The Cost and Outcome Effectiveness of Total Hip Replacement: Technique Choice and Volume-Output Effects Matter; Applied Health Economics and Health Policy; Goldstein, Babikian, Rana, Mackenzie and Millar;
APPENDIX I: Data Construction
The All-Payers Claims Database extract contained facilities claims, professional claims and member (patient) and provider information.
Outcomes
Total costs
Total costs were generated from insurance billing data in the facilities and professional claims data. All costs associated with billings on the day of the operation and for a subsequent 90 day window were summed.
Length of Stay
Facilities claims with associated admissions date equal to the operation date were isolated. The admissions date was subtracted from the discharge date associated with that claim to generate hospital length of stay.
Discharge Status
Each facilities claim had an associated discharge status. The discharge status for the billing corresponding to THR was recoded: discharged home, discharged home with care and rehab facility stay defined as short-term general hospital or skilled nursing facility or swing bed, or impatient rehab with hospital component stays.
Patient Characteristics
Rural
An urban community was defined as any town included within the labor market area (LMA) associated with the seven most populous cities in Maine as defined by the U.S. Department of Labor. Relevant LMAs that bordered with New Hampshire were included. The seven LMAs considered are: Augusta Micropolitan, Bangor Metropolitan, Brunswick, Lewiston-Auburn Metropolitan, Portland – S. Portland – Biddeford Metropolitan and Sanford Micropolitan, Portsmouth N.H. – Maine Metropolitan, and Rochester-Dover N.H.- Maine Metropolitan areas. All other towns were classified as rural. A patient’s home town was used to assign urban and rural designations.
Low Income, Medicare and Medicaid
Patient’s data listed insurance coverage. Patients enrolled in Medicaid were classified as Low Income. Patients covered by Medicare were coded as such. Patients not covered by Medicare, but covered by Medicaid were coded as having Medicaid as a primary payer.
Comorbidities
14 binary comorbidity variables were generated by searching diagnostic codes associated with each patient’s claims during the 90-day period prior to and including the operation date. If a diagnostic code for one of the comorbidities occurred on one or more service dates prior to the operation date, the respective comorbidity variable was assigned a value one. If a diagnostic code appeared for the first time on the day of the operation, a value of .5 was assigned because such an occurrence could not be distinguished as a prior existing comorbidity or as a complication.
The following diagnostic codes were used to determine the presence of a particular comorbidity:
Acute myocardial infarction: 4109, 41091, 41092, 41000, 41002, 41011, 41011, 41071, 41090, 412, 41092, 4104, 4102
Coronary Artery Disease: 414, 4111, 4142, 4148, 4149, 4168, 4169, 41189
Cancer: 140-209, 230-239, 1400-2090, 2300-2399, 14000-20900, 23000-23900 (exclude: 23875)
COPD: 490-496, 4900-4960, 49000-49600 (exclude: asthma)
Diabetes mellitus: 250, 2500-2509, 25000-25093
Hemiplegia: 3429
Hypertension: 4011, 4019, 4010, 4160
Liver disease: 571, 570
Stroke/Cerebrovascular Disease: 436, 43401, 43411, 43491, V1254, 43811, 4380
Ulcer disease: 707, 7070 – 7079, 70700-70725
Dementia: 290, 294, 331, 2900, 3310, 29410, 29420
Peripheral vascular disease: 443, 4439
Congestive heart failure: 428, 4280, 4281, 42822, 4820, 42823, 42830, 42831, 42832, 42833, 42840, 42842
Chronic kidney disease: 5859, 5889, 5851-5856, 58889
Surgeon Variables
The claims for each patient operation with CPT code 27130 were examined. For claims with an orthopedic surgeon as the specialty provider (provider code 207X00000X), the provider id was matched with surgeon information (name, practice etc.). Surgeon years of practice, fellowship, board certification and degree were then obtained from healthgrades.com. Surgeon volume was determined from the professional claims data file by searching the number of 27130 claims made by each service provider during 2011.
In the few cases where a surgeon id could not be matched to a 27130 billing, the surgeon was identified by matching non-surgeon providers who billed the patient, such as a nurse or physician’s assistant, to a given surgeon. Alternatively, billings within a two-week period prior to and after the surgery were examined to see if a surgeon provider specialty (207X00000X) was listed. If a single surgeon was listed, that surgeon was assigned to the patient.
Surgical technique/Surgeon survey
The surgical technique for operations was identified by information provided from a surgeon survey. Surgeons were queried on surgical techniques used for THR performed in 2011 when the diagnosis was one of the three used to define the sample. Surgeons who used more than one technique were asked to provide the percentage of THR performed using each approach. Of the 61 surgeons performing THR, a survey was distributed to 54. Two of three surgeons who left the area could not be contacted, five other surgeons could not be contacted for purposes of sending a survey. Forty-nine surgeons responded to the survey. Thirty-nine of the respondents reported using one surgical technique for all THR (POST-MI (27), LAT-MI (5), ANT-MS (5), direct anterior (1) and other (1)). Ten surveyed surgeons used a mix of two techniques (posterior/ MS anterolateral (6), posterior/direct anterior (2), posterior/modified lateral (2)).
The complete identification of surgical technique by operation from surveyed surgeons who used a single technique constitutes the complete case (CC) sample. The responses of mixed technique surgeons was used to extend the sample as described in the text.
Instruments
All distance variables were calculated as the driving distance in miles between two zip codes (patient-surgeon, patient-hospital) using Google Maps.