Development and validation of the ND10 to measure neck-related functional disability

Neck pain is one of the most common musculoskeletal disorders with one third of all adults experiencing it during the course of one year, and 70% doing so over the course of their lifetime [1]. The severity of disability can range from minor to severely debilitating and the natural history is characterized by episodic reoccurrence [2, 3]. Radiologic [4] or physiologic measures [5] rarely explain neck pain. As a result, accurate measurement of symptom severity and functional disability is essential to targeting treatment and evaluating treatment outcomes. Systematic reviews indicate that baseline pain and disability are the most potent prognostic indicators of future pain and disability outcomes [6, 7].

Musculoskeletal health outcome measures are commonly used to evaluate symptoms, disability, and quality of life, and how this change following an intervention. Neck disorders can cause pain [8, 9]; and disturbances in joint motion [10, 11], sensory function [12‐15], proprioception [16, 17], motor function [18‐20], coordination [5, 21], posture [22, 23], and balance [23, 24]. These can lead to functional disability [2, 25, 26], participation restrictions [27], reduced work capacity [28‐30], and lower quality of life [31, 32]. There are a variety of impairment and disability measures that have been designed to assess these different constructs [23, 33‐38]. A survey of international practice patterns of clinicians with respect to assessing the outcomes for patients with neck pain indicated that the Numeric Pain Rating Scale (NPR) (a single item on pain [37]), the Neck Disability Index (NDI), and the Disabilities of the Arm, Shoulder and Hand (DASH) (developed for upper extremity [39]) are the patient-reported outcome measures (PROMs) most commonly used by clinicians [40].

The two primary features of musculoskeletal conditions, including neck disorders, are pain and disability. Content validity of PROMs requires a clear conceptual foundation with a defined construct [41, 42]. Increasingly, there have been moves to define conceptual frameworks for identification of core constructs as a preliminary step to improving measurement in the field of musculoskeletal disability. A recent international consensus panel [43] identified 6 core domains for whiplash-associated disorders: Physical Functioning, Perceived Recovery, Work and Social Functioning, Psychological Functioning, Quality of Life, and Pain. Many existing measures have not adequately defined a single construct, but rather sample across multiple constructs or domains within a global construct or health condition. A recent international outcome measure core set consensus panel for whiplash disorders concluded that “the content validity of these PROMs has yet to be established… and until this is undertaken, it is not possible to recommend 1 PROM over the other” [35]. Commonly, PROMs for musculoskeletal conditions include items on pain and function and compute total scores—as if these items reflect a single construct. Combining symptoms and disability in a single score from PROMs may not be justifiable on psychometric rounds since these may not represent a single construct. Furthermore, combining different scores together may undermine clinical reasoning or research hypothesis testing since being able to differentiate the impact of interventions on specific constructs is critical to problem-solving and hypothesis testing. Where adequate content validity is not present, measures do not provide accurate information about what aspect of patient status is changing over time [44]. Content validity is a prerequisite to other psychometric properties like factor validity and unidimensionality [45]. Finally, consensus panels have verified that pain and disability are separate constructs that are important core outcomes in health conditions causing neck pain [43].

A variety of PROMs have been previously established. The most of commonly used is the NDI developed by Vernon and Mior [46]. It was constructed based on 5 items adapted from the Oswestry Low Back Pain Index (OLBPI) and an additional 5 new items [46]. The developer published a summary paper in 2008 summarizing a 17-year history with the NDI [47], reflecting its position as the earliest and most commonly used neck-specific PROM. Systematic reviews of the measurement properties of the NDI concluded that there was a deep pool of evidence supporting the NDI as being reliable and responsive, but found validity concerns about the factor structure, and relevance given the number of items left missing in certain populations [33, 48]. Although the NDI is used as if it provides interval-level scaling, Rasch analyses indicate that is not achievable with the original measure [49‐51]. Further, there are substantial differences between the 2 proposed Rasch-based scoring version and the original [52]. A variety of neck-related PROMs have been developed subsequently. An overview of neck-related PROMs [48] found that more limited research on the other neck-related measures: Northwick Park Neck Pain Questionnaire (20 items), Copenhagen Neck Functional Disability Scale (15 items), Neck Bournemouth Questionnaire (7 items), and Neck Pain and Disability scale (20 items).

Construct clarity is important in outcome evaluation. International consensus has concluded that functional disability is an important construct for assessing outcomes in neck-related health problems [43]. However, the wording of many of the current neck PROM items suggests that they measure neck-related pain interference–how much neck pain interferes with function. Pain interference and disability are related but different constructs. It may be problematic when PROMs conflate pain and function or do not specify what they are measuring is pain interference. This construct ambiguity might explain why some factor analyses studies indicate that the NDI contains 2 factors [53‐56]. This is further supported by qualitative studies of experts and patients who suggest that the NDI measures more than physical functioning [44]. Since physical functioning is 1 of the core constructs agreed upon by an international panel [43], a measure that focuses solely on function/disability for people with neck conditions is needed. A recent review of disability measures for whiplash concluded that “the content validity of these PROMs has yet to be established…, and until this is undertaken, it is not possible to recommend 1 PROM over the other for inclusion in (core outcome measure sets)” [35].

Although there are several PROMs used for patients with neck pain, there is no measure that limits its focus to functional disability. Some neck-related PROMs measure symptoms, functional disability, pain interference, and/or quality of life [48]. Surveys [40, 57] suggest that the DASH is frequently used to measure the upper extremity-related components of neck pain which are not covered by the NDI, despite the fact that the DASH was not developed for this purpose. The importance of the upper extremity in neck-related functional disability was emphasized by qualitative studies which found that this was an important component of neck symptoms and disability from the patient perspective [27]. Lab-based studies have demonstrated altered upper extremity neuromuscular functioning in people with neck pain [21], which confirms the importance of considering upper extremity functioning in neck conditions.

The lack of sufficient involvement of patients with neck pain in developing some of the early neck-specific PROMs may have contributed to important gaps in the scope of symptoms or disability included on the NDI. Content validity requires that during development and validation the relevance of items be assessed with respect to the target population [41]. When PROMs fail to address important elements or the full scope of a construct, then content validity is inadequate, regardless of whether the measure demonstrates adequate quantitative psychometric properties.

Therefore, the purpose of this paper is to report the development and validation of a PROM that is designed to measure neck-related disability in patients with neck pain/disorders. Specific objectives are to describe the development process, content validity, readability, potential for floor/ceiling effects, test–retest reliability, and construct validity.

The ND10 was developed as a 10-item functional scale for patients with disorders of the neck. The final version is presented with scoring instructions as Supplementary File 1. Characteristics of participants in the validation studies are listed in Table 2 and indicate a 75/25 female imbalance in gender distribution.

This study provides evidence that the ND10 can provide understandable, relevant, reliable, structurally sound, and discriminative scores representing neck-related disability. While there are multiple PROMs that could be used for people with neck pain, the uniqueness of the ND10 is that it was developed to solely focus on neck-related functional disability, whereas other commonly used measures combine symptoms, pain interference, and other constructs within a single scale. We used iterative quantitative and qualitative work to establish the content validity and usability of the ND10. Content validity analyses was studied using item comparison, classification in ICF and perspective, patient questionnaires and cognitive interviews; and these findings were triangulated during development. The conceptual clarity of the construct being measured, and its constituent items may be the most critical aspect of ND10 development, and is an aspect most sparsely attended to in development of many other neck-related PROMs. COSMIN (COnsensus-based Standards for the selection of health Measurement Instruments) has recently provided more detail highlighting the importance of rigor in content validity [45] and development of item content coding and cognitive interviewing methods provides enhanced methodological support [41].

Although the other PROMs used in the study performed well from a quantitative psychometric view, concerns about content validity of the other neck-related PROMs were apparent in the lack of clear construct definition since items crossed multiple domains and often focused more on pain interference than function. From a health literacy perspective, the readability and relevancy were better than the NDI, and the preference of patients were favourable. Thus, the ND10 may be preferable for clinicians or researchers who wish to distinguish construct of function and pain as recommended by core outcome recommendations [75]. It may also be easier for patients to complete—this is important given the extent to which health literacy is a problem in many clinical contexts. With the move towards identifying core sets of constructs to be measured in musculoskeletal research and practice, the importance of separating pain and disability in separate constructs has become clearer [75‐77]. Overall, the ND10’s psychometric properties were better than other neck-related PROMs in terms of establishing a clear conceptual construct and focusing in functional (dis)ability. It was better than the NDI in terms of patient relevance and health literacy and in avoiding floor effects. The ND10 was similar to the NDI and DASH in terms of its convergent association with other measures and ability to discriminate between known groups. Preliminary factor analysis, based on one sample, supported that the ND10 is unidimensional, which has been problematic in other neck-related PROMs including the NDI [33].

An outcome measure which focuses distinctly on disability can be important where it is the focus of a specific treatment or a specific discipline, e.g., rehabilitation. For example, in patients with chronic pain, treatment programs often target improved function without an expectation of substantial improvements in pain [78]. The development of the ND10 was not to diminish the importance of pain as an outcome measure. Conversely, we think that a brief functional neck-specific measure, like the ND10, allows space in patient contact time for a more thorough multi-dimensional pain assessment using a valid pain-specific outcome measure.

Our findings suggest that some of the limitations in previous measures that we hoped to address were successfully mitigated in our new outcome measure. Our prior work indicated the importance of the upper extremity [27, 58] in neck disorders, concerns about high rates of missingness items due to relevancy issues for some items [79], and the importance of considering health literacy during development. Our qualitative interviews indicated the most consistent concern with the NDI was a lack of clarity in the response options. The previous neck PROMs compared at a content level in Table 3 have response options that are longer, have a great cognitive burden, and are sometimes double-barreled or not mutually exclusive. These issues were commonly noted by patients as reasons that it was difficult to calibrate their responses to the NDI in our cognitive interviews. We designed the ND10 to be very simple and brief (118 words on ND10 versus 783 on ND1). Health literacy and cognitive burden are partially related to the number and complexity of words, but also to the format in which information is presented. Therefore, our use of a consistent response options and icons to represent direction were used to improve health literacy. A few of the respondents noted that the ND10 response options being brief meant that they were more open to interpretation. This is inevitable given the choices made for a streamlined format.

Some ND10 items reflect important issues raised by patients in qualitative interviews and surveys [27, 58] that were not included on the NDI, e.g., lifting and carrying a heavy object, putting something on a high shelf, and overhead work. These items create different types of strain on the neck and represent common tasks of daily life. Other issues we encountered during development indicate that items may have a “shelf-life”. For example, the NDI which was developed more than 2 decades ago asks patients about difficulty reading a book. However, many people now primarily read electronic devices. Although the way people read has changed, the ability to read and communicate with text remains an important human function. Therefore, our rewording of a reading item was designed to be more inclusive of different ways that this function is performed. One of the problematic items due to high rates of missingness on the NDI item is the driving item [33, 56]. Driving tends to leave out specific segments of the population, e.g., in some countries women are not allowed to drive; lower income people may not be able to afford vehicles; age restrictions may limit who can drive; and people with comorbidities may have medical reasons for not being allowed to drive. Thus, the driving item inherently represents a form of selection bias. However, the ability to move around in society is an important human function, and many forms of transportation can be difficult for patients with neck disorders. Therefore, this item was included in a more inclusive format by using “drive or ride” and different modes of transportation as exemplars.

Patients did not indicate concerns about the burden of either the NDI or the ND10, and some felt these PROMs were too short. Patients in a qualitative study [27] and our cognitive interviews wanted PROMs to reflect the full scope of the problems they experience. Several patients commented that the measures did not tap into important impacts of their life. Some of these issues were outside of the target construct of functional disability. This indicates the importance of using multiple PROMs to reflect the different constructs important to patients, particularly when these have been defined by core sets [43]. For example, mental symptoms and social/emotional functioning are important but should be measured in separate well-validated PROMs specific to those constructs. Patients in this study may not have understood that typically we would be measuring a larger suite of PROMs within a clinical study or clinical interaction. CATWAD (Core Outcome Domain Set For Whiplash-Associated Disorders) distinguished pain, recovery, and functional disability as separate constructs [43]. Several issues raised by patients reflected recovery or other domains within quality of life. The Satisfaction and Recovery Index [59, 80] is an example of a measure designed to measure recovery following musculoskeletal trauma. Many of the issues that patients raised as missing constructs from the ND10 and NDI fell within the construct measured on The Satisfaction and Recovery Index (e.g., intimacy, life roles). Patient interviews conducted in this study confirm CATWAD findings about the importance of considering both functional disability status and perceived recovery.

We observed that some patients had unique concerns that they felt were important to communicate, but that were not represented on any of the PROMs evaluated. No outcome measure can capture all issues important to every patient. Patients wanted clinicians to understand the complexity of their neck pain. Listening to patients helped us recognize the importance of allowing space to express individual issues qualitatively when responding on an outcome measure. Therefore, we added an open text box to the ND10 where patients can communicate what they want others to know. Although this does not contribute to the score, it is potentially useful in clinical practice since one of the important consequences of implementing outcome measures should be better communication with patients.

The reliability of the ND10 of 0.87 was high, even though the re-test interval was relatively long for some participants (mean 8.5 days; range 4–25) and did not exclude people under treatment. We attribute this to the measure itself and the chronic nature of the patient’s neck disorder. A minimal detectable change of 7.5 points compares favorably with other PROMs. We speculate that test–retest reliability can be influenced by the re-test interval, the acuity of the condition, and the extent to which the construct being measured is stable and definable by patients. We anticipate that future studies that more rigorously assess whether patients have remained stable and use more consistent test–retest intervals might find an even higher reliability coefficient.

The development of a new PROM is justified when there are no PROMs for an important construct or there are serious flaws in existing PROMs. These rationales apply for the ND10 development since previous PROMs lack conceptual clarity, content validity, or failed to adequately incorporate patient perspectives. The ND10 addresses a core construct recommended by an international consensus as being important for patients with neck pain [43]. Despite all of these favourable findings, we recognize it can be difficult to transition to a new PROM. Although there are conceptual flaws with existing neck PROMs, their long-standing use—particularly with respect to the NDI—means that legacy measures have pools of comparative data and familiarity, which may make some people reluctant to change their current usage patterns.

Although this study reports the findings of a multi-stage process, there are limitations in our work. We did not provide the full suite of psychometric evidence. Important future investigations include fit to the Rasch model and responsiveness studies; as well as widespread cross-cultural translation. Although we found excellent reliability and factor structure, the sample sizes were relatively small for these analyses, and future studies in larger samples are needed for greater precision and confidence. A clear understanding of utility of any new PROMs only becomes apparent over time after it has been tested in multiple contexts and populations.

This study led to the development of a reliable and valid measurement PROM, the ND10, designed specifically for assessing neck-related functional disability. Overall, the findings are supportive of the content validity and suggest strong clinical measurement properties. The ND10 is provided by open access from the developer/copyright owner (J MacDermid: jmacderm@uwo.ca at https://​www.​lawsonresearch.​ca/​hulc/​outcome-measures) so that it is freely available for use where a simple measure of function is needed for patients with neck pain or disability. It should be used in combination with a pain scale and measures of other salient constructs to reflect multiple aspects of health outcomes and quality of life.