Cancer is a significant health problem in older persons.1 It is estimated that 42 % of all incident cases and over 60 % of mortality due to cancer occur in persons aged 70 and over.1,2 With the aging of the population there will be a considerable increase in the number of older adults diagnosed with cancer.1,2 Treatment decisions are based on preferences, estimation of the risks and benefits, and costs. An individual makes a trade-off between the benefits and harms. However, there is less known about risks and benefits for older adults as they are underrepresented in clinical trials, particularly the more frail older adults and those with comorbidities,3–7 which complicates treatment decision-making.
Underuse/nonreceipt of cancer treatment is commonly reported, and is more common in older adults.8–10 Recent studies have shown underuse in 46–49 % of older patients.11–13 Undertreatment can lead to negative outcomes, such as increased cancer recurrence rate and poorer survival. Undertreatment has been most extensively studied in older women with breast cancer.8–10,14,15 Yood et al.15 reported a hazard ratio of 6.25 for breast cancer mortality in older women treated for less than 1 year with hormonal therapy compared with those treated for 5 years while Verkooijen et al.8 showed that older women who declined breast cancer surgery had a hazard ratio of breast cancer mortality of 2.1. Considering the impact of undertreatment on outcomes, it is important to understand for what reasons they would accept or decline cancer treatment. Several narrative reviews of treatment decision-making in older adults had been published16–18 but, until now, no systematic review has been performed. One way to study treatment decision-making is by studying preferences for a certain treatment based on studying the benefits and harms of one treatment compared with other alternatives, or about an individual’s preference for a certain health state compared with a perfect health state.19
Therefore, we conducted a systematic review with the primary objective of synthesizing all factors influencing older adults’ decisions to acceptor decline cancer treatment proposed by their physicians. In particular, we were interested to determine if the factors influencing older adults’ decisions to accept or decline cancer treatment varied by cancer stage, cancer type, cancer treatment, and age (younger-old [65–74] versus older-old [75+]). During the conduct of our systematic review on factors influencing the treatment decisions we noticed important methodologic differences between studies studying actual treatment decisions compared with studies using hypothetical treatment decisions. Additionally, examining hypothetical treatment decision-making removes the acute stresses of making decisions while facing a diagnosis of cancer. Therefore, we decided to report the results on actual situations and hypothetical scenarios separately. In this article, we will report on the results of studies examininghypothetical treatment decisions.
Materials and Methods
Search Strategy and Selection Criteria
This review was based on a systematic, comprehensive search of 10 databases from inception to February 2013 and was conducted by an experienced health sciences librarian (ES). A study (any type of design except case studies and editorials and reviews) was eligible for inclusion if it reported on reasons why older adults with cancer (i.e. mean age study population 65 years or over or if the study mean/median age was 65) accepted or declined cancer treatment and was published in English, Dutch, French, or German.
The final studies included in this review were selected in two steps based on screening of the abstract and full-text review performed independently by two reviewers (MP and BT) (see Figure 1). For all articles for which no mean/median age was reported, we contacted the study authors to obtain details on age. If no response was received after at least three attempts, the article was not included.
The same reviewers who performed the article-selection processconducted data abstraction independently (MP and BT). The abstracted information included study design, aim of study, location of study, sampling method, source of data, recruitment type and timeline, characteristics of study participants, details on cancer diagnosis and treatment, details on how reasons for accepting/declining cancer treatment were collected, and details of statistical analysis, source of funding, and whether or not the authors had declared any conflict of interest. No meta-analysis was conducted as the studies were too heterogeneous with regard to study population and data collected.
Both quantitative and qualitative studies were included in the review. To determine the quality of the studies included in this review the same two reviewers scored the studies using the Mixed Methods Appraisal Tool (MMAT).20 The MMAT scoring system contains five types of mixed methods study components or primary studies, i.e. 1) qualitative; 2) quantitative randomized controlled trials; 3) quantitative nonrandomized; 4) quantitative descriptive; and 5) mixed methodswith each with its own set of methodologic quality criteria based on existing published criteria. For each item the answer categories were ‘yes’, ‘no,’ or ‘can’t tell’ followed by comments. We did not exclude any study based on the quality assessment as we wanted to provide a comprehensive overview of all factors important to older adults reported in the literature.
We reviewed 17,343 titles and abstracts for eligibility in the first step (see Figure 1) in which we selected 545 manuscripts for full text review. Of these 545, 55 manuscripts reporting on 50 unique studies were selected; 40 publications reporting on 38 unique studies examined factors influencing the older adult’s decision to accept or decline treatment examined the actual cancer treatment decision taken by the study participant and reported elsewhere (manuscript submitted). There were 15 manuscripts reporting on 12 unique studies that examined hypothetical treatment decisions and these are included in this review. All manuscripts were written in English.
The quality of the studies can be seen in Table 1. As many studies had been published some time ago when reporting standards were less clear, for most studies there were one or more aspects of the methodology used that were not described in sufficient detail and we tried to contact all study authors for more detail. Response rates were not reported for three out of 12 studies21–23 or were below 60 % in two out of 12 studies,24–27 and the sampling strategy was not reported in two out of 12 studies.21,22 The hypothetical scenarios studied were generally well described. Most scenarios were reflective of actual treatment options (i.e. were realistic) for cancer patients at the time of the study according to the study authors.21,24,25,27–29
Characteristics of the Included Studies See Table 2 for a description of included studies. Two studies were conducted in: Australia,21,30 the US,22,23 and Canada.24–27 One study was conducted in: the Netherlands,28,29 Spain,31 Japan,32 Sweden,33 the UK,34 and one in both the US and France.35 All studies used a cross-sectional quantitative study design. Studies examined hypothetical treatment decisions for colorectal cancer,21,28–30 lung cancer,22,24–27,31,32 prostate cancer,33,34 or for chemotherapy in a mixed-study population.23,35 Sample sizes ranged between 2127 to 511,33 mean age ranged from 6521,24,25,27 to 77 years.31,35
Factors Influencing the Treatment Decisions
Please see Table 3 for factors influencing the treatment decisions included in each study. Below we have summarized the findings for each cancer type treatment decision separately.
Colorectal Cancer Treatment Scenarios
In the three studies examining treatment decisions for colorectal cancer, all participants had been treated for colorectal cancer. Blinman et al.21 included 123 patients who had completed adjuvant chemotherapy for stage 2 to 3 colon cancer and asked the participants using four scenarios about their willingness to undergo treatment in exchange for increasing survival time or survival rate. They showed that 60 % of the sample would choose chemotherapy for 1 additional month of life expectancy. Half of the sample would choose chemotherapy with small benefits (1 day), while a small proportion (<5 %) would never choose chemotherapy. Participants aged 75 and older and those with higher educational levels would only accept chemotherapy for greater benefits. Bossema et al.28,29 used the treatment trade-off and time trade-off method to examine the willingness to be treated with one of two surgical procedures, varying the risk for developing incontinence and permanent stoma as a result of the surgical procedure. All 122 participants had received surgery for stage 1–3 rectal cancer. The previously received surgery impacted the current choice: participants experiencing current incontinence were more willing to give up life years to avoid a permanent stoma (19 %) but not to avoid daily incontinence. Participants without a current stoma had a much stronger preference for the surgery that avoided a permanent stoma even if that meant incontinence monthly or daily. Similar findings were reported by Harrison et al.,30 who reported that 65 % of 103 participants, recruited postoperatively, were willing to give up 34 % of their remaining life expectancy to avoid a stoma. Additionally, site of cancer (colon or rectal) had an effect on the choice, as did level of education, having previously received pre-operative treatment, and knowing someone who had undergone the treatment.
Lung Cancer Treatment Scenarios
Brundage et al.24,25 studied patients who had completed chemotherapy or chemo radiation or radiation treatment for lung (n=22) or prostate (n=34) cancer and solicited preferences for high- versus low- dose radiation and preference for high-dose radiation versus chemo radiation for locally advanced non-small cell lung cancer (NSCLC). They reported that there was a wide range in survival advantage thresholds required before accepting more aggressive treatments (range 0–80 %) and a small number would always decline the aggressive treatments (4–16 %). There was no difference between the lung and prostate cancer patients and no factor (age, sex, education, or preferred role in decision-making) was associated with willingness to accept treatment. Davidson et al.,27 reporting on the same study using the lung patients (n=21), found that essential information needed to make the treatment decision included the regimen, the side effects, the survival information, and the effect of treatment on survival probability gains. Brundage et al.26 enrolled 60 patients treated for other cancers than lung and used the treatment trade-off method to compare when patients would choose best supportive care (BSC) (palliative radiation) versus BSC plus chemotherapy for advanced NSCLC. They showed that just over half would choose BSC plus chemotherapy for a 1-year survival advantage. Those who were older (correlation coefficient 0.30) or with no postsecondary education (correlation coefficient 0.31) had higher survival advantage thresholds before they would accept BSC plus chemotherapy. Girones et al.31 studied 83 patients with lung cancer (all stages) who were given the choice between four treatment options: two chemotherapy regimens with survival as the main goal (a mild regimen and an intensive regimen), one chemotherapy option with no survival benefit but aiming for symptom relief, and BSC. In contrast to all other studies described above, none of the patients had yet received cancer treatment. Girones et al.31 reported that over half of 83 patients choose treatment with survival as the treatment goal, only one-third chose BSC and only 12 % chose symptom relief. Treatment choice was associated with age, performance status, depression and dementia, and frailty. Hirose et al.32 showed similar findings in 73 lung cancer patients who had been previously treated. They reported that the willingness of patients to accept toxicity for survival advantage varied: 19 % would choose intensive and 21 % would choose less-intensive chemotherapy if it would prolong life by 3 months; 73 % would accept intensive chemotherapy for 70 % chance of symptom relief. Although several factors were studied, only age was associated with the choice, with those aged 71–80 less likely to accept chemotherapy with smaller benefits compared with those <70 years. Similar findings, i.e. that willingness to accept more aggressive chemotherapy required larger median survival thresholds and that older patients had higher survival thresholds than younger patients, were reported by Silvestri et al.22 in a study of 81 advanced lung cancer patients who were previously treated with chemotherapy.
Prostate Cancer Treatment Scenarios
Hopfgarten et al.33 examined willingness to trade life expectancy (range 6 months to 5 years) to avoid long-term therapy-related side effects. They enrolled 511 prostate cancer patients of whom most had been previously treated with either surgery, radiation, and/or hormonal therapy. Most men were in the two extreme categories: they would either accept side effects for a small increase in survival probability (percentage of men ranging between 1 % and 64 %) or they would not accept the side effects no matter what the survival benefit of treatment was (percentage of men ranging from 9 % to 41 %). Age 70–80 years, being single, having a lower educational level, and having a smaller social network were associated with a lower likelihood of accepting the treatment if it induced side effects. Sculpher et al.34 reported similar findings for 121 prostate cancer patients, some of whom had been treated with hormonal therapy. They reported that the willingness to undergo treatment was influenced by potential side effects and out-of-pocket costs (the greater the out-of-pocket costs, the less likely that patients preferred that treatment).
Mixed Cancer Population Studies with Chemotherapy Decisions as the Focus
Extermann et al.35 included 195 mixed cancer patients from France and the US and compared willingness to be treated with chemotherapy using an intensive treatment scenario and a milder treatment scenario. Forty-five percent of French patients had received prior chemotherapy versus 86 % of American patients. Most patients in both countries would accept the mild regimen and 70–77 % would accept the intensive regimen. Although several potential influential factors were studied, only self-rated health had an impact on the mild chemotherapy decision (those in better selfrated health were less likely to reject the mild chemotherapy regimen). Yellen et al.23 reported that 42 patients >65 years were less likely to trade quality of life for survival than those <65, and that the choice of treatment was related to patient experience with chemotherapy (e.g. those who had experienced problems with the previous treatment were less aggressive). A quarter of patients had been treated prior to the study.
Summary Factors Influencing the Decision to Accept the Treatment
Table 4 summarizes all factors influencing the decision to accept treatment. With regard to our research question, if older adults’ decisions varied by cancer stage, cancer type, cancer treatment, and age, there are mixed results. There are conflicting findings for the influence of age, education, marital status, and previous treatment experience; in some studies, these factors led to increased likelihood of treatment acceptance whereas in other studies it decreased the likelihood of accepting treatment. Only one study by Brundage et al.24,25 directly compared if the treatment choice varied by cancer type and showed no difference. With regard to differences in preference due to cancer stage, this was not studied. With regard to comparing preferences for the different cancer treatment modalities, most studies examined the preference of older adults for one of two/ three alternative treatments and the type of treatment and potential side effects/toxicity did influence the choice. The findings across all included studies show similar factors to be important, but, as mentioned above, with conflicting findings about the direction of the association. Table 5 shows the results of the willingness to trade survival against different treatment toxicities showing varying willingness to trade survival are described.
To our knowledge, this is the first systematic review focusing on reasons why older adults with cancer accept or decline cancer treatments focusing on studies using hypothetical treatment scenarios. The results showed that the willingness to be treated varied by the benefits of treatment, the particular side effects and previous treatments received/previous treatment experiences. The results also showed conflicting findings in terms of the impact of social status, education, and family situation. None of the reported studies examined the impact of comorbid conditions on treatment preferences but this is be hypothesized to impact older adults’ preferences; persons who have been diagnosed with other life-threatening conditions prior to cancer may be less shocked and more experienced in making treatment decisions. In addition, comorbid conditions do impact cancer specialists’ treatment recommendations.36–41 Only Girones et al.31,42,43 reported the level of frailty in the older participants that is common in older adults. Frailty can impact both treatment tolerability and efficacy44–48 as well as the treatment delivery, as not all older frail patients will be able to attend a clinic or hospital independently to receive treatment, and thus the general level of frailty of older study participants should be described. As the older population is the most heterogeneous in terms of health and functional status, it is important to have a good assessment of the health and well-being (i.e. geriatric assessment) to inform treatment choices as well as to be able to communicate the potential risks and benefits of treatment, which is crucial for patients’ willingness to be treated or not.49 In addition, this information will also help the oncologist with developing treatment recommendations. A recent qualitative study of medical oncologists showed considerable variation in treatment recommendations in patients with advanced cancer, and factors such as the physician’s perception of patient age and life circumstances influenced the treatment recommendation.50 A geriatric assessment will make the decision based on more objective clinical data and less based on clinical judgments. And with a better estimation of risks and benefits, patients can make more informed choices. A recent systematic review of patients’ awareness of disease status did show that this awareness impacted treatment preferences and quality of life.51 With the aging of the population, many older patients have more than one chronic condition for which they face treatment decisions. These treatment recommendations may even be conflicting.52–55 Therefore, the discussion about treatment goals is crucial for older adults to make informed treatment decisions. However, a recent report of the Institute of Medicine Delivering High-Quality Cancer Care56 reported that patients are not always well informed and recommended that patients to be more informed and engaged with the treatment decision-making process as well as more research on older cancer patients and patients with multimorbidities to improve patient centeredness and engagement.
This review showed important gaps in current knowledge. Although we focused on studies with older adults, there were only four studies with a mean age in the seventies (two studies low seventies33,34 and two studies mean of 77 years31,35), and thus little is known about the treatment preferences of the oldest old patients. Similarly, while most studies have focused on lung, colorectal, and prostate cancer, little is known about older adults facing other cancer diagnoses and/or treatments. Furthermore, several of the studies were conducted more than a decade ago. Since then new treatments have been introduced in clinical practice with many of them being oral agents with different risks, benefits, and mode of administration, so it is important to study willingness to be treated with newer, more ‘elder-friendly’ regimens. Similarly, it is important to study the impact of comorbidities, health literacy, and family support on willingness to be treated and also learn more about the oldest old patients. In addition, it is important to understand how we can support older adults facing cancer treatment decisions who have multimorbidity, as this has not been studied. Lastly, there is little known about the role of the family caregiver in the treatment decision-making process for older adults, while many older adults with cancer are dependent on caregivers for support (e.g. cognitive, emotional, practical, such as transportation, and management of side effects) and thus their role on treatment decision-making should be included in future studies.
This systematic review has several strengths. First, a systematic approach to appraise the literature was used with two independent reviewers reviewing the abstracts and manuscripts and abstracting data. Second, our search was very inclusive as we conducted the search in 10 electronic databases and articles published in four languages. Third, we were comprehensive, no study was excluded based on the quality assessment score and studies using both qualitative and quantitative studies were eligible; however, only studies using a quantitative design were found. However, many studies had a response rate below 60 % or it was not reported, thus there may be selection bias that influences the results of these studies. Furthermore, this review also has limitations. As in any review, the findings are limited by the methodologic quality of the included studies and as many studies were reported a while ago when reporting standards were less clear, there was missing information on response rates and sampling strategies used. Even though we tried to contact all study authors using multiple attempts and using the internet to find current contact information, we were not able to contact all authors and obtain all missing methodological details of the included studies. Last, the studies were too heterogeneous to conduct a meta-analysis.
In conclusion, treatment preferences are influenced by age, social support, potential benefits and risks, and previous treatment experiences. More research is needed to study treatment preferences taking into account multimorbidity, including the oldest old populations, and examining willingness to be treated with newer treatments such as oral agents and/ or targeted agents.