Page 181 - 臺大管理論叢第33卷第1期

NTU Management Review Vol. 33 No. 1 Apr. 2023

which one has been reduced owing to the implementation of the CPs. The results indicate
that in the GS division, nine fee items reveal a statistically significant drop (p < 0.05)
after the CPs’ implementation. These items consist of the diagnostic, ward, medicine
service, laboratory, X-ray, therapeutic treatment, special medical supply, medicine and

injection fees. Of these nine items, the ward fee exhibits the largest decrease in terms of
the amount (NT$1,952), and the medicine fee experiences the greatest drop in percentage
terms (54.81%). By contrast, the anesthesia and surgical fees do not exhibit statistically

significant changes. Compared to their counterparts in the ENT division, the fee items in
the GS division that drop significantly after the CPs’ implementation are the diagnostic,
ward, medicine service, laboratory, X-ray, anesthesia, medicine, and injection fees. Among
these eight fee items, the ward fee still has the biggest reduction in terms of the amount
(NT$2,934), and the medicine fee experiences the greatest drop in percentage terms

(79.28%). The surgical, therapeutic treatment and special medical supply fees do not
exhibit statistically significant changes between the GS-ACP and ENT groups (Table 2).

3.2 Multivariate Regression
The results regarding the effects of the CPs on LOS and hospital charges are
presented in Tables 3 and 4, respectively. In Model 1 of Table 3 (with patient-related and
disease-specific characteristics as the independent variables), partial thyroidectomy and
CCI scores have significantly positive influences on LOS. The effect of patient age is

significantly negative, together with the significantly positive effect of patient age squared,
indicating that age exhibits a non-linear relationship with LOS, i.e., the younger or older
a patient is, the longer that LOS will be. In Model 2 of Table 3, basic characteristics

associated with surgeons are also included as independent variables. The results are similar
to those in Model 1. However, we find that surgeon age and its squared term have a
significant influence on LOS. Model 3 of Table 3 further compares the differences for LOS
within each GS division and between the GS-ACP group and ENT group. In particular,
in the GS division, the LOS is significantly longer before the CPs’ implementation (p <

0.01). Further, LOS in the ENT group is significantly longer than in the GS-ACP group.
These results demonstrate that implementing the CPs can effectively reduce LOS, which
is critical for hospitals to remain financially sound. Additionally, the results of Model

3 regarding the rest of the control variables are consistent with those in Models 1 and

173