Study design and patient selection
The study was designed following the COnsensus-based Standards for the selection of health status Measurement INstruments (COSMIN) guidelines for studies on measurement properties and specifically those guidelines regarding construct validity and responsiveness [20]. The study meets COSMIN criteria for these psychometric parameters, though a larger patient population (> 50) would have been preferred. Other psychometric parameters of the PROMs included in the study, including internal consistency and content, structural, and cross-cultural validity were outside the scope of this study but have been reported by previous investigators.
After obtaining approval from the Institutional Review Board, a retrospective analysis was performed of all patients who underwent rTSA for proximal humerus fractures at two American College of Surgeons Level I trauma centers from January 2003 to June 2017. Using our institutions’ Research Patient Data Registry (RPDR), patients were identified using the International Classification of Disease (ICD-9 and ICD-10) codes for the diagnosis of proximal humerus fracture and Current Procedural Terminology (CPT) code for shoulder arthroplasty. Five hundred seventy-six patients were identified from the initial screen. Inclusion criteria were patients 18 years of age or older who underwent rTSA for a proximal humerus fracture between January 1, 2003 and August 1, 2018 and had a minimum of 3 years postoperative follow-up. Exclusion criteria included diagnoses other than proximal humerus fracture, procedures other than rTSA, surgeries performed at an outside institution, and patients without documented postoperative follow-up.
We excluded 264 patients for a diagnosis other than proximal humerus fracture, 74 patients who did not undergo rTSA, and 14 patients who had rTSA for fracture performed at an outside institution. The remaining 224 patients were contacted by telephone at a minimum of 3 years following rTSA for our outcome measures; 71 patients were further excluded for inadequate follow-up, of which 26 were deceased and 55 were unavailable. Furthermore, 113 patients were excluded for an incomplete set of outcome measures at a minimum of 3 years following rTSA, 112 because they were not sent all of the functional outcomes surveys due to their inclusion in different research protocols, and 1 who was contacted for all scores but returned EQ-5D, PROMIS, and DASH but not ASES, SPADI, or SSV. The final cohort included 30 patients (30 shoulders) with proximal humerus fractures treated with rTSA with a complete set of functional outcome measures at least 3 years of follow-up.
Outcome measures
The outcome measures in this study included global health outcome measures, an upper extremity-specific outcome measure, and shoulder-specific outcome measures. Global health outcome measures included the Patient-Reported Outcomes Measurement Information System (PROMIS) Global Health Physical Score Short Form 10b, EuroQol Five-Dimension Score (EQ-5D), and EuroQol Visual Analogue Scale (EQ-VAS). The upper extremity-specific outcome measure was the Disabilities of the Arm, Shoulder and Hand (DASH) score. Shoulder-specific outcome measures included the Shoulder Subjective Value (SSV), American Shoulder and Elbow Surgeons (ASES), and Shoulder Pain and Disability Index (SPADI). All outcome measures were collected through review of the medical records and telephone contact at a minimum of 3 years following rTSA.
The PROMIS Physical Function instrument is a 10-item questionnaire that evaluates physical, mental, and social aspects of health, and all items are evaluated on a 5-point Likert scale [10]. The questionnaire has Physical and Mental Health subscales. The Physical Health subscale score, which was utilized in our study, ranges from 0 to 100 with a mean score of 50 representing the norm for the US general population and a standard deviation of 10. Higher scores indicate better health [16, 21].
The EQ-5D is a health-related quality of life measure with 5 domains: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression [21] There are three options for each response. A score of 1.0 represents perfect health while scores less than zero indicate health status worse than death [21]. The EQ-VAS is a separate quantitative self-assessment of health which has patients evaluate their own health on a visual analogue scale from 0 to 100 where 0 is worst and 100 is best [21].
The DASH score is a 30-item upper-extremity-specific outcome measure which is intended to measure shoulder, elbow, hand, and wrist function in one combined metric [22]. It does not discriminate between the affected and non-affected upper extremity. The DASH assesses multiple domains including physical function, symptoms, and social/psychological function [13]. It has been shown to correlate with shoulder-specific measures, [23]. The DASH is scored from 0 to 100, with lower scores representing better function. Normative values are 22 in females and 13 in males aged 70 to 79 [24]. The minimum clinically important difference (MCID) for shoulder conditions is approximately 10 [this MCID value, and those reported for other metrics below, is not intended to suggest that the MCID in our patient population is identical, but merely to give a sense of the scale on which clinically important differences have been previously reported to occur for other shoulder pathologies] [22].
The SSV is an entirely subjective shoulder-specific measure where the patient is asked to rate their shoulder on a scale from 0 to 100 with some version of the following question: “What is the overall percent value of your shoulder if a completely normal shoulder represents 100 %?” [25] The SSV is considered an adjunct to more complicated shoulder-specific scores and has been shown to be an easily administered, responsive, and valid measure of shoulder function [25].
The ASES is a shoulder-specific score which consists of 10 functional questions and a pain VAS score [22]. The total score ranges from 0 to 100 (higher is better) and is weighted 50 % for function and 50 % for pain. The MCID is considered to be 6.4 for various shoulder pathologies [22].
The SPADI is a self-assessment of symptoms and function of the shoulder designed for any disorder of the shoulder joint [26]. It consists of 13 items (5 for pain symptoms and 8 for disability) which are scored on a 0 to 100 point visual response scale. The domain scores are combined to produce a score from 0 to 100 where 0 is best and 100 is worst [27]. The MCID for the SPADI has been reported to be between 8 and 13.2 [27].
Patient demographics were collected through our institutional RPDR and review of the electronic medical records, including age, sex, hand dominance, body mass index (BMI), Charlson Comorbidity Index (CCI), diabetes mellitus, depression, osteoporosis, smoking status at presentation, and employment status. Injury-related variables included fracture type, presence of ipsilateral upper extremity fracture, open fracture, and associated nerve injury. Treatment-related variables included initial treatment (acute rTSA versus nonoperative treatment), tuberosity repair, complications, and reoperation. The operative techniques and rehabilitation protocols were at the discretion of the treating surgeon.
Statistical analysis
Descriptive statistics were calculated for the study cohort. Measurement of correlations between outcome measures were performed using the Spearman correlation coefficient. To study floor and ceiling effects, the proportion of patients reporting the minimum and maximum scores of each outcome measure, respectively, were determined. Paired comparisons of the proportion of patients at the floor and ceiling of outcome measures were performed using McNemar’s test. The standard significance criterion of α = 0.05 was employed. A convenience sample was used.