Understanding the groups of care transition strategies used by U.S. hospitals: an application of factor analytic and latent class methods

Mays, Glen; Li, Jing; Clouser, Jessica Miller; Du, Gaixin; Stromberg, Arnold; Jack, Brian; Nguyen, Huong Q.; Williams, Mark V.

doi:10.1186/s12874-021-01422-7

Table 4 Associations Between Transitional Care Combinations and Trends in 30-Day Readmissions

From: Understanding the groups of care transition strategies used by U.S. hospitals: an application of factor analytic and latent class methods

	Risk-Adjusted Change in Probability of 30-day
Transitional Care Combination	Readmission, 2010–14
	Estimate	St. Error	P value*
Model 1: Non-Exclusive Combinations Based on Factor Analysis
Factor A: Shared Decision	−0.0056	0.0006	0.0127
Factor B: Identify High Risk	− 0.0046	0.0006	0.2989
Factor C: Medication Reconciliation	−0.0053	0.0006	0.1948
Factor D: Care Plan	−0.0054	0.0006	0.0521
Factor E: Cross-Setting Information Exchange	−0.0060	0.0006	0.0040
No Factor Used (reference)	−0.0049	0.0003
Model 2: Mutually Exclusive Combinations Based on Factor Analysis (ordered by magnitude)
Factors A + B + C + E	−0.0106	0.0017	0.0001
Factors A + C + D + E	−0.0096	0.0024	0.0149
Factors B + C + D + E	−0.0089	0.0020	0.0117
Factors B + C + E	−0.0080	0.0014	0.0007
Factors A + B	−0.0071	0.0030	0.3891
Factors A + B + D	−0.0070	0.0016	0.0665
Factors A + B + D + E	−0.0070	0.0016	0.0870
Factors A + B + C + D + E	−0.0070	0.0012	0.0033
Factors A + B + C + D	−0.0066	0.0011	0.0093
Factors C + D	−0.0065	0.0011	0.0209
Factors A + C + D	−0.0065	0.0010	0.0088
Factor A alone	−0.0064	0.0011	0.0317
Factor D alone	−0.0064	0.0014	0.1272
Factors A + B + C	−0.0059	0.0016	0.3755
Factors B + E	−0.0058	0.0016	0.4562
Factors A + C	−0.0058	0.0011	0.1960
Factors B + D	−0.0056	0.0017	0.6053
Factors A + D	−0.0055	0.0012	0.3749
Factor C alone	−0.0054	0.0011	0.4658
Factors D + E	−0.0048	0.0019	0.9495
Factors B + D + E	−0.0048	0.0021	0.9340
Factors C + E	−0.0041	0.0038	0.8056
Factors B + C + D	−0.0040	0.0014	0.3056
Factors C + D + E	−0.0033	0.0023	0.3752
Factor B alone	−0.0032	0.0024	0.3514
Factors A + D + E	−0.0017	0.0018	0.1180
Factors B + C	0.0014	0.0015	0.0001
No Factors Used (reference)	−0.0049	0.0005
Model 3: Mutually Exclusive Combinations Based on Latent Class Analysis
Class A: Broad Spectrum	−0.0031	0.0013	0.0192
Class B: Intermediate with Care Plan	−0.0057	0.0011	0.9680
Class C: Intermediate with Shared Decision	−0.0100	0.0010	0.0001
Class D: Intermediate with Medication Reconciliation	−0.0025	0.0009	0.0001
Class E: Intermediate with Interdisciplinary Teams	−0.0061	0.0008	0.5966
Class F: Intermediate with Referral to Community Services	−0.0071	0.0005	0.0006
Class G: Narrow Spectrum (Reference)	−0.0058	0.0001
Model 4: Hospital random effects and time trend only
Full sample	−0.0028	0.0001	0.0010
Model Diagnostics	-2Ln(L)	AIC	BIC
Model 1	2,916,257	2,916,261	2,916,268
Model 2	2,916,684	2,916,688	2,916,695
Model 3	2,916,261	2,916,265	2,916,273
Model 4	3,049,854	3,049,858	3,049,866

Note: Estimates are from linear probability models with random hospital effects, adjusting for age, sex, race, dual-eligibility status, disability status, Elixhauser comorbidities, Medicare expenditures in past 180 days, urban/rural designation of hospital, Area Deprivation Index, Health Professions Shortage Area designation, teaching hospital designation, hospital ownership type, and monthly time trend
Mutually exclusive TC combinations are defined such that each hospital is classified into exactly one combination
*P-value indicates the significance of the difference between the coefficient estimate for the current category and the reference category in each model. Bold text indicates p < 0.05

Back to article page

ISSN: 1471-2288

Contact us

General enquiries: journalsubmissions@springernature.com

BMC Medical Research Methodology

Contact us