Assessing national and subnational inequalities in medical care utilization and financial risk protection in Rwanda

Table 2 Significance testing of the trends of the absolute inequality in medical care utilization from 2005 to2016 using Cumming and Finch’s “rule of thumb”

	2005 (N = 6737)				2010 (N = 11,944)				2014 (N = 16,807)				2016 (N = 21,150)				Proportion overlap
	Mean	LB	HB	ME	Mean	LB	HB	ME	Mean	LB	HB	ME	Mean	LB	HB	ME	2005 vs 2010	2010 vs 2014	2014 vs 2016	2005 vs 2016
Poverty	0.084	0.059	0.109	0.025	0.090	0.070	0.110	0.020	0.078	0.062	0.094	0.016	0.068	0.053	0.082	0.015	1.735	1.350	1.301	1.166
Gender	0.020	−0.003	0.043	0.023	−0.004	−0.023	0.016	0.019	− 0.003	− 0.018	0.012	0.015	− 0.011	− 0.025	0.004	0.014	0.874	1.977	1.489	0.339
Education	0.027	0.002	0.053	0.026	0.024	0.001	0.048	0.023	0.045	0.028	0.062	0.017	0.024	0.009	0.040	0.016	1.888	0.968	0.716	1.848
Residence	0.016	−0.014	0.046	0.030	−0.011	− 0.037	0.014	0.026	0.023	−0.001	0.046	0.023	−0.013	−0.038	0.013	0.026	1.008	0.620	0.571	0.963

Note: N is the number of observations, LB is the lower bound of the 95% CIs, HB is the higher bound of the 95% CIs, and ME refers to the margins of error. Margins of error is the distance of either the lower or the higher bound 95% CI from the mean. The proportion overlap is defined as the intervals overlap between the two independent samples, expressed as a proportion of the average margin of error. For example, in the adjusted inequality by poverty from 2005 to 2016 above, 1.166 = (0.082–0.059)/[(0.025 + 0.015)/2]. According to Cumming and Finch [45], when both sample sizes are at least 10, and the margins of error do not differ by more than a factor of two, a proportion overlap less than 0.5 indicates a significant statistical relationship at the 0.05 level (p < 0.05). Therefore, the absolute differences in the two years are not statistically significant

ISSN: 1475-9276