Skip to main content

Advertisement

Table 2 Distribution of glycemic control indicator among predictors, pooled over years and clinics

From: Application of machine learning methodology to assess the performance of DIABETIMSS program for patients with type 2 diabetes in family medicine clinics in Mexico

VariablesHbA1c > =7%HbA1c < 7%MissingAdjusted
p-value
Referred to DIABETIMSS, n (prop.)   < 0.001
 No63284 (0.50)31225 (0.24)33258 (0.26) 
 Yes16254 (0.54)8940 (0.30)4797 (0.16) 
 Missing65391 (0.21)23556 (0.08)224410 (0.72) 
Previous glycemic control, n (prop.)   <0.001
 No69031 (0.60)15161 (0.13)30918 (0.27) 
 Yes12707 (0.26)19724 (0.41)15628 (0.33) 
 Missing63191 (0.21)28836 (0.09)215919 (0.70) 
Age, n (prop.)   < 0.001
 [0,53)39172 (0.55)12795 (0.18)19186 (0.27) 
 [53,62)37784 (0.53)14612 (0.21)18642 (0.26) 
 [62,71)38587 (0.53)18023 (0.25)16378 (0.22) 
 [71, 116]29386 (0.47)18291 (0.29)15395 (0.24) 
 Missing0 (0.00)0 (0.00)192864 (1.00) 
Nutrition status at the beginning of the year, n (prop.)   0.646
 Underweight462 (0.44)190 (0.18)409 (0.39) 
 Normal weight24399 (0.51)10454 (0.22)13326 (0.28) 
 Overweight59249 (0.52)25584 (0.23)28164 (0.25) 
 Obesity60609 (0.53)27360 (0.24)27228 (0.24) 
 Missing210 (0.00)133 (0.00)193338 (1.00) 
Sex, n (prop.)   0.004
 Female86565 (0.52)38609 (0.23)40071 (0.24) 
 Male58364 (0.52)25112 (0.22)29530 (0.26) 
 Missing0 (0.00)0 (0.00)192864 (1.00) 
BMI at the beginning of the year (kg/m2), n (prop.)   0.901
 [11.2, 26.0)36448 (0.51)15636 (0.22)19581 (0.27) 
 [26.0, 28.9)36040 (0.53)15495 (0.23)16969 (0.25) 
 [28.9, 32.4)36242 (0.53)15979 (0.23)16096 (0.24) 
 [32.4, 85.4]35989 (0.52)16478 (0.24)16481 (0.24) 
 Missing210 (0.00)133 (0.00)193338 (1.00) 
Height at the beginning of the year (m), n (prop.)   0.003
 [1.30, 1.50)37877 (0.54)16438 (0.23)16342 (0.23) 
 [1.50, 1.57)39267 (0.53)17393 (0.23)17526 (0.24) 
 [1.57, 1.64)33231 (0.52)14773 (0.23)16363 (0.25) 
 [1.64, 2.10]34344 (0.50)14984 (0.22)18896 (0.28) 
 Missing210 (0.00)133 (0.00)193338 (1.00) 
Weight at the beginning of the year (kg), n (prop.)   < 0.001
 [30, 63)37150 (0.52)15744 (0.22)18099 (0.25)[30, 63)
 [63, 72)36771 (0.52)16188 (0.23)17106 (0.24)[63, 72)
 [72, 82)35912 (0.53)15694 (0.23)16594 (0.24)[72, 82)
 [82, 198]34886 (0.51)15962 (0.23)17328 (0.25)[82, 198]
 Missing210 (0.00)133 (0.00)193338 (1.00) 
Obesity, n (prop.)   0.247
 No24861 (0.50)10644 (0.22)13735 (0.28) 
 Yes119858 (0.53)52944 (0.23)55392 (0.24) 
 Missing210 (0.00)133 (0.00)193338 (1.00) 
Patients with Risk Factors (smoking, hypertension, dyslipidemia), n (prop.)   0.022
 No24358 (0.50)9576 (0.20)14853 (0.30) 
 Yes120571 (0.53)54145 (0.24)54748 (0.24) 
 Missing0 (0.00)0 (0.00)192864 (1.00) 
Smoking Habit, n (prop.)   < 0.001
 No141903 (0.52)62119 (0.23)68196 (0.25) 
 Yes3026 (0.50)1602 (0.27)1405 (0.23) 
 Missing0 (0.00)0 (0.00)192864 (1.00) 
Type of insurance, n (prop.)   0.027
 Others74686 (0.53)31175 (0.22)36123 (0.25) 
 Parents insured/Retired70243 (0.52)32546 (0.24)33478 (0.25) 
 Missing0 (0.00)0 (0.00)192864 (1.00) 
Year, n (prop.)   <0.001
 201228445 (0.30)11297 (0.12)54481 (0.58) 
 201327127 (0.29)10916 (0.12)56180 (0.60) 
 201429070 (0.31)12264 (0.13)52889 (0.56) 
 201530468 (0.32)13582 (0.14)50173 (0.53) 
 201629819 (0.32)15662 (0.17)48742 (0.52) 
Total number of diabetes complications, n (prop.)   <0.001
 077522 (0.50)36812 (0.24)40760 (0.26) 
 145655 (0.54)19015 (0.22)20236 (0.24) 
 >121752 (0.57)7894 (0.21)8605 22) 
  1. The adjusted p-value is derived by tting a generalized estimating equations (GEE) with all the predictors, adjusting for patient ID. Then we did analysis of ‘Wald statistic’ with binomial model and logit link to obtain the p-value. Specifically, the R function is: t = geeglm (formula = indic10 curr diabetimss + edad + sexo + tipo pac + anttab + pesoini + tallaini + imcIni + EdoNutricioIni + facriesg + tot enfcrondiab + SobObes + indic10 prev + year, family = binomial (link = “logit”), data = all complete, id = a l, corstr = “exchangable”, std.err = “san.se” anova (t)