Cumulative Bacta Immersion Time, Timing of First Immersion, and Wound Closure after Penetrating and Blast Injury: A Retrospective Cohort of 357 Patients from Coruscant Treatment Centres, 25–21 BBY

1. Introduction

Bacta immersion and synthflesh are standard medicine across the Republic and well beyond its borders, and the tank is the usual setting for recovery from serious wounds. Force healing exists but is rare and forms no part of routine practice. Given how widely immersion is used, it is surprising how little the clinical literature says about how long a patient should remain in the tank. The published series we could find describe outcomes after immersion without recording its duration in a form that can be analysed (Aldrevik, 27 BBY), and the standing circulars of the Republic Medical Corps leave the length of immersion to the attending physician (Republic Medical Corps, 28 BBY). The consequence is that tank time is allocated by local habit, by bed pressure and by the judgement of individual clinicians.

This paper is written from a vantage of 21 BBY, and the pressure on tank capacity is not an abstract concern. Casualty intake at Coruscant treatment centres has risen since 22 BBY (Republic Health Service, 22 BBY), and every hour of tank time given to one patient is an hour withheld from another. If wound closure improves steadily with each additional hour, the case for long immersion is strong. If the benefit is concentrated early and tails off, the allocation problem looks different. We have no view on which is true. The composition of the tank medium was not examined, and we do not offer a mechanism here. Any account of why a response might saturate would be a hypothesis about tissue biology that these records cannot test.

Two clinical questions are entangled in the existing literature. The first is the shape of the relation between cumulative immersion and time to wound closure. The second is the timing of the first immersion, which earlier work has linked to outcome after penetrating trauma (Sorrenth & Tavendar, 26 BBY). A patient who reaches a tank late and stays a long time may differ from one who reaches it early and stays briefly, and the two exposures cannot be assessed apart. We therefore pursued three aims: to describe the relation of cumulative immersion time to the rate of wound closure; to test whether the relation departs from a straight line, specifically whether returns diminish beyond some number of hours; and to estimate, in the same model, the association of delay to first immersion with closure. A secondary aim was to examine whether the same pattern holds for functional recovery at 90 days after injury.

2. Methods

Design and setting. We conducted a retrospective cohort study using the trauma registry of the Republic Health Service, drawing on records from its Coruscant treatment centres for injuries sustained between 25 BBY and early 21 BBY, so that 90 days of follow-up fell before analysis. Military casualties treated in Grand Army facilities after 22 BBY are not held in this registry and are not represented. Registry clerks recorded admission findings, the time of each entry to and exit from a tank, and the date on which the treating physician recorded full closure of the index wound. All analysis was completed in 21 BBY. Records were held under coded identifiers, and no individual patient is identifiable from anything reported here.

Cohort. We screened 412 records of patients admitted with penetrating injury (blaster bolt, blade or fragment) or blast injury. Thirty-four were excluded because the immersion log was incomplete or the index wound could not be identified, leaving 378. A further 21 were excluded by the landmark rule described below: nine died and 12 reached full closure within ten days of injury. The analytic cohort was 357 patients. Of these, 203 (56.9%) had penetrating injury and 154 (43.1%) blast injury. Humans numbered 263 (73.7%) and members of other species 94 (26.3%). Species other than human were pooled into one group because no single species contributed enough patients to be analysed separately (comparative estimates of repair rate are fragile at small numbers; Ferrowyn, 29 BBY). Median age was 39 standard years (interquartile range 32–46; range 19–64).

Exposure and landmark. Immersion is cumulative and stops when the patient is judged to have healed, so a naive comparison of total tank time with time to closure is biased: patients who close quickly leave the tank early. We addressed this with a landmark analysis (Brannock, 31 BBY). Exposure was the total hours of immersion accrued in the first ten days after injury, a window fixed in advance for every patient, and follow-up for closure began at day ten. Patients who had closed or died before day ten were excluded, since their exposure window was truncated by the outcome. Days to closure are counted from the landmark, so ten days must be added to obtain days from injury. Delay to first immersion was recorded in hours from injury to first tank entry.

Outcomes. The primary outcome was days from the landmark to full closure of the index wound, censored at transfer out of the treatment centres, loss to follow-up, death, or day 80 after the landmark (day 90 after injury). The secondary outcome was the Coruscant Functional Recovery Scale (CFRS), assessed at 90 days after injury. The CFRS is our own instrument, adapted from an earlier outpatient scale (Vessary, 24 BBY). It sums ten everyday tasks covering grip, load-bearing, mobility and self-care, each scored from 0 to 10, and the task list was adapted by the assessor for the anatomy of non-human patients. The CFRS has not been validated across species. It was recorded for 328 of the 357 patients (91.9%).

Covariates. Injury severity was measured with the Coruscant Trauma Severity Index (CTSI), also our own construction, following the general logic of an earlier regional score (Kessarin, 25 BBY). The admitting physician scores six body regions from 0 to 5, and the CTSI is the sum of the squares of the three highest regional scores, giving a range from 0 to 75. Its distribution here was mean 22.8 (SD 8.9; range 4–50). Other covariates were age, species group (human or other), injury type (penetrating or blast) and delay to first immersion.

Analysis. Time to closure was analysed with a Cox proportional hazards model, with ties handled by the Breslow method. A hazard ratio above 1 means that closure occurs faster. Immersion time entered as a linear spline in hours, with two slopes joined at a knot. The knot was chosen by profile partial likelihood over a grid from 24 to 84 hours in steps of 6, and the profile is reported. We tested the difference between the two slopes on the log-hazard scale and report it as the ratio of the per-10-hour hazard ratios above and below the knot. Because the knot is estimated, confidence intervals for the slopes and the spline test are optimistic to an unknown degree, and we report them as nominal. The proportional hazards assumption was examined by fitting the model separately to the first 20 days of follow-up and to patients still open at day 20. Whether the immersion effect differed by species group was examined with an interaction between species group and the lower slope, and the species-specific hazard ratios are taken from that joint model. Calendar period within the window was not recorded in the extract and was not adjusted for. The CFRS was analysed by linear regression with the same covariates and the same knot. Confidence intervals are Wald intervals. The approach to threshold estimation follows Hallowick (32 BBY). All calculations were run on registry extracts held within the Republic Health Service.

3. Results

Of the 357 patients, 338 (94.7%) reached full closure during follow-up and 19 were censored: eight died with the wound still open, seven were transferred out of the treatment centres, three were lost to follow-up, and one remained open at the end of the window. The Kaplan–Meier median time to closure was 20 days after the landmark, which is 30 days after injury. Immersion in the first ten days had a median of 35 hours (interquartile range 25–50; range 9–203). Delay to first immersion had a median of 8 hours (interquartile range 4–15; range about 1–65). Table 1 groups patients by immersion time. Median time to closure fell from 23 days in the group with under 30 hours of immersion to 19 days at 30 to 47 hours and 16 days at 48 to 71 hours, then did not fall further in the crude data; the group at 72 hours or more had a median of 17 days. The four groups were similar in mean CTSI and in the share of blast injuries, and the very long immersions were few. The crude fall from 19 to 16 days between the middle groups is not contradicted by the flat adjusted slope above 30 hours. The adjusted model attributes part of that gap to delay, severity and injury mix, and the gap is compatible with the upper limit of 1.07 per 10 hours.

The adjusted model located the knot at 30 hours. The profile partial log-likelihood was highest there, and only the 30-hour and 36-hour grid points lay within 1.92 units of the maximum, the difference expected at the 95% level for a single parameter. The 24-hour point lay outside that range. The knot at 30 hours is therefore a best estimate with an interval of roughly 30 to 36 hours on this coarse grid, not a fixed threshold. The spline improved the fit over a single straight line by a likelihood-ratio statistic of 13.2 with one degree of freedom, a nominal value that overstates the evidence because the knot was searched for.

Below the knot, each additional 10 hours of immersion raised the rate of closure by a factor of 1.71 (95% CI 1.30–2.23; p < .001). Above the knot the factor was 1.01 (0.95–1.07; p = .80). The ratio of the upper to the lower per-10-hour hazard ratio was 0.59 (0.44–0.79; nominal p < .001). Expressed as a contrast, a patient with 30 hours of immersion had a closure rate 2.23 times (1.49–3.33) that of a patient with 15 hours, all else equal, whereas a patient with 50 hours had a rate 1.02 times (0.90–1.14) that of a patient with 30. Table 2 gives all coefficients.

Delay to first immersion carried an independent association. Each additional 6 hours before first entry to a tank was associated with a closure rate 0.91 times as fast (0.85–0.97; p = .007). Between a first immersion at 6 hours and one at 24 hours, the closure rate ratio was 0.76 (0.62–0.93), computed from the unrounded coefficients. Greater injury severity was associated with slower closure, at 0.88 per 10 CTSI points (0.78–0.997; p = .046), a borderline result. Age per decade (0.92; 0.83–1.02; p = .11), species group (0.96 for other species against humans; 0.75–1.23; p = .74) and injury type (0.96 for blast against penetrating; 0.77–1.20; p = .71) showed no clear association.

Two checks bore on the robustness of the pattern. In a joint model with a species interaction, the lower-slope hazard ratio was 1.69 (1.26–2.27) among the 263 human patients and 1.76 (1.07–2.90) among the 94 patients of other species. The ratio of the two was 1.04 (0.60–1.80; p = .89). With only 94 patients in the second group, that test has little power to detect a difference, and the wide interval for the other-species estimate says as much. Splitting follow-up showed a weaker lower-slope association in the first 20 days (1.61; 1.09–2.38; 176 closures) than afterwards (2.02; 1.39–2.95; 162 closures among 167 patients still open at day 20). The intervals overlap, but the point estimates differ, and the proportional hazards assumption is not perfectly met.

Functional recovery followed a different course (Table 3). Mean CFRS at 90 days was 73.0 (SD 13.0) among the 328 assessed patients. Below the knot, each additional 10 hours of immersion was associated with a CFRS higher by 7.47 points (95% CI 5.07–9.88). Above the knot it was higher by 1.32 points per 10 hours (0.82–1.83), about one sixth of the 7.47 points per 10 hours below it, and the interval did not include zero. Each additional 6 hours of delay to first immersion was associated with a score 1.35 points lower (−2.01 to −0.69), and each additional 10 CTSI points with a score 5.93 points lower (−7.14 to −4.72). Patients of other species scored 4.11 points lower than humans (−6.56 to −1.66; p = .001), a difference that may partly reflect the fit of the scale to non-human anatomy. The model explained 45% of the variance in CFRS (residual SD 9.7 points).

4. Discussion

In this cohort, the rate at which wounds closed rose steeply with cumulative immersion up to about 30 hours in the first ten days, and beyond that point the estimated slope was close to nil. The per-10-hour hazard ratio above the knot was 0.59 times that below it (95% CI 0.44–0.79), and the interval for the upper hazard ratio itself, 0.95 to 1.07, excludes any large gain from additional tank time. The practical reading is narrow. For patients still open at ten days, an hour of tank time beyond about 30 hours bought little additional speed of closure, which is a statement about time to closure in this registry and not a statement about the value of immersion as a whole.

Two features of the result deserve emphasis. First, the knot is an estimate: the profile favoured 30 hours, with 36 hours nearly as likely, and both the knot and its intervals are optimistic because the data chose it. The clinical message is a region, not a threshold. Second, the pattern did not carry over unchanged to function. Functional scores continued to rise beyond the knot, at about one sixth of the earlier rate (1.32 against 7.47 points per 10 hours), with an interval that nominally excludes zero, since the knot was searched for. If that association is real, then the early plateau in closure should not be read as a plateau in benefit. Longer immersion may improve outcomes that the closure date does not capture, such as the quality of the repaired tissue. We cannot say from these records whether this is so, and the functional scale is our own and has not been validated in non-human patients.

Timing of first immersion mattered, but less than duration in the lower range. A first immersion 18 hours later, at 24 hours rather than 6, was associated with a closure rate about three quarters as fast. By comparison, moving from 15 to 30 hours of immersion more than doubled the rate. These are contrasts across different ranges of two exposures, not a like-for-like ranking, and they should not be taken to mean that timing is unimportant. Delay is also shaped by circumstances outside the patient, such as distance from a treatment centre and bed availability. The direction of our estimate agrees with the earlier finding that early first immersion favours outcome after penetrating trauma (Sorrenth & Tavendar, 26 BBY).

The results for species groups are limited by number. The lower-slope estimate for patients of other species was similar to that for humans, with an interval for the ratio of the two, 0.60 to 1.80, too wide to exclude either equal or very different effects. We do not read the absence of a significant interaction as evidence that the immersion response is the same across species, and the comparative literature gives reason to expect differences in repair rate (Ferrowyn, 29 BBY). Studies that sample single non-human species in adequate numbers would be needed.

For allocation, we draw only a cautious implication. Where tank capacity is short, and where a patient has already received something near 30 hours in the first ten days, these records offer no evidence that further hours quicken closure, and the steeper lower slope indicates that closure was faster per additional hour among patients with fewer hours, although confounding by indication limits any causal reading. Whether any change in practice is warranted is a matter for the attending physicians and for the Republic Medical Corps circulars, which currently prescribe no duration. A prospective comparison of allocation policies would be the proper test, and we do not propose a protocol on the strength of a retrospective cohort.

5. Limitations

The study is observational, and the exposure was assigned by clinicians, not by chance. Patients who were doing well may have had their tank time reduced within the ten-day window, which would inflate the apparent benefit of immersion at low hours, and patients who were doing badly may have been kept longer, which would work in the opposite direction. The landmark design removes the bias arising from immersion stopping at closure, but it cannot remove confounding by indication within the window. The CTSI captures severity at admission and cannot reflect complications that appeared later.

Because the knot was chosen from the same data that were used to estimate the slopes, on a coarse grid, the reported intervals for the slopes and the improvement over a straight line are optimistic. The proportional hazards assumption was imperfectly met, with a stronger lower-slope association later in follow-up. We did not model a time-varying effect.

Registry records are clerical documents. Immersion hours came from entry and exit logs, and their precision is unknown, particularly for transfers between centres. Closure was recorded as a physician judgement without a common criterion, and that judgement may have varied between centres. We excluded 34 patients with unusable records and cannot rule out that missingness was related to outcome. Only patients treated in Republic Health Service centres are represented, so military casualties and patients treated elsewhere are outside these findings, and the results describe Coruscant treatment centres over 25–21 BBY, including the first months of the war, and should not be applied to other settings without evidence. The intake mix changed across the window as casualty numbers rose, and period was not adjusted for.

The CFRS and the CTSI are our own instruments and neither has been validated by an independent group. The CFRS task list was adapted by assessors for non-human anatomy, and the lower scores of the other-species group may reflect scale fit as much as recovery. Only 94 patients were of species other than human, and they were pooled, so no statement can be made about any individual species. Finally, this paper does not examine what in the tank medium acts on the wound. Any mechanism for the plateau, such as saturation of a tissue response, is a hypothesis for laboratory work.