Epidemiology: Measuring & Describing Disease 2 Flashcards

1
Q

What are 4 different approaches to classifying research?

A
  • Heirarchyof evidence
  • Research methods: - Qualitative vs Quantitative research
  • Study Design types - Observational research - Intervention research
  • Epidemiological approach types - Descriptive epidemiology - Analytic epidemiology
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Define Qualitative research

A

Smaller numbers of participants, but goes in substantial detail inform research questions used earlier in research process

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What numbers and measures do we use in epidemiology?

A

-Measures of frequency and association - Comparisons and adjusting for differences

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Where do our measures in epidemiology come from?

A
  • Descriptive epidemiology - Observational and interventional study design - Systematic reviews and meta-analysis
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

How are epidemiological findings interpreted

A
  • Association, causation, validity and bias - Confounding and effect modification
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is the name of the transition in which “more resource constrained societies suffer from infectious diseases, which overtime are broadly overcome by improved access to water, sanitation, hygiene, vaccines and antibiotics. In their place, non-communicable diseases take hold: cardiovascular disease, dementia and cancer.”

A

Epidemiological transtion

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Define DALYs

A

DALYs – that’s Disability Adjusted Life Years. - The DALY is a measure of disease burden that combines years of life lost from ill-health, disability or premature death.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Which diseases cause the highest DALYs

A

https://vizhub.healthdata.org/gbd-compare/

IHD - Cardiovascular

and neoplasms highest cause of DALYs

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What are the 3 groups of conditions that cause DALYs?

Which group is most responsible for DALYs

A
  • non communicable diseases
  • communicable, maternal, neonatal and nutritional diseases
  • Injuries
    https: //vizhub.healthdata.org/gbd-compare/
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

What is the leading cause of death in the UK in 2017?

https://vizhub.healthdata.org/gbd-compare/

A

IHD - Ischaemic heart disease

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What’s the leading cause of morbidity (using DALYs as the measure) between ages 15-49

https://vizhub.healthdata.org/gbd-compare/

A

Back pain

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What’s the leading cause of mortality in the uk between ages 15-49?

https://vizhub.healthdata.org/gbd-compare/

A

self-harm

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Which cause of death accounts for the greatest modifiable behavioural risk among 15-49 year olds in the UK?

https://vizhub.healthdata.org/gbd-compare/

A

Drugs

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Measures for continous variables

A

mean, median, mode, range..

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What are the measures of frequency for discrete variables

A
  • Odds
  • Prevalence
  • Cumulative incidence
  • Incidence rate
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Define ratio

A

The term ratio describes a number obtained by dividing one quantity by another.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

Define odds

A

Odds is the ratio of the probability of an event to its complement.

  • it is the ratio of the number of people who have the disease to the number of people who don’t have the disease.

= P/ (P − 1)
P = probability of an event
P – 1 = probability of its complemenT

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

What is the epidemiological definition of odds?

A

The ratio of the probability (P) of an event to the probability of its complement (1-P).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

Calculate the odds:

In a tutor group of 12 students at Imperial College School of Medicine, 10 students have diligently completed their guided online learning before the tutorial, and 2 have not.

What are the odds that any student has diligently completed their guided online learning among this group?

A

5:1

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

Calculation of odds

In a tutor group of 12 students at Imperial College School of Medicine, 10 students have diligently completed their guided online learning before the tutorial, and 2 have not.

What are the odds that any student has not diligently completed their guided online learning among this group?

A

0.20

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

Interpretation of odds

In the context of odds, what is the output number above which it becomes more likely an event takes place than does not take place?

A

1

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

Define prevalence

A

The proportion of individuals in a population who have the disease or attribute of interest at a specific timepoint. (snapshot at a timepoint)

= p/n

  • *P** = no. with an attribute
  • *n** = total no. of individuals in entire population

Dimensionless number – no units. (always specify timepoint)

measured b/w 0-1 or 0%-100% (where 100% means everyone has the disease and 0% means no one has the disease)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

What are the limitations of prevalence as a measure

A
  • Reflects both the occurrence and duration of a disease (can not know whether information
  • Provides no information on new cases of a disease
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

Pros and cons off prevalence and odds as measures

A
  • Can describe the health of a population and monitor trends of disease over time
  • Enable planning of health services and allocation of healthcare resources
  • Less helpful in diseases of short duration and causal inference
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
Q

During the baseline assessment of ICSM Lifestyle Tracking Study in 2019, 22 students (out of 61) reported using the Tube to attend class.

What is the prevalence of Tube use to attend medical school among the first-year medical students responding to the survey?

A

0.36

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
26
Q

Interpretation of prevalence

What are some possible limitations of the previous statistic that X% of medical students used the Tube as their primary means of attending class? Select all that apply.

A
  • The sample (n=61) compared to the population (N=360) is small and therefore there is uncertainty around the point estimate of X%.
  • That non-response bias is possible: insomuch as students who took part in the baseline survey for ICSM LTS may differ systematically from the first-year student body overall.
  • That the finding is only valid for the time point in question: it’s possible that students’ means of travel may change over the course of the year.
  • That the finding may only be valid for first-year students as students in subsequent years may differ systematically in their place of residence and hence their choice of transport modality.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
27
Q

Define cumulative incidence

A

the proportion of the population with a new event during a given time period
=X/Y

  • *x =** no. of new cases during a period of interest
  • *y =** no. of disease-free individuals at the start of this period

Known cases are not included in either the numerator or denominator

Alternative / interchangeable terms

  • Risk
  • Incidence proportion
28
Q

Describe the units of cumulative incidence

A

No units

Dimensionless

State time period

measured b/w 0-1 or 0-100%

29
Q

Describe the requirements of the follow up period in measuring cumulative incidence

A
  • Must be the same for all participants.
  • During the follow up no new participants can join.
  • Inaccuracy occurs if participants are lost to follow up e.g. from death from an unrelated disease.
30
Q

What is the epidemiological definition of cumulative incidence?

A

The proportion of the population with a new event during a given time period.

31
Q

Calculation of cumulative incidence

Let’s assume that infection X is a mild disease, giving rise to life-long immunity.

In a tutor group of 12 students, 10 students arrive at medical school for on-campus teaching. Three of the students already report having recovered from infection X. During the first term, three students of the students attending campus report symptoms of infection X. During the second term, two others report symptoms of infection X.

What is the cumulative incidence of infection X during term 1 among students who are on-campus?

A

43%

32
Q

What is a person-time measurement

A

Person-time is an essential element in calculating incidence rate

Person-time is an essential element in calculating incidence ratePerson-time measures the time participants spend in the study. It stops when diagnosis of interest is made, participant dies or is lost to follow up.

Can be expressed in a range of units, for example

  • Person-years
  • Person-days
  • Person-hours
33
Q

Use the example of ‘classroom sleeping disease’ to calculate person-time contribution to the study

A
34
Q

Define Incidence Rate

A

= X/T

X= no. of new cases during the follow-up period
T = total person-time by disease-free individuals

Can range from 0 - Infinity

ALWAYS state person-time units

Always expressed by unit of person-time

Rates: Can only expressed as new cases per unit of person-time

Accounts for the time of follow up and for time when the new event occurred

35
Q

When is Incidence rate a suitable measure

A

Suitable for studies when:

  • Participants enter and leave the study at different times
  • Participants are lost to follow up
  • There are competing risks

Can be used even when cumulative incidence can not

36
Q

How is incidence rate calculated?

A

The count of new cases during the follow-up period, divided by the total person-time.

37
Q

Calculation of incidence rate

Participant 1: 8 years in study, with disease falling in the period of investigation

Participant 2: 5 years in study, no disease

Participant 3: 5 years in study, no disease

Participant 4: 2 years in study, no disease

What is the incidence rate of disease in the study above?

A

0.05 cases per person-year

(proportion of new cases in a follow up period)

38
Q

Question 1: Your clinical director wants to understand what the likely year-on-year cost of a new drug is going to be, that will replace the need for traditional anti-hypertensive medicines, including in patients on existing anti-hypertensive therapies. Which epidemiological measure would be most useful?

A

Question 1: The correct answers are prevalence and hypertension. Hypertension is a chronic disease and this means that prevalence is a sensible measure.

39
Q

Question 2: Your Director of Public Health wants to undertake cost-effectiveness research into the value of influenza vaccines for persons over the age of 65 years. Using data taken from a select group of GP health records in England, which epidemiological measure would be most useful?

A

Question 2: The correct answers are incidence rate and influenza. In this case because influenza is a disease that occurs for a short time period, incidence is a better indicator of disease burden than prevalence. For example - if seven people have the disease in a week then the incidence will be 7/week. However the prevalence will be much lower as some people may recover in that time - making a cross-sectional measurement prone to under-estimation.

40
Q

Question 3: The Professor of Clinical Infection (who heads your laboratory) has conducted a 24-month study looking at the effectiveness of a new tri-valent influenza vaccine. In a seemingly unique turn of events, all (100%) participants were followed up from day 1 to day 730 of the study. Which outcome measure would best approximate to the effectiveness of the vaccine in the cohort receiving the vaccination?

A

Question 3: If you chose incidence rate for Question 3, then you’re not totally wrong. But cumulative incidence is going to give a more intuitive output statistic than incidence rate. The disease should be influenza.

41
Q

What is the name of the process/approach in which you can adjust measurements of frequency to facilitate comparisons.

A

Standardisation

42
Q

Does incidence give us the full picture alone when comparing

A

No – many other factors to take into consideration so can not compare incidences in populations without standardsation

43
Q

Process of standardisation

A

can adjust measurements of frequency to facilitate comparisons.

eg. We want to understand whether the difference in incidence might be down to their different demography: sex and age.

44
Q

Two types of standardisation

A

There are two different types of standardisation:

Direct standardisation – this gives a similar incidence - eg. 120 strokes per 100k/yr

Indirect standardisation - this gives a ratio out of 100 (or sometimes 1.0)

45
Q

Define direct standardisation

A

Gives comparable incidence/prevalence

  • Allows comparison of like-for-like between populations – normally age or sex

Calculation overview: (won’t be asked to calculate!)
1) Use age/sex specific data from two
populations
2) Calculate crude incidence/prevalence rate
for each age/sex grouping
3) Apply these rates to a standard population
(e.g. the European Standard Population) to
find expected rates for each population
4) Calculate age/sex-specific standardised
incidence/prevalence by dividing each
total expectation over each total
population

* For age Specific standardisation - need to knw age-specific date (of you don’t then its indirec standardisation)

46
Q

how do population pyramids displaying age help us compare populations

A
  • can adjust data to be age-specific using direct-standardisation
47
Q

Implications of direct standardisation

A

When you are looking at any data, and drawing comparisons between two areas, check to see whether the data are standardised, and if so, for what./ (age, sex…)

48
Q

Define Indirect standardisation

A
  • Gives ratio out of 100
  • When direct standardisation can’t be used
  • Use National Statistics (e.g. National Mortality) to find expected values
  • Comparison of observed : expected (e.g. SMR=standardised mortality ratio, SIR=standardised incidence ratio)

Indirect standardisation is useful when we have only high-level data about outcomes, but we can’t make a direct comparison. They are often the first step on a journey of enquiry.

49
Q

How do we calculate SMR/SIR

A

observed count/expected count

50
Q

difference between SMR or SIR

A

Standardised Mortality Ratio (SMR) and Standardised Incidence Ratio (SIR) are used commonly in population health practice.

§SHMI data for hospital performance identifying hospitals that report higher than expected mortality

§Standardised mortality ratio (SMR<75) as a marker of healthy life expectancy

51
Q

When to use standardisation

A
  • When comparing two areas, populations or institutions, look to see whether your data are standardised. And if so – what are they standardised for? Age, sex or something else?
  • If after standardisation there remains a difference, then we can assume that there are other factors which may be affecting the outcome of interest – beyond the variable (or variables standardised for).
  • A range of other statistical approaches are possible and commonly employed in research – in particular where we can look at individual-level patient data. Yet in practice and health service settings, this type of what we call granular data are oftentimes not available.
  • This means that we have to infer what we can from aggregated datasets – where we might only be able to see top-level outcomes for a population – for example how many people died – even though we’d prefer to know which ones!
52
Q

Granular data vs aggregated data

A

granular data = detailed data

Aggregated data = summaries

53
Q

Choose a standardisation approach

Requires that you know variable specific measures – such as age specific incidence in the institution or geography of interest – in order to conduct the standardisation

A

Direct standardisation

54
Q

Which standardidisation requires that you know a benchmark measure – such as national incidence rate – in order to conduct the standardisation by applying the national age-specific incidence against the age structure of the institution or geography of interest.

A

Indirect standardisation

55
Q

Which standardisation is being used if standardising incidence rate, outputs in units of count / time.

A

Direct standardisation

56
Q

Which standardisation is being used If standardising incidence rate, outputs as a standardised incidence ratio

A

Indirect standarisation

57
Q

If an occupation reports a standardised mortality ratio (SMR) of 1.00 (or 100), how can be interpreted?

A

Mortality in the professional group is as expected.

58
Q

If a local area reports a standardised mortality ratio (SMR) of 1.17 (or 117), How can be interpreted?

A

Mortality in the area is 17% higher than expected.

59
Q

If a local hospital reports a standardised mortality ratio (SMR) of 0.93 (or 93), how can be interpreted?

A

Mortality in the area is 7% lower than expected.

So remember when dealing with SMRs, the midpoint is 1.00 (or 100). Anything above this implies additional risk. Anything below this implies reduced risk. But of course it depends what you’re measuring. Increased survival is the same as decreased mortality.

60
Q

using the example of increased hospital deaths, explain the different types of variations and what we can infer

A

e.g. high hospital deaths

  • Unwarranted variation = hospital is dangerous
  • Explained variation = hospital has more high risk procedures and therefore may have more deaths
  • Statistical artefact = hospital is better at recording deaths than others
61
Q

Explain SHMI

A

Summary Hospital-level Mortality Indicator builds on work that’s taken place at Imperial. The SHMI uses the process of indirect standardisation to produce ‘expected’ number of deaths by a series of adjustments taking into account the volume of cases, blend of diagnoses and casemix adjustments for underling demography and health status variation of patients. A range of adjustments for casemix are undertaken and used to calculate an SHMI value (ranging from 0.6 to 1.2).

62
Q

Explain this SHMI Funnel plot

A
63
Q

Whhich 4 measures of disease can allow different epidemiological inferences to be achieved

A

odds, prevalence, cumulative incidence and incidence rate

64
Q

When is standardisation best used

A

Standardisation is commonly applied to descriptive epidemiological data sets enabling more meaningful comparisons to be drawn

65
Q

Why is direct standardisation important and what are the most commonly observed adjustment variables

A

Direct standardisation enables comparison of incidence and prevalence data by adjusting outputs by one or more other variables: age and sex of the most commonly observed adjustment variables use in direct standardisation

66
Q

When is indirect standardisation used

A

Indirect standardisation is an alternative and commonly used standardisation approach where direct standardisation may not be possible