Computer-Based Assessment of Symptoms and Mobility in Palliative Care: Feasibility and Challenges

Article Outline

• Abstract
• Introduction
• Methods
  • Hardware
  • Software Design
  • Software Content
  • Procedure
  • Sampling
  • Data Collection
  • Ethical Considerations
  • Statistical Considerations
• Results
  • Study A
  • Study B
• Discussion
• Conclusions
• Acknowledgments
• Appendix. Items Presented to the Patients in the Computerized Assessment Tool
• References
• Copyright

Abstract 

The aims of the study were to explore the ability of cancer patients who are primarily receiving palliative care to use a touchscreen computer for assessment of symptoms and mobility and to investigate which factors predicted the need for assistance during the assessment. Before the main data collection, a pilot study was conducted to explore the preferences of these patients toward using such a computerized assessment tool. Patients were recruited from nine different inpatient and outpatient palliative care and general cancer clinics in Norway. The patients responded to 60 items on symptoms and mobility directly on the computer. In the pilot study (n=20), 11 patients (55.0%) preferred computerized assessment over paper and pencil, whereas five (25.0%) had no preference. In the main data collection, 370 patients (52.7% men with mean age 62 years and mean Karnofsky Performance Status score of 70) completed the assessment. Eighty-six patients (23.2%) required assistance. Patients requiring assistance were significantly older, had worse performance status, and poorer cognitive function than those not requiring assistance. Predictors for requiring assistance were age (P<0.001) and performance status (P<0.001). Because higher age and worse performance status resulted in more need of assistance, assessment tools should be short and user-friendly to ensure good compliance in frail patients.

Key Words: Palliative care, cancer, computers, symptom assessment, quality of life

Back to Article Outline

Introduction 

Most patients with advanced cancer experience distressing symptoms in addition to a gradual decline in function.¹, ², ³ The main goals of palliative care are to prevent and relieve symptoms and to maintain or improve quality of life (QoL). To achieve these goals, symptoms and functioning must be assessed routinely, and the treatment must be adjusted to each patient's changing condition.⁴, ⁵, ⁶ Inadequate assessment of symptoms and functioning has been recognized as a major obstacle to the provision of good palliative care.⁷, ⁸, ⁹, ¹⁰

Traditionally, patient-reported outcomes (PROs) have been assessed by questionnaires in the paper and pencil (P&P) format.¹¹ The P&P format requires that health professionals distribute and collect the questionnaires, assist in the completion if necessary, and compute the scores. These tasks are generally time consuming, thus preventing the presentation of the results in a timely manner to the clinical decision makers. Questionnaires in the P&P format also have proven difficult to integrate with other data in the medical record. Furthermore, P&P questionnaires “force” the patient to respond to a fixed set of questions, some of which may be less relevant for the individual. All these factors, and probably the P&P format itself, inflate the rates of missing data, especially among the sickest.¹²

Collection of PROs by computers theoretically may improve the assessment of symptoms and functioning by the integration of the results into the electronic medical record and by displaying immediate results to the clinician. In addition, the use of computer-adaptive testing (CAT) can make the assessments/questionnaires shorter, more precise, and better tailored to the individual's symptom profile and level of functioning.

Previous studies investigating the feasibility of computer-based assessment of QoL in cancer populations have concluded that computerized assessment is useful for patient-doctor communication, that compliance over time is good when the tool is integrated in routine inpatient care, and that most of the patients prefer computers over P&P questionnaires.¹³, ¹⁴, ¹⁵, ¹⁶ However, no study has explored the use of computer-based assessments in palliative cancer care, where user-friendliness is critically important, because most of the patients are elderly, frail, and have limited experience with computers. Also, none of the existing studies have explored how many patients are able to complete a computer-based assessment or how many need assistance.

A primary aim of the European Palliative Care Research Collaborative (www.epcrc.org) is to improve the assessment of pain, depression, and cachexia in palliative cancer care by computerized assessments.¹⁷ A preliminary version of a computer-based assessment tool has been developed and was tested in two different populations to investigate the feasibility of the tool. The testing also included analyses of the psychometric performance of the tool's content. Results from these analyses will be reported separately.

The questions addressed by the present study were: How do the patients experience the use of a touchscreen laptop computer for the assessment of symptoms and mobility? Does age, gender, educational level, performance status, or cognitive function predict the requirement for assistance? Is there a difference in time expenditure between patients requiring and patients not requiring assistance when completing the assessment, and which factors influence the time expenditure?

Back to Article Outline

Methods 

Hardware 

For this project, touch-sensitive computers (HP Compaq TC4200^® 12″ tablet PCs made by Hewlett-Packard Development Company L.D.) were used. They were standalone and portable. Patients answered by tapping directly on the screen with an electronic pen.

Software Design 

A focus group consisting of two oncologists and a psychologist worked closely together on the design of the software. Programming was performed by the aforementioned psychologist, as a member of the study group, in Flash 8 and Zinc 2.5 using Actionscript 2. The items were displayed one at a time with large, bold fonts to overcome eventual limitations because of poor vision. In addition to the response alternatives on each item, there were only two other buttons on each page: “next” and “abort.” Return to the previous items was not possible, and each item had to be answered before proceeding to the next. If the patient chose to abort, all data up to the point of aborting was stored.

Software Content 

The patients answered a total of 59 questions plus a body map for pain localization on the laptop computer (Appendix). The selection of pain items was based on the results of a systematic review,¹⁸ in addition to systematically collected responses from patient and expert groups. A similar procedure guided the selection of items on mobility.¹⁹ There were 10 items on pain interference, eight on pain intensity, five on the temporal pattern of the pain, and 24 on mobility. The content was further supplemented with two items on depression, seven items on other prevalent symptoms in palliative care populations, and three items on demographic data (marital status, educational level, and persons in the household). The response alternatives on the different items varied. Sixteen items had 11-point numerical rating scales, 35 items had four-category verbal rating scales (with categories “not at all,” “a little,” “quite a bit,” and “very much”), one item had a six-category verbal rating scale (with categories “none,” “very weak,” “weak,” “moderate,” “strong,” and “very strong”), and four items had “yes/no” as the alternatives.

Procedure 

For practical reasons, background data (age, gender) and medical data (Karnofsky Performance Status [KPS],²⁰ diagnosis, and medication) could be entered into the computer by the research nurse/research assistant either before or after the patient part of the assessment. Additionally, an assessment of cognitive function based on a shortened version of the Mini-Mental State Exam, described in the article by Fayers et al.,²¹ was performed. Health care professionals asked the patients about the present date, month, and year, and to spell the word “sword” backward. Each correct answer for year, month, and date was scored 1, and the spelling of “sword” was scored 1 for each correct letter, giving a maximum cognitive function score of 8 points.

Sampling 

Inclusion criteria were cancer patients receiving palliative care, age≥18 years, and ability to provide written informed consent. The exclusion criterion was the inability to complete the assessment because of physical or obvious cognitive impairment or language problems. Eligible patients were contacted by a research assistant, a nurse, or a doctor and asked if they wanted to participate.

Data Collection 

The study included two data collections. A pilot study (Study A) explored the design and user-friendliness of the computerized tool. In a subsequent multicenter data collection (Study B), data on the requirement of assistance and time expenditure were collected.

Study A was conducted in 2006 at the inpatient and outpatient clinics of the Palliative Medicine Unit at St. Olavs Hospital in Trondheim, Norway. The patients first completed the computerized questionnaire. Their experience with the tool was then explored by a structured interview, including the following six questions: 1) “Do you have experience with using a computer?” (yes/no); 2) “Would you prefer to use pen and paper instead of a monitor?” (yes/no or no preference); 3) “Was the text readable and easy to understand?” (yes/no); 4) “Were the questions relevant for your situation?” (yes/no); 5) “What did you think about the time used to complete the questionnaire?” (too long/ok); and 6) general comments.

Study B was a Norwegian multicenter study conducted from June 2006 to February 2007. Nine different hospital wards (four palliative care units, two oncology wards, and three medical wards) and seven outpatient clinics (three palliative care, one medical, and three oncology clinics) in different parts of Norway participated. In Study B, multiple registrations on the same patient were allowed, with at least seven days between the assessments. Only the first registration for each patient was included in the present article, because the authors assumed that repeated assessments would include an element of learning and, thereby, affect the need for assistance and time expenditure as patients became familiarized with the software.

After giving their written informed consent, the patients were enrolled. A research nurse or research assistant presented the computer to the patients and was present during the assessment. Because the computers were portable, all assessments were conducted at the site of care. Patients in the outpatient clinics completed the assessment in an unoccupied room at their convenience, usually right before or after their appointment with their doctor. Inpatients completed the assessment in their hospital rooms or an unoccupied room. Technical assistance or assistance with filling in the questionnaire was provided on the patient's request. Assistance with filling in implied that a health care professional or family member handled the computer. The questions were read to the patient, and the assistant recorded the patient's answers. The research nurse/research assistant documented the need for assistance directly on the computer after the assessment. The computers automatically registered the time expenditure for the completion of each item. Encrypted backups were made on Universal Serial Bus (USB) memory sticks after each assessment. Every two weeks, these USB memory sticks were sent by mail to the coordination office in Trondheim for data extraction.

Ethical Considerations 

The study was conducted according to the guidelines of the Helsinki Declaration. Data were stored according to regulations set forth by the Norwegian Social Science Data Services. The Regional Committee for Medical Research Ethics, Health Region IV, in Norway, approved the study. Written informed consent was obtained from all participants.

Statistical Considerations 

Pearson's Chi-squared and independent-sample t-tests were performed to explore differences between patients requiring and patients not requiring assistance. A multivariable backward stepwise logistic regression analysis was performed to investigate which factors predicted the patients' need for assistance. Need for assistance was the dependent variable. Multivariable linear regression analyses were performed to explore which factors influenced time expenditure. Time expenditure was the dependent variable. Age, gender, educational level, KPS, and cognitive function score were independent covariates/variables in both regression analyses. Registrations with missing cognitive function scores were omitted. A P-value of 0.01 or less was considered statistically significant, and those more than 0.01 and 0.05 or less were considered to approach statistical significance. The Wilson interval was used for confidence intervals (CIs) in Fig. 1, Fig. 2.²² All analyses were performed with SPSS version 16 for Mac (SPSS Inc., Chicago, IL).

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1 

  Percentage of population within each age group that required assistance when filling in the computerized symptom and mobility assessment.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 2 

  Percentage of population within each Karnofsky Performance Status group that required assistance when filling in the computerized symptom and mobility assessment.

Back to Article Outline

Results 

Study A 

Sixty-one patients attended either the inpatient or outpatient clinic during the four weeks of inclusion. Thirty-five patients were not approached by the health personnel at the clinics who were responsible for asking about willingness to participate. Reasons for not approaching were perceived cognitive impairment, shortage of technical support from the research team, time constraints, and the nursing staff's perception of the patients' situation, reflected in statements, such as “the patient is too ill,” “the patient has too much pain today,” and “the patient is new on the ward.” Of the 26 approached patients (11 patients from the inpatient clinic and 15 from the outpatient clinic), six patients did not want to participate (three from each clinic). Thus, a total of 20 patients were recruited, six women and 14 men, with a mean age of 61 years (range 32–81) and a mean KPS of 70 (range 40–100).

Nine patients had no previous experience with computers. Eleven preferred the computer assessment, five had no preference, and three would have preferred P&P (one did not answer). Thirteen patients regarded the questions as relevant for their situation. None reported problems reading the text, but three had problems hitting the on-screen buttons. One person found the computerized pain body map difficult to understand. Patients were assisted by relatives in three cases. Several patients described filling in the computerized assessment as “fun.” The feedback from the patients resulted in enlargement of the on-screen buttons for the next data collection (Study B).

Study B 

Patient characteristics from Study B are listed in Table 1. The data collection yielded 690 complete registrations from 370 individual patients, 195 (52.7 %) men and 175 (47.3%) women. There was no difference in gender distribution (P=0.298). Eighty-one patients completed multiple registrations, ranging from two to 19 registrations each. A flowchart of the patient recruitment and completion of assessments is presented in Fig. 3. Mean age was 63 years (range 32–88), and mean KPS was 70 (range 20–100) (Table 1).

Table 1. Patient Characteristics (n=370)

Variables	All patients
Gender, n (%)
Men	195 (52.7)
Women	175 (47.3)
Age at assessment
Mean (range)	63 (32–88)
Highest level of education, n (%)
≤9 years	111 (30.0)
10–12 years	152 (41.1)
≤4 years at college/university	66 (17.8)
>4 years at college/university	41 (11.1)
Requirement of assistance when filling in the assessment, n (%)
Assistance required	86 (23.2)
No assistance required	284 (76.8)
Karnofsky Performance Status
Mean (range)	70 (20–100)
Cognitive function
Mean (range)	7.3 (3–8)

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 3 

  Flowchart of the patient recruitment and completion of assessments. CF = cognitive function.

Eighty-six patients (23.2%) required assistance. They were significantly older (mean age: 68 vs. 62 years, P<0.001), had significantly lower KPS (59 vs. 73, P<0.001), and had lower cognitive function scores (6.6 vs. 7.5, P<0.001). There were no differences in gender or educational level (P=0.249 and 0.084, respectively).

The results from the multivariable logistic regression analysis on requirement of assistance are presented in Table 2. Predictors for requirement of assistance were age (odds ratio [OR]: 1.60, 95% CI: 1.27–2.01, P<0.001) and KPS (OR: 0.60, 95% CI: 0.49–0.71, P<0.001). Cognitive function score approached significance (OR: 0.81, 95% CI: 0.70–1.00, P=0.014), whereas gender and education were not significant (P=0.308 and 0.707, respectively). The requirement of assistance in relation to age and KPS is presented in Fig. 1, Fig. 2, respectively. No specific cutoff point for requiring assistance was observed for age or KPS.

Table 2. Logistic Regressions on Factors Predicting the Requirement of Assistance When Completing a Computer-Based Symptom and Mobility Assessment for Palliative Care Cancer Patients

Covariates	No Assistance Required (n=284)	Assistance Required (n=86)	OR (95% CI)	P-value of OR
Univariable regression
Age mean (SD)	62 (10.9)	68 (10.5)	1.60 (1.27–2.01)a	<0.001
Women	139 (49)	36 (42)	1.33 (0.82–2.17)b	0.250
Educational level, n (%)				0.089
≤9 years (0)	76 (27)	35 (41)	1
10–12 years (1)	123 (43)	29 (34)	0.51 (0.29–0.90)
≤4 years at college/university (2)	51 (18)	15 (17)	0.64 (0.32–1.29)
>4 years at college/university (3)	34 (12)	7 (8)	0.45 (0.18–1.11)
Karnofsky Performance Status, mean (SD)	73 (13.8)	59 (17.6)	0.57 (0.48–0.57)c	<0.001
Cognitive function, mean (SD)	7.5 (1.2)	6.6 (1.8)	0.70 (0.60–0.82)d	<0.001
Multivariable regression
Age, mean (SD)	62 (10.9)	68 (10.5)	1.60 (1.25–2.05)a	<0.001
Women	139 (49)	36 (42)	1.34 (0.77–2.34)b	0.308
Educational level, n (%)				0.707
≤9 years (0)	76 (27)	35 (41)	1
10–12 years (1)	123 (43)	29 (34)	0.73 (0.38–1.39)
≤4 years at college/university (2)	51 (18)	15 (17)	0.83 (0.36–1.88)
>4 years at college/university (3)	34 (12)	7 (8)	0.59 (0.20–1.71)
Karnofsky Performance Status, mean (SD)	73 (13.8)	59 (17.6)	0.60 (0.49–0.71)c	<0.001
Cognitive function, mean (SD)	7.5 (1.2)	6.6 (1.8)	0.81 (0.70–1.00)d	0.014

The mean time expenditure for all patients was 15 minutes and 27 seconds (95% CI: 14 minutes and 46 seconds to 16 minutes and nine seconds). There was a statistically significant difference in time expenditure between patients not requiring assistance compared with those who required assistance—14 minutes and 58 seconds (95% CI: 14 minutes and 11 seconds to 15 minutes and 45 seconds) vs. 17 minutes and five seconds (95% CI: 15 minutes and 41 seconds to 18 minutes and 29 seconds) (P=0.01). Uni- and multivariable linear regression analyses on time expenditure are presented in Table 3. For patients not requiring assistance, the multivariable analysis showed that age, education, and KPS significantly influenced time expenditure, with P<0.001, P=0.006, and P<0.001, respectively. Patients spent one minute and 20 seconds (95% CI: 41 seconds to one minute and 59 seconds) more per 10-year increase in age, one minute and six seconds (95% CI: 19 seconds to one minute and 53 seconds) more per level of lower education, and one minute and 18 seconds (95% CI: 45 seconds to one minute and 51 seconds) more per 10-point lower KPS. Gender and cognitive function score did not influence time expenditure (P=0.055 and 0.919, respectively). For patients requiring assistance, only age influenced the time expenditure, with one minute and 58 seconds more per 10-year increase in age (95% CI: 44 seconds to three minutes and 12 seconds, P=0.002). Gender, educational level, KPS, and cognitive function score did not influence the time expenditure (P=0.701, 0.363, 0.425, and 0.889, respectively).

Table 3. Linear Regression Analyses on Time Expenditure When Completing a Computer-Based Symptom and Mobility Assessment for Cancer Patients Receiving Palliative Care

Variables	Regression Coefficient
	Estimate (Seconds)	95% CI	P-value
Patients not requiring assistance (n=284)
Univariable analyses
Agea	94	52–37	<0.001
Womenb	91	−2 to 185	0.056
Education	−89	−138 to −41	<0.001
Karnofsky Performance Statusc	−87	−120 to −55	<0.001
Cognitive functiond	−41	−79 to −3	0.036
Multivariable analyses
Agea	80	41–119	<0.001
Womenb	87	−2 to 175	0.055
Education	−66	−113 to −19	0.006
Karnofsky Performance Statusc	−78	−111 to −45	<0.001
Cognitive functiond	−2	−40 to 36	0.919
Patients requiring assistance (n=86)
Univariable analyses
Agea	120	43–197	0.002
Womenb	236	−148 to 195	0.785
Education	−49	−138 to 39	0.268
Karnofsky Performance Statusc	−15	−63 to 33	0.529
Cognitive functiond	3	−44 to 49	0.912
Multivariable analyses
Agea	118	44–192	0.002
Womenb	32	−134 to 199	0.701
Education	−41	−131 to −49	0.363
Karnofsky Performance Statusc	−19	−68 to −29	0.425
Cognitive functiond	3	−43 to 50	0.889

Back to Article Outline

Discussion 

This study demonstrates that most of the cancer patients receiving palliative care included in this study managed to use the computer without assistance. However, the proportion of patients requiring assistance rose steadily with increasing age and decreasing functional status. In addition, the pilot study suggests that patients either preferred computers to the P&P format or had no preference.

To our knowledge, the need for assistance when using a touchscreen assessment tool in cancer patients receiving palliative care has not been explored previously. Contrary to our expectations, more than three-quarters of the patients managed to fill in the assessment themselves. Higher age and lower performance status resulted in increased need of assistance. This indicates that there may be a limit when patients are no longer able to complete an assessment without assistance. However, there were no indications of distinct cutoff values when computerized assessment was no longer feasible.

Our results are in accordance with the findings of Fielding et al., showing that the proportion of missing P&P QoL forms in a similar population increases as the functional status declines.¹² Consequently, the inability to complete the assessment is not necessarily dependent on the computer itself, but rather on the patients' functional status. This highlights the need for short and user-friendly questionnaires. The length and precision of an assessment can be optimized by the use of CAT and item-response theory.²³, ²⁴ This makes the assessment tool shorter and more precise than current P&P formats and can, therefore, reduce the burden on patients with severe disease. Our instrument contained an abundance of items, because another purpose of the study was to test the psychometric properties of several candidate items for later versions.

The total time expenditure for answering the 60 items was relatively low, about 15 minutes for the patients not requiring assistance and 17 minutes for those who required assistance. On average, patients not requiring and patients requiring assistance used 15 and 17 seconds/item, respectively. Velikova et al. reported a mean time expenditure for general cancer inpatients of 11 seconds/item when completing a computer-based QoL questionnaire.¹⁴ This is faster than both groups in our study, which may be explained by the higher age and poorer performance status of the patients receiving palliative care compared with cancer patients in general. This falls in line with an increase in time expenditure in both groups of patients when age increases and for those not requiring assistance, also when level of education and KPS decrease.

We do not believe answering an equivalent number of items in a P&P format would be easier for a similar population. The items in the computerized tool were displayed one at a time with large fonts and large on-screen buttons. The use of touchscreens and electronic pens probably made the tool somewhat similar to the more familiar P&P format. A corresponding P&P solution would have numerous items in small writing on each page, requiring better vision and dexterity. In all, we believe a computer-based solution is more user-friendly and less confusing than a corresponding P&P assessment.

The generalizability of the present findings is limited for several reasons. First, few data are collected on patients with low KPS. This indicates that our material is selected toward the healthier part of the palliative care population. Low performance status implies a higher symptom burden, and these patients are likely to be perceived as frail. “Gate-keeping” by the health professionals, therefore, may be a significant factor. Results from the pilot study substantiate this; health professionals perceiving the patients as too ill was the main reason for not asking the patients to participate. This is problematic for two reasons: 1) the patients have an ethical right to be asked and to decline participation in studies and 2) because of a “healthy bias,” the symptom burden of the frailest patients was insufficiently assessed, which in turn limits the generalizability. On the other hand, there are ethical considerations related to overburdening very sick patients. Unfortunately, other studies researching computer-based QoL assessment in cancer have reported on the patients' performance status only to a limited degree. Second, we have no complete record of how many patients refused to participate, reasons for declining participation, or the characteristics of these patients. The patients had to be perceived as cognitively intact by the health professionals to be included in the study. This may explain why cognitive function did not predict the need for assistance or influence the time expenditure.

Further research on the design of the graphical user interface (GUI) of a computer-based tool for assessment of symptoms and physical function in palliative care cancer patients is warranted. Specifically, the GUI needs to be developed together with the patients to make it optimally user-friendly and acceptable to as many patients as possible. In addition, there is a need to develop a model that tailors the assessment to each individual patient based on symptom burden and functioning. This enables a tool to be as short and precise as possible.

Back to Article Outline

Conclusions 

Use of computers for self-report of symptoms and functioning is feasible in the population of cancer patients receiving palliative care. Most of the patients in the study were able to complete the assessments without assistance. However, older patients with lower performance status are more likely to need assistance from health professionals or relatives. This highlights the need for computerized assessments to be as short, precise, and user-friendly as possible to ensure good compliance and valid assessments in frail patients.

Back to Article Outline

Acknowledgments 

The authors thank Professor Stian Lydersen and Professor Peter M. Fayers for advice on the statistical analyses.

Back to Article Outline

Appendix. Items Presented to the Patients in the Computerized Assessment Tool 

(P)=pain items; (F)=physical functioning items; (D)=depression items; (S)=items on sociodemographic data; (O)=items on other prevalent symptoms.

Items are presented according to the order of appearance in the tool.

21.(F) Do you have difficulties with carrying more than 5kg a distance of 100 meters?

Back to Article Outline

References 

1. Coyle N, Adelhardt J, Foley KM, Portenoy RK. Character of terminal illness in the advanced cancer patient: pain and other symptoms during the last four weeks of life. J Pain Symptom Manage. 1990;5:83–93
   • View In Article
   • MEDLINE
   • CrossRef
2. Edmonds P, Karlsen S, Khan S, Addington-Hall J. A comparison of the palliative care needs of patients dying from chronic respiratory diseases and lung cancer. Palliat Med. 2001;15:287–295
   • View In Article
   • MEDLINE
   • CrossRef
3. Walsh D, Donnelly S, Rybicki L. The symptoms of advanced cancer: relationship to age, gender, and performance status in 1,000 patients. Support Care Cancer. 2000;8:175–179
   • View In Article
   • MEDLINE
   • CrossRef
4. Detmar SB, Muller MJ, Schornagel JH, Wever LD, Aaronson NK. Health-related quality-of-life assessments and patient-physician communication: a randomized controlled trial. JAMA. 2002;288:3027–3034
   • View In Article
   • MEDLINE
   • CrossRef
5. Higginson IJ, Carr AJ. Measuring quality of life: using quality of life measures in the clinical setting. BMJ. 2001;322:1297–1300
   • View In Article
   • MEDLINE
   • CrossRef
6. Velikova G, Booth L, Smith AB, et al. Measuring quality of life in routine oncology practice improves communication and patient well-being: a randomized controlled trial. J Clin Oncol. 2004;22:714–724
   • View In Article
   • MEDLINE
   • CrossRef
7. Caraceni A, Brunelli C, Martini C, Zecca E, De Conno F. Cancer pain assessment in clinical trials. A review of the literature (1999–2002). J Pain Symptom Manage. 2005;29:507–519
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (152 KB)
   • CrossRef
8. Meuser T, Pietruck C, Radbruch L, et al. Symptoms during cancer pain treatment following WHO guidelines: a longitudinal follow-up study of symptom prevalence, severity and etiology. Pain. 2001;93:247–257
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (152 KB)
   • CrossRef
9. Patrick DL, Ferketich SL, Frame PS, et al. National Institutes of Health State-of-the-Science Conference Statement: symptom management in cancer: pain, depression, and fatigue, July 15–17, 2002. J Natl Cancer Inst. 2003;95:1110–1117
   • View In Article
   • CrossRef
10. Von Roenn JH, Cleeland CS, Gonin R, Hatfield AK, Pandya KJ. Physician attitudes and practice in cancer pain management. A survey from the Eastern Cooperative Oncology Group. Ann Intern Med. 1993;119:121–126
    • View In Article
    • MEDLINE
11. Caraceni A, Cherny N, Fainsinger R, et al. Pain measurement tools and methods in clinical research in palliative care: recommendations of an expert working group of the European Association of Palliative Care. J Pain Symptom Manage. 2002;23:239–255
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (107 KB)
    • CrossRef
12. Fielding S, Fayers PM, Loge JH, Jordhoy MS, Kaasa S. Methods for handling missing data in palliative care research. Palliat Med. 2006;20:791–798
    • View In Article
    • MEDLINE
    • CrossRef
13. Velikova G, Brown JM, Smith AB, Selby PJ. Computer-based quality of life questionnaires may contribute to doctor-patient interactions in oncology. Br J Cancer. 2002;86:51–59
    • View In Article
    • MEDLINE
    • CrossRef
14. Velikova G, Wright EP, Smith AB, et al. Automated collection of quality-of-life data: a comparison of paper and computer touch-screen questionnaires. J Clin Oncol. 1999;17:998–1007
    • View In Article
    • MEDLINE
15. Wright EP, Selby PJ, Crawford M, et al. Feasibility and compliance of automated measurement of quality of life in oncology practice. J Clin Oncol. 2003;21:374–382
    • View In Article
    • MEDLINE
    • CrossRef
16. Clark L, Bardwell WA, Arsenault T, DeTeresa R, Loscalzo M. Implementing touch-screen technology to enhance recognition of distress. Psychooncology. 2009;18(8):822–830
    • View In Article
    • CrossRef
17. Kaasa S, Loge JH, Fayers P, et al. Symptom assessment in palliative care: a need for international collaboration. J Clin Oncol. 2008;26:3867–3873
    • View In Article
    • CrossRef
18. Hølen JC, Hjermstad MJ, Loge JH, Fayers PM. Pain assessment tools: is the content appropriate for use in palliative care?. J Pain Symptom Manage. 2006;32:567–580
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (169 KB)
    • CrossRef
19. Helbostad JL, Hølen JC, Jordhøy MS, et al. A first step in the development of an international self-report instrument for physical functioning in palliative cancer care: a systematic literature review and an expert opinion evaluation study. J Pain Symptom Manage. 2009;37:196–205
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (188 KB)
    • CrossRef
20. Karnofsky DA, Burchenal JH. The clinical evaluation of chemotherapeutic agents in cancer. In:  MacLeod CM editors. Evaluation of chemotherapeutic agents. New York: Columbia University Press; 1949;p. 196
    • View In Article
21. Fayers PM, Hjermstad MJ, Ranhoff AH, et al. Which Mini-Mental State Exam Items can be used to screen for delirium and cognitive impairment?. J Pain Symptom Manage. 2005;30:41–50
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (185 KB)
    • CrossRef
22. Wilson EB. Probable inference, the law of succession, and statistical inference. JASA. 1927;22:209–212
    • View In Article
23. Bichindaritz I, Marling C. Case-based reasoning in the health sciences: what's next?. Artif Intell Med. 2006;36:127–135
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (129 KB)
    • CrossRef
24. Fayers PM, Machin D. Quality of life: The assessment, analysis and interpretation of patient-reported outcomes. 2nd ed.. Hoboken, NJ: John Wiley & Sons; 2007;
    • View In Article