Appendix A1. Diagnose and Procedure Codes for Digit Amputation Injuries, Digit Replantation and Debridement
System Code Definition
ICD-9*
Diagnose Code
885.0 Traumatic thumb amputation
885.1 Traumatic thumb amputation with complication 886.0 Traumatic finger amputation
886.1 Traumatic finger amputation with complication
ICD-9*
Procedure Code
84.21 Thumb reattachment 84.02 Thumb amputation 84.22 Finger reattachment 84.01 Finger amputation
86.22 Excisional debridement of wound, infection, or burn
* International Classification of Disease, Ninth Revision, Clinical Modification
Explanatory and Outcome Variables Dependent Variable
We examined the probability of success digit replantation, among patients who had digit amputation injuries.
Explanatory Variable
We included age, gender, race, type of insurance, median household income state quartile for patient ZIP code, length of stay in hospital, number of comorbidities as patient’s characteristics variables, and hospital annual volume, as a hospital characteristic variable.
Age was a continuous variable. Race was categorized into white, African American, Hispanic, others. Type of insurance includes Medicare, Medicaid, private insurance, self- pay (no insurance) and others. Median household income state quartile for patient ZIP code includes four quartiles with the fourth quartile to be the richest group. Length of stay in the hospital is a continuous variable with day as unit. Number of comorbidities was calculated by Creation of Comorbidity Variables Software (COMORADM), Version 2.1, provided by HCUP, with a maximum value of 29. Specific comorbidities are shown in Appendix A2.
Appendix A2. Comorbidities List for “Number of Comorbidities Variable” in the GEE Model
List of Comorbidities
• Hypertension • Lymphoma
• Congestive heart failure • Metastatic cancer
• Valvular disease • Solid tumor w/out metastasis
• Pulmonary circulation disease • Rheumatoid arthritis/collagen vas
• Peripheral vascular disease • Coagulopthy
• Paralysis • Obesity
• Other neurological disorders • Weight loss
• Chronic pulmonary disease • Fluid and electrolyte disorders
• Diabetes w/o chronic complications • Chronic blood loss anemia
• Diabetes w/ chronic complications • Deficiency Anemias
• Hypothyroidism • Alcohol abuse
• Renal failure • Drug abuse
• Liver disease • Psychoses
• Peptic ulcer Disease x bleeding • Depression
• Acquired immune deficiency syndrome
Steps for getting optimal cutoff value in SAS
1. Used PROC GENMOD procedure to run regression analysis
2. Used predicted probability from PROC GENMOD model as the only variable in PROC LOGISTIC model procedure, with NOINT option
3. Calculate Youden’s J statistic and find the probability level in PROC LOGISTIC model corresponding to the largest J.
4. Calculate the predicted probability in PROC GENMOD model corresponding to the probability level
5. Find the value of hospital annual volume for that predicted probability in PROC GENMOD model, which is the optimal cutoff value