A prospective single-centre cohort study was conducted examining the impact of patient reporting of AEs on PROMs following primary THR. Consecutive patients undergoing a primary THR between January 2012 and January 2013 were screened for eligibility prior to attending a preoperative assessment outpatient appointment. Patients meeting the inclusion criteria were invited to participate at this appointment. Patients were eligible for inclusion if they were due to undergo primary elective THR, were able to provide consent to participate and were able to understand and complete the English language questionnaires. Exclusion criteria were patients undergoing revision arthroplasty, patients who were unwilling or unable to provide consent and patients who were unable to understand or complete the questionnaires [Fig. 1]. Participation was voluntary, and patients provided verbal consent to participate. This study was part of a larger service evaluation project of THR and total knee replacement in this centre and ethical approval was not required according to the National Research Ethics Service guidelines. Participants were asked to complete questionnaires at five timepoints: pre-operatively and at 3, 6, 9 and 12 months post-operatively. Clinical follow-up and post-operative rehabilitation were determined by the treating surgeon and not affected by inclusion in this study. The pre-operative questionnaire assessed work status using the Work Productivity and Activity Impairment (WPAI), [17] hip function using the WOMAC [18] and health-related quality of life using the EuroQol-5D-3 L (EQ5D) [19]. The post-operative questionnaire included all the pre-operative questionnaires as well as the Self-Administered Patient Satisfaction scale (SAPS) [20] and an AE reporting questionnaire. The pre-operative questionnaire was administered and completed at a pre-operative assessment outpatient appointment, no more than 60 days prior to the date of the primary THR. Post-operative questionnaires were posted to participants, and if no response was received in 2 weeks, a reminder was sent.
The WPAI is an instrument to measure impairment in both paid and unpaid work. It can be adapted to measure absence and impairment due to a specific health problem [17]. A score displayed as a percentage may be calculated for work time missed, impairment while working, overall work impairment and activity impairment due to the health problem. The WPAI was completed by participants in employment. The percentage activity impairment represents the degree to which the hip due to be or that had been replaced affected regular activities.
WOMAC is a 24-item questionnaire designed to measure pain, function and stiffness in patients with osteoarthritis of the hip or knee [18]. It uses a 5-point Likert scale from 0 to 4 for each question (giving a total scale of 0–96) with higher scores indicating worse outcomes. A percentage score for each subscale of the WOMAC and a total was calculated, giving a score out of 100 for stiffness, pain, function and total WOMAC score.
EQ5D is a standardised, non-disease specific questionnaire for evaluating health-related quality of life in five dimensions, including mobility, self-care, usual activities, pain or discomfort and anxiety or depression [19]. The EQ5D index is derived from a standardised value set to provide a single value for health status, where one represents full health.
SAPS is a short questionnaire used to evaluate patient satisfaction with total hip and knee replacement [20]. Four items are scored on a 4-point Likert scale with responses from very dissatisfied to very satisfied. The scale score is the unweighted mean of the scores with 100 being most satisfied and 25 the least.
Study-specific questionnaires were used to evaluate AEs. Participants were asked if they had any problem with infection in the joint or wound, a dislocation of the THR or a fracture around the THR, after their surgery. Participants were asked to record details of these events and were asked if any other complications had occurred [Appendix 1]. The first report of an AE episode was counted, in the case of infection, if there was an infection that persisted and hence reported at multiple time points, it was only counted once.
For AEs reported by participants hospital systems and primary care records were interrogated to see if the AE could be confirmed. Hospital medical records were searched for hospital admissions and discharge documentation, outpatient clinic letters, operation records, radiology and microbiology reports. Local picture archiving and communications imaging systems were searched for evidence of dislocation or periprosthetic fracture. Primary care records were searched for documentation of an AE, post-operative general practice attendance or antibiotic prescribing at the time of the recorded AE.
Missing data were handled according to the user guide for each PROM. Data were analysed using STATA (version 13, StataCorp, College Station, TX, USA). Results are reported in three groups of participants:
1) No AE group: participants who reported no AE.
2) Reported group: participants who reported an AE which was not confirmed after searching medical records as described above.
3) Confirmed AE group: participants who reported an AE that was confirmed after medical record search.
Data were checked for normal distribution using a Shapiro-Wilks test and histogram plots. Where data were not normally distributed, central tendency is described as median value with inter-quartile range (IQR). PROMs data were compared among groups at 12 months using a generalised linear model(GLR) with robust standard errors (SE) in order to account for the non-normal distribution of data. The models were adjusted to account for age, gender and body mass index (BMI). The first model describes the relationship of the no AE group to the reported group and the confirmed group. The second model was constrained to assume no difference between the reported and confirmed groups and the Akaike information criteria (AIC) between the models were then compared using a likelihood ratio test (LRtest). A p-value of < 0.05 was considered statistically significant.