Inter-rater reliability of a paediatric outcome measure in Nepal.

ABSTRACT

PIP: The Hospital and Rehabilitation Center for Disabled Children (HRDC) in Nepal identified the need to evaluate their in-hospital and community programs. An instrument was developed to provide information regarding care-givers' attitudes towards disability. Six HRDC field workers traveled in pairs to the childrens' homes and alternated in roles as test administrator and observer. Inter-rater reliability of the measure was tested in a sample of 49 children 3-19 years old (55% male and 45% female) who lived in the Kathmandu Valley, the Terai, or hilly regions. The most common disability types treated at the HRDC were congenital talipes equinovarus or club foot (31.5%), poliomyelitis (15.1%), burn contractures (14.2%), tuberculosis (11.1%), and osteomyelitis (5.6%). Inter-rater reliability was estimated for instrument domains of functional status (self-care, mobility, and social function), use of adaptive equipment, and knowledge and attitudes towards disability. The total scores for each domain were compared between raters using the weighted kappa statistic (Kw). Mean reliability coefficients varied from Kw = 0.60 to Kw = 1.0. All mean values demonstrated substantial to almost perfect correlations, except for the calculated value for self care equipment (Kw = 0.60). However, a Fisher Z transformation revealed no significant difference between the calculated value of 0.60 and the acceptable value of 0.61. Individual reliability coefficients for pairs of raters were generally acceptable, with the exception of one pair (E + F) on the self-care equipment domain. The mean weighted kappa value for younger children was 0.70 +or- 0.02 for the mobility level of assistance scale and 0.72 +or- 0.06 for the mobility functional activities scale. On application of the Fisher Z transformation formula, significant differences were found for this subgroup of younger children (or= 6 years) as compared with the sample population. The instrument demonstrated an acceptable level of inter-rater reliability.