Factors Associated with Unsuccessful Revascularization Surgery in Traumatic Upper-Extremity Amputation

Abstract

Background:

Microsurgical emergency revascularization surgery for traumatic upper-extremity amputations demands high resource use. Injury details and patient characteristics influence the decision of whether to revascularize or revise an amputation involving the upper extremity. Our aim was to study associations between those factors and unsuccessful revascularization to provide information for clinical decision-making regarding amputation injuries.

Methods:

We studied all consecutive patients who had undergone an upper-extremity revascularization at Tampere University Hospital between 2009 and 2019. The primary outcome was the technical success or failure of the operation, which was defined as the survival or non-survival of the amputated tissue. Using logistic regression, we analyzed prognostic factors including age, sex, smoking status, diabetes mellitus, injury mechanism (cut, crush, or avulsion), extent of tissue loss before treatment (number of lost joints), and amputation type (total or subtotal).

Results:

A total of 282 patients (mean age, 47 years; 14% female; mostly White Caucasian) were included. The proportion of successful revascularizations (survival of all reconstructed tissue) was 76% (214 of 282). An avulsion injury mechanism (adjusted odds ratio [aOR], 5.9; 95% confidence interval [CI], 2.5 to 14.2), crush injury mechanism (aOR, 2.8; 95% CI, 1.1 to 7.0]), and total amputation type (aOR, 2.9; 95% CI, 1.5 to 5.8) were the prognostic factors that were associated with the highest risk of unsuccessful revascularizations. We found an S-shaped, nonlinear association between patient age and unsuccessful revascularizations and a U-shaped, nonlinear association between the amount of tissue loss before treatment and unsuccessful revascularizations. There was no evidence of an association between unsuccessful revascularizations and patient sex, smoking, or diabetes mellitus.

Conclusions:

Injury details were the most significant prognostic factors of an unsuccessful upper-extremity revascularization, while age was the only patient characteristic that was associated with this outcome. In particular, total amputation type and avulsion and crush injury mechanisms yielded a higher risk of unsuccessful revascularization. We recommend considering this information when making decisions regarding the treatment of upper-extremity amputation injuries.

Level of Evidence:

Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.

Many factors (age, sex, zone of injury, digit involved, smoking, ischemia time, accident type, and surgeon experience) have been associated with the success rate of revascularizations after traumatic upper-extremity amputations^1-13. However, the evidence regarding the importance of these individual factors is inconsistent, and there is a lack of consensus regarding the existing data^1-23. Meta-analyses of prognostic factors have included studies with small patient samples, different surgical techniques and indications for surgery, selected patient groups, and primarily single-digit operations^14,15,17-19. Thus, credible data on the factors associated with unsuccessful revascularizations is lacking.

Microsurgical emergency revascularization surgery for amputation injuries necessitates high surgical resource use and causes a high burden of care to the individual patient. Therefore, when making treatment decisions, it is important to know how patient characteristics and injury details may impact the probability of the technical success of a revascularization attempt. The aim of this exploratory study was to identify which factors may be associated with unsuccessful revascularization in a consecutive cohort of patients with traumatic upper-extremity amputations.

Methods

Study Design

In this exploratory, retrospective, consecutive cohort study, we included revascularizations performed in patients with a traumatic upper-extremity amputation injury between 2009 and 2019 at Tampere University Hospital, Tampere, Finland. In the study hospital, revascularization is attempted for all multi-finger amputations and isolated thumb amputations unless the amputated or injured part is too severely damaged. The inclusion criteria were upper-extremity amputation injury leading to a fracture and total loss of blood circulation in the thumb or in at least 2 digits (or a more proximal level) and treated with surgical revascularization (vascular repair or reconstruction) to restore perfusion in the injured tissue. We have used the term revascularization to cover all operations, i.e., both totally or subtotally amputated tissue. We used diagnostic and treatment codes from the hospital records to identify eligible patients. The only exclusion criteria were bilateral amputations and amputations in single fingers (except thumbs).

Study Setting, Surgical Technique, and Aftercare

Tampere University Hospital serves as a secondary and tertiary referral center, providing centralized emergency microsurgical care for a population of approximately 3 million people. In Finland, the majority of revascularization surgeries are performed in 2 university hospitals, 1 of which is a study center²⁴.Within the study period, specialized hand surgeons performed revascularization surgeries as emergency operations at the hand surgery unit. One or 2 hand surgeons were on-call all of the time, managing shifts lasting 16 to 24 hours. The surgeons at the unit have expertise ranging from 3 to 20 years in upper-extremity revascularization operations. An experienced specialist always accompanied less-experienced surgeons. Anesthesia during the procedure involved a continuous brachial plexus block, which was maintained for 5 days postoperatively. The standard operating technique included blood vessel and nerve anastomosis or reconstruction under microscope magnification. All surgeons reconstructed 2 veins for each repaired artery when feasible. Postoperative monitoring in the ward continued for 5 to 7 days²⁵.

Patients

During the research period, 282 patients met the inclusion criteria. Adults and children were included. We did not collect specific data about the patients’ race and ethnicity because that was considered nonessential information in this study. Most of the patients were White Caucasian, reflecting the Finnish population, which is predominantly White Caucasian, and 9% of the population were born in a foreign country. Patient-reported long-term outcomes for a large part of the patient cohort were previously reported, while in the present study, we focused only on assessing the factors associated with the technical failure of surgery^26,27.

Variables

The primary outcome was the success or failure of permanently restoring the blood supply in the injured tissue. The outcome was defined as (1) successful if all tissue survived and (2) unsuccessful if not all tissue survived, including complete and partial failure of revascularization resulting in later removal of necrotic tissue in a secondary operation. Revascularization was also considered successful in patients in whom subsequent revascularization(s) was required.

We collected patient characteristics (age, sex, smoking [yes/no], diabetes mellitus [yes/no], and injury details (injury mechanism, extent of tissue loss before treatment, and amputation type). Injury mechanisms were classified as follows: cut (by a sharp edge or cutting blade, for example, an axe or electric circular saw), crush (caused by blunt trauma, for example, compression under a heavy object or by a hydraulic machine), and avulsion (amputation caused by a pulling force, for example, grabbing and pulling by a moving machine or a table drill). To quantify the extent of tissue loss before treatment, we counted the number of lost joints. For instance, amputation of 2 digits at the proximal phalanx level equated to 4 lost joints. Amputation type was defined as subtotal (some tissue, e.g., a strip of skin or a tendon, remained in continuity) or total. The surgeon’s level of expertise was not included as a variable in this study.

Statistical Analysis

We present continuous outcomes as the mean and standard deviation (SD), or as the median and interquartile range (Q1 to Q3) if data were skewed. For the primary analysis, we used multiple logistic regression to assess the association of independent variables with unsuccessful revascularization. The final multivariable analysis included age, sex, smoking status, diabetes mellitus, injury type, extent of tissue loss before treatment, and amputation type, on the basis of clinical experience and results from previous studies^14,15,17-19. Adjusted odd ratios (aORs) are shown. The Wald test was used to assess the importance of the associations of independent variables with the dependent outcome variable. The Wald test also assessed nonlinear associations. Age and tissue loss before treatment were modeled as continuous data using a restricted cubic spline (RCS) function with 4 default knots for age (quantiles: 0.05, 0.35, 0.65, and 0.95, which correspond to 12, 43, 57, and 71 years, respectively)²⁸ and with 3 default knots for the extent of tissue loss before treatment (quantiles: 0.10, 0.50, and 0.90, which correspond to 1 [distal thumb amputation], 4 [2 digits or 1 complete digital ray amputation], and 14 lost joints [all or most digits amputated at the metacarpophalangeal level], respectively)²⁸. The RCS function captured the relationship between a continuous variable (age or tissue loss before treatment) and the dependent outcome in the regression model. Instead of imposing a linear or categorical structure, RCS allows the data to bend and adapt, accommodating potential nonlinear patterns in the continuous data²⁸. To describe these nonlinear associations, we present graphical plots of model predictions. Assumptions of the logistic regression were analyzed and estimated from the data. To address possible multicollinearity in the logistic regression model, a variance inflation factor (VIF) of <1.5 was considered acceptable for variables without RCS modeling. The receiver operating characteristic (ROC) curve describes associations between true-positive (sensitivity) and false-positive (1 – specificity) ratios, in this case indicating the regression model’s discriminatory ability to predict failure of the operation, and the area under the curve (AUC) value quantifies this discriminative ability from 0 to 1 (a value of 0.5 indicates predictive performance no better than chance)²⁹. The Nagelkerke R² coefficient is reported to describe the fit of the regression model³⁰. In a supplementary secondary analysis, we employed the t test or Wilcoxon rank-sum test for continuous variables and the chi-square test or Fisher exact test to compare categorical variables.

Ethics

The study permit was granted by Tampere University Hospital, Finland, the local legal research authority that reviews research permit applications. This study did not require Ethical Committee evaluation or informed consent (waived) by Tays Research Services because of the type of study. This report follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines³¹.

Results

The characteristics of the 282 patients in the study are presented in Table I. The proportion of successful revascularizations was 76% (214 of 282).

TABLE I.

Overview of Patient Characteristics

	Revascularization
	Successful (N = 214)	Unsuccessful (N = 68)	P Value
Age* (yr)	47 ± 17	47 ± 20	0.93
Female sex (no. [%])	31 (15%)	8 (12%)	0.57
Smoking (no. [%])	59 (28%)	19 (28%)	0.95
Diabetes (no. [%])	15 (7%)	5 (7%)	1.00
Injury mechanism (no. [%])			0.01
Cut	160 (75%)	40 (59%)
Crush	30 (14%)	11 (16%)
Avulsion	24 (11%)	17 (25%)
Amputation type: total (no. [%])	115 (54%)	47 (69%)	0.03
Tissue loss before treatment† (no. of lost joints)	4 [2-7]	4 [2-8]	0.76
No. of revascularized joints†	3 [1-6]	0 [0-2]	<0.001
Amputation level (no. [%])			0.25
Thumb only	57 (27%)	15 (22%)
2 digits	65 (30%)	19 (28%)
3 digits	35 (16%)	15 (22%)
4 digits	25 (12%)	11 (16%)
5 digits	9 (4%)	5 (7%)
Proximal to carpus	23 (11%)	3 (4%)

^*The values are given as the mean and standard deviation.

^†The values are given as the median, with the interquartile range in square brackets.

The prognostic factors demonstrating the highest risk of unsuccessful revascularizations were an avulsion injury mechanism, crush injury mechanism, and total amputation type (Fig. 1, Table II). We found a nonlinear, S-shaped association between the age variable and unsuccessful operations, where children and middle-aged patients (approximately 60 years of age) had a higher risk of failure, whereas young adults and the oldest patients had a lower risk of failure (Fig. 2). We found a nonlinear U-shaped association between tissue loss before treatment (number of lost joints) and unsuccessful revascularizations (Fig. 3), whereby the most distal and proximal injuries had the lowest risk of an unsuccessful operation compared with hand-area injuries (approximately 7 lost joints). The ROC curve (Fig. 4) and the AUC value of 0.73 demonstrated moderate discriminative ability of the multivariable model. The Nagelkerke R² coefficient value of 0.17 indicated moderate model fit.

Fig. 1.

Regression analysis of associations between independent variables and unsuccessful revascularization. The Wald test was used to assess the relative importance of each independent variable, with higher Wald test values indicating a stronger association with the likelihood of unsuccessful revascularization. On this plot, the Wald test assessed the overall significance of variables and tested for nonlinear associations. Χ² represents the Wald chi-square (square of [coefficient estimate/standard error] value) and Χ² – df represents the Wald chi-square value minus the degrees of freedom of the coefficient. Amputation type was defined as subtotal amputation (some tissue, for example, a strip of skin or a tendon, remains in continuity) or total amputation (all structures and tissue are separated). Injury mechanisms were classified as follows: cut (by a sharp edge or cutting blade), crush (caused by blunt trauma), and avulsion (amputation caused by a pulling force).

Fig. 2.

Nonlinear association between the age variable and unsuccessful revascularization, adjusted for sex, diabetes, smoking, tissue loss before treatment (number of lost joints), amputation type, and injury mechanism. The age variable was modeled using a restricted cubic spline function with 4 default knots (quantiles: 0.05, 0.35, 0.65, and 0.95, which correspond to 12, 43, 57, and 71 years, respectively). Higher log odds values are indicative of an increased probability of failure. The gray area indicates the 95% confidence interval.

Fig. 3.

Nonlinear association between the variable of tissue loss before treatment (number of lost joints) and unsuccessful revascularization, adjusted for sex, diabetes, smoking, age, amputation type, and injury mechanism. The variable of tissue loss before treatment was modeled using a restricted cubic spline function with 3 default knots (quantiles: 0.10, 0.50, and 0.90, which correspond to 1, 4, and 14 lost joints, respectively). Higher log odds values are indicative of an increased probability of unsuccessful revascularization. The gray area indicates the 95% confidence interval.

Fig. 4.

The receiver operating characteristic (ROC) curve of the regression model represents the association between the true-positive (sensitivity) and the false-positive (1– specificity) ratios indicating the regression model’s discriminatory power to predict failure of the operation. The true-positive ratio (sensitivity) is the ratio of correctly predicted positive observations to the total actual positives. The false-positive ratio (1– specificity) is the ratio of incorrectly predicted negative observations to the total actual negatives. The dashed diagonal line represents an area under the curve (AUC) value of 0.5, indicating a predictive performance no better than chance.

TABLE II.

Adjusted Odd Ratios (aORs) from the Multivariable Logistic Regression Model for Unsuccessful Revascularization

	aOR (95% CI)	P Value
Age, knot 1	0.93 (0.87-0.98)	0.01
Age, knot 2	1.19 (1.06-1.34)	0.004
Age, knot 3	0.29 (0.12-0.71)	0.01
Sex: female*	0.73 (0.30-1.80)	0.50
Smoking: yes†	0.96 (0.49-1.90)	0.92
Diabetes: yes‡	0.90 (0.28-2.85)	0.85
Injury mechanism§
Crush	2.82 (1.14-6.95)	0.02
Avulsion	5.95 (2.50-14.18)	<0.001
Amputation type: total#	2.92 (1.47-5.80)	0.002
Tissue loss before treatment (no. of lost joints), knot 1	1.31 (1.04-1.64)	0.02
Tissue loss before treatment (no. of lost joints), knot 2	0.55 (0.35-0.86)	0.01

^*Compared with sex: male.

^†Compared with smoking: no.

^‡Compared with diabetes: no.

^§Compared with injury mechanism: cut.

^#Compared with amputation type: subtotal.

Discussion

We found in our cohort of 282 consecutive patients who underwent revascularization surgery that avulsion and crush injury mechanisms and total amputation type were prognostic factors with the strongest association with an unsuccessful outcome. Age was the only significant patient-related prognostic factor.

The major limitations of our study were heterogeneous injury details and patient characteristics and a relatively small number of specific amputation types. In the multivariable analysis, we included the most important factors usually considered by surgeons in the emergency room when deciding how to treat amputation injuries. Still, other injury details may have led to residual confounding in our results. Overall, the relatively small number of patients increases uncertainty in our results due to random variation. The ROC curve of our multivariable model indicated moderate discriminative ability in distinguishing binary dependent outcomes, and the Nagelkerke R² coefficient value suggested moderate model fit. However, a substantial amount of unexplained variation in failed revascularizations remained within our study. We consider the risk of selection bias to be low, because our consecutive cohort of patients with amputation injuries was a completely unselected sample of the population within the referral area, all patients received care in the study hospital, and the whole population is covered by universal health care, which includes replantation surgery. We did not collect information about patients’ race and ethnicity, but most of the patients were White Caucasian, reflecting the Finnish population. This must be considered when assessing the generalizability of our study’s findings.

We decided not to include the surgeon’s expertise level in our analyses of prognostic factors. The aim of this study was to focus on patient characteristics and injury details, which can be directly used by the actual on-call surgeons in clinical decision-making on revascularization versus revision amputation, separate from their own experience level. In previous studies, surgeon experience level has been recognized as a prognostic factor for success in some studies^1,11,20, but there is also evidence to the contrary^6,10,12. Determining the true level of experience is complicated, as the amputation procedures often involve 2 surgeons with different levels of expertise. In a retrospective analysis such as the current study, it is impossible to determine which parts of the operations were critical and who performed those. In addition, we did not model and analyze some potentially important prognostic factors, such as ischemia time or specific damaged structures (e.g., the number of repaired or reconstructed vessels). Longer ischemia time has been associated with unsuccessful revascularizations in some studies^1,7,13,21, while other studies have not supported this^{5,6,8,10-12,20}.

A recent meta-analysis suggested that a higher number of anastomoses was associated with a lower failure proportion¹⁷, but there is also contradictory evidence suggesting that it may be more related to the quality of the anastomoses rather than the quantity¹⁶. A further confounding factor in these studies might have been the extent of the tissue damage. In severe injuries, repairing only 1 vein may be the only feasible option, while in a typical injury, 1 to 3 veins per artery are repaired or reconstructed. This can create a perception that unfavorable outcome was related to an intraoperative decision about the vessel repair, while actually, these options may have been limited because of the severity of the injury.

Avulsion and crush injuries have been associated with unsuccessful revascularization operations after upper-extremity amputation in previous meta-analyses^14,15,17-19, and our study confirms this with individual patient-level data. Our data may support the notion to more critically consider the indication for revascularization in amputations if the mechanism of injury has been an avulsion force. Avulsion and crush injury mechanisms are associated with extensive vessel damage and often require grafting techniques. In addition to the risk of insufficient vascular revision, injuries that necessitate the use of a graft are generally associated with a higher risk of thrombosis.

Total amputations were more frequently associated with an unsuccessful operation compared with subtotal amputations, which is in agreement with recent studies^3,12,13,20 and is probably explained by some existing venous return in the tissue that is in continuity in subtotal amputations. We observed a nonlinear U-shaped association with the extent of tissue loss before treatment, suggesting that revascularization of the most distal and proximal amputations is associated less frequently with an unsuccessful procedure. This aligns with findings from some previous studies^1,7,8,10,21, while other studies demonstrated no evidence of an association between the level of amputation and failure of procedures^6,11,13.

Patient age, sex, smoking, and diabetes have been associated with unsuccessful revascularizations after traumatic distal upper-extremity amputations^1,2,6-10,18. We observed a nonlinear association between age and failure of revascularization, with children having a higher probability of unsuccessful revascularizations compared with young adults, which is consistent with 2 previous meta-analyses^14,19. This can be explained by the technically more challenging operations in children because of smaller vascular structures and a pronounced tendency to vasospasm. Some studies have found an association between increasing age and failure of procedures^7,8,21,32, but conflicting evidence also exists^{1,2,5,10-13,20}. In our data, we found a decreased probability of unsuccessful revascularizations among the oldest patients, which we cannot explain. We did not find an association between smoking and unsuccessful revascularizations. Smoking has been associated with failure of operations in some previous studies^5-8,22, but not in all^{1,2,10-13,20,21}, and it likely would be preferable to model smoking history as a continuous variable instead of dichotomization, which decreases statistical power³³. We did not observe an association between sex and unsuccessful revascularizations, in agreement with the previous literature^{1,2,5-7,10-13,20-23}. We also found no association between diabetes mellitus and unsuccessful revascularizations, which aligns with previous studies^2,7,11,13, although an association has been suggested based on a meta-analysis¹⁸.

Our findings quantify the influence of some prognostic factors on the technical failure of revascularization surgery. Clinical decision-making about the treatment of an amputation injury brings together the surgeons’ expertise and the patient’s preferences in a shared decision-making process. The process should also consider the probability of a successful revascularization attempt, and our results suggest that the influence of injury details is more important than patient characteristics. For future studies, we recommend prospective data collection, whereby the inclusion of data regarding the surgeons’ level of expertise and details of technical aspects regarding vessel anastomosis can be justified.