A comparative Rasch analysis and classical test theory short-form derivation and validation of the Nottingham Extended Activities of Daily Living (NEADL) Questionnaire.
Terry Quinn, Alexander Smith, Jonathan Hewitt, Michael Robling, Myzoon Ali & Timothy Pickles

The Nottingham Extended Activities of Daily Living (NEADL) Questionnaire is used extensively in clinical practice and research [1]. However, under scrutiny the NEADL has evidenced psychometric limitations. The initial studies of the psychometric properties of the NEADL proved promising with studies in stroke patients supporting the NEADL’s validity & test-retest reliability [2, 3]. Nevertheless, studies have questioned whether the total summed scored 22 item NEADL is unidimensional [4] and have found the NEADL not to conform to Guttman scaling principles [5]. Across a number of studies, the NEADL has demonstrated high internal consistency [3, 5, 6], which is suggestive of item redundancy and the potential to reduce the number of items in the NEADL [7, 8], whilst improving its psychometric properties.
In clinical practice and in research, time constraints and patient burden are challenges to any form of assessment. Short-forms derived from longer assessment or rating scales have the potential to overcome many of the drawbacks of the original long-form tools due to the shorter administration time and reduced patient burden due to fewer questions to be answered [7, 8]. Furthermore, short-forms allow for psychometric improvements through the eradication of redundant items that are often present in assessment tools. Importantly, short-forms allow for flexibility in that the derived short-forms can be administered as short-forms or the short-form scores derived from the administration of the original long-form.
With this in mind, this analysis proposes to derive a short-form of the NEADL. The resulting short-form NEADL will reduce administration time and patient burden in both busy clinical environments and for research studies where assessment completion is of great importance.

What are the current psychometric properties of the NEADL-22 item and 4 subscale NEADL? - Using Rasch analysis - Using Classical Test Theory (CTT) Item Reduction –
Is it possible to reduce the number of items in the NEADL-22 item and 4 subscale NEADL and improve the psychometrics?
- Using Rasch analysis
- Using CTT

Comparison of Methods –
Which analysis approach to item reduction and short-form derivation provides the most valid and reliable short-form?

We will undertake two separate forms of analysis to understand the current performance of the NEADL and to derive method specific short-forms.


Rasch Analysis

The Rasch analysis will be performed using the Rasch unidimensional measurement model (RUM2030).
NEADL-22 item
Rasch analysis on the summed score of the full 22 items of the NEADL as a general measure of EADLs, with polytomous scoring. Invariance to the scale will be assessed using chi-square interaction term. Item misfit will also be assessed by fit residuals. Person fit residuals will signify misfit to the model. Residual correlation matrices will be examined for positive correlations indicating response dependency. Differential item function (DIF) will be explored for age, sex and stroke type (haemorrhage or infarct).
Short Form Derivation - In contrast to previous Rasch analysis of the NEADL, we will go further in the modification of the NEADL including item reduction to derive a short-form with better fit to the Rasch model than the current original long-form.

4 Sub-Scale NEADL
The Rasch analysis of the 4 subscale NEADL will utilise the methods as outlined previously. Where the approach involves the Rasch analysis of the 4 subscale NEADL, we will attempt to modify the NEADL including a more fundamental re-ordering of items between factors to ensure a good fit to the Rasch model for the derived short-form. In the analysis of the 4 subscale NEADL we will attempt to preserve a 4 factor short-form. Where appropriate Bonferroni correction will be applied.

(Highlighted in the previous Rash analysis was the need to revert to dichotomous scoring we will utilise dichotomous scoring if it is not possible to fit the Rasch model with polytomous scoring. The need to revert to dichotomous scoring we will be apparent early in the initial Rasch analysis and will not impact the quality or significance of the findings. The same Rasch procedure will be adopted for the analysis of dichotomous scoring as the polytomous scoring analysis. However, a dichotomous Rasch model will be utilised as opposed the polytomous Rasch model.)

Classical Test Analysis

NEADL-22 item
Internal consistency – Cronbach alpha
Content Validity – Correlations between individual items relative to total NEADL-22 score, assuming a single domain of EADL will be calculated using Spearman rank correlation.
Explanatory Factor Analysis (EFA) to investigate the number of factors loadings within the NEADL-22 to explore if the NEADL-22 has a single factor structure and to investigate the possibility of a short-form based on a single factor.
Short form derivation – To derive a short-form a correlation matrix will be used between each of the NEADL items score with the total NEADL score. This will be evaluated against the results of the EFA to preserve a single domain structure if the results of the prior EFA indicate the potential for a single factor structure.
4 Sub-Scale NEADL
Internal consistency – Cronbach alpha
Content Validity – Correlations between individual items relative to total domain score, for the item’s corresponding subscale, will be calculated using Spearman rank correlation, assuming that the NEADL has 4 domains.
EFA to investigate the number of factors loadings within the 4 subscale NEADL to confirm if the 4 subscale NEADL has a 4 factor structure and to investigate the possibility of a short-form based on the original 4 factors.
Short form derivation – To derive a short-form a correlation matrix will be used between each of the NEADL items score with the total domain score within its proposed domain. This will be evaluated against the results of the EFA to preserve a 4 domain structure.

Comparison of Short Forms
Rasch Analysis – The short-forms derived from the Classical Test analyses will be subject to the Rasch analysis procedures outlined previously.
Classical Test Analysis – The short-forms derived from the Rasch analyses will be subject to the classical test analysis procedures outlined previously.

1. Wales, K., et al., Functional Assessments Used by Occupational Therapists with Older Adults at Risk of Activity and Participation Limitations: A Systematic Review. PLOS ONE, 2016. 11(2): p. e0147980.
2. Lincoln, N.B. and J.R.F. Gladman, The Extended Activities of Daily Living scale: a further validation. Disability and Rehabilitation, 1992. 14(1): p. 41-43.
3. Nouri, F.M. and N.B. Lincoln, An extended activities of daily living scale for stroke patients. Clinical Rehabilitation, 1987. 1(4): p. 301-305.
4. das Nair, R., B.J. Moreton, and N.B. Lincoln, Rasch analysis of the Nottingham extended activities of daily living scale. J Rehabil Med, 2011. 43(10): p. 944-50.
5. Nicholl, C.R., N.B. Lincoln, and E.D. Playford, The reliability and validity of the Nottingham Extended Activities of Daily Living Scale in patients with multiple sclerosis. Mult Scler, 2002. 8(5): p. 372-6.
6. Gladman, J.R.F., N.B. Lincoln, and S.A. Adams, Use of the Extended ADL Scale with Stroke Patients. Age and Ageing, 1993. 22(6): p. 419-424.
7. MacIsaac, R., et al., Derivation and Validation of a Modified Short Form of the Stroke Impact Scale. Journal of the American Heart Association, 2016. 5(5): p. e003108.
8. MacIsaac, R.L., et al., Use of a 3-Item Short-Form Version of the Barthel Index for Use in Stroke. Stroke, 2017. 48(3): p. 618-623.

Funding has been made available to support these analyses