Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

ORIGINAL RESEARCH

The Epidemiology of Extremity Threat and Amputation after


Vasopressor-Dependent Sepsis
Katherine M. Reitz1,2,3, Jason Kennedy2,4, Caroline Rieser1, Callie Hlavin1, Hayley B. Gershengorn5,6,
Matthew D. Neal1,4, Nicole Bensen2,4, Kelsey Linstrum2,4, Hallie C. Prescott7, Matthew R. Rosengart1,4,
Victor Talisa2,8, Daniel E. Hall1,9,10,11, Edith Tzeng1,3,9, Hannah Wunsch12,13, Sachin Yende2,4,10,
Derek C. Angus2,4, and Christopher W. Seymour2,4
1
Department of Surgery, 4Department of Critical Care Medicine, and 8Department of Biostatistics, University of Pittsburgh, Pittsburgh,
Pennsylvania; 2Clinical Research, Investigation, and Systems Modeling of Acute Illness Center, and 3Division of Vascular Surgery,
University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania; 5Division of Pulmonary, Critical Care, and Sleep Medicine, Miller
School of Medicine, University of Miami, Miami, Florida; 6Division of Critical Care Medicine, Albert Einstein College of Medicine, Bronx,
New York; 7Division of Pulmonary and Critical Care Medicine, University of Michigan, Ann Arbor, Michigan; 9Veterans Affairs Pittsburgh
Healthcare System, and 11Center for Health Equity Research and Promotion, Veterans Affairs, Pittsburgh, Pennsylvania; 10Wolff Center,
University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania; 12Department of Critical Care Medicine, Sunnybrook Health Science
Centre, Toronto, Ontario, Canada; and 13Department of Anesthesia and Interdepartmental Division of Critical Care Medicine, University
of Toronto, Toronto, Ontario, Canada

ORCID IDs: 0000-0002-9397-321X (K.M.R.); 0000-0002-7360-2489 (H.B.G.); 0000-0002-6596-8034 (S.Y.).

Abstract years; male, 6,548 [54%]; sequential organ failure assessment


[SOFA], 10 6 4). Of these, 231 (2%) patients had a threatened
Rationale: Extremity threat and amputation after sepsis is a extremity with 26 undergoing 37 amputations, a risk of 2.2
well-publicized and devastating event. However, there is a paucity of (95% confidence interval [CI], 1.4–3.2) per 1,000, and 205 not
data about the epidemiology of extremity threat after sepsis onset. undergoing amputation, a risk of 17.0 (95% CI, 14.8–19.5) per
1,000. Most amputations occurred in lower extremities (95%),
Objectives: To estimate the incidence of extremity threat with a median (interquartile range) of 16 (6–40) days after sepsis
or without surgical amputation in community sepsis. onset. Compared with patients with no extremity threat,
Methods: Retrospective cohort study of adults with Sepsis-3 patients with threat had a higher SOFA score (11 6 4 vs.
hospitalized at 14 academic and community sites from 2013 to 10 6 4; P , 0.001), serum lactate (4.6 mmol/L [2.4–8.7] vs. 3.1
2017. Vasopressor-dependent sepsis was identified by [1.7–6.0]; P , 0.001), and more bacteremia (n = 37 [37%] vs.
administration of epinephrine, norepinephrine, phenylephrine, n = 2,087 [26%]; P , 0.001) at sepsis onset. Peripheral vascular
vasopressin, or dopamine for more than 1 hour during the 48 disease, congestive heart failure, SOFA score, and
hours before to 24 hours after sepsis onset. Outcomes included the norepinephrine equivalents were significantly associated with
incidence of extremity threat, defined as acute onset ischemia, with extremity threat.
or without amputation, in the 90 days after sepsis onset. The Conclusions: The evaluation of a threatened extremity resulting
association between extremity threat, demographics, comorbid in surgical amputation occurred in 2 per 1,000 patients with
conditions, and time-varying sepsis treatments was evaluated using vasopressor-dependent sepsis.
a Cox proportional hazards model.
Results: Among 24,365 adults with sepsis, 12,060 (54%) were Keywords: sepsis; vasopressor; symmetric peripheral gangrene;
vasopressor dependent (mean 6 standard deviation age, 64 6 16 amputation; peripheral ischemia

(Received in original form May 7, 2021; accepted in final form October 12, 2021)
Supported in part by the National Heart, Lung, and Blood Institute (5T32HL0098036) and National Institute on Aging (L30AG064730) (K.M.R.);
the National Cancer Institute (T32CA113263) (C.R.); and grants for the National Institute of General Medical Sciences (R35GM119519 and
1R35GM119526-01) (M.D.N., D.C.A., and C.W.S.). These funding sources had no role in the design and conduct of the study; collection,
management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit for
publication. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data;
preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Ann Am Thorac Soc Vol 19, No 4, pp 625–632, Apr 2022
Copyright © 2022 by the American Thoracic Society
DOI: 10.1513/AnnalsATS.202105-547OC
Internet address: www:atsjournals:org

Reitz, Kennedy, Rieser, et al.: Extremity Threat and Amputation after Sepsis 625
ORIGINAL RESEARCH

Sepsis is a life-threatening event in which the Pittsburgh Medical Center hospitals in Definition of Vasopressor-
body’s response to an infection results in western Pennsylvania. Hospital EHR data dependent Sepsis
injury to its own tissues and organs (1). (CERNER Co.) was linked with a unique We defined vasopressor-dependent sepsis as
Sepsis occurs in nearly 50 million people per patient identifier to outpatient records before evidence of International Consensus Criteria
year (2), and although deadly, more than and after the incident sepsis encounter (EPIC for Sepsis-3 in the EHR at any point during
four in five patients survive (3, 4). A Systems Co.). All reporting is consistent with the hospital encounter (1), including 1)
particularly devastating and well-publicized the Strengthening the Reporting of evidence of suspected infection, and 2)
complication among these survivors is the Observational Studies statement (25). presence of organ dysfunction. A suspected
loss of extremities (5). infection was identified as the administration
Extremity ischemia after sepsis, of either oral or parenteral antibiotics and
historically termed symmetric peripheral Data acquisition of a body fluid culture specimen.
gangrene, was described over 100 years ago We studied inpatient EHR encounter data Organ dysfunction was defined as at least 2
(6). It has been attributed to a variety of including baseline demographics, admitting SOFA points within the 48 hours before to
causes, including preexisting peripheral hospital, and known factors affecting 24 hours after the suspected infection (29).
vascular disease (7), vasopressor use (8–17), peripheral blood supply and/or coagulation Only the first episode of sepsis was included.
various pathogens (18–21), and (i.e., diabetes, cancer, congestive heart Vasopressor-dependent sepsis was defined as
disseminated intravascular coagulation failure [CHF], peripheral vascular disease the receipt of intravenous epinephrine,
(22–24). Although amputation after sepsis is [PVD], and cardiovascular disease [CVD]) norepinephrine, phenylephrine, vasopressin,
an impactful outcome for patients, the (12, 26, 27) and the Charlson Comorbidity or dopamine for at least 1 hour in the 48
incidence of a threatened extremity with or Index were identified using the hours before to 24 hours after sepsis
without amputation after sepsis is not well International Classification of Diseases, onset (1).
described. Clinical Modification, of the 9th and 10th
In an integrated health system using revision codes (Table E1) (28). We Outcomes
electronic health record (EHR) data, we quantified patient status in the 48 hours The primary outcome was a threatened
quantified the incidence of and risk factors before to 24 hours after sepsis onset extremity (i.e., limb and/or digit) with or
for the surgical evaluation for a threatened using vital signs, laboratory results, and without surgical amputation in the 90 days
extremity with or without amputation in evidence of end organ dysfunction after vasopressor-dependent sepsis
patients with vasopressor-dependent sepsis. measured by the sequential organ failure onset. The EHR was reviewed for
A subset of summarized data maybe assessment (SOFA) score (29). In the 2 documented surgical evaluation within 90
shared upon request to the corresponding days before and 90 days after sepsis onset, days of sepsis onset, including 1) consult
author. we quantified the magnitude of vasopressor notes by orthopedic, vascular, or plastic
exposure each day by the number of surgery; or 2) surgical amputation(s),
vasopressors used and norepinephrine independent of surgical specialty. We used a
Methods equivalents (14). We captured intensive broad definition of extremity threat, which
care unit (ICU), hospital length of stay, includes evidence of acute onset ischemia
The project was approved with a waiver of and outpatient data including smoking with or without large vessel occlusion or
informed consent and Health Insurance status at the most recent clinic visit before iatrogenic external tissue trauma (i.e.,
Portability and Accountability Act incident sepsis. Among all reviewed pressure ulcers). This definition explicitly
authorization by the University of Pittsburgh hospitalizations, the missingness of EHR excludes primarily infected extremities (e.g.,
Human Research Protection Office (Study abstracted variables was measured. Current diabetic foot wound, antecedent traumatic
20040001). Procedural Terminology and intraoperative injury, pressure wounds, etc.) which were the
reports were captured independent of source of sepsis. Two independent reviewers
Study Design inpatient status to capture those completed (K.M.R. and C.R.) evaluated patient
We performed a retrospective cohort study electively, after discharge from the index admission, consult, progress, and operative
among hospitalized adults (age >18 yr) who hospitalization. Mortality was determined notes; laboratory results; imaging, vascular
met International Consensus Criteria for using both inpatient discharge disposition laboratory, and pathology reports; and
Sepsis-3 (1) and received vasopressors from and monthly updated Social Security Death discharge and/or death summaries to
January 1, 2013, to December 31, 2017, at 14 Index linked to the outpatient EHR determine the presence of threatened
community and academic University of (30–32). extremities in all patients with surgical

Author Contributions: Study concept and design: K.M.R., J.K., H.B.G., H.C.P., H.W., D.C.A., and C.W.S. Acquisition of data: K.M.R., J.K., C.R.,
C.H., N.B., K.L., D.E.H., and C.W.S. Interpretation of data: all authors. Drafting of the manuscript: K.M.R. Critical revision of the manuscript for
important intellectual content: all authors. Study supervision: K.M.R., D.C.A., and C.W.S. K.M.R. and C.W.S. take full responsibility for the
integrity of the data and the accuracy of the data analysis.
Correspondence and requests for reprints should be addressed to Katherine M. Reitz, M.D., M.Sc., Division of Vascular Surgery, University of
Pittsburgh School of Medicine, 200 Lothrop Street, F677 Presbyterian Hospital, Pittsburgh, PA 15213. E-mail: reitzkm2@upmc.edu.
This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org.

626 AnnalsATS Volume 19 Number 4 | April 2022


ORIGINAL RESEARCH

evaluations. Interrater agreement was multiple vasopressors and escalating 12,060 (1%) were vasopressor dependent
measured with a kappa statistic. The dosages), as well as the competing risk of (Figure 1). Among those, the mean age was
independent reviewers discussed categorical mortality (29). 64 6 16 years, 54% (n = 6,548) were male,
disagreements. If a consensus could not be Data were presented as mean (6 and the mean SOFA score was 10 6 4.
reached, a third reviewer (C.W.S.) standard deviation) or median (interquartile Among the 8,221 (68%) patients who
determined the categorization. range [IQR]), as appropriate. For survived to discharge, a majority (60%) were
Surgical amputations were identified comparisons, we used t tests, analysis of not discharged to home. Death occurred in
using standard procedural descriptions in the variance, and Kruskal-Wallis tests for 4,806 patients (40%) by 90 days and in 5,747
EHR on mandated and time-stamped continuous data and chi-square or Fischer patients (48%) at 1 year.
intraoperative reports and/or by Current exact tests for categorical data. The
Procedural Terminology code (Table E2). association between extremity threat with or Risk of Extremity Threat with or
Surgical amputations were categorized by without amputation and comorbid without Amputation
anatomic location, surgical specialty, the conditions, sepsis severity at onset, and daily Among 12,060 patients with vasopressor-
number of amputations if multiple, and sepsis treatments were assessed with an dependent sepsis, 1,506 (12%) patients had a
autoamputations (i.e., spontaneous exploratory Cox proportional hazards model surgical evaluation. We excluded 1,275 for
separation of nonviable tissue) occurred. censoring at 90 days and for the competing evaluations unrelated to extremity threat or if
The secondary outcomes were ICU and risk of mortality, with a robust variance the threatened extremity was the source of
hospital length of stay, ICU mortality, estimator (33). Analyses were performed by sepsis. The remaining 231 (2%) patients were
hospital disposition, and 90-day as well as March 20, 2021, with Stata 15.0 (StataCorp) deemed to have a threatened extremity. In
365-day mortality. Outcomes were measured and R 4.0 (the R Foundation), with a the 90 days after sepsis onset, 26 underwent
through January 1, 2019. significance threshold of P , 0.05 in two- amputation, a risk of 2.2 patients (95% CI,
sided tests (34). 1.4–3.2) per 1,000, and 205 had threat but no
Statistical Analysis amputation, a risk of 17.0 (95% CI,
We estimated the proportion of patients with Sensitivity Analysis 14.8–19.5) per 1,000 (Figure 1). Interrater
extremity threat and calculated the incidence We performed sensitivity analyses evaluating agreement for extremity threat generated a
risk (events in the 90 days after sepsis onset, the robustness of the risk of extremity threat. Cohen’s kappa of 0.6 with less than 1%
per 1,000 patients) with binomial exact 95% First, we determined risks after excluding (n = 12) requiring third-party review.
confidence intervals (CI) of threat with or specific groups of patients with known risks A total of 26 patients underwent one or
without amputation in those with for amputation prior to sepsis onset: 1) CHF; more amputation, 7 (27%) had more than
vasopressor-dependent sepsis. To 2) PVD; 3) diabetes; 4) CHF or PVD; or 5) one surgical amputation, and 1 (4%) had a
understand the range of extremity threat, we PVD or diabetes (12, 27). Second, we documented toe autoamputation. Among
stratified the observed risks among those determined the risks using five alternative amputated patients, a total of 37 surgical
with or without amputation by 1) admitting definitions of vasopressor-dependent sepsis amputations occurred, most of which were
hospital; 2) patient age; 3) individual baseline to explore the risk of extremity threat across below- or above-knee amputations (n = 21
risk factors known to affect cardiovascular varying degrees of critical illness: 1) 3 or [56%]) and nearly all occurred in the lower
health and coagulation (i.e., diabetes, cancer, more maximal SOFA score points; 2) 4 or extremities (n = 35 [95%]) (Figure E1). More
smoking, CHF, PVD, or CVD); and 4) the more maximal SOFA score points; 3) no than 90% of extremity amputations for threat
total sum of baseline risk factors; 5) among more than 2 consecutive calendar days of were performed by vascular, orthopedic, or
those who survived to a nonhospice hospital vasopressors for more than 1 hour; 4) serum plastic surgery (Table E3).
discharge. lactate in the 48 hours before and 24 hours Compared with patients with no
We evaluated the clinical course of after sepsis onset; and 5) serum lactate more extremity threat, patients with threat had a
patients with vasopressor-dependent sepsis, than 2.0 mmol/L during the 48 hours before higher SOFA score (mean, 11 6 4 vs. 10 6 4;
stratified by degree of extremity threat, using to 24 hours after sepsis onset (septic shock by P , 0.001), international normalized ratio
estimated length of stay (both ICU and Sepsis-3 [1, 35]). Third, we determined the (median, 1.7 [IQR, 1.3–2.7] vs. 1.5 [IQR,
hospital), hospital discharge disposition, and risks among patients with or without 1.2–2.0]; P , 0.001), serum lactate (median,
mortality in-hospital and within 90 and 365 bacteremia at sepsis onset. Finally, in 4.6 mmol/L [IQR, 2.4–8.7] vs. 3.1 [IQR
days. additional multivariable models, we 1.7–6.0]; P , 0.001); lower fibrinogen
To display the relationship between accounted for hospital-specific variability in (median, 226 mg/dL [IQR, 150–374] vs. 276
vasopressor exposure and extremity threat, the adjusted risk of extremity threat by 1) [IQR, 176–425]; P = 0.039); and more
we present the proportion of patients clustering by hospital to estimate variance bacteremia (n = 37 [37%] vs. n = 2,087 [26%];
exposed to any and each individual using robust sandwich estimators; and 2) P , 0.001) at sepsis onset. Among patients
vasopressor from 2 days before to 14 days using treating hospital as a random effect. with threat, those who underwent
after sepsis onset. We used heatmaps of daily amputations had a lower platelet count than
cardiovascular SOFA points from 2 days those who did not (P = 0.003) (Table 1).
prior to 28 days after sepsis onset to illustrate Results The risks of extremity threat after
the variation in the cumulative of vasopressor-dependent sepsis varied by
cardiovascular dysfunction (i.e., hypotension Among 1,107,255 adult patients hospitalized, hospital (range with amputation, from 1.1 to
[mean arterial blood pressure less than 70 237,764 (21%) were treated for a suspected 4.4 per 1,000; range without amputation,
mm Hg]) and vasopressor support (i.e., infection, 168,103 had sepsis (15%), and from 3.2 to 25.3 per 1,000) and by age (range

Reitz, Kennedy, Rieser, et al.: Extremity Threat and Amputation after Sepsis 627
ORIGINAL RESEARCH

1,107,255 Hospitalized adult


patients

237,764 Suspected infection

168,103 Sepsis

12,060 Vasopressor-dependent
sepsis

1,506 Surgical evaluation 10,554 No surgical evaluation

1,275 Sepsis without extremity threat


1,106 Evaluation not for extremity
threat
169 Extremity cause of sepsis

231 Extremity threat

26 Extremity threat with 205 Extremity threat without


11,829 No extremity threat
amputation amputation

Figure 1. Cohort accrual.

with amputation, from 0.5 to 5.0 per 1,000; significantly different (n = 4 [15%] vs. n = 64 without. There were no differences found in
range without amputation, from 9.8 to 25.5 [31%]; P = 0.095) among those with or these parameters compared with those with
per 1,000) (Figure E2). The risks of extremity without amputation. However, the hospital or without amputation (P , 0.05 for all).
threat with (7.3 [95% CI, 2.0–18.8] per 1,000) length of stay was longer (median, 19 days
or without amputation (35.1 [95% CI, [IQR, 12–32] vs. 16 [IQR, 9–29]; P , 0.001) Sensitivity Analysis
21.2–54.2] per 1,000) were higher among and there was a lower risk of in-hospital After excluding patients with baseline risk
those with baseline PVD (Figure 2). Finally, mortality (n = 5 [19%] vs. n = 82 [40%]; factors for amputation (i.e., CHF or PVD),
the risk of extremity threat among those P = 0.039) among those with amputations the risk of extremity threat with amputation
discharged alive was 2.6 (95% CI, 1.6–3.9) when compared with those without. No was 2.1 (95% CI, 1.1–3.4) and without
per 1,000 with and 15.1 (95% CI, 12.6–18.0) significant differences were detected in the amputation was 14.8 (95% CI, 12.3–17.5) per
per 1,000 without amputation. In an risk of discharge home (n = 3/21 [14%] vs. 1,000 patients (Table E7). Changes in the risk
exploratory Cox proportional hazard model, n = 27/123 [22%]; P = 0.816) among of amputation were not observed using
norepinephrine equivalents, SOFA score at survivors with or without amputation. There alternative definitions of vasopressor-
sepsis onset, and comorbid conditions, were no statistical differences found in the dependent sepsis, ranging from 2.1 to 2.4 per
including CHF and PVD, were associated 90-day (n = 11/26 [42%] vs. n = 97/205 1,000 patients. Restricting the cohort to
with the risk of extremity threat with or [47%]; P = 0.630) or 365-day (n = 18/26 patients with 2 or more days of vasopressor
without amputation (Figure 3, Figures E3 [69%] vs. n = 118/205 [58%]; P = 0.255) exposure or Sepsis-3 septic shock, the risks of
and E4, and Table E5). mortality among those with or without threat without amputation were 23.6 (95%
amputations (Table E6). CI, 19.1–28.8) and 21.4 (95% CI, 18.2–25.0)
Timing of Surgical Evaluations and per 1,000. Among those with bacteremia at
Outcomes among Patients with Vasopressor Use sepsis onset, the risks of threat with
Extremity Threat From the 2 days before until 28 days amputation were higher, 5.1 (95% CI,
The time to surgical evaluations for threat after sepsis onset, the total days of 2.5–9.1) per 1,000 patients (Table E7). Model
after sepsis was not significantly different vasopressor exposure (median days, 5 [IQR, estimates accounting for hospital variability
among those with or without amputations 3–8] vs. 3 [IQR, 2–4]; P , 0.001) (Figure 4), demonstrate similar estimates as the primary
(median days, 2 [IQR, 0–8] vs. 2 [IQR, 0–7]; total number of vasopressors per day of model (Table E8).
P = 0.600), and the median time to exposure (median vasopressors, 1.4 [IQR,
amputation was 16 days (IQR, 6–40) (Figure 1.0–2.0] vs. 1.0 [IQR, 1.0–1.6]; P , 0.001),
E5). For those who underwent amputations, and daily total cardiovascular SOFA score Discussion
the ICU length of stay was shorter (median, (median cardiovascular SOFA, 19 [IQR,
8 days [IQR, 6–24] vs. 10 [IQR, 5–18]; 12–31] versus 11 [IQR, 6–19]; P = 0.001) Among patients with vasopressor-dependent
P , 0.001) than those without. The risk of (Figure E6) were greater among those with sepsis, extremity threat with surgical
ICU mortality was not observed to be extremity threat, compared with those amputation occurred in 2 per 1,000 patients,

628 AnnalsATS Volume 19 Number 4 | April 2022


ORIGINAL RESEARCH

Table 1. Baseline characteristics of patients with vasopressor-dependent sepsis

Extremity Threat Extremity Threat with


No Extremity Threat without Amputation Amputation
Characteristic (n = 11,829) (n = 205) (n = 26) P*

Age, mean (6 SD), yr 65 (616) 61 (618) 58 (618) 0.006


Sex, n (%)
Male 6,420 (54.3) 113 (55.1) 15 (57.7) 0.914
Race/ethnicity, n (%) 0.151
Black 9,668 (81.7) 159 (77.6) 20 (76.9)
White 1,021 (8.7) 21 (10.2) 5 (19.3)
Other† 1,140 (9.6) 25 (12.2) 1 (3.8)
Tobacco use, n (%) 0.354
Never 2,318 (19.6) 40 (19.5) 6 (23.1)
Quit 1,817 (15.4) 24 (11.7) 4 (15.4)
Current 1,438 (12.2) 33 (16.1) 4 (15.4)
Unknown 6,256 (52.8) 108 (52.7) 12 (46.1)
Charlson Comorbidity Index, median (IQR)‡ 1.0 (0.0–3.0) 1.0 (0.0–3.0) 1.5 (0.0–3.0) 0.377
Baseline comorbidities, n (%)
Diabetes 4,667 (39.5) 85 (41.5) 12 (46.2) 0.663
Cancer 744 (15.3) 9 (11.1) 1 (6.2) 0.353
Congestive heart failure 2,895 (24.5) 68 (33.2) 8 (30.8) 0.013
Peripheral vascular disease 519 (4.4) 19 (9.3) 4 (15.4) ,0.001
Cardiovascular disease 1,149 (9.7) 25 (12.2) 2 (7.7) 0.464
Measures of organ dysfunction at sepsis onset§
Maximum SOFA scorejj, mean (6 SD) 9.7 (63.7) 11.0 (64.0) 11.6 (63.9) ,0.001
Norepinephrine equivalents¶, median 0.7 (0.1–3.0) 1.0 (0.3–4.2) 1.8 (0.3–4.35) ,0.001
(IQR) m g/kg/min
Vital signs at sepsis
onset**§, mean (6 SD)
Temperature ( C) 37.8 (61.1) 37.8 (61.1) 38.0 (61.0) 0.417
Oxygen saturation (%) 92.9 (66.3) 92.6 (66.7) 89.0 (612.5) 0.006
Heart rate (beats per minute) 122 (68) 124 (628) 132 (627) 0.068
Systolic blood pressure (mm Hg) 97 (626) 97 (627) 93 (621) 0.721
Glasgow coma scale 10 (65) 9 (64) 10 (65) 0.026
Microbiology at sepsis onset – n (%)
Blood cultures collected†† 5,820 (49) 101 (49) 15 (58) 0.688
Positive blood cultures 2,087 (26) 53 (35) 11 (55) ,0.001
Laboratory values at sepsis
onset**– median (IQR)
Fibrinogen (mg/dL) 276 (176–425) 231 (150–412) 208 (159–250) 0.106
Platelet count (109/L) 170 (116–234) 168 (117–234) 101 (70–179) 0.006
International normalized ratio 1.5 (1.2–2.0) 1.7 (1.3–2.5) 2.0 (1.4–3.3) ,0.001
Serum lactate (mmol/L) 3.1 (1.7–6.0) 4.6 (2.5–8.7) 4.6 (1.8–8.9) ,0.001
Definition of abbreviations: IQR = interquartile range; SD = standard deviation; SOFA = sequential organ failure assessment.
*Kruskal-Wallis, analysis of variance, or chi-square P value as appropriate comparing across all three cohorts. Tests of significance are
completed across all three groups.

Other race corresponds to Chinese, Filipino, Hawaiian, American Indian/Alaskan, Asian, Hawaiian/other Pacific Islander, Middle Eastern, Native
American, not specified, or Pacific Islander.

Charlson is a method of categorizing comorbidities of patients based on the International Classification of Diseases diagnosis codes found in
administrative data, ranging from 0 to 24.
§
Corresponds to minimum or maximum value, as appropriate, within the 48 hours before to 24 hours after sepsis onset.
jj
SOFA score corresponds to the severity of organ dysfunction, reflecting six organ systems each with a score range of 0 to 4 points
(cardiovascular, hepatic, hematologic, respiratory, neurological, and renal), with a total score range of 0 to 24 points.

Total days of vasopressor exposure includes the 48 hours before to 24 hours after sepsis onset.
**Missingness of variables can be found within Table E4.
††
Blood cultures collected within 48 hours before to 24 hours after sepsis onset.

most commonly in the lower extremities. Extremity threat and surgical effectiveness of vasopressors on 28-day
Extremity threat with no amputation amputation are important complications of mortality among nearly 3,000 patients with
occurred in 17 per 1,000 patients. In the sepsis. Common manifestations may include septic shock and included peripheral
United States, these data correspond to an symmetric peripheral gangrene or purpura ischemia, ranging from cold skin to frank
estimated 1,000 surgical amputations fulminans (6, 12), but there is no consensus tissue loss from ischemia, as exploratory
annually after vasopressor-dependent definition or clinical criteria. Three outcomes (13, 16, 17). The range of ischemia
sepsis (36). randomized controlled trials evaluated the was from 9.5 to 24.2 per 1,000 patients. A

Reitz, Kennedy, Rieser, et al.: Extremity Threat and Amputation after Sepsis 629
ORIGINAL RESEARCH

All patients Threat with


amputation
Baseline risk factors
Diabetes Threat without
amputation
Cancer

Smoking a

Heart failure

PVD

CVD

Sum of risk factors


0 Factors

1– 2 Factors

!2 Factors

0 10 20 30 40 50
No. threatened extremities, per 1,000
with vasopressor-dependent sepsis

Figure 2. Risk of extremity threat with or without amputation in 90 days after vasopressor-dependent sepsis (n = 12,060). Risk of extremity threat
with (red) and without (orange) amputation in the 90 days after sepsis onset. The vertical dashed lines indicate the overall risk of patients with
threat or without amputation. All error bars are 95% confidence intervals. aMissingness for smoking status was common among patients with
(n = 10 [38%]) and without (n = 104 [51%]) amputation (Table E4). CVD = cardiovascular disease; PVD = peripheral vascular disease.

recent meta-analysis of observational studies resulting from an extremity source or those however, can worsen peripheral ischemia.
concluded that nearly 66 per 1,000 patients that occurred before sepsis onset. Third, the microbial source of sepsis,
with septic shock had concurrent extremity There are many potential mechanisms through endotoxin release or direct tissue
ischemia (37). These disparate findings for a threat to extremities after sepsis. A infiltration, may directly cause damage to
indicate a paucity of evidence to accurately dysregulated immune response to infection peripheral tissues (21, 40). These processes
estimate a generalizable, and internally valid results in a broad range of physiologic could be exacerbated among patients with
incidence risk (37). We extended this prior changes (35). First, activation of coagulation preexisting risk factors. Patients with PVD or
work by studying nearly 250,000 patients pathways and endothelial dysfunction leads CHF comorbidities may have blood flow-
with a suspected infection and over 12,000 microcirculatory thrombosis or disseminated limiting atherosclerotic disease and even
with vasopressor-dependent sepsis in 14 intravascular coagulation in as many as 60% ventricular dysfunction (41). When
community and academic hospitals. We of patients with shock (38, 39). Second, a loss challenged with sepsis and hypotension,
generated a unified definition of extremity of vascular tone and intravascular volume these derangements in peripheral blood
threat, used clinical adjudication of events in can lead to circulatory failure and supply may contribute to the higher risks of
the EHR to identify extremity threat with or hypotension. Vasoactive medications extremity threat and amputation seen in this
without amputation, and excluded sepsis required to support perfusion to vital organs, study.

Norepinephrine
equivalents
Age, years
Male
Race (White)
Black
Other
SOFA, sepsis onset
7–8
9–11
12+
Comorbid conditions
CHF
PVD
CVD
Diabetes
1 2 3 4 5
Hazard Ratio

Figure 3. Adjusted risk of extremity threat with or without amputation. Forest plot depicting the hazard ratio (dot) contribution to the risk of extremity
threat with or without amputation. All error bars demonstrate 95% confidence intervals. Associated Schoenfeld residuals, cumulative hazard graphs
stratified by subgroups, and full model are available in the online supplement (Figures E6 and E7 and Table E5). CHF = congestive heart failure;
CVD = cardiovascular disease; PVD = peripheral vascular disease; SOFA = sequential organ failure assessment.

630 AnnalsATS Volume 19 Number 4 | April 2022


ORIGINAL RESEARCH

100 Any Vasopressor 100 Norepinephrine 100 Epinephrine


Patients on Any Vasopressor (%)

Patients on Norepinephrine (%)

Patients on Epinephrine (%)


80 No extremity threat 80 80
Threat with amputation
Threat without amputation
60 60 60

40 40 40

20 20 20

0 0 0
0 7 14 0 7 14 0 7 14
Days from Sepsis Onset Days from Sepsis Onset Days from Sepsis Onset

100 Vasopressin 100 Phenylephrine 100 Dopamine


Patients on Phenylephrine (%)
Patients on Vasopressin (%)

80 80 80

Patients on Dopamine (%)


60 60 60

40 40 40

20 20 20

0 0 0
0 7 14 0 7 14 0 7 14
Days from Sepsis Onset Days from Sepsis Onset Days from Sepsis Onset

Figure 4. Vasopressor exposure. Corresponds to the proportion of alive patients who received any and each individual vasopressor daily, from
the 2 days before to the 14 days after sepsis onset. Green corresponds to patients without extremity threat, red to with threat with amputation,
and orange to with threat without amputation in the 90 days after sepsis onset.

An immortal time bias and a clinician’s 2) surgical consultation by an excluded independent of resuscitation fluid volumes.
view that the patient must survive to benefit subspecialty, or 3) autoamputation in those The risks of extremity threat and amputation
from and therefore be considered for both without a surgical amputation. These may be different among less acute sepsis
surgical consultation and amputation may misclassifications would lead to an patients and critical illness among patients
impact these results. Such effects are underestimate of risks of extremity threat in without sepsis. Finally, we studied patients in
demonstrated by the shorter ICU length of this study. We sought to minimize high-resource settings, and these risks may
stay and equivalent ICU mortality among misclassification by using clinical not be generalizable to sepsis in low- or
those with threat who underwent adjudication, enrichment for specialties most middle-income countries.
amputation when compared with those likely to evaluate extremity threat in our
without amputation. The clinical significance healthcare system, and specialty-independent
of threatened extremities to long-term identification of amputations. Second, the Conclusion
outcomes is difficult to discern. In this reported differences in baseline factors and The evaluation of a threatened extremity
cohort, one in three patients did not survive sepsis treatments were exploratory. resulting in surgical amputation occurred in
to hospital discharge after sepsis onset, and Missingness exists and likely reflects 2 per 1,000 patients with vasopressor-
survival at discharge was more common differences in practice patterns of dependent sepsis. 䊏
among those with when compared with measurement. Multivariable models were
Author disclosures are available with the
those without amputation. limited to variables with less than 1% text of this article at www.atsjournals.org.
missingness; however, the associations with
Limitations extremity threat with or without amputation Acknowledgment: The authors thank the staff
This study has limitations. First, patients must be interpreted with caution. Third, we at UPMC Clinical Analytics, the UPMC Wolff
identified retrospectively in the EHR with restricted our analysis to patients with sepsis Center, and Biostatistics and Data
Management Core at the CRISMA Center in the
sepsis and extremity threat may be who were administered vasoactive Department of Critical Care Medicine at the
misclassified for many reasons, such as 1) no medications, quantified as norepinephrine University of Pittsburgh for curating and
surgical consultation completed or reported, equivalents and not individual vasopressors, managing the data.

Reitz, Kennedy, Rieser, et al.: Extremity Threat and Amputation after Sepsis 631
ORIGINAL RESEARCH

References 22 Molos MA, Hall JC. Symmetrical peripheral gangrene and disseminated
intravascular coagulation. Arch Dermatol 1985;121:1057–1061.
1 Seymour CW, Liu VX, Iwashyna TJ, Brunkhorst FM, Rea TD, Scherag A, 23 Berrey MM, van Burik J-A. Images in clinical medicine. Symmetric
et al. Assessment of clinical criteria for sepsis: for the Third peripheral gangrene. N Engl J Med 2001;344:1593.
International Consensus Definitions for Sepsis and Septic Shock 24 Akamatsu S, Kojima A, Tanaka A, Hayashi K, Hashimoto T. Images in
(Sepsis-3). JAMA 2016;315:762–774. anesthesiology: symmetric peripheral gangrene. Anesthesiology 2013;
2 Fleischmann C, Scherag A, Adhikari NK, Hartog CS, Tsaganos T, 118:1455.
Schlattmann P, et al.; International Forum of Acute Care Trialists. 25 von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC,
Assessment of global incidence and mortality of hospital-treated sepsis. Vandenbroucke JP; STROBE Initiative. The Strengthening the
Current estimates and limitations. Am J Respir Crit Care Med 2016; Reporting of Observational Studies in Epidemiology (STROBE)
193:259–272. statement: guidelines for reporting observational studies. Int J Surg
3 Kadri SS, Rhee C, Strich JR, Morales MK, Hohmann S, Menchaca J, 2014;12:1495–1499.
et al. Estimating ten-year trends in septic shock incidence and mortality 26 Jacobs AK, Anderson JL, Halperin JL. The evolution and future of ACC/
in United States academic medical centers using clinical data. Chest AHA clinical practice guidelines: a 30-year journey: a report of the
2017;151:278–285. American College of Cardiology/American Heart Association Task
4 Shankar-Hari M, Harrison DA, Rubenfeld GD, Rowan K. Epidemiology of Force on Practice Guidelines. J Am Coll Cardiol 2014;64:1373–1384.
sepsis and septic shock in critical care units: comparison between 27 Wozniak K, Sleszycka J, Safianowska A, Wiechno W, Domagala-Kulawik
sepsis-2 and sepsis-3 populations using a national critical care J. Systemic inflammation in peripheral arterial disease with or without
database. Br J Anaesth 2017;119:626–636. coexistent chronic obstructive pulmonary disease: analysis of selected
5 Prescott HC, Angus DC. Enhancing recovery from sepsis: a review. markers. Arch Med Sci 2012;8:477–483.
JAMA 2018;319:62–75. 28 Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC, et al.
6 Huchinson J. Notes of uncommon cases: severe symmetrical gangrene Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10
of the extremities. BMJ 1891;2:8–9. administrative data. Med Care 2005;43:1130–1139.
7 Ferreira AC, Fernandes V. Symmetrical peripheral gangrene. Eur J Vasc 29 Vincent J, Moreno R, Takala J, Willatts S, De Mendonca A, Bruining H,
Endovasc Surg 2018;55:391. et al. The SOFA (sepsis-related organ failure assessment) score to
8 Ruffin N, Vasa CV, Breakstone S, Axman W. Symmetrical peripheral describe organ dysfunction/failure. Intensive Care Med 1996;22:
gangrene of bilateral feet and unilateral hand after administration of 707–710.
vasopressors during septic shock. BMJ Case Rep 2018 [accessed 30 Hill ME, Rosenwaike I. The Social Security Administration’s death master
2018 Nov 19]. Available from: https://www.ncbi.nlm.nih.gov/pmc/ file: the completeness of death reporting at older ages. Soc Secur Bull
articles/PMC5836706/pdf/bcr-2017–223602.pdf. 2001-2002;64:45–51.
9 Park JY, Kanzler M, Swetter SM. Dopamine-associated symmetric 31 Navar AM, Peterson ED, Steen DL, Wojdyla DM, Sanchez RJ, Khan I,
peripheral gangrene. Arch Dermatol 1997;133:247–249. et al. Evaluation of mortality data from the Social Security
10 Dong J, Zhang L, Rao G, Zhao X. Complicating symmetric peripheral Administration death master file for clinical research. JAMA Cardiol
gangrene after dopamine therapy to patients with septic shock. 2019;4:375–379.
J Forensic Sci 2015;60:1644–1646. 32 Reitz KM, Marroquin OC, Zenati MS, Kennedy J, Korytkowski M, Tzeng
11 Hayes MA, Yau EH, Hinds CJ, Watson JD. Symmetrical peripheral E, et al. Association between preoperative metformin exposure and
gangrene: association with noradrenaline administration. Intensive Care postoperative outcomes in adults with type 2 diabetes. JAMA Surg
Med 1992;18:433–436. 2020;155:e200416.
12 Jung KJ, Nho JH, Cho HK, Hong S, Won SH, Chun DI, et al. Amputation 33 Therneau T, Grambsch P. Modeling survival data: extending the Cox
of multiple limbs caused by use of inotropics: case report, a report of 4 model. New York: Springer; 2000.
cases. Medicine (Baltimore) 2018;97:e9800. 34 Therneau T. A Package for Survival Analysis in R. The R Foundation;
13 De Backer D, Biston P, Devriendt J, Madl C, Chochrad D, Aldecoa C, 2020 [accessed 2020 Jun 1]. Available from: https://cran.r-project.org/
et al. Comparison of dopamine and norepinephrine in the treatment of web/packages/survival/index.html.
shock. N Engl J Med 2010;362:779–789. 35 Seymour CW, Kennedy JN, Wang S, Chang CH, Elliott CF, Xu Z, et al.
14 Brown SM, Lanspa MJ, Jones JP, Kuttler KG, Li Y, Carlson R, et al. Derivation, validation, and potential treatment implications of novel
Survival after shock requiring high-dose vasopressor therapy. Chest clinical phenotypes for sepsis. JAMA 2019;321:2003–2017.
2013;143:664–671. 36 Paoli CJ, Reynolds MA, Sinha M, Gitlin M, Crouser E. Epidemiology
15 Jiang L, Sheng Y, Feng X, Wu J. The effects and safety of vasopressin and costs of sepsis in the United States-an analysis based on
receptor agonists in patients with septic shock: a meta-analysis and timing of diagnosis and severity level. Crit Care Med 2018;46:
trial sequential analysis. Crit Care 2019;23:91. 1889–1897.
16 Russell JA, Walley KR, Singer J, Gordon AC, He bert PC, Cooper DJ, 37 Levy JH, Ghadimi K, Faraoni D, van Diepen S, Levy B, Hotchkiss R,
et al.; VASST Investigators. Vasopressin versus norepinephrine et al. Ischemic limb necrosis in septic shock: what is the role of
infusion in patients with septic shock. N Engl J Med 2008;358: high-dose vasopressor therapy? J Thromb Haemost 2019;17:
877–887. 1973–1978.
17 Annane D, Vignon P, Renault A, Bollaert PE, Charpentier C, Martin C, 38 Gando S, Levi M, Toh C-H. Disseminated intravascular coagulation. Nat
et al.; CATS Study Group. Norepinephrine plus dobutamine versus Rev Dis Primers 2016;2:16037.
epinephrine alone for management of septic shock: a randomised trial. 39 Saito S, Uchino S, Hayakawa M, Yamakawa K, Kudo D, Iizuka Y,
Lancet 2007;370:676–684. et al.; Japan Septic Disseminated Intravascular Coagulation
18 Shimbo K, Yokota K, Miyamoto J, Okuhara Y, Ochi M. Symmetrical (JSEPTIC DIC) Study Group. Epidemiology of disseminated
peripheral gangrene caused by septic shock. Case Reports Plast Surg intravascular coagulation in sepsis and validation of scoring
Hand Surg 2015;2:53–56. systems. J Crit Care 2019;50:23–30.
19 Philips V. Acute meningococcemia with symmetric peripheral gangrene: 40 Davis MDP, Dy KM, Nelson S. Presentation and outcome of purpura
report of a case with recovery. Ann Intern Med. 1958;48:864–871. fulminans associated with peripheral gangrene in 12 patients at Mayo
20 Modak D, Guha SK. Symmetrical peripheral gangrene: a rare Clinic. J Am Acad Dermatol 2007;57:944–956.
complication of dengue fever. Indian J Med Sci 2012;66:292–295. 41 Benjamin EJ, Virani SS, Callaway CW, Chamberlain AM, Chang AR,
21 Patial T, Sharma K, Jaswal AS, Thakur V, Negi S. Symmetrical peripheral Cheng S, et al. Heart disease and stroke statistics-2018 update: a
gangrene and tuberculosis: a rare kinship. Int J Mycobacteriol 2017;6: report from the American Heart Association. Circulation 2018;137:
407–409. e67–e492.

632 AnnalsATS Volume 19 Number 4 | April 2022

You might also like