Funding has no effect on clinical outcomes of total joint arthroplasty emerging technologies: a systematic review of bibliometrics and conflicts of interest

Abstract

Background: The use of new total joint arthroplasty technologies, including patient-specific implants/instrumentation (PSI), computer-assisted (CA), and robotic-assisted (RA) techniques, is increasing. There is an ongoing debate regarding the value provided and potential concerns about conflicts of interest (COI).

Methods: PRISMA guidelines were followed. PubMed, MEDLINE, and Web of Science databases were searched for total hip and knee arthroplasties, unicompartmental knee arthroplasties (UKA), PSI, CA, and RA. Bibliometric data, financial COI, clinical/functional scores, and patient-reported outcomes were assessed.

Results: Eighty-seven studies were evaluated, with 35 (40.2%) including at least one author reporting COI, and 13 (14.9%) disclosing industry funding. COI and industry funding had no significant effects on outcomes (P = 0.682, P = 0.447), and there were no significant effects of conflicts or funding on level of evidence (P = 0.508, P = 0.826). Studies in which author(s) disclosed COI had significantly higher relative citation ratio (RCR) and impact factor (IF) than those without (P < 0.001, P = 0.032). Subanalysis demonstrated RA and PSI studies were more likely to report COI or industry funding (P = 0.045). RA (OR = 6.31, 95% CI: 1.61-24.68) and UKA (OR = 9.14, 95% CI: 1.43-58.53) had higher odds of reporting favorable outcomes than PSI.

Conclusions: Author COIs (about 40%) may be lower than previously reported in orthopedic technologies/techniques reviews. Studies utilizing RA and PSI were more likely to report COI, while RA and UKA studies were more likely to report favorable outcomes than PSI. No statistically significant association between the presence of COIs and/or industry funding and the frequency of favorable outcomes or study level of evidence was found.

Level of evidence: Level V Systematic Review.

Keywords: Computer-navigated; Conflict of interest; Funding; Patient-specific implant; Robotic-assisted; Total joint arthroplasty.

Fig. 1

PRISMA flow diagram for emerging TJA technologies

Fig. 2

Proportions of studies reporting favorable outcomes (black bar segments) or equivocal/unfavorable outcomes (gray bar segments). Studies were compared for (A) any reported conflicts of interest vs. no reported conflicts of interest, and (B) any reported industry funding vs. no reported industry funding. Neither comparison found significant between-groups differences

Fig. 3

Proportions of studies reporting Level I evidence (black bar segments), Level II evidence (light gray bar segments) or Level III evidence (dark gray bar segments). Studies were compared for (A) any reported conflicts of interest vs. no reported conflicts of interest, and (B) any reported industry funding vs. no reported industry funding. Neither comparison found significant between-groups differences

Fig. 4

Means (circles) and their 95% confidence intervals (error bars) for Relative Citation Ratio (RCR) and Impact Factor (IF). Studies were compared for any reported conflicts of interest vs. no reported conflicts of interest (A: RCR; C: IF), and any reported industry funding vs. no reported industry funding (B: RCR; D: IF). Studies reporting conflicts of interest tended to have significantly higher RCR and IF than those reporting no conflicts. Funded and unfunded studies did not differ significantly for RCR or IF

Fig. 5

A Proportions of studies reporting favorable outcomes (black bar segments) or equivocal/unfavorable outcomes (gray bar segments), with comparison between studies published in open access vs. traditional journals. Outcome proportions were not significantly different between journal types. B Proportions of studies reporting Level I evidence (black bar segments), Level II evidence (light gray bar segments) or Level III evidence (dark gray bar segments), with comparison between studies published in open access vs. traditional journals. Level of evidence distributions were not significantly different between journal types