William B. Saunders, PhD, MPH Program Director, Health Informatics PSM & Certificate Programs Laura J. Dunlap, RN
Background Research Questions Methods Results for North Carolina Results for Specific Counties Next Steps
The US healthcare system is a complex and ever-changing industry Readmission policies recently enacted by CMS, in concert with the Patient Protection and Affordable Care Act (PPACA) of 2010, have demonstrated the US government commitment to risk-sharing with healthcare delivery systems across the country This, along with the historical behavior of commercial insurance payers following suit with CMS actions, have driven healthcare systems to renegotiate contracts with payers. Given this context, we have been examining hospital readmissions in North Carolina using graphical and analytical techniques.
HCUP What s so special about North Carolina?
MI ICD-9 410.xx Patients: 37,341 CCI average: 1.95 median: 1.00 PN ICD-9 codes 480.X, Viral Pneumonia ; 481, Pneumococcal Pneumonia; 482.X, Other bacterial pneumonia; 483.X, Pneumonia, due to other specified organism; 485, Bronchopneumonia, organism unspecified; 486, Pneumonia, unspecified organism Patients: 64,061 CCI average: 5.18 median: 5.00 HF ICD-9 428.xx Patients: 108,861 CCI average: 5.71 median: 5.00
Using HCUP data, determine the predictors, annual burden and cost of care of readmission for selected conditions Illustrate Tableau as a useful tool for examining readmissions by county Present a framework for studying readmissions in light of current policies, data availability, and provider/payer relationships
Focused on heart failure, pneumonia, & acute MI Discharge data from North Carolina hospitals were obtained from the Healthcare Utilization Project (HCUP) (administered by the Agency for Healthcare Research & Quality (AHRQ)) Of particular interest was whether or not high-risk patients were more or less likely to be readmitted given extended travel time to the hospital, or going to a hospital different than the initial admission for the particular condition Tableau was used to display and dig deeper into counties that had higher readmission rates, lower income, and higher average length of stay, examining total costs for these admissions
What is the readmission rate for heart failure (HF) in North Carolina? Myocardial Infarction (MI) Pneumonia (PN) Condition Frequency HF 15,475 MI 8,527 PN 16,879
What is the frequency of readmission for MI in North Carolina (2010)? 30-day all-cause readmissions - 891 / 8527 (10%) MI readmissions - 101 / 8527 (1%) What is the frequency of readmission for PN in North Carolina (2010)? 30-day all-cause readmissions - 1967 / 16879 (12%) PN readmissions - 378 / 16879 (2%)
What is the readmission rate for heart failure (HF) in North Carolina? 30-day all-cause readmissions - 2304 / 15475 (15%) HF readmissions - 2257 / 15475 (15%) Given 9.562 million people in North Carolina, & 8,025,618 in the 67 counties where data was available, the readmission rate was 19.28 per 10k
Bivariate analyses Variable Discharge to SNF vs routine Chi-square p-value % readmitted 16% Discharge to Acute care vs routine 15% <0.0001 9% Discharge to Home Health vs routine 14% Female vs Male 0.0003 12% 9% Other race vs White 0.006 12% 10% Higher income vs 0-25% 0.43 10% 11% LOS 6+ days vs 1-5 <.0001 13% 9% # chronic conditions 8+ vs 1-7 <.0001 14% 9%
Odds Ratio Estimates Effect Point Estimate Discharge to SNF vs routine 1.13 0.90 1.42 95% Wald Confidence Limits Discharge to Acute care vs routine 1.43 0.77 2.68 Discharge to Home Health vs routine 1.05 0.84 1.31 Age (mean=67.3+14.6) 1.02 1.01 1.02 Female vs Male 1.07 0.92 1.25 Other race vs White 1.13 1.04 1.22 Higher income vs 0-25% 0.99 0.92 1.08 LOS 3-5 days vs 1-2 1.60 1.32 1.96 LOS 6+ days vs 1-2 1.64 1.31 2.05 # chronic conditions 5-7 vs 1-4 1.37 1.11 1.70 # chronic conditions 8+ vs 1-4 1.92 1.52 2.42
Bivariate analyses Variable Discharge to SNF vs routine Chi-square p-value % readmitted 14% Discharge to Acute care vs routine 16% <0.0001 10% Discharge to Home Health vs routine 15% Female vs Male 0.12 11% 12% Other race vs White 0.03 12% 11% Higher income vs 0-25% 0.009 11% 12% LOS 6+ days vs 1-5 <.0001 13% 11% # chronic conditions 8+ vs 1-7 <.0001 15% 11%
Odds Ratio Estimates Effect Point Estimate Discharge to SNF vs routine 1.40 1.22 1.60 95% Wald Confidence Limits Discharge to Acute care vs routine 1.46 0.87 2.46 Discharge to Home Health vs routine 1.41 1.24 1.61 Age (mean=67.9+16.7) 0.99 0.99 0.99 Female vs Male 0.91 0.83 1.00 Other race vs White 1.05 0.99 1.11 Higher income vs 0-25% 0.93 0.88 0.99 LOS 3-5 days vs 1-2 1.23 1.07 1.41 LOS 6+ days vs 1-2 1.30 1.13 1.50 # chronic conditions 5-7 vs 1-4 1.35 1.21 1.50 # chronic conditions 8+ vs 1-4 1.62 1.40 1.88
Bivariate analyses Variable Discharge to SNF vs routine Chi-square p-value % readmitted 16% Discharge to Acute care vs routine 21% <0.0001 13% Discharge to Home Health vs routine 18% Female vs Male 0.25 15% 15% Other race vs White 0.22 15% 14% Higher income vs 0-25% 0.018 14% 16% LOS 6+ days vs 1-5 0.0022 16% 14% # chronic conditions 8+ vs 1-7 <.0001 17% 14%
Odds Ratio Estimates Effect Point Estimate 95% Wald Confidence Limits Discharge to SNF vs routine 1.211 1.056 1.388 Discharge to Acute care vs routine 1.719 1.204 2.455 Discharge to Home Health vs routine 1.360 1.213 1.525 Age (mean=71.9+14.3) 1.001 0.998 1.005 Female vs Male 1.024 0.933 1.124 Other race vs White 1.060 1.004 1.118 Higher income vs 0-25% 0.939 0.890 0.992 LOS 3-5 days vs 1-2 1.087 0.970 1.218 LOS 6+ days vs 1-2 1.173 1.034 1.332 # chronic conditions 5-7 vs 1-4 1.184 1.032 1.357 # chronic conditions 8+ vs 1-4 1.410 1.216 1.636
Population data Income data Length of stay data % readmission Readmission rate Cost data
Variable Frequency (%) Female 1,236 (54%) White Race 1,431 (66%) Median household income quartile for patient ZIP 0-25 th %-tile 1,098 (48%) LOS 3-5 days 1,016 (44%) LOS 6+ days 710 (31%) # chronic conditions 5-7 1,303 (57%) # chronic conditions 8+ 681 (30%) Same hospital (readmission) 90%
Variable Mean (SD) Min Max Age 73 (14) 18 102 Cost (initial admission) $10,563 ($14,227) $627 $273,250 Cost (readmission) $10,344 ($12,825) $441 $183,601 Admissions (total) 3.2 (1.9) 2 25
high Readmission rate (used hcounty1) Laura 25% income (hosp county1) LOS 8p Cost2 (Bill to provide LS means for cost2)
Let s look at county-level data
Sample size Migration out of state 30-day overall readmission vs indication-specific Tableau learning curve
Factors related to readmission for HF, MI, & PN are similar Data fluctuates by county County-specific data highlight differences in patient migration related to hospital readmission Tableau is a useful tool for examining data Results need to be expanded to other conditions and compared to other data sources