Preference of Japanese cancer patients for being informed about their prognosis

Introduction

Prognostic information of cancer patients is essential as complex decisions regarding treatment and care are affected by prognosis. In general, prognostic information consists of life expectancy, functional prognosis, and uncertainty (1,2). Previous studies reported that about 80% of advanced cancer patients wanted to be informed of their life expectancy, and this preference persists over time (3-5). Enzinger et al. reported that 71.0% of advanced cancer patients wanted to be told their life expectancy, though only 17.6% recalled receiving a prognostic disclosure by their physician (5). In general, patients desire to maintain their independence to continue essential personal activities and not to be a burden to others (6,7). In terms of the quality of death and dying, the Good Death Inventory (GDI) (8), which is a well-validated tool for assessing the quality of death and dying, consists of items including not being a burden to others and maintaining independence (8). Thus, functional prognosis is important for patients and family as well as life expectancy. A recent longitudinal study reported the personal last wishes of patients with a life-limiting illness, such as travel, activities, and being with family and friends, did not show significant intra-personal changes over time (9).

Although the evidence regarding patient behavior for being informed of prognostic information is limited, a previous study reported that many patients hesitated to ask about their prognosis directly and instead expected clinicians to initiate such a conversation (10). The major reasons patients seek clinical information are to better understand their disease and treatment options, and to aid in decision making when they are dissatisfied with the information provided to them by health professionals (11,12). Yun et al. reported that 56.2% of advanced cancer patients were told their prognosis by the doctor, and 28.5% guessed based on their worsening condition (13). In addition, Nguyen et al. reported that 71.0% of breast cancer patients used the Internet to search for breast cancer information, and 51.1% sought information on prognosis (14).

Previous research indicated that patient satisfaction and the quality of physician-patient communication were improved when physicians asked patients whether they wanted to obtain prognostic information and what that information meant to them (15). Therefore, it is worth exploring the kind of prognostic information that cancer patients want to be informed about, and whether they want to be informed about it. This evidence would help determine desirable prognostic information provision, prognosis communication, and future prognostic research directions. Based on several previous studies and discussion among the authors of this study, we proposed the following hypotheses: (I) more cancer patients want to be informed about their functional prognosis than life expectancy; (II) more cancer patients search for information about their prognosis themselves on the Internet than ask a doctor; and (III) many cancer patients want to be informed about their functional prognosis using the Internet under the assumption that a functional prognosis is available. Thus, we mainly aimed to clarify the prognostic preference of cancer patients and whether the cancer patients want to be informed about their prognosis. We also explored the correlation between the prognostic preference of cancer patients and patients’ characteristics. We present the following article in accordance with the SURGE reporting checklist (available at https://apm.amegroups.com/article/view/10.21037/apm-22-772/rc).

Methods

Study design and setting

We conducted a cross-sectional web-based anonymous survey of cancer patients in January 2022.

Participants and procedure

Participants were recruited in January 2022 through a private web-based survey company (MACROMILL; Tokyo, Japan). The inclusion criteria were cancer patients aged 40 to 75 years old who consented to participate. The definition of cancer patients was patients that were being treated for cancer or those who had regular visits to the hospital for follow-up. Based on a response rate of 50%, sample error of 5%, and confidence level of 95%, our sample size calculation showed that the minimum required number of participants was 73. We planned to recruit 132 cancer patients across Japan, and we determined the number of recruits for each patient category based on age and sex (40–64 years old: 34 men, 32 women; 65–75 years old: 33 men, 33 women), and primary cancer site (breast cancer: 14, prostate cancer: 14, thyroid cancer: 8, others: 96). A survey company recruited potential participants across Japan by convenient sampling and sent questionnaires to them online. Responses to the questionnaire were voluntary, and response was considered consent to participate. Confidentiality and anonymity were maintained throughout all investigations and analyses. The participants received an incentive equivalent to 50 cents from the survey company for completing the questionnaire, and no follow-up was required after the survey completion. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The Institutional Review Board of Seirei Mikatahara General Hospital (Approval No. 21-37) approved the protocol of this study.

Questionnaires

In the absence of specific and validated instruments for evaluating patients’ preference for receiving information about their prognosis and whether they want to be informed, we developed a draft questionnaire based on data from previous studies and discussion among the authors of this study (16-19). The face validity of the draft questionnaires was confirmed by five patients in October 2021, and the questionnaires were revised by researchers.

We asked about background characteristics; age, sex, primary site of cancer, period from cancer diagnosis, presence of recurrence or metastasis, experience of chemotherapy, performance status, maritage status, living with family, education, and experience of having a family member who died from cancer.

We asked the patients to respond with whether they wanted to be informed about their life expectancy and functional prognosis using a 6-point scale (1= Strongly agree, 2= Agree, 3= Agree somewhat, 4= Disagree somewhat, 5= Disagree, 6= Strongly disagree). This consisted of life expectancy, moving freely, complex thinking, eating well, and proper communication. We referred to our previous study, which developed prognostication of walking, talking, and eating to determine the items of functional prognosis (2). We also asked the patients to reply to questions about whether and how often they asked or searched about their prognosis with using a 4-point scale (1= Often, 2= Sometimes, 3= Occasionally, 4= Never). Subsequently, we asked whether the patient wanted to be informed about their prognosis directly from a doctor, and whether they wanted to be informed about their prognosis on the web by entering their information, such as the result of a blood test, using a 6-point scale (1= Strongly agree, 2= Agree, 3= Agree somewhat, 4= Disagree somewhat, 5= Disagree, 6= Strongly disagree) (Supplemental Questionnaire - Appendix 1).

Statistical analysis

We conducted descriptive analysis for the background characteristics of the patients, the prevalence of patients’ preference to be informed about their prognosis, the frequency of asking or searching for information about their prognosis, and the prevalence of patients’ preference of means to be informed about their prognosis.

We defined the answer of whether they wanted to be informed about their prognosis into two categories: Strongly agree/Agree, and others. Subsequently, we investigated the correlation between the prognostic preference and the patient’s characteristics using Spearman’s correlation coefficient. We defined the variables based on the previous research and discussion among the researchers (3-5,13). Probability values were two-sided and statistical significance was P<0.05. Adjustment for multiple comparisons, such as Bonferroni correction, was not performed due to the exploratory nature of this study. All analyses were conducted using SPSS-J (ver. 28.0; IBM, Tokyo, Japan).

Results

A total of 132 patients from all the eight regions of Japan responded to the survey. The mean age was 63.1±9.3, and 67 (50.8%) were men. The most frequent primary tumor was gastrointestinal/hepatobiliary and pancreatic (n=43, 32.6%), followed by urogenital (n=23, 17.4%) and gynecological (n=20, 15.2%). In total, 19 (14.4%) had recurrence or metastases, and 13 (9.8%) were currently undergoing chemotherapy (Table 1).

Table 2 shows the prevalence of whether patients wanted to be informed about their life expectancy and functional prognosis. Among the patients, 26.6% answered “Strongly agree” or “Agree” to wanting to be informed about their life expectancy, and 33.6–46.9% wanted to be informed of their functional prognosis. There was no significant difference in the prevalence of preference of being informed about their prognosis between the patients who had recurrence/metastases and those that did not (data not shown).

The experience of having family member who had died from cancer had a significant positive correlation with the preference to be informed about their prognosis of “Life expectancy” (r_s=0.344, P<0.001), “Moving freely” (r_s=0.210, P=0.016), and “Proper communication” (r_s=0.18, P=0.038) (Table 3).

Table 4 shows the means and frequency of the patients that ask or search for information about their prognosis. Most patients did not actively seek information about their prognosis; however, more patients actively wanted to be informed about their life expectancy prognosis than their functional prognosis.

Table 5 shows the answers to questions about whether the patients wanted to be informed about their prognosis directly from a doctor, or whether they wanted to be informed about their prognosis on the Internet by entering their information. About a third of patients answered “Strongly agree” or “Agree” to wanting to be informed about their prognosis directly from a doctor, and about a fifth of patients wanted to be informed about their prognosis on the Internet by entering their information.

Discussion

To the best of our knowledge, this is the first survey to explore the prognostic preferences of cancer patients and whether cancer patients wanted to be informed about their prognosis.

The first important finding of our study was that the cancer patients were more likely to want to be informed about their functional prognosis than their life expectancy. Our result is valuable as our findings suggest that a certain number of cancer patients wanted to know their functional prognosis as well as time left to be physically and mentally independent, which is a novel finding. Ullrich et al. revealed that the major personal last wishes of people with a life-limiting illness were travel, activities, and being with family and friends (9). Thus, the concept of unfinished business may provide a clue to the interpretation of this result. As cancer patients may have unfinished business, such as engaging in meaningful activities, they may prefer to know their functional prognosis as a guidepost for completing their unfinished business (20). Furthermore, cancer patients may think that their life expectancy does not have the biggest effect on them completing their unfinished business. To clarify this hypothesis, a qualitative study is needed to explore the reasons why cancer patients wanted to know their functional prognosis.

The second important finding was that the majority of patients did not strongly prefer to know their prognosis. This result was inconsistent with a previous study in Japan that indicated that 72.8% of cancer patients would like their physician to explain how their life expectancy will influence their future plans (20). One possible explanation of this discrepancy was the difference in the treatment stages of the participants. A previous study by Umezawa et al. targeted cancer patients deemed by their physician to have received bad news about anticancer treatment cessation at least one week earlier (21). However, most of the participants in the present study were before or after completion of chemotherapy and were in relatively good general health.

The third important finding was that fewer patients actively seek their prognosis compared to the number of patients that want to know their prognosis. Previous studies reported that there were cultural differences regarding preferences for prognostic information (22-24); thus, this result could be interpreted in terms of cultural differences. A previous qualitative study that was conducted in western countries reported that the reason patients did not discuss their prognosis were: concern that discussing prognosis is too emotionally difficult, that the information would not be useful, or that they may not believe that doctors can estimate prognosis accurately (22). Previous research that compared the preference of prognostic disclosure among Japanese Americans and Japanese living in the U.S. revealed that there were group differences regarding the detailed process of initiating prognostic disclosure in cases where the patient had developed an incurable cancer. Japanese living in Japan (J/J) were much more reluctant to ask their physician to initiate prognostic discussion and to inform them about their prognosis in detail compared to Japanese Americans living in America (JA/A) and Japanese living in America (J/A) (23). (J/J: 42%, JA/A: 62%, J/A: 63%)

The fourth important finding was that the experience of family members that had died of cancer had a significant positive correlation with the preference for being informed about life expectancy and functional prognosis. Several previous studies explored associated factors, such as sex, age, marital status, education, and performance status, with the patient’s preference for being informed about their life expectancy (25-27); however, there was no consistent evidence. Thus, our result is novel in terms of identifying a correlation between the preference of prognostic information and having the experience of family bereavement from cancer. We did not evaluate GDI in this study; however, one possible explanation was that the experience of family bereavement from cancer may evoke patients to ask questions such as “Would I burden others?” and “Could I maintain my own independence?”, which are components in GDI (8). Another possible explanation was as health care providers discuss prognostic information with the family rather than the patient in Japan (24), the family may have been more accustomed to such conversations. Therefore, the family may be more likely to consider the pros and cons of telling or not informing the patient of their prognosis in discussions with health care providers (28). Thus, the experience of family members who have died of cancer may facilitate the patient to consider whether to be informed about their prognostic information. Furthermore, many families of cancer patients have received explanations of prognosis and how long they can expect to be able to talk to the patient and have had the experience of acting on the assumption of death (24,29). In addition, a previous study reported that a fourth of the families of terminally ill patients felt distressed because of unfinished business (30). Therefore, patients who had the experience of family members who had died of cancer may understand the importance of prognostic information. In terms of associated factors with the preference for being informed about their life expectancy and functional prognosis, our study did not show a significant correlation between cancer type (colorectal, prostate, and breast cancer) and education level. This result was inconsistent with previous research that found that cancer type and education level were correlated (27). A possible explanation was the difference of the duration from the diagnosis of cancer. Nagler et al. enrolled patients who were diagnosed with cancer within 1 year prior to their survey (15); however, the majority of the participants in the present study were analyzed more than 2 years after diagnosis.

The fifth important finding was that about a third of patients wanted to be informed about their prognosis directly from a doctor. This result was consistent with previous studies in Japan and Australia (31,32). Moreover, another national survey in Japan reported that 34% of physicians would discuss prognosis with a hypothetical patient with newly diagnosed metastatic cancer (33). Although further research is warranted, one possible reason why only a third of patients wanted be informed about their prognosis directly from a doctor was that patients felt that they would be dissatisfied with the prognostic information provided by the doctor (14). Another possible reason was that elderly adults in Japan prefer to leave decisions to a medical expert, and considered it an important component of good death (23,34). As several previous randomized controlled trials suggested it is important not only to “tell and know the life expectancy” but also to “have a dialogue about the prognosis” (35-37), further research is warranted to facilitate the dialogue about prognosis.

Of note, about a fifth of patients wanted to be informed about their prognosis on the Internet by entering their information such as the result of a blood test. This result would suggest new clinical questions such as “What is the most desirable way for cancer patients to be informed about their own prognostic information?”. As a previous cohort study revealed that cancer patients who were informed of their terminal diagnosis had a significantly better quality of life and a lower rate of emotional distress than patients who guessed it from their worsening condition (13), it is important to explore the most desirable way for patients to be informed about their prognostic information. This may require a variety of options to meet individual needs.

The strengths of our study were that is included patients with various cancer types, age categories, an almost equal male/female ratio, respondents from all eight regions of Japan, and no missing data regarding variables and outcomes. However, our study had some limitations. First, the number of participants who were wanted to be informed about their prognostic information was small. Most participants had relatively good performance status and did not receive aggressive chemotherapy; therefore, they may not be representative of incurable cancer patients that need prognostic information. Second, as we applied convenient sampling via the Internet using a private web-based company and analyzed the first 132 responders, we could not extract a response rate or the characteristics of non-responders. Although this sampling method may introduce selection bias, we recruited the patients based on several categories, such as age, sex, and primary cancer site; therefore, we believe that we could minimize the influence of selection bias. In addition, our study mostly targeted patients that were more active online, future studies using a different methodology are needed to assess the preferences of elderly cancer patients. Third, we used questionnaires that had not been clearly validated; thus, we could not strictly compare the patients’ various preferences with clinical and statistical significance. It is necessary to develop reliable and valid survey items to analyze the preference of cancer patients for being informed about their prognosis. Fourth, as 66.7% of the participants had never had chemotherapy and only 14.4% had recurrence or metastases, the majority of the participants could be interpreted as a group with a high probability of cure. Our findings may suggest that the preference of these populations may be closer to the general population. Thus, caution is needed when adapting the results of the present study to all patients undergoing cancer treatment.

Conclusions

Our study found that cancer patients were more likely to want to be informed about their functional prognosis than their life expectancy. In addition, the experience of having a family member that had died of cancer had a significant correlation with the preference for the patient wanting to be informed about their life expectancy and functional prognosis.

Acknowledgments

The authors would like to thank Dr. David Hui from MD Anderson Cancer Center and Dr. Kazuhiro Kosugi from National Cancer Center Hospital East for suggesting the topic of this study.

Funding: This work was supported by JSPS KAKENHI (Grant Nos. 19H03866 and 22H03305). The funder had no role in the design and conduct of the study; the collection, management, analysis, and interpretation of data; the preparation, review, and approval of the manuscript; or the decision to submit the manuscript for publication.

Footnote

Reporting Checklist: The authors have completed the SURGE reporting checklist. Available at https://apm.amegroups.com/article/view/10.21037/apm-22-772/rc

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://apm.amegroups.com/article/view/10.21037/apm-22-772/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The Institutional Review Board of Seirei Mikatahara General Hospital (Approval No. 21-37) approved the protocol of this study. All the participants consented to participate.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bernacki R, Hutchings M, Vick J, et al. Development of the Serious Illness Care Program: a randomised controlled trial of a palliative care communication intervention. BMJ Open 2015;5:e009032. [Crossref] [PubMed]
2. Hiratsuka Y, Yamaguchi T, Maeda I, et al. The Functional Palliative Prognostic Index: a scoring system for functional prognostication of patients with advanced cancer. Support Care Cancer 2020;28:6067-74. [Crossref] [PubMed]
3. Hagerty RG, Butow PN, Ellis PA, et al. Cancer patient preferences for communication of prognosis in the metastatic setting. J Clin Oncol 2004;22:1721-30. [Crossref] [PubMed]
4. Sisk BA, Kang TI, Mack JW. Prognostic disclosures over time: Parental preferences and physician practices. Cancer 2017;123:4031-8. [Crossref] [PubMed]
5. Enzinger AC, Zhang B, Schrag D, et al. Outcomes of Prognostic Disclosure: Associations With Prognostic Understanding, Distress, and Relationship With Physician Among Patients With Advanced Cancer. J Clin Oncol 2015;33:3809-16. [Crossref] [PubMed]
6. Fried TR, Tinetti ME, Iannone L, et al. Health outcome prioritization as a tool for decision making among older persons with multiple chronic conditions. Arch Intern Med 2011;171:1854-6. [Crossref] [PubMed]
7. Paladino J, Lakin JR, Sanders JJ. Communication Strategies for Sharing Prognostic Information With Patients: Beyond Survival Statistics. JAMA 2019;322:1345-6. [Crossref] [PubMed]
8. Miyashita M, Morita T, Sato K, et al. Good death inventory: a measure for evaluating good death from the bereaved family member's perspective. J Pain Symptom Manage 2008;35:486-98. [Crossref] [PubMed]
9. Ullrich A, Hollburg W, Schulz H, et al. What are the personal last wishes of people with a life-limiting illness? Findings from a longitudinal observational study in specialist palliative care. BMC Palliat Care 2022;21:38. [Crossref] [PubMed]
10. Balaban RB. A physician's guide to talking about end-of-life care. J Gen Intern Med 2000;15:195-200. [Crossref] [PubMed]
11. McMullan M. Patients using the Internet to obtain health information: how this affects the patient-health professional relationship. Patient Educ Couns 2006;63:24-8. [Crossref] [PubMed]
12. Dickerson SS, Reinhart A, Boemhke M, et al. Cancer as a problem to be solved: internet use and provider communication by men with cancer. Comput Inform Nurs 2011;29:388-95. [Crossref] [PubMed]
13. Yun YH, Kwon YC, Lee MK, et al. Experiences and attitudes of patients with terminal cancer and their family caregivers toward the disclosure of terminal illness. J Clin Oncol 2010;28:1950-7. [Crossref] [PubMed]
14. Nguyen SK, Ingledew PA. Tangled in the breast cancer web: an evaluation of the usage of web-based information resources by breast cancer patients. J Cancer Educ 2013;28:662-8. [Crossref] [PubMed]
15. Nagler RH, Gray SW, Romantan A, et al. Differences in information seeking among breast, prostate, and colorectal cancer patients: results from a population-based survey. Patient Educ Couns 2010;81:S54-62. [Crossref] [PubMed]
16. Dennison L, Brown M, Kirby S, et al. Do people with multiple sclerosis want to know their prognosis? A UK nationwide study. PLoS One 2018;13:e0193407. [Crossref] [PubMed]
17. Wynn Mon S, Myint Zu WW, Myint Maw M, et al. Awareness of and preference for disease prognosis and participation in treatment decisions among advanced cancer patients in Myanmar: Results from the APPROACH study. Asia Pac J Clin Oncol 2021;17:149-58. [Crossref] [PubMed]
18. Zafar W, Hafeez H, Jamshed A, et al. Preferences regarding disclosure of prognosis and end-of-life care: A survey of cancer patients with advanced disease in a lower-middle-income country. Palliat Med 2016;30:661-73. [Crossref] [PubMed]
19. Uchida M, Sugie C, Yoshimura M, et al. Factors associated with a preference for disclosure of life expectancy information from physicians: a cross-sectional survey of cancer patients undergoing radiation therapy. Support Care Cancer 2019;27:4487-95. [Crossref] [PubMed]
20. Masterson MP, Slivjak E, Jankauskaite G, et al. Beyond the bucket list: Unfinished and business among advanced cancer patients. Psychooncology 2018;27:2573-80. [Crossref] [PubMed]
21. Umezawa S, Fujimori M, Matsushima E, et al. Preferences of advanced cancer patients for communication on anticancer treatment cessation and the transition to palliative care. Cancer 2015;121:4240-9. [Crossref] [PubMed]
22. Ahalt C, Walter LC, Yourman L, et al. "Knowing is better": preferences of diverse older adults for discussing prognosis. J Gen Intern Med 2012;27:568-75. [Crossref] [PubMed]
23. Mori M, Kuwama Y, Ashikaga T, et al. Acculturation and Perceptions of a Good Death Among Japanese Americans and Japanese Living in the U.S. J Pain Symptom Manage 2018;55:31-8. [Crossref] [PubMed]
24. Yamaguchi T, Maeda I, Hatano Y, et al. Communication and Behavior of Palliative Care Physicians of Patients With Cancer Near End of Life in Three East Asian Countries. J Pain Symptom Manage 2021;61:315-322.e1. [Crossref] [PubMed]
25. Elkin EB, Kim SH, Casper ES, et al. Desire for information and involvement in treatment decisions: elderly cancer patients' preferences and their physicians' perceptions. J Clin Oncol 2007;25:5275-80. [Crossref] [PubMed]
26. Franssen SJ, Lagarde SM, van Werven JR, et al. Psychological factors and preferences for communicating prognosis in esophageal cancer patients. Psychooncology 2009;18:1199-207. [Crossref] [PubMed]
27. Fujimori M, Akechi T, Morita T, et al. Preferences of cancer patients regarding the disclosure of bad news. Psychooncology 2007;16:573-81. [Crossref] [PubMed]
28. Yoshida S, Shiozaki M, Sanjo M, et al. Pros and cons of prognostic disclosure to Japanese cancer patients and their families from the family's point of view. J Palliat Med 2012;15:1342-9. [Crossref] [PubMed]
29. Mori M, Yoshida S, Shiozaki M, et al. "What I Did for My Loved One Is More Important than Whether We Talked About Death": A Nationwide Survey of Bereaved Family Members. J Palliat Med 2018;21:335-41. [Crossref] [PubMed]
30. Yamashita R, Arao H, Takao A, et al. Unfinished Business in Families of Terminally Ill With Cancer Patients. J Pain Symptom Manage 2017;54:861-9. [Crossref] [PubMed]
31. Johnson SB, Butow PN, Bell ML, et al. A randomised controlled trial of an advance care planning intervention for patients with incurable cancer. Br J Cancer 2018;119:1182-90. [Crossref] [PubMed]
32. Sanjo M, Miyashita M, Morita T, et al. Preferences regarding end-of-life cancer care and associations with good-death concepts: a population-based survey in Japan. Ann Oncol 2007;18:1539-47. [Crossref] [PubMed]
33. Mori M, Shimizu C, Ogawa A, et al. A National Survey to Systematically Identify Factors Associated With Oncologists' Attitudes Toward End-of-Life Discussions: What Determines Timing of End-of-Life Discussions? Oncologist 2015;20:1304-11. [Crossref] [PubMed]
34. Akechi T, Miyashita M, Morita T, et al. Good death in elderly adults with cancer in Japan based on perspectives of the general population. J Am Geriatr Soc 2012;60:271-6. [Crossref] [PubMed]
35. Epstein RM, Duberstein PR, Fenton JJ, et al. Effect of a Patient-Centered Communication Intervention on Oncologist-Patient Communication, Quality of Life, and Health Care Utilization in Advanced Cancer: The VOICE Randomized Clinical Trial. JAMA Oncol 2017;3:92-100. [PubMed]
36. Paladino J, Bernacki R, Neville BA, et al. Evaluating an Intervention to Improve Communication Between Oncology Clinicians and Patients With Life-Limiting Cancer: A Cluster Randomized Clinical Trial of the Serious Illness Care Program. JAMA Oncol 2019;5:801-9. [Crossref] [PubMed]
37. Bernacki R, Paladino J, Neville BA, et al. Effect of the Serious Illness Care Program in Outpatient Oncology: A Cluster Randomized Clinical Trial. JAMA Intern Med 2019;179:751-9. [Crossref] [PubMed]