Page 242 - 2019_05-HaematologicaMondo-web

N.S. Majhail et al.
Table 2. Analysis for primary and secondary end points. End point*
Mean (Standard Deviation)
Estimate (Standard Error)#
-0.034 (0.028)
0.123 (0.042)
-0.018 (0.013)
0.014 (0.010)
-0.368 (0.638)
-8.907 (3.009)
P# 0.223
0.004
0.182
0.149
0.565
0.003
Confidence in survivorship information†
Cancer and treatment distress‡
Knowledge of transplant exposures†
Health care utilization†
SF12: physical component summary†
SF12: mental component summary†
SCP (N=199)
Routine care (N=199)
SCP (N=199)
Routine care (N=198)
SCP (N=200)
Routine care (N=198)
SCP (N=200)
Routine care (N=198)
SCP (N=200)
Routine care (N=198)
SCP (N=200)
Routine care (N=198)
Baseline
1.44 (0.34)
1.40 (0.38)
0.91 (0.61)
0.91 (0.64)
0.86 (0.18)
0.88 (0.15)
0.80 (0.14)
0.80 (0.14)
46.1 (10.3)
46.0 (9.8)
53.9 (7.6)
53.9 (7.9)
6-months
1.50 (0.34)
1.44 (0.39)
0.78 (0.59)
0.91 (0.69)
0.87 (0.16)
0.86 (0.16)
0.80 (0.15)
0.82 (0.13)
46.2 (10.6)
45.8 (10.1)
54.7 (7.0)
53.4 (8.8)
SCP: Survivorship Care Plan; N: number; SF12: Short Form 12. *N: number who completed both baseline and 6-month assessments. #Estimate and P-value based on analysis of covariance model with center-level random effects where any differences between the treatment groups were measured after adjustment for patients’ baseline measurement; where applicable, estimates were adjusted for demographic variables and/or interactions (see Methods section). †Higher score better. ‡Lower score better.
SCP while patients on the control arm only received an informa- tive letter. SRG then contacted all enrolled patients by phone between 7-28 days of mailing study materials to conduct a health literacy assessment using the Newest Vital Sign.44 During this con- tact, patients on the SCP arm were given the opportunity to address any questions about the content or use of their SCP. No further contact was made till the 6-month phone survey. The Confidence in Survivorship Information (CSI) was the primary end point (Online Supplementary Table S1).45 Secondary end points focused on Cancer and Treatment Distress (CTXD),20,46 as well as measures of Knowledge of Transplant Exposures, Health Care Utilization,26 and HRQOL using the SF-12.47 Patients on the inter- vention arm also received a 12-item assessment for qualitative feedback on SCP utilization. Sample size calculations were per- formed using a standard error formula that allowed for possible variability in treatment effect across centers and considered dropouts from baseline to 6 months. Our enrollment goal was 495 patients, which yielded adequate power to detect standardized effect sizes of ≥0.3, which are considered to be clinically meaning- ful, and anticipated a 10% drop-off from baseline to 6 months. An intention-to-treat approach was followed for analysis. A mixed model with center-level random effects and a fixed treatment effect was used to test whether there was a change in baseline and 6-month response between the treatment and control groups for the primary and secondary end points. The 6-month assessment was used as a response variable and the baseline assessment was used as an explanatory variable in the regression models. If a treat- ment effect was observed, we further evaluated whether the effect was modified by demographic variables or any interactions between variables.
Further details are available in the Online Supplementary Appendix.
Results
Patients' characteristics
Among the 495 patients enrolled, 458 completed the baseline survey and were randomized (SCP=231, con- trol=227); 200 (87%) and 199 (88%) completed 6-month assessments, respectively (Figure 2). The main reasons for dropout were loss to follow up or patients not eligible for follow-up assessment due to interim disease relapse or
progression. A greater proportion of patients who com- pleted the 6-month assessment were White and reported higher health literacy scores; otherwise there were no sig- nificant differences in the demographic characteristics between patients who did and those who did not com- plete the 6-month assessments (Online Supplementary Table S2). Patients' and transplant characteristics (including health literacy scores) were well balanced between the two arms, except for gender (49% males in SCP compared to 60% in controls; P=0.01) (Table 1). Median age was 59 years in both arms and enrolled patients were predomi- nantly White (96% SCP and 92% controls). In the SCP and control arms, 48% and 44% had received allogeneic HCT; among allogeneic HCT recipients 63% and 67% had a history of acute GvHD, and 60% and 66% had a history of chronic GvHD, respectively.
Analyses of primary and secondary end points
Of the 458 patients randomized to the two arms, 399 completed 6-month assessments, including 398 who com- pleted pre- and post-measurements for the primary end point (Table 2). We did not find any association between the SCP intervention and change in CSI scores from base- line to 6-months (P=0.223), even after assessing for the effect of demographic factors and interactions. However, we did observe a significant decrease in CTXD scores (P=0.004) and an increase in HRQOL Mental Component Summary (MCS) scores as assessed by SF-12 (P=0.003) among patients randomized to the SCP arm. There was no association between the SCP intervention and other secondary end points.
We further assessed the effect of demographic variables and interactions for the end points of CTXD and SF-12 MCS, where a significant treatment effect was observed. Age was significantly associated with CTXD scores (regression estimate -0.006, standard error 0.002; P=0.001), with lower distress among older patients. However, there was no significant interaction between age and SCP intervention and adjustment for age did not modify the treatment effect. The decrease in CTXD score for the SCP arm was independent of gender, health litera- cy, diagnosis, transplant type, and GvHD status (including acute and chronic GvHD). We also found a similar effect
1088
haematologica | 2019; 104(5)