Australian Injury Comorbidity Indices (AICIs) to predict burden and readmission among hospital-admitted injury patients

Fernando, Dasamal Tharanga; Berecki-Gisolf, Janneke; Newstead, Stuart; Ansari, Zahid

doi:10.1186/s12913-021-06149-1

Table 3 Performance of selected comorbidity measures in assessing the association between comorbidity and selected outcome measures (Victoria)

From: Australian Injury Comorbidity Indices (AICIs) to predict burden and readmission among hospital-admitted injury patients

Model	LOS^a		LOS^b		Hospital costs^c
Model	AUC (95% CI)	Model fit AIC	McFadden’s Adjusted R²	Model fit AIC	Adjusted R²	Model fit AIC
(i) Baseline model	0.748 (0.745–0.750)	145,468	0.080	579,331	0.312	304,799
(ii) Baseline model + selected comorbidities (individually modelled with binary representation)	0.760 (0.757–0.763)	142,523	0.088	572,274	0.359	297,352
(iii) Baseline model + selected comorbidities (modelled as a weighted summed score)³	0.760 (0.757–0.762)	142,582	0.088	572,673	0.357	297,674
(iv) Baseline model + comorbidity using CCI weights	0.752 (0.749–0.754)	144,554	0.083	577,080	0.328	302,263
(v) Baseline model + comorbidity using Quan weights	0.752 (0.749–0.754)	144,527	0.082	577,610	0.325	302,663
(vi) Baseline model + ECM	0.761 (0.759–0.764)	142,174	0.089	571,537	0.366	296,207
(vii) Baseline model + 23 comorbidities common to LOS and cost outcomes (binary representation)/Baseline model + 8 comorbidities common to readmission outcomes (binary representation)	0.760 (0.757–0.762)	142,603	0.088	448,263	0.358	297,494
Model	All-cause 30-day readmission^d		Non-planned 30-day readmission^e
Model	AUC (95% CI)	Model fit AIC	AUC (95% CI)		Model fit AIC
(i) Baseline model	0.611 (0.606–0.615)	99,275	0.623 (0.618–0.628)		75,455
(ii) Baseline model + selected comorbidities (individually modelled with binary representation)	0.625 (0.621–0.630)	98,445	0.637 (0.632–0.642)		74,950
(iii) Baseline model + selected comorbidities (modelled as a weighted summed score)^f	0.625 (0.620–0.629)	98,473	0.636 (0.631–0.642)		74,946
(iv) Baseline model + comorbidity using CCI weights	0.622 (0.617–0.626)	98,662	0.633 (0.628–0.638)		75,096
(v) Baseline model + comorbidity using Quan weights	0.620 (0.615–0.624)	98,824	0.631 (0.626–0.637)		75,174
(vi) Baseline model + ECM	0.626 (0.622–0.631)	98,457	0.639 (0.634–0.644)		74,912
(vii) Baseline model + 23 comorbidities common to LOS and cost outcomes (binary representation)/Baseline model + 8 comorbidities common to readmission outcomes (binary representation)	0.622 (0.617–0.626)	98,707	0.635 (0.630–0.640)		75,010

Notes
^a Baseline model includes age, sex, injury severity, injury type, body region, geographic region, SEIFA deciles and country of birth; outcome = overnight stay discharge (logistic model)
^b Baseline model includes age, sex, injury severity, injury type, body region, SEIFA deciles and country of birth; outcome = LOS of non-same day discharges with 1–30 days stay (negative binomial model)
^c Baseline model includes age, sex, injury severity, injury type, body region, SEIFA deciles, geographic region (metropolitan and regional) and country of birth; outcome = hospital costs (Ln transformed linear model)
^d Baseline model includes age, sex, injury type, body region, geographic region and country of birth; outcome = all cause 30-day readmission (logistic model)
^e Baseline model includes age, sex, injury severity, injury type, body region, geographic region and SEIFA deciles; outcome = non-planned 30 day readmission (logistic model)
^f See Methods for weight calculation, excludes weights resulting from an OR < 1.2
See Table 4 for selected comorbidities for each outcome

Back to article page

ISSN: 1472-6963

Contact us

General enquiries: journalsubmissions@springernature.com

BMC Health Services Research

Contact us