Patient-hematologist discordance in perceived chance of cure in hematologic malignancies: A multicenter study

Abstract

Background:

Ensuring that patients with hematologic malignancies have an accurate understanding of their likelihood of cure is important for informed decision-making. In a multicenter longitudinal study, we examined discordance in patients’ perception of their chance of cure versus that of their hematologists, whether patient-hematologist discordance changed after a consultation with a hematologist, and factors associated with persistent discordance.

Methods:

Pre- and post-consultation, patients were asked about their perceived chance of cure (options <10%, 10–19%, and up to 90–100% in 10% increments, and “do not wish to answer”). Hematologists were asked the same question post-consultation. Discordance was defined as a difference in response by 2 levels. We used McNemar’s test to compare changes in patient-hematologist prognostic discordance from pre- to post-consultation. We used generalized linear mixed model to examine associations between factors and post-consultation discordance, adjusting for clustering at the hematologist level.

Results:

We included 209 patients and 46 hematologists from 4 sites. Pre-consultation, 61% of dyads were discordant, which improved to 50% post-consultation (p<0·01). On multivariate analysis, lower education [less than college vs. post-graduate; Odds Ratio (OR) 2·24, 95% Confidence Interval (CI) 1·02–4·92], higher social support-affection subscale score (one unit change in score; OR 1.15, 95% CI 1·00–1·32), and discordance pre-consultation (OR 6·17, 95% CI 2.99–12·72) were significantly associated with discordance post-consultation.

Conclusions:

Patient-hematologist concordance in prognostic understanding improved after a hematology consultation, but half of patients’ views of their prognoses remained discordant with those of their hematologists. Interventions are needed to improve prognostic understanding among patients with hematologic malignancies.

Precis:

Patient-hematologist discordance in perceived chance of cure occurred in over 60% of dyads, which decreased to 50% after a hematology consultation. There is a need for interventions to improve prognostic understanding among patients with hematologic malignancies.

Introduction:

Hematologic malignancies are a group of cancers with varying treatment options and prognoses. As opposed to patients with advanced solid tumors, those with advanced hematologic malignancies may still have a chance for a cure of their underlying disease. Despite the potential for cure, many treatments such as chemotherapy and hematopoietic stem cell transplantation (HCT) carry significant associated risks. Because of this, there are associated challenges in communicating prognosis of hematologic malignancies¹ as compared to solid tumors which have a more defined staging delineation and prognosis. Nonetheless, ensuring that patients have an accurate understanding of their likelihood of cure or prognoses is not only important for informed decision-making but also for helping patients to decide on their goals of care, thereby facilitating goal-concordant medical care.

Prior studies show that up to 82% of patients with hematologic malignancies have a different understanding of their prognosis compared to their hematologist.^2–5 Most of this discordance is skewed towards optimism, whereby patients tend to have higher expectations for cure. Discordance may exist due to patient-related (e.g., beliefs, emotional distress, social support), physician-related (e.g., communication, reluctance in discussing prognosis), and societal factors.^6–12 This discordance appears to have significant implications for care received, as patients who are more optimistic are also more likely to opt for aggressive care,¹³ and less likely to accept palliative and comfort-oriented care.^14–16 On the other hand, prognostic discordance has been relatively underexplored in hematologic malignancies, with the sparse existing studies having small sample sizes,^2–4 being conducted at a single site, and being primarily focused on acute myeloid leukemia and myelodysplastic syndromes, or those undergoing allogeneic HCT.

In this analysis of patients with hematologic malignancies, we describe discordance in patients’ perception of their chance of cure versus that of their hematologists. We also examine whether discordance changed after a hematology consultation and explore factors associated with persistent patient-hematologist discordance post-consultation.

Methods:

Study Design, Setting, and Participants

This is a secondary analysis of data from a multicenter longitudinal observational study conducted at four academic cancer centers (Dana-Farber Cancer Institute, Boston, MA; Massachusetts General Hospital, Boston, MA; Fred Hutchinson Cancer Research Center/Seattle Cancer Care Alliance, Seattle, WA; and Massey Cancer Center, Richmond, VA). The parent study focused on physician-patient communication among patients with newly or previously diagnosed hematologic malignancies who were referred for a hematology consultation. Study methods are described in detail elsewhere.^17–19 Briefly, patients aged ≥18 years with hematologic malignancies who were able to communicate in English were recruited from September 2003 to June 2007. Potentially eligible patients were identified through patient intake based on their diagnosis, gender, race, and hematologist. Eligible patients were then approached by mail and subsequent phone contact. After enrollment, patients completed baseline surveys prior to and 1–7 days after their consultation visit with the hematologist. Their hematologists were also enrolled, and asked to complete surveys about their demographics and the patient’s prognosis. All participants provided written informed consent. The study was approved by the Institutional Review Board at each institution.

Measures

All measures were collected via self-administered survey at baseline and/or follow-up (approximately 1–7 days post-consultation).

Patient-hematologist discordance in perceived chance of cure:

Surveys were administered to both patients and hematologists at baseline and post-consultation. Patients were asked about their perceived chance of cure: “Please tell me as a percentage, what do you think is the chance of your disease being cured with the therapies currently available?” Options included <10%, 10–19%, 20–29%, 30–39%, 40–49%, 50–59%, 60–69%, 70–79%, 80–89%, 90–100%, and “do not wish to answer”. This question was adapted from prior studies.^2,9,20 Hematologists were also asked: “What is this patient’s chance of cure with your recommended treatment”. Response options were identical, except for the exclusion of “do not wish to answer.”

Similar to a prior study,¹⁰ discordance in perceived chance of cure was defined as a difference between responses by 2 levels in the patient- hematologist dyads. For example, if a patient answered <10% and his/her hematologist answered 20–29%, this was considered discordant. If a patient answered <10% and his/her hematologist answered 10–19%, this was considered concordant. All responses of “do not wish to answer” were considered to be discordant.

Demographics and clinical variables:

Demographics and clinical variables were collected from patients at baseline. Demographic variables included age, gender (female vs. male), race (white vs. non-white), marital status (married vs. other), education (less than college graduate vs. college graduate vs. post-graduate), work status (working vs. not working), and annual household income (≥$100k vs. <$100k). Clinical variables included type of hematologic malignancy (acute leukemia vs. lymphoma vs. other) and comorbidities (Charlson Comorbidity score).²¹

Patient-reported measures:

Patient-reported measures collected at baseline included decision-making preference, coping, denial, anxiety, depression, and social support.

Decision-making preference:

Patients’ decisional preferences were assessed using the Control Preferences Scale (CPS),²² which asks if they prefer: 1) to leave decisions about their care and treatment up to their doctor, 2) to have their doctor make decisions with significant input from them, 3) to share equally in decisions, 4) to make decisions with significant input from their doctor, or 5) to be the one making decisions about their care. The CPS has been tested and validated in the cancer population.^23,24

Coping:

Coping was assessed using the Brief Coping Orientation to Problems Experienced (COPE), which includes 14 subscales: active coping, planning, positive reframing, acceptance, humor, religion, emotional support, instrumental support, self-distraction, denial, venting, substance use, behavioral disengagement, and self-blame.²⁵ The score for each subscale ranges from 2–8 and a higher score indicates increased utilization of that specific coping strategy.

Denial:

Denial was assessed using the 3-item Denial Scale.¹⁷ Patients were asked the following statements: “I feel there is nothing I can do to help myself”, “I believe it is best not to think of the future when something unpleasant is going to happen, and “I believe what you don’t know won’t hurt you”. Options include strongly agree, agree, disagree, and strongly disagree. The total score ranges from 4–12 and a higher score indicates less denial.

Anxiety and depression:

These were assessed using the 14-item Hospital Anxiety and Depression Scale (HADS).²⁶ The HADS consists of two subscales: anxiety and depression, each ranging from 0–21. A score of >7 is considered abnormal.

Social support:

Social support was assessed using the Medical Outcomes Study (MOS) Social Support survey, which has four subscales assessing specific domains of social support (17-items, total range of 17–85).²⁷ Subscales include: tangible (e.g., someone to help you if you are confined to bed; 4-items, total score range from 4–20), emotional (e.g., someone to give you good advice about a crisis; 7-items, total score range from 7–35), affectionate (e.g., someone to love and make you feel wanted; 3-item, total score range from 3–15), and social interaction (e.g., someone to do something enjoyable with; 3-item, total score range from 3–15).²⁸

Quality of life (QoL):

The 36-item Short Form Survey (SF-36) was used to assess health-related quality of life.²⁸ We combined the physical and mental composite scales into a total score; a higher score indicates better QoL.

Statistical Analyses

We used descriptive statistics to summarize our sample. We categorized patient responses for perceived chance of cure into 0–59%, 60–100%, and “do not wish to answer”. We used 60% because it was the median derived from the study population. We used McNemar’s test to compare changes in perceived chance of cure from pre- to post-consultation. In addition, we used McNemar’s test to compare changes in patient-hematologist prognostic discordance from pre- to post-consultation.

For multivariable modeling, we first identified a priori factors of interest (as mentioned above) that are potentially associated with discordance, based on prior studies^2,3,8,29,30 and expert consensus. We then examined their associations with patient-hematologist discordance at the post-consultation visit. All variables with a P-value of <0·20 on bivariate analyses, in addition to age and sex, were entered into the multivariate models.³¹ Multicollinearity was noted among the social support domains (total score and emotional, affectionate, and social interaction subscales, all r>0·80, p<0·01). Therefore, we assessed the associations of the individual social support domains in separate multivariate models. As patient-hematologist discordance is a dichotomized variable, we implemented the generalized linear mixed model with binary distribution and logit link function to examine the associations between predictors and post-consultation discordance, adjusting for clustering at the hematologist level. We used backward variable selection and retained variables with a P-value of ≤0.05 in the final model, in addition to age and sex.

To ensure the validity of our findings, we conducted two sensitivity analyses. First, we removed patients who selected “do not wish to answer” for perceived cure pre- and post-consultation, and repeated the regression analysis on this sample (N=201). Second, we removed patients who did not provide a response either pre- or post-consultation, and repeated the regression analysis on this sample (N=188). All statistical analyses were conducted using the SAS 9.4 (SAS Institute, Cary, NC).

Results

We included 209 patients (median age 53.7 years, SD 11.4) treated by 46 hematologists (median age 48.9, SD 8.9) in this analysis. The majority were white (92%, 189/209) and married (76%, 158/209); 45% (94/209) were women, 42% (87/209) had less than college education, and 58% (112/209) had an annual household income of <$100k. Approximately 19% had acute leukemia (15% acute myeloid leukemia and 4% acute lymphoblastic leukemia, 38% had lymphoma (26% non-Hodgkin lymphoma, 5% Hodgkin lymphoma, and 7% chronic lymphocytic leukemia/small lymphocytic lymphoma), and 43% had other hematologic malignancies (8% chronic myeloid leukemia, 14% multiple myeloma/amyloidosis, and 21% other myeloid malignancies). Forty-four percent (91/209) of patients have been treated in the past, and 48% (101/209) were receiving treatment at study enrollment. Table 1 shows the demographic and clinical characteristics of the patients as well as the results of bivariate analyses.

Table 1.

Participant demographics and bivariate analyses evaluating factors associated with discordance post-consultation

Variables		All (N=209)	Concordance in perceived chance of cure (N=105)	Discordance in perceived chance of cure (N=104)	Bivariate analyses; P-value
Age, mean		53.7 (11.4)	53.9 (11.3)	53.5 (11.6)	0.78
Gender, N (%)	Male	115 (55.0)	58 (55.2)	57 (54.8)	0.95
	Female	94 (45.0)	47 (44.8)	47 (45.2)
Race, N (%)a	White	189 (91.8)	94 (92.2)	95 (91.4)	0.83
	Non-White	17 (8.3)	8 (7.8)	9 (8.7)
Marital status, N (%)b	Married	158 (76.0)	81 (77.1)	77 (74.8)	0.69
	Non-married	50 (24.0)	24 (22.9)	26 (25.2)
Working, N (%)b	Yes	103 (49.5)	55 (52.4)	48 (46.6)	0.40
	No	105 (50.5)	50 (47.6)	55 (53.4)
Education, N (%)b	Less than college graduate	87 (41.8)	30 (28.6)	57 (55.3)	<0.01
	College graduate	60 (28.9)	38 (36.2)	22 (21.4)
	Post-graduate	61 (29.3)	37 (35.2)	24 (23.3)
Annual household income, N (%)c	≥100k	81 (42.0)	49 (50.5)	32 (33.3)	0.02
	<100k	112 (58.0)	48 (49.5)	64 (66.7)
Charlson comorbidity index, mean (SD)		0.4 (0.9)	0.4 (0.9)	0.4 (0.8)	0.99
Cancer type, N (%)	Acute leukemia	39 (18.7)	19.0 (18.1)	20 (19.2)	0.88
	Lymphoma	80 (38.3)	39.0 (37.1)	41.0 (39.4)
	Other	90 (43.1)	47 (44.8)	43.0 (41.4)
Decision making preference, mean (SD)1	Decision left to doctor or made by doctor with significant patient input	60 (28.9)	25 (24.0)	35 (33.7)	0.24
	Equal and shared decision-making	77 (37.0)	39.0 (37.5)	38 (36.5)
	Decision left to patient or made by patient with significant patient input	71 (34.1)	40.0 (38.5)	31 (29.8)
Denial, mean		10.5 (1.7)	10.5 (1.7)	10.5 (1.7)	0.66
Coping, mean (SD)	Self distraction	4.1 (1.8)	4.1 (1.7)	4.2 (1.8)	0.55
	Active coping (yes)	5.9 (1.9)	6.1 (1.8)	5.8 (2.0)	0.34
	Denial (no)	2.7 (1.4)	2.7 (1.3)	2.8 (1.4)	0.66
	Substance use	2.3 (1.1)	2.3 (1.2)	2.3 (1.0)	0.84
	Emotional support	6.2 (1.8)	6.0 (1.8)	6.3 (1.8)	0.25
	Behavioral disengage	2.3 (0.9)	2.3 (1.0)	2.3 (0.9)	0.67
	Instrumental support	5.1 (2.0)	5.2 (1.9)	5.0 (2.1)	0.55
	Venting	3.6 (1.7)	3.6 (1.7)	3.5 (1.8)	0.80
	Positive reframing (yes)	5.0 (2.1)	5.0 (2.0)	5.0 (2.3)	0.87
	Planning	6.1 (2.0)	6.2 (1.9)	6.0 (2.0)	0.38
	Humor	3.6 (1.8)	3.7 (1.7)	3.6 (1.9)	0.59
	Acceptance (yes)	6.6 (1.8)	6.6 (1.7)	6.5 (1.9)	0.80
	Religion	3.3 (1.4)	3.2 (1.4)	3.3 (1.4)	0.41
	Self-blame (no)	2.4 (1.1)	2.4 (1.0)	2.5 (1.2)	0.44
Anxiety, mean (SD)		9.5 (2.3)	9.3 (2.0)	9.7 (2.6)	0.14
Depression, mean (SD)		8.6 (1.5)	8.8 (1.4)	8.4 (1.6)	0.27
Social support, mean (SD)e	Total	81.2 (13.8)	79.5 (14.7)	82.9 (12.7)	0.08
	Tangible	17.0 (3.4)	16.9 (3.4)	17.1 (3.4)	0.68
	Emotional	29.4 (5.8)	28.8 (6.0)	30.1 (5.1)	0.09
	Affectionate	13.5 (2.5)	13.0 (2.7)	13.9 (2.1)	0.01
	Social interaction	13.0 (2.4)	12.7 (2.6)	13.4 (2.2)	0.03
Quality of life, SF-36 total score, mean (SD)d		91.9 (18.4)	94.2 (18.1)	89.5 (18.5)	0.07
Discordance pre-consultation, N (%)	No	76 (36.4)	59 (56.2)	17 (16.4)	<0.01
	Yes	120 (57.4)	41 (39.1)	79 (76.0)
	Did not fill	13 (6.2)	5 (4.8)	8 (7.7)

^a3 missing

^b1 missing

^c16 missing

^d7 missing

^e2 missing

Abbreviations: SF-36, 36-Item Short Form Survey

Perceived chance of cure and prognostic discordance

Pre-consultation, 51% (101/196) of patients reported a 60–100% chance of cure, which decreased to 39% (86/209) post-consultation (McNemar’s test; p<0.01). Hematologists reported a 60–100% chance of cure post-consultation in just 21% of patients (44/209).

Pre-consultation, 61% (120/196) of patient-hematologist dyads were discordant and 39% (76/196) were concordant. Among the discordant dyads, 70% (84/120) were due to the patient being more optimistic than the oncologist, 24% (29/120) were due to the oncologist being more optimistic than the patient, and 7 patients answered “I don’t know” whereas the oncologist provided a response. Post-consultation, 50% (104/209) of the dyads were discordant (McNemar’s test; p<0.01) (Figure 1). Among the discordant dyads, 72% (75/10) were due to the patient being more optimistic than the oncologist, 26% (27/104) were due to the oncologist being more optimistic than the patient, and 2 patients answered “I don’t know” whereas the oncologist provided a response.

Figure 1:

Changes in discordance in chance of cure from pre- to post consultation

Bivariate analyses

Table 1 shows the bivariate analyses. Education, annual household income, and social support (affection and social interaction subscales) were significantly associated with discordance (all p<0.05). Discordance was more common in dyads where patients had lower education levels (less than college education vs. post-graduate education, 55% vs. 23%, p<0.01) and lower annual household income (<100k vs. ≥100k, 67% vs. 33%, p=0.02). In addition, compared to the concordance dyads, patients in the discordance dyads reported higher scores on the social support-affectionate subscale (13.9 vs. 13.0, p=0.01) and higher social support social interaction subscale (13.4 vs. 12.7, p=0.03)..

In addition, three variables (anxiety, social support-total scale, and social support-emotional subscale) met the pre-specified cut-off for evaluation in the multivariate analyses. Age, gender, race, marital status, working status, comorbidity, cancer type, decision-making preference, denial, coping, depression, and quality of life (SF-36) were not associated with discordance post-consultation.

Multivariate analyses

Lower education [less than college education vs. post-graduate education; Odds Ratio (OR) 2.24, 95% Confidence Interval (CI) 1.02–4·92], higher social support-affection subscale score (one unit change in score; OR 1.15, 95% CI 1.00–1.32), and discordance pre-consultation (OR 6·17, 95% CI 2.99–12.72) were significantly associated with discordance post-consultation (Table 2). Age and gender were not associated with discordance post-consultation. Results were similar in the sensitivity analyses (Supplemental Table 1).

Table 2.

Multivariate analysis evaluating factors associated with discordance post-consultation

Variables (N=206)		AOR	95% CI	Multivariate analysis; P-value
Agea		0.98	0.85–1.14	0.82
Gender	Male	Reference	-
	Female	1.35	0.69–2.67	0.38
Education	Post-graduate	Reference	-	0.02b
	College graduate	0.76	0.32–1.78	0.52
	Less than college graduate	2.24	1.02–4.92	0.04
Social support-affectionate subscalec		1.15	1.00–1.32	0.05
Discordance pre-consultation	No	Reference	-
	Yes	6.17	2.99–12.72	<0.01
	Did not fill	5.98	1.76–20.32	<0.01

^a5 units increase in number of years

^bOverall P-value

^c1 unit increase in social support-affectionate subscale score

Abbreviations: AOR, Adjusted Odds Ratio; CI, Confidence Interval

Discussion

Data on patient and oncologist perceptions of prognosis are limited, and even more so for patients with hematologic malignancies. In this large, multicenter study, we found that almost two-thirds of patient-hematologist dyads were prognostically discordant prior to a hematology consultation, which decreased to 50% post-consultation. Discordance post-consultation was more common in patients with lower education level and higher social support (in the form of affection). In addition, dyads who were discordant pre-consultation were more likely to be discordant post-consultation.

Methods used to assess prognostic understanding vary in published studies; many studies assessed prognostic understanding from patients only³² with some also included oncologist assessment.^8,10 Therefore, the prevalence of discordance differs and comparison across studies is challenging. For example, in a multicenter study, Loh and colleagues found that discordance in beliefs about curability occurred in 60% of dyads involving patients with advanced cancer and their community oncologist.⁸ In a single center study, Lennes and colleagues found that discordance in curability occurred in 28% of patients with advanced cancer and their oncologist.³³ Despite data existing in other cancer types, studying prognostic understanding specifically in hematologic malignancies is important due to the higher uncertainty of prognosis and often higher disease acuity compared to solid tumors.

Discordance in perceived chance of cure is influenced by communication between patients and oncologists,¹⁰ and also by patient beliefs.²⁹ The change in discordance from pre- to post-consultation in our study is likely the effect of communication on patients’ understanding of their prognoses. While our study does not allow us to directly assess why discordance decreased, prior studies have shown that several communication factors influence prognostic understanding.^9,34 These include oncologist exchange of specific information³⁴ and use of pessimistic statements during clinic encounters.⁹ Of note, the percentage of dyads with discordance in our study was lower compared to prior studies in AML and MDS (>75% of dyads were discordant),^2,34 which could reflect the younger age, heterogeneous hematologic malignancies, and higher level of socioeconomic status in our sample, all of which have been shown to influence knowledge of true prognosis.^8,35 Still, the degree of prognostic over-estimation and discordance with the hematologist’s assessment that we found in this cohort are provocative.

We found that higher social support as assessed via the affection subscale of the MOS Social Support survey was associated with discordance post-consultation. The affection subscale assessed whether there was someone to show patients love and affection, someone for the patients to love and make them feel wanted, and someone who hugs the patients. This could reflect that friends or family members may wish to reassure their loved one of a good outcome, and that false gestures of reassurance amid a poor prognosis actually lead to prognostic discordance. Another possible explanation is that patients who have better social support may feel better and have lower symptom and emotional burdens, where may result in an optimistic view of their prognosis and discordance. In a prior study of patients with advanced cancer in their final weeks of life, those who had intensive social contact, defined as having ≥8 family members and >8 contacts per week, were more likely to have poor understanding of their prognosis.⁶ In contrast, among older patients with advanced cancer, those who had impairment in instrumental social support (not having someone to help with doctor appointments, meal preparation, daily chores, or if they were confined to bed) were more likely to believe with 100% certainty that they will be cured.²⁹ Findings from these studies, and ours, are intriguing, and highlight the critical need for more research on the intricate relationships between patients, their social network, and their hematologist/oncologist. While social support has been shown to be associated with outcomes of cancer care,³⁶ our results underscore the important role patients’ families and friends play in influencing patients’ understanding of their prognosis. While some friends and family members may bolster patient self-efficacy and autonomy, others may influence patient prognostic understanding so it is in agreement with their own beliefs.³⁷ Further work should assess the various domains of social support and how they may shape patient prognostic understanding and subsequent treatment decisions.

Consistent with prior studies, discordance post-consultation was more common in patients with lower levels of education.^8,38 There are several potential explanations for this. Patients with lower educational levels are more likely to have lower health literacy,³⁹ which creates challenges in understanding medical words and processes of care (e.g., test results and prognostic terms).⁴⁰ Moreover, these patients may not feel their hematologists are responsive to their concerns.⁴⁰ Finally, patients with lower education levels are more likely to hold fatalistic worldviews (e.g., “it’s in God’s hands”, “whatever happens is just fate one can do little about”).⁴¹ Having a fatalistic worldview may also lead them to have more optimistic views of prognoses, thereby being more discordant with their hematologists when disease is advancing.

Unlike previous studies in hematologic malignancies,^2,3 we did not find significant associations between anxiety or depression and patient-hematologist discordance in perceived chance of cure. In a study of 90 patients with hematologic malignancies undergoing HCT, patient-hematologist concordance was associated with greater depressive symptoms at baseline and over time.² In another study of 100 older patients with AML, patients who reported a lower likelihood of cure compared to their hematologists were more likely to report greater depressive symptoms.³ The difference in findings among these studies with ours may reflect the heterogeneity of study populations and varying follow-up periods. Therefore, the relationships between psychological health and prognostic understanding need to be further evaluated in future studies. Better understanding of these relationships will help to identify patients who require psychological support in order to cope with their illnesses.³⁰

There are several strengths to our study. First, we included a heterogeneous group of hematologic malignancies at four academic centers. Second, to our knowledge, ours is the largest study of prognostic discordance in hematologic malignancies. Our analysis also has several limitations to note. First, we enrolled patients that were referred to academic cancer centers, thus our findings may not represent level of discordance from patients in the community who never make it to an academic center for consultation. That said, a prior study conducted in community oncology practices demonstrated a similarly high discordance rate among patients with advanced solid tumors and their oncologists.⁸ Second, our sample also consisted of patients who had relatively high income, which may limit its generalizability. Third, our study was a secondary analysis and data are relatively old. Therefore, it is unclear if prognostic understanding in patients with hematologic malignancies have changed over the last two decades.³² Fourth, we do not know if prognosis was discussed during the clinic visit, and therefore we could not ascertain if the discordance was due to the hematologist not communicating prognosis or the patient not understanding/accepting the prognosis if it was indeed discussed. Nonetheless, prior research has shown that both of these factors may contribute to discordance.^10,29 Fifth, we did not collect information regarding the aggressiveness of the cancer, the duration of the consultation, specific communication skills used by the oncologist, and whether the patient had a prior hematologist. Future studies should evaluate the associations of the aforementioned factors with discordance. Finally, we focused on discordance rather than the directionality (e.g., patient optimism vs. hematologist optimism), although most of the discordance were due to patient optimism.⁸

In conclusion, patient-hematologist discordance in perceived chance of cure likely decreases after hematology consultation, but in half of patients prognostic discordance persists. Discordance pre-consultation, lower education levels, and higher social support in the form of affection were all associated with discordance post-consultation. To improve prognostic discordance, it is important to study the communication between patients with hematologic malignancies and their hematologists. Interventions are needed to improve prognostic understanding among patients with hematologic malignancies. In addition, future studies are needed to understand how patients’ education level and social support (e.g., what do they hear from their social networks and who do they hear from) influence their prognostic understanding and communication with their hematologists. These will better equip healthcare professionals in gaining patients’ trust and better prepare them on how to reduce prognostic discordance. Studies have shown that social support is associated with better outcomes in patients.^42,43 By understanding the role of social support in prognostic understanding, it will further guide intervention development.