Introduction

Due to diagnostic and therapeutic advances in oncology, many patients diagnosed with cancer today will overcome their disease to become long-term cancer survivors [1, 2]. For a large number of these individuals, secondary cancers, potentially preventable chronic conditions, such as heart disease and diabetes, and late effects of cancer treatment may pose significant threats to their life and wellbeing [3, 4]. To ensure that the survival gains from cancer therapeutic and screening advances are not lost because of competing health risks, posttreatment care of cancer survivors that encompasses prevention, early detection and management of disease recurrences, and late effects has become increasingly recognized as an integral component of comprehensive cancer management [5–7]. A landmark report published by the Institute of Medicine (IOM) emphasized the diverse range of physical and psychosocial issues faced by cancer survivors and underscored the significance of survivorship care as both a clinical and research priority [8, 9].

The aging of the general population is expected to result in growing numbers of cancer survivors, leading to increasing demands on oncologists who are already in short supply and further burdening limited oncology resources [10]. There is also an increasing recognition of the importance of noncancer-related care and general health maintenance for survivors, contributing to growing interest in developing survivorship care models that are based on shared-care principles and do not rely solely on oncologists. These include transferring care to or sharing responsibilities with primary care providers (PCPs), specialized nurses, physician assistants, and other nonspecialist providers. Efforts are needed to minimize fragmentation and reduce duplication of services, especially as patients transition from active treatment to survivorship [11, 12]. While preliminary research has indicated that survival outcomes and patient satisfaction are similar for PCP follow-up and specialist care, the perspectives of these different physician specialties towards providing survivorship care remain largely unknown [13]. We previously reported on the results of a survey that ascertained PCPs’ and oncologists’ preferences for different cancer survivorship care models [14]. In the current analysis, we evaluated how physician attitudes towards and self-efficacy with cancer follow-up affected these preferences. An improved understanding of physician preferred models of survivorship care and the correlates of these preferences can help with survivorship care planning and inform strategies to enhance future models of care.

Methods

Survey of Physician Attitudes Regarding the Care of Cancer Survivors

Sponsored by the National Cancer Institute (NCI) and American Cancer Society, the aims of Survey of Physician Attitudes Regarding the Care of Cancer Survivors (SPARCCS) were to: identify current practices and potential barriers of care to cancer survivors, assess provider needs, and highlight opportunities for improving survivorship experiences for cancer patients and physicians. Survey questions were designed to characterize the perceptions, knowledge, coordination efforts, and practices of PCPs and oncologists with respect to different aspects of posttreatment care for cancer survivors. The overall goal of the survey was to collect information that can be used to enhance the care of survivors, consistent with recommendations of the IOM. Approval was obtained from the NCI’s Institutional Review Board and US Office of Management and Budget prior to the conduct of this study.

Content of the questionnaire

Items were drawn from other published surveys and supplemented with specific questions written for SPARCCS [15]. A modified Delphi approach and an iterative process of pilot testing and revisions were conducted during the development of the questionnaire. In the final version of the survey, respondents were asked to indicate their current attitudes, awareness, and practice patterns in several key areas of cancer survivorship, including: preventive care, monitoring for recurrence and secondary cancers, management of physical and psychosocial late effects resulting from cancer and its therapies, use of treatment summaries and survivorship care plans; coordination of care among different providers, preferences for different cancer follow-up care models, and perceptions about physician roles during survivorship. The survey also collected information on relevant physician demographics and practice level characteristics. Two separate but complementary versions of the survey were administered: one was distributed to PCPs and another to oncologists across the USA. The dual survey design allowed for comparisons of the perceived roles, knowledge, and practices between these two key provider groups. The full survey may be requested from the NCI at http://healthservices.cancer.gov/surveys/sparccs/[15].

For this analysis, which focuses on physician preferences and attitudes, respondents were asked to identify their preferred survivorship care model among the following categories: (1) PCPs have the main responsibility for survivorship care; (2) PCPs share this responsibility with other cancer specialists and providers; (3) oncologists have the main responsibility; (4) follow-up should take place in specialized clinics led by physicians who focus exclusively on cancer survivor care; and (5) follow-up should occur in specialized clinics managed by nurses, nurse practitioners, physician assistants, or other nonphysician providers with specific survivorship training. We subsequently collapsed these into three possible survivorship models: PCP/shared care (categories 1 and 2 together), oncologist, or specialized survivor clinics (categories 4 and 5 together), primarily due to the low number of responses in some of the original categories (Appendix). To evaluate the respondents’ attitudes about specific elements of survivorship care, we included survey items that asked them to rate their self-efficacy in detecting cancer recurrences as well as in managing cancer and its effects for either breast or colorectal cancer, using a 3-point Likert scale (i.e., not at all, somewhat, or very confident). Considering the growing interest in shifting certain aspects of follow-up from specialists to PCPs [11–13], respondents were further asked to indicate their attitudes about PCPs’ skills in conducting routine cancer and noncancer-related follow-up, cancer surveillance, psychosocial support, and assuming primary responsibility of survivorship care for either breast or colorectal cancer separately. Responses to these statements were rated on a 5-point Likert scale (i.e., strongly disagree, somewhat disagree, neutral, somewhat agree, or strongly agree). We subsequently grouped original responses from both the 3-point and 5-point Likert scales into two categories: very confident versus not and somewhat/strongly agree versus all others, respectively, mainly because of sample size considerations and our interest in examining responses for both breast and colorectal cancer together. Finally, we asked PCPs if they were currently involved in surveillance of their patients for cancer recurrence.

Survey sampling and distribution

Using a systematic stratified sampling strategy based on physician type (PCPs vs. oncologists), the American Medical Association files were reviewed to identify a list of potential subjects from which a random sample was drawn. Telephone screening was subsequently conducted by our survey contractor Westat Inc. (Rockville, MD) to make a preliminary determination of study eligibility. Following these steps, an initial survey package that contained an introductory cover letter, a fact sheet about the study, a copy of either the PCP or specialist questionnaire, and a postage-paid envelope for returning the completed survey were mailed to eligible US practicing clinicians from the random sample between March 2009 and December 2009. Regardless of whether or not the physician returned the survey, a monetary check of $50 USD was included with the initial survey package to compensate for any time spent completing the questionnaire and to optimize response rate. Details about sampling methods, survey distribution and reminders, and the mailing of additional survey packages to nonrespondents have been previously described for SPARCCS [14]. For the current analyses, we only included physicians who were actively practicing, were under 70 years of age, and those reporting that they spend more than 20 % of their time involved in patient care. For oncologists, we included only those involved in the treatment of breast and/or colorectal cancer. We excluded respondents who indicated either (a) no preferences for any survivorship care models or (b) more than one preferred survivorship care model. Of note, baseline characteristics of excluded respondents were similar to those of the included sample.

Statistical considerations

Baseline demographics, practice level characteristics, and physician preferences for cancer survivorship care were summarized with descriptive statistics. Clinical and practice characteristics were compared between PCPs and oncologists using Chi-square tests. Multinomial logistic regression models with survivorship care model preference as the dependent variable and physician attitudes about follow-up and self-efficacy with their own skills as the main independent variables were subsequently constructed while controlling for physician and practice characteristics. Models were fit separately for PCPs and oncologists, and interactions between statistically significant attitudinal variables and physician age, sex, region, and training were explored for both models. For the dependent variable, an oncologist-based survivorship model was selected as the reference category since cancer specialists have traditionally conducted the majority of cancer follow-up. To ensure that results from this study were representative of the US physician population from which it was selected, appropriate sampling weights were applied. Stata Version 11 (StataCorp LP, College Station, TX) and SUDAAN software Version 10.0.1 (Research Triangle Institute, Research Triangle Park, NC) were used to conduct the statistical analyses. All tests were two-sided and p ≤ 0.05 was considered statistically significant.

Results

Respondent characteristics

The weighted survey response rate that excludes nonlocatable physicians was 65.1 % (also defined as the “cooperation” rate) [14]. The survey’s absolute weighted response rate, which incorporates unscreened physicians with unknown eligibility, was 57.6 % [14]. This yielded a total survey sample of 2,202 respondents (1,072 PCPs and 1,130 oncologists). Another 134 PCPs and 42 oncologists were excluded, respectively, because they selected either no or multiple survivorship care model preferences, thus leaving 2,026 eligible for analyses. Among these participants, 938 (46 %) were PCPs and 1,088 (54 %) were oncologists. Physicians who were White, male, trained in the USA, and worked in an office-based practice were highly represented, though nonrespondents who formally opted out of survey participation did not differ significantly from respondents with respect to all baseline parameters. When compared with PCPs, significantly more oncologists were affiliated with teaching institutions (51 vs. 42 %, p < 0.001) and received at least some training in cancer and treatment-related effects (93 vs. 65 %, p < 0.001). However, PCPs were more likely to have been trained in the USA than oncologists (79 vs. 65 %, p < 0.001). Physicians in our study mainly cared for patients with health insurance coverage. Baseline demographic and practice characteristics of the physicians are summarized in Table 1.

Table 1 Baseline clinical and practice characteristics of survey respondents (N = 2,026)
Full size table

Physician preferences for cancer survivorship care

Figure 1 illustrates physician preferences for different models of cancer survivorship care. A considerable proportion of PCPs (51 %) supported a PCP or shared model; whereas, the majority of cancer specialists (59 %) endorsed an oncologist-directed approach (p < 0.001). A number of physicians (23 %) favored a model that involved specialized survivor clinics managed by trained providers. Figure 2 describes PCPs’ and oncologists’ self-confidence in providing follow-up to breast and colorectal cancer survivors. While most oncologists were very confident in their knowledge regarding following patients for cancer recurrences as well as managing cancer and its long-term and late effects (80 and 70 %, respectively), PCPs had lower self-efficacy in these areas (34 and 19 %, respectively; both p < 0.001, comparing PCPs vs. oncologists).

Fig. 1
figure 1

Physician preferences for different models of cancer survivorship care. N (center of bar) is based on the number of survey respondents and percents are weighted to the US population of physicians. †p values are calculated using Chi-square tests based on weighted counts. Double dagger, includes clinics operated by nonphysician providers, such as physician assistants

Full size image
Fig. 2
figure 2

Physicians’ self-confidence with their knowledge for providing follow-up care to breast and colorectal cancer survivors. N (center of bar) is based on the number of survey respondents and percents are weighted to the US population of physicians. †p values are calculated using Chi-square tests based on weighted counts. Double dagger, all other responses include any combination of not at all, somewhat, or very confident

Full size image

When PCPs were asked about their actual ability to deliver survivorship care (Fig. 3), most reported that they were better able to provide routine cancer follow-up (57 %), detect and work-up recurrent cancers (74 %), and offer psychosocial support to survivors (50 %). However, only 32 % felt they were able to assume primary responsibility for survivorship care. Of note, a significant proportion of PCPs (43 %) reported already being involved with screening of their breast or colorectal cancer patients for recurrence. Conversely, the majority of cancer specialists did not feel that PCPs were adequately supported or trained to offer routine cancer follow-up (78 %), detect cancer recurrences (62 %), provide psychosocial support (92 %), or take on the principal role for survivorship care (87 %; all p < 0.001).

Fig. 3
figure 3

Views regarding PCPs’ ability to provide cancer follow-up. N (center of bar) is based on the number of survey respondents and percents are weighted to the US population of physicians. †p values are calculated using Chi-square tests based on weighted counts. Double dagger, all other responses include any combination of strongly or somewhat disagree, neither agree nor disagree, somewhat, or strongly agree for breast or colon

Full size image

Predictors of cancer survivorship care preferences

Tables 2 and 3 show the results from the multivariate analyses of PCPs and oncologists, respectively, that explored the relationship among physician and practice factors as well as attitudes and self-efficacy with cancer survivorship care preferences. As these tables highlight, both PCPs (Table 2) and oncologists (Table 3) who were 40 years or older were significantly more likely to prefer an oncologist-based model instead of a PCP/shared care or specialized survivor clinic model than those under 40 years, after controlling for covariates (both overall p < 0.05). In terms of attitudes and self-efficacy, oncologists who indicated that PCPs should be primarily responsible for survivorship care or were better able to offer psychosocial support were more likely to endorse either a PCP/shared care or specialized survivor clinic model (both ORs approximately 2.7; overall, p < 0.001). Among PCPs, those who agreed that PCPs should be responsible for survivorship care were significantly more likely to prefer the PCP/shared care model versus the oncologist model than those who did not agree or were neutral. In addition, female PCPs who agreed were also more likely to prefer the specialized survivor clinic model over the oncologist model than those who did not agree or were neutral; however, this association was not statistically significant among male PCPs.

Table 2 Multinomial analysis of PCPs’ preferences for models of cancer survivorship care (N = 852)
Full size table
Table 3 Multinomial analysis of oncologists’ preferences for models of cancer survivorship care (N = 970)
Full size table

Importantly, PCPs who reported prior or ongoing involvement with cancer surveillance of their patients had significantly higher odds of supporting a PCP/shared care system compared with the oncologist model (OR, 2.08; 95 % CI, 1.34–3.23). No additional physician or practice level characteristics were associated with survivorship care model preferences among PCPs and oncologists.

Discussion

In this analysis, we found that physician attitudes and self-efficacy with follow-up affected preferences for survivorship care among PCPs and oncologists with respect to certain aspects of cancer follow-up: PCPs preferred a PCP/shared-care model whereas cancer specialists preferred an oncologist-based model. This discordance in preferences between PCPs and oncologists may lead to poor coordination of care among providers, supporting prior research showing that the current quality of follow-up care is suboptimal [16], and that many survivorship outcomes are frequently overlooked by the healthcare system [17–20]. Differences in patient and physician expectations, lack of clarity surrounding provider roles, and debates about the best strategy for follow-up can pose significant challenges to delivering quality survivorship care [21, 22]. As a result, there has been substantial interest in developing survivorship care plans and establishing more effective and efficient models of care which can help to streamline access to appropriate follow-up care and increase guideline-concordant surveillance practices [23–25]. The perspectives of PCPs and oncologists about who can and should provide this care is fundamental in creating survivorship care models that are effective in optimizing outcomes, feasible to implement on a broad scale, and acceptable to both patients and physicians.

While PCPs indicated that they were better able to perform routine follow-up, self-confidence with their own knowledge levels about cancer recurrence and long-term and late treatment effects was not as strong. Likewise, their willingness to assume primary responsibility for such follow-up care was low. This contrasts with previous studies of non-US physicians that revealed PCPs were highly agreeable to assuming primary responsibility for cancer follow-up, providing that supports such as access to cancer information and specialist services were readily available, if necessary [26, 27]. Because SPARCCS did not specifically ask respondents about their level of satisfaction with survivorship resources, it is possible that US physicians would be equally willing to play a main role in cancer follow-up if adequate information and supports were available to them. Nonetheless, there are significant differences between US and non-US healthcare systems in terms of infrastructure, physician training, allied health support, and remuneration that may potentially explain this disparity between US and non-US physicians. Furthermore, the apparent discrepancy between PCPs’ preference for a PCP/shared care model and their willingness to actually accept full responsibility for survivorship care suggests that there may be complex health system barriers (e.g., reimbursement disincentives and fear of litigation) discouraging PCPs from fully taking on this role. Future studies should focus on identifying and resolving these barriers so that more PCPs can become involved in cancer follow-up.

The observation that oncologists largely favored a specialist-directed approach to care despite evidence showing comparable survival outcomes and patient satisfaction between PCP and specialist follow-up presents an opportunity for increased awareness among oncologists [13]. Cancer follow-up performed solely by oncologists is unlikely to be sustainable or cost-effective given the anticipated shortage of specialists in the face of increasing numbers of survivors [28]. Studies also show that survivors are more likely to receive appropriate preventive care if a PCP is involved [17–20]. Furthermore, cancer survivors may be less inclined to seek help from their oncologists over time because many of the issues faced by survivors are not cancer related (e.g., general preventive health and chronic disease management). Thus, PCPs and nonphysician providers are likely better positioned to address these problems, referring back to specialists when necessary [29]. Moreover, burdening cancer clinics with routine follow-up may limit oncologists’ ability to respond to patients who actually require specialized medical attention [29]. Importantly, our results support prior research showing that PCPs who reported active involvement in cancer surveillance were significantly more receptive towards a PCP/shared care-based approach to managing cancer survivors [30]. Presumably, such experience shows them that cancer surveillance is generally not difficult. This emphasizes the potential benefits of oncology related educational programs for PCPs that may promote their early engagement in the care and follow-up of cancer patients.

Strengths of the current study are the large, population-based sample that is representative of US physicians, the inclusion of oncologists’ views, the assessment of attitudes and self-efficacy as well as the dual survey design that allowed for comparisons between PCPs and specialists. However, there are some limitations to our findings. First, we surveyed only US physicians and limited our analyses to oncologists who treat either breast or colorectal cancers. Thus, our results may not be generalizable to other health care systems or other cancer types. While we were able to demonstrate that PCPs who already participated in cancer surveillance of their patients tended to support a PCP/shared care model for follow-up, it is unclear which aspects of the PCPs’ involvement were responsible for this effect. Moreover, we were unable to account for respondents’ familiarity with or prior exposure to specific models of care, which could have significantly affected their preferences. Third, some of our survey questions of physician attitudes and preferences explored similar or overlapping constructs which may have limited the scope of our analysis. Fourth, the survey mainly focused on overall follow-up care and specific survivorship domains that primarily included surveillance and psychosocial support. It also only asked participants to select their most preferred model of care rather than to rank the various choices in order of preference, which would have yielded further insights. To this end, future studies are needed to explore the views and self-efficacy of physicians regarding other pertinent survivorship issues, such as employment, insurance, and sexual concerns [31–33]. Qualitative studies are also recommended in order to provide a richer understanding of physician attitudes and serve as a basis for developing interventions that may modify physician preferences [34]. As our study demonstrates, even oncology providers’ knowledge of late effects is suboptimal, underscoring the importance to examine the feasibility of alternative models of care so that physicians, particularly those who are used to more conventional models, will become increasingly receptive towards newer models of care.

In summary, PCPs and oncologists have different preferences for models of cancer survivorship care, underscoring the apparent disconnect among different providers with respect to their roles during posttreatment follow-up. Importantly, PCPs who are already involved with cancer management appear to be most willing to assume some responsibility for survivorship care. This is relevant given the expected increase in the survivorship burden and the growing interest in transferring at least part of routine follow-up from specialized cancer clinics to nonspecialist care. In light of these findings, promoting early engagement of PCPs in cancer follow-up may improve PCP-oncologist communication, help to clarify provider roles during survivorship, and enhance acceptance of additional PCP involvement with cancer survivorship care. Continued efforts to examine alternative models of survivorship care and their acceptance by physicians and patients are also imperative as impending physician workforce issues will necessitate a change in the way that survivorship care is delivered.