• No results found

Literature review

Numbers 1 to 6 represent the 6-team members ' identities The notation "X” across this table shows the involvement o f each team member across the major activities which were implemented during the

V. This study should be taken as a baseline to test the sensitivity/specificity in the

3.10 Summary of “Part A” methods

4.1.7 Performance of “Part 2” “Section C” of the “Questionnaire” screen

“Section C” of the “Questionnaire” screen has 10 questions and two observation tests for each child for ages 60-72 months (n=272). These data obtained from 272 children were analysed to determine the specific questions in this section of the screen and the results are presented in Table 4.19.

Table 4.19: Part 2, ‘Section C ’ (Specific age group o f children aged 60-72 months) o f the “Questionnaire” screen compared with the pure-tone screen results (n=272)

“Part 2” “ “Section C”

questions

Pure-tone screen P-value

Sensitivity % Specificity % Predictive value o f positive % Predictive value of negative % Overall % Q36 2 99 13 96 95 0.77 Q37 15 87 3 96 84 0.35 Q38 99 9 5 98 13 0.99 Q39 40 63 4 96 62 0.03 Q40 85 11 4 96 14 0.78 Q41 30 80 4 96 77 0.41 Q42 99 5 4 98 9 0.94 Q43 32 77 5 96 75 0.87 Q44 15 92 5 96 88 0.66 Q45 1 97 3 96 93 0.99 OBI 1 99 13 96 95 0.93 0B 2 1 99 6 96 94 0.76 Average 35 68 6 96 67 0.71 Notes:

Q=Question, O B =Observation test, --- this broken line separates the questions (Qs) and the

observation tests (0 3 ).

The summary of Table 4.19 of the performance of “Section C” of ‘Part 2’ (specific age group of the child) of the “Questionnaire” screen with 10-questions had a sensitivity of 35% has a specificity of 68%; the predictive value for positive was 6%; the predictive value for negative was 96%, and the overall performance of this section in classifying children with/out hearing loss correctly was 67%.

The results of the performance of each question in “Section C” of Part 2 of this screen compared with the pure-tone screen (n=272) show the following outcomes:

a) Sensitivity

• Three questions (Qs: 38, 40 and 42) have high sensitivity of >70%

• Three questions (Qs: 39, 41 and 43) have marginal moderate sensitivity between 30 and 49%.

• Four questions (Qs: 36, 37, 44 and 45) have very low sensitivity of <30%.

b) Specificity

• But six questions (Qs: 36, 37, 41, 43, 44 and 45) have high specificity. • And only three questions (Qs: 38, 40, 42) have very low specificity. • The two observation tests (OBI and OB2) have high specificity

Ten questions and two observations of “Section C” (Qs: 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, OBI and 0B2) of the “Questionnaire” screen identified 67% children with/out hearing loss in a sample group of 233 children (n=233).

As mentioned already, there are two observation tests for each child performed by the interviewer. The interviewer on each section of the “Questionnaire” screen records these observation tests. S/he is required to summarise her/his observations on the scale provided on the top copy of this tool. The scale is graded from 0 to 10 (0 being no response to sound and 10 meaning the child always responded).

The results of the scale of the “Questionnaire” screen are presented in Table 4.20 below.

Table 4.20: “Questionnaire” screen scale compared with the pure-tone screen (n=747)

Questionnaire Screen Pure-tone screen

Observational scale Sensitivity % Specificity % Predictive value of positive % Predictive value o f negative % Overall % P-value >4 50 91 39 100 93 0.60 <4 3 99 0 99 99 0.50 Average 27 95 20 99 96 0.55

The results of the performance of the observation scale of the “Questionnaire” screen presented in Table 4.20 above are summarised as follows:

• Scale >4 are: sensitivity/specificity = 50% and 91% respectively; predictive values for positive/negative = 39% and 100% respectively; overall: 93%

• Scale <4 are: sensitivity/specificity = 3% and 99% respectively; predictive values for positive/negative = 0% and 99% respectively; overall: 96%

The “Questionnaire” screen’s performance is evaluated as follows: the sensitivity and specificity were 79% and 96% respectively; the predictive values for positive and negative were 75% and 96% respectively; and the overall performance of this screen was 79.8%. The results of the performance on the questions of the “Questionnaire” screen in identifying deaf children when compared with the pure-tone screen were analysed and have been described in the previous subsections of this chapter.

The next section classifies the performance of the individual questions of the “Questionnaire” screen for “Part 1” and “Part 2” “Sections: A, B and C” for the purpose of eliminating questions with low sensitivity and specificity of the recommended questions (See Appendix XVIII).

4.1.8 Classifîcation of the performance of the questions of the “Questionnaire’’ screen for “Part 1” and “Part 2” “Sections: A, B and C ”

This section attempts to classify the performance of the questions of the “Questionnaire” screen and group them according to their levels of sensitivity and specificity in identifying deaf children with bilateral permanent hearing loss in excess of 50dBHL compared with the pure-tone screen.

The performance on the questions of the “Questionnaire” screen for “Part 1 ” and “Part 2” (Sections: A, B and C) was evaluated as follows:

• Questions with high sensitivity and specificity

• Questions with high and medium sensitivity or high and medium specificity • Questions with high and low sensitivity or high and low specificity

• Questions with medium sensitivity and specificity

• Questions with medium and low sensitivity or medium and low specificity • Questions with low sensitivity and specificity

The classification of the questions contained in the “Questionnaire” screen was performed to separate the questions with high, moderate and low performance.

This analysis assists to refine the “Questionnaire” screen further to optimise its performance to identify bilateral permanent hearing loss in children. The data obtained from the first “Questionnaire” and pure-tone screens were analysed and classified according to the categories spelt out above. The classifications are presented in Tables 4.21-4.25 of subsequent subsections.

The individual questions contained in Part 1” (for every child) of the “Questionnaire” screen were classified according to the categories mentioned above on how each question performed in i^ntifying deaf children in the sample group. ThdSe results are presented in the next subsection.

4.1.9 Classification of the performance of “Part 1” of the “Questionnaire” screen