Lactate Clearance Vs Central Venous Oxygen Saturation As Goals of Early Sepsis Therapy a Randomized Clinical Trial

CARING FOR THE CRITICALLY ILL PATIENT

Lactate Clearance vs Central Venous Oxygen Saturation as Goals of Early Sepsis Therapy A Randomized Clinical Trial

Alan E. Jones, MD Context Goal-directed resuscitation for severe sepsis and septic shock has been re- Nathan I. Shapiro, MD, MPH ported to reduce mortality when applied in the emergency department. Stephen Trzeciak, MD, MPH Objective To test the hypothesis of noninferiority between lactate clearance and cen- Ryan C. Arnold, MD tral venous oxygen saturation (ScvO2) as goals of early sepsis resuscitation. Heather A. Claremont, BFA Design, Setting, and Patients Multicenter randomized, noninferiority trial involv- ing patients with severe sepsis and evidence of hypoperfusion or septic shock who Jeffrey A. Kline, MD were admitted to the emergency department from January 2007 to January 2009 at for the Emergency Medicine Shock 1 of 3 participating US urban hospitals. Research Network (EMShockNet) Interventions We randomly assigned patients to 1 of 2 resuscitation protocols. The Investigators ScvO2 group was resuscitated to normalize central venous pressure, mean arterial pressure, and ScvO2 of at least 70%; and the lactate clearance group was resuscitated to HE RATE OF SEVERE SEPSIS HOS- normalize central venous pressure, mean arterial pressure, and lactate clearance of at pitalizations has doubled dur- least 10%. The study protocol was continued until all goals were achieved or for up to ing the last decade with esti- 6 hours. Clinicians who subsequently assumed the care of the patients were blinded mates indicating that at least to the treatment assignment. 750T 000 persons are affected annually Main Outcome Measure The primary outcome was absolute in-hospital mortal- in the United States.1-3 Approxi- ity rate; the noninferiority threshold was set at ⌬ equal to −10%. mately, 500 000 patients with severe Results Of the 300 patients enrolled, 150 were assigned to each group and sepsis in the United States annually are patients were well matched by demographic, comorbidities, and physiological fea- initially treated in emergency depart- tures. There were no differences in treatments administered during the initial 72 ments.4 The Surviving Sepsis Cam- hours of hospitalization. Thirty-four patients (23%) in the ScvO2 group died while in paign international consensus guide- the hospital (95% confidence interval [CI], 17%-30%) compared with 25 (17%; 95% CI, 11%-24%) in the lactate clearance group. This observed difference between lines recommend protocol-driven mortality rates did not reach the predefined −10% threshold (intent-to-treat analysis: treatment that uses quantitative resus- 95% CI for the 6% difference, −3% to 15%). There were no differences in citation for emergency department pa- treatment-related adverse events between the groups. tients with severe sepsis and septic 5 Conclusion Among patients with septic shock who were treated to normalize cen- shock. tral venous and mean arterial pressure, additional management to normalize lactate Quantitative resuscitation refers to clearance compared with management to normalize ScvO2 did not result in signifi- the use of an explicit protocol that tar- cantly different in-hospital mortality. gets predefined physiological or labo- Trial Registration clinicaltrials.gov Identifier: NCT00372502 ratory goals to be achieved within the JAMA. 2010;303(8):739-746 www.jama.com first several hours. This concept was pioneered by Shoemaker et al6 to treat high-risk surgical patients. Results of Author Affiliations: Department of Emergency Medicine, University Hospital, Camden, New Jersey. Carolinas Medical Center, Charlotte, North Carolina Corresponding Author: Jeffrey A. Kline, MD, Depart- a recent meta-analysis indicated a sur- (Drs Jones and Kline and Ms Claremont); Department of ment of Emergency Medicine, 1000 Blythe Blvd, MEB vival benefit associated with the use of Emergency Medicine and Center for Vascular Biology Re- 304, Carolinas Medical Center, Charlotte, NC 28203 search, Beth Israel Deaconess Medical Center, Boston, ([email protected]). Massachusetts (Dr Shapiro); and Departments of Medi- Caring for the Critically Ill Patient Section Editor: Derek See also p 777 and Patient Page. cine, Division of Critical Care Medicine (Dr Trzeciak), and C. Angus, MD, MPH, Contributing Editor, JAMA Emergency Medicine (Drs Trzeciak and Arnold), Cooper ([email protected]).

Downloaded From: https://jamanetwork.com/ on 09/29/2021 LACTATE CLEARANCE IN SEPSIS RESUSCITATION

an early and quantitative resuscitation oxygen delivery during early quantita- http://www.jama.com). Each group strategy applied to heterogeneous popu- tive resuscitation of severe sepsis and received structured quantitative resus- lations of patients with sepsis.7 septic shock. The trial took place from citation while in the emergency depart- The optimal goals for quantitative January 2007 to January 2009 in the ment. The ScvO2 group was resusci- resuscitation of sepsis remain uncer- emergency departments of 3 large ur- tated by sequentially providing therapy tain. It is generally accepted that hemo- ban medical centers in the United States. needed to meet thresholds of central ve- dynamic targets should include some The research protocol was approved by nous pressure, followed by mean arte- measure of the adequacy of cardiac pre- the local institutional review boards and rial pressure, and then ScvO2. The lac- load, such as central venous pressure, and performed in accordance with Good tate clearance group had similarly perfusion pressure, such as mean arte- Clinical Practice guidelines. targeted thresholds in central venous rial pressure.8 A more controversial issue pressure, followed by mean arterial pres- is the method of determining tissue oxy- Participants sure, and then lactate clearance instead gen delivery. Citing a single-center study, Patients with severe sepsis or septic of ScvO2. Standard measures were used the Surviving Sepsis Campaign guide- shock were assessed for inclusion, to ensure appropriate concealment of lines recommend the use of central which required that patients be older group assignment until after informed venous oxygen saturation (ScvO2)or than 17 years with confirmed or pre- consent was obtained. The group as- mixed venous oxygen saturation to assess sumed infection, have 2 or more sys- signment sequence was generated by an the balance of tissue oxygen delivery and temic inflammatory response crite- independent statistician using a paral- consumption9; however, since its pub- ria,17 and have hypoperfusion evidenced lel design, balanced randomization lication in 2001 a substantial amount of by either a systolic blood pressure lower schedule (1:1 ratio of cases and con- controversy about this single-center study than 90 mm Hg after a minimum of 20 trols), using the PROC PLAN function has been generated in the scientific com- mL/kg rapid volume challenge or a in SAS incorporating a sample size of munity.10-12 Additionally, recently pub- blood lactate concentration of at least 300, block sizeequal to 10, with a lished practice surveys have indicated that 36 mg/dL (4 mmol/L). The criteria for seed of 6 457 149 (SAS Institute Inc, the time, expertise, and specialized equip- exclusion from the study were preg- Cary, North Carolina). After written in- ment required to measure ScvO2 collec- nancy, any primary diagnosis other than formed consent was obtained, study staff tively pose a major barrier to the imple- sepsis, suspected requirement for im- opened an opaque sealed envelope con- mentation of protocol-driven quantitative mediate surgery within 6 hours of di- taining the randomization assignment. resuscitation programs.13,14 In contrast, agnosis, an absolute contraindication to Study staff then enforced the study pro- lactate clearance, derived from calculat- chest or neck central venous catheter- tocol. By design, the clinical staff in the ing the change in lactate concentration ization, cardiopulmonary resuscita- emergency departments could not be from 2 blood specimens drawn at differ- tion, transfer from another institution blinded to group assignment; however, ent times, potentially represents a more with a sepsis-specific resuscitative the clinical staff (physicians and nurses) accessible method to assess tissue oxy- therapy underway, and advanced di- who assumed subsequent care of the pa- gen delivery.15,16 rective orders that would restrict the tients in the intensive care units (ICUs) To address the potential utility of lac- study procedure. Using a 24-hour day, were unaware of group assignment. Prior tate clearance as a substitute for ScvO2, 7-day-week method that was previ- to entry in the study, patients were cared we conducted a multicenter, random- ously established for the routine clini- for by emergency physicians, who pro- ized trial among patients presenting to cal care of sepsis patients at each of the vided basic care processes according to the emergency department with se- participating institutions, an alert was participating institutional standards. vere sepsis and septic shock, with the sent to inform clinical care resources primary hypothesis that early resusci- when patients were identified as can- Treatment Interventions tation targeting lactate clearance as the didates for early aggressive resuscita- Appropriate specimens were taken for marker of adequacy of oxygen deliv- tion. This alert was also received by culture, and antibiotics were adminis- ery was noninferior to the currently rec- study staff who responded and screened tered as soon as practical. Blood pres- ommended ScvO2 monitoring for the the patients for study enrollment. Each sure was monitored by either noninva- outcome of in-hospital mortality. enrolled patient or the patient’s le- sive automated cuff sphygmomanometer gally authorized next of kin provided or arterial catheter according to the clini- METHODS written informed consent prior to col- cal team’s preference. All patients re- Study Design lection of data. Patients or family mem- ceived chest or neck central venous cath- This study was a prospective random- bers self-identified their race. eter capable of measuring continuous ized, parallel group, nonblinded, clini- ScvO2 (PreSep, Edwards Lifesciences, Ir- cal trial designed to assess the noninfe- Treatment Assignment vine, California). riority of lactate clearance vs ScvO2 as the Patients were randomly assigned to 1 Patients randomized to the ScvO2 protocol goal that evaluated adequacy of of 2 groups (eFigure, available at group had their central venous cath-

Downloaded From: https://jamanetwork.com/ on 09/29/2021 LACTATE CLEARANCE IN SEPSIS RESUSCITATION

eters connected to a computerized spec- transfused to achieve a hematocrit of at whom the clinicians enacted the study trophotometer (Edwards Lifesciences) least 30%. If the lactate clearance group crossover methods were as- that displayed continuous ScvO2 read- remained lower than 10% after the sessed using an intent-to-treat analy- ings. The patients were then cared for ac- hematocrit was at least 30%, dobuta- sis of the original group assignment. cording to the prespecified treatment mine was initiated and titrated in plan. First, isotonic crystalloid was ad- attempts to achieve a lactate clearance Assessments and Outcome Measures ministered in boluses to achieve a cen- of at least 10%. When treatment was During the study resuscitation treat- tral venous pressure of 8 mm Hg or continued due to lactate clearance less ment period, the patient’s physiologi- higher. Second, the mean arterial pres- than 10%, subsequent lactate measure- cal parameters were measured rou- sure goal of 65 mm Hg or higher, if not ments were performed at a minimum tinely. All data needed to calculate the achieved with fluid administration, was of 1-hour intervals and repeat lactate Simplified Acute Physiology Score targeted by initiating and titrating vaso- clearance calculated. Lactate measure- (SAPS) II,19 Sequential Organ Failure pressors (dopamine or norepineph- ments were performed using venous Assessment (SOFA) score,20 and the rine) to achieve this desired blood pres- whole blood samples using US Food and Mortality in Emergency Department 21 sure goal. Finally, the ScvO2 goal of 70% Drug Administration–approved devices Sepsis were collected. After ICU ad- or higher was targeted after central ve- performed either at the point of care or mission, patients were assessed daily for nous and mean arterial pressure goals in the central hospital laboratory, 72 hours and detailed data were col- were met. If the ScvO2 was lower than according to participating institu- lected. Patients were followed up un- 70% and the hematocrit was lower than tional standards.18 The lactate clear- til hospital discharge or death. 30%, packed red blood cells were trans- ance goal was met by a lactate clear- The primary end point was abso- fused to achieve a hematocrit of at least ance of at least 10% or if both the initial lute in-hospital mortality rate. Second- 30%. If the ScvO2 remained lower than and delayed lactate concentrations were ary end points were ICU length of stay, 70% after the hematocrit was 30% or not elevated (Յ18 mg/dL [2 mmol/L]). hospital length of stay, ventilator-free higher, dobutamine was initiated and ti- Study patients were treated in the days, and new onset multiple organ fail- trated in attempts to achieve an ScvO2 of emergency department during the entire ure. Other end points assessed were the at least 70%. study treatment period, from random- number of resuscitative goals achieved, Patients randomized to the lactate ization to either of the 2 study termina- administered treatments, and pre- clearance group received an identical tion criteria: all treatment goals were defined protocol-related serious ad- central venous catheter capable of mea- achieved or 6 hours had elapsed. Patients verse events. suring continuous ScvO2. The primary were then transferred to an ICU where intervention consisted of the act of not the critical care physicians, unencum- Statistical Analysis connecting the catheter to the comput- bered by the study protocol in any way, The sample size calculation and pri- erized spectrophotometer thus prevent- assumed the care of all patients. The study mary analytical plan centered on the hy- ing display of ScvO2 at any time in the investigators did not provide care for the pothesis of noninferiority of lactate emergency department. These patients patients or influence their care in the ICU. clearance vs ScvO2. For the sample size were cared for according to an identi- As a safety measure, clinical physi- estimate, previously published data sug- cal prespecified treatment plan for cen- cians could elect study group cross- gested a 25% primary outcome event tral venous and mean arterial pressure over. For patients assigned to the ScvO2 rate for the ScvO2 group; we predicted targets as outlined for the ScvO2 group. group, clinicians could order a second no difference between the groups for the However, in the lactate clearance group, lactate concentration to calculate the primary outcome event rate, and we set clinicians used lactate clearance instead lactate clearance. In the lactate clear- the noninferiority margin (⌬)at 22-24 of ScvO2 as the last resuscitation goal ance group, the clinician could re- −10%. The −10% margin was cho- in the protocol and targeted a lactate quest to connect the central venous sen because it represented a two- 15,16 clearance of at least 10%. The lac- catheter to monitor ScvO2. To execute thirds proportion of the active com- tate clearance was defined by the equa- 1 of these options, the clinician was re- parator’s (ScvO2 group) established 9 tion [(lactateinitial −lactatedelayed)/ quired to indicate clinical deteriora- superiority in a similar clinical trial and ϫ lactateinitial] 100%, for which lactate tion based on 1 of the following crite- large databases indicate that the mor- initial was the measurement at the start ria: (1) falling systolic blood pressure tality rate for severe sepsis can be ex- of the resuscitation and lactate delayed or inadequate urine output (Ͻ0.5 pected to vary up to 10% between both was another measurement after a mini- mL/kg per hour); (2) worsening ven- state-of-the art medical facilities and re- mum of 2 hours after resuscitation was tilatory status based on either clinical gions of the United States.25 Using a initiated. If the lactate clearance was not (respiratory rate, oxygen saturation, or 1-sided test of noninferiority, assum- at least 10% at the first delayed mea- oxygen requirement), arterial blood gas, ing a control group mortality rate of surement and the hematocrit was less or mechanical ventilator parameters; or 25% and ␣ = .05, a sample size of 150 than 30%, packed red blood cells were (3) worsened mental status. Patients for per group gave 71% power to deter-

Downloaded From: https://jamanetwork.com/ on 09/29/2021 LACTATE CLEARANCE IN SEPSIS RESUSCITATION

(IQR, 1-48 minutes) in the lactate clear- Figure. Study Flow Diagram ance group and 13 minutes (IQR, 1-55 O 452 Patients were assessed for eligibility minutes) in the Scv 2 group (P=.72). Prior to enrollment in the study, the 152 Excluded mean (SD) amount of intravenous fluid 64 Met specific exclusion criteria 88 Had other reasons administered was 2.3 L(1.4 L) in the 36 Declined to participate lactate clearance group and 2.4 L (1.4 L) 51 Could not provide consent or contact next of kin in the ScvO2 group (P=.37). 1 Was already enrolled in another interventional study TABLE 3 shows the physiological and severity of illness variables during the 300 Randomized first 72 hours of hospitalization. Dur- ing the first 24 hours, both groups of o 150 Randomized to lactate 150 Randomized to Scv 2 group patients tended to show a trend to- clearance group 147 Received Scvo2 protocol 147 Received lactate clearance 3 Did not receive protocol ward slightly worsening severity of ill- protocol 2 Unable to place central line ness in the form of lower systolic blood 3 Did not receive protocol 1 Family withdrew care prior 2 Unable to place central line to protocol completion pressures and higher SOFA scores. The 1 Family withdrew care prior to protocol completion mean initial lactate concentrations were 35.1 mg/dL (3.9 mmol/L) in the lac- 150 Included in the primary analysis 150 Included in the primary analysis tate clearance group and 37.8 mg/dL (4.2 mmol/L) in the ScvO2 group ScvO2 indicates central venous oxygen saturation. (P=.39). The mean (SD) lactate concentration measured at 2 hours in the mine the intervention did not increase had the crossover enacted by a physi- lactate clearance group was 23.4 (23.3) mortality by more than 10%. cian after all study goals had been met. mg/dL (2.6 [2.59] mmol/L) and the me- When appropriate, categorical data The ScvO2 value in this patient was nor- dian lactate clearance at 2 hours was were compared using a ␹2 test or Fisher mal, so no treatment changes were 40% (IQR, 18%-64%). After the initial exact test and continuous variables were made. Because the study treatment pe- 24 hours, survivors in both groups compared with a Mann-Whitney U test riod had ended, this patient was ana- manifested improvements in their or unpaired t test, 2-sided with PϽ.05 lyzed in the lactate clearance group for physiological and severity of illness considered significant (StatsDirect v both the intent-to-treat and per- scores. We observed that lactate mea- 2.7.7, Cheshire, England). Two pre- protocol analyses. None were lost to fol- surements did not worsen over the ini- planned blinded interim safety analy- low-up or voluntarily withdrew from tial 72 hours (Table 3), suggesting that ses were performed after one-third and the study, leaving 150 patients in each the initial lactate levels were often the two-thirds of the participants were en- group for the intent-to-treat analysis most abnormal. rolled, and these results were re- and 147 in each group for the per- There were no differences in the ad- viewed by an independent safety moni- protocol analysis. ministered treatments through the ini- tor who had the authority to terminate The baseline characteristics of the 2 tial 72 hours of hospitalization as shown the study for safety concerns. No sample groups are shown in TABLE 1 and in in TABLE 4. During the emergency de- size adjustments were planned during TABLE 2. There were no significant dif- partment–based 6-hour resuscitation the interim analyses. We planned both ferences between the groups in demo- period, patients received approxi- intent-to-treat and per-protocol analy- graphics, comorbid conditions, sever- mately 4.5 L of crystalloid, 221 pa- ses after study completion. ity of illness scores, or suspected site tients (74%; 95% confidence interval of infection. The lungs were the most [CI], 68%-79%) required vasopres- RESULTS common source of infection, and 38% sors for hypotension, and 79 patients Of 452 patients who underwent screen- of patients had a blood specimen that (26%; 95% CI, 21%-32%) required me- ing for eligibility, 300 underwent ran- yielded growth of bacteria, whereas 84% chanical ventilation. Notably, only 29 domization (FIGURE). Of these 300 pa- had at least 1 culture specimen that was patients (10%; 95% CI, 7%-14%) re- tients, the same pattern of protocol positive. The median time from emer- quired either dobutamine infusion or noncompletion was observed in each gency department triage to eligibility packed red blood cell transfusion group: 2 because a chest or neck cen- was 111 minutes (interquartile range during the initial 6 hours of treat- tral venous catheter could not be placed [IQR], 56-192 minutes) in the lactate ment. Activated protein C administra- and 1 because family members had de- clearance group and 105 minutes (IQR, tion was only administered in 5 patients cided to withdraw support before the 60-175 minutes) in the ScvO2 group (2%; 95% CI, 1%-4%). research procedure was completed. One (P=.67); the median time from eligi- Among the 294 (147 per group) pa- patient in the lactate clearance group bility to study entry was 14 minutes tients included in the per-protocol analy-

Downloaded From: https://jamanetwork.com/ on 09/29/2021 LACTATE CLEARANCE IN SEPSIS RESUSCITATION

sis, the central venous pressure goal was Table 1. Patient Demographics and Clinical Characteristicsa achieved in 133 patients (91%; 95% CI, No. (%) of Patients 85%-95%) in the lactate clearance group and 133 (91%; 95% CI, 85%-95%) in the Lactate Clearance Group ScvO2 Group Variable (n = 150) (n = 150) ScvO2 group (P=.99); the mean arterial Age, mean (SD), y 59.8 (17.6) 61.6 (17.6) pressure goal was achieved in 142 pa- Race tients (97%; 95% CI, 92%-99%) in the White 88 (59) 77 (51) lactate clearance group and 142 (97%; Black 47 (31) 56 (37) 95% CI, 92%-99%) in the ScvO2 group Sex (P=.99); and the lactate clearance goal Men 83 (55) 80 (53) was met in 139 patients (95%; 95% CI, Women 67 (45) 70 (47) Comorbidities 90%-98%) in the lactate clearance group Diabetes mellitus 45 (30) 57 (38) and the ScvO2 goal was met in 136 (93%; Chronic obstructive pulmonary disease 25 (17) 25 (17) 95% CI, 87%-96%) patients in the ScvO2 Human immunodeficiency virus infection 12 (8) 13 (9) group (P=.67). The median time from End-stage renal disease 15 (10) 14 (9) patient triage in the emergency depart- Active malignancy 42 (28) 32 (21) ment to first antibiotic administration Organ transplant 5 (3) 4 (6) was 115 (IQR, 62-180) minutes in the Indwelling vascular line 6 (4) 10 (7) lactate clearance group and 115 (IQR, 66- Nursing home resident 28 (19) 28 (19) 170) minutes in the ScvO group (P=.98). Disease severity 2 SAPS II score 44.8 (18.4) 44.1 (17.3) The primary and secondary study SOFA score 6.7 (3.6) 6.6 (3.5) outcome analysis is outlined in TABLE 5. MEDS score 10.9 (3.9) 10.6 (3.4) In the intent-to-treat analysis the in- Suspected source of infection hospital mortality rate was 17% (25 of Pulmonary 48 (32) 54 (36) 150 [95% CI, 11%-24%]) in the lac- Urinary tract 40 (27) 39 (26) tate clearance group compared with Intra-abdominal 34 (23) 24 (16) 23% (34 of 150 [95% CI, 17%-30%]) Skin/soft tissue 19 (13) 23 (15) Blood 8 (5) 9 (6) in the ScvO2 group. The difference in these mortality rates was 6% (95% CI, Unknown 13 (9) 9 (6) Features of sepsis −3% to 15%). The lower limit of this CI Lactate Ն4 61 (41) 56 (37) is well above the −10% predefined non- Shockb 121 (81) 123 (82) inferiority threshold, confirming the Culture positive 123 (82) 127 (85) primary hypothesis of noninferiority be- Blood culture positive 62 (41) 53 (35) tween the lactate clearance and ScvO2 Gram positive 33 (22) 36 (24) groups for in-hospital mortality. These Gram negative 29 (19) 17 (11) results did not change substantially in Abbreviations: MEDS, mortality in emergency department sepsis; SAPS, Simplified Acute Physiology Score; ScvO2, central venous oxygen saturation; SOFA, Sequential Organ Failure Assessment. the per-protocol analysis. aContinuous data are compared using an unpaired t test; categorical data, using the ␹2 test. There were no differences in the ob- bShock is defined as a systolic blood pressure of 90 mm Hg or less after receiving a 20 mL/kg-fluid bolus. served rates of predefined protocol- related serious adverse events be- Table 2. Systemic Inflammatory Response Criteria and Dysfunctional Organ Systems tween the lactate clearance (9 of 150 No. (%) of Patients [6%; 95% CI, 3%-11%]) and ScvO2 (11 Lactate Clearance Group ScvO2 Group of 150 [7%; 95% CI, 4%-13%]) groups Variable (n = 150) (n = 150) (P=.81). SIRS criteria Abnormal white blood cell count 117 (78) 104 (69) COMMENT Elevated heart rate 100 (67) 108 (72) The results of this large multicenter ran- Elevated respiratory rate 96 (64) 89 (59) domized controlled trial of 2 resuscita- Abnormal body temperature 61 (41) 64 (43) Organ dysfunction tion protocols for early sepsis resuscita- Respiratory 85 (57) 86 (57) tion indicate that a protocol targeting Liver 43 (29) 43 (29) lactate clearance of at least 10% as evi- Neurological 47 (31) 58 (39) dence of adequate tissue oxygen deliv- Coagulation 51 (34) 39 (26) ery produces a similar short-term sur- Cardiovascular 132 (88) 128 (85) vival rate as a protocol using ScvO2 Renal 110 (73) 109 (73)

monitoring. Patients in the group resus- Abbreviations: ScvO2, central venous oxygen saturation; SIRS, systemic inflammatory response syndrome.

Downloaded From: https://jamanetwork.com/ on 09/29/2021 LACTATE CLEARANCE IN SEPSIS RESUSCITATION

citated to a lactate clearance of 10% or ence, −3% to 15%) exceeding the −10% nous blood as a safe and efficacious al- higher had 6% lower in-hospital mor- predefined noninferiority threshold. ternative to a computerized spectropho- tality than those resuscitated to an ScvO2 These data support the substitution of tometric catheter in the resuscitation of of at least 70% (95% CI for this differ- lactate measurements in peripheral ve- sepsis. To our knowledge, this is the larg- est randomized trial of emergency department–based early quantitative re- Table 3. Physiological and Severity of Illness Measurements suscitation for sepsis conducted to date Lactate Clearance Group ScvO2 Group P and the first such trial to investigate the Variable by Study Time Point, ha (n = 150) (n = 150) Valueb Systolic blood pressure, mm Hg relative value of different goals of early, 0 91 (24.6) 92 (21.0) .62 emergency department–resuscitation 24 73 (20.8) 79 (15.3) .01 strategies. 48 94 (22.1) 95 (21.3) .91 The physiological basis for lactate 72 103 (19.9) 103 (19.1) .87 clearance presumes that circulatory Heart rate, beats/min shock causes inadequate oxygen deliv- 0 103 (23.6) 106 (24.4) .36 ery, resulting in mitochondrial hy- 24 117 (23.6) 119 (21.9) .37 poxia. Under hypoxic conditions, mito- 48 106 (19.7) 107 (20.5) .51 chondrial oxidative phosphorylation 72 105 (22.1) 103 (20.1) .56 fails, and energy metabolism becomes de- Central venous pressure, mm Hg 26 0 11 (6.5) 11 (6.2) .55 pendent on anaerobic glycolysis. An- 24 16 (7.8) 15 (6.6) .47 aerobic glycolysis sharply increases the 48 13 (6.4) 14 (6.5) .45 production of cellular lactate, which dif- 72 12 (6.5) 14 (8.3) .14 fuses into the blood during prolonged cell Central venous oxygen saturation, % hypoxia. In patients with a clinical pic- 0 74 (12.3) ture of severe infection, the blood lac- 24 65 (19.9) 64 (13.1) .71 tate concentration varies in proportion 48 70 (16.6) 68 (14.8) .53 to the ongoing deficit in tissue oxygen- 72 ation, and the ability of the patient to re- Lactate level, mg/dLc duce the blood lactate concentration in- 0 35.1 (28.1) 37.8 (27.7) .39 dicates restoration of oxygen delivery 24 with resuscitation.27 Previous work has 48 found that a lactate clearance of 10% or 72 35.1 (30.3) 36.9 (29.6) .67 more predicts survival from septic shock, SOFA score, median (IQR) 15,16 0 6 (4-9) 6 (4-9) .71 providing the rationale for this goal. 24 8 (5-11) 7 (5-11) .98 In addition, we constructed the protocol 48 4 (2-7) 5 (2-7) .90 to consider 2 normal lactate levels (Յ18 72 3 (1-6) 3 (1-6) .62 mg/dL [2 mmol/L]) at least 2 hours apart SAPS II score as evidence of ongoing adequate tissue 0 44.8 (18.4) 44.1 (17.3) .69 oxygenation. The rationale for includ- 24 ing this criterion was that clinically it 48 would make no sense to attempt to clear 72 33.4 (14.1) 34.6 (17.2) .54 a value that is already normal and that 2 MEDS score 0 10.9 (3.9) 10.6 (3.4) .46 normal values provide a reasonable clini- 24 cal signal that effective resuscitation has 48 prevented worsening of tissue oxygen- 72 8.4 (4.2) 8.4 (4.5) .93 ation and anaerobic metabolism. Glasgow coma scale We have previously documented that 0 13 (4.1) 13 (3.7) .67 many clinicians perceive a significant 24 12 (4.3) 12 (3.9) .68 degree of technical difficulty associ- 48 13 (3.7) 13 (3.5) .91 ated with the use of computerized spec- 72 15 (3.1) 14 (4.0) .04 trophotometric catheters to monitor Abbreviations: IQR, interquartile range; MEDS, Mortality in Emergency Department Sepsis; SAPS, Simplified Acute Physi- 28 ScvO2. These devices require equip- ology Score; ScvO2, central venous oxygen saturation; SOFA, Sequential Organ Failure Assessment. SI conversion factor: to convert lactate concentration from mg/dL to mmol/L, multiply by 0.111. ment and expertise that are not avail- aValues represent the mean (SD) measurements at enrollment (0) and the most abnormal values at each hour of measurement, except for the SOFA score. Lactate values and SAPS and MEDS scores were not recorded at 24 and 48 hours. able in many tertiary care emergency b ␹2 13 Continuous data are compared using an unpaired t test; categorical variables, using the test. departments. Use of ScvO2 monitor- cLactate levels at 72 hours represent worst value over the initial 72 hours of hospitalization. ing catheters requires preplanned train-

Downloaded From: https://jamanetwork.com/ on 09/29/2021 LACTATE CLEARANCE IN SEPSIS RESUSCITATION

ing and real-time calibration and vidual treatment goals during the study suggested that the potential difference troubleshooting that can divert atten- treatment period. Third, this study was in protocol actions directly attribut- 28 tion from the patient. We thus sub- conducted at 3 institutions that had es- able to using lactate clearance vs ScvO2 mit that the need exists for a simpler tablished emergency department– was small, because only 10% of pa- and more generalizable method to based quantitative resuscitation pro- tients went on to receive dobutamine monitor the adequacy of tissue oxy- grams for sepsis prior to initiation of the or packed red blood cell transfusion. gen delivery as a research imperative in study. Therefore, our results may not Fifth, we did not have a mechanism to the treatment of patients with severe in- be generalizable to centers that do not query ICU admission for potentially fection. Our results address this un- routinely perform early quantitative re- missed cases; thus, we may have missed met need by providing data that jus- suscitation. Fourth, if other influ- patients who met criteria because a tify the use of lactate clearance instead ences on care were ignored, it could be clinical alert was not activated. Finally, of continuous ScvO2 monitoring. The sample size assumed a mortality rate of 25% in the active comparator Table 4. Administered Treatments and Resuscitation Goals (ScvO2) group. The observed mortality No. (%) of Patients was 23% in the ScvO2 group, indicating Lactate Clearance Group ScvO2 Group P that we maintained approximately 80% Intervention, h (n = 150) (n = 150) Valuea power to detect a true difference. Al- Crystalloid volume, mean (SD), L though our observed overall mortality 0-Ͻ6 4.5 (2.36) 4.3 (2.21) .55 rate (20%) is lower than the 37% over- 6-72 12.4 (6.15) 11.8 (6.41) .44 Vasopressor administration all rate reported in the sentinel emer- 0-Ͻ6 108 (72) 113 (75) .60 gency department–based resuscitation 6-72 100 (67) 108 (72) .45 9 study, it is nearly identical to mortality Dobutamine administration rates that both individual investigators 0-Ͻ6 5 (3) 8 (5) .57 from our group22-24 and others29 have pre- 6-72 10 (7) 13 (9) .66 viously reported in effectiveness stud- PRBC transfusion ies conducted in more heterogeneous 0-Ͻ6 11 (7) 5 (3) .20 populations. We believe that that the 6-72 35 (23) 31 (21) .78 Mechanical ventilation mortality rate represents a contempora- 0-Ͻ6 40 (27) 39 (26) .99 neous and accurate estimate of the true 6-72 69 (46) 75 (50) .56 mortality rate for patients with severe Activated protein C sepsis and septic shock treated with an 0-Ͻ600 early quantitative resuscitation proto- 6-72 3 (2) 2 (1) .68 col in the emergency department. Parenteral corticosteroids Ͻ Several limitations of our study war- 0- 6 18 (12) 26 (17) .25 rant discussion. By its design, the 6-72 59 (39) 51 (34) .40 Abbreviations: PRBC, packed red blood cell; ScvO2, central venous oxygen saturation. groups could not be blinded, allowing aContinuous variables are compared using unpaired t test; categorical variables, using ␹2 test except activated protein C for possible treatment bias. Our proto- which was analyzed using Fisher exact test. col was designed with safeguards to minimize this potential effect. For ex- Table 5. Hospital Mortality and Length of Stay ample, every participant received iden- Proportion tical central venous catheters so that Lactate Clearance Difference (95% Group ScvO2 Group Confidence P group assignment would not be easily Variable (n = 150) (n = 150) Interval) Valueb identifiable. Also, investigators in- In-hospital mortality, No. (%)a volved with the study were not al- Intent to treat 25 (17) 34 (23) 6 (−3 to 15) lowed to provide care for the partici- Per protocol 25 (17) 33 (22) 5 (−3 to 14) pants or influence their care in the ICU. Length of stay, mean (SD), d Second, we did not have a method to ICU 5.9 (8.46) 5.6 (7.39) .75 assess whether an indicated therapeu- Hospital 11.4 (10.89) 12.1 (11.68) .60 Hospital complications tic action was performed in response to Ventilator-free days, mean (SD) 9.3 (10.31) 9.9 (11.09) .67 a parameter below the intended goal Multiple organ failure, No. (%) 37 (25) 33 (22) .68 (eg, if central venous pressure was 4 Care withdrawn, No. (%) 14 (9) 23 (15) .15

mm Hg, we did not record whether a Abbreviations: ICU, intensive care unit; ScvO2, central venous oxygen saturation. fluid bolus was given). Rather, we only aPrimary study end point. bContinuous data are compared using an unpaired t test; categorical variables, using the ␹2 test. assessed for compliance with indi-

Downloaded From: https://jamanetwork.com/ on 09/29/2021 LACTATE CLEARANCE IN SEPSIS RESUSCITATION

it is possible that knowledge of the thoughtful critique of this report and who received no 14. Carlbom DJ, Rubenfeld GD. Barriers to imple- remuneration. We are indebted to the emergency de- menting protocol-based sepsis resuscitation in the study prompted clinical care provid- partment and ICU nurses and physicians at Carolinas emergency department—results of a national survey. ers to have a heightened awareness and Medical Center, Beth Israel Deaconess Medical Cen- Crit Care Med. 2007;35(11):2525-2532. ter, and Cooper University Hospital who willingly as- 15. Arnold RC, Shapiro NI, Jones AE, et al. Multi- provide differential treatment pat- sisted in the care of the study patients. Finally, we hum- center study of early lactate clearance as a determi- terns (ie, a Hawthorne-like effect). bly express our deepest gratitude to the patients who nant of survival in patients with presumed sepsis. Shock. In conclusion, in this randomized so willingly participated in this study. 2009;32(1):35-39. 16. Nguyen HB, Rivers EP, Knoblich BP, et al. Early trial, we found no difference in mor- REFERENCES lactate clearance is associated with improved out- tality for patients with severe sepsis and come in severe sepsis and septic shock. Crit Care Med. 1. Angus DC, Linde-Zwirble WT, Lidicker J, Clermont 2004;32(8):1637-1642. septic shock resuscitated with a proto- G, Carcillo J, Pinsky MR. Epidemiology of severe sep- 17. Bone RC, Balk RA, Cerra FB, et al; The ACCP/ col that used lactate clearance com- sis in the United States: analysis of incidence, out- SCCM Consensus Conference Committee. American pared with a protocol that used ScvO come, and associated costs of care. Crit Care Med. College of Chest Physicians/Society of Critical Care 2 2001;29(7):1303-1310. Medicine. Definitions for sepsis and organ failure and as the method of measuring total body 2. Dombrovskiy VY, Martin AA, Sunderram J, Paz HL. guidelines for the use of innovative therapies in sep- oxygen metabolism. Rapid increase in hospitalization and mortality rates sis. The ACCP/SCCM Consensus Conference for severe sepsis in the United States: a trend analysis Committee. Chest. 1992;101(6):1644-1655. Author Contributions: Dr Jones had full access to all from 1993 to 2003. Crit Care Med. 2007;35(5): 18. Shapiro NI, Fisher C, Donnino M, et al. The fea- of the data in the study and takes responsibility for 1244-1250. sibility and accuracy of point-of-care lactate measure- the integrity of the data and the accuracy of the data 3. Martin GS, Mannino DM, Eaton S, Moss M. The ment in emergency department patients with sus- analysis. epidemiology of sepsis in the United States from 1979 pected infection [published online ahead of print Study concept and design: Jones, Shapiro, Trzeciak, through 2000. N Engl J Med. 2003;348(16):1546- September 1, 2009]. J Emerg Med. doi:10.1016/j Kline. 1554. .jemermed.2009.07.021. Acquisition of data: Jones, Shapiro, Trzeciak, Arnold, 4. Wang HE, Shapiro NI, Angus DC, Yealy DM. Na- 19. LeGall JR, Lemeshow S, Saulnier F. A new Sim- Claremont, Kline. tional estimates of severe sepsis in United States emer- plified Acute Physiology Score (SAPS II) based on a Analysis and interpretation of data: Jones, Shapiro, gency departments. Crit Care Med. 2007;35(8): European/North American multicenter study. JAMA. Trzeciak, Kline. 1928-1936. 1993;270(24):2957-2963. Drafting of the manuscript: Jones, Kline. 5. Dellinger RP, Levy MM, Carlet JM, et al; Interna- 20. Vincent JL, Moreno R, Takala J, et al. The SOFA Critical revision of the manuscript for important in- tional Surviving Sepsis Campaign Guidelines Committee; (Sepsis-related Organ Failure Assessment) score to de- tellectual content: Jones, Shapiro, Trzeciak, Arnold, American Association of Critical-Care Nurses; Ameri- scribe organ dysfunction/failure. Intensive Care Med. Claremont, Kline. can College of Chest Physicians; American College of 1996;22(7):707-710. Statistical analysis: Jones, Kline. Emergency Physicians; Canadian Critical Care Society; 21. Shapiro NI, Wolfe RE, Moore RB, Smith E, Burdick Obtained funding: Jones, Kline. European Society of Clinical Microbiology and Infec- E, Bates DW. Mortality in Emergency Department Sep- Administrative, technical, or material support: Jones, tious Diseases; European Society of Intensive Care sis (MEDS) score: a prospectively derived and vali- Shapiro, Arnold, Claremont, Kline. Medicine; European Respiratory Society; International dated clinical prediction rule. Crit Care Med. 2003; Study supervision: Jones, Kline. Sepsis Forum; Japanese Association for Acute Medicine; 31(3):670-675. Financial Disclosures: Dr Jones reports receiving re- Japanese Society of Intensive Care Medicine; Society of 22. Jones AE, Focht A, Horton JM, Kline JA. Prospec- search support from Critical Biologics Corp and Critical Care Medicine; Society of Hospital Medicine; Sur- tive external validation of the clinical effectiveness of Hutchinson Technology and having served on an ad- gical Infection Society; World Federation of Societies of an emergency department-based early goal-directed visory board for Brahms and Siemens in 2009. Dr Intensive and Critical Care Medicine. Surviving Sepsis therapy protocol for severe sepsis and septic shock. Trzeciak reports that he receives research support from Campaign: international guidelines for management of Chest. 2007;132(2):425-432. Ikaria, Novo Nordisk, serves as a consultant to Spec- severe sepsis and septic shock: 2008. Crit Care Med. 23. Trzeciak S, Dellinger RP, Abata NL, et al. Trans- tral Diagnostics, and has received 3 honoraria from 2008;36(1):296-327. lating research to clinical practice: a 1-year experi- Edwards Lifesciences prior to 2005, but has not re- 6. Shoemaker WC, Appel PL, Kram HB, Waxman K, ence with implementing early goal-directed therapy ceived any personal remuneration from any commer- Lee TS. Prospective trial of supranormal values of sur- for septic shock in the emergency department. Chest. cial interest since 2005. Dr Shapiro reports that he has vivors as therapeutic goals in high-risk surgical patients. 2006;129(2):225-232. received research support from Biosite, Hutchinson Chest. 1988;94(6):1176-1186. 24. Shapiro NI, Howell MD, Talmor D, et al. Imple- Technology, and Eli Lilly. Dr Kline reported that he is 7. Jones AE, Brown MD, Trzeciak S, et al; Emer- mentation and outcomes of the Multiple Urgent Sep- the inventor on US patent 7,083,574, a breath- gency Medicine Shock Research Network Investigators. sis Therapies (MUST) protocol. Crit Care Med. 2006; based device that monitors patient response to resus- The effect of a quantitative resuscitation strategy on 34(4):1025-1032. citation. Dr Arnold and Ms Claremont reported no fi- mortality in patients with sepsis: a meta-analysis. Crit 25. Agency for Healthcare Research and Quality nancial conflicts. Care Med. 2008;36(10):2734-2739. Healthcare Cost and Utilization Project National Emer- Funding/Support: This project was supported by grant 8. Hollenberg SM, Ahrens TS, Annane D, et al. Prac- gency Department Sample 2006. http://hcupnet.ahrq GM76652 from the National Institutes of Health (NIH), tice parameter for hemodynamic support of sepsis in .gov/HCUPnet.jsp?Id=CA7BFB8F2750623A&Form National Institute of General Medical Sciences (Dr adult patients: 2004 update. Crit Care Med. 2004; =SelCROSSTAB&JS=Y&Action=%3E%3ENext Jones). Dr Trzeciak was supported by grant GM83211 32(9):1928-1948. %3E%3E&_Oneway=Yes. Accessed December 30, from the NIH. Dr Shapiro was supported by grants 9. Rivers E, Nguyen B, Havstad S, et al; Early Goal- 2009. HL091757 and GM076659 from the NIH. Directed Therapy Collaborative Group. Early goal- 26. Watts JA, Kline JA. Bench to bedside: the role of Role of the Sponsors: The sponsor of this study re- directed therapy in the treatment of severe sepsis and mitochondrial medicine in the pathogenesis and treat- viewed and approved the study design but had no role septic shock. N Engl J Med. 2001;345(19):1368- ment of cellular injury. Acad Emerg Med. 2003; in the conduct of the study; in the collection, man- 1377. 10(9):985-997. agement, analysis, or interpretation of the data; or in 10. Peake S, Webb S, Delaney A. Early goal-directed 27. Weil MA, Afifi AA. Experimental and clinical stud- the preparation, review, or approval of the article. therapy of septic shock: we honestly remain skeptical. ies on lactate and pyruvate as indicators of the sever- Online-Only Material: The eFigure is available at http: Crit Care Med. 2007;35(3):994-995. ity of acute circulatory failure (shock). Circulation. 1970; //www.jama.com. 11. Ho BC, Bellomo R, McGain F, et al. The inci- 41(6):989-1001. Additional Contributions: We thank J. Lee Garvey, MD, dence and outcome of septic shock patients in the ab- 28. Jones AE, Shapiro NI, Roshon M. Implementing who functioned as the study monitor; Jackeline Hern- sence of early-goal directed therapy. Crit Care. 2006; early goal-directed therapy in the emergency set- andez, MD, and Nikita Young, BS (all from Carolinas 10(3):R80. ting: the challenges and experiences of translating re- Medical Center, Charlotte, North Carolina), and David 12. Perel A. Bench-to-bedside review: the initial he- search innovations into clinical reality in academic and Lundy, MD (Cooper University Hospital, Camden, New modynamic resuscitation of the septic patient accord- community settings. Acad Emerg Med. 2007;14 Jersey), for their invaluable assistance with monitoring ing to Surviving Sepsis Campaign guidelines—does one (11):1072-1078. the study and enrolling patients; Charlie Johnson, BS size fit all? Crit Care. 2008;12(5):223. 29. Nguyen HB, Corbett SW, Steele R, et al. Imple- (paid consultant from Studymaker LLC, Newton, Mas- 13. Jones AE, Kline JA. Use of goal-directed therapy mentation of a bundle of quality indicators for the early sachusetts) for assistance with database manage- for severe sepsis and septic shock in academic emer- management of severe sepsis and septic shock is as- ment. We also thank R. Phillip Dellinger, MD (Cooper gency departments. Crit Care Med. 2005;33(8): sociated with decreased mortality. Crit Care Med. 2007; University Hospital, Camden, New Jersey), for his 1888-1889. 35(4):1105-1112.

Downloaded From: https://jamanetwork.com/ on 09/29/2021