Improving event prediction using general practitioner clinical judgement in a digital risk stratification model: a pilot study

Table 1 Characteristics of those not identified by risk factors and therefore not escalated, and of those escalated then triaged by GP clinical judgment into No Concern v Concern. Statistical analysis of the 3 groups was by ANOVA or Chi square tests for the comparison of means or proportions, respectively

	Not escalated	Escalated, GP No Concern	Escalated, GP Concern	3 Group comparison
	n = 27,424 (87.4)	n = 3, 450 (10.9)	n = 518 (1.7)
Age (years): mean (SD)	36.0 ± 21.2	68.1 ± 17.7	78.3 ± 13.9	F = 4554.3, p < 0.001
Gender (male)	51.6	44.9	44.4 *	X²= 63.1, p < 0.001
Ethnicity (white)	53.5	70.9	69.7 *	X² = 283.1, p < 0.001
IMD: mean (SD)	30.5 ± 15.7	26.7 ± 14.7	26.9 ± 14.8 *	F = 104.8, p < 0.001
A&E attendances > = 3 in 12 months	0	7.1	8.7 *	X² = 2,041.9, p < 0.001
Non-elective admission > = 3 in 12 months	0	2.9	8.30	X² = 1,284.6, p < 0.001
PARR score > = 80%	0	2.8	11.00	X²= 31,237, p < 0.001
Comorbidities > = 3	0	77.2	84.60	X² = 23,825.1, p < 0.001
Electronic Frailty Index (moderate or severe)	0	43.1	64.50	X² = 13,737.1, p < 0.001
On End-of-Life register	0	7.8	26.40	X² = 4,069.9, p < 0.001
Nursing home resident	0	12.5	15.3 *	X² = 3,597.8, p < 0.001
Any risk flag	0	all	all	-
Total number of flags (SD)	-	1.5 ± 0.8	2.2 ± 1.1	F = 297.1, p < 0.001

Figures are column percentages unless otherwise stated. IMD Indices of Multiple Deprivation; A&E Accident and Emergency; PARR Patient’s At Risk of Re-admission
Post hoc analysis for between group differences were all p < 0.001 except where * indicated non significance for GP No Concern vs Concern

ISSN: 1472-6947