The effect of two lottery-style incentives on response rates to postal questionnaires in a prospective cohort study in preschool children at high risk of asthma: a randomized trial

Table 3 Results of the analysis

Crude effects of intervention on outcome
Intervention	Daytrip
Outcome	RD	95% CI	NNT
All questionnaires returned (100% returned)	−0.008	−0.07 – 0.06	126 (NNH*)
All questionnaires returned without reminder	0.019	−0.03 – 0.07	51
Non-response	0.0004	−0.03 – 0.03	2,359
Withdrawal	0.05	−0.01 – 0.09	19
Outcome	Median-difference intervention group-control group	p-value of ranksum-test
Number of reminders sent	0.5	0.753
Percentage of all questionnaires returned	0	0.803
Intervention	Gift voucher
Outcome	RD	95% CI	NNT
All questionnaires returned (100% returned)	0.017	−0.50 – 0.08	60
All questionnaires returned without reminder	0.015	−0.04 – 0.07	67
Non-response	−0.017	−0.05 – 0.02	60 (NNH*)
Withdrawal	0.008	−0.03 – 0.05	118
Outcome	Median-difference intervention group-control group	p-value of ranksum-test
Number of reminders sent	1	0.045
Percentage of all questionnaires returned	0	0.374
Outcome	Odds Ratio	p-value of multilevel logistic regression	95% CI
Response Rate per questionnaire	1.29	0.26	0.83 – 2.01

Table 3 presents the results of the analysis on “all questionnaires returned, “without reminders” versus “not all questionnaires returned or reminders sent”. Risk Differences (RD) with 95% confidence intervals and Number Needed Treat (NNT = reciprocal of RD) for are presented.
CI 95%-level of NNT is not given, because all outcomes are not significant. In case of a non-significant treatment effect at 5%, the CI95% for the risk difference will include zero, and thus the 95% confidence interval for the number needed to treat will include infinity (∞). For example the effect of daytrip on non-responders gives an RD of 0.0004 (CI95% -0.03 – 0.03; including zero). The NNT is 2359 (1/0.0004), if we calculate CI95% on the same way, it must include infinity, thus from 32 to ∞ and from minus 31 to minus ∞ [6].
* Number needed to harm (NNH in case of a negative risk difference).

ISSN: 1471-2288