TABLE 2.
Multiple Regression Results With Multiply Imputed Data Sets For Predictors of Male Infant Circumcision: Nationwide Inpatient Sample, 2004
| ba (95% CI) | P | |
| Intercept | 1.276 (0.629, 1.923) | <.001 |
| State Medicaid program covers routine male circumcision | 0.866 (0.375, 1.357) | <.001 |
| Hospital sizeb (Ref = large) | .866 | |
| Small | 0.068 (–0.186, 0.323) | .599 |
| Medium | 0.021 (–0.168, 0.210) | .827 |
| Rural hospital location | 0.213 (–0.182, 0.607) | .290 |
| Teaching hospital | −0.162 (–0.447, 0.124) | .267 |
| Region (Ref = South) | <.001 | |
| Northeast | −0.549 (−0.859, −0.239) | <.001 |
| Midwest | 0.210 (–0.085, 0.504) | .164 |
| West | −0.813 (−1.189, −0.436) | <.001 |
| Hospital ownershipc | .122 | |
| Income quartile of patient's zip coded (Ref = lowest) | .373 | |
| Proportion in second quartile | 0.341 (–0.176, 0.857) | .196 |
| Proportion in third quartile | 0.038 (–0.560, 0.637) | .900 |
| Proportion in highest quartile | −0.107 (–0.634, 0.421) | .692 |
| Length of stay (LOS; Ref = 2 to 5 d) | <.001 | |
| Proportion with LOS = 1 d, centered | −1.487 (−2.343, −0.632) | <.001 |
| Proportion with LOS = 1 d, squared | −4.601 (−8.497, −0.706) | .021 |
| Proportion with LOS ≥ 6 d, centered | −5.267 (−8.009, −2.525) | <.001 |
| Proportion with LOS ≥ 6 d, squared | 4.303 (–0.824, 9.431) | .1 |
| Primary payer (Ref = private insurance) | <.001 | |
| Proportion with Medicaid as primary payer | −2.986 (−3.758, −2.215) | <.001 |
| Medicaid coverage × Medicaid proportion | 2.080 (1.165, 2.996) | <.001 |
| Proportion with no insurance | −1.561 (−2.625, −0.496) | .004 |
| Race/ethnicity (Ref = non-Hispanic White) | <.001 | |
| Proportion African American | 1.324 (–0.258, 2.905) | .099 |
| Medicaid coverage × African American proportion | −1.029 (–2.814, 0.756) | .251 |
| Proportion Hispanic | −2.095 (−2.846, −1.344) | <.001 |
| Medicaid coverage × Hispanic proportion | −0.961 (−1.834, −0.087) | .031 |
| Proportion other race/ethnicity | 0.550 (–0.994, 2.095) | .445 |
| Medicaid coverage × other race proportion | −0.428 (–2.111, 1.255) | .615 |
Note. CI = confidence interval.
Logit-transformed scale.
Determined based on the number of beds. Classifications were relative to a hospital's region, urban or rural location, and teaching status.10
In sampling, hospitals were first assigned to strata on the basis of their region, location, and teaching status and then broken down further by hospital ownership. Number of categories used to define ownership depended on the size of these initial strata. Because of these context-specific aggregation levels, this variable is difficult to interpret, but was included in the multivariate analyses as a means of control for stratification. Hospital size was also used in stratification.
This was based on zip codes with median income of $1 to $35 999 (lowest quartile), $36 000 to $44 999 (second quartile), $45 000 to $58 999 (third quartile), and $59 000 or higher (highest quartile). This was calculated from Nationwide Inpatient Sample data on median household income quartile for each patient's zip code.10