If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address correspondence to: Heike Schmidt, MD, Medical Faculty, Institute for Health and Nursing Science, Martin-Luther-University Halle-Wittenberg, Magdeburger Strasse 8, Halle 06097, Germany.
Cancer patients frequently suffer from various symptoms often impairing functional status and quality of life. To enable timely supportive care, these symptoms must be assessed adequately with reliable tools.
Objectives
This study aimed to validate the German version of the M. D. Anderson Symptom Inventory (MDASI).
Methods
This was a multicenter, cross-sectional, observational study. At five German university hospitals, 697 cancer patients aged from 18 to 80 years undergoing active anticancer treatment were recruited to participate in the study. For the validation, reliability (Cronbach's alpha), construct validity (factor analysis), known group validity (Eastern Cooperative Oncology Group Performance Status), and convergent divergent analyses were calculated.
Results
Of the 980 patients who were eligible, 697 patients were included and agreed to participate in the study (71%). Reliability analysis showed good internal consistencies for the MDASI set of symptoms (Cronbach's alpha coefficient = 0.82; 95% CI = 0.78, 0.84) and for the set of interference items (Cronbach's alpha coefficient = 0.857; 95% CI = 0.484, 0.87). Factor analysis resulted in a one-factor solution (general symptoms; eigenvalue = 4.26) with a psychological (distress and sadness) and a gastrointestinal subscale (nausea and vomiting). Convergent and divergent analyses showed significant correlations between symptom burden and distress and global health-related quality of life (subscale of the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-C30 Version 3.0.).
Conclusion
The MDASI-German version is a valid tool for measuring patient-reported symptom severity and symptom interference in German cancer patients. It is easily applicable and can be used by German clinicians and researchers for screening and monitoring purposes and the comparison of international data.
Cancer patients often suffer from various disease- or treatment-related symptoms that may impair their functional status and result in high symptom burden. Unrelieved symptoms can limit therapy options and reduce quality of life.
In clinical practice, symptoms might persist unrecognized and undertreated because, when not asked, patients might not report their symptoms exhaustively. Furthermore, considerable numbers of clinicians underestimate symptom intensity.
Therefore, to enable tailored supportive care measures, all relevant symptoms have to be assessed correctly and frequently including the patients' perceptions. Patient-reported outcomes (PROs) are more and more accepted as significant measures of symptom intensity and interference as well as of health-related quality of life.
Patients report improvements in continuity of care when quality of life assessments are used routinely in oncology practice: secondary outcomes of a randomised controlled trial.
However, comprehensive standardized assessments are still not widely used in daily clinical practice despite the many available and valid questionnaires. To facilitate implementation of PRO measures in everyday practice, it is important not only to provide valid but also feasible questionnaires that assess relevant symptoms.
The M. D. Anderson Symptom Inventory (MDASI) is a comparatively short self-administered questionnaire that was developed and validated to measure symptom intensity and interference in cancer patients.
It comprises 19 numeric rating scales (NRSs) regarding the presence and intensity of common symptoms and functional restrictions. The MDASI assesses the severity of 13 symptoms at their worst in the last 24 hours on a zero to 10 NRS, with zero being “not present” and 10 being “as bad as you can imagine.”
The symptom scales represent two underlying structures, namely a general symptom severity factor (pain, fatigue, disturbed sleep, distress [emotional], shortness of breath, drowsiness, dry mouth, sadness, difficulty remembering, and numbness or tingling) and a gastrointestinal factor (nausea and vomiting); lack of appetite loads on both the factors.
Symptom interference with daily activities is measured by six functional scales regarding general activity, mood, work, relations with others, walking, and enjoyment of life. Interference is also rated on a zero to 10 NRS, zero being “did not interfere” and 10 being “interfered completely.” The mean of the interference items can be used to represent overall symptom distress. Symptom burden is defined as the sum of symptom severity and symptom interference.
Because scores of single items can be directly understood and implied in daily care without further computing, the MDASI can be used easily to screen or to monitor symptoms throughout the course of the treatment. It has been translated and validated in many languages including French, Taiwanese, and Russian.
The main objective of this study was to gather representative information about symptom severity, symptom interference, and symptom burden within a large heterogeneous population of cancer patients undergoing active anticancer treatment in different settings. To test and provide a relatively short disease-specific instrument with a short recall period feasible for screening and monitoring for future use in clinical routine, we decided to use the German version of the MDASI (MDASI-G), which had already been linguistically validated, and to perform a psychometric validation of this tool.
Methods
Study Design
The study was designed as a multicenter, cross-sectional, observational study to investigate symptom severity and symptom interference in a heterogeneous population of cancer patients in different settings. The validation study was carried out alongside the large descriptive study. Recruitment took place at the oncology departments of five German university hospitals and aimed to fulfill convenience samples of 150 inpatients and outpatients per center.
Participants
Inclusion Criteria
Patients aged between 18 and 80 years, diagnosed with cancer, and undergoing active anticancer treatment with an Eastern Cooperative Oncology Group Performance Status (ECOG PS) of three or lower who gave written informed consent were eligible to participate in the study.
Exclusion Criterion
Patients lacking sufficient knowledge of the German language were not eligible.
Data Sources
To perform the psychometric validation, we used the linguistically validated MDASI-G provided by the M. D. Anderson Cancer Center. In addition, patients filled out the Distress Thermometer (DT), which is a short screening tool for measuring self-reported distress on a zero to 10 NRS,
and the two general questions of the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-C30 (EORTC QLQ-C30), Version 3.0, regarding global health-related quality of life.
The European Organization for Research and Treatment of Cancer QLQ-C30: a quality-of-life instrument for use in international clinical trials in oncology.
Demographic and disease-related data (age, gender, marital status, education level, disease type, model of care, and type of treatment) also were collected.
Ethical Considerations
As study data were not available for the physicians treating the participants, there was no direct advantage, for example application of supportive measures, for the participants. To reduce the burden on the patients, we limited the number of items to be answered and did not use the problem checklist of the DT or another comparable reference questionnaire assessing symptoms and functional impairments. Patients who were not willing to participate also did not give permission to save any data. Therefore, reasons for nonparticipation could not be elicited. The study was approved by the local ethics committees of the participating university hospitals.
Statistical Analysis
Scoring of the symptom severity and symptom interference scales including the handling of missing values was carried out as described in the MDASI User Guide.
Descriptive statistics were used to give an account of symptom prevalence, severity, and interference. In accordance with the methodology used in the original English language validation study and studies validating the MDASI for foreign languages, our validation analysis plan included examination of reliability, known-group validity, and analysis of convergence and divergence. To establish reliability, we examined the internal consistency (Cronbach's alpha coefficient). To establish convergent validity, we performed a convergent and divergent analysis by testing the correlation of symptom burden (sum of the means of symptom severity and symptom interference) with DT and global health-related quality of life (EORTC QLQ-C30 global scale). To examine the underlying constructs that the MDASI-G is supposed to measure, we performed an exploratory factor analysis on our validation sample. Because other language versions of the MDASI showed various factor solutions,
we chose exploratory factor analysis over confirmatory factor analysis to better understand the constructs being assessed by the MDASI-G. All analyses were carried out using SPSS Version 18 (SPSS, Inc., Chicago, IL).
Results
Participants
Of the 980 patients who were eligible, 697 patients were included and agreed to participate in the study (71%). Recruitment sites were internal medicine (n = 268, 38.5%), gynecology (n = 146, 20.9%), surgery (n = 104, 14.9%), radiotherapy (n = 93, 13.3%), urology (n = 58, 8.3%), head and neck (n = 19, 2.7%), and dermatology (n = 7, 1.0%) wards. Recruitment rates are shown in Table 1; demographic and disease-related data are shown in Table 2. As nonparticipants did not give permission to collect any data, the reasons for nonparticipation could not be examined.
Mean age had to be computed for n=588 participants because one center documented age groups (n=109). However, no statistical difference between age groups at the recruiting centers was found, P=0.7.
Patients may have received one or more treatments.
Operation
407 (58.4)
Chemotherapy
511 (73.3)
Radiotherapy
195 (28.0)
ECOG PS=Eastern Cooperative Oncology Group Performance Status.
a Mean age had to be computed for n=588 participants because one center documented age groups (n=109). However, no statistical difference between age groups at the recruiting centers was found, P=0.7.
b Patients may have received one or more treatments.
Eleven patients did not complete the required seven items of symptom severity and two patients did not complete the required four items of symptom interference.
Following the National Comprehensive Cancer Network guidelines for the assessment of pain
we defined symptom severity as mild if it was rated between one and three, moderate if it was rated between four and six, and severe if it was rated equal or greater than seven on the zero to 10 NRS. Descriptive results for means, symptom prevalence, and severity are presented in Table 3.
Table 3Descriptive Results of MDASI-G Referring to the Last 24 Hours
We examined internal consistency by calculating the Cronbach's alpha coefficient. Following the rule of thumb by George and Mallery, alpha values greater than 0.9 are rated excellent, greater than 0.8 as good, greater than 0.7 as acceptable, and values lower than 0.6 as doubtful.
Analysis showed good internal consistencies for the MDASI set of symptom items (Cronbach's alpha coefficient = 0.82; 95% CI = 0.80, 0.84) and for the set of interference items (Cronbach's alpha coefficient = 0.84; 95% CI = 0.82, 0.86; Table 3).
Construct Validity
Construct validity was assessed by factor analysis regarding the symptom scales. The data were suitable for factor analysis (Kaiser-Meyer-Olkin criterion = 0.80). The correlation matrix with Z-standardized values showed highest correlations between distress and sadness (r = 0.78; 95% CI = 0.74, 0.82) and nausea and vomiting (r = 0.67; 95% CI = 0.58, 0.74). Lowest correlations were found between vomiting and difficulty remembering (r = 0.087; 95% CI = −0.007, 0.18) and between poor appetite and numbness (r = 0.07; 95% CI = −0.02, 0.16).
The number of factors was identified using eigenvalues together with the scree plot and parallel analysis. The scree plot shows the factors against the respective eigenvalues (Fig. 1). Parallel analysis “involves extracting eigenvalues of random data sets that parallel the actual data set with regard to the number of cases and variables. Factors are retained as long as the ith eigenvalue from the actual data is greater than the ith eigenvalue of the random data set.”
Results of parallel analysis, eigenvalues, and the scree plot resulted in a possible three-factor solution. Principal axis factor analysis with varimax rotation was carried out for the 13 MDASI-G symptom items. The eigenvalues of the three factors were 4.26, 1.41, and 1.20, respectively explaining 52.9% of the variance (32.8%, 10.9%, and 9.2%, respectively). Factor 1 included affective symptoms with distress, sadness, and sleep disturbance. Factor 2 included general symptoms with fatigue, drowsiness, shortness of breath, dry mouth, difficulty remembering, and numbness. Factor 3 included gastrointestinal symptoms with nausea, vomiting, and lack of appetite. Pain loaded with 0.4 on the first factor, 0.31 on the second, and 0.28 on the third factor.
In analyzing the factor loadings, it must be noted that only distress and sadness and nausea and vomiting had factor loadings higher than 0.8. Factor loadings are shown in Table 4. Reliability analysis for the suggested factor solution showed a Cronbach's alpha coefficient for the first factor of 0.73 (95% CI = 0.69, 0.76), for the second factor of 0.73 (95% CI = 0.68, 0.75), and for the third factor of 0.69 (95% CI = 0.65, 0.73). The first factor, however, showed an increase of Cronbach's alpha to 0.88 if disturbed sleep was deleted and only distress and sadness were tested. For the third factor, Cronbach's alpha increased to 0.78 if poor appetite was deleted. Taking these results into account, we decided on a one-factor solution (general symptoms) with an affective and a gastrointestinal subscale.
Table 4Factor Loadings for Symptom Intensity Items
Symptom Item
Factor Loadings
Factor 1
Factor 2
Factor 3
Distress
0.90
0.08
0.08
Sadness
0.90
0.05
0.15
Disturbed sleep
0.50
0.31
0.11
Pain
0.40
0.31
0.28
Shortness of breath
0.02
0.64
0.15
Fatigue
0.33
0.59
0.35
Drowsiness
0.21
0.59
0.29
Difficulty remembering
0.20
0.59
−0.03
Numbness
0.01
0.57
0.01
Dry mouth
0.18
0.47
0.33
Vomiting
0.05
0.02
0.86
Nausea
0.12
0.13
0.86
Poor appetite
0.22
0.30
0.59
Method: Principal axis factor analysis with varimax rotation.
In addition, we performed a hierarchical cluster analysis to explore the symptom patterns. Results are presented in the dendrogram (Fig. 2). The cluster analysis again reveals high interdependencies between single symptoms, leading to sparsely selective clusters and low intracluster distances
In accordance with the factor analysis, the affective and the gastrointestinal symptom domains are moderately prominent. Thus, the results of the cluster analysis are partly consistent with the factor analysis.
Known-group validity (sensitivity) was examined by comparing the MDASI-G total scores between patients with low functional status (ECOG PS score ≥2) and patients with high functional status (ECOG PS score ≤1). As expected, the total MDASI-G scores for symptom severity, symptom interference, and symptom burden were significantly higher for patients with a low functional status (Table 5).
Table 5Known-Group Validity
Parameters
ECOG PS 0–1 N
ECOG PS 2–4 N
ECOG 0–1 Mean (SD)
ECOG 2–4 Mean (SD)
Mean Difference (95% CI)
P-value
Symptom severity
282
307
20.9 (16.7)
34.7 (18.6)
−13.8 (−16.7, −10.9)
<0.01
Symptom interference
270
297
10.5 (10.0)
26.1 (12.7)
−15.6 (−17.5, −13.7)
<0.01
Symptom burden
270
297
31.6 (24.9)
61.2 (28.5)
−29.5 (−33.9, −25.1)
<0.01
ECOG PS=Eastern Cooperative Oncology Group Performance Status.
To examine convergent validity, we calculated the correlations between symptom burden (mean = 5.2, SD = 3.5), distress (mean = 5.1, SD = 2.7), and global health-related quality of life (mean = 49.4, SD = 22.8). Pearson's correlation coefficient (r) between symptom burden and global health-related quality of life was r = 0.66 (95% CI = −0.70, −0.62) and between symptom burden and distress r = 0.60 (95% CI = −0.56, −0.65). Both were significant two-sided correlations.
Discussion
The study demonstrates that the MDASI-G is a valid and reliable tool for assessing symptom intensity and interference in German cancer patients, consistent with the psychometrically validated versions in other languages. The MDASI-G is applicable for patients with different diagnoses and in different treatment settings. The very small number of “missings” suggests a high degree of compliance and good feasibility for everyday practice. Analysis showed good internal consistencies for the MDASI set of symptom items and for the set of interference items. The calculated values for Cronbach's alpha, with 0.82 for the MDASI set of symptom items and 0.84 for the set of interference items, are comparable with other validation studies.
as 0.91. Construct validity was assessed by factor analysis. After careful consideration, we decided on a one-factor solution with gastrointestinal and affective subscales. This result is consistent with other validation studies identifying underlying constructs of a gastrointestinal factor,
the known-group validity was satisfactory, showing significant differences in symptom burden for patients with “good” and “poor” ECOG PS.
To limit the burden for patients, we did not apply a detailed reference questionnaire to establish concurrent validity but carried out a convergent and divergent analysis with global scores for distress and global health-related quality of life. It would have been interesting, however, to compare the results of a questionnaire not using a zero to 10 scale, for example, the EORTC QLQ-C30 with the MDASI, as was done in other studies.
The descriptive results show to what extent patients are still suffering from symptom burden despite supportive therapy options. Although, corresponding to other studies, the observed low rates of vomiting indicate the success of antiemetic treatment. Comparable with other studies,
fatigue was among the most prevalent and most severe symptoms, a finding that might motivate clinicians to perform screening and offer guideline-based treatment. In addition, distress, sadness, and sleep disturbance were reported frequently with moderate and severe intensity. If assessed on a regular basis, these symptoms could be taken care of early by psycho-oncologic counseling. The result that 20.4% of the patients reported moderate pain should likewise trigger efforts to optimize symptom management. Furthermore, the assessment of symptoms in connection with functional impairments is of importance to anticipate possible supportive needs after discharge and the planning of after-care.
In summary, these findings emphasize the importance of integrating PRO measures in everyday clinical practice. However, “assessment is not enough” to optimize supportive therapy.
Assessment is not enough: a randomized controlled trial of the effects of HRQL assessment on quality of life and satisfaction in oncology clinical practice.
It is of great importance that relevant scores of symptoms, for example, mild and severe are followed by special diagnosis and treatment if necessary. Pathways of diagnosis and treatment have to be established and implemented. Health care professionals have to be trained in interpreting results and acting accordingly. Our findings of symptom severity and functional impairments despite existing supportive options might be interpreted against the background that this process in Germany is still evolving.
The study had several limitations. The cross-sectional design did not allow for examining sensitivity to change. Test-retest reliability also was not addressed. Differing recruitment rates in the participating centers could not be explained because information about reasons for refusal was not available. For our analysis of convergence and divergence, we only measured the two global questions of the EORTC QLQ-C30 and distress. Presuming that symptom burden and global health-related quality of life and distress depict similar underlying constructs, this design was chosen to limit the number of questions for the patients. However, it would have been desirable to examine full convergence and divergence with respect to single symptoms of different instruments. The main strengths of the study were the rather large sample size comprising a comprehensive sample of cancer patients and the multicenter design.
Future research using longitudinal designs could examine sensitivity to change, provide further information about symptom burden of cancer patients over time in different settings and stages, and should include studies to verify levels and cutoff scores for symptoms and functional impairment.
In conclusion, by measuring not only prevalence but also symptom intensity and interference, the MDASI-G can provide a feasible method for screening and monitoring symptom burden in German-speaking countries, thus further facilitating international efforts to implement routine assessment of PROs into clinical practice.
Disclosures and Acknowledgments
This work was supported in part by ProKID (German Cancer Information Service). The authors declare no conflicts of interest.
The authors thank the patients for participating in the study and cooperating so generously; the nursing departments of the five participating University hospitals who carried out the study; and Professor Tito Mendoza, Dipl. Psych., Dirk Rennert, and Dr. rer. Nat. Christine Lautenschläger for their fruitful discussion.
References
Cleeland C.S.
Mendoza T.R.
Wang X.S.
et al.
Assessing symptom distress in cancer patients: the M.D. Anderson Symptom Inventory.
Patients report improvements in continuity of care when quality of life assessments are used routinely in oncology practice: secondary outcomes of a randomised controlled trial.
The European Organization for Research and Treatment of Cancer QLQ-C30: a quality-of-life instrument for use in international clinical trials in oncology.
Assessment is not enough: a randomized controlled trial of the effects of HRQL assessment on quality of life and satisfaction in oncology clinical practice.
00255-3&id=gr1.jpg){kind=link}
00255-3&id=gr2.jpg){kind=link}