Table 2.

Analysis of Self-Rated Oral Health (in Rows) versus Dentist-Rated Oral Hygiene (in Columns) among Adults Aged 50 Years and Older with Teeth Who Completed a Baseline Questionnaire and Were Clinically Examined by a Dentist (n = 566): The ElderSmile Program, New York, NY, August 2006 to March 2009

Frequency		Dentist-rated oral hygiene
Row percent		Excellent	Good	Fair	Poor	Total
Column percent
Self-rated oral health	Excellent	1	6	1	1	9
		11.11%	66.67%	11.11%	11.11%
		4.76%	2.47%	0.45%	1.23%
	Good	8	105	66	22	202
		3.98%	52.24%	32.84%	10.95%
		38.10%	43.21%	29.86%	27.16%
	Fair	6	71	82	22	181
		3.31%	39.23%	45.30%	12.15%
		28.57%	29.22%	37.10%	27.16%
	Poor	6	61	72	36	175
		3.43%	34.86%	41.14%	20.57%
		28.57%	25.10%	32.58%	44.44%
	Total	21	243	221	81	566

Agreement measured by Cohen’s kappa statistic
Simple kappa statistic
Value	Standard error	95% CI	Testing H0: kappa = 0
			Standard error under H0	Z-score	One-sided P-value	Two-sided P-value
0.1086	0.0288	(0.0520, 0.1651)	0.0276	3.9384	<0.0001	<0.0001

Weighted kappa statistic
Value	Standard error	95% CI	Testing H0: kappa = 0
			Standard error under H0	Z-score	One-sided P-value	Two-sided P-value
0.1312	0.0304	(0.0716, 0.1908)	0.0294	4.4624	<0.0001	<0.0001

A positive value of kappa indicates inter-rater agreement; kappa = 1 stands for perfect agreement, kappa = 0 means lack of agreement (agreement by pure chance), and a negative value of kappa indicates negative agreement. In the present analysis, kappa (both simple and weighted) is “significantly greater than zero,” implying some agreement between self-rated oral health and dentist-rated oral hygiene, although not very high agreement. “Weighted kappa may be more suitable here because the categories are ordered.”

CI, confidence interval.