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

NURSES’ RETENTION AND HOSPITAL CHARACTERISTICS IN NEW SOUTH WALES

Notes: * The reference group is defined as female, Australian born, with no post basic qualifications, engaged in
clinical activity in the entry job classification. The coefficients on the missing values dummies are not reported to save
on space but briefly, for both specifications, the dummies for missing age, hours and job classification are significant
and negative while the others (missing information on post basic qualification and activity) are not significant. Although
these results are difficult to interpret by their very nature, they do suggest that individuals who are perhaps
disillusioned with their work and not willing to complete all the questions on the labour force questionnaire are more
likely to leave nursing within the year. ** The proportion correctly predicted is constructed as follows. A count is made
of those observations with predicted probabilities exceeding (less than) the average predicted probability over the
sample; these are correct predictions for those nurses (not) working in 1997. The table presents the proportion over the
observed number in each cell.

Age, years of registration and hours of work are included in a quadratic specification. All three
variables show significant negative coefficients on the squared term, indicating a concave relationship
between each variable and the predicted retention rate.¹⁴ While we might expect strong correlation
between age and years of registration, the two variables are both strongly significant. Treating age
independently of registration, the retention probability is maximised with respect to age at
approximately 39 years. Beyond this, increasing age tends to reduce participation in nursing. Use of a
cubic function of age does not alter these results substantially. We do not find evidence of a U-
shaped retention probability with a trough at the most likely child-bearing years. Instead we find the
lowest retention rates at the very beginning of the nursing careers. This is consistent with the results
found in Shields and Ward (2001), who report the highest quitting intentions for nurses aged less than
29. Also, since the average age in our sample is just under 39 years, a further aging of the cohort of
nurses will contribute to a reduction in retention ceteris paribus.

The maximum predicted probability of retention with respect to years of registration occurs at 21
years, while the sample average is 15 years. This is consistent with the results on age if we consider
a nurse first registering in her early 20’s although the maximum occurs a bit later in terms of years
of registration. Differences between the age and years of registration suggest that those nurses who
registered later in life are more likely to stay as working nurses later in their lifecycle.

In the restricted model the effect of the unemployment rate (matched to the region of the workplace)
exerts a significant positive effect on retention. Nurses are less likely to quit when unemployment in
the region is higher. This effect becomes insignificant when the hospital variables are included. This
suggests that variations in the hospital characteristics are correlated with regional variations in
labour markets conditions.

Increased hours of work are found to be associated with increased retention over the range of
reasonable work hours. The maximum with respect to work hours occurs at 60 hours, while the
average is 36 hours. We do not find evidence that long hours of work contribute to RN quit rates. Of
course this does not capture the presence of shift work nor the intensity of the work. Given the lack
of direct information on these factors, they are likely captured to some extent by the variables
representing hospital characteristics. We turn to these next.

As discussed above the hospital characteristics are jointly very significant. They are also mostly
individually significant. Generally the results suggest that size of the hospital tends to increase
retention probabilities (number of separations) while complexity and intensity tend to reduce retention
(ANDRG weight, high cost procedures, separations per nurse, wait time). Staff levels reduce retention
controlling for other characteristics while expenditures (excluding VMOs) increase the likelihood of
staying on as a nurse. Since the characteristics enter nonlinearly, it is more useful to look at the
impacts of hospital characteristics through the marginal effects.

14 Since the predicted retention probability is generally greater than 0.5, the concave relationship with the index also
indicates a concave relationship between the variable and the predicted probability.

<<   12   13   14   15   16   17   18   >>

More intriguing information