Whether or Not Longer Durations of Breastfeeding Caused Anemia in Young Children - Assignment Example

? EPID: Epidemiology B QUESTION 9 marks) A study aimed to assess the prevalence of anemia in children from 1 to 2 years in a country. Haemoglobin was measured by a Hemocue photometer and anaemia was defined as haemoglobin less than 8 g/100 mL blood. Duration of breastfeeding was determined by asking parents or carers. There were 4500 mothers or careers and their children who were asked to participate in the survey, of whom 457 refused to participate, and 243 had an incomplete questionnaire and/or test results. Among 1300 children breastfeed for less than 1 year, 520 had anaemia compared to 500 children with anaemia in the group of 2500 children breastfeed for one year or more (note: these are hypothetical figures only). Answer the following questions: i) What design was used in this study? What are the main limitations of this study design? How would you use this study to address the question of whether or not longer durations of breastfeeding caused anemia in young children? [1.5 marks] This is a cohort retrospective study. Since, its mode of study entails probing for more information regarding the children’s breastfeeding history from their mothers or caregivers. Limitations of the study The study exclusively relies on the subjects’ information where there is high probability of recording wrong information or data (Aschengrau & Seage, 2008). Since, not all members will be able to remember exact durations, which they breastfed their children. The researcher does not have any control over the outcome assessment because he or she relies on hearsay information to prepare final data. ii) What sampling method should be used in this study? What would be the advantages and disadvantages of the method selected? [1.5 marks] Quota sampling is appropriate for this study. Advantages: The researcher will be able to ascertain in each of the two categories (Breastfed over 1 year and those breastfed for less than a year) individual anemia prevalence and feature them in percentages. It is extremely easy in arranging and developing of diverse categories from the entire population. iii) What measurement issues should be considered? [2 marks] Measurement bias – Researchers may face the temptation of manipulating results due to the knowledge they have concerning benefits of prolonged breastfeeding (Aschengrau & Seage, 2008). This is especially when the mothers or caregivers seem not to remember exactly the durations, which they breastfed their children (Aschengrau & Seage, 2008). Therefore, they end up suggesting dates to the mothers or caregivers with the intention of calculating implied durations where they may wrong. iv) What effect measure would you use to estimate the association between anaemia and breastfeeding? Calculate this measure of association. Please show your calculation, including a 2x2 table. [2 marks] The most suitable effect of measure is the relative risk. Table 1 Breastfeeding Duration Anaemia + Anaemia - Total Over 1 year 500 2,000 2,500 Less than 1 year 520 780 1,300 Total 1,020 2,780 3,800 Exposed group < 1 year Non exposed group > 1 year RR = (500/2,500)/ (520/1300) = 0.5 v) What biases might occur in the study and how would you minimize them? [2 marks] Researchers’ preconception (informational Bias), where some hold to the notion anaemia prevalence is extremely high among children breastfed for less than one year (Aschengrau & Seage, 2008). Consequently, this prompts some researchers to manipulate figures especially when a child’s mother or caregiver seems not to recall vividly the duration she breastfed. Reflectional bias Measurement bias Solution: Therefore, researchers always should avoid prejudgements or suggesting to the participants what they want and instead probe further for reliable information. QUESTION 2 (9 marks) In a study examining the possible association between chronic unspecified headache and the risk of stroke, 1699 randomly selected Finnish men aged 25 to 64 years were recruited into cardiovascular risk factor surveys in 1972, 1977, 1982 and 1987. Self-reported headache, smoking, diabetes, blood pressure, weight, height, and serum cholesterol level were recorded at baseline. During the follow-up, 118 incident stroke events were ascertained with computer-based record linkage. No of participants No of stroke cases Person years follow up Headache 148 24 3267 No headache 1551 94 34 115 Total 1699 118 37 382 Answer the following questions: i) What kind of study design was used? What are the advantages and disadvantages of this study design? [2 marks] Study design used in this scenario is Cross-sectional survey (Aschengrau & Seage, 2008). Advantages This design is very cheap compared to other methods of research because it entails fewer resources meant to attain the intended population’s information. In addition, it is the best way to ascertain the prevalence of any disease in a population, within a given time without making a follow up (Aschengrau & Seage, 2008). Offers string and reliable information regarding causation of disease Has a wide applications especially when it comes to multiple outcomes Allows the researcher to ascertain diseases’ incidence Disadvantages The major predicament with this method is to ascertain cause and effect of the issue under study. This is because the design’s mode of study is similar to “snapshot analysis” which entails disregarding information concerning diverse short causes of the problem (Aschengrau & Seage, 2008). Reliable only when the history of the subject under study is known ii) What study exposure measurement issues should be considered by the investigator? [2 marks] The selected sample representing generalized population under study should be sufficiently large to allow effective estimation’s precision (Aschengrau & Seage, 2008). Sample selection should ensure effective representative of the entire population bearing the intended prevalence meant for study. Conceptualizing the appropriate construct of dose Conceptualize the actual active agent that is hypothesized to be causally related to disease iii) What effect measure would you use to estimate the association between headache and stroke in men? Calculate this measure of association, showing your calculations. [3 marks] I would utilize Relative Risk (RR) Exposed group < 1 year Non exposed group > 1 year RR = (24/148)/ (94/1551) = 2.674 iv) What bias might occur and how to avoid them? [2 marks] Selection bias The remedy to this bias entails capturing large target population to increase the study’s accuracy (Aschengrau & Seage, 2008). This will minimize errors, which emanate from having limited or small population and excluding high prevalence found in other sections not included in the sample. Question 3. (9 marks) Myringotomy (an incision in the ear drum membrane) with the insertion of tympanostomy tubes (small tubes to keep the ear drum incision open and allow fluid from the middle ear to drain) is the most common operation among children beyond the newborn period in the United States; an estimated 280,000 children younger than three years of age underwent the operation in 1996. A main indication for the procedure of insertion of tympanostomy tubes in infants and young children is persistent otitis media with effusion (infection of the middle ear with excess fluid accumulation), and concern that this condition may cause lasting impairments of speech, language, cognitive, and psychosocial development. However, evidence of such relations is inconclusive, and evidence is lacking that the insertion of tympanostomy tubes prevents development impairment. A recently reported randomised controlled clinical trial was designed to address whether prompt insertion of tympanostomy tubes protects against or minimises subsequent developmental impairment. Subjects for the trial were recruited from eight public and private clinics. Below is a figure reporting the flow of participants in the trial. i) What inclusion and exclusion criteria would you use to select subjects for the trial? Explain how using these criteria to include or exclude subjects would improve the design of the trial. [2 marks] Inclusion and Exclusion Criteria Participants should be under 3 years old and free from other fatal health complications (ear' predicaments) that may worsen during the study. In addition, subjects meant for operation should not encompass those who may be having ear complications (Aschengrau & Seage, 2008). These criteria aid in avoiding cases that may lead to the inaccuracy of the research due other bodily complications, for example, ear illnesses. ii) What randomisation scheme would you use and why? [1 mark] Stratified randomization scheme The scheme helps in capturing diverse and numerous groups effectively where I will be able to attain exact and detailed results. iii) The treatment in this trial, early versus late insertion of tympanostomy tubes, cannot be blinded. What types of biases might arise because of this lack of blinding? What would you do to minimise these biases? [2 marks] Assessor bias Laboratory bias I will advise researchers to record exact results despite being aware of both the subjects’ condition and what the study seeks to ascertain (Aschengrau & Seage, 2008). That is, they should avoid preconceptions that will lead to compromised results. iv) Explain what is meant by analysing data in clinical trials using the intention-to-treat principle? [2 marks] It is a data method analysis, which entails examination of all randomized categories regardless whether they received treatment, or not (Aschengrau & Seage, 2008). Groups found under certain arm of the study normally receive similar evaluation with no consideration whether they did complete the dose or received the necessary regimen (Aschengrau & Seage, 2008). v) Which trial subjects in each treatment group you would include in the final analysis keeping in mind the timing of the insertion of their tympanostomy tubes? [2 marks] Within 30 days after randomization Within 60 days after randomization Within 180 days after randomization At the age of 3 years QUESTION 4 (8 marks) Below is a graphical summary of the results of a meta analysis by Rahman et al which was published in the International Journal of Epidemiology in 2010. The graph presents the odds ratio and relative risks and 95% confidence intervals [CI] for colon cancer with exposure to chlorination disinfection by-products in water. Answer the following questions. i) How would you interpret the first study conducted by Wiilkinson JR (reference number 1)? How would you interpret the last study by King (reference number 10)? What might account for the different results in these two studies? [2 marks] Wiilkinson JR study has a very small weighting compared to the entire Meta analysis results, hence implying its influence is insignificant. In addition, the study is less accurate due to its wide CI. King WD’s study has a significant influence compared to the Meta analysis results. Difference between the two studies emanate from their varying weighting and the CI width. ii) How would you interpret the overall result of this meta analysis based on the graphical summary? [2 marks] The graph’s overlapping CIs imply the entire study is homogenous. iii) One journal article reports that the Q-test for heterogeneity yields p=0.2. What is this test and how do you interpret this value? [2 marks] There is no heterogeneity. This is because heterogeneity occurs when p-value < 0.05 (< 0.10), hence we can allow null hypothesis. For heterogeneity, the statement is vice versa. Fig 3: Funnel plot for colon cancer iv) What is publication bias and how would you check for it? By looking at the above funnel plot, could you please comment if there is any publication bias present or not? [2 marks] This is the tendency of reporting study’s findings towards a preferred direction. In ascertaining publication bias, I will check the distribution of diverse studies depicted in the above funnel plot (Aschengrau & Seage, 2008). There is no publication bias in the above figure because the distribution of the studies is even. QUESTION 5 (5 marks) State 'T' (true) or 'F' (false): i) Stratification of the study subjects mainly improves the precision, not the generalisability of the study findings. [1 mark] T. ii) Case-control studies always allow the temporal sequence between exposure and disease to be clearly established. [1 mark] F. iii) Random error usually shifts the odds ratio towards null. [1 mark] T. iv) When a variable is found to be both confounding and an effect modifier, adjustment for this variable is contraindicated. [1 mark] T. v) Inadequate concealment of the randomisation sequence is a form of information bias. [1 mark] T. References Aschengrau, A. & Seage, G. R. (2008). Essentials of epidemiology in public health. Sudbury, Mass: Jones and Bartlett Publishers. Read More