Long-Term Complications of Full-Neural-Interface Cybernetic Limbs: A Five-Year Retrospective Cohort of Civilian Recipients at the Coruscant Medical Academy, 30–22 BBY

1. Introduction

A cybernetic replacement limb with a full neural interface is standard care in the Republic for civilians who lose an arm or leg, and recipients commonly regain grip, gait and fine manipulation that approach those of the natural limb. Because the device couples directly to residual peripheral nerves, it also carries problems that a passive prosthesis does not. Interface pain, persistent phantom sensation, gradual loss of calibration between intended and executed movement, and inflammatory reactions at the coupling site have each been described in case series (Harrow & Quell, 34 BBY; Oddane, 29 BBY). How often they occur over years of ordinary use, and in whom, remains an open question.

Much of the existing literature follows patients for a few months after fitting, when attendance at the fitting clinic is routine. Beyond that window, recipients return only when something goes wrong, and those later presentations are scattered across outpatient files. Published calibration protocols assume that drift is roughly constant over the life of the device (Solvane, 31 BBY), an assumption that has never been tested against long-term records.

We therefore assembled a five-year retrospective cohort of civilian recipients fitted at a single Coruscant centre. Our aims were to estimate the frequency of four complications, to describe how calibration drift changes with time since fitting, to test whether proximal fittings carry more interface pain than distal ones, and to report, with appropriate caution, how non-human recipients in the same cohort fared.

2. Methods

Setting and cohort. The Department of Prosthetic and Cybernetic Medicine at the Coruscant Medical Academy fits full-neural-interface limbs for civilian patients referred from across the Core. We identified every patient fitted between 30 and 27 BBY from the department's fitting registers and extracted outcomes from outpatient files through the close of follow-up in 22 BBY, when the last patients fitted reached five years. Extraction began in 23 BBY, and the analysis reported here was completed in 22 BBY. Of 231 patients fitted in the period, 11 with bilateral fittings and 6 with incomplete fitting records were excluded, leaving 214. The Academy review board approved the study under its standards for secondary use of patient registries, with records coded before analysis (Coruscant Medical Academy, 27 BBY).

Exposures. Fitting level was classified as proximal (above-elbow or above-knee; n = 86) or distal (below-elbow, wrist, below-knee or ankle; n = 128). Covariates were age at fitting, limb (upper or lower), species (human or non-human), cause of limb loss, and an interval of more than six standard months between limb loss and fitting. Before fitting, all residual limbs had been managed to the Republic standard of bacta immersion and, where soft-tissue cover was deficient, synthflesh grafting (Republic Medical Corps, 32 BBY).

Outcomes. An interface pain episode was a clinical contact for pain localised to the coupling site or referred along an interfaced nerve. A calibration-drift event was a documented divergence between intended and executed movement that required a recalibration session under the department's protocol (Solvane, 31 BBY). Phantom sensation was any recorded report of sensation attributed to the absent natural limb, painful or not. An interface-site tissue reaction was erythema, granulation or skin breakdown at the coupling site that required treatment. Sensory feedback in upper-limb recipients was assessed at the twelve-month review as the two-point discrimination threshold on the prosthetic index pad, measured with blunt-tipped callipers under an adaptive staircase of one- and two-point trials; the contralateral natural fingertip served as reference. A companion study in this journal has since applied the same kind of calliper staircase to the Twi'lek lekku.

Analysis. Pain episodes and drift events were modelled by negative binomial regression with log person-years as offset, because both counts were overdispersed (dispersion parameter 1.3 for pain and 0.6 for drift). Change in drift over time was modelled with year of follow-up as a continuous term, and a quadratic term was tested by likelihood-ratio test. Phantom sensation was analysed by logistic regression. Time to first tissue reaction was estimated by the Kaplan–Meier method and modelled by proportional-hazards regression. All adjusted models included the covariates listed above, and confidence intervals used robust variance estimates (Kesh, 26 BBY). Patients who relocated off-world or died of unrelated causes were censored at last contact.

3. Results

The 214 patients had a median age at fitting of 41 years (interquartile range 29–53). Limb loss followed industrial accidents in 97, transport collisions in 61, vascular or infective disease in 34, and congenital absence in 22. There were 131 upper-limb and 83 lower-limb fittings; 88 patients waited more than six months for fitting. The cohort comprised 178 humans, 15 Twi'leks, 11 Zabraks and 10 Rodians. Of the 23 patients who did not complete five years, 17 relocated off-world and 6 died of causes unrelated to the prosthesis. Total follow-up was 1,002.5 person-years.

Interface pain was the commonest complication: 988 episodes were recorded, a rate of 0.99 per person-year, and 154 patients (72.0%) had at least one. Most episodes were mild; 919 (93.0%) resolved with recalibration, analgesia or socket adjustment alone, and 7 patients underwent surgical revision of the interface. The rate was 1.41 per person-year after proximal fittings and 0.70 after distal fittings, an adjusted incidence rate ratio (IRR) of 1.87 (95% CI 1.41–2.48; p < .001; Table 1).

Calibration drift followed a different course. Of 412 drift events (0.41 per person-year), 181 (43.9%) occurred in the first year. Annual rates were 0.86, 0.47, 0.31, 0.22 and 0.16 per person-year across years one to five, an adjusted IRR of 0.65 per additional year (0.59–0.72; p < .001). A quadratic term improved fit (likelihood-ratio χ²(1) = 4.3, p = .04), indicating that the decline was steepest early and flattened from the third year. Drift did not differ by fitting level (IRR 1.12, 0.91–1.38; p = .29).

Phantom sensation was reported by 131 patients (61.2%). It was not significantly associated with proximal fitting (odds ratio 1.44, 0.80–2.59; p = .22) but was more likely when fitting was delayed beyond six months (OR 2.10, 1.13–3.90; p = .02). Two-point thresholds were available for 124 of the 131 upper-limb recipients. On the prosthetic index pad the median was 9 mm (IQR 7–13), against 3 mm (2–4) on the contralateral fingertip. Coarser thresholds showed a weak, non-significant association with phantom sensation (OR 1.18 per 2 mm, 0.98–1.42; p = .08).

Thirty-eight patients (17.8%) had at least one interface-site tissue reaction, a Kaplan–Meier five-year cumulative incidence of 18.6% (95% CI 13.7–24.9). Lower-limb fittings carried a higher hazard than upper-limb fittings (22 of 83, 26.5%, against 16 of 131, 12.2%; adjusted hazard ratio 2.20, 1.15–4.21; p = .02), whereas fitting level did not (HR 1.31, 0.69–2.49; p = .41). Almost all reactions settled with topical bacta or a short immersion; 3 patients needed surgical revision with synthflesh coverage.

Among the 36 non-human patients the picture was mixed, and every estimate is imprecise. Rodians had the highest pain rate (64 episodes in 46.1 person-years, 1.39 per person-year; adjusted IRR against humans 1.38, 0.82–2.31; p = .22) and a higher drift rate (28 events, 0.61 per person-year; adjusted IRR against humans 1.47, 0.93–2.32; p = .10). Zabraks had the lowest pain rate (31 episodes in 52.9 person-years, 0.59) and no tissue reactions in 11 patients. Twi'leks had 58 episodes in 70.9 person-years (0.82) and 2 tissue reactions in 15 patients. No lekku prosthesis was fitted during the study period.

4. Discussion

Over five years, most recipients of a full-neural-interface limb experienced at least one complication, yet very few required more than an outpatient visit. Interface pain was frequent and almost always settled with conservative measures. Drift requiring recalibration was concentrated in the first year and fell steeply before levelling off. Tissue reactions affected fewer than one patient in five and responded to bacta in nearly every case. For a device that couples to living nerve, this is a reassuring profile.

The course of calibration drift is the finding with the clearest practical consequence. Current protocols schedule recalibration at fixed intervals throughout the life of the limb, on the premise that drift accrues at a constant rate (Solvane, 31 BBY). Our data do not support that premise. More than two-fifths of all drift events fell in the first year, and the rate had dropped to about a quarter of its initial level by the fourth year and below a fifth by the fifth. One plausible explanation is that the recipient's nervous system and the interface converge on a stable mapping during the first months of use; once that mapping settles, it needs little correction. That account is our hypothesis and was not tested here. Whatever the mechanism, front-loading review in the first year and extending intervals thereafter would match clinic resources to need.

Proximal fittings nearly doubled the rate of interface pain without increasing drift or tissue reactions. Above-elbow and above-knee devices couple to larger nerve trunks carrying more mixed motor and sensory fibres, and the coupling must drive more joints. Either feature could plausibly increase the chance of aberrant sensory signalling at the site. Earlier case series reported the same direction of effect in far smaller groups (Harrow & Quell, 34 BBY). We would advise that proximal recipients be told to expect more pain episodes and that their pain be reviewed at every recalibration visit.

Phantom sensation behaved differently from interface pain. It was tied to delay before fitting and not to fitting level, which fits the view that the central representation of the absent limb persists or reorganises during the interval without input (Oddane, 29 BBY). Sensory feedback through the prosthetic pad remained roughly three times coarser than the natural fingertip, and coarser feedback showed only a weak trend towards more phantom sensation. Early fitting may matter more than feedback resolution, though our study was not designed to separate the two.

The higher hazard of tissue reactions after lower-limb fitting most likely reflects load. A weight-bearing socket transmits force and shear across the coupling site with every step, and skin breakdown under such conditions has been described across several species (Venn-Adaro, 28 BBY). Socket design, not the neural interface, may be the more promising target for prevention.

Results for non-human recipients are reported for completeness, not for inference. Rodian recipients showed higher point estimates for pain and drift, and Zabrak recipients fared comparatively well, but none of these differences reached significance, and with ten to fifteen patients per species a single outlier can shape a rate. Earlier work at Lessu raised similar questions about interface tolerance in Rodian and Twi'lek recipients (Talvorra, 27 BBY). The absence of lekku prostheses in this cohort deserves comment. Lekku replacement lies outside the scope of this study, yet the companion study of 60 healthy adults in Lessu, published in this journal, found a mean distal lekku threshold of 2.3 mm, finer than the 2.9 mm recorded at the index fingertip of the same volunteers, and linked distal acuity to the comprehension of fine lekku signs. Our prosthetic pads achieved a median of 9 mm, three times coarser than the natural fingertip and roughly four times coarser than the distal lekku. A sensory-feedback lekku that restored communicative function would therefore be a considerably harder target than a hand.

5. Limitations

The design is retrospective, and outcomes depend on what clinicians chose to record. Pain or drift managed at home would not appear in the files, so these are rates of clinical presentation, not occurrence. Recording practice may also have changed across the eight years spanned by fittings and follow-up.

All patients were fitted at one Core centre using the department's own protocols. Recipients elsewhere, particularly in the Outer Rim where follow-up is less accessible, may do worse. Patients who relocated off-world were censored, and if they left because the device was troubling them, complication rates are underestimated.

The non-human subgroups are small and heterogeneous, and species-specific norms for pain reporting and two-point thresholds are poorly established. Species was entered as a single human or non-human term in the adjusted models, which cannot capture differences among the three non-human species. They are a prompt for multi-centre work with centres that treat larger non-human populations, not estimates to guide individual care.