Criteria abstracted from The Users' Guide to Medical Literature, from the Health Information Research Unit and Clinical Epidemiology and Biostatistics, McMaster University

Highlighted lines and questions below provide links to the pertinent description of criteria in The EBM User's Guide, now available at the Canadian Centres for Health Evidence

Article Reviewed:

Please visit the new Evidence Based Journal Club Reviews

Early hyperlactataemia in critically ill children

Hatherill M, McIntyre AG, Wattie M, Murdoch IA.

Intensive Care Med 2000; 26:314-318. [abstract]

Reviewed by Robert Graham, MD, Children's Hospital, Boston MA

Review posted August 9, 2001

I. What is being studied?

Study objective:

The objective was to examine the relationship between early hyperlactatemia, acidosis, organ failure, and mortality in children admitted to the intensive care unit.

Study design

This was a prospective, observational study.

II. Are the results of the study valid?

Note: These questions follow from Randolph AG et al. Understanding articles describing clinical prediction tools. Crit Care Med 1998;26:1603-1612. [abstract]

1. Was a representative group of patients completely followed up?

The investigators screened 705 pediatric intensive care admissions over an eighteen month period for hyperlactatemia. After excluding post-operative patients and those with inherited metabolic diseases, i.e., scenarios which might independently contribute to elevated blood lactate levels, they enrolled 50 children with hyperlactatemia. All other diagnostic categories were included; thus the study population should be representative of the heterogeneous ICU population. The investigators achieved 100% follow-up of their target population.

2. Were all potential predictors included?

The scope of this investigation was limited to serum lactate and acid-base parameters as predictors for organ failure and mortality. There is significant debate about the physiologic significance of hyperlactatemia and its relation to oxygen supply dependency (1); several factors affecting either the clearance of lactate or augmentation of pyruvate flux may confound the interpretation of hyperlactatemia. The investigators recognized the potential limitations of lactate as a marker of tissue hypoxia. By studying a heterogeneous group of critically ill children, however, they attempted to establish an association between early hyperlactatemia and degrees of organ failure and mortality, irrespective of its origin. Causality, obviously, was not implied as lactate is a physiologic marker.

Lactate measurements are a continuous variable, posing the issues of establishing multiple ranges or levels as independent variables. The investigators defined hyperlactatemia as greater than 2 mmol/L. Additionally, by reporting the cumulative average lactate levels as well as differentiating persistent (> 24hrs) versus non-sustained hyperlactatemia, they established their own standard for a predictive lactate level and incorporated time as a factor to account for variable significance between predictive ranges.

3. Did the investigators test the independent contribution of each predictor variable?

Yes, the investigators tested the independent contribution of admission lactate, pH, and base excess as well as peak lactate and cumulative average lactate levels against mortality. Only admission and peak lactate levels, however, were tested against multiorgan system failure (MOSF) score.

4. Were outcome variables clearly and objectively defined?

Yes, mortality and morbidity, as assessed by the MOSF score, were the primary outcome measures. PRISM scores, median duration until death, and length of stay were also reported but no correlations were made with the predictor variables.

III. What are the results?

1. What is(are) the prediction tool(s)?

The investigators demonstrated a statistically significant relationship between greater MOSF score, higher peak serum lactate levels, and greater cumulative average lactate levels as predictive indices of ICU mortality. Neither admission lactate (if already > 2 mmol/L), pH, nor base excess differed between survivors and nonsurvivors.

2. How well does the model categorize patients into different levels of risk?

The model is compelling when categorizing patients into hyperlactactemia (> 2 mmol/L) versus normolactate controls. Comparison of pre and post-test probabilities of death revealed a likelihood ratio of 15. Post-test likelihood of mortality given admission hyperlactatemia is 64% versus 6% with a normal lactate, suggesting a high diagnostic importance of this test when categorizing patients.

Of the patients with admission hyperlactatemia, those with persistently elevated levels (> 2 mmol/L) at 24 hours also had increased mortality with a likelihood ratio of 7. This is seemingly less significant; however, the higher incidence of mortality in this group makes discrimination of high versus higher risk patients more difficult, as is evident with a positive predictive value of 93% and negative predictive value of only 70%. The area under the receiver operating characteristic (ROC) curve for hyperlactatemia after 24 hours was 0.86 (95% CI: 0.73, 0.99), giving the probability that the nonsurvivor will have a higher lactate than the survivor. The authors do not specify how they arrived at a cutoff of > 2 mmol/L. An ROC plot for all lactate levels both at admission and 24 hours would be interesting and might help identify an even better level for discriminating high risk patients.

While the authors did not report whether any patients with lactate < 2 mmol/L at admission subsequently developed hyperlactatemia at 24 hours, calculating an LR for the 24 hour hyperlactate group for the entire screened sample is even more impressive; A potential likelihood ratio of 113 (95% CI: 27, 469) with PPV and NPV both 93% (2). Clearly, a very high risk subgroup. Analysis of risk based upon admission lactate level and underlying diagnosis, septic versus non-septic, revealed no significant difference between groups and no difference between survivors and non-survivors within each category. However, interpretation of this data is limited because of the inclusion of admission lactate levels < 2mmol/L and due to the variability in lactate levels. The investigators also only reported median admission lactate when comparing etiology as opposed to peak and cumulative levels, which were significant in the overall sample. The conclusion that hyperlactatemia represents an equivalent risk of mortality, regardless of disease state, cannot be drawn.

The patients requiring hemodynamic support at the time of ICU admission showed a trend toward higher initial lactate levels. While expected due to relationship between tissue hypoperfusion and potential need for inotropes as well as possible catecholamine-induced glycolysis, the investigators did not follow-up on these, likely, high-risk patients. Subgroup analysis might reveal that hyperlactatemia and the need for inotropes were not independent variables.

3. How confident are you in the estimates of the risk?

The use of lactate as a predictive index, whether isolated or serial measurements, has its limitations. The range of hyperlactatemia is quite broad. While the upper range, peak > 10 mmol/L and average > 5 mmol/L, is seemingly isolated to non-surviving patients, there is notable overlap at the lower levels, where larger numbers of patients would be categorized and clinical uncertainty is also greater (e.g., peak lactate level ranges in survivors versus non-survivors were 2 - 9.3 mmol/L vs. 2.3 - 22 mmol/L). This is further illustrated when one considers the confidence intervals for the likelihood ratios and the positive predictive values. Hyperlactatemia at admission yields an estimate of risk with a LR of 15 (95% CI: 9.6, 27.2) while the PPV is only 64% (95% CI: 50, 77%). Whereas persistent hyperlactatemia at 24 hours yields a LR 7 (95% CI: 1.9, 26.3) and PPV of 93% (95% CI: 82, 99%) (2). Confidence in estimates of risk are greater with admission lactate yet the low overall mortality diminishes confidence, but not diagnostic importance, when utilizing lactate as a predictive tool.

IV. Will the results help me in caring for my patients?

1. Does the tool maintain its prediction power in a new sample of patients?

Presumably, "persistent" or "cumulative" serum hyperlactatemia should maintain its prediction power in a new sample of patients, but such testing was not performed. The findings should be generalizable to a heterogeneous ICU population with the exception of post-operative patients, cardiac surgery patients and children with inborn errors of metabolism as noted by the investigators. Further analysis, or repeat sampling, is necessary before concluding that this holds across septic and non-septic diagnoses or independently of the need for hemodynamic support.

2. Are your patients similar to those patients used in deriving and validating the tool(s)?

Yes, the sample population would reflect admissions to a noncardiac multidisciplinary intensive care unit. Of note, however, post-operative patients often comprise a substantial proportion of admission for which the measurement of lactate was not tested or validated.

3. Does the tool improve your clinical decisions?

No, the serum lactate alone will not likely alter clinical decisions. Used in conjunction with the other variables, including clinical scenario, vital sign trends, physical examination, and other indices, it may be used for general prognostic tool or to spur further investigation of underlying pathology. It cannot, however, be used in isolation to indiscriminately increase care measures or, alternatively, prompt the redirection or withdrawal of therapy.

References:

1. Mizock BA. The hepatosplanchnic area and hyperlactatemia: a tale of two lactates. Crit Care Med 2001;29(2):447-449. [citation]
2. UBC Bayesian Calculator at http://www.healthcare.ubc.ca/calc/bayes.html

 

Comments

Back to the EB Journal Club Index