Study of Clinical Outcomes Associated With the Pulmonary Artery Catheter (PAC) in Cardiac Surgery Patients

A retrospective study will be conducted using prospectively collected hospital inpatient data over a duration of over 5 years (Jan. 1, 2010 - June 30, 2015) using a large US electronic health database (Cerner HealthFacts; Kansas City, MO). Patients who underwent a qualifying cardiac surgery (verified through use of selected valid International Classification of Diseases-9 procedure codes and/or Current Procedural Terminology [CPT] codes) will be included. Each patient's cohort designation will be defined based upon whether he/she did or did not receive a pulmonary artery catheter (PAC) for monitoring purposes. Propensity scores, which take into account patient and hospital demographics, patient comorbidities, surgical type (isolated coronary artery bypass graft [CABG], valve, aortic surgery, multi-procedures, other complex nonvalvular procedures and heart transplants), and pre-operative condition (via an adapted EuroSCORE II) will be utilized to "match" patients who received a PAC for monitoring purposes with those who did not, to form a matched study cohort. Clinical outcomes will be monitored through index visit discharge and up to 90 days post index visit discharge.

Inclusion criteria

• Patient undergoes a qualifying cardiac surgery between Jan 1, 2010 and January 1, 2015. If multiple qualifying surgeries are present, the first in database will be utilized
• Inpatient with a LOS of at least 48 hours
• Treated arm receives a PAC for monitoring purposes within admission date and qualifying cardiac surgical day plus one via specified ICD-9 or CPT-4 codes, or EHR recorded PAC readings

Exclusion criteria

• Cardiac surgery patients with age <18 years on index procedure date
• Non-treated arm derived from an institution which does not have database documented use of ICD-9 or CPT-4 PAC placement codes for monitoring purposes [Lessens the likelihood that the untreated arm is indeed treated by ensuring that the patient would likely be coded if he/she had a PAC in place for monitoring purposes]
• Patient record must have the demographics populated of age, gender, and race. ICD-9 diagnosis and procedure codes must be present in record for index visit, as well as medications administered over index visit
• Patient must be treated at a hospital which performs a minimum of 100 qualifying cardiac procedures per year