Fracture of the wrist is a common injury and is especially prevalent in older post-menopausal women. Wrist fracture is associated with variable outcome, for example, approximately 50% of cases fail to recover pre-operative function [1] and 20% of patients report moderate to severe pain at 6 month post-fracture [2]. A recent series of Cochrane reviews of the effectiveness of commonly applied interventions for wrist fracture have highlighted the continued unsatisfactory outcome for many patients [3]. Further, there is an association between wrist fracture and long-term disability and increased risk of dependency [4]. Any measure of outcome from wrist fracture needs to be able to index the impact of fracture, not only on the impairment itself but also on the level of associated disability.

The development of core measurement sets for patients with musculoskeletal conditions or for acute hospital and early post-acute rehabilitation are available [8–10], and these may be relevant to patients with wrist fracture. Whilst, the core measurement sets detail the categories within each health outcome that need to be assessed for a given health condition, they do not identify the actual items to be used to measure each category. The ability of existing health outcome measures to operationalise the ICF is an area of active research [11–13] and it is possible that existing wrist fracture outcome measures may contain items that measure each ICF outcome. That said, in order for the ICF to be tested as a model of the outcomes from wrist fracture it is not only necessary to have measures of each ICF outcome but those measures need to show discriminant validity. Such pure measures are necessary to ensure that any observed relationship between the three outcomes is not simply due to measurement confound.

The method of Discriminant Content Validity (DCV) is able to establish the discriminant validity of measurement items [13, 14]. DCV examines the relationship between individual measurement items and all constructs within a theoretical model thereby establishing the content validity of a measurement item against all constructs within a given theory. Rather than judge items against a single theoretical construct, the DCV method establishes whether each theoretical construct can be measured discriminately because the method asks judges to indicate the extent to which an item matches each theoretical construct of interest, in this case the three main outcomes identified by the ICF. This method has been used to establish the content validity of items within existing orthopaedic and chronic pain measures [13, 14]. In this study we used DCV to identify ICF outcomes measured by the Disabilities of the Arm, Shoulder and Hand Questionnaire (DASH) [15].

The DASH contained 27 pure construct measures and can, therefore, be used to measure each ICF construct without measurement confound.

A recently published study used 4 judges to link the ICF to the DASH items [21]. In this study the ICF coding taxonomy was employed and individual DASH items were assigned one or more ICF code. This strategy enables the identification of I items but does not enable A to be distinguished from P because the ICF coding taxonomy fails to discriminate between these two theoretically distinct constructs. Nonetheless, these findings were consistent with the DCV analysis; with the exception of one item, all items identified as measuring I in the current study were assigned to ICF codes within the body functions category in the linkage study. The exception was item 29 which was not classified in the current study due to a lack of agreement between judges; in the Drummond et al study item 29, which concerns sleeping difficulties, received a body function coding. The I items within the DASH primarily focus on pain and muscle stiffness or weakness, previous studies have identified pain items as measures of the impairment construct [13].

All other items (except item 30) were coded as "activities and participation" by Drummond et al. In the current study these items were also coded as A or P or AP. However, the current data enables A and P items to be distinguished and enables mixed items to be clearly identified. This has the advantage of enabling the discriminant measurement of each type of outcome.

The availability of discriminant outcome measures is especially important in assessing outcomes in intervention studies. Being able to assess the impact of an intervention on I, A and P outcomes, in a manner that is free of measurement confound, provides the means to assess the success of an intervention within each outcome domain. For example, in the case of wrist fracture, x-ray may indicate the fracture (I) to be healed but the patient may not have regained full function (A) and may remain afraid of going outside (P). Measurement of outcome in this patient would be over positive if only the impairment outcome were measured. The DASH produces a single score; consequently it does not distinguish between the three ICF health outcomes as it is currently used. However, its content is capable of producing an impairment score; an activity limitations score and a participation restriction score as well as the standard single score. The standard method of scoring the DASH generates a simple mean score across all items and converts this mean score to a 0–100 scale. The standard method of scoring the questionnaire could be applied to the items measuring each health outcome, such that, a respondent would generate a score between 0–100 for each health outcome [see Additional files 1 and 2]. We do not advocate the standard method of scoring the DASH be replaced. Rather, the availability of the DCV analysis of the DASH for clinicians and researchers enables the instrument to be individualised to the requirements of a given situation, be that the particular needs of an individual patient or the requirements of a specific research question. Consequently, it enables the existing evidence base to be developed cumulatively.

This study adds to previous research in identifying pure measurement items for each of the three ICF outcomes. It provides a means to generate three distinct outcome measures for arm, shoulder and hand trauma, namely impairment, activity limitations and participation restrictions. As a consequence, the impact of interventions, such as surgery, on each health outcome can be identified.