Abstract
Objective
Longitudinal validity refers to the ability of an outcome measure to detect clinically meaningful change in status, also referred to as responsiveness. Most recently the Department of Health has been a proponent of increased use of patient-reported outcome measures in evaluating the effectiveness of interventions to improve health. The Disability of the Arm, Shoulder and Hand (DASH) questionnaire is a comprehensive self-report measure that has been used well in rehabilitation. More recently a shortened version – the Quick DASH has been developed.
Methods
This study compared the longitudinal validity of the DASH and Quick DASH in a cohort of musculoskeletal outpatients with hand trauma and degenerative hand conditions attending an outpatient occupational therapy (OT) rehabilitation service. Responsiveness was calculated using effect sizes (ES) and standardized response mean (SRM) and paired t-tests to test for statistically significant differences.
Results
Twenty-two patients completed questionnaires. They received a mean of 12 weeks (SD = 7.5) of OT treatment. Mean baseline and discharge scores for DASH were 50.20 points (SD = 22.27) and 19.43 points (SD = 15.61), respectively and for the Quick DASH 50.43 (SD = 21.22) and 19.70 (SD = 16.01), respectively. Both the full DASH and Quick Dash yielded similar ES and SRM statistics (DASH: ES = 1.38; SRM = 1.93) (QDASH: ES = 1.51; SRM= 1.77) and t-tests showed statistically significant changes from baseline to discharge DASH (t = 9.06; P < 0.01); Quick DASH (t = 8.30, P < 0.01).
Conclusions
The Quick DASH and DASH demonstrated similarly high responsiveness in this patient population.
Introduction
The ability of an outcome measure to detect clinically meaningful change over time is defined as longitudinal validity, also referred to as responsiveness. This concept is always linked to the unit of measurement, the level of data collected the range of values and the minimal detectable change available on a scale. Thus nominal and ordinal data is likely to be less responsive to change than interval data. However, the more discrete the interval scale the more likely the measure will detect ‘noise’ (variation or measurement error) and potentially reduce signal to noise ratio. 1 Conversely, a more sensitive interval scale could therefore have poorer responsiveness (as it is more sensitive to measurement error) than a scale with fewer measurement or outcome options (which is less sensitive to measurement error). It is therefore difficult to predict which measures will perform best over time.
Patients' and health professionals' perspectives of disease and functional ability differ; 2,3 so self-report outcome measures have a clear place in recording the effectiveness of hand therapy in rehabilitation programmes. It has been increasingly important to gain patients' views on therapy outcome 4 and most recently the Department of Health have issued recommendations to increase the use of patient-reported outcome measures in documenting effectiveness of health services. 5,6
A review of the ability of two commonly used self-report outcome measures to detect and record patient-reported change following outpatient hand therapy was timely and appropriate. Both the full DASH 7 and Quick DASH 8 questionnaire have shown sound responsiveness over a 12-month rehabilitation period following upper limb surgery, 9 but to date there has been no detail regarding shorter time periods that are more reflective of UK rehabilitation practices in outpatient hand therapy departments.
The aim of this study was to compare the longitudinal validity of the full Disability of the Arm, Shoulder and Hand (DASH) questionnaire and the Quick DASH in a hand therapy outpatient population receiving hand therapy over three months.
Methods
The study was designed as a longitudinal cohort study recruiting consecutive musculoskeletal occupational therapy (OT) outpatients treated by an extended scope OT practitioner (KW). The treating therapist approached consecutive patients referred by consultant hand surgeons, the Hampshire Primary Care Trust multi-professional triage team practitioners and local general practitioners and asked them to complete the two questionnaires. There were no exclusion criteria for this study and every patient referred to OT was included. The study was approved as an audit of the responsiveness of current self-reported outcome measures routinely used within the department and therefore patient consent to take part was not required. Patient treatment was unaffected by the study and these forms were currently in routine use in the department there with no alteration to patient care. Full permission to conduct this responsiveness analysis was given by the NHS Trust's Audit Department.
DASH and Quick DASH questionnaires were completed and scored at time of referral to OT (baseline) and on discharge from treatment. Only the disability/symptom section of the full DASH (30 items scored from 1 to 5) and Quick DASH (11 items scored from 1 to 5) were used. It was hypothesized that DASH scores would improve (i.e. reduce) in all cases.
Statistical analysis
Distribution-based methods were used for analysis. A distribution approach does not use an external criterion to define what constitutes clinically meaningful change. For example if patients or clinicians were to provide their own rating of what they felt was a meaningful change in hand function status this would act as an external criterion or reference on which to measure or ‘anchor’ meaningful change. Without this ‘anchor’ or external reference, distribution methods are in part limited in their ability to interpret change. However, what distribution methods do achieve is an estimate of ‘signal to noise’ ratio. Distribution-based methods allow calculation of how much change is measurement error or variation and how much is true change in status. They are linked to measurement error and the variation of the sample population. As such responsiveness statistics are population-specific. 10 If an outcome measure has a large amount of measurement error, then these methods would be less likely to detect large effect sizes (ES). They simply provide an estimate of the magnitude of treatment effect and Cohen 11 provides useful methods for interpreting ES, whereby an ES of 0.2 represents a small clinically important effect, 0.5 a medium effect and 0.8 a large effect.
Standardized response mean (SRM), ES and paired t-tests were used to compare responsiveness between the two questionnaires. SRM and ES were calculated using the following formulae: SRM was calculated by dividing the mean change in scores by the SD of that change and ES calculated by dividing the mean change in scores by the SD of baseline scores. The higher the SRM or ES, the greater the level of responsiveness.
Results
Forty-eight patients were given the DASH and Quick DASH questionnaires to complete at the start of their treatment. Of these, 22 (46%) patients completed the questionnaire to enable scores to be calculated. Patients were referred for therapy following trauma, surgery, degenerative osteoarthritis and non-specific wrist and hand pain. Outpatient OT intervention lasted for a mean of 12 weeks and patients received a mean of nearly 2.15 hours of total treatment time with the treating occupational therapist. There was a range of interventions provided. Table 1 shows the demographic and intervention details.
Demographic and intervention details for the patients attending outpatient occupational therapy (OT)
The DASH and Quick DASH both showed high responsiveness levels. Table 2 shows the baseline and follow-up scores of both questionnaires with respective SRM, ES and significance values for differences between baseline and follow-up scores.
Baseline and discharge DASH and Quick DASH scores with responsiveness statistics and significance levels
DASH, Disability of the Arm Shoulder and Hand questionnaire; SRM, standardized response mean; ES, effect sizes
Discussion
The Quick DASH was as responsive as the full DASH in this small sample of musculoskeletal OT outpatients treated during a short term. Both questionnaires demonstrated high levels of responsiveness as defined by Cohen. 11 The full DASH had a higher SRM and the Quick DASH a higher ES. While the SRM and ES differed between measures this has already been reported as not uncommon. 12 The results of this study indicate that the Quick DASH is as effective at detecting change as the full DASH in this setting with this group of patients.
Gummesson et al.'s 9 study of longitudinal validity on patients assessed over a 12-month period showed lower rates of responsiveness in both the full DASH and Quick DASH. However, their sample also showed lower mean disability at presurgery baseline and less improvement or change in upper limb disability over 12 months in comparison with our sample. As all responsiveness statistics are context- and setting-specific, 10 it is understandable that different clinical populations treated in different settings may produce different results.
The Quick DASH is quicker and easier to complete and score for both the patient and the clinician, respectively, so arguably has higher clinical utility than the full DASH without losing its ability to discriminate between longitudinal change over time. Clinicians should have confidence that the Quick DASH can detect clinical change over a short 12-week rehabilitation period in patients attending outpatient therapy. However, the full DASH is reported to provide greater precision, 13 so may be better at monitoring individual progress rather than group change.
This study is not without limitations. The results can only be applied to a similar population to this sample. Inferences cannot be made to other patient populations or other types of settings. The cohort of patients recruited for this study was a convenience sample, so a larger more diverse sample may produce different results.
Limitations
While this study measured only a small number of outpatients, the small sample size should not have adversely affected the statistical analysis. Both SRM and SE do not utilize standard error to calculate summary statistics so therefore should not be adversely affected by small sample sizes. The therapy provided for these individuals was given by an expert-extended scope practitioner, so arguably these could be the ‘best case scenario’ for improvement in status over time. Finally, there were no ‘anchor-based’ statistical methods for judging the clinical meaning for this change; only distribution-based statistical assessment was used. This is acknowledged as a limitation and one that could be easily rectified in subsequent studies by requesting patients to complete their own evaluation of whether they felt that they had improved their hand function or not after therapy.
Footnotes
Acknowledgements
Mr Warwick and Mr Hargreaves, Hand Surgeons at the Southampton University Hospitals Trust and Mr Richards from The Royal Bournemouth Hospital for patient referral.
Competing interests
None declared.