Predictors of 30-day mortality in major lower limb amputations: insights from a five-year retrospective study in a South Asian LMIC

Abstract

Background

Major lower limb amputation (MLLA) is a life-saving procedure with significant morbidity and mortality. The objectives of this study were to determine the incidence of 30-day mortality, perioperative complications, and predictive risk factors of mortality in MLLA patients.

Methods

This retrospective cohort study included patients undergoing MLLA at The Aga Khan University Hospital, Pakistan, from January 2018 to December 2022. Data was collected using chart reviews and analyzed via Stata 15. Demographic, clinical, and procedural variables were examined. Predictive factors for 30-day mortality were assessed using Firth logistic regression.

Results

Among 286 patients, the mean age was 49.4 ± 20.9 years, and 79.4% were male. The leading indications for MLLA were diabetic foot (43.4%) and trauma (23.8%). Thirty-day mortality was 6.6%. Most deaths occurred in patients with diabetic foot (68.4%), chronic kidney disease (31.6%), or ischemic heart disease (36.8%). Below-knee amputation accounted for 68.4% of cases. Postoperative infections (11.2%) and stump necrosis (4.2%) were the most common complications. Multivariable analysis identified chronic kidney disease (OR: 3.613; 95% CI: 1.112–11.739; p = 0.033) and postoperative local wound infection (OR: 3.416; 95% CI: 1.036–11.267; p = 0.044) as significant predictors of 30-day mortality in this cohort.

Conclusion

MLLA is associated with considerable short-term mortality, particularly among patients with chronic kidney disease and postoperative surgical site infections. These findings emphasize the need for proactive identification of high-risk individuals, optimization of comorbid conditions, especially renal function, and strengthening perioperative infection control practices to improve patient outcomes in resource-limited settings.

Introduction

Major lower limb amputation is the surgical removal of a portion or the entire limb, involving the cutting of bone or joint proximal to the ankle, including above, below, through knee amputations and hip disarticulation [1]. MLLA is typically performed as a life-preserving intervention in cases of non-salvageable trauma, peripheral arterial disease, necrotizing infections, and diabetes-related complications. Although it is performed to improve patient survival, it is associated with considerable morbidity and represents a critical clinical event. The worldwide incidence of MLLA ranges from 3.6 to 68.4% [2], with diabetic foot ulcers identified as the leading cause [3]. According to a study conducted at Dow University of Health Sciences, Pakistan, diabetes was identified in 21.4% of all lower limb amputations, whereas non-diabetics accounted for 78.6% of cases, with trauma being the primary cause [4].

MLLA has significant morbidity and mortality reported worldwide. These complications range from major adverse cardiac events (MACE), infection, proximal amputation and prolonged Intensive Care Unit (ICU) stay. The reported incidence of inpatient morbidity remains more than 50% [5]. Similarly the 30 day mortality also is reported to be around 4–22% [6]. Jolissaint et al. identified predictors of 30-day mortality in patients undergoing MLLA [7]. Despite these findings, there is limited regional data on predictive factors associated with 30-day mortality and the associated mortality rates in patients undergoing MLLA.

A regional study conducted a prospective demographic survey among military individuals undergoing traumatic lower limb amputations, providing valuable insight into that specific population. However, its findings may have limited applicability to the general civilian population due to differences in etiology and healthcare setting [8]. There is limited data on the patterns and outcomes of MLLA in lower- and middle- income countries (LMICs), particularly South Asia. In LMICs such as Pakistan and Nigeria, trauma, infection, and diabetic foot are commonly reported causes of MLLA [1, 4]. In contrast, in developed settings, diabetes mellitus and peripheral arterial disease are consistently reported as the leading etiologies for non-traumatic MLLA [3]. Consequently, global incidence rates and outcome predictors may not be fully generalizable to LMIC populations. Given the limited availability of regional data, we conducted a retrospective audit of all the MLLA performed at our unit.

Due to a lack of regional data and variation in reasons for amputation, it is vital to understand MLLA mortality predictors for clinical planning and resource allocation, especially in resource-limited settings. Therefore, the objectives of this study were to determine the incidence of 30-day mortality, perioperative complications, and predictive risk factors of mortality in MLLA patients from a low middle income country like our region.

Materials and methods

This single center, retrospective audit was conducted using non-probability consecutive sampling technique and it adheres to STROBE guidelines for reporting observational studies. All patients who underwent major lower limb amputations at The Aga Khan University Hospital (AKUH) from 1st January 2018 till 31st December 2022, following Ethics Review Committee Approval, were included. Patients admitted at AKUH after index procedure performed at another institution/hospital and with unavailable complete medical records were excluded. Data was collected via chart review and included demographic, clinical, and procedural details using a predesigned proforma. Patients were identified using the ICD-9 coding system. The sample size was calculated using OpenEpi software version 3.0. Based on reported 30-day mortality rates for MLLA ranging from 12% to over 30% [7], a minimum of 163 patients was required at a 95% confidence interval, with an α value of 0.05 and a 10% margin of error.

Major lower limb amputation was defined as the surgical removal of a portion or the entire limb, involving the cutting of bone or joint proximal to the ankle. This includes above, below, through knee amputations and hip disarticulation [1]. Chronic kidney disease (CKD) is diagnosed when there is persistent kidney damage or a sustained reduction in glomerular filtration rate below 60 mL/min/1.73 m2 for a duration of at least three months [9]. Hypertension is diagnosed when blood pressure is consistently ≥ 130 and/or ≥ 80 mm Hg at least two office measurements on at least two separate occasions [10, 11]. Ischemic heart disease can be diagnosed based on a history of myocardial infarction or coronary revascularization (PCI or CABG), the presence of typical angina confirmed by diagnostic testing (e.g., stress testing or coronary angiography), or evidence of coronary artery obstruction during evaluation for acute coronary syndromes [12]. A surgical site infection (SSI) is defined as an infection occurring near the surgical incision within 30 days of the operative procedure, or within 90 days if prosthetic material or an implant is involved [13]. Diabetes Mellitus is diagnosed at an A1C of greater than or equal to 6.5%, fasting blood glucose of greater than or equal to 126 mg/dl or two-hour blood glucose of greater than or equal to 200 mg/dl [14].

Data analysis was conducted on Stata 15 software. Categorical data like gender, smoking status, comorbidities, mode of admission, indications of MLLA and post-operative complications were presented as frequency and percentages. Continuous data such as age were presented as mean and standard deviation/median and interquartile range. Associations between predictive factors for MLLA and 30-day mortality were determined via univariate and multivariable firth logistic regression. Predictor selection in the multivariable model was done using a modified stepwise approach. While p < 0.20 in univariate analysis was used as a general threshold, inclusion was most influenced by clinical relevance. Certain variables were analyzed despite not meeting the statistical threshold due to their importance in clinical practice. A p-value less than 0.05 was considered significant in the final model.

The Aga Khan University Ethics Review Committee (ERC), Pakistan approved this study and ERC Exemption was obtained before its commencement (Reference # 2023-9401-27386). Only laboratory values and medical information were retrieved from the hospital information management system at the Aga Khan University Hospital, Pakistan, with no direct contact with participants and the need for written informed consent waived by the AKUH ERC. This study adheres to the WMA Declaration of Helsinki. Data was de-identified after collection and stored on a password-protected computer that was accessible only by the authors.

Results

A total of 286 patient records were included in this study. Most of the patients were male (79.4%) with a mean age of 49.4 ± 20.9 years. Comorbidities included diabetes mellitus in 169 patients (59.1%), hypertension in 123 patients (43.0%), ischemic heart disease in 56 patients (19.6%), and CKD in 29 patients (10.1%). Diabetic foot was the most common indication for amputation (43.4%), followed by trauma (23.8%), acute limb ischemia (12.2%), necrotizing infection/wet gangrene (9.4%), and bone tumor (8.0%). Thirty-two patients developed surgical site infection (11.2%), 12 patients developed stump necrosis (4.2%), 4 patients developed wound dehiscence (1.4%), and one patient reported phantom limb pain (0.3%). Vascular surgery was the specialty performing the most amputation surgeries (58.0%). Most amputations were primary procedures (86.7%), and 241 patients were operated on under general anesthesia (84.3%). Demographic data is reported in Table 1.

Table 1.

Demographic, clinical, and procedural characteristics of patients undergoing major lower limb amputation (n = 286)

Demographic		Frequency (%)
Age (Mean ± SD)		49.4 ± 20.9
Gender	Male	227 (79.4)
	Female	59 (20.6)
Comorbidities	Hypertension	123 (43.0)
	Diabetes	169 (59.1)
	Ischemic Heart Disease	56 (19.6)
	Chronic kidney disease	29 (10.1)
	Peripheral arterial disease	15 (5.2)
	Cerebrovascular accident	2 (0.7)
	Smoking	11 (3.8)
Indication for amputation	Diabetic feet	124 (43.4)
	Trauma	68 (23.8)
	Acute limb ischemia	35 (12.2)
	Necrotizing infection/wet gangrene	27 (9.4)
	Non-reconstructible peripheral arterial disease	7 (2.4)
	Bone tumor	23 (8.0)
	Other	2 (0.7)
Local wound complications	Infection	32 (11.2)
	Stump necrosis	12 (4.2)
	Wound dehiscence	4 (1.4)
	Phantom limb pain	1 (0.3)
Surgical specialty performing the amputation	Vascular Surgery	166 (58.0)
	Orthopedic Surgery	101 (35.3)
	General Surgery	18 (6.3)
	Pediatric Surgery	1 (0.3)
Level of amputation	Below knee	184 (64.3)
	Above knee	92 (32.2)
	Through knee	9 (3.1)
	Hip disarticulation	1 (0.3)
Type	Primary	248 (86.7)
	Revision	38 (13.3)
Side	Right	149 (52.1)
	Left	133 (46.5)
	Bilateral	3 (1.0)
Anesthesia	General	241 (84.3)
	Spinal	32 (11.2)
	Monitored Anesthesia Care	8 (2.8)
	Block	5 (1.7)
Mortality		19 (6.6)

Among the 27 patients who presented with necrotizing infection preoperatively, 5 (18.5%) developed postoperative SSI. In contrast, 27 out of 259 patients (10.4%) without preoperative necrotizing infection developed postoperative SSI. A higher proportion of SSI was observed in patients with necrotizing infection, but the association was not statistically significant (p = 0.204). The distribution of postoperative SSI by preoperative infection status is detailed in Table 2.

Table 2.

Association between preoperative necrotizing infection and postoperative surgical site infection (SSI) (n = 286)

Preoperative Necrotizing Infection	Postoperative SSI present (%)	No Postoperative SSI (%)	Total
Present	5 (18.5)	22 (81.5)	27
Absent	27 (10.4)	232 (89.6)	259
Total	32 (11.2)	254 (88.8)	286

chi² = 1.612; p = 0.204

In our study, the 30-day mortality rate was 6.6% (19 patients). Among patients who died within 30 days, diabetic foot was the most common indication for MLLA (68.4%), and infection was the only local wound complication post operatively (26.3%). Sixteen patients were diabetic (84.2%), fifteen patients were hypertensive (78.9%), six patients had CKD (31.6%), and seven patients had ischemic heart disease (36.8%). Below-knee amputation was the most common type of amputation (68.4%). Further demographic data for the mortality cohort is reported in Table 3.

Table 3.

Demographic, clinical, and procedural characteristics of the 30-Day mortality cohort (n = 19)

Demographic		Mortality (%)
Age (Mean ± SD)		57.6 ± 15.2
Gender	Male	16 (84.2)
	Female	3 (15.8)
Comorbidities	Hypertension	15 (78.9)
	Diabetes	16 (84.2)
	Ischemic Heart Disease	7 (36.8)
	Chronic kidney disease	6 (31.6)
	Peripheral arterial disease	0 (0.0)
	Cerebrovascular disease	1 (5.3)
	Smoking	1 (5.3)
Indication for amputation	Diabetic feet	12 (63.2)
	Trauma	1 (5.3)
	Acute limb ischemia	3 (15.8)
	Necrotizing infection/wet gangrene	3 (15.8)
	Non-reconstructible peripheral arterial disease	0 (0.0)
	Bone tumor	0 (0.0)
	Other	0 (0.0)
Local wound complications	Infection	5 (26.3)
	Stump necrosis	0 (0.0)
	Wound dehiscence	0 (0.0)
	Phantom limb pain	0 (0.0)
Surgical specialty performing the amputation	Vascular Surgery	16 (84.2)
	Orthopedic Surgery	1 (5.3)
	General Surgery	2 (10.5)
	Pediatric Surgery	0 (0.0)
Level of amputation	Below knee	13 (68.4)
	Above knee	6 (31.6)
	Through knee	0 (0.0)
	Hip disarticulation	0 (0.0)
Type	Primary	19 (100.0)
	Revision	0 (0.0)
Anesthesia	General	18 (94.7)
	Spinal	1 (5.3)
	Block	0 (0.0)
	Monitored Anesthesia Care	0 (0.0)

Firth logistic regression (Table 4) determined predictive factors associated with 30-day mortality in MLLA patients. After adjustment, OR of CKD (OR: 3.613; 95% CI: 1.112, 11.739; p = 0.033) and surgical site infection (OR: 3.416, 95% CI: 1.036, 11.267; p = 0.044) were statistically significant on multivariable analysis. Therefore, our analysis indicates that there is a significant association of CKD and surgical site infection with the occurrence of 30-day mortality in MLLA patients. The statistical analysis of all the variables is reported in detail in Table 4.

Table 4.

Predictive factors of 30-day mortality in MLLA patients (n = 286)

Factors	Group-specific Mortality (n/N)	Unadjusted Odds Ratio (95% CI)	p value	Adjusted Odds Ratio (95% CI)	p value
Chronic Kidney Disease
Yes	6/29	5.009 (1.794–13.990)	0.002	3.613 (1.112–11.739)	0.033*
No	13/257	Reference		Reference
Local Wound Complications
Infection	5/32	3.083 (1.070–8.884)	0.037	3.416 (1.036–11.267)	0.044*
Stump necrosis	0/12	0.617 (0.035–10.941)	0.742	0.479 (0.024–9.731)	0.632
Wound dehiscence	0/4	1.713 (0.088–33.370)	0.723	0.890 (0.034–23.584)	0.944
Phantom limb pain	0/1	5.138 (0.200–131.767)	0.323	1.444 (0.037–35.304)	0.939
No complications	14/237	Reference		Reference
Reason for amputation
Trauma	1/68	1.044 (0.041–26.531)	0.979	2.314 (0.062–86.703)	0.650
Acute limb ischemia	3/35	5.061 (0.249–102.720)	0.291	13.199 (0.326–534.314)	0.172
Non-reconstructible peripheral arterial disease	0/7	3.133 (0.057–171.962)	0.576	4.492 (0.040–501.253)	0.532
Necrotizing infection/wet gangrene	3/27	6.714 (0.329–137.136)	0.216	11.020 (0.357–340.278)	0.170
Diabetic foot	12/124	5.222 (0.299–91.319)	0.258	12.607 (0.317–501.085)	0.177
Other	0/2	9.400 (0.151–585.104)	0.288	42.692 (0.444–4108.902)	0.107
Bone tumor	0/23	Reference		Reference
Level of amputation
Above knee	6/92	0.955 (0.362–2.520)	0.925	2.337 (0.658–8.297)	0.189
Through knee	0/9	0.669 (0.037–12.119)	0.785	6.170 (0.212–179.222)	0.290
Hip disarticulation	0/1	4.235 (0.164–109.015)	0.384	1.830 (0.050–66.820)	0.742
Below knee	13/184	Reference		Reference
Diabetes
Yes	16/169	3.517 (1.082–11.427)	0.037	1.098 (0.133–9.053)	0.930
No	3/117	Reference		Reference
Ischemic heart disease
Yes	7/56	2.648 (1.018–6.892)	0.046	1.453 (0.496–4.258)	0.495
No	12/230	Reference		Reference
Age
50 years and above	15/168	2.569 (0.875–7.544)	0.086	1.359 (0.256–7.214)	0.718
Lesser than 50 years	4/118	Reference		Reference
Anesthesia
General	18/241	1.738 (0.316–9.565)	0.703	1.611 (0.290–8.957)	0.586
Block	0/5	1.909 (0.069–53.140)	0.525	2.184 (0.068–70.276)	0.659
MAC	0/8	1.235 (0.046–33.127)	0.900	0.501 (0.017–14.938)	0.690
Spinal	1/32	Reference		Reference
Gender
Male	16/227	1.259 (0.383–4.140)	0.704	1.812 (0.488–6.729)	0.374
Female	3/59	Reference		Reference
Smoking status
Smoker	1/8	2.816 (0.458–17.301)	0.264	1.200 (0.285–14.000)	0.486
Non-smoker	18/278	Reference		Reference
Potassium
Hypokalemia	3/24	2.676 (0.755–9.490)	0.127	4.084 (0.979–17.046)	0.054
Hyperkalemia	4/45	1.783 (0.577–5.507)	0.315	1.243 (0.341–4.537)	0.742
Normal	12/217	Reference		Reference
Sodium
Hyponatremia	13/139	2.858 (0.955–8.548)	0.06	1.512 (0.485–4.716)	0.476
Hypernatremia	2/23	3.114 (0.621–15.619)	0.167	6.003 (0.811–43.434)	0.079
Normal	4/124	Reference		Reference
Primary/Revision procedure
Primary	19/248	6.542 (0.387–110.622)	0.193
Revision	0/38	Reference
Hemoglobin
Low	15/240	0.717 (0.213–2.407)	0.590
High	1/7	2.407 (0.299–19.395)	0.409
Normal	3/39	Reference
Serum CRP
High CRP	17/268	0.459 (0.111–1.892)	0.281
Low CRP	2/18	Reference
WBC Count
High WBC	17/198	3.336 (0.865–12.865)	0.080
Low/normal WBC	2/88	Reference
Procalcitonin
High risk of sepsis	14/239	0.497 (0.177–1.398)	0.185
Low risk of sepsis	5/47	Reference

*Significant at p value < 0.05 on multivariate analysis

CRP: C-Reactive Protein; WBC: White Blood Cell, MAC: Monitored Anesthesia Care

CI: Confidence Interval

Discussion

The observed 30-day mortality rate reinforces the gravity of MLLA in LMIC settings. The significant impact of CKD and wound infection on early mortality highlights modifiable perioperative risk factors that could inform targeted interventions.

Renal insufficiency has been shown to exhibit this association in several existing studies, albeit mostly from the developed part of the globe. A recent retrospective study noted a significantly higher 5-year mortality among patients with CKD that underwent MLLA [15]. Moreover, dialysis patients had a 4 times increased risk of mortality post-amputation in a prospective study [16]. Assi et al. reported that patients with CKD represented a sub-group that were at a greater risk of mortality [17]. Furthermore, a study conducted in South Texas, USA found that more patients with CKD or on dialysis underwent AKA and BKA compared to patients without renal disease [18]. CKD predicts higher mortality because anemia, secondary to decreased erythropoietin synthesis, compromises oxygen delivery, therefore hindering post-operative recovery [18]. Renal disease is also known to increase the rate of atherosclerosis, increasing the risk of cardiovascular disease which negatively affects prognosis [19]. To the authors’ knowledge, no studies were found that directly contradict this finding. Our results make valuable contributions to the evidence of the effect of CKD on mortality particularly in an LMIC setting. We recommend prospective studies to be carried out in LMICs to be able to better understand and reinforce this association, as well as carry out subgroup analysis in studies with larger datasets.

Surgical site infection is a known predictor of unfavorable outcomes in most surgical procedures, including MLLA as indicated by multivariable analysis in this study. A retrospective study of 342 patients in Brazil concluded that post-operative infection significantly predicted mortality in patients who underwent major amputation surgery [20]. Surgical site infection was noted to be a common complication after amputation surgery in a recent systematic review [21]. Furthermore, a 10-year retrospective study studied traumatic amputations and reported wound infection to be the most common complication (56.6%) and was the major contributor to the incidence of mortality [22]. However, it is possible that a uremia-induced immune dysfunction secondary to CKD may be highly associated with the increased risk of infection [23]. It’s possible that infection acts as a partial mediator in the causal pathway between CKD and mortality, rather than being a completely independent predictor. While our analysis shows a clear connection between local wound infections and higher mortality rates, more research is needed to determine whether this effect stands on its own or is partly influenced by the presence of CKD. To better understand this relationship, future studies should consider using causal mediation analysis. This approach could help clarify how much of CKD’s impact on mortality is direct and how much is indirectly driven by an increased risk of infection.

We found that diabetes, although significantly associated with mortality on univariate analysis, was not significantly associated with higher risk of mortality in this cohort when adjusted for other factors on multivariable analysis. This finding is congruent with some existing literature e.g. a prospective study found a significantly increased risk of mortality in dialysis patients irrespective of diabetic status [16]. Another study also observed that rates of 180-day mortality in amputation patients were similar among patients with and without diabetes, but also observed higher mortality in diabetic patients in more than 180 days post-operatively [24]. Lilja et al. conducted nationwide propensity score adjusted analysis and concluded that there were no differences in mortality in patients with or without diabetes [25]. However, several studies contradict these findings. A meta-analysis conducted in 2017 reported diabetes, among other comorbidities, to cause at least a two-fold increase in mortality rates in lower limb amputation [26]. Given the conflicting evidence, we recommend prospective studies to be carried out with a high focus on diabetes as a predictor of mortality to help resolve this discrepancy, especially in our population.

This study is sufficiently powered, surpassing the required sample size, thereby enhancing the reliability of our findings. It addresses a significant gap in regional data by analyzing factors contributing to mortality after MLLA and will potentially aid the post-operative management and monitoring of lower limb amputees. These methodological strengths improve the generalizability of our results.

This is a single center study; therefore, results cannot be generalized over the population. Further multi-center studies will need to be conducted to provide findings that can be representative of the overall population. Moreover, we could not conduct a formal mediation analysis, which makes it difficult to ascertain the direct contribution of CKD to mortality as opposed to its contribution through elevated risk of infection. We did not assess interactions because of the small number of deaths, which limited how many variables we could include in the model. However, this possible interaction could influence mortality risk and should be explored in future studies with larger sample sizes. Medical records with data on CKD staging and dialysis status were not consistently available, which limited our ability to assess the association of CKD severity or dialysis dependence on mortality. While the study gives some insights, future works that incorporate mediation analysis and consider the severity of CKD could improve understanding of the phenomena. A modified stepwise approach that combines statistical and clinical considerations was used to develop our multivariable model. This flexible method is often used in retrospective studies where strict statistical thresholds may not fully capture clinically meaningful relationships. While not all variables included met the univariate significance threshold, inclusion was based on clinical importance and relevance to our patient population. This strategy helped us balance data-driven insights with practical clinical value, though future studies with larger samples and prospective designs will be helpful to validate these findings.

Conclusion

This study found a 30-day mortality rate of 6.6% among patients undergoing MLLA. CKD and surgical site infection were identified as significant predictors of 30-day mortality. By addressing a regional gap in data, our findings highlight the need for early identification of high-risk patients, management of CKD, and optimization of perioperative infection control protocols. These measures are especially critical in LMIC settings, where delayed presentations and limited resources may increase risk.

Abbreviations

MLLA

Major Lower Limb Amputation

CKD

Chronic Kidney Disease

ICU

Intensive Care Unit

MACE

Major Adverse Cardiac Events

AKUH

Aga Khan University Hospital

ERC

Ethics Review Committee

ICD-9

International Classification of Diseases, 9th Revision

OR

Odds Ratio

CI

Confidence Interval

LMIC

Low-Middle Income Country

AKA

Above-Knee Amputation

BKA

Below-Knee Amputation

Author contributions

M.Z. collected data, conducted data analysis, authored the initial draft of the manuscript and included revisions to make the final manuscript. Z.U.R. conceptualized the study, supervised all the stages of the study and revised the manuscript. M.A. developed the methodology and sought Ethics Review Committee Approval. H.S. and A.Q.M. made major contributions to data collection. N.A.S. and F.S. revised and reviewed the initial draft of the manuscript. All authors read and approved the final manuscript.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or non-profit sectors.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The Aga Khan University Ethics Review Committee (AKUH ERC), Pakistan approved this study and ERC Exemption was obtained before its commencement (Reference # 2023-9401-27386). Only laboratory values and medical information were retrieved from the hospital information management system at the Aga Khan University Hospital, Pakistan, with no direct contact with participants and the need for written informed consent waived by the AKUH ERC. This study adheres to the WMA Declaration of Helsinki. Data was de-identified after collection and stored on a password-protected computer that was accessible only by the authors.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.