Measures

Information extracted from the EHR prior to surgery included age, sex, self-identified race, body mass index (BMI), tobacco use history, and comorbidity count. We also extracted surgical encounter information related to the TJA including preoperative pain rating (0–10 scale) within 30 days prior to surgery, the date of surgery, and the number of TJA or TJA-related surgeries performed within the data extraction period. These variables were selected based on having clinical relevance (e.g., preoperative pain intensity, number of surgeries) and/or prior association with TJA outcomes (e.g., BMI, age, tobacco use, comorbidity count, and sex). The length of time from first surgery to survey completion was selected to account for the variability in postoperative time.

Information from the survey responses included PROMIS short forms for physical function, pain interference, sleep disturbance, anxiety, and dyspnea. PROMIS measures have been validated for use in the general population [11, 12, 21, 22] as well as orthopedic surgery populations [14, 23–25]. Each PROMIS measure is standardized, such that a T score for each patient is given that has a mean score of 50 in the general US population and a standard deviation of 10 which enables comparison for clinical and research applications.

Information collected from the survey responses also included chronic pain status from the Graded Chronic Pain Scale-Revised which has been previously described in detail [26, 27]. Briefly, we used the Graded Chronic Pain Scale-Revised, which has 2 questions: (1) In the past 3 months, how often did you have pain? and (2) Over the past 3 months, how often did pain limit your life or work activities? These responses were combined with PROMIS pain interference score [21] to categorize patients into 4 pain states: (1) chronic pain absent (Question 1 response: “never,” or “some” days in pain); (2) mild chronic pain (Question 1 response: “most” or “every” day(s) in pain; Question 2 response: “never” or “some” days with impact on daily activities; Pain Interference Score < 50th percentile); (3) bothersome chronic pain (Question 1 response: “most” or “every” day(s) in pain; Question 2 response: “never” or “some” days with impact on daily activities; Pain Interference Score ≥ 50th percentile); and (4) high impact chronic pain (Question 1 response: “most” or “every” day(s) in pain; Question 2 response: “most” or “every” day(s) with impact on daily activities). This scale has been used recently to categorize rates of chronic pain in different populations, including the general population [28], those enrolled in a health care plan [26], those with spinal pain [29], and following TJA [30].