Nurses' retention and hospital characteristics in New South Wales, CHERE Discussion Paper No 52

NURSES’ RETENTION AND HOSPITAL CHARACTERISTICS IN NEW SOUTH WALES

Table 5 presents marginal effects for personal, job, and hospital characteristics. For discrete
variables, the marginal effect is the shift in probability when placing the person in the group
represented by the variable versus the reference group. For continuous variables, the derivative of the
predicted probability is taken. In either case, all other variables are kept at their observed values.
Marginal effects are computed for each observation and averaged over the sample. Standard errors
around these means are computed with a bootstrap estimator using 500 replications.

Generally, the size of the hospital as measured by separations and bed days does not have a
significant effect on retention. The signs suggest however that the number of non acute cases
increases retention while acute cases have conflicting effects depending on whether they are
measured by separations or bed days and whether they involve babies. The ambiguous effects of
acute cases is not surprising as these can present more challenging situations for the nurses (see
below) but they can also involve shorter patient stays and less longer-term contact with the patients
and their families. Acute cases can also mean more stress for the nurses since they are often
removed from the decision-making level of the health care in these situations.

In surveys of job satisfaction, nurses often report positively on the challenging nature of some
aspects of the job. This could be an explanation for retention rising with the number of admissions
from emergency. The retention probability rises by almost 2 ppts for each 1000 admissions. However
not all measures of complexity are positively related to retention. The number of high cost
procedures and the ANDRG weight suggest that in some dimensions, more complexity leads to
higher quit rates. An increase in the ANDRG weight of 1 causes the retention probability to fall by 3.5
ppts. (As an example of the range of values of the ANDRG weight, a tonsillectomy gets a weight of
0.5, while a liver transplant corresponds to a value of 28.6.) Given the large range of values taken by
this variable, the marginal effect is evidence of a very strong relationship. It would be interesting to
relate these different measures to stress levels, learning possibilities and other aspects of the job.

11

<<   13   14   15   16   17   18   19   >>

More intriguing information