Comparing the Retrospective Reports of Fatigue Using the Fatigue Severity Index With Daily Diary Ratings in Women Receiving Chemotherapy for Gynecologic Cancer

Abstract

Context

Fatigue, one of the most common side effects of chemotherapy, is typically assessed via retrospective recall (e.g., over the past week). It is unknown how such retrospective recall of fatigue correlates with daily ratings among people receiving chemotherapy.

Objectives

The current study compared fatigue recorded in daily diaries with retrospective ratings using the Fatigue Severity Inventory (FSI) in patients receiving chemotherapy for gynecologic cancer.

Methods

During the week before and the week after their first infusion of chemotherapy, patients completed daily diaries at 10am, 2pm, and 6pm and the FSI at the end of each week.

Results

FSI and diary ratings of peak, lowest, and average fatigue were significantly correlated (P-values<0.001). When peak, end, average, and variance diary ratings were regressed separately on the average FSI item, each was significant pre-chemotherapy (P-values<0.01) and post-chemotherapy (P-values<0.05). However, when entered into a stepwise regression model, only the average fatigue diary rating was retained, explaining 52% of the variance pre-chemotherapy and 54% of the variance post-chemotherapy average FSI item (P-values<0.001).

Conclusion

The FSI keyed to the past week accurately reflects daily ratings of fatigue among patients receiving chemotherapy. This study has important implications, as completing retrospective ratings of fatigue may be less burdensome for cancer patients than daily assessments.

Introduction

Fatigue is one of the most common symptoms among cancer patients.^1,2 When assessing fatigue, researchers and clinicians typically rely on patients’ retrospective recall. For example, the Fatigue Severity Inventory (FSI)^3,4 asks patients to retrospectively recall fatigue in the past week. Prior work has determined the FSI to be reliable and valid using comparisons with existing retrospective measures of fatigue.³ The FSI has not been validated with ratings of fatigue obtained using daily diaries. Understanding how fatigue assessed retrospectively using the FSI compares with daily ratings is important, as research suggests retrospective recalls of symptoms may be inflated compared to averaged diary ratings.⁵

Momentary self-reports assess “the self in the moment” and retrospective self-reports assess “the self through time.”⁶ Although conceptually distinct, to be valid, retrospective ratings should correspond to the mean of momentary ratings.⁶ Research on associations between momentary assessments and retrospective recall of pain and fatigue has been conducted in rheumatology patients^5,7,8 and has been based on “peak-end” heuristics, which suggest people attend to the most intense and most recent levels of symptoms.⁹ In studies of pain and fatigue in rheumatology patients, retrospectively recalled pain and fatigue were higher than diary ratings;⁵ however, whereas peak and end levels created a bias in recall of pain, they did not bias recall of fatigue.⁸ Higher variability in real-time pain has been associated with higher levels of pain recalled retrospectively, but was not examined for fatigue.⁷

Whether daily ratings of fatigue are associated with weekly recall is particularly relevant to patients receiving chemotherapy because symptoms “roller coaster” during treatment. Fatigue rises in the days following an infusion, then subsides prior to the next infusion.^10-12 Although momentary assessments of fatigue have been studied in cancer patients,^13-15 it has not been evaluated whether peak, end, or variability in daily ratings of fatigue bias retrospective recall in chemotherapy patients.

Our aim was to compare ratings of fatigue assessed though daily diaries with retrospective ratings using the FSI in a sample of gynecologic cancer patients in the week before and after their first cycle of chemotherapy. Based on previous research,⁵ we hypothesized that retrospective recall of fatigue would be higher than daily diary ratings. We also explored how much of the variance in the item of the FSI that assesses patients’ average fatigue level over the week was accounted for by peak, end, average, and variance in diary fatigue ratings

Methods

Participants and Procedures

Participants were recruited from September 2007 to July 2009 as part of a larger institutional review board-approved study examining side effects of platinum-based chemotherapy for gynecologic cancer on symptoms of sleep disturbances, depression, and fatigue.¹⁶ Eligibility criteria were: 18 years of age or older, scheduled to receive intravenous platinum-based chemotherapy for gynecologic cancer at Moffitt Cancer Center, no chemotherapy treatment for two or more months, free of documented or observable psychiatric or neurologic disorders that could interfere with participation, able to speak and read English, and able to provide informed consent.

Informed consent was obtained before the start of chemotherapy. Participants were provided with actigraphs, which they continuously wore on their non-dominant wrist one week before and one week after their first chemotherapy infusion. Actigraphs beeped daily at regularly scheduled intervals when participants were likely to be awake (i.e., 10am, 2pm, 6pm) to remind patients to complete the diaries. Questionnaires were completed on the day they received chemotherapy (pre-chemotherapy) and one week later (post-chemotherapy).

Eighty women agreed to participate. Three participants did not provide any data, one was missing diaries at both time points, four were missing diaries at one time point, two were missing both diaries and FSI at one time point, and two were missing FSI at one time point; 68 had complete data.

Measures

Demographic information was assessed via self-report and clinical information was obtained via medical chart review. Daily ratings of fatigue were assessed using diaries. At 10 am, 2 pm, and 6 pm during assessment days (i.e., three ratings per day for eight days), participants were asked to rate “How much fatigue are you experiencing now?” (0=no fatigue at all, 10=as fatigued as I could be). From this information, we identified the highest (i.e., peak), lowest, and last (i.e., end) diary fatigue ratings. We also calculated the mean of the 24 diary ratings to obtain an average diary rating; the standard deviation (SD) of diary ratings was calculated to obtain the variance in diary ratings for each participant.

Retrospective ratings of fatigue were measured using the FSI.³ For comparisons with diary data, the three FSI items of interest were: “Rate your level of fatigue on the day you felt most fatigued during the past week” (i.e., peak), “Rate your level of fatigue on the day you felt least fatigued during the past week” (i.e., lowest), and “Rate your level of fatigue on average during the past week” (i.e., average). FSI items were rated on the same 11-point scale (0=no fatigue at all, 10=as fatigued as I could be).

Analyses

To evaluate associations between diary and FSI ratings, correlations and 95% confidence limits were computed using Pearson correlation coefficients and Fisher’s r-to-z transformation.¹⁷ Discrepancies between diary and FSI ratings of peak, lowest, and average fatigue were determined by subtracting diary scores from FSI score, such that higher scores indicated the FSI ratings were higher than the diary ratings. Participants were then classified as having FSI scores higher, lower, or the same as diary ratings, and Chi square tests were used to test for differences from pre- to post-chemotherapy. Regression was used to determine whether peak, end, or variability in diary ratings were associated with the FSI item that asked participants to rate their average level of fatigue during the past week. Analyses were based on participants with complete data (N=68). Data were analyzed using in SAS v9.2 (SAS Institute Inc., Cary, NC). Results were considered significant if P<0.05 (two-tailed).

Results

Participants (N=68) were women (ages 33–87 years; mean±SD=64±11 years) receiving intravenous chemotherapy for stages I (21%), II (13%), III (51%), or IV (15%) gynecologic cancer. Median time since diagnosis was 79.5 days (range 13–4868, SD=1028) and time since surgery was 44 days (range 13–4868, SD=765). Most participants were being treated with a taxane (85%); 28% had experienced a disease recurrence and 19% had received radiotherapy. Most participants were non-Hispanic (97%) and Caucasian (96%); 26% had graduated college.

Means and SDs of FSI and diary ratings are presented in Table 1. Correlations between FSI and diaries for ratings of peak, lowest, and average fatigue were all significant (P-values<0.001; Table 2). Evaluating whether participants were statistically more likely to report higher, lower, or the same scores on the FSI from pre-chemotherapy to post-chemotherapy (Table 3), there were significant changes for the lowest fatigue rating (P=0.007), but no changes in concordance rates of peak or average fatigue ratings (P-values>0.05). More participants reported lower lowest fatigue rating on the FSI relative to diaries post-chemotherapy (18%) than pre-chemotherapy (4%) (χ²=8.91, P=0.003). There were no changes in the percentages of participants who reported higher or the same FSI scores compared to diary ratings of lowest fatigue pre- to post-chemotherapy (P-values>0.05).

Table 1.

Means, Standard Deviations, and Ranges of FSI and Daily Diary Ratings (N=68)

	FSI		Daily Diary
Variables	M (SD)	Range	M (SD)	Range
Pre-chemotherapy
Peak Fatigue	4.76 (2.77)	0 – 10	5.43 (2.82)	0 - 10
Lowest Fatigue	1.96 (1.89)	0 – 7	1.09 (1.63)	0 – 7
Average fatigue	3.00 (2.13)	0 – 9	2.83 (2.27)	0 - 9
Post-chemotherapy
Peak Fatigue	6.26 (2.65)	0 – 10	6.20 (2.64)	0 - 10
Lowest Fatigue	1.85 (1.98)	0 – 7	1.38 (1.82)	0 - 6
Average fatigue	3.81 (2.22)	0 – 8	3.72 (2.27)	0 - 8

FSI = Fatigue Severity Inventory.

Table 2.

Calculation and Test of Correlations Between FSI and Daily Diary Data (N=68)

		95% CI
Variables	r	Lower Limit	Upper Limit
Peak Fatigue
Pre-chemotherapy	0.69a	0.54	0.80
Post-chemotherapy	0.70 a	0.56	0.80
Lowest Fatigue
Pre-chemotherapy	0.67 a	0.51	0.78
Post-chemotherapy	0.63 a	0.46	0.75
Average fatigue
Pre-chemotherapy	0.72 a	0.58	0.82
Post-chemotherapy	0.73 a	0.60	0.83

FSI = Fatigue Severity Inventory; CI = confidence interval.

^aP<0.001.

Table 3.

Concordance and Discordance Rates Between FSI and Daily Diaries (N=68)

	Concordance	Pre-Chemotherapy Week	Post-Chemotherapy Week	χ2
Peak Fatigue (%)				2.46
	-	49	35
	0	29	46
	+	22	19
Lowest Fatigue (%)				14.17 a
	-	4	18
	0	49	41
	+	47	41
Average Fatigue (%)				5.13
	-	25	37
	0	37	29
	+	38	34

FSI = Fatigue Severity Inventory; - = FSI lower than daily diary ratings, 0 = FSI same as daily diary ratings, + = FSI higher than daily diary ratings.

^aP<0.01.

To evaluate how much of the variance in the retrospectively recalled average FSI item was accounted for by peak, end, average, and variance in diary ratings of fatigue, we ran univariate regression analyses. Analyses were conducted for pre- and post-chemotherapy ratings. Peak diary fatigue rating explained 36% (P<0.001) and 39% (P<0.001) of the variance in average FSI item pre-chemotherapy and post-chemotherapy, respectively. End diary rating of fatigue explained 28% (P<0.001) and 34% (P<0.001) of the variance in average FSI item pre-chemotherapy and post-chemotherapy, respectively. Average diary ratings explained 52% (P<0.001) and 54% (P<0.001) of the variance in average FSI item pre-chemotherapy and post-chemotherapy, respectively. Variance in diary fatigue ratings explained 9% of the variance in average FSI item both pre-chemotherapy (P=0.010) and post-chemotherapy (P=0.011) (Table 4).

Table 4.

Univariate Regression Analyses for Diary Rating of Fatigue on Average FSI

Variable	R2	β	SE	t
Pre-chemotherapy
Peak diary rating	0.36	0.45	0.07	6.14 a
End diary rating	0.28	0.39	0.08	5.12 a
Average diary rating	0.52	0.70	0.08	8.46 a
Variance in diary rating	0.09	0.84	0.32	2.66 b
Post-chemotherapy
Peak diary rating	0.39	0.52	0.08	6.45 a
End diary rating	0.34	0.43	0.07	5.89 a
Average diary rating	0.54	0.72	0.08	8.80 a
Variance in diary rating	0.09	0.87	0.33	2.61 c

Diary = Daily Diary; FSI = Fatigue Severity Inventory; R²=squared multiple correlation; β=standardized beta weight; SE=standard error.

Note: In stepwise regression models, only average diary rating was retained both pre- and post-chemotherapy.

^aP<0.001.

^bP<0.01.

^cP<0.05.

To determine which of the diary ratings accounted for the greatest amount of variability in average FSI item in multivariate analysis, peak, end, average, and variance ratings were entered into a stepwise regression model. Only average diary rating was retained in the pre- (R²=0.52, β=0.70, P<0.001) and post-chemotherapy (R²=0.54, β=0.72, P< 0.001) model.

Discussion

Overall, retrospective ratings of fatigue as assessed by the FSI showed good concordance with diary ratings. Diary ratings of fatigue were significantly correlated with their corresponding FSI scores. Evaluating how much of the variance in the average FSI item was accounted for by peak, end, average, and variance in diary ratings of fatigue, in univariate regression analyses each diary rating explained a significant proportion of the variance in average FSI item. However, when entered into a stepwise regression model, only the average of the 24 weekly diary ratings was retained in the model.

Prior research demonstrated retrospective ratings of pain and fatigue are inflated related to averaged real-time ratings.⁵ The current study found significant correlations between the peak fatigue level, lowest fatigue level, and average fatigue level recorded in daily diaries and the corresponding ratings from the retrospectively recalled FSI. The pattern of significant correlations was the same pre- and post-chemotherapy. Concordance rates of diary and FSI ratings were consistent pre- to post-chemotherapy for peak and average ratings. However, participants were more likely to report lower levels of lowest fatigue on the FSI relative to diaries post-chemotherapy than pre-chemotherapy. Because fatigue levels are higher post-chemotherapy than pre-chemotherapy (as we reported previously¹⁶), the higher fatigue post-treatment may bias patients’ retrospective recall of their lowest fatigue level. Alternatively, patients’ lowest levels of fatigue may not have occurred during the times they were asked to complete the dairy.

Prior work with rheumatology patients found more variability in momentary measures of pain predicted retrospectively recalled pain, but fatigue was not assessed in that study.⁷ A study of rheumatology patients that included measures of both pain and fatigue found peak and end ratings of pain contributed to retrospectively recalled pain; however, peak and end ratings were not associated with retrospectively recalled fatigue.⁸ In the current study, univariate analyses indicated that higher peak diary rating, higher end diary rating, higher average diary ratings, and greater variance in diary ratings were each associated with higher average FSI item. However, using stepwise regression, because of collinearity only the average daily fatigue rating was retained in the model; it explained 52% of the variance in pre-chemotherapy average FSI item and 54% of the variance in post-chemotherapy average FSI item. Our finding that the mean of diary ratings corresponds to retrospective recall lends support to the validity of the FSI.⁶

Our data suggest that the FSI is an acceptable substitute for diary ratings of fatigue in patients receiving chemotherapy. Additional support for using a weekly measure of fatigue comes from a study of cancer patients that found the seven-day Functional Assessment of Chronic Illness Therapy-Fatigue was more precise and informative than the four-day measure.¹⁸ As receiving chemotherapy can be stressful for patients, it is important to identify symptom measures that minimize patient burden.

Study strengths include the use of a fatigue measure designed for cancer patients and a diary format using the same 0-10 rating scale, focus on gynecologic cancer patients beginning chemotherapy (for whom fatigue is likely to be a significant problem¹⁹), and inclusion of pre- and post-chemotherapy assessments. Limitations include the lack of a time and date stamp for diaries. Prior research suggests some patients may complete paper assessments at a later time,²⁰ leading to greater congruence between diary ratings and retrospective recall. However, our actigraph beeped to remind patients to complete diaries, which may have increased the likelihood of real-time ratings. Momentary assessments using electronic devices have higher compliance rates²⁰ and have been useful at assessing fatigue in people receiving hematopoietic stem cell transplantation.¹³ Future researchers should replicate these results using an electronic device that documents the exact time of rating.

FSI scores keyed to the past week are a good approximation of fatigue ratings reported in daily diaries in the week prior to and following chemotherapy. Findings support the use of this retrospective measure of fatigue for patients receiving chemotherapy, which may be less burdensome than daily ratings.