AI-assisted 3D preoperative planning in primary total hip arthroplasty for secondary hip osteoarthritis: etiology-specific perioperative burden and early recovery in DDH versus post-septic sequelae.

BACKGROUND

Computed tomography (CT)-based, artificial intelligence (AI)-assisted three-dimensional (3D) preoperative planning is increasingly used to standardize component sizing and positioning in total hip arthroplasty (THA). In complex secondary hip osteoarthritis (HOA), perioperative burden and early recovery are particularly relevant to older or medically complex patients, yet it remains unclear whether etiology still shapes these outcomes under a unified planning pathway and single implant system.

METHODS

This single-center retrospective cohort analyzed 95 consecutive cementless primary THAs for developmental dysplasia of the hip-associated HOA (DDH-HOA, n = 52) or post-septic HOA (n = 43), all performed by one senior surgeon using the same CT-based AI-assisted 3D planning pathway and one cementless implant system. The primary endpoint was the Harris Hip Score (HHS) at 24 months; secondary outcomes included operative time, intraoperative blood loss (IBL), femoral osteotomy, planning-to-implant concordance, radiographic targets, patient-reported outcome measures (PROMs), and complications.

RESULTS

Planning performance was comparable (cup match 59.62% vs. 65.12%; stem match 61.54% vs. 67.44%), with similar Lewinnek/Callanan safe-zone attainment and a median leg-length discrepancy (LLD) of 1 mm in both groups. Intraoperative blood loss was lower in DDH-HOA than in post-septic HOA (212.69 ± 62.31 vs. 243.72 ± 58.64 mL, P = 0.015), while operative time was comparable (72.31 ± 14.11 vs. 77.91 ± 14.56 min, P = 0.061) and femoral osteotomy rates did not differ. At 1 month, DDH-HOA showed better pain and function [visual analogue scale (VAS) P = 0.035; HHS P < 0.001; Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) P = 0.004], whereas by 24 months HHS was comparable (P = 0.062). Two sciatic nerve injuries occurred in post-septic HOA; no reoperation or revision was recorded through 24 months.

CONCLUSION

Under a unified CT-based AI-assisted 3D planning pathway and single cementless system, sizing concordance and radiographic target attainment were similarly high across etiologies. By 24 months, PROMs were broadly comparable between groups. Post-septic sequelae were associated with greater intraoperative blood loss and less favorable pain and function at 1 month, supporting etiology-informed counseling and targeted perioperative risk mitigation.