Table 2.

Associations between turnover rates of NHS clinical staff and risk adjusted patient outcomes at 30 days

	Mortality risk (95% CI)			Unplanned emergency readmission risk after elective surgery (95% CI)
	All cause admissions	Emergency admissions	Elective admissions
Turnover rate (standardised)
Nurses	0.035*** (0.024 to 0.045)	0.052*** (0.037 to 0.067)	0.004 (−0.009 to 0.016)	−0.013 (−0.051 to 0.025)
Senior doctors	0.014*** (0.005 to 0.024)	0.019*** (0.006 to 0.033)	−0.002 (−0.009 to 0.005)	0.017 (−0.014 to 0.047)
Control covariates
No of nurses (standardised)	0.088 (−0.029 to 0.205)	0.096 (−0.076 to 0.268)	−0.021 (−0.097 to 0.056)	0.112 (−0.145 to 0.369)
No of senior doctors (standardised)	−0.049 (−0.155 to 0.056)	−0.047 (−0.197 to 0.104)	0.006 (−0.047 to 0.059)	−0.140 (−0.334 to 0.054)
Quarter 2: Apr-Jun	−0.460*** (−0.480 to −0.441)	−0.629*** (−0.656 to −0.602)	−0.044*** (−0.069 to −0.019)	0.207*** (0.151 to 0.263)
Quarter 3: Jul-Sep	−0.595*** (−0.618 to −0.573)	−0.805*** (−0.837 to −0.773)	−0.046*** (−0.062 to −0.031)	0.250*** (0.186 to 0.313)
Quarter 4: Oct-Dec	−0.199*** (−0.216 to −0.182)	−0.289*** (−0.313 to −0.264)	−0.028*** (−0.043 to −0.014)	0.113*** (0.059 to 0.168)
Constant	3.004*** (2.990 to 3.017)	4.516*** (4.497 to 4.535)	0.495*** (0.484 to 0.507)	6.199*** (6.163 to 6.235)
NHS hospital trust-by-year fixed effects	Yes	Yes	Yes	Yes
R2	0.330	0.296	0.004	0.007

CI=confidence interval.

Ordinary least square estimates of the baseline model specification. Observations: Number of NHS hospital trusts (N)=148; NHS hospital trusts×months (N×T)=14 768 for all regressions. Heteroskedasticity and autocorrelation consistent standard errors clustered at hospital level. Reference quarter is NHS financial year quarter 1 (January to March). Turnover rates and staff number levels (nurses and senior doctors) are standardised and represent a 1 SD change in the variables of interest.

^*P<0.1; **P<0.05; ***P<0.01.