Answers to the Quiz on Evidence Based Health (with explanations)

Image for post
Image for post

Q1. Which of the following activities is the best example of evidence based health practice?

  • A heart surgeon conducted a heart transplant surgery
  • A university student critically appraised a journal article on an RCT of a new drug (X)
  • A lawyer presented evidence in defence of his client on medical malpractice
  • A professor lectured on the philosophical aspects of the theory of disease causation
  • An Epidemiologist conducted an investigation of a diarrhoea outbreak after a Sunday feast in a local church

Explanation: Critical appraisal is the most important aspect of evidence based health.

Q2. You have conducted a study on the association between weekly fast food consumption and prevalence of obesity among university students at the University of Canterbury (UC). Which of the following will limit the external validity of your study?

  • The association between fast food consumption and obesity found in UC students may not be the same for a wider NZ population (X)
  • UC students are not representative of the wider NZ population
  • As it was a small study, you did not find a statistically significant association between fast food consumption and obesity
  • You could not control for all potential confounding variables
  • You cannot establish if fast food consumption caused obesity among the students

Explanation: External validity is about whether the relationship between the exposure and the outcome or the intervention and the outcome will hold in another population, not so much as whether the population where you conduct the study is representative of the wider population or not.

Q3. You conducted a study (N = 200) among patients admitted to the Christchurch hospital on the association between self-reported cigarette smoking status and heart disease; you found that smokers in your sample had higher risk of heart disease (Odds Ratio = 2.5) and the results were statistically significant (p < 0.001 and 95% Confidence Interval: 1.26–3.25). Which of the following statements is a limitation of the internal validity of your study?

  • You were unable to rule out random error in your study
  • Narrowing of blood vessels that resulted from smoking and led to heart disease was a potential confounding variable you could not control
  • You do not know if cigarette smoking caused the heart disease or whether those with heart disease started smoking more
  • As you relied on the self-reported smoking status, you were unable to eliminate response bias (X)
  • Christchurch hospital patients are not representative of the general NZ population

Explanation: Internal validity is achieved when you rule out chance, eliminate bias, and control for confounding variables in the study. In this case, you were able to rule out chance (p < 0.001), narrowing of blood vessels cannot be confounding as this resulted from smoking and led to heart disease. Knowledge whether smoking caused heart disease or heart disease resulted in smoking is irrelevant here because the question is about internal validity. Hence, potential response bias will interfere with internal bias. Christchurch hospital patients are not representative of NZ population but this is more about external validity if at all.

Q4. Which of the following statements about the sample size for a study is correct?

  • A large sample size is always a good use of resources and time
  • With a large sample size, a statistically significant result may be irrelevant (X)
  • A tiny sample size helps to precisely point to the effect size
  • A study with smaller sample size will lead to a narrower confidence interval than a study with larger sample size
  • Detection of large effect needs a large sample size

Explanation: A large sample size will result in even small differences turning out to be statistically significant, so it does not tell us anything about the relevance of the study. Large sample size require lots of resources and time to manage; a tiny sample size will lead to a large confidence interval, so cannot be precise; and detection of large effect size requires small sample size.

Q5. You have conducted a double blind randomised controlled trial of a new antidepressant medicine, comparing with treatment as usual, on university students who were diagnosed with depression. You found that the new medicine had better outcomes than treatment as usual in controlling depression among university students (p = 0.002). Which of the following statements is correct about limitation of your study?

  • Your results are based on participants you selected from a university setting, this may limit the external validity of your study (X)
  • You were unable to control for selection bias in your study design
  • You were unable to control for potential confounding variables in your study design
  • You were unable to control for random error in the study
  • You cannot claim efficacy of this new medicine based on this study

Explanation: Randomised controlled trials are limited as they are not well suited to address external validity of studies. RCTs are good study designs to control for selection bias. RCTs are also good study designs to control for confounding variables that are seen and unseen. As your study has a small p-value, it suggests you were able to control for the play of chance, and this also will help you to claim for establishment of efficacy in your study. Hence the correct answer is the first choice, it will restrict external validity.

Q6. To investigate if wood splitting with heavy axe can lead to heart attacks (acute myocardial infarction, aka AMI), you conducted a case control study at a hospital with 200 patients who suffered heart attack (“cases”), and 200 patients who were admitted for some other conditions (“controls”). For “wood splitting with heavy axe” and AMI, you found Odds Ratio = 0.87, 95% Confidence Interval: 0.78–2.20. Which ONE of the following statements is true about this study?

  • The study found that wood splitting with heavy axe increases risk of heart attack
  • Using case control study, you were able to control for all possible confounding variables
  • A case control study is the best study design to establish causal linkage between “wood splitting with heavy axe” and heart attack
  • Your study suffered from selection bias against the cases as several people who died at home while splitting wood with heavy axes or died on the way to hospital, therefore did not reach hospital (X)
  • A randomised controlled trial with larger sample size would likely to find a significantly LOWER risk estimate

Explanation: The study finding of 0.87 as OR and associated confidence interval that straddled 1.0 suggests that you cannot be definite about the association. As this was a case control study, you could not control for all potential confounding variables, it is also not be “best” study design. A better study design would be a cohort study. Selection bias is a problem. An RCT that would eliminate selection bias would find a larger risk estimate.

Q7. A company advertised a new “turmeric tablet” in the market. In their advertisement, they showed a famous doctor who claimed that he tested the “turmeric tablets” twice daily on 200 arthritis patients who had severe pain & who were selected at random from his clinics; 150/200 (75%) patients reported that they were pain free after three months of treatment. the company claimed that their “turmeric” medication was effective treatment of arthritic pain. Which ONE of the following statements about this claim is CORRECT?

  • Their claim is correct as majority of the patients became pain-free after three months (75% patients, or 150/200)
  • Their claim is wrong as it failed to relieve pain for at least 50 patients (25% participants)
  • Their claim is correct as a famous doctor conducted the study, & turmeric is also known to be good for health
  • Their claim is wrong as there was no valid comparison group in the study (X)
  • Their claim is correct as about 3 times as many people were better with turmeric (75% vs 25%) than not.

Q8. You are about to assess the evidence on the effectiveness of screening programmes for cervical cancer. Which of the following statements is CORRECT about systematically developing an evidence base?

  • You should start with background reading, identifying experts, and developing a set of PICO formatted questions (X)
  • You should start with seeking an appointment with an information specialist
  • You should start with picking up a random article on screening and appraising it
  • You should find the latest case study on cervical cancer screening and base your analysis on it
  • All you need to do is to find the latest guideline on the topic and summarise the findings

Q9. You want to find out if smoking increases heart failure risk. You select 100 heart failure patients admitted to the local hospital and recruited 100 other patients in the same hospital but in different wards. You then took their histories of smoking. What kind of study did you conduct?

  • Case-control study (X)
  • Retrospective Cohort study
  • Cross-sectional survey
  • Case series
  • Randomised controlled trial

Q10. You are a health protection officer at a factory and you suspect some workers are exposed to mercury in the plant that results in nerve damage to these workers. You decide to study old health records of the workers and decide to follow them for the previous 10 years. You learn about workers who started working healthy, then got exposed to mercury in the plant and subsequently developed nerve disease. You also reviewed similar health records of clerks who worked in other sections of the factory and were never exposed to mercury, for comparison. What is your study design?

  • Retrospective cohort study ( X)
  • Prospective cohort study
  • Case control study
  • Cross-sectional survey
  • Randomised controlled trial

Q11. A rare disease is defined as one whose age-adjusted annual rate is less than 1 per 10, 000. Age-adjusted annual rate of the cancer of the pancreas is 11 per 100, 000 population. You want to find out if heavy alcohol consumption causes pancreatic cancer. Which of the following study designs is best suited for such a research?

  • Case control study (X)
  • Cross-sectional survey
  • Prospective cohort study
  • Randomised control trials
  • Case series or disease surveillance

Explanation: Case control studies are better suited than other observational epidemiological study designs to study rare diseases

Q12. You want to find out the prevalence of high blood pressure (“hypertension”) among university students in New Zealand. You invite all UC students to participate in the study and those who were willing to participate, you asked them to report to the UC health centre on the last day of the term for their blood pressures to be taken by a nurse. You will use the blood pressure measurements to estimate the prevalence of hypertension in university students. Which of the following best describes your study design?

  • Cross-sectional survey (X)
  • Case control study
  • Case series or disease surveillance
  • Cohort study
  • Randomised controlled trial

Q13. You work at the Canterbury District health board and they want you to report all new cases of gastroenteritis in healthy adults in Canterbury between 2018–2019. What study design should you set up to report gastroenteritis cases in the district between 2018–2019?

  • Disease surveillance using case series (X)
  • Cross-sectional survey
  • Prospective cohort study
  • Randomised controlled trial
  • Individual case studies

Q14. You are testing a drug to treat high blood pressure and you are conducting a double blind randomised controlled trial. What of the following is correct about double blinding?

  • Both you and your study participants do not know the conditions to which the study participants have been assigned (X)
  • Only you do not know the conditions to which the study participants have been assigned
  • Only your study participants do not know to which the study participants have been assigned
  • You are conducting the study without knowing if it will be funded
  • You are unaware as to what study findings will be

Q15. You want to study the effectiveness of lifestyle modification in prevention of heart disease. If you have to select one method to study the association, which of the following is MOST appropriate for you to consider when you select your methodology?

  • Is the method sophisticated enough for the problem?
  • Will the method generate enough quantitative data?
  • Is the method feasible to to be applied to the given problem? (X)
  • Is the method internationally recognised for solving the problem?
  • Will the method generate enough qualitative data to address the problem?

Q16. You have read a report that stated, “For men with heart disease and high cholesterol, use of the drug Lipicor as opposed to Placebo reduces the five-year chance of death from 9 percent to 5 percent”. What is the relative risk reduction (in % value) in reduction of heart disease related deaths if patients were to use Lipicor?

  • 4%
  • 44.4% (X)
  • 5%
  • 55.5%
  • Insufficient information in the data to assess

Explanation: Relative risk reduction is the reduction of risk relative to the baseline risk. The reduction of risk in this case was 4%, and the baseline risk was 9%, hence relative risk reduction was 4/9 = 44.4%

Q17. You have read a report that stated, “For men with heart disease and high cholesterol, use of the drug Lipicor as opposed to Placebo reduces the five-year chance of death from 9 percent to 5 percent”. Based on this data, how many people would have to be treated (NNT) with Lipicor in order to prevent death in one person?

  • 25 people (X)
  • 20 people
  • 2 people
  • 10 people
  • Insufficient information in the data to assess

Explanation: NNT is calculated as 1/ARR. ARR is absolute risk reduction and stated as the reduction of risk from baseline. In this case, the reduction from baseline risk was 4% (9% — 5%); hence 1/4% is 100/4 = 25

Q18. The following figure shows association between exercise and serum cholesterol levels (X axis = hours of exercise, Y axis = serum cholesterol level). On the left hand side, data shown for the entire population, on the right hand side graph, the same graph is stratified for 10-yearly age groups. The two figures are same, except the regression lines are different. Which of the following statements is CORRECT?

  • Age is confounding the association between exercise and cholesterol levels (X)
  • The more you exercise, the higher will be your cholesterol levels
  • Exercise and cholesterol levels are unrelated to each other
  • Younger people have a sharper drop in cholesterol levels than older people
  • Older people have sharper drop in cholesterol levels with exercise than younger people

Q19. Which of the following study designs is best suited to study causes of rare diseases such as cancers?

  • Case-control studies (X)
  • Cohort studies
  • Randomised controlled trials
  • Cross-sectional surveys
  • Case series or disease surveillance

Q20. You are testing the effectiveness of a newly developed drug (“D”) versus treatment as usual (“TAU”) on blood pressure control for people with high blood pressure (“hypertensives”). You obtain data from hypertensives who had high blood pressure at baseline (“high BP group”) and relatively lower levels of blood pressure at baseline (“low BP group”) and obtained the following data (refer to the table below):

```
Levels of blood pressure control in a study on the effectiveness of drug D versus TAU

| Condition | TAU group (N = 350) | D group (N = 350) |
|-----------|---------------------|-------------------|
| Low BP | 81/87 (93%) | 234/270 (87%) |
| High BP | 192/263 (73%) | 55/80 (69%) |
| Combined | 273/350 (78%) | 289/350 (83%) |
```

Based on the data presented above, which of the following statements is CORRECT?

  • We should recommend the treatment for all people (X)
  • We should recommend the treatment ONLY to the low BP group
  • Baseline blood pressure is a valid confounding variable in this study
  • We should recommend the treatment ONLY to the high BP group
  • We should NOT recommend the treatment to EITHER the low or the high BP group

Explanation: We should recommend the treatment for all people because when combined, the drug D was found to be beneficial. Baseline blood pressure is a bogus confounding variable as this is unrelated to the dose administered, and if the goal is reduction of blood pressure, then it does not matter if the blood pressure at baseline was high or low.

Q21. In your community, the prevalence of cigarette smoking among adults is 20%. You know the relative risk of cigarette smoking for lung cancer is about 20.0 (i.e., RR = 20.0). Assuming cigarette smoking is the ONLY cause of lung cancer, if you were to achieve 100% smoking cessation in your community, by how much would the incidence of lung cancer reduce? (Calculate the population attributable fraction of smoking cessation for lung cancer). Select the CORRECT value from the following figures

  • 79.167% (X)
  • 40%
  • 20%
  • 10%
  • 35%

Explanation: PAF (population attributable fraction) is given by the following formula:

Prevalence of exposure * (RR-1) / ((prevalence of exposure * RR — 1) + 1)

Here, this will be 0.20 * 19/ (0.20 * 19 + 1) = 3.80/4.80 = 79.167%

Q22. In your community, 10% of the population are exposed to Radon gas, 10% exposed to asbestos, and 20% people smoke. Exposure to all three (Radon, asbestos, and smoking) are high risks of lung cancer, with RRs between 10–20. You found that when you add up the population attributable fractions, they amount to 174% of the burden of lung disease. Which of the following statements is a CORRECT explanation of why the three exposures add up to174% of the burden of lung cancer?

  • Some people may have been exposed to radon and asbestos and smokers as well — they have multiple exposures (X)
  • These figures are overestimation of the risk
  • These figures are underestimate the risk
  • This figure is not correct as not all people in my community will have RR of 20 for smoking and lung cancer
  • These figures are artificially inflated measurement of risk

Q23. You have conducted a meta-analysis of randomised controlled trials to study whether aerobic exercise can prevent heart disease among people over 50 years of age, and found that regular aerobic exercise can prevent heart disease (pooled Odds Ratio = 6.3, 95%CI: 5.2–8.5). Considering you used only meta-analysis of RCTs with high effect size, which of the following is the CORRECT listing to support causality?

  • Strength, Consistency, Temporality (X)
  • Strength, Biological Gradient, Plausibility
  • Biological gradient, specificity, plausibility
  • Plausibility, coherence, analogy
  • Biological gradient, plausibility, analogy

Explanation: By convention, 6.3 is strong effect; as this is based on meta analysis this addresses that the studies were consistent in that direction, and as this estimate is based on a meta-analysis of RCTs, therefore the pooled estimate as it is based on RCTs, also satisfies the temporality criterion. We do not have any evidence of biological gradient, and we do cannot claim about plausibility or analogy either.

Q24. Which of the following can be used as an argument against “Temporality” criterion of Hill’s criteria?

  • Consumption of fried food can cause depression (X)
  • Infection with influenza virus causes “Flu”
  • Touching a hot barbecue griddle can cause burns to your hand
  • Drunk driving of the driver can cause death of the driver from motor vehicle accidents
  • Overdosing with drugs can lead to instant death

Explanation: The only selection where either fried food can lead to depression or depression can lead to increased fried food consumption, hence we are not sure which preceded which. For all the other choices, this is not the case. Flu cannot lead to influenza virus infection, if you burn your hand already, you cannot claim to touch the griddle and heat it with that, and if you are dead, then you cannot go back to drink or get overdosed. These things would have to happen before you burn your hand, or you die.

Q25. Sherlock Holmes advised Dr Watson in “Sign of Four”, “… when you have eliminated the impossible, whatever remains, however improbable, must be the truth”. Which ONE of the Hill’s criteria fits best with this statement?

  • Plausibility (X)
  • Strength
  • Consistency
  • Temporality
  • Biological gradient

Q26. You want to investigate whether cigarette smoking causes lung cancer. You know that people above 50 years of age, and men are more likely to both smoke and have lung cancer. You also know that smoking causes changes in lung tissue you can measure that in turn lead to lung cancer. Which of the following will be a confounding variable in your study between lung cancer and smoking?

  • Age (X)
  • Number of packs of cigarette smoked
  • Lung tissue changes
  • Residence of the person
  • Treatment received

Q27. Observational studies have suggested that menopausal hormone therapy (HRT) are associated with lower the risk of heart attack and stroke for women with heart disease. But RCTs now shows that women with heart disease should not take it. Which of the following statements is CORRECT about explaining the differences?

  • Selection biases may explain the cardiovascular benefits found in observational studies (X)
  • There would be no difference between women who used HRT and those who did not use HRT in terms of their other risks for heart diseases
  • Healthier, wealthier, and more educated women would avoid using HRT in earlier studies
  • RCTs are interventions while observational studies are based on observations of behaviours only, hence the difference
  • Findings that come later in time ALWAYS update findings in the initial phases of studies

Explanation: observational epidemiological studies have selection biases; randomised controlled trials overcome this limitation because of randomisation. None of the other choices are feasible or practical or can be deduced from the information presented and hence rejected.

Q28. You conducted a case-control study on the association between coffee consumption and pancreatic cancer with adequate sample size; pancreatic cancer cases were selected from the cancer ward of the hospital and healthy controls were general members of the public. You took history of their coffee consumption by asking them about their coffee consumption patterns. Which of the following is a threat to internal validity?

  • Chance might be a threat to the internal validity
  • The study would have low power
  • The findings may not be consistent
  • A dose-response relationship would not be established
  • The study is subject to recall bias (X)

Q29. You conduct a survey of students at the University of Canterbury to find out the prevalence of low back pain among students using clinical examination. Which of the following statements is correct?

  • You may underestimate the true prevalence as students with low back pain who are too sick to attend the university on that day are excluded (X)
  • You will overestimate the prevalence of low back pain in the student community
  • You should set up a surveillance for low back pain among students to estimate the prevalence of low back pain
  • Your study will suffer from recall bias
  • Your study will suffer from low sample size

Q30. Which of the following statement is CORRECT about controlling for confounding variables in a study:

  • Randomisation cannot control for unobserved confounding variables
  • You can control for confounding variables ONLY at the study design phase
  • Matching will also allow for elimination of recall bias
  • Multivariate analysis only allows for a single confounding variable
  • Using stratified analysis you can control for more than one confounding variables (X)

Q31. Which of the following statements is correct about Randomised controlled trials

  • Randomisation only controls for known confounders
  • Intention to treat analysis is useful for addressing problems that arise due to cross-over of the participants (X)
  • In intention to treat analyses, the principle is, “once randomised, never allowed to cross over”
  • Randomised controlled trials are excellent study designs for maintaining external validity
  • In RCT, participants are selected on the basis of their disease status

Explanation: The goal of intention to treat analysis in RCTs is to ensure “once randomised, always analysed”

Q32. You are appraising a case-control study on the association between smoking and myocardial infarction (heart attack) and found Odds Ratios for ex-smokers as 2.4 (95% Confidence interval: 1.3–3.2), and for current smokers 4.7 (95% Confidence Interval: 3.4–6.8); your reference category was “never smokers” (hence OR 1.00). You adjusted for age, education, cholesterol, high blood pressure, and BMI. Which of the following statements is CORRECT about the study findings:

  • Smoking is a risk factor for heart attack in your study (X)
  • Cigarette smoking protects against heart attack
  • In this study you did not find any evidence of dose response effect between smoking and heart attack
  • While smoking increased the risk of heart attack, the results were not statistically significant
  • You were able to control for unknown confounders in the study as well

Q33. Which of the following sequences correctly orders the hierarchy of evidence in terms of highest to lowest in the order of evidence (best to the least)?

  • Meta-analysis of RCTS (highest) >> RCTs >> cohort studies >> case control studies >> descriptive studies (lowest) (X)
  • Case control studies (highest) >> descriptive studies >> anecdotal evidence >> meta-analyses (lowest)
  • Meta-analysis of RCTS (highest) >> case control studies >> cross-sectional surveys >> anecdotal evidence >> descriptive studies (lowest)
  • RCTs (highest) >> Meta-analysis of RCTs >> Case control studies >> Cohort studies (lowest)
  • Cross-sectional surveys (highest) >> RCTs >> Cohort studies >> Case control studies (lowest)

Q34. Franz Messerli (2012) published a study in the New England Journal of Medicine and reported a significant linear correlation (r=0.791, P<0.0001) between per-capita chocolate consumption and the number of Nobel laureates per 10 million persons in a total of 23 countries. When Sweden was excluded, the correlation coefficient increased to 0.862. They stated, “The slope of the regression line allows us to estimate that it would take about 0.4 kg of chocolate per capita per year to increase the number of Nobel laureates in a given country by 1”. Based on this, which of the following statements is correct?

  • The more chocolates you eat, the higher will be your chance of winning the Nobel Prize
  • As the results of the study were in aggregated form, it does not mean that an individual consuming chocolate will also have a higher chance to win the Nobel Prize (X)
  • Correlation may imply causation here as the regression slope indicates
  • The results suggest a dose-response relationship between chocolate consumption and individual’s chance of winning the Nobel Prize
  • As the study was published in the New England Journal of Medicine, the study has high level of evidence about the benefit of chocolate consumption

Explanation: You cannot arrive at conclusions about individual exposures and risks from research that was conducted at aggregate levels (ecological fallacy)

Q35. Which of the following is true about biases in randomised controlled trials:

  • Randomised controlled trials are free of biases
  • Single and double blinding are done to eliminate selection and response biases (X)
  • Allocation concealment is not possible using sealed envelopes
  • Randomisation is when subsequent patients are allocated to control or the interventions
  • Blinding of the investigators is needed for control of confounding variables

Associate Professor of Epidemiology and Environmental Health at the University of Canterbury, New Zealand. Also in: https://refind.com/arinbasu

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store