Risk Factors for Depression in Patients Undergoing Hematopoietic Cell Transplantation

Abstract

Despite the prevalence and known adverse impacts of depression after hematopoietic cell transplantation (HCT), little is known about the trajectory of depression following HCT, or which pre-transplant risk factors might help predict new or worsening depression post-HCT. This secondary analysis was conducted to evaluate the relationships between pre-transplant patient-reported outcomes and demographic characteristics and post-transplant depression. 228 adult HCT patients were evaluated pre-transplant (T1) and 6 to 7 weeks post-transplant (T2), using touch-screen computers in the transplant clinic during participation in a larger trial. Measures included the Symptom Distress Scale, the EORTC QLQ-C30 for quality of life, a single-item Pain Intensity question, and the PHQ-9 for measurement of depression. At T1, rates of depression were quite low with only 6% of participants endorsing moderate or higher depression. At T2, however, 31% had moderate or higher depression. We observed a strong linear relation in PHQ-9 scores between T1 and T2 (p<.0001). T1 depression score was a significant predictor of depression scores at T2 (p=.03), as was poorer emotional function at T1 (p<.01). Results indicate that depression is common post-HCT, even for patients with low depression pre-transplant. Frequent screening for depressive symptoms at critical time points, including 6–7 weeks post-HCT, are needed in this population, followed by referrals to supportive care as appropriate.

Introduction

Success rates for hematopoietic cell transplantation (HCT) have continued to improve as the procedure has been refined [1,2]. Despite this significant progress, HCT remains an extraordinarily stressful procedure physically, mentally, and emotionally[3,4]. One significant, and potentially limiting, symptom associated with HCT is depression. Depression is one of the most common psychiatric conditions during and after cancer treatment. Prevalence estimates of depression across cancer patients range from 3% to over 50% depending on the timing and method used to measure the symptoms [5]. Studies have indicated that depression is prevalent in patients undergoing HCT, with estimates that a quarter to a third of HCT patients experience depression during the first 100 days or in recovery from their transplant [3,6–9].

Depression has many potential negative psychosocial and physical impacts in persons undergoing HCT. It can interfere significantly with quality of life, physical, social, and recreational activities, and overall health, and can be comorbid with other significant concerns, such as post-traumatic stress disorder and suicidal ideation in HCT survivors [3,10,11]. Depression can also interfere with cancer treatment adherence, and is associated with negative health behaviors such as tobacco and alcohol use [12,13]. Depression is well-known to be associated with increased mortality in the general population [14–16], as well as in cancer patients [17]. Depression may be an independent risk factor for survival after HCT, over and above its status as a potential indicator of poorer health status [7,18]. National accreditation bodies, including the National Comprehensive Cancer Network [19] and the Commission on Cancer [20], have mandated that distress screening be completed during a patient’s treatment. For patients with clinical evidence of moderate or severe distress, the oncology team must “assess the psychological, behavioral, and social problems of patients that may interfere with their ability to participate fully in their health care and manage their illness and its consequences”[20]. Patients must then be referred to appropriate supportive care, and a follow up plan determined. Thus, for HCT clinicians, early identification of depression is a critical element of comprehensive HCT care, along with appropriate referral and intervention to address symptoms. Understanding the risk factors and clinical course of depressive symptoms after HCT will help inform with whom and at what time points screening should occur [11].

The time course of depressive symptoms may vary significantly between HCT patients. In some, depression may occur before HCT begins and persist (or worsen) throughout the course of treatment; in other patients, depression may not appear until weeks or months after the transplant occurs, remaining a long-term concern for patients undergoing HCT. In one study of HCT survivors 1–3 years after transplant, 15% reported moderate to severe depressive symptoms. In that study, allogeneic HCT recipients (vs. those receiving autologous transplant) and those with poorer functional status reported higher levels of depression [10]. Another long-term study of recovery post-HCT found that 19% of patients 5 years post-HCT continued to experience depressive symptoms [6].

Despite the prevalence and known adverse impacts of depression after HCT, little is known about the trajectory of depression immediately following HCT, or which pre-transplant risk factors might help predict new or worsening depression post-HCT. This analysis was conducted to evaluate the relationships between pre-transplant patient-reported outcomes and demographic characteristics and post-transplant depression. Variables of interest included symptom distress, quality of life, demographics, and social roles (vocational status, relationship status, etc.). The purpose of the analysis is to aid clinicians in identifying patients who might be at high risk for depression in the early post-transplant period, facilitating early detection and thus more effective intervention for those patients.

Materials and Methods

Sample

Research participants for the larger Electronic Self-Report Assessment - Cancer (ESRA-C) study [21], from which these data were collected, were recruited from the Seattle Cancer Care Alliance (SCCA), a consortium between the University of Washington Medical Center, Fred Hutchinson Cancer Research Center, and Seattle Children’s Hospital in Seattle, Washington, USA. The SCCA cared for 3,609 new patients in 2006, when these data were collected, with the majority (85%) originating from Washington State. Eligibility criteria for the analytic sample included the following: new patients who were being evaluated for HCT, at least 18 years of age, able to communicate in English, and competent to understand the study information and give informed consent. Participants were included irrespective of the presence of diagnosis or treatment of psychiatric conditions, as long as they met criteria for HCT. Between April 2005 and November 2006, 228 eligible HCT patients enrolled in the study.

Procedures

Details of the methods and procedures of the ESRA-C study are described elsewhere [21]. In brief, baseline assessments (T1) were administered via touch-screen computer at a clinic visit prior to beginning HCT conditioning. At the first ambulatory visit post-transplant (+6 to 7 weeks), patients were surveyed a second time (T2), using the same methodology. The technical aspects and navigability of the ESRA-C program have been described elsewhere [22–24]. The ESRA-C has been shown to be well-received by patients [21,25]. During the T1 session, patients were presented with an introductory screen followed by demographic questions. Patients then were presented with four validated questionnaires during both T1 and T2 survey sessions:

• Symptom Distress Scale (SDS) [26]. The 13-item SDS assesses the level of symptom distress for eleven symptoms including nausea, appetite, insomnia, pain, fatigue, concentration, and others. Each item is scored from 1–5 with descriptive options. The score is a sum of all item scores.

• The European Organization for Research and Treatment of Cancer (EORTC) QLQ-C30 v.3 [27]. The EORTC QLQ-C30 is a questionnaire developed to assess the quality of life of cancer patients, including those undergoing HCT [28]. The QLQ-C30 incorporates nine multi-item scales: five functional scales (physical, role, cognitive, emotional, and social); three symptom scales (fatigue, pain, and nausea and vomiting); and a global health and quality-of-life scale. Each subscale is a multi-item index which results in a 0–100 score, with higher scores indicating better function for the functional scales (e.g. emotional function), and more symptoms for the symptom scales (e.g. fatigue). Emotional function questions include symptoms such as irritability, worry, tension, and depression.

• A single-item Pain Intensity Numerical Scale from 0–10, where 0 is no pain and 10 is the worst pain imaginable.

• The nine-item Patient Health Questionnaire (PHQ-9) depression scale [29]. The PHQ-9 has been validated for in-person self-report or interviewer administration as well as administration over the telephone [30–33] Standard PHQ-9 depression scores were categorized as follows: no (≤4), mild (5–9), moderate (10–14), moderately severe (15–19), or severe (≥20) depression. A score of 10 has been found to be the optimal cutoff for identifying probable major depressive disorder (sensitivity, 0.88; specificity, 0.88) in primary care patients [29]. For analyses, we categorized patients into “no/mild” depression (PHQ-9 total score <10) and “moderate or higher” depression (PHQ-9 total score 10 or higher). To minimize patient burden, we used the presence of at least one cardinal symptom of depression on at least half the days in the last two weeks, either (a) anhedonia or (b) depressed mood, as a trigger for completing the remaining seven items of the PHQ-9. Initial screening in this manner, known as the PHQ-2, has been validated in medical populations [34–36]. Participants who did not trigger the full PHQ-9 were categorized as having “no/mild” depression. To provide additional data among distressed subjects, the full PHQ-9 was also triggered in the case of specific responses on the QLQ-C30 (scores ≤50 on the scale of 0–100 for the Emotional Function or Cognitive Function subscales) or SDS (scores ≥3 on the response range of 1–5 for the fear/worry, concentration, or sleep disturbance items).

Analysis

Baseline patient socioeconomic factors and quality of life measures were compared between dropouts and those completing the study with a t-test for continuous variables (age) or a Fisher exact/Chi-Square test for categorical variables. A generalized McNemar test was used to check potential pattern change in depression between T1 and T2. Logistic regression was used to predict moderate or higher depression at T2, with a list of pre-selected baseline variables, including minority (or not), income, education, working status, computer use, partnered (or not), transplant type and baseline measures on QLQ-C30, SDS, pain, and the PHQ-9. The factors were first checked individually adjusting for T1 depression status, and then factors with p<0.2 were included in the multivariable model. A backward model selection was used for variable selection and all variables with p<0.1 were retained in the final model. Odds ratios and two-sided p-values were calculated. The analysis was performed with SAS (version 9.2) and R (version 2.15.0).

Results

Two-hundred twenty-eight HCT participants were enrolled in the study. Thirty-six of the 228 participants did not complete assessments at both time points; attrition was mainly due to death or illness. Therefore, the final analytic sample included 192 participants. There were no significant differences in demographic characteristics between non-completers and those completing the study, although there were trends for non-completers to be older (p=0.06) and less likely to be working at T1 (p=0.07) than those who remained in the analysis. However, non-completers did have significantly lower global QOL scores (subscale of the QLQ-C30) at T1 compared to those remaining in the analysis (p=0.04).

Demographic and clinical characteristics of the 192 study participants are shown in Table 1. The sample was 59% male, and the majority were married, had 2 or more years of college education, and were familiar with using computers at either home or work. Participants were mainly Caucasian and non-Hispanic/Latino (91%). The majority of participants (59%) were working, on medical leave or students.

Table 1.

Baseline demographics and clinical characteristics, N=192

	N (%)
Age: Median (range)	51 (19–75)
Gender
Male	114 (59%)
Minoritya	17 (9%)
Married or partnered	139 (72%)
College education or more	147 (77%)
Income ≥ $55K/year	103 (54%)
Working at T1b	114 (59%)
Computer userc	177 (92%)
Cancer diagnosis
Leukemias	92 (48%)
Lymphomas	59 (31%)
Myelomas	38 (20%)
Other	3 (2%)
Transplant type
Autologous	76 (40%)
Allogeneic	116 (60%)

^a“Minority” is defined as non-Caucasian race and/or Hispanic/Latino ethnicity, by self-report.

^b“Working at T1” is defined as working full-time, part-time, on full or partial medical leave, or being a student. “Not working” is defined as being unemployed or fully retired.

^cUses a computer sometimes or often at home or work.

Changes in PHQ-9 depression categories from T1 to T2 are presented in Table 2. At T1, rates of depression were low, with 11 (6%) endorsing moderate or higher depression. However, at T2, 60 (31%) had moderate or higher depression. This difference in percentage of patients experiencing moderate or higher depression at T2 compared with T1 was significant (p<.001 by Generalized McNemar test).

Table 2.

Change in PHQ-9 depression scores from T1 to T2. Number of participants (with percent of total N) meeting criteria for no/mild and moderate or higher depression categories, based on PHQ-9 scores, are listed at each time point.^a

		T2
		No/Mild	Moderate or Higher
T1	No/Mild	131 (68.2%)	50 (26.1%)	181 (94.3%)
	Moderate or Higher	1 (0.5%)	10 (5.2%)	11 (5.7%)
		132 (68.7%)	60 (31.3%)	192 (100%)

^aCategories: no/mild depression= PHQ-9 score <10, moderate or higher= PHQ-9 score ≥10

We observed a strong linear relation in PHQ-9 category scores between T1 and T2 (p<.0001). Results of univariate analyses are shown in Table 3a. Adjusting for T1 depression status, depression at T2 was associated with poorer emotional function (p<.01) and greater symptom distress (p=.02) at T1. Transplant type (autologous vs. allogeneic) was not a significant predictor of elevated depression at T2. The prevalence of moderate or higher depression at T1 ant T2 in the allogeneic HCT patients was 7% (8/116) and 34% (40/116), and in the autologous HCT patients was 4% (3/76) and 26% (20/76). These differences were not statistically significant (P=0.53 at T1, P=0.27 at T2). Results of multivariate analyses are shown in Table 3b. T1 depression score remained a significant predictor of depression scores at T2 (p=.03), as did poorer emotional function at T1 (p<.01). In multivariate analysis, however, elevated symptom distress was no longer significant (p=.18) in predicting elevated depression at T2. T1 working status trended toward significance (p=.07), such that those who were working at T1 (even if on medical leave for their treatment) had higher rates of depression at T2. All possible two-way interactions in the multivariate model were tested, and none were significant.

Table 3a.

Univariate logistic regression predicting moderate or higher depression (PHQ-9 scores ≥10) at T2, adjusting for depression status at baseline. Depression status at T1 is significantly associated with T2 status (p<.0001). Bold text indicates significant predictors.

T1 Predictor	β	Odds Ratio	p value
Male	−0.12	0.89	0.73
Minoritya	−0.97	0.38	0.18
Working at T1b	0.54	1.72	0.12
College education or more	0.58	1.78	0.18
Married or partnered	0.53	1.69	0.19
Income ≥ $55K/year	0.17	1.18	0.63
Computer userc	0.01	1.01	0.98
Global quality of life (EORTC QLQ-C30)	−0.01	0.99	0.15
Physical function (EORTC QLQ-C30)	−0.01	0.99	0.43
Role function (EORTC QLQ-C30)	0	1	0.56
Emotional function (EORTC QLQ-C30)	−0.04	0.96	<.01
Cognitive function (EORTC QLQ-C30)	0	1	0.66
Social function (EORTC QLQ-C30)	0	1	0.45
Fatigue (SDS)	0.01	1.01	0.15
Nausea/vomiting (SDS)	0	1	0.96
Pain (SDS)	0	1	0.94
Total symptom distress (SDS)	0.08	1.08	0.02
Transplant type (allogeneic)	0.33	1.39	0.33
Impact on sexual activities and interest)	0.11	1.12	0.32
Fever/chills	−0.06	0.94	0.89

^a“Minority” is defined as non-Caucasian race and/or Hispanic/Latino ethnicity, by self-report.

^b“Working at T1” is defined as working full-time, part-time, on full or partial medical leave, or being a student. “Not working” is defined as being unemployed or retired at T1.

^cUses a computer at least sometimes at home or work.

Discussion

This exploratory investigation adds to the growing body of literature indicating that depression is common following HCT, and provides information that will help identify patients at risk for depression post-transplant. While few patients (6%) met criteria for moderate or higher depression before their transplants, nearly one-third (31%) met criteria for moderate or higher depression when assessed at 6–7 weeks post-HCT. Many participants who had elevated depression scores at T2 were not depressed at T1, indicating that even if a patient seems to be doing well emotionally pre-HCT, there may be a significant risk of depression post-transplant.

At our center, as at many others, all HCT patients get a comprehensive social work evaluation pre-transplant. Several factors place patients at elevated risk of depression post-transplant, and these factors should help clinicians identify patients who merit closer monitoring as they proceed through the transplant process. Depression scores at T1, unsurprisingly, were strongly associated with T2 scores. Thus, patients who enter the HCT process with symptoms of depression should be monitored closely for worsening of their symptoms post-HCT. Another risk factor for elevated depression at T2 was lower self-reported emotional function at T1 a subscale of the EORTC QLQ-C30. This effect was significant even controlling for depression score at T1, indicating that clinicians need to attend not only to symptoms of depression per se, but also for other broad indicators of emotional distress pre-HCT, such as anxiety, tension and irritability. Others also have found that pre-transplant distress (usually defined as anxiety or depression) have been the strongest predictors of post-HCT anxiety or depression [6,37].

Overall symptom burden pre-transplant was another predictor of post-transplant difficulty with depression. Elevated SDS scores predicted high depression post-HCT in our univariate analysis (controlling for baseline depression), indicating that it is important for clinicians to help patients cope with their myriad symptoms before they begin HCT and to re-screen patients with high baseline symptom burden for depression one to two months post-transplant. The impact of symptom distress at T1 was not significant, however, when analyzed in a multivariate model, which may indicate that emotional distress (which may result in part from symptom burden), is a stronger factor predicting future depression.

An interesting, but non-significant, trend noted was that patients who had a vocational status that implied continuing responsibility for this role (including those who were working full or part-time, on full or part-time leave, or were students) at T1 had higher depression at T2. One potential explanation for this is that being employed or in school carries with it a set of expectations and stress that continues into transplant, even if the patient is on leave. It is possible that the added stress of these vocational demands contributes to depression after HCT. Indeed, recent study findings in a non-HCT cancer sample indicated that work-related stress is common, even for those on medical leave, during cancer treatment [38]. Patients who are working or in school, in some capacity, during the transplant process may require additional support and strategies on managing vocational demands and stress in the context of diminished medical and psychosocial reserves. This issue merits exploration in future studies.

Strengths of this study include use of reliable and validated self-report instruments at two time points in a well-established HCT program. Thus, we were able to determine several pre-transplant predictors associated with depression at T2. The relative weakness of the associations we found, however, indicates that there likely are other predisposing and precipitating factors which may contribute to depression post-HCT. The time course of depression symptoms from T1 to T2 likely reflects the uniqueness of HCT as compared to other cancer treatments. Due to the eligibility criteria for transplantation, it is possible that transplant patients report that they are more optimistic and hopeful at baseline (i.e., before their transplant) than other patients starting cancer treatment, who are often extremely distressed shortly following diagnosis and before they have started their initial treatments [8]. In addition, our T2 assessment occurred at 6–7 weeks post-HCT, which is a time when most patients are discharged from the hospital and have had several weeks to recover from deconditioning and other lingering acute symptoms, thus minimizing the direct effects of acute hospitalization on depressive symptoms. However, many adverse physical symptoms are often still present for HCT patients at 6 to 7 weeks post-HCT; it is possible that these impacted reports of patients’ mood. Future studies should examine the course and longitudinal predictors of depression at additional post-HCT time points. In our analysis, transplant type (allogeneic vs. autologous) was not a predictor of depression post-HCT, but this is an area that could be evaluated more thoroughly in future studies as these groups can differ significantly in medical course post-transplant.

An additional strength of this investigation was the electronic collection of data via touch-screen computer. The majority of participants in our study had familiarity with computers. Previous studies have shown that measurement of depression and symptom distress with touch screen computers is feasible and reliable [8]. Moreover, computer-administered symptom screens are clinically useful due to the ease of providing real-time scores and symptom information to clinicians, which can then be discussed with patients in a timely manner [21]. As our data were collected only via computer, however, findings should be replicated in populations less acquainted with this modality.

Participants in this study were relatively well-educated and affluent, and predominantly Caucasian and non-Hispanic/Latino. It is possible that samples of more diverse groups might have differing rates of, and risk factors for, depression post-HCT. Future studies should address these concerns. Finally, a broad gap in knowledge is to evaluate the effectiveness of brief interventions to address depression and enhance coping that can be implemented in the HCT setting.

Organizations such as the American College of Surgeons Commission on Cancer [20] are mandating that screening for distress (including depression, anxiety, and other psychosocial difficulties) occur at least once for all patients treated at participating cancer centers. However, findings from our study suggest that screening for distress just once at pre-transplant may be insufficient, and that 6–7 weeks post-HCT may be an important assessment point. Depression screening has been shown, in this study and others, to be feasible and acceptable by both patients and medical providers in a busy HCT clinic setting [8,11]. Our findings underscore the importance of assessment of emotional distress in HCT clinics at multiple time points during the course of treatment. Periodic depression screening at pivotal points along the trajectory of treatment and recovery provides opportunities for earlier referral and intervention for those patients either experiencing depression symptoms already or at risk for developing depression. Proactive efforts to manage depression in HCT patients may improve their overall recovery and post-transplant quality of life.

Table 3b.

Multivariate logistic regression predicting moderate or higher depression (PHQ-9 scores ≥10) at T2, adjusting for depression status at baseline. Bold text indicates significant predictors; italic text indicates predictors trending toward significance.

T1 Predictor	β	Odds Ratio	p value
Moderate or higher depression (PHQ-9 ≥ 10)	2.44	11.49	0.03
Working at T1a	0.65	1.92	0.07
Emotional function (EORTC QLQ-C30)	−0.034	0.96	<0.01

^a“Working at T1” is defined as working full-time, part-time, on full or partial medical leave, or being a student. “Not working” is defined as being unemployed or fully retired at T1.