Bifalgorithm post · BMJ Open Diabetes Research & Care is an open access · Results

Regression Models

Regression Models (BMJ Open Diabetes Research & Care is an open access · Results) — "Linear regression models demonstrated significant associations between OGT…

Ideas in this flowchart

  • OGTT glucose, body-size measures, and insulin requirements show significant but limited predictive relationships
  • Models 1 and 2 examined the initial insulin dose required to achieve the target
  • OGTT fasting glucose was significantly associated with the initial insulin dose requirement
  • Pregnancy weight was included in model 1 as an associated variable
  • BMI was included in model 2 as an associated variable
  • The performance of the initial-dose models was limited, as shown in table 3
  • Models 3 and 4 predicted maximum insulin dose requirement
  • Similar associations were observed for maximum insulin dose requirement
  • The maximum-dose models performed even less well than the initial-dose models
  • Models 5 to 8 compared variable association strength using standardized variables
  • OGTT fasting glucose was the strongest predictor of initial insulin dose requirement
  • OGTT fasting glucose, BMI, and pre-pregnancy weight had similar associations with maximum insulin dose requirement
  • Sensitivity analyses adjusted the models for metformin use
  • Adjustment for metformin did not substantially alter the relationships between OGTT glucose, BMI, and insulin requirements
  • Correlations between fasting and 120-minute OGTT glucose and initial and peak insulin requirements are shown in supplemental figures S2–S5
Feed/

Regression Models

No cover

Channel

BMJ Open Diabetes Research & Care is an open access

1 page

"Linear regression models demonstrated significant associations between OGTT fasting glucose, pregnancy weight (model 1) or BMI (model 2), and insulin dose required initially to achieve target, although performance of the model was limited (table 3). Similar associations were seen in models predicting maximum insulin (models 3 and 4; table 3); however, model performance was even lower than in those predicting initial insulin dose required to reach the target. The correlation between fasting and 120-minute OGTT glucose results and initial and peak insulin requirements are shown in online supplemental figures S2–S5.\n\nIn models which compared the strength of variable association with insulin dose requirement via variable standardization (models 5 to 8; table 4), OGTT fasting glucose is the strongest predictor of initial insulin dose requirement, and OGTT fasting glucose, BMI, and pre-pregnancy weight had a similar strength of association with maximum insulin dose requirement. In sensitivity analyses, adjustment for metformin use did not substantially alter the relationships between OGTT glucose values, BMI, and insulin requirements (online supplemental table S1)."

Flowchart preview
👍
0
0 comments0 shares

Sign in to view and post comments.