Purposeful variable selection and stratification to impute missing Focused Assessment with Sonography for Trauma data...

Area of Science:

• Trauma research
• Medical informatics
• Biostatistics

Background:

• The Focused Assessment with Sonography for Trauma (FAST) examination is crucial for retrospective trauma studies.
• Missing FAST data can compromise study validity due to non-random missingness.
• Developing an imputation method is essential to address missing FAST data.

Purpose of the Study:

• To devise an imputation method for missing FAST examination data.
• To overcome limitations of excluding patients with missing FAST values in trauma research.
• To improve the validity of retrospective trauma studies with incomplete FAST data.

Main Methods:

• Imputation was performed assuming data were missing at random.
• Patients with missing FAST data were grouped based on injury characteristics and associated variables.
• Alternate clinical variables were used to classify missing FAST values as positive or negative.

Main Results:

• The imputation method achieved 87.2% correct classification of non-missing FAST results.
• All 407 missing FAST values were successfully imputed, with 109 classified as positive.
• Specific injury types (severe head/neck, extremity, abdominal) were significantly associated with missing FAST data.

Conclusions:

• Purposeful imputation for missing FAST data using variable interactions is a valuable tool.
• This method can enhance sensitivity analyses for different missing data mechanisms.
• The approach shows potential for broad application in clinical and translational research, warranting further validation.