Abstract
Background Public reporting of surgical outcomes must adjust for patient risk. However, whether patient sociodemographic status (SDS) should be included is debatable. Our objective was to empirically compare risk-adjustment models and hospital ratings with or without SDS factors for patients undergoing coronary artery bypass grafting. Study Design This is a retrospective analysis of the California Coronary Artery Bypass Grafting Outcomes Reporting Program, 2011–2012. Outcomes included 30-day or in-hospital mortality, perioperative stroke, and 30-day readmission. Sociodemographic status factors included race, language, insurance, ZIP code-based median income, and percent that were a college graduate. The c-statistic and goodness-of-fit were compared between models with and without SDS factors. Differences in hospital performance rating when adjusting for SDS were also compared. Results None of the SDS factors predicted mortality. Income, education, and language had no impact on any outcomes. Insurance predicted stroke (MediCal vs private insurance, odds ratio [OR] = 1.91; 95% CI, 1.11–3.31; p = 0.020) and readmissions (Medicare vs private insurance, OR = 1.36; 95% CI, 1.16–1.61; p < 0.001; MediCal vs private insurance, OR = 1.56; 95% CI, 1.26–1.94; p < 0.001). Race also predicted stroke (Asian vs white, OR = 2.26; p < 0.001). Adding SDS factors improved the c-statistic in readmission only (0.652 vs 0.645; p = 0.008). Goodness-of-fit worsened when adding SDS factors to mortality models, but was no different in stroke or readmissions. Hospital performance rating only changed in readmissions; of 124 hospitals, only 1 hospital moved from “better” to “average” when adjusting for SDS. Conclusions Adjusting for insurance improves statistical models when analyzing readmissions after coronary artery bypass grafting, but does not impact hospital performance ratings substantially. Deciding whether SDS should be included in a patient's risk profile depends on valid measurements of SDS and requires a nuanced approach to assessing how these variables improve risk-adjusted models.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 221-230 |
| Number of pages | 10 |
| Journal | Journal of the American College of Surgeons |
| Volume | 223 |
| Issue number | 2 |
| DOIs | |
| State | Published - Aug 1 2016 |
Fingerprint
ASJC Scopus subject areas
- Surgery
Cite this
Should Risk Adjustment for Surgical Outcomes Reporting Include Sociodemographic Status? A Study of Coronary Artery Bypass Grafting in California. / Anderson, Jamie; Li, Zhongmin; Romano, Patrick S; Parker, Joseph; Chang, David C.
In: Journal of the American College of Surgeons, Vol. 223, No. 2, 01.08.2016, p. 221-230.Research output: Contribution to journal › Article
}
TY - JOUR
T1 - Should Risk Adjustment for Surgical Outcomes Reporting Include Sociodemographic Status? A Study of Coronary Artery Bypass Grafting in California
AU - Anderson, Jamie
AU - Li, Zhongmin
AU - Romano, Patrick S
AU - Parker, Joseph
AU - Chang, David C.
PY - 2016/8/1
Y1 - 2016/8/1
N2 - Background Public reporting of surgical outcomes must adjust for patient risk. However, whether patient sociodemographic status (SDS) should be included is debatable. Our objective was to empirically compare risk-adjustment models and hospital ratings with or without SDS factors for patients undergoing coronary artery bypass grafting. Study Design This is a retrospective analysis of the California Coronary Artery Bypass Grafting Outcomes Reporting Program, 2011–2012. Outcomes included 30-day or in-hospital mortality, perioperative stroke, and 30-day readmission. Sociodemographic status factors included race, language, insurance, ZIP code-based median income, and percent that were a college graduate. The c-statistic and goodness-of-fit were compared between models with and without SDS factors. Differences in hospital performance rating when adjusting for SDS were also compared. Results None of the SDS factors predicted mortality. Income, education, and language had no impact on any outcomes. Insurance predicted stroke (MediCal vs private insurance, odds ratio [OR] = 1.91; 95% CI, 1.11–3.31; p = 0.020) and readmissions (Medicare vs private insurance, OR = 1.36; 95% CI, 1.16–1.61; p < 0.001; MediCal vs private insurance, OR = 1.56; 95% CI, 1.26–1.94; p < 0.001). Race also predicted stroke (Asian vs white, OR = 2.26; p < 0.001). Adding SDS factors improved the c-statistic in readmission only (0.652 vs 0.645; p = 0.008). Goodness-of-fit worsened when adding SDS factors to mortality models, but was no different in stroke or readmissions. Hospital performance rating only changed in readmissions; of 124 hospitals, only 1 hospital moved from “better” to “average” when adjusting for SDS. Conclusions Adjusting for insurance improves statistical models when analyzing readmissions after coronary artery bypass grafting, but does not impact hospital performance ratings substantially. Deciding whether SDS should be included in a patient's risk profile depends on valid measurements of SDS and requires a nuanced approach to assessing how these variables improve risk-adjusted models.
AB - Background Public reporting of surgical outcomes must adjust for patient risk. However, whether patient sociodemographic status (SDS) should be included is debatable. Our objective was to empirically compare risk-adjustment models and hospital ratings with or without SDS factors for patients undergoing coronary artery bypass grafting. Study Design This is a retrospective analysis of the California Coronary Artery Bypass Grafting Outcomes Reporting Program, 2011–2012. Outcomes included 30-day or in-hospital mortality, perioperative stroke, and 30-day readmission. Sociodemographic status factors included race, language, insurance, ZIP code-based median income, and percent that were a college graduate. The c-statistic and goodness-of-fit were compared between models with and without SDS factors. Differences in hospital performance rating when adjusting for SDS were also compared. Results None of the SDS factors predicted mortality. Income, education, and language had no impact on any outcomes. Insurance predicted stroke (MediCal vs private insurance, odds ratio [OR] = 1.91; 95% CI, 1.11–3.31; p = 0.020) and readmissions (Medicare vs private insurance, OR = 1.36; 95% CI, 1.16–1.61; p < 0.001; MediCal vs private insurance, OR = 1.56; 95% CI, 1.26–1.94; p < 0.001). Race also predicted stroke (Asian vs white, OR = 2.26; p < 0.001). Adding SDS factors improved the c-statistic in readmission only (0.652 vs 0.645; p = 0.008). Goodness-of-fit worsened when adding SDS factors to mortality models, but was no different in stroke or readmissions. Hospital performance rating only changed in readmissions; of 124 hospitals, only 1 hospital moved from “better” to “average” when adjusting for SDS. Conclusions Adjusting for insurance improves statistical models when analyzing readmissions after coronary artery bypass grafting, but does not impact hospital performance ratings substantially. Deciding whether SDS should be included in a patient's risk profile depends on valid measurements of SDS and requires a nuanced approach to assessing how these variables improve risk-adjusted models.
UR - http://www.scopus.com/inward/record.url?scp=84991208141&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84991208141&partnerID=8YFLogxK
U2 - 10.1016/j.jamcollsurg.2016.05.008
DO - 10.1016/j.jamcollsurg.2016.05.008
M3 - Article
VL - 223
SP - 221
EP - 230
JO - Journal of the American College of Surgeons
JF - Journal of the American College of Surgeons
SN - 1072-7515
IS - 2
ER -