Result – Same proportion of children symptomatic in both group at all time pointsDuration of hospital stay was shorter by 9 hours in the intervention group but statistically insignificantConclusion –
Not a big loss – both arms Could have imputedReason for loss to follow up not mentioned
Baseline characteristics not balanced, no mention of who dispensed the drug to the patientCallling it as intention to treatP values in baseline characteristicsNot stratified, shudhavminimisedPost hoc subgroup analysis
HAS REDUCED THE POWER OF THE STUDYCONFIDENCE INTERVAL POSITIVE, SO TREND IS POSITIVE DEFINITELYBASELINE P VALUES – NOT REQUIRED- ATTRIBUTED TO CHANCE
A CRITICAL APPRAISALAnuradha Bansal – Veena R. Parmar – Srikant Basu – Jasbinder Kaur– Suksham Jain – Abhijeet Saha – Deepak Chawla Presented at Workshop on RANDOMIZED CONTROL TRIALS- DESIGN, ANALYSIS AND REPORTING Indian Institute of Public Health (Delhi)
1. Summary of the article2. Basic Study Design3. Section-wise critical appraisal Title and abstract Introduction Material and Methods Results Discussion Conclusion4. Internal Validity and External Validity
Objectiv To evaluate the efficacy of zinc supplementation one duration of illness in children with severe ALRTI Children age group 2-24 months with severe ALRTI (InclusionP criteria) In a teaching referral hospital in North India 20mg of elemental zinc per day for 5 daysI Control group received an equal amount of placebo withC appropriate taste, smell, color and consistency similar to zinc mixture Primary - Time to be asymptomatic (Composite)O Secondary – Duration on hospital stay, Resolution of danger signs, respiratory distress, tachypnea and hypoxia in room air Superiority objectiveT
Randomized Control Trial, prospective in nature
Assessed for eligibility N=120 14 excluded Enrollment (parental refusal or exclusion Process criteria) 106 Randomized 53 - zinc 53 -control Randomization group group 52 - zinc group(1 left against medical 53 received Intervention advice) placebo Follow up 52 analyzed 52 analyzed (1 lost to follow up)
ITT? The study claims to have followed an intention to treat analysis whereas actually the trial profile shows a minor inconsistency In the intervention arm, one patient left after being randomized – the final result has EXCLUDED the patient and considered n=52 in this arm In the control arm, the loss to follow up has been INCLUDED in the size of the control group as n=53 here. (The n given in the baseline and outcome tables is different)
TITLE Mentions study design, intervention and population Does not mention the area from which the population has been studied ABSTRACT Objectives, methods, findings and conclusions have been well summarised No mention of the method of randomization INTRODUCTION Justification for the need for study and area mentioned (North India) has been clearly outlined and reflects clinical equipoise
Method Description CommentRANDOMIZATION Simple Randomization through a random Suspicious asTECHNIQUE number table exactly same size of control and intervention arm has been achievedALLOCATION Achieved by dispensing the drug and placebo Justified- but noCONCEALMENT in identically looking bottles mention of who dispensed the drug to the patientPREPARATION OF Placebo prepared to give similar taste, smell, Mentioned sourcesPLACEBO color and consistency compared to original and details zinc mixtureBLINDING Repeated mention of the trial being of a No mention of ‘triple blind design’ ‘WHO’ exactly was blindedDISCHARGE Asymptomatic for at least 24 hours. Objective in natureCRITERIA FORPATIENTS
SAMPLE SIZE Pilot study revealed that No detailed mention ofCALCULATION for 80% power and 0.05 way of arriving at sample two sided alpha, a sample size of 400 was needed which -effect size was brought down to 106 -change in primary due to time constraints outcomeSAMPLE SIZE OF THE N=106 Small to give any decisiveSTUDY Identified by the study as a result, does not resolve limitation the inconsistency of results established by studies before – Precision compromised therefore WIDE CONFIDENCE INTERVAL OBTAINED.
Time to be asymptomatic Median hours with IQR – range more or less same, statistically insignificant No. of children symptomatic Similar and statistically insignificant value Duration of hospital stay Shorter by 9 hours in the zinc group – result insignificant Time to resolution of respiratory distress, tachypnea, dangers signs and hypoxia Similar in both the groups, results statistically insignificant.
Sub Group Analysis Post Hoc is questionable, should have been a priori Multiplicity problem Secondary outcomes are correlated Primary and secondary also correlated Duration of stay- degree of subjectivity, not mentioned about blinded also Components of composite endpoint have equal importance in terms of relevance to the patient - more or less the same, least important is duration of stay
Zinc supplementation did not reduce recovery time and duration of hospital stay Not enough evidence found to reject null hypothesisThe conclusions are plausible.
Internal Validity Selection bias – Doubtful since the number in both the arms has been found to be equal despite simple randomisation Performance bias – Triple blind claimed but no mention of who all were blinded Attrition bias – no problem Detection bias – no mention of who assessed the outcome and who did what? Reporting bias – inconclusive
External Validity Given the inclusion and exclusion criteria which is generally applicable Inclusion of both wheezy and non wheezy children Very less proportion of population of slum areas, less even in intervention group -Pragmatism? Good proportion of anaemic patients, confounding variable Small sample size
Group 2 :Dr Sabyasachi Das, Aleesha Mary Joseph, Jalnidh Kaur, Anjali Sinha, Mr Suraj Parab, Dr Sunny Chaudhary, Dr Bhawna Shukla, Dr Rupal Shah