ORIGINAL ARTICLE
The Effect of Pain on Physical Functioning After Breast Cancer Treatment Development and Validation of an Assessment Tool Kenneth G. Andersen, MD,*w Karl B. Christensen, MSc, PhD,z Henrik Kehlet, MD, PhD,* and Pernille E. Bidstup, MSc, PhDy Objectives: Persistent postsurgical pain, musculoskeletal pain, sensory disturbances, and lymphedema are major clinical problems after treatment for breast cancer. However, there is little evidence on how these sequelae affects physical function. The aim this study was to develop and validate a procedure-specific tool for assessing the impact of pain and other sequelae on physical function after breast cancer treatment. Methods: A literature review, patient and expert interviews were used to identify dimensions of physical function and sequelae. A questionnaire was developed and tested using cognitive interviews, and field tested among 389 patients treated for primary breast cancer without recurrence (response rate 81%). Median follow-up was 14 months. Using item response theory we identified 5 cause scales of reduced physical functioning: pain after surgery, musculoskeletal pain, sensory disturbances, lymphedema, and other causes. Convergent validity was assessed using the QuickDsability of Arm, Shoulder, and Hand Scale (Q-DASH). Results: About half of the patients reported decreased physical function. All 5 scales displayed good fit, unidimensionality, monotonicity, local independence, and lack of differential item functioning. Cronbach coefficient a ranged from 0.88 (other causes) to 0.96 (sensory disturbances) for the 5 scales. For the Q-DASH a was 0.91. Each scale revealed different information on causes of reduced function. Discussion: The present scales displayed good psychometric qualities, and may be used to evaluate the impact of specific sequelae after breast cancer treatment on physical functioning, as well as to monitor and target interventions to optimize pain treatment and rehabilitation. Key Words: breast cancer, physical function, sequelae, persistent pain, neuropathy, lymphedema, Rasch analysis Received for publication April 26, 2014; revised October 1, 2014; accepted August 28, 2014. From the *Section for Surgical Pathophysiology; wDepartment of Breast Surgery, Rigshospitalet, University of Copenhagen; zDepartment of Biostatistics, Institute of Public Health, University of Copenhagen; and yDepartment of Survivorship, Danish Cancer Research Center, Danish Cancer Society, Copenhagen, Denmark. Supported by a grant from the Danish Cancer Society, Copenhagen, Denmark, and the study is part of the Europain Collaboration, which has received support from the Innovative Medicines Initiative Joint Undertaking, under grant agreement no 115007, resources of which are composed of financial contribution from the European Union’s Seventh Framework Programme (FP7/20072013) and EFPIA companies’ kind contribution. http://www. imieuropain.org/. The authors declare no conflict of interest. Reprints: Kenneth G. Andersen, MD, Section for Surgical Pathophysiology, 4074, Rigshospitalet, University of Copenhagen, Blegdamsvej 9, Copenhagen DK-2100, Denmark (e-mail: [email protected]). Supplemental Digital Content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Website, www.clinicalpain.com. Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. DOI: 10.1097/AJP.0000000000000156
(Clin J Pain 2015;31:794–802)
S
urvival after breast cancer treatment has increased over the past years, and 5-year relative survival is now 80% to 90% in the Nordic countries and North America,1 emphasizing the need for more attention on late sequelae and their rehabilitation. Such sequelae are common, including persistent pain, sensory disturbances, lymphedema, reduced arm function, and side effects of medications, all of which may impair physical function, psychological well-being, and quality of life. Persistent pain after breast cancer treatment (PPBCT) affects between 25% and 60% of the patients depending on treatment,2–4 but the effect of PPBCT on functional impairment per se is not well described. Tightly connected to PPBCT are sensory disturbances, which is frequent in the area of surgery2 as well as in hands and feet5 due to chemotherapy-induced neuropathy. Sensory disturbances such as allodynia and hyperalgesia may cause discomfort when wearing clothes, handling small objects, and in affected feet cause problem during walk. The prevalence of lymphedema among patients with breast cancer was 15% to 30% in a large meta-analysis,6 and may substantially affect both physical and psychological aspects of quality of life.7,8 Reduced arm mobility due to pain or fibrotic tissue in the axilla and musculoskeletal pain problems as a side effect of antihormonal treatment are also common and may contribute to reduced physical functioning.9 Unfortunately, many of these problems are not always acknowledged by health care providers.10 Within the patient-reported measures, the construct of functioning—and especially physical functioning— play a key role as it may capture the impact and severity of the sequelae, which otherwise may be perceived differently by the individual patients. To make progress in preventive and interventional strategies for alleviating sequelae, there is a need for a validated specific and sensitive tool to describe the specific sequelae causing reduced physical function.11 The aim of this study was to construct and validate such a tool.
MATERIALS AND METHODS Theoretical Model Physical function may be understood in a biopsychosocial context, with interactions between disease or disorder, impairment of body functions and structures, reduction of ability to perform daily activities, participation in work and social activities, environmental factors, and patient-related factors.12 At the core of this context, ability to perform daily activities is the most concrete outcome and simple to measure. Either a patient can perform an activity, perform it despite difficulties, or the activity cannot be performed. Inability to perform daily activities may be caused by several different factors including breast cancer–specific
794 | www.clinicalpain.com Clin J Pain Volume 31, Number 9, September 2015 Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clin J Pain
Volume 31, Number 9, September 2015
sequelae or other reasons such as comorbidity. Thus, it is possible not only to measure whether or not there is a functional impairment, but also what the underlying cause is.
Literature Review Using searches in PubMed as well as reviews2,13,14 on the area, the literature was examined for available scales and questionnaires used in the assessment of PPBCT, QoL, and functional impairment of upper extremities; items and domains were entered in an item database, and labeled according to the construct each item was designed to measure. This was conducted to get an overview of the dimensions of sequelae and possible constructs.
Semistructured Interviews On the basis of the identified dimensions in the literature search, an interview guide was developed. Semistructured interviews were conducted to assure that all relevant dimensions were covered. Number of patients needed to interview was decided by sampling to redundancy. Ten breast cancer patients covering all treatment modalities were recruited from the breast surgery and breast oncology departments and 10 health care providers from the breast surgery, breast oncology, and rheumatology departments were interviewed. All interviews were recorded and transcribed pragmatically extracting dimensions and items, which were added to the item database.
Construction of a Preliminary Questionnaire The questionnaire for assessing sequelae after breast cancer treatment was constructed by single items covering persistent pain, sensory disturbances, lymphedema, and items covering physical function intended for scale development. Items concerning persistent pain and lymphedema were extracted from a previous study,3,15 in which the content validity had been examined through personal patient interviews. Items on sensory disturbances were inspired and developed from a review specifically examining scales relating to neuropathic pain.16 All items entered in the database were assessed, and then synthesized into the questionnaire to cover the identified dimensions.
Construction and Testing of Content Validity The preliminary questionnaire was then sequentially tested among breast cancer patients and health care providers by cognitive interviews of understanding and relevance of each item. After each dimension, the participants were asked if the questions covered all areas. Additional items were supplemented where needed. After interviewing 8 breast cancer patients and 8 health care providers, no additional information was provided. Two additional breast cancer patients were interviewed and finally the eighth revision was considered ready to be tested in a larger population. Twenty items regarding ability to perform daily activities, and reasons for reduced ability were included for Rasch analysis. Furthermore, the questionnaire consists of items related to sequelae, but not included in the Rasch analysis. (Supplementary Questionnaire, Supplemental Digital Content 1, http://links.lww.com/CJP/A131).
Convergent Validity As a measure for upper arm function, the QuickDisability of Arm, Shoulder, and Hand (Q-DASH) questionnaire was choosen17 to assess convergent validity. The Q-DASH is a validated shortened version of the DASH Outcome Measure, with 11 items to measure impaired Copyright
r
Assessing Physical Function After Breast Cancer Treatment
physical function and symptoms in the upper body, and is validated in the breast cancer population.18 The DASH has been widely used in the breast cancer population, and is the recommended scale to measure upper body after breast cancer treatment.19 Furthermore, convergent validity was examined with correlations with items of self-assessed pain, sensory disturbances, and lymphedema developed in parallel to the cause scales.
Testing in a Breast Cancer Population The study was performed in a cohort of consecutive breast cancer survivors treated at Rigshospitalet between August 13, 2009 and July 30, 2010, using a cross-sectional design. Data were collected in April 2011 at Rigshospitalet, Copenhagen, Denmark. Patients eligible for inclusion were primary breast cancer patients undergoing treatment according to the standard national Danish treatment protocol. Exclusion criteria were: death, disease in the nervous system, bilateral disease, recurrence, or other malignancy. Disease characteristics were provided by the Danish Breast Cancer Cooperative Group (DBCG), which prospectively collects clinical and pathologic data,20,21 and includes details on type of surgery of breast and axilla, radiation therapy, chemotherapy, and endocrine treatment. The study was performed in accordance with Danish law (The Committees on Health Research Ethics for the Capital Region of Denmark, H-4-2013-FSP-045 and the Danish Data Protection Agency 2007-58-0015/30-0545).
Statistics We addressed a specific type of construct validity, called criterion-related construct validity, which concerns the relationship between the index, the individual items included in the index, and exogenous variables, which characterize the population such as age and prognostic or treatment-related factors. For each domain the fit of the items to an item response theory (IRT) model was examined.22 The simplest IRT model, the Rasch model23–25 was chosen. All items were assessed using the Rasch model, with a systematic and iterative process discarding items not fitting the model. In the Rasch model the sum of all items is a sufficient statistic for the underlying latent variable, and this means that when a scale fits the Rasch model, the simple summation of the items within this scale can be used as a measure of the underlying latent variable. The analysis assessed overall fit of the model by Andersen conditional likelihood ratio test,26 testing homogeneity of 2 score groups. The fit of each individual item to the Rasch model was assessed by comparison of observed and expected correlations between separate item scores and sum of the remaining items.27 Briefly, this item fit statistic evaluates whether the observed rank correlation between the item and the sum of the remaining items in the scale differs from the expected value correlation. The fit of each individual item to the Rasch model was also assessed by conditional infit and outfit test statistics.28 These compare observed and expected item responses weighted together as a sum of squared residuals. We tested local dependence and the absence of differential item functioning (DIF) using loglinear Rasch models.29 DIF was examined with respect to the variables chemotherapy (yes/no), follow-up time (under/over 12 mo), age (over/under 60 y), and axillary surgery (sentinel lymph node biopsy or axillary clearance). Additional dichotomizations of the variable “follow-up time” (10, 14, and 16 mo) were also evaluated. Analyses were carried out using the DIGRAM software (Department of Biostatistics, University of Copenhagen, Denmark). The significance of
2014 Wolters Kluwer Health, Inc. All rights reserved.
www.clinicalpain.com |
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
795
Clin J Pain
Andersen et al
Volume 31, Number 9, September 2015
Daily activities We are interested to know how treatment for breast cancer affects your daily activities. Think about how last week had been for you in relation to the activites below: If you have difficulty, experiencing discomfort, or simply just cannot do some of these activities, please describe your difficulties in the box provided below each question.
Scoring of the cause scales:
Scoring stage 1:
Do laundry Not relevant, I don’t do this = 0 Not relevant, I don’t do this
I can do this without any problem = 0 I can do this without any problem I can do this, but with difficulties because (please explain)
I can do this, but with difficulties = 1
I can’t do this due to
I can’t do this due to = 2
(please explain)
Scoring stage 2: Yes = multiply score from stage 1 with 1 No = multiply score from stage 1 with 0
Is the cause of your difficulty due to: Pain in the area you had surgery? Muscle or joint pain? Sensory disturbances, such as pins and needles, sleeping or stabbing sensations? Swelling (lymphedema) Other cause
Yes Yes
No No
Yes
No
Yes Yes
No No
Example: Pain after surgery scale: 2 x 1 = 2 Musculoskeletal pain scale: 2 x 0 = 0 Sensory disturbance scale: 2 x 1 = 2
Description:___________________________________
Lymphedema scale: 0 x 2 = 0 Other causes scale: 0 x 2 = 0
Scores are summated in the 5 cause scales and transformed to a 0-100 scale.
FIGURE 1. Scoring of the cause scales with an example.
item fit statistics was assessed by the Benjamini-Hochberg procedure controlling the false discovery rate at 0.05. Cronbach a-coefficients30,31 were calculated as measures of reliability, and values between 0.7 and 0.8 were regarded as satisfactory.32 Criterion validity related to Q-DASH scores and measures of pain, lymphedema, and sensory disturbances were evaluated using Spearman rank correlations.
Scoring and Construction of Scales Each of the 20 items included in the initial item set had 2 scoring stages: an initial rating of difficulty or discomfort experienced in relation to the activity (response options “Not relevant, I don’t do this,” “I can do this without any problem,” “I can do this, but with difficulties,” and “I can’t do this”) and a secondary rating where respondents attributed the difficulty or discomfort to either pain after surgery, musculoskeletal pain, sensory disturbances, lymphedema, or other causes, respectively. On the basis of these responses, a total of 100 items (20 in each dimension) were coded as outlined in Figure 1. The scores were summated for each of the 5 scales, and transformed into a 0 to 100 scale.
RESULTS Participants: The Breast Cancer Test Population Between August 13, 2009 and July 30, 2010, 500 consecutive patients were identified in DBCG database as eligible. One patient had died, 3 were not reachable at the address, and 2 had migrated. Of 494 questionnaires, 402 questionnaires were returned (response rate: 81%). A total of 13 patients were excluded: 4 patients returned the questionnaire and stated they did not wish to participate, 3 were excluded due to mainly unanswered questionnaires, 2 with
metastatic disease, 1 with hemiparesis after apoplexia, 1 with disseminated sclerosis, 1 with reconstructive surgery, and 1 with a recent shoulder fracture. Final analysis included 389 patients. Patient characteristics are presented in Table 1. There were no differences in age, follow-up, or treatment among responders and nonresponders. Prevalence of pain, sensory disturbances, lymphedema, and function related to treatment groups are presented in Supplementary Tables 1 to 3 (Supplemental Digital Content 2–4, http://links.lww.com/ CJP/A132, http://links.lww.com/CJP/A133, http://links. lww.com/CJP/A134). The frequencies of patients affected according to the cause scales and the median scores are shown in Table 2, and percentages of the population affected by each activity are shown in Figure 2.
The Cause Scales: Rasch Analysis The cause scales “pain after surgery,” “musculoskeletal pain,” “sensory disturbances,” “lymphedema,” and “other causes” were constructed from the 20 general items and their accompanying cause item, using item reduction. The items included in each scale are shown in Table 3, together with the item fit statistics, which indicate an overall good fit across all the 5 proposed scales. Two items concerning difficulties with “sexual activity” and “getting out of bed” were not included in any scale. DIF with regard to surgery in the axilla, chemotherapy, age, or follow-up time, was not found for the any of 5 scales (Supplementary Tables 4–8, Supplemental Digital Content 5–9, http://links.lww.com/CJP/A135, http://links. lww.com/CJP/A136, http://links.lww.com/CJP/A137, http:// links.lww.com/CJP/A138, http://links.lww.com/CJP/A139). Regarding local dependence no clinically relevant evidence was disclosed for the 5 proposed scales. The scales fitted the Rasch model, and thus supported the use of simple
796 | www.clinicalpain.com Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clin J Pain
Volume 31, Number 9, September 2015
Assessing Physical Function After Breast Cancer Treatment
TABLE 1. Population Characteristics of Breast Cancer Patients Enrolled in Interviews and Questionnaire Study
Breast Cancer Test Population N (%) Age (median, IQR) (y) Follow-up (median, IQR) (mo) Surgical treatment (N [%]) BCS + SLNB BCS + ALND Ma + SLNB Ma + ALND Adjuvant treatment (N [%]) Chemotherapy Yes No Radiation therapy No radiation therapy BRT AT/LRRT Endocrine therapy No endocrine therapy Tamoxifen Aromatase inhibitors
Semistructured Interview
Questionnaire Interview
Responders
Nonresponders
Total
10 (100) 57 (54-59) 18 (12-44)
10 (100) 65 (60-73) 27 (23-56)
389 (81) 60 (53-66) 14 (11-17)
92 (19) 59 (50-67) 15 (11.5-17)
481 (100) 60 (52-66) 14 (11-17)
179 99 34 77
38 19 10 25
217 118 59 87
2 2 2 4
(20) (20) (20) (40)
2 2 2 3
(20) (20) (20) (30)
(46) (25) (9) (20)
(41) (21) (11) (27)
(45) (25) (12) (18)
9 (90) 1 (10)
5 (50) 4 (40)
186 (48) 203 (52)
52 (47) 40 (43)
238 (49) 243 (51)
3 (30) 2 (20) 5 (50)
4 (40) 3 (30) 2 (20)
44 (11) 208 (54) 137 (35)
11 (12) 41 (45) 40 (43)
55 (11) 249 (52) 177 (37)
1 (10) 2 (20) 6 (60)
1 (10) 4 (40) 5 (50)
104 (27) 59 (15) 226 (58)
26 (28) 15 (16) 51 (56)
130 (27) 74 (15) 277 (58)
Population characteristics for the patients enrolled in the semistructured interviews, the questionnaire interviews, and the breast cancer population. Follow-up: time from surgery to participation in interview or answering the questionnaire. Missing data: semistructured interview—endocrine therapy: 1. Questionnaire interview—treatment, surgery: 1, chemotherapy: 1. radiation therapy: 1. ALND indicates axillary lymph node dissection; AT/LRRT, anterior thoracic wall radiation therapy and locoregional radiation therapy corresponding to periclavicular, axillary level 3 and for right-sided cancer the internal mammary lymph nodes; BCS, breast conservative surgery; BRT, breast radiation therapy corresponding to residual breast tissue; IQR, interquartile range; SLNB, sentinel lymph node biopsy.
summation of the items within each scale.23 Andersen conditional likelihood ratio test demonstrated homogeneity of the scales: pain after surgery, P = 0.67; musculoskeletal pain, P = 0.06; sensory disturbances, P = 0.44; lymphedema, P = 0.09; and other causes, P = 0.62. The correlations between the cause scales and items in the questionnaire are shown in Table 4. Means, SD and range for the 5 scales are shown in Table 5.
The Cause Scales: Reliability Analysis Cronbach coefficient a indicated high reliability for all scales: “pain after surgery,” a = 0.89; “musculoskeletal pain,” a = 0.92; “sensory disturbances,” a = 0.96; “lymphedema,” a = 0.91; and “other causes,” a = 0.88.
Q-DASH Validity and Correlation to Cause Scales The number of missing responses was low for most items; Q-DASH question 6 (recreational activities) and 10 (tingling, pins, and needles in arm, shoulder, and/or hand) having the highest number of missing responses. Cronbach a was 0.91 and the interitem correlations ranged from 0.27 to 0.77. The item fit was poor (Supplementary Table 9, Supplemental Digital Content 10, http://links.lww.com/ CJP/A140). There was evidence of local dependence for 10 of the 55 item pairs (results not shown). Regarding DIF, this was found for question 1 (open a tight or new jar) for follow-up time and for questions 8 (limitations in work or other daily activities) and 9 (arm, shoulder, or hand pain) for axillary surgery (P < 0.01). Q-DASH correlation with the cause scales ranged from 0.35 to 0.6, lowest for the sensory disturbances scale and highest for the musculoskeletal pain scale (Table 4). Copyright
r
DISCUSSION A thorough development process including a literature review and patient and expert interviews in 2 stages resulted in an initial item bank of 20 items on physical functioning related to daily activities in patients after breast cancer treatment. We grouped items based on the patient-reported causes of reduced physical functioning (including pain in the area of surgery, musculoskeletal pain, sensory disturbances, lymphedema, and other causes) resulting in 5 cause scales. Initial validation of the cause scales indicated that the common criteria for validity, monotonicity, unidimensionality, homogeneity, local independence, and lack of DIF were all achieved. Internal consistency expressed as Cronbach a was high ranging from 0.88 to 0.96. Furthermore, the cause scale scores correlated satisfactorily with other items in the questionnaire measuring corresponding dimensions. Thus, overall psychometric properties for all scales indicate proof of concept as a measuring tool of the causes of reduced physical function in the breast cancer population. The approach of scoring reduced physical function through 2 steps is novel, and the first attempt to provide information as to what causes reduced physical function— whether it is breast cancer–specific sequelae or other factors such as comorbidity. Patient-reported outcome measurements assessing physical functioning often applied in the breast cancer population include general quality of life scales such as SF-3633 and quality of life scales with diseasespecific and symptom-specific modules such as the EORTC c3034 with BR2335 and FACT-G with FACT-B.36 SF-36 measures ability to perform daily activities, but we have not been able to find specific data of psychometric properties in a breast cancer population. EORTC c30/BR23 and FACTB measure several dimensions of Health-related Quality of
2014 Wolters Kluwer Health, Inc. All rights reserved.
www.clinicalpain.com |
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
797
8 (4) 171 (96) 12.24 (6.2522.65)
12 (7) 167 (93) 11.81 (5.9023.61)
50 (28) 129 (72) 15 (9.0925)
56 (28) 147 (72) 13.07 (7.29-25)
94 (51) 92 (49) 8.33 (4.5516.67)
38 (20) 148 (80) 16.03 (9.0926.92)
15 (7) 193 (93) 10 (4.1723.08)
17 (8) 191 (92) 11.11 (6.25-25)
60 (29) 148 (71) 15 (9.0926.39)
21 (48) 96 (46) 23 (52) 112 (54) 5.56 (4.55- 4.69 (4.5510) 16.67)
3 (7) 41 (93) 7.69 (4.1746.15)
AT/LRRT
71 (52) 66 (48) 10 (5-25)
33 (24) 104 (76) 16.67 (11.54-25)
18 (13) 119 (87) 19.05 (11.1135.71)
48 (35) 89 (65) 25 (8.7136.93)
38 (18) 40 (29) 170 (82) 97 (71) 9.09 (5.56- 13.07 (8.3315) 34.09)
BRT
Radiotherapy
44 (42) 60 (58) 11.25 (519.09)
9 (9) 95 (91) 16.67 (1023.08)
9 (9) 95 (9) 12.50 (11.11-30)
26 (25) 78 (75) 20.42 (9.0933.33)
22 (21) 82 (79) 13.07 (6.2531.82)
37 (63) 22 (37) 7.69 (4.5511.54)
13 (22) 46 (78) 12.50 (11.5420.83)
6 (10) 53 (90) 13.89 (1016.25)
18 (31) 41 (69) 10 (4.5527.27)
20 (34) 39 (66) 10 (6.2517.92)
TAM
AI
107 (47) 119 (53) 8.33 (4.5518.18)
29 (13) 197 (87) 16.67 (7.6926.92)
26 (12) 200 (88) 16.67 (5.5628.57)
72 (32) 154 (68) 16.67 (8.3334.17)
41 (18) 185 (82) 10 (4.5527.27)
Endocrine Therapy None
188 (48) 201 (52) 8.33 (4.5518.18)
51 (13) 338 (87) 15.38 (8.33-25)
41 (11) 348 (89) 16.67 (6.2528.57)
116 (30) 272 (70) 16.67 (8.7133.33)
83 (21) 306 (79) 10 (5.5627.27)
Total
Number and percentages of patients scoring 0 (indicated as “no”) and above 0 (indicated as “yes”) and median scores of patients scoring above 0. AI indicates aromatase inhibitors; ALND, axillary lymph node dissection; AT/LRRT, anterior thoracic wall radiation therapy and locoregional radiation therapy corresponding to periclavicular, axillary level 3 and for right-sided cancer the internal mammary lymph nodes; BCS, breast conservative surgery; BRT, breast radiation therapy corresponding to residual breast tissue; IQR, interquartile range; SLNB, sentinel lymph node biopsy; TAM, tamoxifen.
94 (46) 109 (54) 9.09 (518.18)
13 (6) 190 (94) 11.54 (8.3319.23
6 (14) 38 (86) 15.71 (6.2522.22)
8 (18) 36 (82) 9.55 (4.7731.06)
5 (11) 39 (89) 27.27 (4.1727.27)
None
46 (46) 53 (54) 8.33 (5.015.38)
22 (22) 77 (78) 16.67 11.5422.73)
21 (21) 165 (89) 22.22 (11.1135.71)
60 (32) 126 (68) 20.83 (9.0941.88)
32 (16) 51 (27) 171 (84) 135 (73) 9.09 (4.55- 13.64 (8.33-40) 15.91)
Chemotherapy
Chemotherapy None
18 (18) 20 (10) 81 (82) 183 (90) 21.83 (7.14- 9.13 (5.5637.50) 21.83)
38 (38) 61 (62) 20.42 (8.3338.89)
29 (29) 71 (71) 18.18 (1041.67)
ALND
BCS
30 (17) 149 (83) 8.71 (4.55 13.64)
SLNB
Clin J Pain
43 (56) 83 (46) 34 (44) 96 (54) 10 (4.55-25) 8.33 (4.1718.18)
16 (47) 18 (53) 5.28 (4.3612.69)
Other Yes (N [%]) No (N [%]) Scores of those affected (median [IQR])
7 (9) 70 (91) 11.11 (5.5616.67)
22 (29) 55 (71) 29.29 (13.6440.91)
20 (26) 57 (74) 9.55 (6.2529.55)
2 (6) 19 (25) 32 (94) 58 (75) 26.92 (7.69- 15.38 (8.3346.15) 27.27)
4 (12) 30 (88) 21.83 (15.7136.11)
6 (18) 28 (82) 9.55 (5.016.67)
4 (12) 30 (88) 27.27 (15.7238.64)
ALND
Mastectomy
SLNB
Lymphedema Yes (N [%]) No (N [%]) Scores of those affected (median [IQR])
Sensory disturbances Yes (N [%]) No (N [%]) Scores of those affected (median [IQR])
Musculoskeletal pain Yes (N [%]) No (N [%]) Scores of those affected (median [IQR])
Pain in area of surgery Yes (N [%]) No (N [%]) Scores of those affected (median [IQR])
Cause Scales
TABLE 2. Physical Function in a Breast Cancer Population by Surgical and Adjuvant Treatment Groups
Andersen et al Volume 31, Number 9, September 2015
798 | www.clinicalpain.com Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clin J Pain
Volume 31, Number 9, September 2015
Assessing Physical Function After Breast Cancer Treatment
FIGURE 2. Percentages of patients with reduced capacity to perform daily activities by cause. Numbers in rows are percentages of total population.
Life (HRQoL), where physical function is expressed as physical well-being. The present cause scales measure a different and more concrete dimension of physical functioning, using the ability to perform well-defined daily activities. Generic pain impairment questionnaires, such as the Brief Pain Inventory,37 may not discern unspecific pain reports from specific PPBCT. The use of procedure-specific questionnaires to measure pain-related physical impairment is supported after other surgical interventions such as hernia repair38 and thoracic surgery.39 Scales measuring upper body morbidity such as the Q-DASH have been validated18; however, there is a need for further validation.19 A previous study among patients with upper limb disorder indicated problems with unidimensionality in the Q-DASH.40 Furthermore, our psychometric analyses of the Q-DASH among women with breast cancer indicated multidimensionality, poor item fit, as well as local dependence and DIF. Summarizing, the present cause scales are distinct from existing scales and fill a gap in measuring physical function and sequelae in the breast cancer population. Fit of data to an IRT model guarantees that items are appropriate for assessment of a single underlying variable. The benefits of using IRT methods compared with classic psychometric theory are summarized by the IMMPACT initiative.41 IRT models exhibit a number of desirable properties, namely: (1) unidimensionality, (2) monotonicity, (3) local independence, and (4) lack of DIF.42 These characteristics are often neglected, but may identify serious measurement problems in the scales. Unidimensionality is a key component in summarizing scales, and multidimensionality may introduce Copyright
r
bias. The potential implications of departures from the Rasch model such as local dependence and DIF include poor power and bias, for example, in a logistic regression model resulting in nonlinear and diminished effects.43 However, there are some methodological issues that need to be taken into consideration when interpreting the present study. The items “getting out of bed” and “having sex” had poor fit possibly due to few patients reporting having problems with these activities and high number of missing values, for example, due to unwillingness to report problems with sex life or because patients may have other reasons for not having sex than their physical functioning. Thus, these items were not included in any of the cause scales. These items may be omitted in future use of the scales, as the information they carry is lost, but they may be used as single items. Swimming was also problematic and was only found fit for the scale on sensory disturbances and similar potential explanation could be applied. Despite a thorough development phase, we cannot exclude that there are items which should have been included to cover a larger spectrum of reduced physical function. Arguably, items covering more delicate motor skill such as buttoning could have been included to better cover chemotherapy-induced neuropathy. Furthermore, items were selected on basis of interviews within a Danish population, and may not readily be generalizable to other populations. We tested for DIF and LD for age, axillary surgery, and chemotherapy as these are known to influence sequelae3,5,15 and follow-up time. However, the range of follow-up is limited and it is not possible to generalize to patients with a shorter or longer time since surgery. Furthermore, we cannot exclude
2014 Wolters Kluwer Health, Inc. All rights reserved.
www.clinicalpain.com |
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
799
0.95 0.86 0.94
1.0 0.98
0.94 0.97
0.99 0.86 0.82 0.99 0.94
Not Not Not
Not
Not
Not Not Not
0.94 0.90 0.92 0.94 0.91 included included included 0.92 0.94 included 0.94 0.94 included 0.93 0.94 0.94 included included included
Expected
0.68 0.39 0.46
0.49 0.46
0.76 0.58
0.36 0.36 0.05 0.36 0.54
P
Pain in Area of Surgery
Observed
0.97
0.88
0.94 0.95 0.96 0.92 0.79
0.99 0.84
0.99 0.94 0.99
Not Not Not
Not
Not
Not Not
Not
0.93 0.95 0.96 included 0.96 0.94 included included 0.95 0.93 0.95 0.93 0.91 included 0.94 included 0.93 included included included
Expected
P
0.30
0.08
0.71 0.63 0.66 0.88 0.13
0.14 0.00
0.16 0.81 0.19
Musculoskeletal Pain Observed
0.99
0.97 0.97 1.0 1.0 1.0
1.0 1.0 1.0
1.0
Not Not Not Not Not Not
Not
Not Not
Not
0.98 included 0.99 0.99 0.98 included included 0.99 0.98 0.98 0.99 0.98 included 0.99 included included included included included included
Expected
P
0.66
0.12 0.63 0.41 0.51 0.41
0.51 0.51 0.41
0.41
Sensory Disturbances Observed
1.0 0.97
0.97
0.98
0.97
0.98 0.98 0.95 0.98
Not included 0.97 0.98 0.97 0.99 Not included Not included Not included 0.97 Not included 0.98 Not included Not included Not included 0.96 Not included 0.96 0.96 Not included Not included
Expected
Lymphedema Observed
0.09 0.58
0.53
0.49
0.64
0.77 0.91 0.28 0.51
P
0.89
0.87 0.85
0.99 0.98 1.0 0.95 0.86 0.77 0.83 0.98
Expected
Other Not included 0.89 0.89 0.89 0.89 0.90 0.92 0.91 0.89 Not included 0.90 0.89 Not included 0.89 Not included Not included Not included Not included Not included Not included
Observed
0.02
0.56 0.65
0.43 0.27 0.30 0.37 0.61 0.04 0.27 0.24
P
Clin J Pain
All examined items are listed, and those included in each scale are reported with observed and expected item-rest score correlations and P-values. Items not fitting and thus excluded from the scale are indicated in the table as not included. *Numbers refer to supplemental questionnaire.
Laundry (item 40) Reach up (item 41) Carry grocery (item 42) Work in kitchen (item 43) House cleaning (item 44) Walk 1 km (item 45) Bicycle 1 km (item 46) Swimming (item 47) Open door (item 48) Drive car (item 49) Lift children (item 50) Put on socks (item 51) Wash hair (item 52) Bend down/kneel (item 53) Put on sweater (item 54) Put on bra (item 55) Put on coat (item 56) Sleep (item 57) Get out of bed (item 58) Sex (item 59)
Item*
TABLE 3. Item Fit Statistics for Items Selected for the Different Cause Scales
Andersen et al Volume 31, Number 9, September 2015
800 | www.clinicalpain.com Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clin J Pain
Volume 31, Number 9, September 2015
Assessing Physical Function After Breast Cancer Treatment
TABLE 4. Spearman Correlation Coefficients Between the Cause Scales, Q-DASH, and Single Items Concerning Sequelae
Pain in Area of Surgery
Musculoskeletal Pain
Sensory Disturbances
Lymphedema
Other Causes
Q-DASH Score
0.46 0.56
0.30 0.41
0.21 0.29
0.30 0.38
0.2 0.13
0.52 0.63
0.51 0.17
0.43 0.43
0.29 0.20
0.27 0.17
0.17 0.16
0.63 0.38
0.30 0.23 0.33 0.22 0.27 0.21 0.26 0.09 0.21
0.21 0.17 0.22 0.07 0.22 0.16 0.28 0.28 0.23
0.20 0.16 0.10 0.09 0.20 0.01 0.17 0.28 0.20
0.26 0.14 0.16 0.24 0.18 0.02 0.15 0.14 0.15
0.16 0.17 0.12 0.02 0.14 0.03 0.09 0.09 0.21
0.39 0.35 0.33 0.28 0.33 0.22 0.38 0.36 0.32
0.38 0.32 0.24
0.26 0.21 0.19
0.23 0.24 0.16
0.39 0.46 0.48
0.20 0.13 0.13
0.49 0.42 0.33
Pain Pain, surgery (y/n) Pain, surgery NRS (maximum/ weekly)* Pain in movement NRS Pain elsewhere NRSw Sensory disturbances Pins and needles Electric shock/jumping Burning Numbness Light touch: pain Cold: pain Pain and itch In hands In feet Lymph edema Sensation Visible In treatment Q-DASH score Full population Subpopulationsz
0.51 0.46y
0.60 0.648
0.35 0.36#
0.44 0.51z
0.34
Correlation coefficients are shown with negative values for dichotomized questions. *Pain, surgery NRS (maximum, weekly): highest NRS score in the 4 locations in patients experiencing pain on a weekly basis. wPain in locations outside the area of surgery. zCorrelation between cause scales and Q-DASH scores are shown for the full cohort and for the subpopulations according to the following single items. yPain in the breast, side of chest, axilla, or arm: NRSZ4. 8Pain in locations outside area of surgery: NRSZ4. #Sensory disturbances: pins and needles in the area of surgery. zLymphedema: in treatment for lymphedema. NRS indicates numerical rating scale; Q-DASH, Quick-Disability of Arm, Shoulder, and Hand.
that other factors such as analgesic or other treatments may introduce bias. Finally, the test-retest reliability was not assessed due to the study design where we focused on construction and validation of the scales. Therefore, the scales should be further validated and test-retest reliability and responsiveness to change be described in future studies, including other breast cancer populations and calibration of the scales could be provided using objective measurements of physical function and sequelae.
Clinical and Research Perspectives Applying the 5 scales in the current population indicated that almost half of the women with breast cancer had reduced physical functioning due to their breast cancer disease and treatment across the cause dimensions. The sequelae that most patients reported causing reduced physical functioning was musculoskeletal pain, where 23%, for example, reported having problems bending down due
TABLE 5. Means, SDs, and Range of the Cause Scales of Breast Cancer Patients With Reduced Physical Function Pain surgical Musculoskeletal pain Sensory disturbances Lymphedema Other causes
Copyright
r
N
Mean
SD
Range
83 116 41 51 188
18.59 21.47 19.73 18.83 13.02
16.30 15.97 14.64 13.40 11.72
4.17-65.00 4.17-66.67 5.56-57.14 3.85-54.55 3.85-58.33
to this type of pain. Musculoskeletal pain is a known complication of endocrine therapy9 as well as an age-related problem in the general population.44 Potentially, scales such as those described in the current study could be applied as initial screening tools, providing the oncologist an initial impression of the causes of impaired physical functioning in the individual patient. Patients with impaired physical functioning could be monitored for more specific treatments at pain clinics or by lymphedema therapists. Furthermore, the present cause scales can be used in research to monitor patients in various contexts, for example, in trials examining adverse effects after surgical and adjuvant treatments or in trials examining analgesics and other treatments.
CONCLUSIONS Across the cause dimensions, almost half of the women with breast cancer indicated having reduced physical functioning due to their breast cancer disease and/or its treatment. Each scale revealed different information on causes of reduced function. The present breast cancer sequelae cause scales can be used as a measurement tool to evaluate the impact of specific sequelae after breast cancer treatment on physical functioning. These scales may be applied for research and in follow-up in the breast oncology setting to systematically monitor and manage sequelae, as well as facilitate targeted clinical interventions to optimize rehabilitation.
2014 Wolters Kluwer Health, Inc. All rights reserved.
www.clinicalpain.com |
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
801
Clin J Pain
Andersen et al
ACKNOWLEDGMENT The authors thank Maj-Britt Jensen, MSc, Danish Breast Cancer cooperative Group, Copenhagen, Denmark; for assistance in providing data from the DBCG database, Per Sjøgren, MD, PhD, Section for palliative medicine, Rigshospitalet, Copenhagen, Denmark; for thoughtful comments on the manuscript, and Sebastian Jud, MD, Universta¨tsklinikum Erlangen, Germany; for the drawings used in the questionnaire. REFERENCES 1. Youlden DR, Cramb SM, Dunn NA, et al. The descriptive epidemiology of female breast cancer: an international comparison of screening, incidence, survival and mortality. Cancer Epidemiol. 2012;36:237–248. 2. Andersen KG, Kehlet H. Persistent pain after breast cancer treatment: a critical review of risk factors and strategies for prevention. J Pain. 2011;12:725–746. 3. Gartner R, Jensen MB, Nielsen J, et al. Prevalence of and factors associated with persistent pain following breast cancer surgery. JAMA. 2009;302:1985–1992. 4. Mejdahl MK, Andersen KG, Gartner R, et al. Persistent pain and sensory disturbances after treatment for breast cancer: six year nationwide follow-up study. BMJ. 2013;346:f1865. 5. Andersen KG, Jensen MB, Kehlet H, et al. Persistent pain, sensory disturbances and functional impairment after adjuvant chemotherapy for breast cancer: cyclophosphamide, epirubicin and fluorouracil compared with docetaxel + epirubicin and cyclophosphamide. Acta Oncol. 2012;51:1036–1044. 6. DiSipio T, Rye S, Newman B, et al. Incidence of unilateral arm lymphoedema after breast cancer: a systematic review and meta-analysis. Lancet Oncol. 2013;14:500–515. 7. McWayne J, Heiney SP. Psychologic and social sequelae of secondary lymphedema: a review. Cancer. 2005;104:457–466. 8. Paskett ED, Dean JA, Oliveri JM, et al. Cancer-related lymphedema risk factors, diagnosis, treatment, and impact: a review. J Clin Oncol. 2012;30:3726–3733. 9. Din OS, Dodwell D, Wakefield RJ, et al. Aromatase inhibitorinduced arthralgia in early breast cancer: what do we know and how can we find out more? Breast Cancer Res Treat. 2010;120: 525–538. 10. Groenvold M, Fayers PM, Petersen MA, et al. Breast cancer patients on adjuvant chemotherapy report a wide range of problems not identified by health-care staff. Breast Cancer Res Treat. 2007;103:185–195. 11. Hidding JT, Beurskens CH, van der Wees PJ, et al. Treatment related impairments in arm and shoulder in patients with breast cancer: a systematic review. PLoS One. 2014;9:e96748. 12. World Health Organization. International Classification of Functioning, Disability and Health. Geneva, Switzerland: World Health Organization; 2001. 13. Montazeri A. Health-related quality of life in breast cancer patients: a bibliographic review of the literature from 1974 to 2007. J Exp Clin Cancer Res. 2008;27:32. 14. Rietman JS, Dijkstra PU, Hoekstra HJ, et al. Late morbidity after treatment of breast cancer in relation to daily activities and quality of life: a systematic review. Eur J Surg Oncol. 2003;29:229–238. 15. Gartner R, Jensen MB, Kronborg L, et al. Self-reported armlymphedema and functional impairment after breast cancer treatment—a nationwide study of prevalence and associated factors. Breast. 2010;19:506–515. 16. Bennett MI, Attal N, Backonja MM, et al. Using screening tools to identify neuropathic pain. Pain. 2007;127:199–203. 17. Kennedy CA, Beaton DE, Solway S, et al. Disabilities of the Arm, Shoulder and Hand (DASH). The DASH and QuickDASH Outcome Measure User’s Manual. Toronto, ON, Canada: Institute for Work & Health; 2011. 18. Leblanc M, Mao JJ, Stineman M, et al. Validation of QuickDASH outcome measure in breast cancer survivors for upper extremity disability. Arch Phys Med Rehabil. 2014;95: 493–498.
Volume 31, Number 9, September 2015
19. Harrington S, Michener LA, Kendig T, et al. Patient-reported upper extremity outcome measures used in breast cancer survivors: a systematic review. Arch Phys Med Rehabil. 2014;95:153–162. 20. Moller S, Jensen MB, Ejlertsen B, et al. The clinical database and the treatment guidelines of the Danish Breast Cancer Cooperative Group (DBCG); its 30-years experience and future promise. Acta Oncol. 2008;47:506–524. 21. Blichert-Toft M, Christiansen P, Mouridsen HT. Danish Breast Cancer Cooperative Group—DBCG: history, organization, and status of scientific achievements at 30-year anniversary. Acta Oncol. 2008;47:497–505. 22. Linden WJ, Hambleton RK. Handbook of Modern Item Response Theory. New York: Springer Verlag; 1997. 23. Christensen KB, Kreiner S, Mesbah M. Rasch Models in Health. London, UK: Wiley-ISTE; 2013. 24. Fischer G, Molenaar IW. Rasch Models. New York: Springer Verlag; 1995. 25. Rasch G. Probabilistic Models for Some Intelligence and Attainment Tests. Copenhagen, Denmark: Nielsen & Lydiche; 1960. 26. Andersen EB. A goodness of fit test for the Rasch model. Psychometrika. 1973;38:123–140. 27. Kreiner S. A note on item-rest score association in Rasch models. Appl Psychol Meas. 2011;35:557–561. 28. Kreiner S, Christensen KB. Exact evaluation of bias in Rasch model residuals. Volume 12. In: Baswell AR, ed. Advances in Mathematics Research. New York: Nova Publishers; 2012:19–40. 29. Kelderman H. Loglinear Rasch model tests. Psychometrika. 1984;49:223–245. 30. Cronbach L. Coefficient alpha and the internal structure of tests. Psychometrika. 1951;16:297–334. 31. Streiner DL, Norman GR. Health Measurement Scales. Oxford, UK: Oxford University Press; 2008. 32. Bland JM, Altman DG. Cronbach’s alpha. BMJ. 1997;314:572. 33. Ware JE Jr., Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med Care. 1992;30:473–483. 34. Aaronson NK, Ahmedzai S, Bergman B, et al. The European Organization for Research and Treatment of Cancer QLQC30: a quality-of-life instrument for use in international clinical trials in oncology. J Natl Cancer Inst. 1993;85: 365–376. 35. Sprangers MA, Groenvold M, Arraras JI, et al. The European Organization for Research and Treatment of Cancer breast cancer-specific quality-of-life questionnaire module: first results from a three-country field study. J Clin Oncol. 1996;14:2756–2768. 36. Brady MJ, Cella DF, Mo F, et al. Reliability and validity of the functional assessment of cancer therapy-breast quality-oflife instrument. J Clin Oncol. 1997;15:974–986. 37. Daut RL, Cleeland CS, Flanery RC. Development of the Wisconsin Brief Pain Questionnaire to assess pain in cancer and other diseases. Pain. 1983;17:197–210. 38. McCarthy M Jr., Jonasson O, Chang CH, et al. Assessment of patient functional status after surgery. J Am Coll Surg. 2005;201:171–178. 39. Ringsted TK, Wildgaard K, Kreiner S, et al. Pain-related impairment of daily activities after thoracic surgery: a questionnaire validation. Clin J Pain. 2013;9:791–799. 40. Franchignoni F, Ferriero G, Giordano A, et al. Psychometric properties of QuickDASH—a classical test theory and Rasch analysis study. Man Ther. 2011;16:177–182. 41. Turk DC, Dworkin RH, Burke LB, et al. Developing patientreported outcome measures for pain clinical trials: IMMPACT recommendations. Pain. 2006;125:208–215. 42. Rosenbaum PR. Criterion-related construct-validity. Psychometrika. 1989;54:625–633. 43. Hansen PE, Siersma V, Ross L, et al. Psychometric properties of brief indexes designed to measure social-cognitive predictors of smoking initiation. Drug Alcohol Depend. 2007;88:64–74. 44. Kurita GP, Sjogren P, Juel K, et al. The burden of chronic pain: a cross-sectional survey focussing on diseases, immigration, and opioid use. Pain. 2012;153:2332–2338.
802 | www.clinicalpain.com Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
The Effect of Pain on Physical Functioning After Breast Cancer Treatment Development and Validation of an Assessment Tool Kenneth G. Andersen, MD,*w Karl B. Christensen, MSc, PhD,z Henrik Kehlet, MD, PhD,* and Pernille E. Bidstup, MSc, PhDy Objectives: Persistent postsurgical pain, musculoskeletal pain, sensory disturbances, and lymphedema are major clinical problems after treatment for breast cancer. However, there is little evidence on how these sequelae affects physical function. The aim this study was to develop and validate a procedure-specific tool for assessing the impact of pain and other sequelae on physical function after breast cancer treatment. Methods: A literature review, patient and expert interviews were used to identify dimensions of physical function and sequelae. A questionnaire was developed and tested using cognitive interviews, and field tested among 389 patients treated for primary breast cancer without recurrence (response rate 81%). Median follow-up was 14 months. Using item response theory we identified 5 cause scales of reduced physical functioning: pain after surgery, musculoskeletal pain, sensory disturbances, lymphedema, and other causes. Convergent validity was assessed using the QuickDsability of Arm, Shoulder, and Hand Scale (Q-DASH). Results: About half of the patients reported decreased physical function. All 5 scales displayed good fit, unidimensionality, monotonicity, local independence, and lack of differential item functioning. Cronbach coefficient a ranged from 0.88 (other causes) to 0.96 (sensory disturbances) for the 5 scales. For the Q-DASH a was 0.91. Each scale revealed different information on causes of reduced function. Discussion: The present scales displayed good psychometric qualities, and may be used to evaluate the impact of specific sequelae after breast cancer treatment on physical functioning, as well as to monitor and target interventions to optimize pain treatment and rehabilitation. Key Words: breast cancer, physical function, sequelae, persistent pain, neuropathy, lymphedema, Rasch analysis Received for publication April 26, 2014; revised October 1, 2014; accepted August 28, 2014. From the *Section for Surgical Pathophysiology; wDepartment of Breast Surgery, Rigshospitalet, University of Copenhagen; zDepartment of Biostatistics, Institute of Public Health, University of Copenhagen; and yDepartment of Survivorship, Danish Cancer Research Center, Danish Cancer Society, Copenhagen, Denmark. Supported by a grant from the Danish Cancer Society, Copenhagen, Denmark, and the study is part of the Europain Collaboration, which has received support from the Innovative Medicines Initiative Joint Undertaking, under grant agreement no 115007, resources of which are composed of financial contribution from the European Union’s Seventh Framework Programme (FP7/20072013) and EFPIA companies’ kind contribution. http://www. imieuropain.org/. The authors declare no conflict of interest. Reprints: Kenneth G. Andersen, MD, Section for Surgical Pathophysiology, 4074, Rigshospitalet, University of Copenhagen, Blegdamsvej 9, Copenhagen DK-2100, Denmark (e-mail: [email protected]). Supplemental Digital Content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal’s Website, www.clinicalpain.com. Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. DOI: 10.1097/AJP.0000000000000156
(Clin J Pain 2015;31:794–802)
S
urvival after breast cancer treatment has increased over the past years, and 5-year relative survival is now 80% to 90% in the Nordic countries and North America,1 emphasizing the need for more attention on late sequelae and their rehabilitation. Such sequelae are common, including persistent pain, sensory disturbances, lymphedema, reduced arm function, and side effects of medications, all of which may impair physical function, psychological well-being, and quality of life. Persistent pain after breast cancer treatment (PPBCT) affects between 25% and 60% of the patients depending on treatment,2–4 but the effect of PPBCT on functional impairment per se is not well described. Tightly connected to PPBCT are sensory disturbances, which is frequent in the area of surgery2 as well as in hands and feet5 due to chemotherapy-induced neuropathy. Sensory disturbances such as allodynia and hyperalgesia may cause discomfort when wearing clothes, handling small objects, and in affected feet cause problem during walk. The prevalence of lymphedema among patients with breast cancer was 15% to 30% in a large meta-analysis,6 and may substantially affect both physical and psychological aspects of quality of life.7,8 Reduced arm mobility due to pain or fibrotic tissue in the axilla and musculoskeletal pain problems as a side effect of antihormonal treatment are also common and may contribute to reduced physical functioning.9 Unfortunately, many of these problems are not always acknowledged by health care providers.10 Within the patient-reported measures, the construct of functioning—and especially physical functioning— play a key role as it may capture the impact and severity of the sequelae, which otherwise may be perceived differently by the individual patients. To make progress in preventive and interventional strategies for alleviating sequelae, there is a need for a validated specific and sensitive tool to describe the specific sequelae causing reduced physical function.11 The aim of this study was to construct and validate such a tool.
MATERIALS AND METHODS Theoretical Model Physical function may be understood in a biopsychosocial context, with interactions between disease or disorder, impairment of body functions and structures, reduction of ability to perform daily activities, participation in work and social activities, environmental factors, and patient-related factors.12 At the core of this context, ability to perform daily activities is the most concrete outcome and simple to measure. Either a patient can perform an activity, perform it despite difficulties, or the activity cannot be performed. Inability to perform daily activities may be caused by several different factors including breast cancer–specific
794 | www.clinicalpain.com Clin J Pain Volume 31, Number 9, September 2015 Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clin J Pain
Volume 31, Number 9, September 2015
sequelae or other reasons such as comorbidity. Thus, it is possible not only to measure whether or not there is a functional impairment, but also what the underlying cause is.
Literature Review Using searches in PubMed as well as reviews2,13,14 on the area, the literature was examined for available scales and questionnaires used in the assessment of PPBCT, QoL, and functional impairment of upper extremities; items and domains were entered in an item database, and labeled according to the construct each item was designed to measure. This was conducted to get an overview of the dimensions of sequelae and possible constructs.
Semistructured Interviews On the basis of the identified dimensions in the literature search, an interview guide was developed. Semistructured interviews were conducted to assure that all relevant dimensions were covered. Number of patients needed to interview was decided by sampling to redundancy. Ten breast cancer patients covering all treatment modalities were recruited from the breast surgery and breast oncology departments and 10 health care providers from the breast surgery, breast oncology, and rheumatology departments were interviewed. All interviews were recorded and transcribed pragmatically extracting dimensions and items, which were added to the item database.
Construction of a Preliminary Questionnaire The questionnaire for assessing sequelae after breast cancer treatment was constructed by single items covering persistent pain, sensory disturbances, lymphedema, and items covering physical function intended for scale development. Items concerning persistent pain and lymphedema were extracted from a previous study,3,15 in which the content validity had been examined through personal patient interviews. Items on sensory disturbances were inspired and developed from a review specifically examining scales relating to neuropathic pain.16 All items entered in the database were assessed, and then synthesized into the questionnaire to cover the identified dimensions.
Construction and Testing of Content Validity The preliminary questionnaire was then sequentially tested among breast cancer patients and health care providers by cognitive interviews of understanding and relevance of each item. After each dimension, the participants were asked if the questions covered all areas. Additional items were supplemented where needed. After interviewing 8 breast cancer patients and 8 health care providers, no additional information was provided. Two additional breast cancer patients were interviewed and finally the eighth revision was considered ready to be tested in a larger population. Twenty items regarding ability to perform daily activities, and reasons for reduced ability were included for Rasch analysis. Furthermore, the questionnaire consists of items related to sequelae, but not included in the Rasch analysis. (Supplementary Questionnaire, Supplemental Digital Content 1, http://links.lww.com/CJP/A131).
Convergent Validity As a measure for upper arm function, the QuickDisability of Arm, Shoulder, and Hand (Q-DASH) questionnaire was choosen17 to assess convergent validity. The Q-DASH is a validated shortened version of the DASH Outcome Measure, with 11 items to measure impaired Copyright
r
Assessing Physical Function After Breast Cancer Treatment
physical function and symptoms in the upper body, and is validated in the breast cancer population.18 The DASH has been widely used in the breast cancer population, and is the recommended scale to measure upper body after breast cancer treatment.19 Furthermore, convergent validity was examined with correlations with items of self-assessed pain, sensory disturbances, and lymphedema developed in parallel to the cause scales.
Testing in a Breast Cancer Population The study was performed in a cohort of consecutive breast cancer survivors treated at Rigshospitalet between August 13, 2009 and July 30, 2010, using a cross-sectional design. Data were collected in April 2011 at Rigshospitalet, Copenhagen, Denmark. Patients eligible for inclusion were primary breast cancer patients undergoing treatment according to the standard national Danish treatment protocol. Exclusion criteria were: death, disease in the nervous system, bilateral disease, recurrence, or other malignancy. Disease characteristics were provided by the Danish Breast Cancer Cooperative Group (DBCG), which prospectively collects clinical and pathologic data,20,21 and includes details on type of surgery of breast and axilla, radiation therapy, chemotherapy, and endocrine treatment. The study was performed in accordance with Danish law (The Committees on Health Research Ethics for the Capital Region of Denmark, H-4-2013-FSP-045 and the Danish Data Protection Agency 2007-58-0015/30-0545).
Statistics We addressed a specific type of construct validity, called criterion-related construct validity, which concerns the relationship between the index, the individual items included in the index, and exogenous variables, which characterize the population such as age and prognostic or treatment-related factors. For each domain the fit of the items to an item response theory (IRT) model was examined.22 The simplest IRT model, the Rasch model23–25 was chosen. All items were assessed using the Rasch model, with a systematic and iterative process discarding items not fitting the model. In the Rasch model the sum of all items is a sufficient statistic for the underlying latent variable, and this means that when a scale fits the Rasch model, the simple summation of the items within this scale can be used as a measure of the underlying latent variable. The analysis assessed overall fit of the model by Andersen conditional likelihood ratio test,26 testing homogeneity of 2 score groups. The fit of each individual item to the Rasch model was assessed by comparison of observed and expected correlations between separate item scores and sum of the remaining items.27 Briefly, this item fit statistic evaluates whether the observed rank correlation between the item and the sum of the remaining items in the scale differs from the expected value correlation. The fit of each individual item to the Rasch model was also assessed by conditional infit and outfit test statistics.28 These compare observed and expected item responses weighted together as a sum of squared residuals. We tested local dependence and the absence of differential item functioning (DIF) using loglinear Rasch models.29 DIF was examined with respect to the variables chemotherapy (yes/no), follow-up time (under/over 12 mo), age (over/under 60 y), and axillary surgery (sentinel lymph node biopsy or axillary clearance). Additional dichotomizations of the variable “follow-up time” (10, 14, and 16 mo) were also evaluated. Analyses were carried out using the DIGRAM software (Department of Biostatistics, University of Copenhagen, Denmark). The significance of
2014 Wolters Kluwer Health, Inc. All rights reserved.
www.clinicalpain.com |
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
795
Clin J Pain
Andersen et al
Volume 31, Number 9, September 2015
Daily activities We are interested to know how treatment for breast cancer affects your daily activities. Think about how last week had been for you in relation to the activites below: If you have difficulty, experiencing discomfort, or simply just cannot do some of these activities, please describe your difficulties in the box provided below each question.
Scoring of the cause scales:
Scoring stage 1:
Do laundry Not relevant, I don’t do this = 0 Not relevant, I don’t do this
I can do this without any problem = 0 I can do this without any problem I can do this, but with difficulties because (please explain)
I can do this, but with difficulties = 1
I can’t do this due to
I can’t do this due to = 2
(please explain)
Scoring stage 2: Yes = multiply score from stage 1 with 1 No = multiply score from stage 1 with 0
Is the cause of your difficulty due to: Pain in the area you had surgery? Muscle or joint pain? Sensory disturbances, such as pins and needles, sleeping or stabbing sensations? Swelling (lymphedema) Other cause
Yes Yes
No No
Yes
No
Yes Yes
No No
Example: Pain after surgery scale: 2 x 1 = 2 Musculoskeletal pain scale: 2 x 0 = 0 Sensory disturbance scale: 2 x 1 = 2
Description:___________________________________
Lymphedema scale: 0 x 2 = 0 Other causes scale: 0 x 2 = 0
Scores are summated in the 5 cause scales and transformed to a 0-100 scale.
FIGURE 1. Scoring of the cause scales with an example.
item fit statistics was assessed by the Benjamini-Hochberg procedure controlling the false discovery rate at 0.05. Cronbach a-coefficients30,31 were calculated as measures of reliability, and values between 0.7 and 0.8 were regarded as satisfactory.32 Criterion validity related to Q-DASH scores and measures of pain, lymphedema, and sensory disturbances were evaluated using Spearman rank correlations.
Scoring and Construction of Scales Each of the 20 items included in the initial item set had 2 scoring stages: an initial rating of difficulty or discomfort experienced in relation to the activity (response options “Not relevant, I don’t do this,” “I can do this without any problem,” “I can do this, but with difficulties,” and “I can’t do this”) and a secondary rating where respondents attributed the difficulty or discomfort to either pain after surgery, musculoskeletal pain, sensory disturbances, lymphedema, or other causes, respectively. On the basis of these responses, a total of 100 items (20 in each dimension) were coded as outlined in Figure 1. The scores were summated for each of the 5 scales, and transformed into a 0 to 100 scale.
RESULTS Participants: The Breast Cancer Test Population Between August 13, 2009 and July 30, 2010, 500 consecutive patients were identified in DBCG database as eligible. One patient had died, 3 were not reachable at the address, and 2 had migrated. Of 494 questionnaires, 402 questionnaires were returned (response rate: 81%). A total of 13 patients were excluded: 4 patients returned the questionnaire and stated they did not wish to participate, 3 were excluded due to mainly unanswered questionnaires, 2 with
metastatic disease, 1 with hemiparesis after apoplexia, 1 with disseminated sclerosis, 1 with reconstructive surgery, and 1 with a recent shoulder fracture. Final analysis included 389 patients. Patient characteristics are presented in Table 1. There were no differences in age, follow-up, or treatment among responders and nonresponders. Prevalence of pain, sensory disturbances, lymphedema, and function related to treatment groups are presented in Supplementary Tables 1 to 3 (Supplemental Digital Content 2–4, http://links.lww.com/ CJP/A132, http://links.lww.com/CJP/A133, http://links. lww.com/CJP/A134). The frequencies of patients affected according to the cause scales and the median scores are shown in Table 2, and percentages of the population affected by each activity are shown in Figure 2.
The Cause Scales: Rasch Analysis The cause scales “pain after surgery,” “musculoskeletal pain,” “sensory disturbances,” “lymphedema,” and “other causes” were constructed from the 20 general items and their accompanying cause item, using item reduction. The items included in each scale are shown in Table 3, together with the item fit statistics, which indicate an overall good fit across all the 5 proposed scales. Two items concerning difficulties with “sexual activity” and “getting out of bed” were not included in any scale. DIF with regard to surgery in the axilla, chemotherapy, age, or follow-up time, was not found for the any of 5 scales (Supplementary Tables 4–8, Supplemental Digital Content 5–9, http://links.lww.com/CJP/A135, http://links. lww.com/CJP/A136, http://links.lww.com/CJP/A137, http:// links.lww.com/CJP/A138, http://links.lww.com/CJP/A139). Regarding local dependence no clinically relevant evidence was disclosed for the 5 proposed scales. The scales fitted the Rasch model, and thus supported the use of simple
796 | www.clinicalpain.com Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clin J Pain
Volume 31, Number 9, September 2015
Assessing Physical Function After Breast Cancer Treatment
TABLE 1. Population Characteristics of Breast Cancer Patients Enrolled in Interviews and Questionnaire Study
Breast Cancer Test Population N (%) Age (median, IQR) (y) Follow-up (median, IQR) (mo) Surgical treatment (N [%]) BCS + SLNB BCS + ALND Ma + SLNB Ma + ALND Adjuvant treatment (N [%]) Chemotherapy Yes No Radiation therapy No radiation therapy BRT AT/LRRT Endocrine therapy No endocrine therapy Tamoxifen Aromatase inhibitors
Semistructured Interview
Questionnaire Interview
Responders
Nonresponders
Total
10 (100) 57 (54-59) 18 (12-44)
10 (100) 65 (60-73) 27 (23-56)
389 (81) 60 (53-66) 14 (11-17)
92 (19) 59 (50-67) 15 (11.5-17)
481 (100) 60 (52-66) 14 (11-17)
179 99 34 77
38 19 10 25
217 118 59 87
2 2 2 4
(20) (20) (20) (40)
2 2 2 3
(20) (20) (20) (30)
(46) (25) (9) (20)
(41) (21) (11) (27)
(45) (25) (12) (18)
9 (90) 1 (10)
5 (50) 4 (40)
186 (48) 203 (52)
52 (47) 40 (43)
238 (49) 243 (51)
3 (30) 2 (20) 5 (50)
4 (40) 3 (30) 2 (20)
44 (11) 208 (54) 137 (35)
11 (12) 41 (45) 40 (43)
55 (11) 249 (52) 177 (37)
1 (10) 2 (20) 6 (60)
1 (10) 4 (40) 5 (50)
104 (27) 59 (15) 226 (58)
26 (28) 15 (16) 51 (56)
130 (27) 74 (15) 277 (58)
Population characteristics for the patients enrolled in the semistructured interviews, the questionnaire interviews, and the breast cancer population. Follow-up: time from surgery to participation in interview or answering the questionnaire. Missing data: semistructured interview—endocrine therapy: 1. Questionnaire interview—treatment, surgery: 1, chemotherapy: 1. radiation therapy: 1. ALND indicates axillary lymph node dissection; AT/LRRT, anterior thoracic wall radiation therapy and locoregional radiation therapy corresponding to periclavicular, axillary level 3 and for right-sided cancer the internal mammary lymph nodes; BCS, breast conservative surgery; BRT, breast radiation therapy corresponding to residual breast tissue; IQR, interquartile range; SLNB, sentinel lymph node biopsy.
summation of the items within each scale.23 Andersen conditional likelihood ratio test demonstrated homogeneity of the scales: pain after surgery, P = 0.67; musculoskeletal pain, P = 0.06; sensory disturbances, P = 0.44; lymphedema, P = 0.09; and other causes, P = 0.62. The correlations between the cause scales and items in the questionnaire are shown in Table 4. Means, SD and range for the 5 scales are shown in Table 5.
The Cause Scales: Reliability Analysis Cronbach coefficient a indicated high reliability for all scales: “pain after surgery,” a = 0.89; “musculoskeletal pain,” a = 0.92; “sensory disturbances,” a = 0.96; “lymphedema,” a = 0.91; and “other causes,” a = 0.88.
Q-DASH Validity and Correlation to Cause Scales The number of missing responses was low for most items; Q-DASH question 6 (recreational activities) and 10 (tingling, pins, and needles in arm, shoulder, and/or hand) having the highest number of missing responses. Cronbach a was 0.91 and the interitem correlations ranged from 0.27 to 0.77. The item fit was poor (Supplementary Table 9, Supplemental Digital Content 10, http://links.lww.com/ CJP/A140). There was evidence of local dependence for 10 of the 55 item pairs (results not shown). Regarding DIF, this was found for question 1 (open a tight or new jar) for follow-up time and for questions 8 (limitations in work or other daily activities) and 9 (arm, shoulder, or hand pain) for axillary surgery (P < 0.01). Q-DASH correlation with the cause scales ranged from 0.35 to 0.6, lowest for the sensory disturbances scale and highest for the musculoskeletal pain scale (Table 4). Copyright
r
DISCUSSION A thorough development process including a literature review and patient and expert interviews in 2 stages resulted in an initial item bank of 20 items on physical functioning related to daily activities in patients after breast cancer treatment. We grouped items based on the patient-reported causes of reduced physical functioning (including pain in the area of surgery, musculoskeletal pain, sensory disturbances, lymphedema, and other causes) resulting in 5 cause scales. Initial validation of the cause scales indicated that the common criteria for validity, monotonicity, unidimensionality, homogeneity, local independence, and lack of DIF were all achieved. Internal consistency expressed as Cronbach a was high ranging from 0.88 to 0.96. Furthermore, the cause scale scores correlated satisfactorily with other items in the questionnaire measuring corresponding dimensions. Thus, overall psychometric properties for all scales indicate proof of concept as a measuring tool of the causes of reduced physical function in the breast cancer population. The approach of scoring reduced physical function through 2 steps is novel, and the first attempt to provide information as to what causes reduced physical function— whether it is breast cancer–specific sequelae or other factors such as comorbidity. Patient-reported outcome measurements assessing physical functioning often applied in the breast cancer population include general quality of life scales such as SF-3633 and quality of life scales with diseasespecific and symptom-specific modules such as the EORTC c3034 with BR2335 and FACT-G with FACT-B.36 SF-36 measures ability to perform daily activities, but we have not been able to find specific data of psychometric properties in a breast cancer population. EORTC c30/BR23 and FACTB measure several dimensions of Health-related Quality of
2014 Wolters Kluwer Health, Inc. All rights reserved.
www.clinicalpain.com |
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
797
8 (4) 171 (96) 12.24 (6.2522.65)
12 (7) 167 (93) 11.81 (5.9023.61)
50 (28) 129 (72) 15 (9.0925)
56 (28) 147 (72) 13.07 (7.29-25)
94 (51) 92 (49) 8.33 (4.5516.67)
38 (20) 148 (80) 16.03 (9.0926.92)
15 (7) 193 (93) 10 (4.1723.08)
17 (8) 191 (92) 11.11 (6.25-25)
60 (29) 148 (71) 15 (9.0926.39)
21 (48) 96 (46) 23 (52) 112 (54) 5.56 (4.55- 4.69 (4.5510) 16.67)
3 (7) 41 (93) 7.69 (4.1746.15)
AT/LRRT
71 (52) 66 (48) 10 (5-25)
33 (24) 104 (76) 16.67 (11.54-25)
18 (13) 119 (87) 19.05 (11.1135.71)
48 (35) 89 (65) 25 (8.7136.93)
38 (18) 40 (29) 170 (82) 97 (71) 9.09 (5.56- 13.07 (8.3315) 34.09)
BRT
Radiotherapy
44 (42) 60 (58) 11.25 (519.09)
9 (9) 95 (91) 16.67 (1023.08)
9 (9) 95 (9) 12.50 (11.11-30)
26 (25) 78 (75) 20.42 (9.0933.33)
22 (21) 82 (79) 13.07 (6.2531.82)
37 (63) 22 (37) 7.69 (4.5511.54)
13 (22) 46 (78) 12.50 (11.5420.83)
6 (10) 53 (90) 13.89 (1016.25)
18 (31) 41 (69) 10 (4.5527.27)
20 (34) 39 (66) 10 (6.2517.92)
TAM
AI
107 (47) 119 (53) 8.33 (4.5518.18)
29 (13) 197 (87) 16.67 (7.6926.92)
26 (12) 200 (88) 16.67 (5.5628.57)
72 (32) 154 (68) 16.67 (8.3334.17)
41 (18) 185 (82) 10 (4.5527.27)
Endocrine Therapy None
188 (48) 201 (52) 8.33 (4.5518.18)
51 (13) 338 (87) 15.38 (8.33-25)
41 (11) 348 (89) 16.67 (6.2528.57)
116 (30) 272 (70) 16.67 (8.7133.33)
83 (21) 306 (79) 10 (5.5627.27)
Total
Number and percentages of patients scoring 0 (indicated as “no”) and above 0 (indicated as “yes”) and median scores of patients scoring above 0. AI indicates aromatase inhibitors; ALND, axillary lymph node dissection; AT/LRRT, anterior thoracic wall radiation therapy and locoregional radiation therapy corresponding to periclavicular, axillary level 3 and for right-sided cancer the internal mammary lymph nodes; BCS, breast conservative surgery; BRT, breast radiation therapy corresponding to residual breast tissue; IQR, interquartile range; SLNB, sentinel lymph node biopsy; TAM, tamoxifen.
94 (46) 109 (54) 9.09 (518.18)
13 (6) 190 (94) 11.54 (8.3319.23
6 (14) 38 (86) 15.71 (6.2522.22)
8 (18) 36 (82) 9.55 (4.7731.06)
5 (11) 39 (89) 27.27 (4.1727.27)
None
46 (46) 53 (54) 8.33 (5.015.38)
22 (22) 77 (78) 16.67 11.5422.73)
21 (21) 165 (89) 22.22 (11.1135.71)
60 (32) 126 (68) 20.83 (9.0941.88)
32 (16) 51 (27) 171 (84) 135 (73) 9.09 (4.55- 13.64 (8.33-40) 15.91)
Chemotherapy
Chemotherapy None
18 (18) 20 (10) 81 (82) 183 (90) 21.83 (7.14- 9.13 (5.5637.50) 21.83)
38 (38) 61 (62) 20.42 (8.3338.89)
29 (29) 71 (71) 18.18 (1041.67)
ALND
BCS
30 (17) 149 (83) 8.71 (4.55 13.64)
SLNB
Clin J Pain
43 (56) 83 (46) 34 (44) 96 (54) 10 (4.55-25) 8.33 (4.1718.18)
16 (47) 18 (53) 5.28 (4.3612.69)
Other Yes (N [%]) No (N [%]) Scores of those affected (median [IQR])
7 (9) 70 (91) 11.11 (5.5616.67)
22 (29) 55 (71) 29.29 (13.6440.91)
20 (26) 57 (74) 9.55 (6.2529.55)
2 (6) 19 (25) 32 (94) 58 (75) 26.92 (7.69- 15.38 (8.3346.15) 27.27)
4 (12) 30 (88) 21.83 (15.7136.11)
6 (18) 28 (82) 9.55 (5.016.67)
4 (12) 30 (88) 27.27 (15.7238.64)
ALND
Mastectomy
SLNB
Lymphedema Yes (N [%]) No (N [%]) Scores of those affected (median [IQR])
Sensory disturbances Yes (N [%]) No (N [%]) Scores of those affected (median [IQR])
Musculoskeletal pain Yes (N [%]) No (N [%]) Scores of those affected (median [IQR])
Pain in area of surgery Yes (N [%]) No (N [%]) Scores of those affected (median [IQR])
Cause Scales
TABLE 2. Physical Function in a Breast Cancer Population by Surgical and Adjuvant Treatment Groups
Andersen et al Volume 31, Number 9, September 2015
798 | www.clinicalpain.com Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clin J Pain
Volume 31, Number 9, September 2015
Assessing Physical Function After Breast Cancer Treatment
FIGURE 2. Percentages of patients with reduced capacity to perform daily activities by cause. Numbers in rows are percentages of total population.
Life (HRQoL), where physical function is expressed as physical well-being. The present cause scales measure a different and more concrete dimension of physical functioning, using the ability to perform well-defined daily activities. Generic pain impairment questionnaires, such as the Brief Pain Inventory,37 may not discern unspecific pain reports from specific PPBCT. The use of procedure-specific questionnaires to measure pain-related physical impairment is supported after other surgical interventions such as hernia repair38 and thoracic surgery.39 Scales measuring upper body morbidity such as the Q-DASH have been validated18; however, there is a need for further validation.19 A previous study among patients with upper limb disorder indicated problems with unidimensionality in the Q-DASH.40 Furthermore, our psychometric analyses of the Q-DASH among women with breast cancer indicated multidimensionality, poor item fit, as well as local dependence and DIF. Summarizing, the present cause scales are distinct from existing scales and fill a gap in measuring physical function and sequelae in the breast cancer population. Fit of data to an IRT model guarantees that items are appropriate for assessment of a single underlying variable. The benefits of using IRT methods compared with classic psychometric theory are summarized by the IMMPACT initiative.41 IRT models exhibit a number of desirable properties, namely: (1) unidimensionality, (2) monotonicity, (3) local independence, and (4) lack of DIF.42 These characteristics are often neglected, but may identify serious measurement problems in the scales. Unidimensionality is a key component in summarizing scales, and multidimensionality may introduce Copyright
r
bias. The potential implications of departures from the Rasch model such as local dependence and DIF include poor power and bias, for example, in a logistic regression model resulting in nonlinear and diminished effects.43 However, there are some methodological issues that need to be taken into consideration when interpreting the present study. The items “getting out of bed” and “having sex” had poor fit possibly due to few patients reporting having problems with these activities and high number of missing values, for example, due to unwillingness to report problems with sex life or because patients may have other reasons for not having sex than their physical functioning. Thus, these items were not included in any of the cause scales. These items may be omitted in future use of the scales, as the information they carry is lost, but they may be used as single items. Swimming was also problematic and was only found fit for the scale on sensory disturbances and similar potential explanation could be applied. Despite a thorough development phase, we cannot exclude that there are items which should have been included to cover a larger spectrum of reduced physical function. Arguably, items covering more delicate motor skill such as buttoning could have been included to better cover chemotherapy-induced neuropathy. Furthermore, items were selected on basis of interviews within a Danish population, and may not readily be generalizable to other populations. We tested for DIF and LD for age, axillary surgery, and chemotherapy as these are known to influence sequelae3,5,15 and follow-up time. However, the range of follow-up is limited and it is not possible to generalize to patients with a shorter or longer time since surgery. Furthermore, we cannot exclude
2014 Wolters Kluwer Health, Inc. All rights reserved.
www.clinicalpain.com |
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
799
0.95 0.86 0.94
1.0 0.98
0.94 0.97
0.99 0.86 0.82 0.99 0.94
Not Not Not
Not
Not
Not Not Not
0.94 0.90 0.92 0.94 0.91 included included included 0.92 0.94 included 0.94 0.94 included 0.93 0.94 0.94 included included included
Expected
0.68 0.39 0.46
0.49 0.46
0.76 0.58
0.36 0.36 0.05 0.36 0.54
P
Pain in Area of Surgery
Observed
0.97
0.88
0.94 0.95 0.96 0.92 0.79
0.99 0.84
0.99 0.94 0.99
Not Not Not
Not
Not
Not Not
Not
0.93 0.95 0.96 included 0.96 0.94 included included 0.95 0.93 0.95 0.93 0.91 included 0.94 included 0.93 included included included
Expected
P
0.30
0.08
0.71 0.63 0.66 0.88 0.13
0.14 0.00
0.16 0.81 0.19
Musculoskeletal Pain Observed
0.99
0.97 0.97 1.0 1.0 1.0
1.0 1.0 1.0
1.0
Not Not Not Not Not Not
Not
Not Not
Not
0.98 included 0.99 0.99 0.98 included included 0.99 0.98 0.98 0.99 0.98 included 0.99 included included included included included included
Expected
P
0.66
0.12 0.63 0.41 0.51 0.41
0.51 0.51 0.41
0.41
Sensory Disturbances Observed
1.0 0.97
0.97
0.98
0.97
0.98 0.98 0.95 0.98
Not included 0.97 0.98 0.97 0.99 Not included Not included Not included 0.97 Not included 0.98 Not included Not included Not included 0.96 Not included 0.96 0.96 Not included Not included
Expected
Lymphedema Observed
0.09 0.58
0.53
0.49
0.64
0.77 0.91 0.28 0.51
P
0.89
0.87 0.85
0.99 0.98 1.0 0.95 0.86 0.77 0.83 0.98
Expected
Other Not included 0.89 0.89 0.89 0.89 0.90 0.92 0.91 0.89 Not included 0.90 0.89 Not included 0.89 Not included Not included Not included Not included Not included Not included
Observed
0.02
0.56 0.65
0.43 0.27 0.30 0.37 0.61 0.04 0.27 0.24
P
Clin J Pain
All examined items are listed, and those included in each scale are reported with observed and expected item-rest score correlations and P-values. Items not fitting and thus excluded from the scale are indicated in the table as not included. *Numbers refer to supplemental questionnaire.
Laundry (item 40) Reach up (item 41) Carry grocery (item 42) Work in kitchen (item 43) House cleaning (item 44) Walk 1 km (item 45) Bicycle 1 km (item 46) Swimming (item 47) Open door (item 48) Drive car (item 49) Lift children (item 50) Put on socks (item 51) Wash hair (item 52) Bend down/kneel (item 53) Put on sweater (item 54) Put on bra (item 55) Put on coat (item 56) Sleep (item 57) Get out of bed (item 58) Sex (item 59)
Item*
TABLE 3. Item Fit Statistics for Items Selected for the Different Cause Scales
Andersen et al Volume 31, Number 9, September 2015
800 | www.clinicalpain.com Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clin J Pain
Volume 31, Number 9, September 2015
Assessing Physical Function After Breast Cancer Treatment
TABLE 4. Spearman Correlation Coefficients Between the Cause Scales, Q-DASH, and Single Items Concerning Sequelae
Pain in Area of Surgery
Musculoskeletal Pain
Sensory Disturbances
Lymphedema
Other Causes
Q-DASH Score
0.46 0.56
0.30 0.41
0.21 0.29
0.30 0.38
0.2 0.13
0.52 0.63
0.51 0.17
0.43 0.43
0.29 0.20
0.27 0.17
0.17 0.16
0.63 0.38
0.30 0.23 0.33 0.22 0.27 0.21 0.26 0.09 0.21
0.21 0.17 0.22 0.07 0.22 0.16 0.28 0.28 0.23
0.20 0.16 0.10 0.09 0.20 0.01 0.17 0.28 0.20
0.26 0.14 0.16 0.24 0.18 0.02 0.15 0.14 0.15
0.16 0.17 0.12 0.02 0.14 0.03 0.09 0.09 0.21
0.39 0.35 0.33 0.28 0.33 0.22 0.38 0.36 0.32
0.38 0.32 0.24
0.26 0.21 0.19
0.23 0.24 0.16
0.39 0.46 0.48
0.20 0.13 0.13
0.49 0.42 0.33
Pain Pain, surgery (y/n) Pain, surgery NRS (maximum/ weekly)* Pain in movement NRS Pain elsewhere NRSw Sensory disturbances Pins and needles Electric shock/jumping Burning Numbness Light touch: pain Cold: pain Pain and itch In hands In feet Lymph edema Sensation Visible In treatment Q-DASH score Full population Subpopulationsz
0.51 0.46y
0.60 0.648
0.35 0.36#
0.44 0.51z
0.34
Correlation coefficients are shown with negative values for dichotomized questions. *Pain, surgery NRS (maximum, weekly): highest NRS score in the 4 locations in patients experiencing pain on a weekly basis. wPain in locations outside the area of surgery. zCorrelation between cause scales and Q-DASH scores are shown for the full cohort and for the subpopulations according to the following single items. yPain in the breast, side of chest, axilla, or arm: NRSZ4. 8Pain in locations outside area of surgery: NRSZ4. #Sensory disturbances: pins and needles in the area of surgery. zLymphedema: in treatment for lymphedema. NRS indicates numerical rating scale; Q-DASH, Quick-Disability of Arm, Shoulder, and Hand.
that other factors such as analgesic or other treatments may introduce bias. Finally, the test-retest reliability was not assessed due to the study design where we focused on construction and validation of the scales. Therefore, the scales should be further validated and test-retest reliability and responsiveness to change be described in future studies, including other breast cancer populations and calibration of the scales could be provided using objective measurements of physical function and sequelae.
Clinical and Research Perspectives Applying the 5 scales in the current population indicated that almost half of the women with breast cancer had reduced physical functioning due to their breast cancer disease and treatment across the cause dimensions. The sequelae that most patients reported causing reduced physical functioning was musculoskeletal pain, where 23%, for example, reported having problems bending down due
TABLE 5. Means, SDs, and Range of the Cause Scales of Breast Cancer Patients With Reduced Physical Function Pain surgical Musculoskeletal pain Sensory disturbances Lymphedema Other causes
Copyright
r
N
Mean
SD
Range
83 116 41 51 188
18.59 21.47 19.73 18.83 13.02
16.30 15.97 14.64 13.40 11.72
4.17-65.00 4.17-66.67 5.56-57.14 3.85-54.55 3.85-58.33
to this type of pain. Musculoskeletal pain is a known complication of endocrine therapy9 as well as an age-related problem in the general population.44 Potentially, scales such as those described in the current study could be applied as initial screening tools, providing the oncologist an initial impression of the causes of impaired physical functioning in the individual patient. Patients with impaired physical functioning could be monitored for more specific treatments at pain clinics or by lymphedema therapists. Furthermore, the present cause scales can be used in research to monitor patients in various contexts, for example, in trials examining adverse effects after surgical and adjuvant treatments or in trials examining analgesics and other treatments.
CONCLUSIONS Across the cause dimensions, almost half of the women with breast cancer indicated having reduced physical functioning due to their breast cancer disease and/or its treatment. Each scale revealed different information on causes of reduced function. The present breast cancer sequelae cause scales can be used as a measurement tool to evaluate the impact of specific sequelae after breast cancer treatment on physical functioning. These scales may be applied for research and in follow-up in the breast oncology setting to systematically monitor and manage sequelae, as well as facilitate targeted clinical interventions to optimize rehabilitation.
2014 Wolters Kluwer Health, Inc. All rights reserved.
www.clinicalpain.com |
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
801
Clin J Pain
Andersen et al
ACKNOWLEDGMENT The authors thank Maj-Britt Jensen, MSc, Danish Breast Cancer cooperative Group, Copenhagen, Denmark; for assistance in providing data from the DBCG database, Per Sjøgren, MD, PhD, Section for palliative medicine, Rigshospitalet, Copenhagen, Denmark; for thoughtful comments on the manuscript, and Sebastian Jud, MD, Universta¨tsklinikum Erlangen, Germany; for the drawings used in the questionnaire. REFERENCES 1. Youlden DR, Cramb SM, Dunn NA, et al. The descriptive epidemiology of female breast cancer: an international comparison of screening, incidence, survival and mortality. Cancer Epidemiol. 2012;36:237–248. 2. Andersen KG, Kehlet H. Persistent pain after breast cancer treatment: a critical review of risk factors and strategies for prevention. J Pain. 2011;12:725–746. 3. Gartner R, Jensen MB, Nielsen J, et al. Prevalence of and factors associated with persistent pain following breast cancer surgery. JAMA. 2009;302:1985–1992. 4. Mejdahl MK, Andersen KG, Gartner R, et al. Persistent pain and sensory disturbances after treatment for breast cancer: six year nationwide follow-up study. BMJ. 2013;346:f1865. 5. Andersen KG, Jensen MB, Kehlet H, et al. Persistent pain, sensory disturbances and functional impairment after adjuvant chemotherapy for breast cancer: cyclophosphamide, epirubicin and fluorouracil compared with docetaxel + epirubicin and cyclophosphamide. Acta Oncol. 2012;51:1036–1044. 6. DiSipio T, Rye S, Newman B, et al. Incidence of unilateral arm lymphoedema after breast cancer: a systematic review and meta-analysis. Lancet Oncol. 2013;14:500–515. 7. McWayne J, Heiney SP. Psychologic and social sequelae of secondary lymphedema: a review. Cancer. 2005;104:457–466. 8. Paskett ED, Dean JA, Oliveri JM, et al. Cancer-related lymphedema risk factors, diagnosis, treatment, and impact: a review. J Clin Oncol. 2012;30:3726–3733. 9. Din OS, Dodwell D, Wakefield RJ, et al. Aromatase inhibitorinduced arthralgia in early breast cancer: what do we know and how can we find out more? Breast Cancer Res Treat. 2010;120: 525–538. 10. Groenvold M, Fayers PM, Petersen MA, et al. Breast cancer patients on adjuvant chemotherapy report a wide range of problems not identified by health-care staff. Breast Cancer Res Treat. 2007;103:185–195. 11. Hidding JT, Beurskens CH, van der Wees PJ, et al. Treatment related impairments in arm and shoulder in patients with breast cancer: a systematic review. PLoS One. 2014;9:e96748. 12. World Health Organization. International Classification of Functioning, Disability and Health. Geneva, Switzerland: World Health Organization; 2001. 13. Montazeri A. Health-related quality of life in breast cancer patients: a bibliographic review of the literature from 1974 to 2007. J Exp Clin Cancer Res. 2008;27:32. 14. Rietman JS, Dijkstra PU, Hoekstra HJ, et al. Late morbidity after treatment of breast cancer in relation to daily activities and quality of life: a systematic review. Eur J Surg Oncol. 2003;29:229–238. 15. Gartner R, Jensen MB, Kronborg L, et al. Self-reported armlymphedema and functional impairment after breast cancer treatment—a nationwide study of prevalence and associated factors. Breast. 2010;19:506–515. 16. Bennett MI, Attal N, Backonja MM, et al. Using screening tools to identify neuropathic pain. Pain. 2007;127:199–203. 17. Kennedy CA, Beaton DE, Solway S, et al. Disabilities of the Arm, Shoulder and Hand (DASH). The DASH and QuickDASH Outcome Measure User’s Manual. Toronto, ON, Canada: Institute for Work & Health; 2011. 18. Leblanc M, Mao JJ, Stineman M, et al. Validation of QuickDASH outcome measure in breast cancer survivors for upper extremity disability. Arch Phys Med Rehabil. 2014;95: 493–498.
Volume 31, Number 9, September 2015
19. Harrington S, Michener LA, Kendig T, et al. Patient-reported upper extremity outcome measures used in breast cancer survivors: a systematic review. Arch Phys Med Rehabil. 2014;95:153–162. 20. Moller S, Jensen MB, Ejlertsen B, et al. The clinical database and the treatment guidelines of the Danish Breast Cancer Cooperative Group (DBCG); its 30-years experience and future promise. Acta Oncol. 2008;47:506–524. 21. Blichert-Toft M, Christiansen P, Mouridsen HT. Danish Breast Cancer Cooperative Group—DBCG: history, organization, and status of scientific achievements at 30-year anniversary. Acta Oncol. 2008;47:497–505. 22. Linden WJ, Hambleton RK. Handbook of Modern Item Response Theory. New York: Springer Verlag; 1997. 23. Christensen KB, Kreiner S, Mesbah M. Rasch Models in Health. London, UK: Wiley-ISTE; 2013. 24. Fischer G, Molenaar IW. Rasch Models. New York: Springer Verlag; 1995. 25. Rasch G. Probabilistic Models for Some Intelligence and Attainment Tests. Copenhagen, Denmark: Nielsen & Lydiche; 1960. 26. Andersen EB. A goodness of fit test for the Rasch model. Psychometrika. 1973;38:123–140. 27. Kreiner S. A note on item-rest score association in Rasch models. Appl Psychol Meas. 2011;35:557–561. 28. Kreiner S, Christensen KB. Exact evaluation of bias in Rasch model residuals. Volume 12. In: Baswell AR, ed. Advances in Mathematics Research. New York: Nova Publishers; 2012:19–40. 29. Kelderman H. Loglinear Rasch model tests. Psychometrika. 1984;49:223–245. 30. Cronbach L. Coefficient alpha and the internal structure of tests. Psychometrika. 1951;16:297–334. 31. Streiner DL, Norman GR. Health Measurement Scales. Oxford, UK: Oxford University Press; 2008. 32. Bland JM, Altman DG. Cronbach’s alpha. BMJ. 1997;314:572. 33. Ware JE Jr., Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med Care. 1992;30:473–483. 34. Aaronson NK, Ahmedzai S, Bergman B, et al. The European Organization for Research and Treatment of Cancer QLQC30: a quality-of-life instrument for use in international clinical trials in oncology. J Natl Cancer Inst. 1993;85: 365–376. 35. Sprangers MA, Groenvold M, Arraras JI, et al. The European Organization for Research and Treatment of Cancer breast cancer-specific quality-of-life questionnaire module: first results from a three-country field study. J Clin Oncol. 1996;14:2756–2768. 36. Brady MJ, Cella DF, Mo F, et al. Reliability and validity of the functional assessment of cancer therapy-breast quality-oflife instrument. J Clin Oncol. 1997;15:974–986. 37. Daut RL, Cleeland CS, Flanery RC. Development of the Wisconsin Brief Pain Questionnaire to assess pain in cancer and other diseases. Pain. 1983;17:197–210. 38. McCarthy M Jr., Jonasson O, Chang CH, et al. Assessment of patient functional status after surgery. J Am Coll Surg. 2005;201:171–178. 39. Ringsted TK, Wildgaard K, Kreiner S, et al. Pain-related impairment of daily activities after thoracic surgery: a questionnaire validation. Clin J Pain. 2013;9:791–799. 40. Franchignoni F, Ferriero G, Giordano A, et al. Psychometric properties of QuickDASH—a classical test theory and Rasch analysis study. Man Ther. 2011;16:177–182. 41. Turk DC, Dworkin RH, Burke LB, et al. Developing patientreported outcome measures for pain clinical trials: IMMPACT recommendations. Pain. 2006;125:208–215. 42. Rosenbaum PR. Criterion-related construct-validity. Psychometrika. 1989;54:625–633. 43. Hansen PE, Siersma V, Ross L, et al. Psychometric properties of brief indexes designed to measure social-cognitive predictors of smoking initiation. Drug Alcohol Depend. 2007;88:64–74. 44. Kurita GP, Sjogren P, Juel K, et al. The burden of chronic pain: a cross-sectional survey focussing on diseases, immigration, and opioid use. Pain. 2012;153:2332–2338.
802 | www.clinicalpain.com Copyright r 2014 Wolters Kluwer Health, Inc. All rights reserved. Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
