Skip to main content
. 2021 Apr 7;71(707):e423–e431. doi: 10.3399/BJGP.2020.0486

Table 4.

Final mixed-effects regression models of patient and consultation factors against consultation duration in the development and validation data sets;a random effects for patient and practice

2013/2014 Development (N = 725 616 consultations) 2017/2018 Validation (N = 234 447 consultations)


Change, minutes 95% CI P-value Change, minutes 95% CI P-value
Consultation factorsa
Mean duration of consultation 9.78 9.56 to 10.00 <0.001 9.81 9.38 to 10.24 <0.001
About drug/alcohol use 4.19 3.68 to 4.69 <0.001 3.73 2.54 to 4.92 <0.001
About chronic pain 0.98 0.86 to 1.10 <0.001 1.00 0.79 to 1.21 <0.001
About dementiab 1.42 0.90 to 1.94 <0.001 NA NA NA
Results in emergency hospital admission 7.76 7.09 to 8.43 <0.001 4.69 3.47 to 5.92 <0.001
About learning disability/autism 3.84 3.17 to 4.52 <0.001 3.05 1.98 to 4.12 <0.001
About mental health problems 3.85 3.70 to 3.99 <0.001 3.83 3.58 to 4.08 <0.001
≥2 diagnoses from unique Read chapters recorded 2.54 2.42 to 2.67 <0.001 2.86 2.61 to 3.12 <0.001
≥3 unique substances prescribed 1.73 1.67 to 1.80 <0.001 1.82 1.69 to 1.94 <0.001
≥2 preventive/routine tasks carried out 3.73 3.55 to 3.92 <0.001 4.81 4.44 to 5.19 <0.001
First consultation after diabetes diagnosis 3.05 2.48 to 3.62 <0.001 2.39 1.60 to 3.19 <0.001

Patient factorsa
Mean duration of consultation 10.02 9.80 to 10.25 <0.001 10.03 9.60 to 10.46 <0.001
Drug/alcohol abuse in previous yearb 1.89 1.44 to 2.33 <0.001 NA NA NA
Chronic pain in previous year 0.72 0.57 to 0.87 <0.001 0.75 0.51 to 1.00 <0.001
Domestic violence in last year 1.43 0.37 to 2.49 0.008 2.37 0.99 to 3.74 <0.001
Homelessness in previous yearb 1.36 0.39 to 2.33 0.006 NA NA NA
No English noted in last 3 years 0.98 0.25 to 1.72 0.009 1.01 0.12 to 1.91 0.026
≥3 long-term conditionsb 0.30 0.21 to 0.40 <0.001 NA NA NA
Personality/disruptive disorder (ever)b 0.51 0.12 to 0.89 0.01 NA NA NA
a

Based on separate regressions for consultation and patient factors.

b

These variables were retained in the complexity measure despite low prevalence or lack of significant effect in the 2017–2018 model.