An international review of the patterns and determinants of health service utilisation by adult cancer survivors

The US health-care system permits individuals to choose to an extent which health care provider provides their care in relation to their insurance plan restrictions or absence of an insurance plan. Specialists are included in this section on primary care as data indicated that there is a limited set of services used by patients in the US health care system that may be provided by a specialist such as an oncologist or by a primary care physician. There was a consistent association between contact with primary care and predisposing characteristics such as age, ethnicity, gender and time since diagnosis. Contact with services for cancer-related problems and all symptoms and illnesses are included within the results. Studies were inconsistent regarding whether younger or older CSs were more likely to visit an oncologist or a primary care physician. Younger colorectal CSs were more likely to visit both a primary care physician and an oncologist, whereas older colorectal CSs were more likely to visit only a primary care physician [9, 10]. Moreover, younger breast CSs from the US were more likely to visit an oncologist [11], whereas older breast CSs from Denmark were most likely to visit their primary care physician [12]. Differences in the pattern of primary care utilisation by ethnicity were found across cancer sites. Black colorectal CSs were more likely to receive care from physicians other than a primary care physician, whereas, white colorectal CSs were more likely to receive care from a primary care physician [10]. Furthermore, white breast CSs were more likely to utilise health services compared to black CSs [13]. Compared to male colorectal CSs, female colorectal CSs were more likely to receive care from a primary care physician and an oncologist than either one of these physicians only [9]. Only one study (conducted in the Netherlands) addressed time since diagnosis and the use of primary care services. Dutch endometrial CSs diagnosed between 10 and 15 years previously were less likely to visit a primary care physician than CSs diagnosed within 10 years or very long-term survivors at least 15 years post-diagnosis [14].

Area of residence, participation in an intervention, primary care physician-orientated follow-up and marital status (as a proxy for social support) acted as significant enablers for primary care service use. French colorectal CSs from the Saône and Loire regions were more likely to have made regular contact with a primary care physician than CSs from other French regions [15]. One UK study demonstrated that the orientation of follow-up cancer care was important for health service use. CSs who received hospital-oriented follow-up had lower health-care utilisation rates compared to CSs whose follow-up care was oriented by their primary care physician [16]. A small proportion of Canadian breast cancer survivors whose follow-up care was provided by a primary care provider made visits to an oncologist over a 12 month period [17]. A reduction in health service use was observed for breast CSs who participated in a psychotherapy-based intervention to reduce stress [18]. One study from the Netherlands found that a lack of social support (i.e. CSs who were single or divorced) led to less use of health services compared to CSs who were married [14].

A number of studies found that co-morbidities were associated with visits to primary care. Breast CSs with a co-morbidity, or self-reported poor functioning or high depressive mood had greater service utilisation and health-care costs than breast CSs without a co-morbidity, or who reported better functioning or had low scores low depressive mood [13]. Similarly in the Netherlands health service utilisation rates were two times greater for cancer survivors with a co-morbidity compared to CSs without a co-morbidity [14]. It is not clear from the findings whether CSs were more or less likely to utilise primary care services compared to a non-cancer control population. Compared to individuals without cancer, US and UK colorectal CSs, UK prostate CSs and Dutch endometrial survivors [14, 19, 20], were significantly more likely to visit their primary care physician. However, Dutch CSs and Danish breast CSs had similar primary care physician use as non-cancer controls [12, 21, 22]. In terms of the level of contact made with primary care patterns of utilisation changed over time. Contact with oncologists largely decreased over time, whereas visits to a primary care physician largely increased over time. Annual visits to primary care physicians and oncologists were made by 51 % and 27 % of breast CSs respectively [11]. Within the first year of survivorship there was a high health service utilisation rate for breast CSs; an average of 14 visits per individual was made to a medical provider including a primary care physician or an oncologist [13]. Approximately 50 % of CSs visited an oncologist alongside other physicians, whereas 8 % of CSs made visits to only an oncologist [19]. One cross-sectional study reported that 52 % of breast CSs made visits to both an oncologist and a primary care physician, whereas 41 % visited a primary care physician only and 4 % visited an oncologist only [23]. The level of contact made to both primary care physicians and oncologists decreased over a three year period, from 70-42 % and 30-17 %, respectively. One longitudinal study found that in the first year of follow-up, the majority of breast CSs made visits to both physicians; over time the number of visits made to a primary care physician decreased and the number of visits made to an oncologist increased [24]. In contrast, further studies reported an increase in the number of visits to a primary care physician and a decrease of visits to an oncologist and other physicians over time for colorectal and breast CSs respectively [9, 10, 13, 20, 25‐27].

Follow-up cancer surveillance includes any test which is used to screen for recurrence or metastases of the primary cancer. Each study which addressed the predisposing characteristics associated with CSs who utilised follow-up cancer surveillance procedures was conducted in the USA. Predisposing characteristics found to be significantly related to follow-up cancer surveillance were age, ethnicity, gender and health beliefs. Older age was associated with receipt of less follow-up cancer surveillance procedures. Compared to their younger counterparts, older colorectal CSs were consistently less likely to receive surveillance procedures within 3 years of diagnosis [28], 5 years of diagnosis and treatment [29, 30], and they were also less likely to receive follow-up surveillance in adherence to government follow-up guidelines [31]. At the time of the study a number of guidelines for colorectal cancer follow-up care had been developed but none had been widely implemented. Therefore, the authors amalgamated these guidelines to create a minimum number of service and procedure receipt recommendations. These recommendations included: at least two visits to a physician per year; the receipt of at least two Carcino-Embryonic Antigen (CEA) tests within each of the first two years of survivorship; and the receipt of at least one colonoscopy within the first three years of survivorship. Excess of government guidelines included receiving the minimum level of care in addition to the receipt of at least one CT scan and/or at least one PET scan [31]. Older breast CSs were also less likely to receive mammograms compared to their younger counterparts [25, 27, 32].

The majority of the evidence regarding ethnicity and uptake of follow-up surveillance found that white CSs were more likely to receive follow-up screening [25, 27, 28, 30, 32, 33] and to adhere to the guidelines specified above [31] than CSs of other ethnicities. A study based at a ‘safety-net hospital’ which provided care to underserved populations such as ethnic minorities found that black colorectal CSs were more likely to receive follow-up colonoscopies within 3-years of curative resection than CSs of other ethnicities [34]. The extent to which gender predisposes follow-up cancer surveillance is unclear. Within the first three years of diagnosis, females were more likely to utilise colorectal screening compared to male colorectal CSs, [28] but were less likely than males to receive either colonoscopy or sigmoidoscopy within 1 and 3 years of treatment [34]. Colorectal CSs perceived a greater absolute and comparative risk for developing cancer leading to greater receipt of screening compared to non-cancer controls [35].

Factors which enabled receipt of follow-up surveillance included visits to specific health-care providers, frequency of health-care contact, area of residence, physician recommendation of test and previous cancer diagnosis detected via screening procedure. Variations in colorectal cancer surveillance uptake by French colorectal CSs were dependent on type of physician; 21 % of all colorectal surveillance procedures within 3 years of curative surgery were delivered by a primary care physician and 41 % by a gastroenterologist or an oncologist [15]. An increased number of outpatient visits led to a greater likelihood of receiving colonoscopy or sigmoidoscopy within 3 years of treatment for colorectal CSs [34]. A significant geographical variation in receipt of surveillance procedures was also observed for colorectal CSs [29]. An explanation for this variation given by the authors was the influence of local practice on testing. Moreover, a physician recommendation of follow-up procedures increased the likelihood of procedure uptake for breast CSs and prostate CSs [35, 36]. Receiving a previous breast cancer diagnosis detected by a mammogram was significantly and positively associated with subsequent receipt of mammogram in the survivorship period [36].

Co-morbid illnesses, cancer stage and treatment history were significant need characteristics associated with receipt of follow-up surveillance. Presence of a co-morbid illness led to a lower likelihood of receiving follow-up surveillance. Colorectal CSs with a co-morbidity were less likely to receive colonoscopy and sigmoidoscopy than CSs without a co-morbidity in the first year of survivorship and were less likely to receive CEA testing to the standard recommended by US follow-up guidelines which included receiving at least two CEA tests per year [29, 31]. This finding is further supported by the breast cancer survivorship literature. Breast CSs with co-morbidities compared to CSs without co-morbidities were less likely to receive a mammogram [25, 27, 32]. Colorectal CSs with later stage and an undifferentiated cancer were more likely to exceed recommended guidelines which included receipt of at least one CT and/or PET scan, in addition to minimum recommendations [31]. Further factors associated with mammography receipt for breast CSs were a more recent diagnosis, a secondary cancer, a large tumour and no history of adjuvant radiotherapy [11]. Comparisons of surveillance procedure receipt between CSs and non-cancer controls were largely consistent. Breast CSs had significantly greater use of mammogram compared to controls when adjustments were made for age, race and access to health-care [26, 35]. A similar result was found for PSA testing among prostate CSs [35]. However, one study found that having a cancer history did not lead to differential receipt of Faecal Occult Blood Test (FOBT) or colonoscopy compared to the general population [28]. Rates of follow-up surveillance receipt varied across studies and cancer sites, overall rates of uptake were low to moderate. Eleven percent of colorectal CSs received at least one surveillance procedure each year [28]. Over a three year period, 58 % of colorectal CSs in the USA received on average 2.8 colonoscopies, a lower percentage (19 %) received on average 2.0 sigmoidoscopies and there was an observed decrease in the receipt of barium enema and sigmoidoscopy [29, 31]. In contrast to this finding some studies reported an increase over time in the receipt of a number of colorectal cancer surveillance procedures following treatment specifically colonoscopy, CEA testing and metastatic disease testing [28‐31, 37].

For breast CSs the receipt of mammography decreased over time, with the exception of one study. Sixty-two percent of breast CSs received a mammography in both the first and second years of survivorship whereas 23 % of breast CSs received a mammogram in either year of the first two years of survivorship [13, 32]. A further two studies reported a decrease over time in mammogram receipt [27, 38] whereas another study reported an increase over a two year period [36]. A few studies assessed the level of surveillance receipt in comparison to government-recommended guidelines. The majority of CSs did not meet recommended levels of surveillance receipt; in one study 17 % of colorectal CSs met the guidelines, 23 % exceeded the guidelines and 60 % failed to meet the guidelines. Guidelines have been described above [31]. In the first year of survivorship between 11-59 % of breast CSs received surveillance procedures such as a chest x-ray which were not recommended by American Society of Clinical Oncology (ASCO) guidelines [11].

Significant predisposing variables for receipt of preventive care included ethnicity and age. White CSs were consistently more likely to receive preventive care than non-white CSs. Colorectal CSs who were non-white were less likely to receive preventive care, particularly influenza vaccination [9, 19] than white CSs. Furthermore, white breast and uterine CSs were more likely to receive preventive care compared to black CSs [23, 39]. The evidence was consistent regarding age and receipt of preventive care, whereby older CSs were less likely to receive preventive care (with the exception of influenza vaccination) than their younger counterparts. Older colorectal CSs were less likely to receive preventive care including cholesterol testing, cervical examination, bone densitometry and mammography, [9, 10, 19, 40] but were more likely to receive influenza vaccinations than younger cancer survivors [10]. Older breast and uterine CSs were less likely to receive preventive care compared to younger CSs [23, 39]. A UK-based study found that colorectal, breast and prostate CSs over the age of 65 were more likely to receive influenza vaccination than younger CSs [38].

Enabling characteristics associated with the receipt of preventive care included visits made to specific physicians, frequency of health-care contact, area of residence, socio-economic status, overnight hospitalisations and health insurance. Visits to more than one type of health-care provider (i.e. primary care physician and oncologist) were more likely to facilitate receipt of preventive care. This finding was consistent for both colorectal and breast CSs respectively [9, 10, 23, 24, 26]. Visits made by colorectal CSs to either a primary care physician or an oncologist facilitated receipt of preventive care, but not at the same level as visits made to both health-care providers; whereas CSs who did not visit either health care provider received the lowest levels of preventive care receipt [19]. Moreover, general preventive care (i.e. bone densitometry) was more likely to be delivered by a primary care physician than any other type of physician. Oncologists and gynaecologists were more likely to deliver cancer-related preventive care such as mammography to colorectal and uterine CSs respectively [9, 40]. Moreover, an increasing number of visits to a health-care provider was the strongest predictor for receipt of preventive care for a sample of UK CSs [38]. A study of uterine CSs in the US quantified this amount as 5 or more visits to a physician [39]. Living in an urban area was significantly associated with greater receipt of mammography, influenza vaccination and cervical smear among colorectal CSs [9, 40] and general preventive care among breast CSs [23]. One study reported that breast CSs who had lower socio-economic status were less likely to receive preventive care compared to breast CSs with higher socio-economic status [23]. Receipt of preventive health services among uterine CSs who had not been hospitalised was greater compared to CSs who had been hospitalised. Furthermore, colorectal CSs with private health insurance were more likely to receive a mammogram compared to colorectal CSs with government-funded health insurance [39].

Colorectal CSs with a co-morbidity were less likely to receive lipid testing, cervical screening and bone densitometry, but were more likely to receive influenza vaccination, cholesterol testing and mammography compared to CSs without a co-morbidity [9, 10, 19]. Conversely, breast CSs with a co-morbidity were more likely to receive overall preventive care than CSs without a co-morbidity [23]. Evaluation and Management meetings refer to service contact which is not for the intention of procedures or tests. Survivors with a co-morbidity who attended for Evaluation and Management meetings were more likely to receive a mammography than CSs who did not attend for Evaluation and Management meetings [40]. Rates of preventive care receipt were not consistent between CSs and non-cancer controls. Breast and uterine CSs were more likely than controls to receive preventive care such as colorectal cancer screening [23, 39]. An UK-based study found comparative rates of cholesterol testing and blood pressure monitoring between CSs (including prostate and breast cancers) and the general population, but a 19 % increased likelihood of PSA testing for colorectal CSs compared to the general population [38]. Comparisons of preventive care receipt between breast CSs and healthy controls demonstrated that CSs were less likely to receive preventive care, particularly lipid and cholesterol testing [19, 24, 26]. However, they were more likely to receive bone densitometry or general preventive care than non-cancer controls with co-morbidities [24, 38]. Colorectal CSs receipt of mammography and cervical screening decreased over time, receipt of bone densitometry remained low, whereas rates of influenza vaccination fluctuated over time [9].

Support was found for age, employment and student status as predisposing characteristics associated with the use of mental health services. Younger age (<65 year olds vs. >65 year olds) was significantly associated with seeking mental health or supportive care services among breast CSs in two Canadian studies [41, 42] and among survivors of breast, lymphoma, colorectal, melanoma and other cancers in one US study [43]. Breast CSs who reported that they were currently employed or a student were more likely to utilise professional supportive care services [41].

Two US studies reported on the patterns of inpatient hospitalisations among breast CSs. Only one of these studies addressed the predisposing characteristics associated with being hospitalised. Ambulatory-Care-Sensitive Hospitalisations or preventable hospitalisations were associated with older age, being widowed or divorced and lower likelihood was associated with being of an ethnicity other than white and black [44].

Education level, household income, health insurance and residential area were enabling factors associated with the use of mental health services. CSs with a high level of education were more likely to utilise mental health services compared to CSs with a lower level of education; this was supported by a study of endometrial, prostate and lymphoma CSs in the Netherlands [14] and breast CSs in Canada [41, 42]. Household income which is strongly associated with educational level was a significant enabling characteristic for mental health service utilisation; both Canadian studies found that higher household income resulted in greater likelihood of using mental health services [41, 42]. Further results from these two studies found that additional health insurance compared to government-funded insurance was associated with increased mental health service use [41, 42]. One US study addressed the impact of urban or rural residence on receipt of mental health services among breast, colorectal and haematological CSs and found no significant difference in service receipt despite rural CSs being less likely to have psychiatric services within 30 miles of their home [45].

Enabling characteristics associated with inpatient hospitalisations included socio-economic status of residential area and previous visits to a physician. CSs from an impoverished area compared to a more affluent area were more likely to experience an Ambulatory-Care-Sensitive Hospitalisation. However, if CSs had visited a physician recently the risk of being hospitalised was reduced [44].

Experiencing a psychological disorder and expressing an explicit need for mental health services was associated with use of mental health or supportive care services. CSs reported a greater need for and use of mental health services compared to the general population without cancer, due to a higher prevalence of anxiety and sleep disorders [43]. Canadian breast CSs were asked if they were in need of, or could not access mental health services; breast CSs who were younger, had additional health insurance or a high level of education, were working or studying were more likely to report an explicit need for services [42]. Rates of mental health service utilisation were low among CSs ranging from 1 % to18%. Between 1 and 10 % of CSs in the Netherlands utilised psychology services; survivors of lymphoma had the greatest use of psychology services compared to survivors of endometrial and prostate cancers [14]. There were low utilisation rates of both psychiatric and psychology services among Canadian breast CSs (4 % and 5 % respectively) [41]. Eighteen percent of USA CSs made on average 2 or 3 visits to a mental health professional and breast CSs were the highest users of services [43]. Between-study variation in rates may be due to different modes of access to services across health care systems.

Need-related factors associated with inpatient hospitalisations included having at least one co-morbidity, experiencing menopausal symptoms and higher scores on the Centre for Epidemiology Studies-Depression scale (CES-D). Between 13 % and 25 % of breast CSs had at least one overnight hospital stay [44, 46].

The use of other health services such as social services or dietetics ranged from 0-11 % of CSs [22, 42]. Danish breast CSs had greater utilisation of allied health care professionals than the general population, [12] and US CSs had an average number of 3 visits per survivor to a physiotherapist or an occupational therapist [13].