TABLE 3.
Association between hospital characteristics and hospital publication volumes using negative binomial regression
| Variables | Rate ratio | 95% confidence interval | p value |
|---|---|---|---|
| Teaching status | |||
| COTH teaching | 24.29 | 15.03–38.86 | <0.0001 |
| Non‐COTH teaching | 3.09 | 2.44–3.94 | <0.0001 |
| Nonteaching hospitals | ref | ref | ref |
| Hospital ownership | |||
| Not‐for‐profit | 1.39 | 1.04–1.86 | 0.03 |
| Government‐owned | 2.59 | 1.75–3.78 | <0.0001 |
| Investor‐owned | ref | ref | ref |
| Metropolitan location | |||
| Yes | 1.16 | 0.87–1.54 | 0.32 |
| No | ref | ref | ref |
| Magnet status in 2015–2016 | |||
| Yes | 1.36 | 0.97–1.90 | 0.07 |
| No | ref | ref | ref |
| Hospital beds (per 10 beds increase) | 1.02 | 1.01–1.03 | <0.0001 |
| Percent of Medicaid admissions among all inpatient admissions, 1% increase | 0.99 | 0.98–0.998 | 0.01 |
| Regions (Census divisions) | |||
| Mid/South Atlantic | 1.02 | 0.66–1.58 | 0.92 |
| Pacific | 1.99 | 1.23–3.19 | 0.01 |
| East North/South Central | 0.99 | 0.63–1.55 | 0.97 |
| West North/South Central | 1.42 | 0.90–2.23 | 0.12 |
| New England | 3.90 | 2.16–6.96 | <0.0001 |
| Mountain | ref | ref | ref |
Note: Negative binomial regression for number of publications with an offset for the annual admissions was used to estimate the association between hospital characteristics and expected publication rates. The rate ratio refers to the ratio of expected rate of publications per annual admission relative to the reference group after adjustment for the other variables listed. Significant predictors of high publication rates included teaching intensity, number of hospital beds, New England or Pacific region, not‐for‐profit or government ownership. Percentage of Medicaid admissions was negatively associated with publication rates.