The use of quality of life data in clinical practice
-
Upload
jenny-morris -
Category
Documents
-
view
216 -
download
4
Transcript of The use of quality of life data in clinical practice
Quality of Life Research, 7, pp. 85–91
© 1998 Rapid Science Publishers Quality of Life Research . Vol 7 . 1998 85
The use of quality of life data in clinical practice
Jenny Morris*, David Perez and Bronwen McNoe
Hugh Adam Cancer Epidemiology Unit, Department of Preventive and SocialMedicine, University of Otago, Dunedin, New Zealand (J. Morris, B. McNoe);Department of Medicine, Healthcare Otago, Dunedin, New Zealand (D Perez)
A large amount of quality of life (QoL) information hasbeen and is being collected in the oncology settingbut it is unclear how such data influence decisionsabout the management of individual patients. Aquestionnaire designed specifically for the study wasmailed to 260 senior oncologists to investigate howQoL data are being used outside the context ofcancer clinical trials; replies were received from 154(59%). Approximately 80% believed QoL informationshould be collected prior to the commencement oftreatment, but less than 50% actually did so.Similarly, less than 50% assessed QoL as a method ofmonitoring the responses to treatment even when thetreatment goal was palliation. The barriers tocollecting such data were time and resourceconstraints, perceived lack of an appropriateinstrument and a belief that QoL assessments wereunnecessary. Other than making a subjectiveassessment based on examination and history, 73(47%) used either standardized questionnaires or asystem derived in their unit to assess the QoL of theirpatients. Given an appropriate instrument themajority believed that QoL data could be collected ona routine basis. The main barriers to collecting QoLdata are logistic and the challenge remains todevelop a method of collecting and analysing QoLinformation in a manner which enhances decisionmaking.
Keywords: QoL; oncology; decision making; physicianbehaviour.
Introduction
Psychosocial research has revealed a substantial levelof psychosocial morbidity associated with a diagnosisof cancer and the side-effects of the various modes oftreatment: surgery, chemotherapy and radiotherapy.
This research has been undertaken mainly in specialistcentres in the UK, USA and Europe studying a varietyof patient groups. The findings have been consistentrevealing that a significant minority of patients(23%–40%) experience clinical levels of anxietyand/or depression following treatment,1 the side-effects from chemotherapy produce a conditionedresponse termed anticipatory nausea and vomitingaffecting some 25% of patients treated withchemotherapy,2 and feelings of helplessness, loss ofself-esteem and impairment of social and sexualrelationships have also been reported.3,4
This body of research has been influential in thatthere have been notable changes in clinical practiceespecially in specialist centres. For example, specialistnurses have been trained in order to reduce thepsychosocial morbidity associated with diagnosis andtreatment as well as to offer practical care,5,6 commu-nication skills programmes have been developed forhealth care professionals working with cancerpatients7,8 and cognitive behavioural programmeshave been designed to aid psychosocial adaptation inthose patients most at risk of developing psychosocialproblems.9 A fourth key area has been the devel-opment of health-related quality of life (QoL)measures to standardize the way in whichpsychosocial data are collected and to aid compar-isons between groups of patients with regard to thelevel of psychosocial morbidity. Health-related QoL istaken to be a multidimensional construct whichconsists of four core domains: psychologicalfunctioning (well-being and emotional status), socialfunctioning, physical status and disease andtreatment-related symptoms.10 A number of instru-ments have been designed to measure QoL in cancerpatients11,12 and, consequently, there has been anincrease in the number of phase III cancer clinicaltrials in which QoL data are collected.13–15 In a generalreview of the development of QoL research over thepast decade, Ganz16 emphasized that QoL outcomedata are needed to facilitate clinical decision making
*To whom correspondence should be addressed at HughAdam Epidemiology Unit, Department of Preventive andSocial Medicine, University of Otago, Dunedin, NewZealand.
and that such data are likely to be incorporated intothe decisions of policy makers.16
The extent to and manner in which such datainfluence decisions about treatment, however, remainrelatively unexplored. Gough and Dalgleish17
explored the relative value of QoL compared withtraditional measures of outcome as assessed by oncol-ogists, GPs and oncology nurses in Queensland,Australia and found that all three groups rated QoL asbeing the most important objective of palliativechemotherapy for advanced cancer.17 This study didnot, however, investigate the circumstances in whichmedical and nursing staff would use QoL informationto influence decisions about treatment in non-palliative cases and the factors which would hinderuse of QoL data.
While it is clear that some specialist centres haveused QoL information to improve the management ofcancer patients, it is not clear how such data are beingused by oncologists in routine practice. The aim of thisstudy is to explore how QoL data are being used byoncologists and the perceived barriers to using suchinformation.
Methods
Design and sample
A questionnaire survey of senior medical andradiation oncologists currently working in NewZealand, Australia and Hong Kong was carried out.Lists of senior oncologists working in New Zealandand Australia were obtained from the relevant profes-sional organizations. In Hong Kong, questionnaireswere sent to the senior oncologists working in the fivemajor public sector hospitals which have recognizedoncology services.
Questionnaire
Interviews with oncologists at Healthcare Otago inNew Zealand were undertaken to develop thequestionnaire which was designed to focus on threemain areas: (1) the use of QoL information inindividual patient care prior to the commencement oftreatment; (2) the use of QoL information inindividual patient care as a means of monitoringresponse to treatment and (3) the use of QoL infor-mation in the evaluation of new treatments. Health-related QoL was defined as affecting four coredomains: physical functioning (e.g. energy levels andthe ability to undertake everyday activities), the
symptoms caused by both the disease and treatment,psychological functioning (emotional state and well-being) and social functioning. The questions relatedto decisions about the management of adult patientswith cancer, and considered curative/life-extendingand palliative situations. Questions regarding age,sex, the number of years in oncology practice and theresearch experience of the oncologists were alsoincluded. The questionnaire was then reviewed bythe medical oncologists at Healthcare Otago and alsoby those working in the Department of ClinicalOncology at the Chinese University of Hong Kong.The final version of this questionnaire was sent to allsenior medical and radiation oncologists in NewZealand and Australia and also the senior medicaland radiation oncologists working in the five majorpublic sector hospitals in Hong Kong. A reminderletter plus an additional copy of the questionnairewas sent to non-responders.
Data analysis
The data from the open-ended questions were codedand checked by two independent raters; the categoriesused for coding are available upon request. The datawere entered onto a VAX mainframe using doubleentry verification. The data were analysed using theSPSS package.
Results
Response rate
Replies were received from 27 of the 34 oncologistsmailed in New Zealand (79%), 112 of the 191 oncolo-gists in Australia (59%) and 15 of the 35 oncologists inHong Kong (43%). The responses to the questions donot always total 154 due to missing data.
Characteristics of the sample
The characteristics of those participating in the surveyare summarized in Table 1. We do not have descriptiveinformation about the non-responders other thancurrent position (consultant or equivalent/seniorregistrar or equivalent) and, therefore, cannot makecomparisons between responders and non-responders.
The mean number of years working as an oncol-ogist was 13 years (SD 7.6 years). One hundred and forty-four (93.5%) belonged to professional
J. Morris et al.
86 Quality of Life Research . Vol 7 . 1998
Table 2. Use of QoL information in the management of individual patients prior to commencement of treatment
All Most Somecircumstances circumstances circumstances Never
n % n % n % n %
Do you believe an assessment of QoL should be made prior to the commencement of treatment?
Treatment considered curative/life-extending 69 45 53 35 29 19 2 1Treatment considered palliative 87 57 45 30 19 12 1 1
Do you formally assess and record QoL prior to the commencement of treatment?
Treatment considered curative/life-extending 41 27 26 17 56 38 26 17Treatment considered palliative 44 30 28 19 50 34 26 18
associations concerned specifically with cancerresearch, clinical trials or standards of oncologypractice. Ten percent of the oncologists did not enrollany patients into clinical cancer trials in 1994 and 28%entered 50 or more patients. There were no significantdifferences between the oncologists in the threecountries according to age, sex and number of years inpractice as a medical or radiation oncologist. More ofthe oncologists in Hong Kong, however (50%compared with 8% in New Zealand and 6% inAustralia, χ2 = 38.7, p < 0.0001), did not enter anypatients into cancer clinical trials in 1994. In addition,fewer of the oncologists in Hong Kong belonged toprofessional associations concerned specifically withcancer research, clinical trials or standards of oncologypractice (60% compared with 100% in New Zealandand 96% in Australia, χ2 = 31.2, p < 0.0001).
Use of QoL information in individual patientcare prior to commencement of treatment
Table 2 illustrates the circumstances in which theoncologists would use QoL information prior to thecommencement of treatment.
It can be seen that the majority believe anassessment of QoL should be made ‘in all/mostcircumstances’ when treatment is considered eithercurative (80%) or palliative (87%), but less than 50%formally assessed and recorded QoL ‘in all/mostcircumstances’ (44% in curative/life-extending situa-tions and 49% when treatment was consideredpalliative). Only 17%, however, stated they neverformally assessed and recorded QoL information priorto the commencement of treatment.
The use of QoL data in clinical practice
Quality of Life Research . Vol 7 . 1998 87
Table 1. Characteristics of the sample
n %
SexFemale 21 14Male 133 86
Age group30–39 years 60 3940–49 years 66 4350–59 years 22 1460–69 years 6 4
Current positionConsultant/hospital specialist/
senior lecturer 143 93Senior registrar/senior medical
officer 9 6Research fellow 1 1
CountryNew Zealand 27 17Australia 112 73Hong Kong 15 10
Sector in which practisedPublic 66 43Private 8 5Public and private 80 52
Number of patients enrolled intocancer clinical trials in 1994
0 15 101–5 25 176–10 22 1511–20 42 2921–50 29 2050+ 12 8
The main reasons given for not formally assessingand recording QoL when treatment was consideredcurative/life-extending were time and resourceconstraints (24%), lack of an appropriate instrument(22%), a combination of these two factors (8%), a beliefthat a doctor’s subjective assessment of QoL was suffi-cient (23%) and a belief that treatment decisionsoutweighed QoL information (22%).
The reasons given when treatment was consideredpalliative were a belief that a doctor’s subjectiveassessment of QoL was sufficient (25%), lack of anappropriate instrument (24%), time and resourceconstraints (20%) or a combination of the two (11%)and a belief that QoL information would not altertreatment decisions (14%).
The most common circumstances when QoL infor-mation would influence the oncologist’s decisionsabout the treatment of a particular patient were indecisions about palliative treatment (31%), in discus-sions about choice of treatment (18%), a combinationof these two factors (9%) and in the assessment oftreatment outcome (17%).
Use of QoL information in individual patientcare as a means of monitoring response totreatment
The results are summarized in Table 3 from which itcan be seen that the majority (60% when treatmentwas considered curative/life-extending and 54%when treatment was considered palliative) formallyassess and record QoL information ‘some of the time’or ‘never’. When treatment is considered palliative,46% use QoL information to monitor the response totreatment and approximately 20% never use suchinformation when treatment is considered eithercurative or palliative. The reasons for not collectingsuch information are similar to those stated in theprevious section.
Methods of assessing QoL
Twenty-eight percent of the oncologists stated thatthey used standarized questionnaires to assess theQoL of patients in their unit, although 64% of thisgroup indicated that this was in the context of clinicaltrials only. Twenty percent indicated that they used asystem derived in their unit; such ‘systems’ includedquestionnaires, informal recording in their notes andgeneral questions not necessarily recorded in theirnotes. It was clear from some of the comments made,however, that performance status rather than QoL wasbeing assessed, e.g. ‘record performance status’,‘performance status, assessment of circumstances’ and‘0 asymptomatic, 1 symptomatic but fully active,2 < 50% time spent resting, 3 > 50% time spent restingand 4 bedridden’.
Relative importance of outcome measures
The oncologists were asked to rank several measuresof outcome to illustrate their relative importance indeciding whether to continue treatment. They wereasked to assume that QoL and doctor-assessed perfor-mance status were assessed using a standardizedquestionnaire. The results are illustrated in Table 4.
When treatment was considered curative/life-extending, 86% ranked tumour response firstcompared with only 6% for QoL and 9% for doctor-assessed performance status. Furthermore, 14% didnot rank doctor-assessed performance status and 11%did not rank patient-assessed QoL; these percentagesare considerably higher than the figures relating to thetraditional clinical measures of outcome. Whentreatment was considered palliative, the differencesbetween the measures in terms of relative importancewere less obvious. The most frequently endorsedmeasure was patient-assessed QoL which was rankedfirst by 46%. As might be expected, side-effects of
J. Morris et al.
88 Quality of Life Research . Vol 7 . 1998
Table 3. Use of QoL information to monitor response to treatment
All of the Most of the Some of the Nevertime time time
n % n % n % n %
Do you formally assess and record QoL as a means of monitoring response to treatment?
Treatment considered curative/life-extending 30 20 28 19 59 40 30 20Treatment considered palliative 37 25 30 21 48 33 30 21
treatment were ranked in the top three by 89% of theoncologists. The lack of agreement about the relativeimportance of doctor-assessed performance status isnotable with similar numbers ranking this most andleast important.
The oncologists were given the opportunity toinclude additional criteria to those described in Table 4. The criteria included were a doctor’ssubjective assessment of QoL (4%), survival (4%),symptom control (4%), subjective assessment ofperformance status (3%) and discussion with patients(2%).
The ranking exercise was used to determine therelative importance of particular types of information
when evaluating new treatments (e.g. when reviewingpublished results from clinical trials and literatureproduced by pharmaceutical companies); the resultsare presented in Table 5.
When treatment was considered curative/life-extending, tumour response was ranked first mostfrequently (83%), the side-effects of treatment wereranked second (70%) and patient-assessed QoL third(46%). When treatment was considered palliative, therelative importance of the criteria are less obvious.Patient-assessed QoL was considered most importantby 46%, the side-effects of treatment were rankedsecond by 43% and doctor-assessed performancestatus ranked third (23%). Tumour response was
The use of QoL data in clinical practice
Quality of Life Research . Vol 7 . 1998 89
Table 4. Relative importance of outcome measures in the effectiveness of treatment (figures expressed as % ofoncologists)
Patient- Doctor-assessedTumour Side-effects of assessed Laboratory performance
response treatment �QoL a results statusa
Rank C P C P C P C P C P
1b 86 18 6 25 6 46 9 2 9 182 8 14 26 39 7 17 46 5 5 143 2 23 40 25 21 11 11 7 18 254 1 28 19 8 31 8 12 23 21 195 1 9 5 1 22 9 16 44 30 126 – – – – 1 1 – 6 3 3Not ranked 2 8 3 1 11 8 6 12 14 10
C, treatment considered curative/life extending (n = 147) and P, treatment considered palliative (n = 145).aSpecified using a standardized questionnaire.bRank of 1 = most important.
Table 5. Relative importance of outcome measures in the evaluation of new treatments (figures expressed as % ofoncologists)
Patient Doctor-assessedTumour Side-effects assessed performance
response of treatment QoLa statusa
Rank C P C P C P C P
1b 83 31 7 27 7 46 5 72 9 13 70 43 8 25 7 173 3 17 16 25 46 19 18 234 2 28 5 5 29 8 50 435 1 3 1 – 5 1 9 56 – 1 – – – – 3 1Not ranked 3 7 1 1 5 1 8 3
C, treatment considered curative/life-extending (n = 147) and, P, treatment considered palliative (n = 145).aSpecified using a standardized questionnaire.bRank of 1 = most important.
ranked first by 31% and fourth by 28% showing someinconsistency in the relative importance of thismeasure of effectiveness.
General QoL issues
The oncologists were asked whether they believedhealth-related QoL data could be collected routinely intheir unit if they had a 20-item questionnaire whichtook patients only 5 minutes to complete. Sixty-fivepercent believed QoL information could be collectedfrom 75–100% of their patients, 14% from 50%, 5%from 25%, 12% from a highly selected group only andonly 3% believed such data could not be collected fromany of their patients. Fifty-six percent of the oncolo-gists provided reasons why such data could not becollected routinely on 100% of their patients. Thesewere a lack of resources to collect and/or analyse theinformation (40%), patient compliance (22%, e.g.‘patients do not want it’ and ‘patient factors – difficultyin understanding the forms for some, patients notinterested’), language difficulties in a number ofpatients (15%), the QoL questionnaire being of nobenefit (14%, e.g. ‘because the subtleties of theindividual’s physical, psychological and social well-being are too complex to be reduced to a shortquestionnaire that would produce useful information.. . .’, ‘it remains a research tool’ and ‘subjectiveassessment sufficient’), patients incapable ofcompleting questionnaires (6%) and unlikely that onequestionnaire would be suitable for all patients (2%).
At the end of the questionnaire, the oncologistswere invited to add any other relevant commentsabout QoL information in relation to the managementof patients. Comments were received by 29%, 36% ofwhich were positive reflecting the perceived impor-tance of the contribution of QoL information, 33% ofwhich highlighted methodological issues and 11% ofwhich emphasised that QoL measurement was mostrelevant in the context of clinical trials. Two commentswere negative: ‘I am concerned at the drive bypsychologists to simplify quality of life etc. to astandard questionnaire. . . I would rather have acompetent psychologist available and a nursecounsellor here than a bundle of questionnaires. . .’and ‘for routine care I do not think a well trainedconsiderate clinician who is prepared to listen and isnot overworked needs a questionnaire. . .’. Theremainder of the comments were general.
Subgroup analyses
In addition to the analyses reported above, the datawere grouped according to the country in which the
oncologists practised, the age and sex of the oncolo-gists, whether or not they belonged to professionalassociations concerned with the cancer research orcancer clinical trials and the number of patientsenrolled into cancer clinical trials. There were nosignificant differences between the results for any ofthese subgroups.
Discussion
The results from this survey indicate that QoL infor-mation could have an important part to play in themanagement of individual patients. Approximately80% believed an assessment of QoL should be madeprior to the commencement of treatment, both whentreatment is considered curative/life-extending andalso when treatment is considered palliative. Inpractice, however, less than 50% formally assessedand recorded QoL information prior to thecommencement of treatment and also as a means ofmonitoring response to treatment.
Thirty (20%) of the oncologists reported that theyused a system derived in their unit for assessinghealth-related QoL. From the information provided,however, it was clear that the development of such‘systems’ varied considerably. One-third of the oncol-ogists used questionnaires developed in their unit tomeasure QoL in their patients, 23% stated that theyasked QoL-specific questions but did not describehow such information was recorded in the notes, 20%indicated that they assessed performance status andthe remainder of the responses could not be grouped.It is unlikely that these informal systems of measuringQoL could be used in any systematic way to assesspatients prior to the commencement of treatment or asa means of monitoring responses to treatment. Inaddition, merely recording QoL information inpatients’ notes without a structured proforma isunlikely to be of use to anyone other than the clini-cians within the given unit.
With regard to the practical application of QoL as ameasure of outcome, it is clear that biomedicalmeasures such as tumour response and laboratoryresults are considered more important than QoL whentreatment is considered curative/life-extending.When treatment is palliative, however, there is morevariability in the perceived relative importance of theoutcome measures with QoL being ranked mostimportant most often but by less than 50% of thesample. It is encouraging to note that patient-assessedQoL is considered a more important measure ofoutcome than doctor-assessed performance statuswhen the treatment is considered palliative. It is clear
J. Morris et al.
90 Quality of Life Research . Vol 7 . 1998
from the percentage of oncologists who did not rankpatient-assessed QoL or doctor-assessed performancestatus (11% and 14%, respectively) that incurative/life-extending situations there is a preferencefor traditional clinical measures of outcome.
While there are logistical reasons (such as time andresource constraints and the perceived lack of asuitable questionnaire) for collecting QoL informationroutinely, these reasons were given by a relativelysmall subset and were similar in number to those whobelieved that treatment decisions based on traditionalclinical criteria outweighed QoL information and thata doctor’s subjective assessment of QoL is sufficient.
During the period in which this study was beingundertaken, results were published from a Canadianstudy which are relevant to this issue.18 Consistentwith the results reported here, the data from theCanadian study illustrated substantial variation in theextent to which the oncologists were willing to collectand use QoL information and that many felt QoLassessments were most appropriate within the contextof clinical trials. Furthermore, it was noted that someoncologists emphasized the physical rather than thepsychosocial domain of QoL which is consistent withthe results reported here, indicating that some oncolo-gists were confusing performance status with QoL.
In conclusion, there is conflict in that many oncolo-gists appreciate the relevance of QoL data in assessingthe outcome from treatment, particularly in thecontext of palliative care and yet few collect andrecord such information on a routine basis. It is clearthat some oncologists doubt the usefulness of QoLinformation, a view that may both reflect thepreference for biomedical measures consistent withmedical training and also a relative lack of knowledgeand expertise in the area of questionnaire design andpsychometric issues in general. In addition, there arepractical barriers related to data collection andanalysis. The QoL instruments currently availablehave mainly been used in clinical trials and there is noconsensus on which QoL instrument to use, how bestto collect and record QoL data for use in routineclinical care and how to interpret the data and thusmake key decisions about therapy based on thematerial present.
Acknowledgements
We would like thank Dr S. F. Leung, lecturer and Dr N.Wickham, Senior lecturer, Department of ClinicalOncology at the Chinese University of Hong Kong fororganizing the data collection in Hong Kong, the
oncologists who participated in the survey, and theCancer Society of New Zealand who funded JennyMorris during the period of this study.
References
1. Hughes J. Cancer and Emotion. Psychological Preludes andReactions to Cancer. Chichester: John Wiley & Sons, 1987.
2. Morrow G, Dobkin P. Anticipatory nausea and vomitingin cancer patients undergoing chemotherapy treatment:prevalence, etiology and behavioral interventions. ClinPsychol Rev 1988; 8: 517–556.
3. Greer S. Psycho-oncology: its aims, achievements andfuture tasks. Psycho-Oncology 1994; 3: 87–101.
4. Sprangers M, Te Velde A, Aaronson N, Taal B. Quality oflife following surgery for colorectal cancer: literaturereview. Psycho-Oncology 1993; 2: 247–259.
5. Maguire P, Brooke M, Tait A, Thomas C, Sellwood R. Theeffect of counselling on physical disability and socialrecovery after mastectomy. Clin Oncol 1983; 9: 319–324.
6. Watson M, Denton S, Baum M, Greer S. Counsellingbreast cancer patients: a specialist nurse service. CounselPsychol Quart 1988; 1: 25–34.
7. Maguire P, Faulkner A. Improve the counselling skills ofdoctors and nurses in cancer care. BMJ 1988; 297:847–849.
8. Maguire P, Faulkner A. Communicate with cancerpatients: 1. Handling bad news and difficult questions.BMJ 1988; 297: 907–909.
9. Greer S, Moorey S, Baruch J et al. Adjuvant psychologicaltherapy for patients with cancer: a prospectiverandomised trial. BMJ 1992; 304: 675–680.
10. Aaronson N. Quality of life: what is it? How should it bemeasured? Oncology 1988; 2: 69–74.
11. Anderson R, Aaronson N, Wilkin D. Critical review of theinternational assessments of health-related quality of life.Qual Life Res 1993; 2: 369–395.
12. Measuring the effect of cancer on health-related qualityof life. PharmacoEconomics 1995; 7: 308–319.
13. van Holten-Verzantvoort A, Zwinderman A, Aaronson Net al. The effect of supportive pamidronate treatment onaspects of quality of life of patients with advanced breastcancer. Eur J Cancer 1991; 27: 544–549.
14. Hurny C, Bernhard J, Gelber R et al. Qualify of lifemeasures for patients receiving adjuvant therapy forbreast cancer: an international trial. Eur J Cancer 1992; 28:118–124.
15. Morris J, Goddard M. Economic evaluation and qualityof life assessment in cancer clinical trials: the CHARTtrial. Eur J Cancer 1993; 29A: 766–770.
16. Ganz P. Quality of life and the patient with cancer.Individual and policy implications. Cancer 1994; 74:1445–1452.
17. Gough I, Dalgleish L. What value is given to quality oflife assessment by health professionals consideringresponse to palliative chemotherapy for advancedcancer? Cancer 1991; 68: 220–225.
18. Taylor K, Macdonald K, Bezjak A, Ng P, DePetrillo A.Physicians’ perspective on quality of life: an exploratorystudy of oncologists. Qual Life Res 1996; 5: 5–14.
(Received 15 April 1997; accepted 25 July 1997)
The use of QoL data in clinical practice
Quality of Life Research . Vol 7 . 1998 91