Skip to main content

Table 4 Parameter estimates for parsimonious models across outcomes*

From: A randomised fractional factorial screening experiment to predict effective features of audit and feedback

Parameter Primary outcome: intention
Est (p-value)
Proximal intention Comprehension
Est (p-value)
User experience
Est (p-value)
Attention
Est (p-value)
Goals
Est (p-value)
Action plan
Est (p-value)
Review performance
Est (p-value)
Intercept 1.829 (< 0.001) 1.812 (< 0.001) 1.738 (< 0.001) 1.608 (< 0.001) 1.335 (< 0.001) 2.011 (< 0.001) 1.870 (< 0.001)
Block 0.09 (0.107) 0.082 (0.120) 0.026 (0.681) 0.017 (0.799) 0.093 (0.176) 0.036 (0.390) 0.042 (0.310)
NCA (vs NDA)
 MINAP 0.211 (0.317) 0.521 (< 0.001) 0.445 (0.069) −0.482 (0.049) 0.089 (0.607) 0.140 (0.203) 0.090 (0.407)
 NCABT −0.893 (< 0.001) 0.253 (0.115) −0.638 (0.006) −0.603 (0.011) −0.565 (0.008) 0.135 (0.292) 0.118 (0.356)
 PICANet 0.361 (0.270) 0.755 (0.002) 0.164 (0.657) 0.263 (0.491) 0.186 (0.548) 0.312 (0.110) 0.247 (0.206)
 TARN 0.003 (0.989) 0.304 (0.050) 0.005 (0.984) 0.275 (0.291) 0.246 (0.208) 0.022 (0.858) 0.014 (0.908)
Non-clinical (vs clinical) −0.867 (< 0.001)   −0.489 (0.030) −0.571 (0.015) −0.290 (0.043)   
A:Effective comparator 0.038 (0.498) 0.015 (0.778) 0.082 (0.190) 0.018 (0.837) 0.019 (0.784) −0.091 (0.029) −0.087 (0.036)
B:Multimodal feedback 0.018 (0.807) 0.016 (0.766) 0.061 (0.575) 0.064 (0.313) 0.052 (0.436) 0.054 (0.198) 0.048 (0.251)
C:Specific actions 0.082 (0.141) 0.044 (0.400) 0.065 (0.285) 0.075 (0.240) 0.118 (0.075)   0.017 (0.679)
D:Optional detail 0.017 (0.816)   0.050 (0.415) 0.051 (0.434) 0.093 (0.174) 0.022 (0.603) 0.056 (0.176)
E:Patient voice 0.078 (0.161)   0.059 (0.343) 0.064 (0.327) 0.088 (0.201)   0.057 (0.172)
F:Cognitive load 0.008 (0.890) 0.126 (0.016) 0.049 (0.656) 0.044 (0.708) 0.042 (0.533) 0.103 (0.014)  
A * B 0.011 (0.844) 0.083 (0.111) 0.081 (0.190) 0.073 (0.256) 0.115 (0.089)   
A * C    0.041 (0.500) 0.080 (0.210)    
A * D    0.069 (0.266) 0.095 (0.132)   0.078 (0.068) 0.085 (0.043)
A * E 0.014 (0.798)   0.046 (0.458) 0.044 (0.496) 0.096 (0.153)   
A * F    0.025 (0.696) 0.032 (0.616)    
B * C   0.078 (0.138)      
B * D −0.112 (0.047)   0.114 (0.067)    −0.114 (0.008) −0.115 (0.006)
B * E 0.035 (0.537)   0.013 (0.832) 0.026 (0.690)    
B * F   0.087 (0.089) 0.057 (0.607)   0.119 (0.085)   
C * D     0.107 (0.099) 0.119 (0.078)   
C * E        0.066 (0.116)
C * F 0.093 (0.09)   0.059 (0.338) 0.050 (0.438)    
D * E    0.045 (0.469)     0.085 (0.039)
D * F 0.093 (0.089)     0.123 (0.065) 0.073 (0.079)  
E * F    0.008 (0.894)     
Additional interactions
A * B * E = C * D * F 0.101 (0.072)   0.121 (0.053) −0.137 (0.033)    
A * C * F = B * D * E    0.099 (0.109) 0.159 (0.015)    
A * D * E = B * C * F    0.096 (0.136)     
A * E * F = B * C * D    0.092 (0.132)     
Non-clinical * MINAP 0.453 (0.117)   0.721 (0.030) 0.695 (0.040)    
Non-clinical * NCABT 1.312 (0.011)   0.817 (0.146) 0.829 (0.160)    
Non-clinical * PICA 0.783 (0.141)   −1.360 (0.021) 0.708 (0.248)    
Non-clinical * TARN 0.017 (0.959)   0.310 (0.406) 0.135 (0.728)    
A * Non-clinical     0.185 (0.160)    
B * Non-clinical 0.196 (0.083)       
D * Non-clinical 0.195 (0.087)       
B * NCABT    0.218 (0.247)     
B * MINAP    0.101 (0.526)     
B * PICANet    0.607 (0.035)     
B * TARN    0.052 (0.775)     
F * NCABT    0.135 (0.477) 0.038 (0.846)    
F * MINAP    0.094 (0.561) 0.022 (0.894)    
F * PICANet    0.080 (0.782) 0.461 (0.132)    
F * TARN    0.357 (0.049) 0.407 (0.033)    
B * F * NCABT    0.170 (0.372)     
B *F * MINAP    −0.400 (0.013)     
B *F * PICANet    0.569 (0.046)     
B *F * TARN    0.036 (0.840)     
Parameter estimates  
• +ve/−ve: positive or negative effect on outcome
Modification interactions + estimate = synergistic/ve estimate = antagonistic
   
  1. *The columns identify promising detected effects for each outcome, whilst rows identify consistent effects identified across outcomes. Blank cells represent parameters not included in the model. Parameter estimates are all on the same scale of −3 “completely disagree” to +3 “completely agree”. The model intercept represents the overall predicted mean outcome in the NDA and clinical recipient reference groups, averaged across all possible combinations of modifications. Parameter estimates for NCA and role represent the deviation from the predicted mean outcome for the alternative audit and non-clinical recipients. Positive estimates represent an improvement in outcome compared to the reference NDA and clinical recipients, whereas a negative parameter estimate represents a detrimental effect on outcome. Positive parameter estimates for the main effect of each modification represent an improvement in outcome when the modification is ON (+1) and a negative effect on outcome when the modification is OFF (−1). Conversely, negative parameter estimates represent a negative effect on outcome when the modification is ON (+1) and an improvement in outcome when the modification is OFF (−1). Parameter estimates for interactions between modifications represent the additional deviation from the predicted mean outcome