Engagement, not personal characteristics, was associated with the seriousness of regulatory adjudication decisions about physicians: a cross-sectional study

Table 3 Results of the three simultaneous comparisons (grouped ORs and p values) in the final partial proportional odds model (model 12 in Table 2)

			All other versus no impairment		Suspension or erasure versus		Erasure versus all other
					no impairment or impairment
Variable	Base	Categories	OR [95% CI]	p	OR [95% CI]	p	OR [95% CI]	p
Age			1.01 [0.99 to 1.02]	0.29	-	-	-	-
Allegation	Other	Health	2.11 [0.95 to 4.72]	0.07	1.18 [0.61 to 2.30]	0.63	0.57 [0.29 to 1.14]	0.11
		Performance	0.58 [0.35 to 0.98]	0.04 ^†	-	-	-	-
		Misconduct	0.89 [0.50 to 1.55]	0.67	1.20 [0.70 to 2.06]	0.51	1.02 [0.57 to 1.80]	0.96
		Conviction	1.17 [0.59 to 2.29]	0.66	1.70 [0.90 to 3.23]	0.10	2.06 [1.10 to 3.88]	0.03 ^†
Area of practice	Neither (not in training)	Neither (in training)	0.79 [0.49 to 1.28]	0.34	-	-	-	-
		GP	0.87 [0.64 to 1.19]	0.39	-	-	-	-
		Specialist	0.71 [0.48 to 1.06]	0.09	0.51 [0.35 to 0.74]	<0.001^†	0.85 [0.58 to 1.27]	0.43
Attended	Yes	No	5.28 [3.89 to 7.18]	<0.001^†	-	-	-	-
Legally represented	Unknown	No	1.87 [1.34 to 2.60]	<0.001^†	-	-	-	-
		Yes	0.85 [0.64 to 1.14]	0.28	-	-	-	-
PMQ region	UK	EEA	1.25 [0.86 to 1.81]	0.23	-	-	-	-
		ROW	1.33 [0.99 to 1.79]	0.06	-	-	-	-
Race	Not recorded	BME	1.01 [0.74 to 1.37]	0.96	-	-	-	-
		White	1.11 [0.80 to 1.55]	0.53	-	-	-	-
Sex	Female	Male	1.20 [0.86 to 1.68]	0.29	-	-	-	-
Source	Other	Public	0.46 [0.28 to 0.74]	<0.01 ^†	0.46 [0.28 to 0.73]	<0.01 ^†	0.82 [0.48 to 1.40]	0.46
		Employer	0.68 [0.49 to 0.96]	0.03 ^†	-	-	-	-
		Another physician	0.60 [0.38 to 0.95]	0.03 ^†	-	-	-	-
		Self-referral	1.60 [0.74 to 3.46]	0.23	1.25 [0.65 to 2.44]	0.50	0.38 [0.19 to 0.77]	<0.01 ^†
		Police	0.91 [0.50 to 1.65]	0.76	-	-	-	-
		Regulator	0.78 [0.44 to 1.37]	0.38	-	-	-	-
Constant			2.09 [0.89 to 4.94]	0.09	1.07 [0.46 to 2.49]	0.88	0.15 [0.06 to 0.35]	<0.001 ^†

Comparisons gave ORs for physicians receiving outcomes more serious than ‘no impairment’ rather than a ‘no impairment’ outcome; receiving ‘suspension’ or ‘erasure’ rather than ‘no impairment’ or ‘impairment’; or receiving erasure, instead of milder outcomes
^†Significant at α=0.05. In this model, it was unnecessary to automatically relax the parallel-line assumption for most features. For clarity, ORs and p values for this majority are shown only at the leftmost comparison. They are equal for the rightmost two (dashes filled). See Table 1 for the data’s descriptive statistics

ISSN: 1741-7015