitals may differ. Due to the small numbers of patients who refused, statistical analyses to compare the characteristics of the refusers to the participants were not possible. Nevertheless, many of the lessons learned in our study (same interviewer, having participants choose the place and time of interview and identifying a proxy during the informed consent procedure) are most likely applicable to other sites when trying to recruit older newly-diagnosed cancer patients. The feedback obtained shows that older newly-diagnosed cancer patients are willing to participate in observational studies while receiving cancer treatment. There is a need to refine the recruitment and retention strategies for conducting observational research with older cancer patients.