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

Mortality and the nature of metabolic acidosis in children with shock.

Hatherill M, Waggie Z, Purves L, Reynolds L, Argent A.

Intensive Care Med. 2003 Feb;29(2):286-91 [abstract]

Reviewed by Richard Lambert MD and Arno Zaritsky MD, University of Florida Shands Children's Hospital at Gainesville, FL

Review posted April 29, 2004

I. What is being studied?

Study objective:

Examine the relationship between base excess (BE), hyperlactatemia, hyperchloremia, 'unmeasured' strong anions (UA), and mortality in children with shock.

Study design

Prospective observational study set in a 22 bed multi-disciplinary Pediatric Intensive Care Unit in Cape Town, South Africa.

Outcomes assessed:

1. Predicted mortality was calculated from the pediatric index of mortality (PIM) score. This was then compared to the observed mortality in the investigators' pediatric ICU using the standardized mortality ratio (SMR); calculated by dividing the number of observed deaths by the number of expected deaths.
2. pH, BE, serum lactate, corrected chloride, UA, and strong ion gap (SIG) were measured or calculated, as appropriate, at admission and 24 hours.

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? Was follow-up sufficiently long and complete?

Yes. A total of 55 children were enrolled with the admission diagnosis of shock, defined by the investigators as hypotension or prolonged capillary refill requiring either fluid resuscitation or inotropic support. Nine of these children were excluded from the outcome analysis due to meeting exclusion criteria (two had congenital heart defects and one was brain dead on admission), or lacking complete laboratory data necessary for the required calculations. This left 46 children for outcome analysis. Patients were relatively young with only a median age of 6 months (1.5 - 14.4 months), but the spectrum of diagnoses causing shock was representative of a common population in a tertiary pediatric intensive care unit in the United States. Septic shock was present in over 50% of the sample. Children admitted after trauma or cardiac surgeries were not eligible for enrollment.

Data was collected at admission through 24 hours or death for analysis of the investigators' primary objective, i.e. examining the relationship between BE, hyperlactatemia, hyperchloremia, UA and overall mortality. I was surprised that half, or 8/16, of the total deaths occurred during the first 24 hours. In addition, I would be interested to know more details about the timing and circumstances of the 8 deaths that occurred after discharge from their PICU.

2. Were all potential predictors included?

Yes. The investigators examined acid-base parameters to determine if mortality was more closely related to the specific nature of an existing metabolic acidosis then its overall magnitude. They chose pH, BE, lactate, corrected chloride and UA (utilizing the Fencl-Stewart method of acid-base analysis) (1) as their acid-base parameters. While these measured and calculated parameters certainly have a role in assessing the metabolic status of a patient in shock, I believe that there are additional factors that may be important predictors of outcome. These include serum albumin, HCO3, AG, and strong ion difference (SID), the difference between strong cations and strong anions in serum. The investigators collected this data to calculate UA (also known as strong ion gap (SIG)) but did not report it in their analysis.

The three commonly accepted methods of evaluating acid-base status are: BE, serum HCO3 and AG, and SID (1). Of these three, the authors only reported data analysis for BE. While SID is inherent within the calculation of UA, I believe that including it as an additional acid-base parameter for analysis would have been interesting. In addition, the pervasive incidence of hypoalbuminemia in the ICU also may play a role in masking previously unnoticed metabolic acidosis, as the authors mentioned in their discussion. I would be interested in knowing the median serum albumin level within this sample of patients.

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

No. The investigators reported results of individual ROC curves for each predictor individually (calculating sensitivity, specificity and likelihood ratios) for survivors versus nonsurvivors. Given the inter-relationships between these acid-base variables, multi-variable regression analysis might not have been appropriate or possible.

4. Were outcome variables clearly and objectively defined?

Yes. All outcome variables were objective and unlikely to be open to subjective interpretation.

III. What are the results?

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

The investigators found that mortality in children with shock was more closely related to the nature and not the magnitude of a metabolic acidosis. Out of the four parameters evaluated, only admission hyperlactatemia and not BE, had an association with mortality. Occult anionemia did not predict poor outcome better than hyperlactatemia.

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

Only admission hyperlactatemia was able to discriminate between patients who survived and those that did not. Although the investigators used a lactate value of > 2 mmol/l as upper range for normal, a statistical association with increased mortality did not occur until admission lactate levels reached 5 mmol/l. Between survivors and non-survivors, the median value of serum lactate was 3.3 mmol/l and 11.6 mmol/l respectively, with a P value of 0.0003. All non-survivors in the study had an admission lactate of > 3 mmol/l. The area under the mortality ROC curve for lactate (using a cutoff value of 5 mmol/l) was 0.83 with a 95% confidence interval of 0.71-0.95. In comparison with the risk of mortality predicted by the PIM score, lactate correlated significantly (P=0.0016) but with weak prognostic value (r²=0.2). An admission lactate level > 5 mmol/l only predicted ICU mortality with a likelihood ratio (LR) of 2.0, while no relationship existed with lactate levels in patients alive at 24 hours (LR = 0.9).

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

The PPV for a LR of 2.0 provides little support for using lactate > 5 mmol/l as a stratifier between high risk and low risk patients (PPV 52% (95% CI 31-72%) and NPV 82% (95% CI 67-98%))(2). Using the investigators' accepted value of 2.0 mmol/l for the upper range of normal lactate would likely yield an even poorer prognostic data set due to the high number of survivors that had a lactate > 2 (median value of 3.3 mmol/l). Conversely, in the group with admission lactate > 5, there was no significant difference in mortality between patients who still had a lactate level > 5 at 24 hours and those whose levels were < 5 mmol/l.

There are a number of causes other than decreased tissue perfusion, which may contribute to hyperlactatemia. These include any significant causes of increased glycolytic flux without compensatory increases in oxygen delivery, such as may occur in a patient on an adrenaline infusion; 24% of the children received adrenaline infusions. Decreased hepatic clearance, inhibition of the phosphate dehydrogenase complex, and increased production from inflammatory cells has also been documented to elevate lactate levels in patients without evidence of compromised tissue perfusion (3-7). The wide variability of lactate levels in patients with shock directly contributes to a 'wider' confidence interval, and hence, less confidence in the estimate of attributable risk. In addition, the standard resuscitation fluid in this PICU is lactated Ringers. Thus, it is hard to know how large volume fluid resuscitation may contribute to the plasma lactate concentration.

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?

There was no additional testing done on an independent population. Therefore, the results should be restricted to the pediatric intensive care setting not inclusive of post-op cardiac surgery patients and children whom are victims of trauma.

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

Yes. The spectrum of diagnoses in children presenting with shock in this study was clinically diverse and approximates the type of child admitted to our pediatric ICU, albeit the median age was only 6 months. However, the prediction cannot be applied to trauma or post-op cardiac surgical patients.

3. Does the tool improve your clinical decisions?

No. The results in this study serve to validate findings in previous studies while underscoring the significance of not relying on one clinical parameter or laboratory value to make a medical decision or treatment choice, especially in a disease process as complex as shock. The authors chose to examine the relationship between well-described indicators of metabolic acidosis and mortality. They did not intend to make recommendations for selecting or avoiding therapy. Their findings are interesting and, I hope, will encourage multi-institutional studies that can utilize a much larger patient population to further define the role of metabolic acidosis in pediatric shock.

4. Are the results useful for reassuring or counseling patients?

The most prominent finding in this study is the relationship between an admission lactate level > 5 and mortality. This data is not new, but it does reinforce the importance of understanding the mechanisms that contribute to hyperlactatemia, as well as other causes of acidosis. I do not think that these results are irrefutable enough to engender a feeling of confidence that would allow me to offer advice to a family concerning their child's welfare without considering the multitude of additional factors that play a role in the pathology and evolution of pediatric shock.

References:

1. Fencl V, Jabor A, Kazda A, Figge J. Diagnosis of metabolic acid-base disturbances in critically ill patients. Am J Respir Crit Care Med 2000;162(6):2246-51. [abstract]
2. UBC Bayesian Calculator: www.healthcare.ubc.ca/calc/bayes.html.
3. De Backer D. Lactic acidosis. Intensive Care Med 2003;29(5):699-702. [citation]
4. Gore DC, Jahoor F, Hibbert JM, DeMaria EJ. Lactic acidosis during sepsis is related to increased pyruvate production, not deficits in tissue oxygen availability. Ann Surg 1996;224(1):97-102. [abstract]
5. Levy B, Bollaert PE, Charpentier C, Nace L, Audibert G, Bauer P, et al. Comparison of norepinephrine and dobutamine to epinephrine for hemodynamics, lactate metabolism, and gastric tonometric variables in septic shock: a prospective, randomized study. Intensive Care Med 1997;23(3):282-7. [abstract]
6. Vary TC, Hazen SA, Maish G, Cooney RN. TNF binding protein prevents hyperlactatemia and inactivation of PDH complex in skeletal muscle during sepsis. J Surg Res 1998;80(1):44-51. [abstract]
7. Routsi C, Bardouniotou H, Delivoria-Ioannidou V, Kazi D, Roussos C, Zakynthinos S. Pulmonary lactate release in patients with acute lung injury is not attributable to lung tissue hypoxia. Crit Care Med 1999;27(11):2469-73. [abstract]

 

Comments

Back to the EB Journal Club Index