Background
Variation in medical practice is a common phenomenon [
1,
2] and although understudied, is likely to be large among patients with advanced serious illnesses (ASI), such as those with advanced cancers or advanced organ failures, for which treatment options are largely restricted to life extension or palliation of symptoms. Such variation can be appropriate and beneficial if it occurs due to differences in illness characteristics or patient preferences. However if the variation in treatment recommendations for these patients, such as whether or not to recommend a life extending treatment, is independent of patient characteristics or preferences, and is due to physician characteristics or their practice style, then this raises questions about whether patients are receiving the best possible care at the end of life (EOL) and whether healthcare resources are being used optimally.
Research documents variation in physician practice styles or recommendations for a range of services and conditions [
3‐
11]. Such variation may relate to differences in cost, [
12] patient age, [
13] presence of cognitive impairment, [
14] health care systems, [
3,
4] cultural or health care beliefs, [
5] and treatment guidelines [
6]. The relationship with patient’s cognitive status may be due to caregiver or physician attitudes regarding perceived benefit of treatment to patients with dementia. Equivalent research for ASI patients in Singapore (the setting for this study), though presently lacking, is likely to indicate that variation in physician practice styles will be large as patients (and caregivers) often defer to their physicians for final decisions regarding treatment [
7] and much of the care is paid out-of-pocket [
8]. The latter point is especially salient as the lack of a third party payer affords physicians greater flexibility in treatment recommendations, with fewer concerns about coverage, oversight, and reimbursement guidelines.
Variation in treatment of patients with ASI may also result from variation in knowledge/attitude and perceived roles of physicians in managing such patients. In Singapore, most EOL care is provided by non-palliative care physicians who have little training in palliative care [
9,
10]. Poor knowledge due to absence of training may result in poor management of patients. Furthermore, control of pain and other symptoms has traditionally received lower priority in care of patients with cancer and other ASI as compared to extending patient’s life [
11]. This may be related to (non-palliative care) physicians’ perception of their role in primarily providing life-extending treatments as opposed to managing their symptoms. If so, these physicians may be less likely to provide palliative care to patients.
It is challenging to identify the causes of variation in treatment patterns for patients with ASI in the real world given differences in health status, patient and caregiver preferences, socio-demographics and other factors. Therefore, we attempted to address this issue via a carefully designed survey of physicians who treat patients with ASI in Singapore. The survey aimed to ascertain the variation in physician recommendations for life-extending treatments for ASI patients, their knowledge/attitude regarding palliative care provision and their perceived roles regarding provision of EOL care. We hypothesized that there will be a large variation in physician recommendations, which will systematically vary by their characteristics (age, gender, experience, country of medical training and training in palliative care), that they will lack appropriate knowledge/attitude about palliative care provision and that extending patient’s life for as long as possible will be the most important perceived role of physicians. If confirmed, then this might raise concerns that patients are receiving sub-optimal care at the EOL.
Discussion
Results show considerable variation in physician recommendations for EOL care, some of which can be predicted by physician characteristics. If these results extend to real life, it raises significant concerns as to whether patients are receiving the most appropriate treatment at the EOL. Unfortunately, there are no real world studies that we are aware of to gauge the veracity of the results. Results further show that lack of appropriate knowledge/attitude in some aspects of palliative care, reinforcing the need to focus on palliative care education among non-palliative care physicians.
More experienced physicians were more likely to recommend life-extending treatments to patients without cognitive impairment, possibly due to their greater expertise in use of these treatments. However, with more experience, there is also a realistic appreciation of patient’s prognosis and realization of the apparent futility of such treatments, especially among those with cognitive impairment. Thus experienced physicians become less likely to recommend life-extending treatments to patients with cognitive impairments. Alternatively, this relationship could denote a cohort effect, due to younger physicians being trained in a different medical culture as compared to older physicians. However, physicians trained outside Asia were more likely to recommend these treatments to patients with cognitive impairment. The reason for this is not known. Nevertheless, study results suggest that if all decision making were left to physicians, there would likely be large variation in treatment recommendations even for similarly situated patients. By eliciting and respecting patient preferences, it may be possible to minimize variations in EOL treatments as a result of differences in physician practice styles or their personal characteristics. Formulation of standards or guidelines for care and their promotion by the government or third party payers, perhaps tied to pricing and reimbursement, may also help to reduce practice variation, as would greater levels of utilization review.
As with our findings, previous studies have also found that patients with dementia are less likely to be treated for cancer [
20‐
23]. With an increase in prevalence of dementia, physicians can expect to face these decisions more frequently. It is unclear whether life-extending treatments offer any survival benefit to patients with moderate to severe cognitive impairment, [
23] and thus it is difficult to define what course of treatment will achieve the best possible medical outcomes for these patients. This is further complicated by the challenge of eliciting patient preferences, as most patients with serious cognitive impairment lack capacity for decision making and even minor diagnostic and treatment procedures may cause significant distress to the patient. Future studies on this topic and development of guidelines for treatment of ASI among patients with cognitive impairment may help physicians provide clearer guidance to caregivers and work to reduce variation in physician recommendations.
Despite some evidence that age per se does not influence the efficacy of many life-extending treatments, [
24‐
26] we find that physicians are more reluctant to recommend these treatments to older patients as compared to younger patients, given the same prognosis and treatment costs. This is consistent with previous literature reporting an age bias in physician recommendation of cancer treatments and life-sustaining treatments [
27,
28]. It may be that older patients are at a greater risk of adverse drug reactions, [
29] that physicians believe that older patients may not want as much life-extending treatment because of fatalist attitudes, or because physicians themselves attach lower value to extending life in an older patient compared to a younger patient. Therefore, efforts should be made to encourage physicians to discuss patient’s treatment preferences without being biased by their age.
Treatment costs also influence physician treatment recommendations; however, physicians did not seem to view discussing treatment costs with EOL patients as an important role, possibly because they feel that they do not have the skills, knowledge, or time to do so or because they do not see it as their role.
We further show that despite the presence of well-established guidelines regarding treatment of severe pain, only 50 % of the physicians surveyed responded correctly to a statement regarding appropriate opioid prescription. Furthermore, Singapore trained physicians scored lower compared to physicians trained in many other developed countries. Insufficient exposure to palliative care in the undergraduate medical curriculum in Singapore may have contributed to this. Medical students in Singapore receive only 4 days of training in palliative care in their entire undergraduate medical curriculum, as compared to an average of 14 days in United States [
10,
30,
31]. Basic palliative care education and training for medical students, residents, and practicing physicians is needed to fill this lacuna, and one needs to look into the best models for incorporating palliative care in training curricula at various levels.
The study has several limitations. First, it is hypothetical and results may not generalize to real patients. There is also a lack of clinical detail upon which to make decisions; more detail may have generated greater consistency. Moreover, despite promising confidentiality, it is possible that physician responses may have been influenced if they felt that they were being evaluated. It is also possible that framing of some of the statements in the best-worst tasks could have influenced physician responses. For instance, the statement ‘extending patient’s life for as long as possible’ is worded quite strongly and as a result could have been consistently underrated by physicians. Further, despite every effort to expand this survey to all major departments in major public hospitals, some of the departments did not allow or did not have time for the study team to present in their department meetings. However, we have no reason to believe these departments would respond differently. Lastly, due to the absence of a sampling frame of physicians, we invited physicians to participate in the survey by advertising in department meetings and through department heads and colleagues. Due to different modalities of invitation and advertisement, it was not possible to calculate how many physicians received the invitation to do the survey and hence the response rate.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
CM and EAF conceptualized the study. NC, JZ and HBD participated in acquisition of data and designing the survey questionnaire. All authors contributed to interpretation of data, drafting of manuscript and revising it critically. All authors read and approved the final manuscript.