Article Text
Abstract
Background People with ILD, currently have less access to SPC and there is no validated needs assessment tool (NAT). We adapted the NAT:PD-cancer for use in ILD and conducted psychometric testing.
Aim To test the construct validity of NAT:PD-ILD.
Methods ILD clinicians in four hospitals were trained to use the NAT:PD-ILD. After a consultation, the clinician completed the NAT:PD-ILD, patients completed the St. George’s Respiratory Questionnaire (SGRQ-I) and carers completed the Carer Strain Index (CSI) and Carer Support Needs Assessment Tool (CSNAT).
Kendall’s Tau-b correlation coefficient (and associated p-value) was calculated to determine the correlation between the NAT: PD-ILD items relating to patient wellbeing, and a total score for a subset of SGRQ-I questions identified a priori as measuring similar constructs. The prevalence and bias adjusted kappa (PABAK), Cohen’s kappa and percentage of agreement were used to assess whether responses were similar between the NAT: PD-ILD items relating to the ability and wellbeing of the carer and appropriate CSI and CSNAT items which were considered to measure similar concerns/support needs.
Results A total of 68 patients were recruited. The average age of participating patients was 66 years (range 34 to 87) and 62% were male. Forty-five (66%) patients had a carer of whom 27 completed the CSI (mean 4.4, SD 3.0, median 4, range 0–11) and 29 completed at least one item of the CSNAT.
Items 2, 3, 5 and 6 of the NAT: PD-ILD statistically significantly positively correlated with their comparator SGRQ-I scores (ρ range 0.24 to 0.36, p < 0.05). PABAK values comparing the NAT: PD-ILD items with appropriate CSI and CSNAT items show most items have PABAK positive values (range from 0.04 to 0.57, with a minimum of 52% agreement). However, NAT:PD-ILD items 11 and 13 have negative PABAK values (Inter-personal relationships and Grief topics – Psychosocial Dimension).
Conclusion The NAT: PD-ILD has adequate construct validity for most domains. However, agreement is poor for physical symptoms and spiritual concerns. This may indicate that clinicians identify concerns with symptoms less well unless they are severe.