SMA REACH UK One Year On: The Evolution of Robust Functional

Outcome Measures for Spinal Muscular Atrophy Type 2 & 3

D. Ramsey1, M.Scoto1, A. Mayhew2, M. Main1, I. Wilson1, E. Mazzone3, J. Montes4, K. Bushby2, R. Finkel5, E. Mercuri1,3, F. Muntoni1
1

Dubowitz Neuromuscular Centre, UCL & Great Ormond Street Hospital, London, 2Institute of Human Genetics, Newcastle University,3Universita Cattolica Roma, Italy,
4
Department of Neurology, Columbia University Medical Center, New York, USA,
5
Nemours Childrens Hospital, University of Central Florida College of Medicine, Florida, USA

Background:

Pictures reproduced from HFMSE and ULM for SMA manuals demonstrating items from these scales
(HFMSE1, ULM2)

Increasing momentum in the field of Spinal Muscular Atrophy (SMA) has resulted in
potentially promising experimental therapies entering clinical trials. Robust
functional outcome measures are required to demonstrate efficacy of experimental
therapies and for their regulatory approval.

SMA REACH UK (Spinal Muscular Atrophy REsearch And Clinical Hub UK) has been
working collaboratively with the Italian SMA Network and the PNCR Network USA to
develop and refine functional outcome measures for use in SMA, to meet
requirements of regulatory authorities.
70 subjects have been recruited to SMA REACH UK (55 London GOSH , 15
Newcastle).

Figure 4: Revised Hammersmith Scale for SMA

(RHS)
Figure 2a: RHS Item Targeting (Rasch Measurement Model)

Aim:

To develop psychometrically robust functional outcome measures for use in

type 2 and 3 SMA.

Methods:
SMA REACH UK hosted 2 international workshops (2013, 2014), national meetings
and numerous teleconferences, where expert physicians and physiotherapists
meticulously discussed functional outcome measures for SMA. Utilising item
response theory, expert opinion and experience from clinical trials, the
Hammersmith Functional Motor Scale Expanded and Upper Limb Module for SMA
were used as the foundation for the development of new scales. Two revised scales
have been prospectively piloted and further refined over the last twelve months.
The process described in figure 1 has also been applied to revision of the Upper Limb
Module for SMA.

Figure 2b: RHS Item Thresholds (Rasch Measurement Model)

Figure 1: Development of the Revised Hammersmith Scale for SMA

Figure 3a: ULM/RULM Item Targeting (Rasch Measurement Model)

Results of Phase 2 Pilot:

The Revised Hammersmith Scale for SMA (RHS):
Following three iterative revisions we outline the final proforma and Rasch analysis
ofItem
the Fit
phase two pilot. DF
Person Fit
PSI
Mean -0.0743 SD 0.6357 60

Mean -0.1108 SD 0.5015

-0.97173

Figure 3b: ULM/RULM Item Targeting (Rasch Measurement Model)

Subjects: n = 70 (Female = 37, Male = 33), Type 3 SMA = 32, Type 2 SMA = 38
(Ambulant 50, Non-ambulant 20). 16 patients had had spinal surgery and
approximately 50% were taking Salbutamol. 2 invalid results, 3 extremes.
RHS Rasch Analysis:

Item Fit: excellent, all fit residuals within +2.5 (3 significant chi square values).
Person Fit: Good spread of items, with no floor or ceiling effect.
Disordered Thresholds: 11/30 discrepancies in these items were also identified
during discussion of clinical meaningfulness.
Following the application of clinical sensibility in addition to Rasch the expert panel
of physiotherapists further revised the RHS with the final definitive version currently
being piloted (phase 3), see figure 4. Initial testing using the Rasch Measurement
Model identified fit was excellent with no miss fitting items, however further testing
of this scale in a larger cohort is required.
The RHS currently is a scale suitable for the assessment of type 2 and 3 SMA and
Item Fitof 36 Items with DF
Fit69 (17.03.2015).
PSI
consists
a total availablePerson
score of
Mean 0.0119 SD 0.8683

52

Mean -0.1954 SD 0.6221

0.96222

The Revised Upper Limb Module (RULM):

Conclusion:

Subjects: n = 42 (Female 22, Male 20), Type 3 SMA = 18, Type 2 SMA = 24
(Ambulant 9, Non-ambulant 33). 11 had spinal surgery, and 30 were taking
Salbutamol. 1 invalid result, 6 extremes.

Initial findings of the RHS and RULM indicate that they will make promising alternatives to the
original scales whilst also meeting the rigorous psychometric requirements of regulatory
authorities.

RULM/ULM Rasch Analysis:

The RHS and RULM are undergoing further responsiveness and psychometric testing in a wider
international cohort to endorse its content and the cohesiveness of items. In addition inter-rater
and intra-rater reliability testing of the RHS is to be conducted in UK in April 2015. Face validity of
the RHS will be discussed with patients and parents/carers at a focus group being held by SMA
REACH UK in March 2015.
Correspondence: danielle.ramsey.12@ucl.ac.uk www.SMAREACHUK.com
For further detail on the complex psychometric analysis please contact Dr Anna Mayhew
anna.mayhew@newcastle.ac.uk

Disordered Thresholds: 6/17

ULM Psychometric analysis has shown good fit of all items and suitable targeting;
however a ceiling effect still exists for the stronger type 3 patients. Following the
application of clinical sensibility in addition to the Rasch analysis the scoring of the
RULM was simplified and harder items introduced to reduce the ceiling effect. The
new scoring demonstrates excellent fit when preliminarily assessed with the Rasch
Measurement Model however this requires further testing in the phase 3 pilot.

8th Annual UK Neuromuscular Translational Research Conference 19 th - 20th March 2015, Centre for Life, Newcastle