Reductions in anaemia and fatigue are associated with improvements in productivity in cancer patients receiving chemotherapy

Pharmacoeconomics 2005; 23 (5): 505-514ORIGINAL RESEARCH ARTICLE 1170-7690/05/0005-0505/$34.95/0

© 2005 Adis Data Information BV. All rights reserved.

Reductions in Anaemia and Fatigueare Associated with Improvements inProductivity in Cancer PatientsReceiving ChemotherapyErnst Berndt,1 Joel Kallich,2 Anne McDermott,3 Xiao Xu,3 Howard Lee4 andJohn Glaspy4

1 Massachusetts Institute of Technology Sloan School of Management, Cambridge,Massachusetts, USA

2 Amgen, Inc., Thousand Oaks, California, USA3 Covance Health Economics and Outcomes Services Inc., Gaithersburg, Maryland, USA4 School of Medicine, University of California Los Angeles, Los Angeles, California, USA

Objective: Cancer-related anaemia is associated with fatigue that adverselyAbstractaffects patients’ everyday functioning and wellbeing. We explore the impact offatigue on patient productivity and caregiver burden.Methods: The analyses are based on data from a randomised, open-label,active-controlled, dose-finding trial of darbepoetin alfa among solid-tumour can-cer patients with anaemia, who are receiving chemotherapy. Fatigue is assessedwith the Functional Assessment of Cancer Therapy (FACT)-Fatigue subscalescore. Productivity and caregiver outcomes include time (hours) missed fromusual activities, amount of assistance (hours) needed from others, overall ability toperform desired activities and ability to perform family responsibilities. Theseoutcomes are assessed at baseline and the end of the 12-week treatment period.ANOVA and linear regression models are used to evaluate associations.Results: Patients (n = 300) were aged 61 years on average, with a mean (SD)baseline haemoglobin of 9.9 (0.9) g/dL. FACT-Fatigue subscale score improve-ments were significantly (p = 0.003) associated with haemoglobin improvements.Over a 2-week period, after controlling for age, sex and disease progression,one-point improvements in FACT-Fatigue subscale scores corresponded to a1-hour (95% CI 0.5, 1.5) gain in productive time, 0.7-hour (95% CI 0.4, 1.0)reduction in caregiver time and 1.6% (95% CI 1.4, 1.7) improvement in overallactivity.Conclusions: Reducing fatigue is associated with gains in productive time,reductions in caregiver burden and enhanced ability to perform activities. Theseoutcomes may have broader implications for patients’ wellbeing and for thesocietal impact of cancer-related fatigue and anaemia.

506 Berndt et al.

Anaemia, an important determinant of fatigue, is of any published studies reporting the impact ofa frequent adverse effect of cancer and its treat- cancer patients’ fatigue on caregiver requirements,ment.[1-5] Estimates of the prevalence of fatigue or of any prospective data regarding the impact ofamong cancer patients receiving chemotherapy anaemia treatment on any productivity outcome.range from 59% to 82%.[6] Recently, the Fatigue To explore the impact of fatigue on productivityCoalition in the US conducted a national survey of and caregiver burden, a secondary analysis of dataindividuals with cancer who had undergone chemo- from a dose-finding study of darbepoetin alfatherapy.[7] Seventy-six percent of respondents re- (Aransep® 1: Amgen, Inc., Thousand Oaks, CA,ported experiencing fatigue at least a few days each USA) was conducted among anaemic patients withmonth during their treatment. The most common non-haematologic malignancies undergoing chemo-physical manifestations of fatigue were diminished therapy. These data were used to examine the rela-energy (81%), the need to slow down from a normal tionship between anaemia and fatigue, as well as thepace (81%), sluggishness or tiredness (79%) and the impact of fatigue on patients’ productivity andneed for more sleep or rest (78%). These symptoms caregiver burden. Four outcomes were evaluated: (i)had a significant adverse impact on everyday activi- time missed from work or usual activities; (ii)ties, with 88% of respondents reporting that their amount of assistance needed from others to performfatigue caused them to alter their daily routine and usual activities (caregiver time); (iii) overall ability75% of those employed reporting that they changed to perform desired activities; and (iv) ability totheir work status because of their fatigue.[7] Guide- perform usual family roles and responsibilities.lines for the management of fatigue include advisingpatients about the likelihood of experiencing fatigue Methodsduring the course of their treatment and suggestingthat patients conserve their energy where possible Data for these analyses were collected during aand seek assistance from others in performing their randomised, open-label, active-controlled, dose-usual activities.[8] finding study of darbepoetin alfa administered sub-

cutaneously every 1 or 2 weeks during cancer chem-Few studies have documented the specific impactotherapy.[11] Anaemic (haemoglobin ≤11 g/dL), sol-of anaemia-related fatigue on cancer patients’ pro-id-tumour cancer patients were randomised to re-ductivity, namely missed time from work or otherceive darbepoetin alfa or epoetin alfa (Procrit®;usual activities and the need for assistance fromOrtho Biotech, Raritan, NJ, USA) for 12 weeks. Theothers in completing usual activities. In the Fatiguedoses and schedules of darbepoetin alfa studied inCoalition survey, individuals receiving chemothera-this trial were 0.5, 1.0, 1.5, 2.25, 4.5, 6.0 and 8.0 μg/py reported using an average of 4.2 sick or vacationkg/week and 3.0, 5.0, 7.0 and 9.0 μg/kg every 2days per month as a result of their fatigue during,weeks. The dosages and schedules of epoetin alfaand shortly after, their cancer treatment.[7] In a retro-were 150 U/kg three times per week and 40 000 U/spective analysis of medical claims data and disabil-week, with dose increases permitted for patientsity records from a large employer, Lyman et al.[9]

with inadequate haemoglobin responses. The analy-found that anaemic cancer patients missed an addi-ses reported here are based on pooled data from alltional 5.3 days from work in the 2 months followingtreatment regimens.anaemia diagnosis relative to non-anaemic cancer

patients. In another retrospective analysis of an em- The major inclusion criteria were anaemia asployer database, anaemic cancer patients averaged 5 indicated by a haemoglobin concentration ≤11 g/dL;more short-term disability leave days per month no evidence of iron, folate or vitamin B12 (cyanoco-relative to non-anaemic cancer patients and in- balamin) deficiency; adequate renal and liver func-creased medical costs.[10] The authors are not aware tion; Eastern Cooperative Oncology Group (ECOG)

1 The use of trade names is for product identification purposes only and does not imply endorsement.

© 2005 Adis Data Information BV. All rights reserved. Pharmacoeconomics 2005; 23 (5)

Anaemia-Related Fatigue and Productivity 507

performance status of 0–2; currently receiving cyc- and fatigue (table I). The productive time lost andlic chemotherapy; no erythropoietic therapy within caregiver time questions were as follows.8 weeks of randomisation; and no more than two red • During the past 2 weeks, for about how manyblood cell transfusions within 2 weeks of randomis- hours were you unable to work because of youration. The study was conducted in the US. The condition (productive time lost)?institutional review boards of the participating cen- • During the past 2 weeks, for about how manytres approved the protocol and all patients gave hours did someone (family, friend or other per-written informed consent before any study-specific son) care for you or help you with your dailyprocedures were carried out. Patients aged <18 years tasks (e.g. shopping, child care or cleaningwere excluded from the study. house) [caregiver time]?

To assess patient-perceived functioning and All patients were instructed to answer the ques-wellbeing, patient-reported health-related quality of tion about productive time lost, independent of theirlife (HR-QOL) and productivity outcomes were in- employment status. Because 76% of patients includ-cluded as exploratory endpoints in the trial. Patients ed in the analyses were not employed at the time ofcompleted the questionnaire before the administra- the study, the productive time lost outcome reflectstion of the study drug (darbepoetin alfa or epoetin time missed from a broad set of activities that pa-alfa) at the beginning of each chemotherapy cycle tients consider their usual responsibilities, such asand at the end of the treatment period. The outcomes housework and volunteer work. Because change isfrom the questionnaire included in these analyses calculated as the follow-up number of hours less theare described below. The analyses were limited to baseline hours, negative change represents gains inpatients in the study who completed at least 4 weeks productivity and reductions in caregiver require-of treatment and reported the outcomes at baseline ments (that is, positive outcomes) in the analyses.and at least once after 4 weeks of treatment. Base- Patients were asked to rate their ability to doline and last observed values were used to assess everything they would like to have done during thechange over time, where the last observed value past 2 weeks on a numeric rating scale (NRS) from 0occurred any time after week 4 and before the end of to 10, where 0 represents ‘able to do nothing’ and 10the 12-week treatment period. represents ‘able to do everything’. To facilitate in-

terpretation, the 0–10 scale was transformed to aPatient-Reported Outcomes 0–100 scale. As higher ratings indicate greater abili-

ty to perform desired activities, positive changescores represent improvements.The outcomes in these analyses included mea-

sures of productive time lost, caregiver time, ability The ability to meet one’s family needs was as-to do usual activities, ability to meet family needs sessed using a single question from the Functional

Table I. Study outcome measures

Study outcome Descriptiona Scale Direction

Productive time lost Self-reported time unable to work (paid or unpaid) because of health Hours –condition

Caregiver time Self-reported amount of care or help received from others Hours –

NRS activity Numeric rating scale of ability to perform usual activities 0–100 +

Ability to meet family needs Single item assessing trouble meeting family needs because of physical 0–4 –condition

Fatigue FACT-Fatigue subscale (includes 13 items, such as ‘I feel fatigued’ and ‘I 0–52 +feel weak all over’)

a The reference period for all study variables was ‘during the past 2 weeks’.

FACT = Functional Assessment of Cancer Therapy; NRS = numeric rating scale; + higher values indicate better outcomes (e.g. greaterability to do usual activities); – higher values indicate poorer outcomes (e.g. more productive time lost).


508 Berndt et al.

Assessment of Cancer Treatment (FACT)-Anaemia considered: age, sex, baseline performance status,survey’s physical wellbeing subscale (see following tumour type and end-of-study disease progression.paragraph). For this question, patients were asked to Baseline performance status was measured using thereport how true the following statement was for ECOG rating. For inclusion in the study, patientsthem: ‘Because of my physical condition, I have were required to have an ECOG rating of 0, 1 or 2trouble meeting the needs of my family’. Response defined as follows: fully active, able to carry out alloptions included: not at all true (0), a little bit true pre-disease performance without restriction (0); re-(1), somewhat true (2), quite a bit true (3) and very stricted in physically strenuous activity, but ambula-much true (4). This outcome was analysed as a tory and able to carry out work of a light or sedenta-categorical variable, with change classified as ‘im- ry nature (1); and ambulatory and capable of all self-proved’, ‘stayed the same’ or ‘worsened’. care, but unable to carry out any work activities (2).

Tumour type was categorised as breast, gastrointes-The FACT-Anaemia assessment was included intinal, lung, gynaecologic or other. Disease progres-this study. It measures general HR-QOL areassion was defined as complete response, partial re-(physical, social, emotional and functional wellbe-sponse, stable disease or progressive disease, asing) and fatigue- and anaemia-specific concerns.[12]

reported by the investigator.In these analyses, the focus was on the Fatiguesubscale, which includes 13 statements, such as ‘Ifeel fatigued’ and ‘I feel weak all over’. Patients Statistical Methodsrespond to each statement on a five-point scale from

To describe the study sample, means, standard0 (not at all true) to 4 (very much true) based on theirdeviations and ranges of continuous variables andexperience during the past 2 weeks. For the calcula-frequency distributions of categorical variables weretion of the subscale score, many of the questioncomputed. To evaluate the association between fa-response categories are reversed so that when alltigue and haemoglobin, the authors used anquestions are summed to create the total score,ANOVA model of change in FACT-Fatigue sub-which ranges from 0 to 52, higher scores representscale scores as a function of haemoglobin change:less fatigue. When there are <50% of questions<0 g/dL, 0 to <2 g/dL and ≥2 g/dL. To evaluate themissing, the subscale score is computed by sum-association between continuous outcomes (produc-ming the non-missing questions, multiplying by 13tive time lost, caregiver time, NRS activity) and(the total number of questions in the scale) andfatigue, mean change scores across levels of fatiguedividing by the number of non-missing questions.improvement were compared using ANOVA mod-The subscale score was recorded as missing whenels. Clinically meaningful improvement in fatigue≥50% of questions were not answered.[13] Thewas defined as a change in FACT-Fatigue subscaleFACT-Fatigue subscale has been used extensivelyscores of three or more points. To evaluate thein studies of cancer patients with anaemia and hasassociation between the categorical outcome (abilitybeen administered to a general sample population into meet family needs) and fatigue, we used Chi-the US to provide normative data.[14] Based on asquare tests. Statistical significance was evaluated atprevious study, we classified patients as having clin-the 0.05 level.ically meaningful improvements in fatigue if they

To explore potential explanatory variables, linearhad an improvement of three or more points inregression models of change in the continuous out-FACT-Fatigue subscale scores.[15]

comes were analysed. Specifically, change wasmodelled in each outcome as a function of change inExplanatory VariablesFACT-Fatigue subscale scores, age, sex and end-of-

When evaluating the relationship between fa- study disease progression (progressive versus stabletigue and the productivity and caregiver burden out- or responsive). To allow for regression to the mean,comes, several potential explanatory variables were baseline levels of the outcome and FACT-Fatigue



subscale scores were also included in each model.The statistical significance of parameter estimateswas evaluated using t-tests. The parameter esti-mates, associated 95% confidence intervals and theestimated variation explained by each model (R2)

are also reported. The authors note that change inoutcomes was relatively normally distributed andthat an analysis of regression diagnostics from themodels did not reveal any major violations of theunderlying statistical assumptions of the model.

Results

Of the 389 patients who completed 4 weeks oftreatment, 300 (77%) were included in the analysissample (table II). These patients were aged 61 years,on average, and were predominantly female (70%)and White (80%). The predominant tumour typeswere breast (31%), gastrointestinal (21%) and lung(19%). With respect to performance status, mostpatients were either fully active (34%) or restrictedin physically strenuous activities but otherwise am-bulatory (59%). The mean baseline haemoglobinwas 9.9 g/dL, with a range of 6.2–12.2 g/dL. Thesepatients were similar to patients in the study whowere not included in the analyses because of missingoutcomes data (n = 89). In a logistic regressionmodel of inclusion in the analysis sample as a func-tion of baseline characteristics, only age was statisti-cally significant (p = 0.012). Patients included in theanalysis sample were 4.2 years younger, on average,than the 89 study patients who were not included.

At baseline, the mean (SD) FACT-Fatigue sub-scale score was 25.8 (12.5). Patients reported losinga mean (SD) of 20.9 (36.4) hours of productive timeand requiring 16.2 (33.2) hours of caregiving fromothers during the previous 2 weeks. This corre-sponds to an approximate average of 1.5 hours perday of lost productivity and 1.2 hours per day ofcaregiving. Patients’ mean (SD) NRS activity ratingwas 45.8 (25.4). Thus, on average, patients reportedtheir overall activity level as slightly below the

Table II. Baseline characteristics of the analysis sample

Characteristic Analysis Patients with missingsamplea outcomes datab

(n = 300) (n = 89)

Age (years)

mean (SD) 60.8 (12.3) 65 (12.3)

range 20–91 27–87

Age group, n (%)

≥65 years 130 (43.3) 52 (58.4)

≥75 years 36 (12.0) 23 (25.8)

Sex, n (%)

male 91 (30.3) 32 (36.0)

female 209 (69.7) 57 (64.0)

Race, n (%)

Asian 10 (3.3) 3 (3.4)

Black 24 (8.0) 13 (14.6)

Hispanic 24 (8.0) 8 (9.0)

White 239 (79.7) 64 (71.9)

other 3 (1.0) 1 (1.1)

Haemoglobin g/dL

mean (SD) 9.9 (0.9) 9.7 (0.8)

range 6.2–12.2 7.7–11.2

Haemoglobin, n (%)

<10 g/dL 149 (49.7) 50 (56.2)

10–10.9 g/dL 110 (36.7) 35 (39.3)

≥11 g/dL 41 (13.7) 4 (4.5)

Tumour type, n (%)

breast 92 (30.7) 19 (21.4)

lung 57 (19.0) 25 (28.1)

gastrointestinal 64 (21.3) 23 (25.8)

gynaecologic 39 (13.0) 9 (10.1)

other 48 (16.0) 13 (14.6)

ECOG, rating n (%)c

0 100 (33.8) 19 (21.6)

1 173 (58.5) 57 (64.8)

2 21 (7.1) 11 (12.5)

3 2 (0.7) 1 (1.1)

a Patients (n = 300) were included in the analysis if they wererandomised, received at least 4 weeks of study treatment andreported outcomes at baseline and at least once after 4weeks of treatment.

b Patients (n = 89) who were randomised and received at least4 weeks of study treatment, but did not report outcomes atbaseline or at least once after 4 weeks of treatment.

c Four patients were missing ECOG results in the analysissample and one patient was missing ECOG results in thesample with missing outcomes data.

ECOG = Eastern Cooperative Oncology Group.

middle of the range of the 100-point scale, where100 represents the ability to perform all desired reported experiencing at least some difficulty meet-activities. Furthermore, most of the patients (78%) ing family needs owing to their physical health.


510 Berndt et al.

served, but this difference did not reach statisticalsignificance (p = 0.063). With respect to NRSactivity ratings, patients with clinically meaningfulimprovements in fatigue reported 19.2% enhance-ments in their ability to perform everyday activitiesrelative to 9.4% decrements among patients withoutimproved fatigue (p < 0.0001) over 12 weeks oftreatment. Furthermore, a higher percentage of pa-tients with improved fatigue reported statisticallysignificant improvements in their ability to meettheir family’s needs (54.1% vs 27.7%; p < 0.001).

Table III. Mean change in FACT-Fatigue subscale scores by levelof change in haemoglobin

Change in Change in FACT-Fatigue subscale score

haemoglobin category na mean (95% CI)b

<0 g/dL 55 –1.1 (–4.3, 2.1)

0 to <2 g/dL 121 3.1 (1.0, 5.2)

≥2 g/dL 121 5.5 (3.4, 7.7)

a One patient was missing change in the FACT-Fatiguesubscale score and two patients were missing change inhaemoglobin values from baseline to the end of treatment.

b p = 0.0032 (one-way ANOVA model testing overall statisticalsignificance of mean change in FACT-Fatigue subscale scoreby level of change in haemoglobin).

FACT = Functional Assessment of Cancer Therapy.Finally, to explore potential explanatory vari-

ables, the authors modelled change in the continu-Over time (change from baseline to the end of ous outcomes as a function of change in FACT-

treatment), patients included in these analyses re- Fatigue subscale scores and controlled for age, sexported mean (SD) improvements in FACT-Fatigue and end-of-study disease progression. We exploredsubscale scores of 3.2 (12.3). Improvements of at the correlation among these potential explanatoryleast three points on the FACT-Fatigue subscale are variables to mitigate effects of multicollinearity.considered clinically meaningful.[15] Half (50.2%) The correlation coefficients ranged from 0.00 toof the patients experienced this level of improve- 0.15 (p-values not significant). To allow for regres-ment. sion to the mean, we also added baseline levels of

FACT-Fatigue subscale scores and productivity out-Changes in FACT-Fatigue subscale scores werecomes as regressors.significantly (p = 0.003) associated with level of

change in haemoglobin (table III). Patients with a Change in FACT-Fatigue subscale scores washaemoglobin improvement of at least 2 g/dL report- significantly associated (p < 0.001) with change ined a mean (95% CI) improvement in FACT-Fatigue each of the productivity outcomes (table V). Aftersubscale scores of 5.5 (3.4, 7.7), while patients with controlling for age, sex and disease progression,a decline in haemoglobin reported a mean decline of over a 2-week period, one-point improvements in–1.1 (–4.3, 2.1) in FACT-Fatigue subscale scores. FACT-Fatigue subscale scores corresponded to aAmong the 121 patients who had a ≥2 g/dL im- 1-hour (95% CI 0.5, 1.5) gain in productive time,provement in haemoglobin, 72 (59.5%) also had an 0.7-hour (95% CI 0.4, 1.0) reduction in caregiverimprovement of more than three or more points in time and 1.6% (95% CI 1.4, 1.7) improvement intheir FACT-Fatigue subscale score. overall activity. Age, sex and disease progression

were not significantly associated with productiveClinically meaningful improvements in FACT-time lost or caregiver time. In the model of changeFatigue subscale scores (three or more points) werein NRS activity rating, however, females (n = 165)significantly associated with improvements in threereported 6.4% (95% CI 2.2, 10.6) greater improve-of the four productivity outcomes (table IV). Pa-ments in activity level (p = 0.003) relative to malestients with clinically meaningful improvements in(n = 77), and patients with progressive diseasefatigue reported reductions in caregiver burden,(n = 80) reported 4.8% (95% CI 0.6, 9.0) greaterwhile the converse was true for patients who did notreductions in activity level (p = 0.026) relative toexperience improved fatigue. The magnitude of thepatients with stable (n = 95) or responsive diseasedifference in mean reported changes between these(n = 67). Adjusted R2 values for productive timepatient subgroups was 11 hours of caregiver timelost, caregiver time and NRS activity models were(p = 0.007) over a 2-week period. Improvements in0.37, 0.14 and 0.73, respectively.productive time of almost 11 hours were also ob-



Discussion (p < 0.01) except for gains in productive time. Theimpact of fatigue on these outcomes highlights theimportance of current practice guidelines for screen-Cancer patients with anaemia included in theseing, continuous monitoring and prompt treatment ofanalyses reported levels of fatigue (mean baselinefatigue.[18] Patients with improved fatigue reportedFACT-Fatigue subscale scores of 26) that were sub-11 more hours of productive time and 11 fewerstantially worse than average levels observed inhours of caregiver time over a 2-week period rela-cancer patients without anaemia (40) and in thetive to those who did not experience improved fa-general population (44).[14] Over an observation pe-tigue. The difference in productive time was basedriod during which they were receiving chemothera-on 5-hour gains among patients with improved fa-py and treatment for anaemia, these patients exper-tigue and 6-hour losses among patients who did notienced levels of fatigue improvement that were con-experience fatigue improvement. The difference insidered clinically meaningful.[15] Clinical evidencecaregiver time was based on 2-hour reductionsfor the superior amelioration of anaemia-related fa-among those with improved fatigue and 9-hour in-tigue with erythropoeitic growth factors comparedcreases among those without improved fatigue. Forwith transfusions alone has been shown in threeevery hour of lost productivity in the fatigued group,recently completed randomised, placebo-controlled,1.5 hours of caregiving were required. Thus,double-blind clinical trials.[3,16,17] Furthermore, con-caregiver time does not appear to be a one-for-onesistent with previous clinical trials,[2-5] improve-substitute for reductions in patient productivity; fail-ments in haemoglobin observed in this study wereure to mitigate patients’ fatigue appears to imposesignificantly associated with improvements in fa-an even greater burden on caregivers.tigue.

These analyses also demonstrated that clinically With respect to patients’ overall ability to per-meaningful improvement in fatigue was substantial- form their usual activities, those experiencing clini-ly associated with gains in productive time, reduc- cally meaningful improvements in fatigue reportedtions in caregiver requirements and enhanced ability approximately 30% greater improvements thanto perform usual activities and family responsibili- those who did not have a clinically meaningfulties, and the associations are statistically significant improvement. Similar rating scales of activity have

Table IV. Change in productivity outcomes by change in fatigue

Productivity outcomes No clinically meaningful improvement Clinically meaningful improvement inin FACT-Fatigue subscale scores FACT-Fatigue subscale scores

(Δ three or more points) (Δ three or more points)

Continuous outcomesa n mean Δ (95% CI) n mean Δ (95% CI) p-value

Productive time lost (h) 112 6.3 (–3.5, 16.2) 103 –4.5 (–10.5, 1.4) 0.0625

Caregiver time (h) 149 9.1 (2.4, 15.7) 150 –1.9 (–6.2, 2.4) 0.0068

NRS activity 145 –9.4 (–13.3, –5.6) 148 19.2 (15.0, 23.4) <0.0001

Categorical outcomeb n % (95% CI) n % (95% CI) p-value

Ability to meet family needs 0.0006c

Improved 26 27.7 (18.7, 36.7) 53 54.1 (44.2, 64.0)

Stayed the same 39 41.5 (31.5, 51.5) 30 30.6 (21.5, 39.7)

Worsened 29 30.9 (21.6, 40.2) 15 15.3 (8.2, 22.4)

a One patient was missing change in the FACT-Fatigue subscale score, 84 patients were missing change in productive time lost andsix patients were missing change in NRS activity scores from baseline to the end of treatment.

b One patient was missing change in the FACT-Fatigue subscale score and 107 patients were missing change in the family needsassessment from baseline to the end of treatment.

c p-Value from Chi-square test indicating the overall association between the ability to meet family needs and clinically meaningfulimprovement in FACT-fatigue subscale scores.

FACT = Functional Assessment of Cancer Therapy; NRS = numeric rating scale; Δ = change.


512 Berndt et al.

been used in previous studies of cancer patients withanaemia and have demonstrated significant associa-tions with haemoglobin improvement.[3-5] The anal-yses presented here provide a direct association withfatigue. In the multivariate analysis, clinicallymeaningful improvements in fatigue were associat-ed with a 5% improvement in activity level. As such,5% may represent a meaningful improvement inpatients’ overall ability to perform activities and cantherefore serve as a benchmark in future applica-tions of the measure.

In the multivariate analyses of productive timelost and caregiver time, clinically meaningful im-provements in fatigue (improvements of three ormore points) corresponded to 3-hour gains in pro-ductive time and 2-hour reductions in requiredcaregiver time. The authors note that there were afew patients (n = 7) with extreme changes in produc-tive and caregiver time, apparently primarily owingto progressive disease. When these observationswere removed from the models, the qualitative rela-tionship between fatigue and these outcomes, andthe level of statistical significance was unchanged,although the effect size was somewhat muted.

As these findings were based on exploratory out-comes reported by cancer patients with anaemiareceiving chemotherapy and erythropoietic therapyin a clinical trial setting, they should be confirmed instudies of broader patient populations. For example,the relationship between fatigue and productivitymay vary according to patient characteristics, suchas baseline haemoglobin level, performance statusand whether they are employed. The sample sizewas insufficient to perform meaningful subgroupanalyses. Furthermore, the trial was not powered toexamine treatment effects related to productivityoutcomes. A larger, randomised, controlled trial ofpatients with fatigue would facilitate further explo-ration of these effects.

The ability to resume normal roles and activitiesmay have broader implications for patients. Forexample, patients may often become anxious, frus-trated or depressed with their reduced productivityand increased reliance on others for assistance.Therefore, alleviating fatigue leads to enhanced ac-


Tab

le V

. M

ultiv

aria

te r

egre

ssio

n m

odel

of

chan

ge in

pro

duct

ivity

Stu

dy v

aria

ble

Pro

duct

ive

time

lost

(h)

Car

egiv

er t

ime

(h)

NR

S a

ctiv

ity[n

= 1

78,

adju

sted

R2

= 0

.37]

a[n

= 2

45,

adju

sted

R2

= 0

.14]

b[n

= 2

42,

adju

sted

R2

= 0

.73]

c

para

met

er e

stim

ate

p-va

lue

para

met

er e

stim

ate

p-va

lue

para

met

er e

stim

ate

p-va

lue

(95%

CI)

(95%

CI)

(95%

CI)

Inte

rcep

t64

.6 (

18.1

, 11

1.1)

0.00

6719

.2 (

–6.3

, 44

.7)

0.13

931.

5 (–

13.2

, 16

.1)

0.84

39A

ge–0

.4 (

–0.8

, 0.

1)0.

0987

–0.0

(–0

.3,

0.2)

0.90

07–0

.1 (

–0.2

, 0.

1)0.

4599

Fem

ale

7.8

(–4.

4, 1

9.9)

0.20

841.

0 (–

6.1,

8.1

)0.

7776

6.4

(2.2

, 10

.6)

0.00

31P

rogr

essi

ve d

isea

se–4

.1 (

–16.

5, 8

.2)

0.51

062.

8 (–

4.4,

10.

1)0.

4400

–4.8

(–9

.0,

–0.6

)0.

0256

Bas

elin

e pr

oduc

tive

time

lost

–0.8

(–0

.9,

–0.6

)<

0.00

01N

AN

AN

AN

AB

asel

ine

care

give

r tim

eN

AN

A–0

.3 (

–0.4

, –0

.2)

<0.

0001

NA

NA

Bas

elin

e N

RS

act

ivity

NA

NA

NA

NA

–0.8

(–0

.9,

–0.7

)<

0.00

01B

asel

ine

FA

CT

-Fat

igue

sub

scal

e sc

ore

–1.1

(–1

.6,

–0.5

)0.

0002

–0.6

(–0

.9,

–0.3

)0.

0002

1.4

(1.1

, 1.

6)<

0.00

01C

hang

e in

FA

CT

-Fat

igue

sub

scal

e sc

ore

–1.0

(–1

.5,

–0.5

)<

0.00

01–0

.7 (

–1.0

, –0

.4)

<0.

0001

1.6

(1.4

, 1.

7)<

0.00

01a

The

mod

el w

as b

ased

on

178

patie

nts

with

non

-mis

sing

cha

nge

in p

rodu

ctiv

e tim

e lo

st a

nd n

on-m

issi

ng v

alue

s of

the

cov

aria

tes

(age

, se

x, d

isea

se p

rogr

essi

on,

base

line

prod

uctiv

e tim

e lo

st,

base

line

FA

CT

-Fat

igue

sub

scal

e sc

ore

and

chan

ge in

FA

CT

-Fat

igue

sub

scal

e sc

ore)

. N

egat

ive

coef

ficie

nt m

eans

impr

ovem

ent

in p

rodu

ctiv

e tim

e lo

st.

bT

he m

odel

was

bas

ed o

n 24

5 pa

tient

s w

ith n

on-m

issi

ng c

hang

e in

car

egiv

er t

ime

and

non-

mis

sing

val

ues

of t

he c

ovar

iate

s (a

ge,

sex,

dis

ease

pro

gres

sion

, ba

selin

eca

regi

ver

time,

bas

elin

e F

AC

T-F

atig

ue s

ubsc

ale

scor

e an

d ch

ange

in F

AC

T-F

atig

ue s

ubsc

ale

scor

e).

Neg

ativ

e co

effic

ient

mea

ns im

prov

emen

t in

car

egiv

er t

ime.

cT

he m

odel

was

bas

ed o

n 24

2 pa

tient

s w

ith n

on-m

issi

ng c

hang

e in

NR

S a

ctiv

ity a

nd n

on-m

issi

ng v

alue

s of

the

cov

aria

tes

(age

, se

x, d

isea

se p

rogr

essi

on,

base

line

NR

Sac

tivity

, ba

selin

e F

AC

T-f

atig

ue s

ubsc

ale

scor

e an

d ch

ange

in F

AC

T-f

atig

ue s

ubsc

ale

scor

e).

FA

CT

= F

unct

iona

l Ass

essm

ent

of C

ance

r T

hera

py;

NA

= n

ot a

pplic

able

; N

RS

= n

umer

ic r

atin

g sc

ale.


Ernst Berndt was responsible for the conceptual and sta-tivity levels and patients may, in turn, have reducedtistical approach to the data and analysis, writing of sectionsanxiety. In a pooled analysis of the psychologicaland overall responsibility for content and analysis. Dr Berndt

outcomes reported in this trial and in a similar lung has had a consulting contract with Amgen, Inc. Anne McDer-cancer trial, improvements in fatigue were, in fact, mott and Xiao Xu were responsible for the statistical ap-significantly associated with reductions in anxiety proach, writing of sections of the manuscript and completion

of statistical analyses. Neither have individual conflicts ofand depression.[19]

interest but both are employed by Covance Incorporated,Productivity in this study was defined as a broadwhich has contracts with Amgen, Inc. Howard Lee and John

set of activities that patients consider their usual Glaspy were responsible for the clinical and conceptual ap-responsibilities. When broadly defined, losses in proach to the data and analysis, writing of sections and

editorial responsibilities. Drs Lee and Glaspy have no con-productivity have both indirect and direct cost impli-flicts of interest. Joel Kallich was responsbile for the designcations. In a retrospective analysis of disability data,and execution of the studies/patient-reported outcome dataBarnett et al.[20] estimated that it costs the employercollection, conceptual and statistical approach to analysis and

$US133 for every person-day of missed work, lead- writing of sections. Dr Kallich is an employee of Amgen,ing to an estimated $US1117 (year 1996–98 values) Inc., Thousand Oaks, California, USA.in additional annual employer costs for an employedcancer patient with anaemia relative to an employed Referencescancer patient without anaemia. Furthermore, the 1. Groopman J, Itri L. Chemotherapy-induced anemia in adults:

incidence and treatment. J Natl Cancer Inst 1999; 91: 1616-34economic burden of anaemia has been documented2. Glaspy J, Bukowski R, Steinberg D, et al. Impact of therapywith respect to direct medical costs. Cancer patients

with epoetin alfa on clinical outcomes in patients with nonmy-with anaemia use considerably more inpatient and eloid malignancies during cancer chemotherapy in community

oncology practice. J Clin Oncol 1997; 15: 1218-34outpatient services relative to cancer patients with-3. Littlewood TJ, Bajetta E, Nortier JWR, et al. Effects of epoetinout anaemia.[9] To provide evidence-based, cost-

alfa on hematologic parameters and quality of life in cancereffectiveness analyses of anaemia therapy associat- patients receiving nonplatinum chemotherapy: results of a

randomized, double-blind, placebo-controlled trial. J Clined with cancer, it is essential that reliable estimatesOncol 2001; 19: 2865-74of total costs, both direct medical expenditures and

4. Demetri GD, Kris M, Wade J, et al. Quality-of-life benefit inindirect costs to society associated with lost produc- chemotherapy patients treated with epoetin alfa is independent

of disease response or tumour type: results from a prospectivetivity on the part of the patient and caregivers, becommunity oncology study. J Clin Oncol 1998; 16: 3412-25

developed and validated. 5. Gabrilove JL, Cleeland CS, Livingston RB, et al. Clinical evalu-ation of once-weekly dosing of epoetin alfa in chemotherapypatients: improvements in haemoglobin and quality of life areConclusionsimilar to three-times-weekly dosing. J Clin Oncol 2001; 19:2875-82

While economic considerations are striking, re- 6. Richardson A. Fatigue in cancer patients: a review of the litera-ductions in fatigue associated with the treatment of ture. Eur J Cancer Care (Engl) 1995; 4: 20-32

7. Curt G, Breitbart W, Cella D, et al. Impact of cancer-relatedanaemia in cancer patients receiving chemotherapyfatigue on the lives of patients: new findings from the Fatigue

have repeatedly been associated with improvements Coalition. Oncologist 2000; 5: 353-60in measured HR-QOL.[3,17,21] The findings reported 8. Cella D, Peterman A, Passik S, et al. Progress toward guidelines

for the management of fatigue. Oncology 1998; 12: 369-77in this study, linking decreases in fatigue to objec-9. Lyman GH, Berndt ER, Kallich JD, et al. The economic burden

tive improvements in productive time and reduced of anemia in cancer patients receiving chemotherapy. ValueHealth 2005; 8 (2): 149-56caregiver burden, can help in transforming abstract

10. Berndt E, Crown W, Kallich J, et al. The impact of anaemia andconcepts of HR-QOL into more tangible and under-its treatment on employee disability and medical costs.

standable real and quantifiable impacts on patients PharmacoEconomics 2005; 23 (2): 183-9211. Glaspy J, Jadeja J, Justice G, et al. Darbepoetin alfa given everyand their caregivers.

1 or 2 weeks alleviates anaemia associated with cancer chemo-therapy. Br J Cancer 2002; 87: 268-76

Acknowledgements 12. Cella D. The Functional Assessment of Cancer Therapy-Ane-mia (FACT-An) Scale: a new tool for the assessment of

This research was supported by Amgen, Inc., Thousand outcomes in cancer anemia and fatigue. Semin Hematol 1997;34 Suppl. 2: 13-9Oaks, California, USA.


514 Berndt et al.

13. Cella D. Manual of the Functional Assessment of Chronic [online]. Available from URL: http://www.nccn.org/profes-Illness Therapy (FACIT) Scales, Version 4. Evanston (IL): sionals/physician_gls/PDF/fatigue.pdf [Accessed 2005 MarCenter on Outcomes Research and Education, Evanston North- 29]western Healthcare and Northwestern University, 1997

19. Kallich J, Tchekmedyian NS, Damiano A, et al. Psychological14. Cella D, Lai J, Chang C, et al. Fatigue in cancer patientsoutcomes associated with anemia-related fatigue in cancercompared with fatigue in the general United States population.

Cancer 2002; 94: 528-38 patients. Oncology (Huntingt) 2002 Sep; 16 (9 Suppl. 10):15. Cella D, Eton D, Lai J, et al. Combining anchor and distribution 117-24

based methods to derive minimal clinically important differ-20. Barnett A, Cremieux PY, Fendrick AM, et al. Anemia-relatedences on the Functional Assessment of Cancer Therapy

costs for cancer patients. J Managed Care Med 2002; 6: 20-8(FACT) Anemia and Fatigue Scales. J Pain Symptom Manage2002; 24: 547-61 21. Osterborg A, Brandberg Y, Molostova V, et al. Randomized,

16. Hedenus M, Adriansson M, San Miguel J, et al. Efficacy and double-blind, placebo-controlled trial of recombinant humansafety of darbepoetin alfa in anaemic patients with

erythropoietin, epoetin beta, in hematologic malignancies. Jlymphoproliferative malignancies: a randomized, double-Clin Oncol 2002; 20: 2486-94blind, placebo-controlled study. Br J Haematol 2003; 122:

394-40317. Vansteenkiste J, Pirker R, Massuti B, et al. Double-blind, place-

bo-controlled, randomized phase III trial of darbepoetin alfa in Correspondence and offprints: Dr Ernst Berndt, MIT Sloanlung cancer patients receiving chemotherapy. J Natl Cancer School of Management, Room E52-452, 50 Memorial Drive,Inst 2002; 94: 1211-20

Cambridge, MA 02142, USA.18. National Comprehensive Cancer Network, Inc. Clinical practiceguidelines in oncology: cancer-related fatigue. Version 1, 2004 E-mail: [email protected]


Documents

Reductions in anaemia and fatigue are associated with improvements in productivity in cancer patients receiving chemotherapy