SYSTEMATIC REVIEW The Use of Colloids and Crystalloids in Pediatric Dengue Shock Syndrome: a Systematic Review and Meta-analysis* Sharon Lyn R. Jalac,1 M. D. , Michelle de Vera,1 M. D. , and Marissa M. Alejandria,2 M. D. ABSTRACT Background: The cornerstone of therapy for dengue shock syndrome (DSS) remains to be the prompt and aggressive restoration of circulating plasma volume. Two common types of intravenous uids currently used in DSS are crystalloids and colloids. The debate is ongoing on whether one is superior to the other in critically ill patients.
Objective: This systematic review aims to compare the therapeutic effects of colloids versus crystalloids in reducing the recurrence of shock, the requirement for rescue uids, the need for diuretics, the total volume of intravenous uids given, the hematocrit level and pulse rates, and mortality rates of children with DSS. Search strategy: We searched Medline 1966 to August 2008, the Cochrane Controlled Trials Register (The Cochrane Library Issue 1, 2008), and the Philippines Herdin database 1964 to 2008. Local journals were handsearched and reference lists were examined.
Researchers and experts in the eld were contacted for unpublished or ongoing trials. Selection criteria: The studies included were randomized controlled trials comparing colloids versus crystalloids in children less than 18 years with a diagnosis of DSS. Data collection and analysis: Two reviewers independently did trial selection and assessment of methodologic quality using the Cochrane Infectious Diseases Group criteria. Two reviewers independently extracted the data and analyzed it using Review Manager Version 5.
Results: Of eight studies initially reviewed, four studies ful lled the selection criteria (N=694, colloids=410, crystalloids=284). Colloids and crystalloids did not differ signi cantly in decreasing the risk for recurrence of shock (RR 0. 92, 95% CI 0. 62 to 1. 38), the need for rescue uids (RR 0. 90, 95% CI 0. 70 to 1. 16), total volume of intravenous uids given (WMD 0. 80, 95% CI -1. 68 to 3. 28) and the need for diuretics (RR=1. 17, 95% CI 0. 84 to 1. 64). However, there were signi cant improvements from baseline in the hematocrit levels (WMD -3. 37, 95% CI -5. 94 to -0. 0) and pulse rates (WMD -3. 37, 95% CI -5. 94 to -0. 8) of patients who were given colloids. Tests for heterogeneity were not signi cant. Allergic type reactions were seen in patients given colloids. One child died in the colloid group. Conclusion: Colloids decreased the hematocrit and pulse rates of children with DSS after the rst two hours of uid resuscitation. However, no signi cant advantage was found over crystalloids in reducing the recurrence of shock, the need for rescue colloids, the total amount of uids, the need for diuretics, and in reducing mortality.
KEYWORDS: Meta-analysis, Crystalloids, Colloids, Pediatric, Dengue shock syndrome Department of Pediatrics, The Medical City Department of Clinical Epidemiology, UP College of Medicine 1 2 *Best Paper, The Medical City Interdepartmental Research Forum, 2008 14 Philippine Journal of Microbiology and Infectious Diseases Vol. 39, Issue 1, January-June 2010 Colloids of lifesaving F luid resuscitation is oneshock thecritically ill cornerstones in managing in patients. Physicians have an array of intravenous uids to choose from, however, the debate continues on which type of uid would be most bene cial.
One condition that would bene t most from smart uid choice is dengue shock syndrome (DSS). DSS is the most serious manifestation of dengue hemorrhagic fever (DHF). The burden of disease is greatest in Asia, where in many countries dengue is a leading cause of pediatric hospitalization. An estimated 500,000 patients with severe dengue require hospitalization each year, a large proportion of which are children. At least 2. 5% of patients die, although case fatality could be twice as high. 1 The major pathophysiological event seen in DSS is the acute increase in vascular permeability leading to the loss of plasma from the vascular compartment. Thus, the key to its management is the rapid restoration of the circulating intravenous blood volume. Two common types of intravenous uids currently used in DSS are crystalloids and colloids. Crystalloids are uids based on a solution of sterile water incorporated with electrolytes that are formulated to be hypertonic, hypotonic, or isotonic in comparison to the human plasma. The most common formulations include 0. 9% isotonic saline which is designed to approximate closely the human plasma, and lactated Ringer’s solution, almost similar to isotonic saline, but with the addition of lactate. Colloidal uids are also water- and electrolytecontaining solutions, but with the addition of a substance that does not easily diffuse through a semipermeable membrane owing to its high molecular weight. Albumin preparations contain human serum albumin, the largest component of colloid osmotic pressure in the human blood. The hetastarches, which are 6% hetastarch and 10% pentastarch, are derived from a polysaccharide that is incorporated in a uid with different molecular weights. Dextran solutions use synthetically derived dextran proteins.
Gelatin solutions have true gelatin component designed to be biologically compatible. 3 The current guidelines in the management of DSS include administration of crystalloid intravenous uids at a rate of 10-20mL/kg over an hour upon presentation of unstable vital signs or manifestations of shock. Shifting to colloidal solution is warranted if there is evidence of hemoconcentration despite administration of about 1 liter of initial intravenous uids, as shown in Appendix 1. Once improvement is seen, colloids should be shifted back to crystalloids. The increased vascular permeability in DSS, combined with the capability of crystalloid uids to seep through easily, predisposes a patient to receive massive volumes until hemodynamic stability is achieved. This led to the belief that colloids, with their higher molecular weight and ability to maintain or improve the colloid osmotic pressure, will require less amount to achieve improvement, and can thus be a better alternative to crystalloids during the early resuscitative phase in DSS. The studies that have investigated the bene ts of colloids over crystalloids in patients with shock are limited.
The large, multi-center, double-blind, randomized controlled trial Saline vs Albumin Fluid Evaluation conducted in 2004 compared the colloid albumin and the crystalloid isotonic saline for uid resuscitation in 6,997 intensive care unit patients. In the said trial, albumin and saline were comparable in terms of 28-day all cause mortality. 4 Likewise, a systematic review of the use of colloids and crystalloids for uid resuscitation in critically ill patients found no evidence that colloids, compared with crystalloids, reduced the risk of dying. Despite the increasing numbers of patients with DSS who are mostly children, there are only a few studies that compared the use of colloids and crystalloids in these patients. 6-9 OBJECTIVES The purpose of this systematic review is to synthesize the available data from clinical trials comparing colloids and crystalloids in resuscitating children with DSS. It aims to estimate the effects of colloids compared to crystalloids on the following parameters in children with DSS: • Recurrence of shock. Requirement for rescue intravenous uids. • Mortality. • Total volume of intravenous uid given. 15 Philippine Journal of Microbiology and Infectious Diseases Vol. 39, Issue 1, January-June 2010 Colloids • Need for diuretics. • Change in hematocrit level. • Change in pulse rate. This systematic review aims to compare the adverse effects associated with the administration of colloids versus crystalloids in children with DSS. experts, and researchers in the eld for additional information on unpublished and ongoing trials.
There was no language restriction. Methods of the review: Two reviewers independently assessed the eligibility of potentially relevant trials based on the selection criteria. The reviewers independently assessed the methodologic quality of each trial using the Cochrane Infectious Diseases Group criteria. Studies were assessed as “high quality” if randomization and allocation concealment were adequate, blinding maneuvers were performed, and intent-to-treat analysis was done.
Studies were assessed as “fair quality” if any subtle biases were present: unclear allocation concealment; absence of blinding; and no intent-to-treat analysis. Studies were considered “low-quality” if any of the following biases was seen: inadequate randomization and allocation concealment and/or signi cant differences between the treatment group and control group in terms of known predictors of outcome; obvious differences in the general quality of care received by subjects in both groups; marked difference in drop-out rates; and outcome detection methods were different for both groups.
Two reviewers independently abstracted data from the studies including baseline characteristics of patients, inclusion and exclusion criteria, types of intervention or study uids, and dosing regimen. The information on each outcome measure and the number of affected patients were also obtained. Data were analyzed using the Review Manager (RevMan) Version 5. 0. For dichotomous data, the risk ratio, or the probability that an event will occur, was determined for each comparison.
The need for diuretics, need for rescue uid, and recurrence of shock were all considered dichotomous outcomes. For continuous data, the weighted mean differences were determined. The decrease in hematocrit and pulse rate from baseline and the total volume of intravenous uids administered were all analyzed as continuous data. Missing data such as standard deviations were derived using the 95% con dence intervals provided. SEARCH METHODOLOGY
Criteria for considering studies for this review: Studies included in this review were randomized controlled trials in which children less than 18 years of age were given either colloid or crystalloid intravenous uid within the rst 2 hours after being diagnosed with DSS. The diagnosis of DSS was based on the World Health Organization (WHO) criteria, as shown in Appendix 2, which included hemoconcentration associated with the characteristic plasma leakage, presence of circulatory compromise as evidenced by narrowing of pulse pressure, hypotension, decreased perfusion, and decreased urine output.
The primary outcome measures included the number of episodes of shock, the requirement for rescue intravenous uids, and mortality. The secondary outcome measures included the change from the baseline in pulse rate, the change from the baseline of hematocrit, the need for diuretics, the total volume of intravenous uid given, and adverse effects. Search strategy: We searched the following databases for relevant studies: Medline 1966 to June 2008; the Cochrane Central Register of Controlled Trials (The
Cochrane Library Issue 1, 2008); and the Herdin database of the Philippines 1964-2008. The search strategy combined the terms “pediatrics”, “child”, “children”, “dengue”, “dengue hemorrhagic fever”, “dengue shock”, and “intravenous uids” with the Cochrane Highly Sensitive Search Strategy phases one and two as contained in the Cochrane Reviewer’s Handbook, as shown in Appendix 3. Conference proceedings and local journals were handsearched for potentially relevant trials. The reference lists of articles were examined to identify additional relevant studies.
We also searched institutional libraries, including libraries in universities, hospitals, and pharmaceutical companies. We contacted colleagues, 16 Philippine Journal of Microbiology and Infectious Diseases Vol. 39, Issue 1, January-June 2010 Colloids Study Flow Diagram Potentially relevant randomized controlled trials identified through Medline, Herdin, and CENTRAL searches (n=4) Trials excluded (n=2) Included patients with septic shock (n=1) Assessed difference between clinical and laboratory diagnosis of dengue (n=1) Articles obtained from colleagues (n=2)
Trial excluded (n=1) Used both colloids and crystalloids in a single patient (n=1) Links to related articles and other references examined (n=2) Trial excluded (n=1) Compared colloid with blood plasma (n=1) Randomized controlled trials comparing colloids and crystalloids in pediatric dengue shock (n=4) The formula for grouping data and imputing group means and group standard deviations were obtained from the Cochrane Handbook of Systematic Reviews. 0 We assessed heterogeneity using the chi-square test for heterogeneity (P value 50% indicated moderate degree of heterogeneity; a value of >75% indicated large degree of heterogeneity across trials. For outcome measures with signi cant heterogeneity, the sources of heterogeneity were explored by doing sensitivity analysis according to methodologic quality. excluded two trials: one trial was on the assessment of the differences between the clinical and laboratory diagnosis of dengue, and the other trial included patients with septic shock.
On review of the reference lists of relevant articles, two more trials were identi ed: one ful lled the selection criteria while the other was excluded since it used both colloids and crystalloids for each study participant. We obtained two additional studies from colleagues. One trial was excluded because it compared a colloid with blood plasma. Included studies: Four randomized controlled trials on the use of colloids and crystalloids in children with DSS were included with a combined sample size of 694 (colloids=410, crystalloids=284). -9 All the trials used WHO criteria for the diagnosis of DSS. 2 Participants: The patients were under 18 years old, with ages ranging from 1-15 years. All patients were Asians: 655 were Vietnamese and 39 were Indonesian. In all the studies, none of the patients received any intravenous uids prior to the study. 17 RESULTS Search results: As shown in the study ow diagram, we identi ed four potentially relevant randomized controlled trials on the use of colloids and crystalloids through Medline and CENTRAL searches. We Philippine Journal of Microbiology and Infectious Diseases Vol. 9, Issue 1, January-June 2010 Colloids Table 1. Characteristics of Included Studies Author, Year Nhan 2001 7 Study Population 230 Vietnamese children clinically diagnosed DHF DHF grade III = 222 DHF grade IV = 8 1-15 years old Intervention Study uids Lactated Ringer’s solution, isotonic saline, dextran, gelatin Fluid rate DHF grade III: 20mL/kg for 1 hr DHF grade IV: 20ml/kg for 15min, then 20mL/kg over the following hour Study uids Lactated Ringer’s solution, starch, dextran Fluid rate 15mL/kg for 1 hr, then 10mL/kg for the 2nd hr
Outcomes Data presented are those of DHF grade III patients only: PPRT, reshock rate, time to 1st episode of reshock, change from baseline of hematocrit and pulse rate, volume of uid infused, requirement for rescue uid, volume of rescue uid used, requirement for diuretic, mortality Change from baseline of hematocrit; total volume of rescue uid used after initial resuscitation, total volume of uid given, requirement for further uid resuscitation, requirement for diuretic, mortality, length of hospital stay; depth of pleural effusion, new bleeding after study entry, clinical uid overload, volume of ascites Change from baseline of hematocrit, cardiac index, pulse rate and pulse pressure, requirement for further uid resuscitation, requirement for diuretic, duration of shock, episodes of reshock, mortality Change from the baseline of hematocrit, hemoglobin, pulse rate and pulse pressure; total volume of uid given, episodes of reshock, mortality Adverse Reactions None stated Wills 20056 512 Vietnamese children with clinical DSS Moderate shock = 383 Severe shock = 129 2-15 years old Allergic type reactions, transient high fever and rigors without cardio-respiratory compromise; urticarial rashes Dung 19998 50 Vietnamese children with clinical DSS 5-15 years old
Study uids Lactated Ringer’s solution, isotonic saline, dextran, gelatin Fluid rate 20mL/kg for 1 hr, then 10mL/kg for the 2nd hr Study uids Lactated Ringer’s solution, gelatin Fluid rate 20mL/kg as initial volume of uid resuscitation None stated Prasetyo20089 39 Indonesian children with clinical DSS 1-13 years Adverse reactions in coagulation, liver and renal functions and acid-base equilibrium; severe allergic type reactions DHF: dengue hemorrhagic fever DSS: dengue shock syndrome PPRT: pulse pressure recovery time One study included both DHF grades III and IV patients9, while two studies presented data on DHF grade III patients only. 7-8 The study by Wills (2005) grouped patients into those with moderate and severe shock. The subset of patients with severe shock was not given the chance to receive crystalloid uids and was not included in this review.
Children were excluded in the study if they had severe hemorrhagic manifestation for which transfusion was likely7, chronic disorders7, malnutrition9, and heart, renal, liver, and lung diseases. 9 Interventions: The colloids used were dextran 706-8, gelatin7-8, and hydroxyethyl starch. 6,9 The crystalloids used were lactated Ringer’s solution69 and isotonic saline. 7-8 In three studies6-8, the study uids were administered initially at 20mL/kg for 1 hr for DHF grade III patients and 20ml/kg for 15min. Subsequently, uids were given depending on each study’s protocol: 20mL/kg over the following hour for DHF grade IV patients7; 15mL/kg for 1 hour, then 10mL/kg for the 2nd hour6; and 20mL/kg for 1 hour, then 10mL/kg for the 2nd hour. The study by Prasetyo9 administered an initial volume of 20mL/kg to all patients. Table 2. Characteristics of Excluded Studies Author, Year Hung 2006 Martinez-Vega 2006 Tatura 2008 Upadhyay 2005 Reason for Exclusion The study used both colloids and crystalloids in all patients with DSS. The study did not compare colloids and crystalloids; it was based on the assessment of the difference between clinical and laboratory diagnosis of dengue. The study compared gelatin solution with blood plasma. The study participants were children with septic shock. 18 Philippine Journal of Microbiology and Infectious Diseases Vol. 39, Issue 1, January-June 2010 Colloids Table 3. Methodological Quality of Included Studies
Author Year Dung 19998 Randomization Adequate Allocation Concealment Adequate-treatment pack numbers contained in opaque envelopes Adequate-treatment pack numbers contained in opaque envelopes Computer generated random numbers, treatment packs prepared by independent staff Not stated Blinding Double-blind each treatment bottle covered in black tape Double-blind Double-blind Intent-to-treat Analysis Not stated Nhan 20017 Wills 20056 Sequential randomization Adequate, strati ed by pulse pressure Random assignment, strati ed to DHF 3 or 4 Yes Yes Prasetyo 20089 Not stated Not stated Outcomes: The outcomes measured included the following: changes from baseline of hematocrit6-9 and pulse rate7-9; total volume of rescue uid used after initial resuscitation6,7 and requirement for further uid resuscitation6-8; total volume of intravenous uids given6,7,9; requirement for diuretics6-8; and number of patients with recurrences of shock. -9 Other parameters evaluated were the following: change in hemoglobin9, pulse pressure8,9 and pulse pressure recovery time7; cardiac index8; duration of shock8; time to rst episode of reshock7; length of hospital stay in days6; and mortality. 6-9 The study by Prasetyo further determined disorders of blood coagulation, liver and renal function parameters, and acid-base equilibrium. 9 The study by Wills reported depth of pleural effusion, new bleeding after study entry, clinical uid overload, and volume of ascites. 6 Two studies determined severe allergic type reactions after intravenous uid infusion. 6,9 The characteristics of the included studies are summarized in Table 1. Excluded studies: Trials that did not compare rystalloids and colloids or were not used on pediatric patients with DSS were excluded from this review. 11-14 The characteristics of the excluded studies are summarized in Table 2. Methodological quality of included studies: Two studies were assessed as “high quality. ”6. 7 The study by Prasetyo had unclear allocation concealment and did not mention blinding or intent-to-treat analysis, as shown in Table 3. The study by Dung also did not state whether intent-to-treat analysis was done. Both were rated as fair quality. Despite randomization, the study of Nhan had unequal distribution of more severely ill patients: dextran group had fewer patients with very low pulse pressure, but had patients with lowest mean pulse rate at presentation.
Primary outcomes: The meta-analysis of the trials showed no statistically signi cant difference in reducing the risk for recurrence of shock (3 trials, RR 0. 92, 95% CI 0. 62 to 1. 38) and the need for rescue colloids (2 trials, RR 0. 90, 95% CI 0. 70 to 1. 16) after the initial resuscitation. Test for heterogeneity was not signi cant for both outcomes, as shown in Figures 1 and 2. Figure 1. Number of Patients Who Had Recurrence of Shock After the Initial Fluid Resuscitation Study or Subgroup Dung 1999 Nhan 2001 Prasetyo 2008 Total (95% CI) Colloid Crystalloid Risk Ratio Events Total Events Total Weight M-H, Fixed, 95% CI 2 31 1 25 111 19 155 2 32 3 25 5. 4% 111 86. 7% 20 7. 9% 156 100. 0% 1. 00 [0. 15, 6. 55] 0. 97 [0. 64, 1. 47] 0. 35 [0. 04, 3. 09] 0. 92 [0. 62, 1. 38]
Risk Ratio M-H, Fixed, 95% CI Total events 34 37 Heterogeneity: Chi? = 0. 82, df = 2 (P = 0. 66); I? = 0% Test for overall effect: Z = 0. 40 (P = 0. 69) 0. 005 0. 1 1 10 200 Favours colloids Favours crystalloids Favours experimental Favours control 19 Philippine Journal of Microbiology and Infectious Diseases Vol. 39, Issue 1, January-June 2010 Colloids Figure 2. Number of Patients Who Needed Rescue Colloids After the Initial Fluid Resuscitation Study or Subgroup Nhan 2001 Wills 2005 Total (95% CI) colloid crystalloid Risk Ratio Events Total Events Total Weight M-H, Fixed, 95% CI 32 74 111 255 366 37 40 111 41. 0% 128 59. 0% 239 100. 0% 0. 86 [0. 58, 1. 28] 0. 93 [0. 7, 1. 28] 0. 90 [0. 70, 1. 16] Risk Ratio M-H, Fixed, 95% CI 77 Total events 106 Heterogeneity: Chi? = 0. 08, df = 1 (P = 0. 78); I? = 0% Test for overall effect: Z = 0. 81 (P = 0. 42) 0. 01 0. 1 1 10 100 Favours colloids Favours crystalloids Favours experimental Favours control All study participants recovered fully, except for one child in the colloid group (starch) who died of profound shock and gastrointestinal bleeding. 6 Secondary outcomes: There was no signi cant difference between patients who received colloids and crystalloids in terms of the total volume of intravenous uids given during resuscitation (WMD 0. 80 ml/kg, 95% CI -1. 68 to 3. 8) and the need for diuretics (RR=1. 17, 95% CI 0. 84 to 1. 64). The test for heterogeneity was not signi cant, as shown in Figures 3 and 4. Hemodynamic variables: On the other hand, there was a signi cant improvement in hematocrit levels from the baseline after the rst two hours of uid resuscitation (WMD -7. 87, 95% CI -8. 52 to 7. 22), but with a large degree of heterogeneity, as shown in Figure 5. Sensitivity analysis removing the study of Wills (2005), which used median instead of mean, removed the heterogeneity and a statistically signi cant reduction in hematocrit levels remained (WMD -3. 37, 95% CI -5. 94 to -0. 80), as shown in Figure 5a.
The study of Wills (2005) also showed a signi cant reduction in the median hematocrit levels two hours after the initial resuscitation (25% for the dextran group versus 9% for the Ringer’s lactate group, p 35% Improvement IV therapy by crystalloid, successively reducing the ow from 10 to 6, 6 to 3ml/kg/hr Discontinue after 24-48 hrs Philippine Journal of Microbiology and Infectious Diseases Vol. 39, Issue 1, January-June 2010 25 Colloids Appendix 2. World Health Organization Criteria for the Diagnosis of Dengue Fever and Dengue Hemorrhagic Fever DF/DHF DF Grade* Symptoms Fever with two or more of the following signs, headache, retro-orbital pain, myalgia, arthralgia I Above signs plus positive tourniquet test Above signs plus spontaneous bleeding Laboratory Leukopenia occasionally.
Thrombocytopenia, may be present, no evidence of plasma loss Thrombocytopenia < 100,000, Hct rise > 20% Thrombocytopenia < 100,000, Hct rise > 20% Thrombocytopenia < 100,000, Hct rise > 20% DHF DHF II DHF III Above signs plus circulatory failure (weak pulse, hypotension, restlessness) Profound shock with undetectable blood pressure and pulse DHF IV Thrombocytopenia < 100,000, Hct rise > 20% *DHF Grade III and IV are also called as Dengue Shock Syndrome (DSS) 26 Philippine Journal of Microbiology and Infectious Diseases Vol. 39, Issue 1, January-June 2010 Colloids Appendix 3. Search Strategy Philippine Journal of Microbiology and Infectious Diseases Vol. 39, Issue 1, January-June 2010 27