This 4-year balanced panel data included 120 records. The equations (1), (2) and (3) were tested in turn to explore the moderation effect of workload on the relationship between medical professionals and CRPA rate.
Correlation analyses between the variables
As shown in Table 3, workload was significantly and positively correlated with CRPA rate(r=0.363; p=0.000).However, the association between numbers of physicians, registered nurses, pharmacists, clinical technicians and CRPA rate were not statistically significant, respectively (r=0.120, r=0.109, r=0.095, r=0.136; p=0.194, p=0.236, p=0.301, p=0.140).
Table 3. Correlation coefficients among medical professionals, CRPA rate and workload
Variable
|
1
|
2
|
3
|
4
|
5
|
6
|
7
|
1.ln PHY
|
1.000
|
|
|
|
|
|
|
2.ln NUR
|
0.788***
|
1.000
|
|
|
|
|
|
3.ln PHA
|
0.788***
|
0.794***
|
1.000
|
|
|
|
|
4.ln MCB
|
0.731***
|
0.849***
|
0.766***
|
1.000
|
|
|
|
5.ln WORKL
|
0.302***
|
0.388***
|
0.486***
|
0.225**
|
1.000
|
|
|
6.ln HOS
|
0.109
|
0.128
|
0.086
|
-0.037
|
0.158*
|
1.000
|
|
7. ln CRPA
|
0.120
|
0.109
|
0.095
|
0.136
|
0.363***
|
0.461***
|
1.000
|
*** p<0.01, ** p<0.05, * p<0.1
Relationship between Medical Professionals and CRPA
As shown in Table 4, there were significant negative correlations between physicians, registered nurses, pharmacists, clinical technicians and CRPA rate, respectively (Coef.= -0.889, -0.775, -0.1.176, -0.822;P= 0.003, 0.003, 0.011, 0.007). That meant the increase in medical professionals was associated with lower CRPA rate. Hypothesis 1 was supported.
Table 4. Relationship between medical professionals and CRPA rate
Variable
|
Model (1)
|
Model (2)
|
Model (3)
|
Model (4)
|
ln PHY
|
-0.889***
|
|
|
|
|
(0.276)
|
|
|
|
ln NUR
|
|
-0.775***
|
|
|
|
|
(0.240)
|
|
|
ln PHA
|
|
|
-1.176**
|
|
|
|
|
(0.434)
|
|
ln MCB
|
|
|
|
-0.822***
|
|
|
|
|
(0.283)
|
ln HOS
|
-0.418***
|
-0.322**
|
-0.465***
|
-0.445***
|
|
(0.120)
|
(0.143)
|
(0.125)
|
(0.126)
|
Constant
|
5.302***
|
5.034***
|
3.560***
|
3.922***
|
|
(0.354)
|
(0.403)
|
(0.959)
|
(0.791)
|
R-squared
|
0.431
|
0.457
|
0.404
|
0.412
|
F
|
66.77***
|
64.89***
|
50.88***
|
55.55***
|
*** p<0.01, ** p<0.05, * p<0.1, cluster-robust standard errors in parentheses.
The Moderation Effect Test of Medical Professionals Workload
Model (5), model (7), model (9) and model (11) displayed that there were significant positive correlations between the number of physicians, registered nurses, pharmacists, clinical technicians and CRPA rate after the addition of workload , respectively(Coef.= -0.888,-0.757, -1.125, -0.785; P=0.005, 0.004, 0.017, 0.008). Model (6), model (8), model (10) and model (12) showed that the interaction terms of workload and medical professionals (ln WORKL* ln PHY, ln WORKL* ln NUR ln WORKL* ln PHA, ln WORKL* ln PHY, ln WORKL* ln MCB) were significantly positive with CRPA rate, respectively (Coef.= 1.270, 1.400, 2.210, 1.634; P=0.004, 0.001, 0.035, 0.003). This indicated that workload moderated the impact of the number of medical professionals on CRPA rate, and the increase in workload may lead to CRPA rate increased. Therefore, hypothesis 2 was supported (Table 5).
Table 5. Workload ' moderation effect on the relationship between medical professionals and CRPA rate
Variable
|
Model (5)
|
Model (6)
|
Model (7)
|
Model (8)
|
Model (9)
|
Model (10)
|
Model (11)
|
Model (12)
|
ln PHY
|
-0.888**
|
-0.781**
|
|
|
|
|
|
|
|
(0.295)
|
(0.291)
|
|
|
|
|
|
|
ln NUR
|
|
|
-0.757***
|
-0.683***
|
|
|
|
|
|
|
|
(0.241)
|
(0.221)
|
|
|
|
|
ln PHA
|
|
|
|
|
-1.125**
|
-0.980**
|
|
|
|
|
|
|
|
(0.445)
|
(0.403)
|
|
|
ln MCB
|
|
|
|
|
|
|
-0.785***
|
-0.739**
|
|
|
|
|
|
|
|
(0.278)
|
(0.272)
|
ln WORKL
|
0.001
|
0.298
|
0.181
|
0.771*
|
0.398
|
0.702
|
0.296
|
0.702
|
|
(0.503)
|
(0.526)
|
(0.469)
|
(0.434)
|
(0.602)
|
(0.640)
|
(0.546)
|
(0.570)
|
lnWORKL* ln STAFF
|
|
1.270***
|
|
1.400***
|
|
2.210**
|
|
1.634***
|
|
(0.403)
|
|
(0.358)
|
|
(0.999)
|
|
(0.500)
|
ln HOS
|
-0.418***
|
-0.443***
|
-0.316**
|
-0.306**
|
-0.443***
|
-0.438***
|
-0.433***
|
-0.426***
|
|
(0.121)
|
(0.118)
|
(0.143)
|
(0.133)
|
(0.130)
|
(1.618)
|
(0.130)
|
(0.132)
|
Constant
|
5.289***
|
4.736***
|
4.646***
|
3.379
|
2.764*
|
2.250
|
3.345**
|
2.569*
|
|
(1.228)
|
(1.237)
|
(1.167)
|
(1.057)
|
(1.608)
|
(1.618)
|
(1.463)
|
(1.481)
|
R-squared
|
0.431
|
0.452
|
0.458
|
0.500
|
0.410
|
0.439
|
0.415
|
0.447
|
F
|
55.98***
|
51.39***
|
49.46***
|
57.19***
|
39.76***
|
39.20***
|
43.98***
|
43.20***
|
*** p<0.01, ** p<0.05, * p<0.1, cluster-robust standard errors in parentheses.
ln STAFF refers to ln PHY in model(5) and model(6), ln NUR in model(7) and model(8), ln PHA in model(9) and model(10) , ln MCB in model(11) and model(12).