Validity and reliability of the questionnaire
Face validity was determined by health educators. A pilot
test of the questionnaire was conducted with 10 NGT pri-
mary caregivers. The difficulty and appropriateness of
wording in the questionnaire were evaluated and modified.
Content validity was reviewed by specialists in neurology,
gastroenterology and nursing. The reviewers provided con-
structive comments and comprehensive assessment of the
appropriateness, necessity and clarity of wording. The
reviewers scored each questionnaire item on a scale from
1–5: 1 – very inapplicable, should be deleted; 2 – inapplica-
ble; 3 – requires modification before use; 4 – applicable; 5
– very applicable, must be included. The individual content
validity index of each question reached 80%, indicating sat-
isfactory validity. Cronbach’s a of 0967 for the knowledge
scale and 0926 for the skill scale indicated satisfactory reli-
ability of the questionnaire.
Incidence of complications
The complication rate is considered to be a significant indi-
cator of patient outcome (Johnson & Sandford 2005) and
was used to evaluate the effectiveness of the systematic
nursing intervention. Three months after subjects were dis-
charged from the hospital, researchers retrieved records of
the treatment received by subjects within the first three
months posthospital discharge to identify and evaluate the
incidence of complications. Twenty-three patients did not
complete three-month follow-up data due to death (n = 2)
or imminent death (n = 21).
Statistical analysis
Caregiver age, which was normally distributed, was pre-
sented as mean standard deviation. All other continuous
data were presented as median and interquartile range due
to non-normal distributions. Categorical variables were pre-
sented as count and percentage. Differences between experi-
mental and control groups were tested with the
nonparametric Mann–Whitney U test and the independent
two samples t-test for continuous variables, and the Fisher’s
exact test for the categorical variables. The Wilcoxon
signed-rank test was performed to evaluate changes
between pretest and post-test knowledge and skill scores.
Simple and multiple linear regression models were per-
formed to evaluate the impact of independent factors on
knowledge and skill scores. Variables which showed statis-
tical significance in the simple linear regression model were
stepwise selected into the multiple linear regression model
by both forward selection and backward elimination. p-val-
ues < 005 were considered statistically significant. Analyses
were performed using SPSS version 15.0 statistics software
(SPSS for Windows, Version 15.0. Chicago, SPSS Inc.).