Office of Evidence Based Practice Specific Care Question: Dexamethasone vs. Prednisone for a Pediatric Asthma Exacerbation

Transcription

1 Specific Care Question : Is dexamethasone better than prednisolone/prednisone for a pediatric asthma exacerbation Question Originator: Jeff Michael, DO, FAAP Plain Language Summary from The Office of Evidence Based Practice: Children with an asthma exacerbation present to the Emergency Department/Urgent Care Center (ED/UCC) with wheezing, tightness in their chest and difficulty breathing. Glucocorticosteroids (CS) are a first line medication used to reduce inflammation and reduce the symptoms of asthma. However, CS can have side effects, and reducing the side effects of a treatment is an important goal. The recommendation of the Asthma in the ED CPG (202) team is to use prednisone/prednisolone over dexamethasone. It is based on the recommendation of The National Asthma Education and Prevention Program, Expert Panel Report: Guidelines for the Diagnoses and Management of Asthma (NAEP-EPR-3, 2007). The recommendations of various national and professional society guideline creators were compiled (See Table ). There is no consensus across the guidelines on which CS to use for an asthma exacerbation. There are two major types of CS that can be used. One is dexamethasone and the other is either prednisone (table) prednisolone (syrup). The medications are similar in how well and how quickly they decrease asthma symptoms. Prednisone and/or prednisolone have been the preferred medications to treat acute asthma because it is believed there are fewer side effects, such as hyperactivity, nausea, and reduced growth. Dexamethasone is a long acting steroid medication that is 5 times stronger than prednisone/prednisolone and has a longer half-life (Hendeles, 2003). A smaller, less frequent dose of the stronger medication may increase the ability of the child to take the medication. Therefore, the choice of using one CS over the other is based not on efficacy, but on outcomes that are not well studied, such as believed compliance to treatment and adverse events including (a) vomiting in the ED, (b) vomiting at home, and (c) adrenal suppression in children with frequent asthma exacerbations. Since the previous systematic review and abbreviated meta-analysis on this question, three studies ( RCT and 3 meta-analyses) have been published (Cronin et al., 205; Keeney et al., 204; (Meyer, Riese, & Biondi, 204); Normansell, Kew, & Mansour, 206). They are added to this critically appraised topic. Review by Outcome For the comparison PO dexamethasone vs. PO prednisone/olone, (Normansell et al., 206) includes the studies in (Keeney et al., 204) and includes a new study (Cronin et al., 205). Therefore, Normansell et al. (206) will be used in this analysis. For the comparison IM dexamethasone vs. PO prednisone/olone (Keeney et al., 204) reported on IM plus PO dexamethasone versus PO prednisone/olone. For this analysis, only PO dexamethasone versus PO prednisone will be included. The Risk of Bias Summary is found in Figure.

2 The following outcomes are included in this analysis. PO dexamethasone vs. PO prednisone/olone a. Hospital admission at initial presentation b. Relapse or new exacerbation during the follow up period c. Vomiting 2. IM dexamethasone vs. PO prednisone/olone a. Relapse or new exacerbation during the follow up period b. Vomiting in the ED c. Vomiting at home PO Dexamethasone vs. PO Prednisone/olone Hospital admission at initial presentation. Three studies are included for this outcome (Altamimi et al., 2006; Cronin et al., 205; Qureshi, Zaritsky, & Poirier, 200). The quality of the evidence for this outcome is low. The studies are downgraded for risk of bias (selection, performance, and detection bias). The studies are also downgraded for imprecision since the confidence intervals for the estimate of the effect cross the line of no effect, and the number of events is low(see Table 2). There is no difference in the odds of being hospitalized at the initial presentation at the ED for an asthma exacerbation whether treated with PO dexamethasone of PO prednisolone, OR=.08, 95% CI [0.74,.58] (See Figure 2.). Relapse or new exacerbation during the follow up period. Four studies are included for this outcome (Altamimi et al., 2006; Cronin et al., 205; Greenberg, Kerby, & Roosevelt, 2008; Qureshi et al., 200). The quality of the evidence for this outcome is very low. The studies are downgraded for risk of bias (selection, performance, and detection bias). The studies are also downgraded for inconsistency as the follow periods ranged from 5-30 days, and imprecision since the confidence intervals for the estimate of the effect cross the line of no effect, and the number of events is low (See Table 2). There no difference in the odds of having a relapse or new exacerbation during the follow- up periods whether treated with PO dexamethasone or PO prednisone/olone, OR = 0.85, 95% CI [0.54,.34] (See Figure 3). Vomiting. Three studies are included for this outcome (Cronin et al., 205; Greenberg et al., 2008; Qureshi et al., 200). The quality of evidence for this outcome is very low. The studies are downgraded for risk of bias (selection, performance, and detection bias). Since the studies included vomiting in the ED and at home, the studies are also downgraded for inconsistency. Finally, the studies are downgraded for imprecision since the estimate of the effect crosses the line of no effect, and the number of events is low. There is no difference between vomiting after treatment with PO dexamethasone or PO prednisone/olone. OR = 0.33, 95% CI [0.09,.4] (see Figure 4). IM dexamethasone vs. PO prednisone/olone Relapse or new exacerbation during the follow up period. Three studies are included in for this outcome (Gordon, Tompkins, & Dayan, 2007; Gries, Moffitt, Pulos, & Carter, 2000; Klig, Hodge, & Rutherford, 997). The quality of evidence for this outcome is low. The studies are downgraded for risk of bias (selection, detection, and performance bias). There are no events of relapse or unscheduled visits to a clinic, ED, or hospital in 2 of the three studies, therefore due to the low number of events; the evidence is downgraded for imprecision; the estimate of the effect crosses the line of no effect, and the number of events is low. There is no difference in relapse or unscheduled visits after treatment with IM dexamethasone or PO prednisone/olone, OR= 0.7, 95% CI [0.29,.74] (See Figure 5). 2

3 Vomiting in the ED. One study is included for this outcome (Gordon et al., 2007) Unable to grade this evidence as there is only one study for this outcome, and the number of events is low. Gordon et al. (2007) shows that the odds of vomiting in the ED significantly less with dexamethasone OR = 0.05, 95% CI [0.00, 0.87] (See Figure 5). Our confidence in the estimate of the effect is low due to only one study being included in the body of evidence for vomiting in the ED. Note that Keeney et al (204) included Altamini (2006) for this outcome. It is excluded from this analysis because data for this outcome was not reported. Vomiting at home. Two studies are included for this outcome (Gordon et al., 2007 and Altamimi et al., 2006). T he quality of this evidence is low. The studies are downgraded for risk of bias (performance and attrition bias). There are no events of vomiting at home in the dexamethasone group and only one event of vomiting at home in the prednisone/olone group in each included study. Therefore, due to the very low number of events, the evidence is downgraded for imprecision. There is no difference in vomiting at home in children treated with IM dexamethasone or PO prednisone/olone, OR = 0.32, 95% CI [0.03, 3.08] (See Figure 6). EBP Scholar s responsible for analyzing the literature: Helen Murphy BHS RRT AE-C EBP team member responsible for reviewing, synthesizing, and developing this literature: Nancy H Allen, MS, MLS, RD, LD Search Strategy and Results: ("Steroids"[Mesh] OR "Glucocorticoids"[Mesh] OR "Glucocorticoids"[Pharmacological Action] OR "dexamethasone 2-phosphate"[Supplementary Concept] OR "Dexamethasone"[Mesh] OR "Prednisone"[Mesh] OR "Prednisolone"[Mesh] OR "Methylprednisolone"[Mesh]) AND "Asthma/drug therapy"[majr] AND (("204/0/0"[PDAT] : "206/2/3"[PDAT]) AND English[lang] AND ("infant"[mesh Terms] OR "child"[mesh Terms] OR "adolescent"[mesh Terms])) 04 results Studies included in this review: (Keeney et al., 204) Altamimi et al., 2006 Gordon, Tompkins, & Dayan, 2007 Greenberg, Kerby, & Roosevelt, 2008 Klig, Hodge, & Rutherford, 997 Qureshi, Zaritsky, & Poirier, 200 (Meyer et al., 204) Altamimi et al., 2006 Gordon, Tompkins, & Dayan, 2007 Gries, et al., 2000 Greenberg, Kerby, & Roosevelt, 2008 Klig, Hodge, & Rutherford, 997 3

4 Qureshi, Zaritsky, & Poirier, 200 (Normansell et al., 206) Altamimi et al., 2006 Cronin et al., 205 Gordon et al Greenberg, Kerby, & Roosevelt, 2008 Qureshi, Zaritsky, & Poirier, 200 Hendeles (2003) is used as background information Studies not included in this review with rationale for exclusion: Study Cross 20 Goleva 202 Shefrin 2009 Sutherland 2008 (Watnick, Fabbri, & Arnold, 206) Reason for Exclusion Narrative review Cohort study of steroid resistant vs. steroid sensitive Narrative review Does not answer the question. Outcome include non-smoking adult lung function, BMI, and spirometric response to medication Cohort study Method Used for Appraisal and Synthesis: The Cochrane Collaborative computer program, Review Manager (RevMan 5.3.5), was used to synthesize the XXX included studies. Updated: May ; June

5 Table Guideline Comparison: Dexamethasone vs. Prednisone for a Guidelines Prednisone Dexamethasone EPR Recommendation based on Hendeles (2003) Prefers prednisone over dexamethasone Does not mention dexamethasone for an exacerbation. For long term control medications it states dexamethasone is likely to be as effective as prednisolone. It goes on to say: -- or patients unable to tolerate the liquid preparations, dexamethasone syrup at 0.4 mg/kg/d may be an alternative. Studies are limited, however, and the longer duration of activity increases the risk of adrenal suppression Canadian guideline Prednisone/prednisolone (oral) Dose:-2 mg/kg (max 50 mg) Duration: 3 to 5 days Does not mention dexamethasone Colorado 20 Treat with Prednisone(or equivalent) Dose: 2 mg/kg orally (80 mg max for inpatient) Dose: 2 mg/kg orally (60 mg max for outpatient) Duration 3-0 days Does not mention dexamethasone 5

6 Seattle Children s Hospital Recommendation based on Keeney et al. (204) and Meyer et al. (204) Children s Hospital of Philadelphia Emergency Department Recommendation made based on Keeney et al. (204) Children s Hospital of Philadelphia Inpatient Prednisone or equivalents for Inpatient use Prednisone Dose:2 mg/kg/d PO QD Max60 mg Duration: 5-0 days of total steroids Prednisolone Dose :2 mg/kg/d PO QD Max 60 mg Duration: 5-0 days of total steroids Methylprednisolone Dose: mg/kg. IV q 6 (Max 60 mg per dose) Duration: not stated Severe: ESI -2 Prednisone/olone/Methylprednisolone Dose: Prednisone/ Methylprednisolone: 2 mg/kg p.o./iv MAX 60 mg Duration: not stated Steroid management: prednisone/prednisolone q 2 hours after initial systemic corticosteroid administration discontinued inhaled corticosteroid. Severe: Continue prednisone/prednisolone/ Methylprednisolone Moderate: Continue prednisone/prednisolone Mild: Continue prednisone Dexamethasone for ED use Dose: 0.6 mg/kg/d PO QD (6 mg max dose) Duration not stated Should be given within one hour of entering the ED Mild: ESI Triage 4 Consider dexamethasone tablet (alternative prednisone/olone) Moderate: ESI Triage 3 Dexamethasone tablet (alternative prednisone/olone) Dose: Dexamethasone: Mild to moderate flare,, 5-8 years 4 mg >8-2 6 mg years > 2 8 mg years Duration: repeat in hours Does not mention dexamethasone 6

7 Global Initiative for Asthma (GINA) 205 British Thoracic Society/ Scottish Intercollegiate Guidelines Network (BTS/SIGN) 204 For children, an OCS dose of 2 mg/kg up to a maximum of 40 mg/day is adequate. Oral prednisone is the steroid of choice for asthma attacks in children unless the patient is unable to tolerate the dose 2-5 years 20 mg/kg >5 years mg Intravenous hydrocortisone (4 mg/kg, repeated four hourly) should be reserved for severely affected children who are unable to retain oral medication Oral dexamethasone for 2 days can also be used, but there are concerns about metabolic side effects if it is continued beyond 2 days Insufficient evidence to suggest that dexamethasone offers an advantage over prednisolone 7

8 Table 2 Grade Summary of Normansell et al. (206) Question: Should dexamethasone be used in acute asthma exacerbation in children Studies included in the meta-analysis: (Altamimi et al., 2006; Cronin et al., 205; Gordon et al. 2007; Greenberg, Kerby, & Roosevelt, 2008; Qureshi, Zaritsky, & Poirier, 200) Quality assessment of patients Effect of studies Study design Risk of bias Inconsistency Indirectness Imprecision Other considerations Relative Oral dexamethasone Prednisone (95% CI) Absolute (95% CI) Quality Hospital admission at initial presentation 3 randomized trials serious not serious not serious serious 2 none 63/508 (2.4%) 58/499 (.6%) OR.08 (0.74 to.58) 8 more per,000 (from 28 fewer to 56 more) LOW New exacerbation during the follow-up period 4 randomized trials serious not serious not serious serious 2 none 39/475 (8.2%) 49/506 (9.7%) OR 0.85 (0.54 to.34) 3 fewer per,000 (from 29 more to 42 fewer) LOW Vomiting 3 randomized trials serious not serious not serious serious 2 none /446 (2.5%) 32/42 (7.6%) - 50 fewer per 000 (from (0.69 fewer to 0 more) LOW CI: Confidence interval; OR: Odds ratio; MD: Mean difference 8

9 . Downgraded once for risk of bias, Cronin et al. (205) 8% of subjects were enrolled twice. Quereshi et al. (200) was considered to be at high or unclear risk of selection bias, while performance and detection bias were noted in both Qureshi et al. (200) and Cronin et al. (205). 2. Confidence cross the line of no effect; unsure if a clinically important event is present, and which treatment it favors 9

10 Table 3 Grade Summary of (Keeney et al. (204) Question: Should dexamethasone be used in acute asthma exacerbation in children Studies included in the meta-analysis: (Altamimi et al., 2006; Gordon, Tompkins, & Dayan, 2007; Greenberg, Kerby, & Roosevelt, 2008; Klig, Hodge, & Rutherford, 997; Qureshi, Zaritsky, & Poirier, 200) Summary of findings Quality assessment No of patients Effect Relative Importance No of Other Dexamethasone control (95% Absolute Quality Design Limitations Inconsistency Indirectness Imprecision studies considerations CI) Relapse rate 5 days (follow-up 5 days; number who returned for treatment of exacerbation) 4 randomized serious,2 serious 3,4 no serious no serious none trials indirectness imprecision 4/67 (8.4%) 5/69 (8.9%) Relapse rate 0-4 days (follow-up 0-4 days; number who returned for treatment of exacerbation) 3 randomized serious,2 serious 3,4 no serious no serious none trials indirectness imprecision Vomiting in the ED (follow-up 2-4 hours) 4 randomized serious,2 serious 3,4 no serious trials indirectness no serious imprecision none 47/39 (2%) 6/506 (.2%) 40/372 (0.8%) 27/5 (5.3%) RR 0.90 (0.46 to.78) RR.3 (0.77 to.67) RR 0.29 (0.2 to 0.69) 9 fewer per 000 (from 48 fewer to 69 more) 4 more per 000 (from 25 fewer to 72 more) OO LOW OO LOW 38 fewer per 000 OO (from 6 LOW fewer to 46 fewer) There is poor reporting of allocation concealment in half of the studies. 2 One third of the studies blinded the participants, and in only one study were outcome assessors blinded. 3 The I2 statistic low (0-8.7) but there were differences in the treatments, Three studies used a single IM dose of dexamethasone, one study used a single oral dose and two studies used multiple oral doses. 4 The dose of IM dexamethasone ranged from 0.3 mg/kg with a mix of 5 mg IM X dose to.7 mg/kg.7 mg/kg (max 36 mg) IM X dose. The range of oral dexamethasone ranged from 0.3 mg/kg (max 5 mg) PO X dose to 0.6 mg/kg (max 6 mg) PO once daily X 2 days. CRITICAL CRITICAL IMPORTANT 0

11 Random sequence generation (selection bias) Allocation concealment (selection bias) Blinding (performance bias and detection bias) Incomplete outcome data (attrition bias) Selective reporting (reporting bias) Other bias Office of Evidence Based Practice Specific Care Question: Dexamethasone vs. Prednisone for a Figures: Altamimi Cronin 205 Gordon Greenberg Gries 2000 Klig Qureshi 200 Figure. Risk of bias summary

12 Dexamethasone Predinisone/olone Odds Ratio Odds Ratio Risk of Bias Study or Subgroup Weight M-H, Random, 95% CI M-H, Random, 95% CI A B C D E F Altamimi % 0.63 [0.2,.89] Cronin %.4 [0.55, 2.35] Qureshi %.04 [0.63,.7] (95% CI) %.00 [0.68,.47] events Heterogeneity: Tau² = 0.00; Chi² = 0.8, df = 2 (P = 0.67); I² = 0% Test for overall effect: Z = 0.00 (P =.00) PO Dexamethasone PO Prednisone/olone Risk of bias legend (A) Random sequence generation (selection bias) (B) Allocation concealment (selection bias) (C) Blinding (performance bias and detection bias) (D) Incomplete outcome data (attrition bias) (E) Selective reporting (reporting bias) (F) Other bias Figure 2. Oral dexamethasone vs. oral prednisone/olone, Outcome: Hospital admission at initial presentation Dexamethasone Predinisone/olone Odds Ratio Odds Ratio Risk of Bias Study or Subgroup Weight M-H, Random, 95% CI M-H, Random, 95% CI A B C D E F Altamimi % 3.86 [0.42, 35.62] Cronin % 0.98 [0.48, 2.03] Greenberg % 2.7 [0.54, 8.80] Qureshi %.07 [0.55, 2.07] (95% CI) %.7 [0.75,.85] events Heterogeneity: Tau² = 0.00; Chi² = 2.6, df = 3 (P = 0.54); I² = 0% Test for overall effect: Z = 0.70 (P = 0.48) Dexamethasone Prednisone/olone Risk of bias legend (A) Random sequence generation (selection bias) (B) Allocation concealment (selection bias) (C) Blinding (performance bias and detection bias) (D) Incomplete outcome data (attrition bias) (E) Selective reporting (reporting bias) (F) Other bias Figure 3. Oral dexamethasone vs. oral prednisone/olone, Outcome: New exacerbation during the follow up period 2

13 Dexamethasone Predinisone/olone Odds Ratio Odds Ratio Risk of Bias Study or Subgroup Weight M-H, Random, 95% CI M-H, Random, 95% CI A B C D E F Cronin % 0.03 [0.00, 0.5] Greenberg % 0.48 [0.4,.65] Qureshi % 0.5 [0.9,.4] (95% CI) % 0.33 [0.09,.4] events 32 Heterogeneity: Tau² = 0.62; Chi² = 4.29, df = 2 (P = 0.2); I² = 53% Test for overall effect: Z =.76 (P = 0.08) PO Dexamethasone PO Prednisone/olone Risk of bias legend (A) Random sequence generation (selection bias) (B) Allocation concealment (selection bias) (C) Blinding (performance bias and detection bias) (D) Incomplete outcome data (attrition bias) (E) Selective reporting (reporting bias) (F) Other bias Figure 4. Oral dexamethasone vs. oral prednisone/olone, Outcome: Vomiting 3

14 Dexamethasone Prednisone/olone Odds Ratio Odds Ratio Risk of Bias Study or Subgroup Weight M-H, Fixed, 95% CI M-H, Fixed, 95% CI A B C D E F Gordon % 0.05 [0.00, 0.87] (95% CI) % 0.05 [0.00, 0.87] events 0 9 Heterogeneity: Not applicable Test for overall effect: Z = 2.06 (P = 0.04) IM Dexamethasone PO Predinsione/olone Risk of bias legend (A) Random sequence generation (selection bias) (B) Allocation concealment (selection bias) (C) Blinding (performance bias and detection bias) (D) Incomplete outcome data (attrition bias) (E) Selective reporting (reporting bias) (F) Other bias Figure 5. IM Dexamethasone vs. oral prednisone/olone Outcome: Vomiting in the ED Dexamethasone Prednisone/olone Odds Ratio Odds Ratio Risk of Bias Study or Subgroup Weight M-H, Fixed, 95% CI M-H, Fixed, 95% CI A B C D E F Altamimi 2006 Gordon % 50.0% 0.32 [0.0, 7.92] 0.32 [0.0, 7.92] (95% CI) % 0.32 [0.03, 3.08] events 0 2 Heterogeneity: Chi² = 0.00, df = (P =.00); I² = 0% Test for overall effect: Z = 0.99 (P = 0.32) IM Dexamethasone PO Predinsione/olone Risk of bias legend (A) Random sequence generation (selection bias) (B) Allocation concealment (selection bias) (C) Blinding (performance bias and detection bias) (D) Incomplete outcome data (attrition bias) (E) Selective reporting (reporting bias) (F) Other bias Figure 6. Figure 5. IM Dexamethasone vs. oral prednisone/olone Outcome: Vomiting at home 4

15 References Altamimi, S., Robertson, G., Jastaniah, W., Davey, A., Dehghani, N., Chen, R.,... Colbourne, M. (2006). Single-dose oral dexamethasone in the emergency management of children with exacerbations of mild to moderate asthma. Pediatr Emerg Care, 22(2), doi:0.097/0.pec [pii] Cronin, J. J., McCoy, S., Kennedy, U., An Fhaili, S. N., Wakai, A., Hayden, J.,... O'Sullivan, R. (205). A Randomized Trial of Single-Dose Oral Dexamethasone Versus Multidose Prednisolone for Acute Exacerbations of Asthma in Children Who Attend the Emergency Department. Ann Emerg Med. doi:0.06/j.annemergmed Gordon, S., Tompkins, T., & Dayan, P. S. (2007). Randomized trial of single-dose intramuscular dexamethasone compared with prednisolone for children with acute asthma. Pediatr Emerg Care, 23(8), doi:0.097/pec.0b03e3828f [pii] Greenberg, R. A., Kerby, G., & Roosevelt, G. E. (2008). A comparison of oral dexamethasone with oral prednisone in pediatric asthma exacerbations treated in the emergency department. Clin Pediatr (Phila), 47(8), doi:0.77/ [pii] Gries, D. M., Moffitt, D. R., Pulos, E., & Carter, E. R. (2000). A single dose of intramuscularly administered dexamethasone acetate is as effective as oral prednisone to treat asthma exacerbations in young children. J Pediatr, 36(3), doi:0.067/mpd Hendeles, L. (2003). Selecting a systemic corticosteroid for acute asthma in young children. J Pediatr, 42(2 Suppl), S Keeney, G. E., Gray, M. P., Morrison, A. K., Levas, M. N., Kessler, E. A., Hill, G. D.,... Jackson, J. L. (204). Dexamethasone for acute asthma exacerbations in children: a meta-analysis. Pediatrics, 33(3), doi:0.542/peds Klig, J. E., Hodge, D., 3rd, & Rutherford, M. W. (997). Symptomatic improvement following emergency department management of asthma: a pilot study of intramuscular dexamethasone versus oral prednisone. J Asthma, 34(5), Lougheed, M. D., Lemiere, C., Ducharme, F. M., Licskai, C., Dell, S. D., Rowe, B. H.,... Canadian Thoracic Society Asthma Clinical, A. (202). Canadian Thoracic Society 202 guideline update: Diagnosis and management of asthma in preschoolers, children and adults: Executive summary. Canadian Respiratory Journal : Journal of the Canadian Thoracic Society, 9(6), e8-e88. Meyer, J. S., Riese, J., & Biondi, E. (204). Is dexamethasone an effective alternative to oral prednisone in the treatment of pediatric asthma exacerbations Hosp Pediatr, 4(3), doi:0.542/hpeds NAEP-EPR-3. (2007). Expert Panel Report 3: Guidelines for the diagnosis and management of asthma. Bethesda MD: U.S. Department of Health and Human Services, National Institutes of Health. Normansell, R., Kew, K. M., & Mansour, G. (206). Different oral corticosteroid regimens for acute asthma. Cochrane Database Syst Rev, 5, CD080. doi:0.002/ cd080.pub2 Qureshi, F., Zaritsky, A., & Poirier, M. P. (200). Comparative efficacy of oral dexamethasone versus oral prednisone in acute pediatric asthma. The Journal of pediatrics, 39(), doi:0.067/mpd Watnick, C. S., Fabbri, D., & Arnold, D. H. (206). Single-dose oral dexamethasone is effective in preventing relapse after acute asthma exacerbations. Ann Allergy Asthma Immunol, 6(2), doi:0.06/j.anai

16 Appendix Understanding GRADE (Grading of Recommendations Assessment, Development, and Evaluation) The following concepts are utilized when applying GRADE to a literature analysis that answers a clinical question. The quality of the evidence (High, Medium, Low, or Very Low) reflects the confidence in the estimate of effect for an outcome to support the recommendation. When the evidence is only from RCTs the GRADE starts as High quality evidence. The RCTs are read and assessed upon the criteria below. Risk of Bias, looks specifically for: o Randomization o Concealment of allocation o Blinding (participants, providers, outcome assessor) o How subjects who did not complete the study were handled o Selective reporting Inconsistency (heterogeneity) is present when: o Variation of point estimates across the included studies is reported o The confidence intervals of the included studies do not overlap, or have minimal overlap o The p value of the statistical tests for heterogeneity (Chi 2 ) is low (p< 0.05) o The I 2 statistic is roughly interpreted as (note the overlap in the ranges below; I 2 is not a precise measure): 0%-40% the heterogeneity might not be a factor 30-60% may indicate moderate heterogeneity 50-90% may indicate substantial heterogeneity 75-00% indicates considerable heterogeneity Indirectness is present when: o Studies with a direct comparison are not available, and the included studies are combined to indirectly look at the comparison of interest. Example: The Migraine in the Emergency Department Clinical Practice Guideline compares dexamethasone to magnesium sulfate IV for the outcome pain relief at 2 hours. Studies have been identified that compare (a) dexamethasone to magnesium sulfate (IV), and (b) prochlorperazine to magnesium sulfate (IV). If they are used to compare dexamethasone to prochlorperazine the evidence is indirect. (See Figure A) o Studies predominantly include (a) participants, (b) methods- such as doses, (c) outcomes that are different from the question asked by the guideline panel. In pediatrics, this is most often seen when adult population studies are applied to children. Imprecision is present when: o The included studies have less than ~ 400 subjects cumulatively (this is a rule of thumb) o The included studies have a small number of events. For example, for the outcome mortality, if there are only 2 deaths across all studies, the evidence will be downgraded for imprecision. o The 95% confidence interval of the cumulative effect includes the line of no effect and the confidence intervals of the included study are wide 6

17 Publication Bias is evident when: o Literature is systematically over- or underestimate of the treatment effect due to which studies are selected for publication Studies with small sample sizes are less likely to be published Negative studies are less likely to be published All the studies have strong links to commercial enterprises that might benefit from the results o Publication bias is assessed by evaluating funnel plots EXAMPLE For each criterion, if present, the quality of evidence is reduced by one. For example, four studies have been located for the comparison dexamethasone versus magnesium sulfate (IV) to treat refractory migraine in the ED. The outcome is Hospitalization. The studies start as High quality for grading summary (See Table A). The studies have: Risk of Bias- the study is methodologically strong- Low risk of bias- not downgraded Inconsistency- confidence intervals overlap and the I 2 is 0%- Low risk of inconsistency- not downgraded (See Figure A2) Indirectness- no direct comparison, extrapolated from comparison to another drug- High risk of indirectness- downgraded to Moderate Imprecision- there are four studies, the cumulative number of subjects is 299, and the number of events (number of subjects hospitalized is ). Imprecision is high because (a) the effect size crosses the line of no effect, and (b) the number of subjects and the number hospitalized are low, therefore the evidence has High risk of imprecision- downgraded to Low (See Figure A2.) Publication Bias- unable to assess, only four small studies are included- Stays at Low For this outcome, using the GRADE criteria, the quality of evidence for this comparison is Low. Table A. Grade Summary for the Example Quality of Evidence for this Outcome GRADE High Starting point Moderate Downgrade to Moderate due to Indirectness Low Downgrade to Low due to Imprecision Very Low Note: this evidence was not downgraded for risk of bias, inconsistency or publication bias. 7

18 Dexamethasone vs. Magnesium Sulfate IV Prochlorperazine vs. Magnesium Sulfate IV Dexamethasone vs. Prochlorperazine Indirect comparison Figure A. Example of Indirectness TREATMENT A TREATMENT B Risk Ratio Risk Ratio Study or Subgroup Weight M-H, Fixed, 95% CI M-H, Fixed, 95% CI STUDY %.20 [0.08, 8.38] STUDY % 0.39 [0.04, 4.] STUDY %.30 [0.2, 3.62] STUDY %.88 [0.8, 20.3] (95% CI) %.02 [0.32, 3.25] events 6 5 Heterogeneity: Chi² = 0.95, df = 3 (P = 0.8); I² = 0% Test for overall effect: Z = 0.03 (P = 0.98) TREATMENT A TREATMENT B Figure A2. Meta-analysis example for Grade Reference: Schünemann H, Santesso, N. (206). How to GRADE the Evidence. Retrieved from 8