Background

Cancer survivors experience various difficulties in their daily lives, including physical, mental, social, and economic issues [1,2,3,4]. However, caregivers of cancer survivors (including family members, friends, and other people concerned) also encounter difficulties, and studies have shown that they can experience even greater anxiety than the survivors do [5,6,7,8,9]. Further, survivors of cancer, even when they are considered to be cured, often continue to experience difficulties associated with the disease, such as sequelae of treatment or psychosocial problems such as cancer stigma and changes in relationships with their acquaintances, which are medically difficult to solve [10,11,12,13]. Additionally, survivors and caregivers occasionally encounter difficulties in receiving adequate consultation time from medical providers, mainly due to time constraints, a desire not to trouble medical professionals, or communication problems. Thus, survivors and their caregivers have a range of unmet needs [14, 15].

Some studies have suggested that individuals’ backgrounds (age, sex, race, marital status, cancer type, treatment progress, treatment situation, etc.) are related to the specific types of unmet needs they experience [12, 16, 17]. This would be an important development for cancer care. If individuals’ backgrounds can be proven to predict specific types of unmet needs, we could determine the needs of survivors and caregivers in advance, and consequently provide adequate medical, psychosocial, and economic support. Such a predictive ability would also encourage survivors and caregivers to report their unmet needs and allow us to improve the efficiency of the care provided and make decisions based on survivors’/caregivers’ individual preferences. A few studies suggested a relationship between cancer type and specific unmet needs [16, 18,19,20], while the present study focused on the relationship. In clinical practice, cancers are conventionally dealt with by each category for the affected organ. Understanding how unmet needs vary by the type of cancer would make it possible to address those needs immediately and on an individual-basis, even under the current caregiving system.

Telephone consultation services are a promising means of identifying unmet needs. Such services allow medical professionals to respond to unmet needs that cannot be addressed through daily clinical practice [21,22,23,24]; for instance, these services allow survivors and caregivers to contact medical professionals even if they are geographically distant, are physically restricted, or do not wish to engage in face-to-face counseling [22, 25, 26]. Given their high utility, such services are widely used worldwide. Telephone consultations operate under a framework different from that of ordinary medical consultations, as the services are provided by professional consultation staff in locations remote from those of the survivors/caregivers, who can be contacted at any time.

Considering this, it is clear that records of telephone consultations are worth investigating because they potentially contain rich information regarding survivors’ and caregivers’ unmet medical needs that would otherwise be unattainable [6, 18, 27, 28]. There were some studies aiming to identify their medical needs from telephone counseling services [26, 29,30,31,32]; nevertheless, to our knowledge, no study was conducted to reveal the relationship between their unmet medical needs and individual background.

In this study, we qualitatively and quantitatively investigate the records of a cancer-focused telephone consultation service that was based in the Kanagawa Cancer Center Research Institute (KCCRI) in Japan. It was possible to conduct large-scale analysis of these data because a large amount of telephone consultation data was accumulated by the service, which allowed us to clarify the influence of individuals’ backgrounds on specific types of unmet needs.

Methods

Definitions

Some articles suggested various definitions of cancer survivors [12, 33]. In this study, we defined cancer survivors as people who were diagnosed with cancer and were still living, not including family members, friends, and caregivers. Caregivers were defined as people who have taken care of a cancer survivor, including family members, friends, and other people concerned.

Telephone consultation service

In 2006, the Kanagawa Cancer Clinical Research Information Organization (KCCRIO), based in Yokohama, Japan, initiated a telephone counseling resource for cancer survivors and their caregivers. The KCCRIO was founded by public and private enterprises in October 2006 and remained in operation until May 2014, when the service was concluded on being integrated into similar telephone counseling service that newly launched. Although this telephone counseling service was based in Kanagawa and, therefore, it is likely that it was used primarily by Kanagawa residents, the service was also widely utilized by cancer survivors and caregivers located outside Kanagawa, both within Japan and overseas.

At KCCRIO, which was located in the KCCRI, three nurses with high levels of clinical experience worked in shifts to provide consultation and support via telephone, and the callers could consult a physician four days a week if necessary. The callers remained anonymous (i.e., they did not provide identifying information) and the consultation was provided free of charge (although standard telephone fees applied). Prior to the consultation, nurses obtained consent verbally to record (in written form) the content of the counseling for research purposes, and in some cases, survivors/caregivers provided an evaluation of the quality of the counseling they had received at the end of the call. While providing telephone counseling, nurses tallied survey items on a specialized form and recorded the content of the counseling. The forms recorded age (both that of the survivor in question and the caller, if applicable), sex (both that of the survivor in question and the caller, if applicable), the caller’s relationship with the survivor (if applicable), the time of the consultation, the number of previous consultations the caller had received, the survivor’s cancer type, the site of the survivor’s cancer, treatment progress, presence of symptoms, survivor’s (and caller’s, if applicable) area of residence, the time it would take the survivor to travel to a medical institute, and how the caller learned of this consultation service. The main topics of the counseling were recorded in more detail by the nurse after the counseling had finished. Microsoft Excel was used to store the data, with 13,962 consultations being provided by the service between October 2006 and May 2014. For the qualitative and quantitative analysis, we randomly extracted 3000 of the 13,962 consultations using a random number table generated by Microsoft Excel.

For the present study, to analyze these data, we chose to adopt a mixed-methods approach, combining qualitative analysis (involving classifying consultation content into specific themes regarding needs) with quantitative analysis (comprising statistical analysis of the correlation between needs and callers’ specific backgrounds).

Informed consent was obtained from all participant verbally since the subject was anonymous telephone counseling service. This study and the procedure for verbal consent were approved by the institutional review board of KCC (H27 epidemiology-12).

Qualitative analysis

We identified individuals’ needs from the consultation records for each case and, by referring to the full transcript of each consultation, classified all needs into 16 need-related themes (physical, financial, education/information, personal control, system of care, resources, emotional/mental health, social support, communications, provider relationship, cure, body image, survivor identity, employment, and existential) to clarify the unmet medical needs of cancer survivors, which was proposed by Burg et al.; it was significant as a case of qualitative classification [12]. We calculated the total number of needs mentioned in each call, which was defined as the number of themes detected in the record for each call/case, in addition to calculating the frequency by which each theme appeared across all records. If we experienced difficulty classifying a need into an existing theme, or if there was no relevant theme, we decided to create new themes or redefine existing themes. For such difficult-to-classify needs, we held regular discussions on the best means of proceeding until a consensus was reached. We initially performed a pilot study in which we analyzed 1000 cases, and then, after discussions regarding the method, we added a further 2000 cases. All qualitative analyses were conducted by two authors (KW and HN).

Quantitative analysis part

Following the qualitative analysis, we conducted quantitative analysis. Using a statistical approach, we examined the relationships among the mean values; frequency of occurrence of the needs; and the survivors’/caregivers’ demographic, psychosocial, and medical information.

Regarding the number of needs, we performed a Mann-Whitney U test and Kruskal-Wallis test to determine the differences in the mean number of unmet needs associated with each caller’s characteristics. We performed multivariable logistic regression analysis to determine the relationship between themes with a frequency greater than 1% and each characteristic. We primarily found that cancer type affected specific unmet needs. As a secondary analysis, and conducted a similar analysis of age, sex, relationship with survivor, and treatment course. At this point, to determine the differences in the unmet needs between survivors and caregivers, which were previously reported, and to prevent an influence of confounding factors, we divided eligible cases into survivors and other callers (mainly caregivers) for analysis, and focused on the survivors.

Statistical analyses were performed using EZR on R commander, version 2.13.0. P values less than 0.05 were considered to indicate statistical significance [34].

Results

A total of 13,962 counseling cases were recorded by KCCRIO between October 2006 and May 2014; of these, 10,896 were first-consultation cases. We randomly sampled 3000 of the former cases. Given the aim of clarifying the distribution of unmet needs during telephone consultations, repeat callers’ consultations were considered inappropriate due to duplication of content. Therefore, after excluding 619 cases that were second (or higher) consultations and 443 cases whose context could not be interpreted due to lack of detailed information, we analyzed the remaining 1938 first-consultation cases. Table 1 shows the characteristics of the 1938 cases analyzed, as well as basic details regarding the 10,896 first-consultation cases. No items showed a significant difference (p < 0.05).

Table 1 Characteristics of the analysis subset in comparison to those of the full survey sample
Full size table

Themes of unmet needs

In the qualitative analysis, as a result of slight differences between the needs-based themes we identified and the theme definitions proposed by Burg et al. [12], we made minor modifications to three of Burg et al.’s theme definitions (“communication,” “system of care,” and “cure”), and applied the remaining 13 directly. The modified points are shown in Additional file 1. Table 2 shows the modified three and the original 13 theme definitions and the frequency of occurrence of each need theme. The mean number of needs mentioned per consultation was 1.58 (standard deviation (SD) = 0.86; Table 3). Regarding types of cancers (Table 1), callers with multiple primary cancers had the largest mean number of needs 1.75 (SD = 1.20). Callers whose consultations were over 15 min in length had the highest mean number of needs 1.89 (SD = 1.04).

Table 2 Unmet needs themes (frequency and description)
Full size table
Table 3 Average number of unmet needs stratified by each characteristics
Full size table

Differences in specific cancer types in survivors

In the multivariate analysis, we excluded survivor’s sex as a result of its significant correlation (in the chi-square test, p < 0.001) with caller’s sex.

In the primary analysis, for callers who were survivors, there were no significant differences in needs frequency among cancer types, except that individuals who were never diagnosed with cancer had a lower frequency of “physical” and “financial” needs than did survivors with breast cancer (Table 4).

Table 4 Logistic regression analysis (survivors)
Full size table

Effects of other factors on survivors

In the secondary analysis, it is noteworthy that survivors with symptoms showed a higher frequency of “physical” and “emotional/mental health” needs than did survivors without symptoms (Table 4). Moreover, survivors with symptoms showed a lower frequency of “education/information”, “resource”, and “cure” needs than did survivors without symptoms.

Other statistically significant results are as follows: about sex, female participants showed a higher frequency of “physical” and “emotional/mental health” needs and a lower frequency of “resources” needs than male participants. About age, older age groups showed a lower frequency of “employment” needs than did the younger age group. About treatment course, compared to survivors with pretreatment, survivors with treatment ongoing and completed showed a higher frequency of “physical” needs. Survivors with ongoing treatment had a lower frequency of “resources” needs. Those with completed treatment had a lower frequency of “cure” needs. Other results are in Table 4.

We also conducted the similar analysis in callers except survivors (Additional file 1 and Table A1), and performed three additional, though slightly modified, logistic regression analyses. The results of the additional analyses are shown in Additional file 1, Table A2, A3, A4, A5 and A6.

Discussion

Small effects of specific cancer types

The primary findings of this study are that cancer type (and cancer site) has a small effect on survivors’ unmet needs and that their backgrounds—such as survivors’ age, sex, treatment course, and presence of symptoms—play more important roles in this regard as independent factors. This result is not consistent with those of previous studies, which suggested strong relationships between specific cancer types and specific unmet needs. For example, several studies have reported that breast-cancer survivors tend to have psychosocial needs [12, 18,19,20].

While cancer type may be merely a confounding factor for the influence of other backgrounds, three explanations for this inconsistency must be considered.

First, in this multivariate analysis, only the frequency of occurrence of each need was analyzed; the strength of each need was not. Therefore, the effect of strength might be underestimated in our results. Consequently, further studies that measure strength of needs are required. “The Supportive Care Needs Survey” is worth investigating in this regard [35].

Second, longer disease duration, which is usually related to type of cancer, might foster more serious “financial,” “personal control,” “social support,” and “employment” problems, as reported in previous studies [36,37,38]. The low frequency of these needs in this study might have resulted in insufficient statistical power to detect the impact of differing cancer types. There is a possibility that cancer types and sites affect needs, as a result of their influence on disease durations, but some needs themes may have been insufficiently frequent to show statistically significant differences in this regard.

Third, in clinical practice, different cancers cause different symptoms; for example, lymph edema after axillary lymph node dissection for breast cancer, decreased respiratory function after lung-cancer dissection, gastrointestinal obstruction for digestive cancer, and hot flashes and sexual function disorders as a result of hormone therapy. In this study, all needs regarding symptoms were integrated into the “physical” needs theme. This might have resulted in an underestimation of the differences among cancer types.

Further, this study indicated that differing needs among cancer types might be an indication of differences in people’s backgrounds in regard to major needs such as “education/information,” “resources,” and “emotional/mental health,” which we considered to have a sufficient frequency for analysis. As a result, we were able to examine these needs based on survivors’ individual backgrounds. However, for less-frequent needs themes and the “physical” theme (as mentioned above), we were unable to draw definite conclusions. Consequently, future large-scale studies are needed.

Other interpretations of the results

There are more interpretations that are worth considering. Older survivors seem to have more physical problems, since they may have age-related decline in physical function and frailness. Moreover, they also seem to have lesser needs in terms of employment given their retirement from work. The findings support the hypothesis that older survivors tend to have more “physical” needs, while they have less “employment” needs.

Given the lack of enough data in this study, future studies should explore other possibilities based on the findings. For example, although previous studies have shown the relationships among sexual needs and cancer survivor’s background (sex and age) [39, 40], the present study could not consider this sexual needs, as those are rarely discussing in telephone consultations. This is likely because it is not common to talk about sexual concerns publicly in Japan. This in itself may indicate a social barrier and further research is needed to gain a better understanding. Further, although the relationships between emotional/mental health needs and social network have also been reported in a previous study [41], survivor/caregivers’ social support was not measured in the present study because of the passivity from the telephone consultation service. In our study, “emotional mental health” need was observed in 7 out of 16 cases (43%) of survivor/caregivers who had “social support” needs. We performed a Fisher’s exact test for the presence of these needs and found no significant difference (p = 0.056). Future studies should investigate this in more detail.

Limitations

This study had some limitations. First, available data were obtained from telephone consultations; in these consultations, all calls were instigated in a spontaneous manner by survivors or caregivers. As a result, we could not evaluate unmet needs that were not mentioned in the consultations, or the needs of people who did not receive a consultation. This may have led to a selection bias, for some deference of results such as relationships between unmet needs and age, in contrast to a previous study, which included general patient population, although the proportions of sex and age of the sample were similar to those in our study [12]. In a study by Burg et al., the most frequent theme of unmet needs was “physical,” followed by “financial” and “education/information,” while in the present study, the most frequent one was “resources,” followed by “education/information” and “cure.” This suggests that the findings of the present study may have been strongly affected by the nature of telephone consultations as avenues to acquire information. In addition, approaching telephone consultation requires activeness, and individual personality traits may have affected the distribution of unmet needs in the present study. A previous study showed that personality traits may influence one’s ability to cope with living with cancer [42].

Second, we excluded repeat callers’ consultations from our analysis. Therefore, we may not have obtained an accurate evaluation of individuals who have many and severe unmet needs. Nevertheless, as the purpose of this study was to clarify what unmet needs survivors and their caregivers generally had; individuals who have many and severe unmet needs should be studied in the future.

Third, the qualitative data we analyzed were summaries recorded by each consultant, and these were not recorded word-for-word. Consequently, detailed context might have been lost, or may have been affected by the perspectives of the consultants.

Finally, some minor themes appeared too infrequently to be examined sufficiently; large-scale research that can investigate such minor themes is consequently needed.

Conclusions

In conclusion, this study showed that cancer type was not a significant factor for specific unmet needs, and that individuals’ backgrounds and presence of symptoms play a more important role. Through this study, it was found that instruments to predict people’s needs and a system to provide individualized cancer care across cancer types should be developed in the future.