Skip to main content

Table 3 Evidence for adding functioning information into casemix systems

From: Capturing patients’ needs in casemix: a systematic literature review on the value of adding functioning information in reimbursement systems

Author and year Study characteristics a Setting & sample size b Type of casemix Model(s) Key results c
Costs
 Covinsky et al. (1997) a1, b3, c1, d3 General medical service at a teaching hospital n = 823 DRG hospitalization costs: hospitalization costs (measured in units):
Model 1*: Dependent in 0 ADL Model 1: 100
Model 2: Dependent in 1–3 ADLs Model 2: 112 (99–126)
Model 3: Dependent in 4–5 ADLs Model 3: 142 (125–162)
Model 4: Dependent in 6 ADLs Model 4: 150 (131–172)
* all models controlled for Acute Physiology Score, Charlson score, age, race, gender, admission from nursing home and diagnosis related group cost weight  
 Evers et al. (2002) a2, b2, c1, d2 Hospital n = 731 DRG Explained variance in costs directly related to medical care: Explained variance costs directly related to medical care (R2):
Total costs: Total costs:
Model 1: DRGs Model 1: 0.338
Model 2: DRGs + Need factors (includes functioning information among others) Model 2: 0.547
Model 3: DRGs + Need factors + Enabling factors + Predisposing factors + First order interactions (includes interactions between functioning and gender) Model 3: 0.611
Diagnostic costs: Diagnostic costs:
Model 1: DRGs Model 1: 0.168
Model 2: DRGs + Need factors (includes functioning information among others) Model 2: 0.362
Model 3: DRGs + Need factors + Enabling factors + Predisposing factors + First order interaction (includes interactions between functioning and gender) Model 3: 0.407
Therapeutic costs: Therapeutic costs:
Model 1: DRGs Model 1: 0.377
Model 2: DRGs + Need factors (includes functioning information among others) Model 2: 0.483
Model 3: DRGs + Need factors + Enabling factors + Predisposing factors + First order interactions (includes interactions between functioning and gender) Model 3: 0.533
 Chuang et al. (2003) a2, b3, c2, d3 General medical service at a teaching hospital n = 1612 DRG Hospital costs: Hospital costs (in $):
All patients: All patients:
Model 1: Independent in ADL on admission Model 1: $4,060
Model 2: Dependent in ADL on admission Model 2: $5,300
DRG weight <0.9: DRG weight <0.9:
Model 1: Independent in ADL on admission Model 1: $3,090
Model 2: Dependent in ADL in admission Model 2: $4,130
DRG weight 0.9-1.0: DRG weight 0.9-1.0:
Model 1: Independent in ADL on admission Model 1: $3,560
Model 2: Dependent in ADL on admission Model 2: $4,440
DRG weight 1.0-1.2: DRG weight 1.0–1.2:
Model 1: Independent in ADL on admission Model 1: $3,940
Model 2: Dependent in ADL on admission Model 2: $4,840
DRG weight >1.2: DRG weight >1.2:
Model 1: Independent in ADL on admission Model 1: $6,560
Model 2: Dependent in ADL on admission Model 2: $8,250
All patients adjusted for DRG weight: All patients adjusted for DRG weight:
Model 1: Independent in ADL on admission Model 1: $4,140
Model 2: Dependent in ADL on admission Model 2: $5,240
All patients adjusted for age, race, sex, Charlson Comorbidity score, APACHE II score, admission from nursing home and DRG weight: All patients adjusted for age, race, sex, Charlson Comorbidity score, APACHE II score, admission from nursing home and DRG weight:
Model 1: Independent in ADL on admission Model 1: $4,220
Model 2: Dependent in ADL on admission Model 2: $5,200
 Pietz et al. (2004) a2, b3, c3, d1 VA medical centers primary care patients n = 35337 ACG-based ADGs Model’s ability to predict costs for FY 1998 and FY 1999: Models ability to predict costs for FY 1998 and FY 1999 (R2):
Cost 1998: Cost 1998:
Model 1: ACGs Model 1: 0.277
Model 2: age, gender, ADGs, PCS, MCS, Model 2: 0.294
Model 3: age, gender, ADGs, 8 items Model 3: 0.298
Cost 1999: Cost 1999:
Model 1: ACGs Model 1: 0.070
Model 2: age, gender, ADGs, PCS, MCS, Model 2: 0.085
Model 3: age, gender, ADGs, 8 items Model 3: 0.087
MAPE for costs 1999: MAPE for 10th decile for costs 1999:
Model 1: age, gender, ADGs Model 1: $23440
Model 2: age, gender, ADGs, 8 items Model 2: $23204
Length of stay
 Dunstan et al. (1996) a1, b2, c1, d3 Geriatric Medicine Service n = 400 development of new system (ACME) Explained variance for Length of Stay: Explained variance for Length of Stay (%):
Model: Model:
Model 1: CMIX* Model 1: 19.5 %
Model 2: Presenting Illness (PI) + Functional Status (FX) Model 2: 19.2 %
Model 3: PI Model 3: 13.0 %
Model 4: FX Model 4: 14.1 %
Model + center: Model + center:
Model 1: CMIX* Model 1: 25.2 %
Model 2: PI + FX Model 2: 25.0 %
Model 3: PI Model 3: 19.6 %
Model 4: FX Model 4: 19.3 %
Model + center + age + sex: Model + center + age + sex:
Model 1: CMIX* Model 1: 25.2 %
Model 2: PI + FX Model 2: 25.0 %
Model 3: PI Model 3: 20.1 %
Model 4: FX Model 4: 19.4 %
*CMIX is a three-level score calculated by simple addition of the 0 and 1 scores of PI and FX.  
 Sahadevan et al. (2004) a2, b1, c1, d2 Acute care hospital Department of Geriatric Medicine & General Medicine Department
n = 232
DRG Variance explained in actual Length of Stay: Variance explained in actual Length of Stay (adjusted R2):
  Analysis with outliers: Analysis with outliers:
Length of stay (all subjects): Length of stay (all subjects):
Model 1: DRG’s trimmed ALOS Model 1: 8 %
Model 2: Functional status at discharge, total number of referrals to therapists, trimmed ALOS Model 2: 28 %
Interdepartmental differences in Length of stay (subjects with common DRG): Interdepartmental differences in Length of stay (subjects with common DRG):
Model 1: Department factor + DRG’s trimmed ALOS Model 1: 23 %
Model 2: Functional profile at discharge, total number of referrals to therapists, trimmed ALOS, department factor Model 2: 31.4 %
Analysis without outliers: Analysis without outliers:
Length of stay (all subjects): Length of stay (all subjects):
Model 1: DRG’s trimmed ALOS Model 1: 23.8 %
Model 2: Overall functional profile at admission, total number of therapy referrals, trimmed ALOS Model 2: 31.4 %
Interdepartmental differences in Length of stay (subjects with common DRG): Interdepartmental differences in Length of stay (subjects with common DRG):
Model 1: Department factor, DRG’s trimmed ALOS Model 1: 28.1 %
Model 2: Overall functional profile at admission, trimmed ALOS, referrals to medical social worker, department factor Model 2: 34.5 %
 Carpenter et al. (2007) a2, b2, c1, d2 Hospital n = 1685 HRG (equivalent to DRG) Difference in actual Length of Stay & predicted Length of Stay: Difference in actual Length of Stay & predicted Length of Stay (Ratio & 95 % CI)
All patients: All patients:
Model 1: low and medium ADL score Model 1: 1
Model 2: high ADL score Model 2: 1.40 (1.26–1.56)
Stroke: Stroke:
Model 1: low and medium ADL score Model 1: 1
Model 2: high ADL score Model 2: 1.67 (1.23–2.26)
Acute respiratory infection: Acute respiratory infection:
a) a)
Model 1: medium ADL score Model 1: 1
Model 2: high ADL score Model 2: 1.44 (1.16–1.80)
b) b)
Model 1: low ADL score Model 1: 1
Model 2: medium ADL score Model 2: 1.37 (1.01–1.85)
Chronic obstructive pulmonary disease: Chronic obstructive pulmonary disease:
Model 1: low and medium ADL score Model 1: 1
Model 2: high ADL score Model 2: 1.21 (1.04–1.53)
Falls: Falls:
Model 1: low and medium ADL score Model 1: 1
Model 2: high ADL score Model 2: 1.68 (1.23–2.28)
* all models controlled for healthcare resource group length of stay, hospital, discharge destination, admission source and age
 Herwig et al. (2009) a2, b2, c2, d1 University hospital, Psychiatry
n = 613
development of new system based on AMDP Predicted variation in Length of Stay: Predicted variation Length of Stay (%):
Model 1: AMDP Syndromes (Psychopathological Syndromes)* Model 1: 5,9 %
Model 2: AMDP Syndromes + Age at admission + Global assessment of functioning + clinical global impressions + voluntary admission + own apartment** Model 2: 19,8 %
*n = 998  
**n = 613
 Warner et al. (2004) a2, b3, c2, d1 Inpatient & Outpatient Veterans
n = 5888
ACG & DCG Predicting inpatient, outpatient and total days of care: Predicting inpatient, outpatient and total days of care (R2):
DCG: DCG*:
Model 1: Age/sex + HCCs Model 1: Inpatient days of care (IP): 0.36; Outpatient days of care (OP): 0.33; Both: 0.30
Model 2: Functionally enhanced* Model 2: IP: 0.36; OP: 0.33; Both: 0.30
ACG: ACG*:
Model 1: Age/sex + ADGs Model 1: IP: 0.15; OP: 0.28; Both: 0.20
Model 2: Functionally enhanced* Model 2: IP: 0.19; OP: 0.28; Both: 0.22
* Functionally enhanced: ACG/DCG + age, gender + self-reported functional measure * n = 2347 for inpatient days of care and n = 5888 for outpatient days of care
Resource provision
 Phillips & Hawes (1992) a1, b3, c2, d3 Nursing care units n = 1792 RUG-II Explained variation in resource provision by time: Explained variation in resource provision by time (R2):
Licensed time: Licensed time:
Model 1: RUG-II Model 1: 0.14
Model 2: RUG-II with cognitive variables Model 2: 0.16
Aide time: Aide time:
Model 1: RUG-II Model 1: 0.39
Model 2: RUG-II with cognitive variables Model 2: 0.39
Total time: Total time:
Model 1: RUG-II Model 1: 0.40
Model 2: RUG-II with cognitive variables Model 2: 0.40
  1. ACG Adjusted Clinical Groups, ACME Admission Case-Mix System for the Elderly, ADG Adjusted Diagnostic Groups, ADL activities of daily living, ALOS average length of stay, AMDP Arbeitsgemeinschaft für Methodik und Dokumentation, CI confidence interval, DCG Diagnostic Costing Groups, DRG Diagnosis Related Groups, FX functional status, HCC Hierarchical Condition Categories, HMO health maintenance organization, HRG Healthcare Resource Groups, IP inpatient days of care, MAPE Mean Absolute Predicted Error, MCS Mental Component Score, ns not specified, OP outpatients days of care, PCS Physical Component Score, PI presenting illness, RUG-II Resource Utilization Groups Version II, VA veteran affairs
  2. aSee Table 1 study characteristics
  3. bInformation in the table is presented as stated by author
  4. cPresentation of figures of key results for each model are aligned with the presentation of results by authors of the study