Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
246 Cards in this Set
- Front
- Back
- 3rd side (hint)
Population at risk |
The population susceptible to a disease, defined by either environment or by demographics |
|
|
Prevalance |
Number of cases (old and New)/ Population at risk |
|
|
Point Prevalance |
P = C/N where C is number of observed cases and N is the population sixe at risk at that time Snapshot at that time |
|
|
Period prevalance |
PP = C+I/N where C is old cases in the population I is the new cases within a period N is the size of population at risk |
|
|
Factors contributing to INCREASED PREVALANCE |
* Long duration of illness with a. Low cure rate and b. Low case fatality * Immigration of diseased population * Emigration of healthy population * Increased new cases * Increased detection |
|
|
Factors contributing to DECREASED PREVALANCE |
* Short duration of illness due to a. High cure rate, b. High Case Fatality * Decreased new cases * Improved cure rate * Immigration of healthy population * Emigration of diseased population |
|
|
Uses of prevalance data |
1. Assessing health care needs 2. Planning health services 3. Measuring occurance of conditions with gradual onset 4. Study chronic disease |
|
|
Incidence |
The number of new cases in a a given period in the specified population Time period must be specified Measures rapidity with which new cases are occuring in the population Can be expressed as A) absolute numbers B) Cumulated Incidence C) Incidence Density |
|
|
Cumulated incidence |
CI = No of new cases/ Population @ risk at the beginning *10 to the power of n
Attack Rate
Assumes that the entire population at risk was followed up for the time period of observation |
|
|
Risk (incidence risk) |
Probablity that an individual will experience a health status change over a specified follow-up period Assuming that a. The individual does not have the disease under study at the beginning of the period b. Did not die for other causes during that period Corresponds to Cumulated Incidence |
|
|
Incidence Desnsity @ Incidence Rate |
ID = Number of new cases/ Total 'person-time' of observation * 10 to the power of n * More accurate * Describes trends * Evaluates the impact of primary prevention programmes |
|
|
Prevalance Relationship |
P = Incidence * Duration Changes in prevalance from time to time is due to change in incidence or duration of measurement or both. |
|
|
Case Fatality |
CF = No of deaths due to the disease/ No. of persons with that disease Reflects severity of the disease It is not a Rate, but a proportion or Ratio Differentiate from Mortality = No of deaths/ population @ risk |
|
|
Types of descriptive studies |
1. Case reports 2. Case series 3. Ecological Studies 4. Cross-sectional studies |
|
|
Case Reports |
Single case New presentation Unfamiliar pattern or disease Rare manifestations Generate hypothesis regarding pathophysiological mechanisms |
|
|
Case Series |
Study a larger than one group eg. >10 * to assess play of chance * to delineate the clinical picture No comparision group is a drawback |
|
|
Ecological study |
* A type of descriptive study *Group as the unit of analysis * No individual level information is taken * Relate exposure to disease across populations |
|
|
Cross-sectional surveys |
Type of Descriptive study * Observation of a cross section of a population at a single point in time * Also known as prevalence studies *Recruitment of study participants from population or population sample * Observe for the presence of one or more Exposures or one or more outcomes |
|
|
Analytical Study |
* It is a type of epidemiological study in which there is a comparision group * But the investigator has NOT assigned the exposure * Investigator merely measures the exposure and the outcome/disease in both groups to arrive at conclusions |
|
|
Descriptive Study |
* A type of epidemiological study
* Studies a health event in terms of Time, Place and Person
* No comparison groups or assignment of exposure |
|
|
Cohort |
Cohort: a group of individuals sharing same common attributes. Eg. Birth cohort, the persons sharing the same birth D/M/Y |
|
|
Cohort study |
* Exposure to Outcome * Select exposed and unexposed cohorts * Follow up to see outcome in both * Measure incidence of disease in both * Compare incidences using Relative Risk (a measure of association) |
|
|
Types of Cohort Studies |
1. Prospective Study>Exposure>Outcome 2. Retrospective Exposure>Outcome>Study 3. Ambispective (Bi-directional) Exposure>Study>Outcome |
|
|
Steps of Cohort Studies |
1. Selection of study population 2. Gathering of baseline info 3. Follow-up 4. Analysis |
|
|
Cohort: Selection of study pop |
1. General pop or a subset of it 2. Special exposure cohorts (eg. Occupational groups) |
|
|
Cohort: Baseline info |
Sources: Records, Interviews, Examinations, Measurement of environment Objectives: * Valid assessment of exposure status of members * Identification data * Exclude those with outcome disease at baseline * Define individuals @ risk for outcome disease * Obtain data on co-variables (other significant exposures) |
|
|
Cohort: Choice of Comparision Group |
1. Internal comparision group, unexposed persons from the same locality, occupation, factory
2. External comparison group, when internal comparison group is unavailable, use the national data |
|
|
Cohort: Follow-up |
1. Uniform and complete fu of ALL cohorts 2. Complete ascertainment of Exp & Outcomes 3. Using standardised diagnosis of outcome measures |
|
|
Relative risk |
In the analysis of a cohort study. It is the ratio between 1. The exposed with disease / total exposed 2. The unexposed with disease / total unexposed RR= 1 Exp not associated with outcome RR< 1 Exp is positively associated with outcome RR > 1 Exp is negatively associated with outcome |
|
|
Cohort: Strengths |
1. Incidence can be calculated due to FU 2. Examines multiple outcomes for one exposure 3. Clarity in temporal sequence of events 4. Good for investigating rare exposures |
|
|
Cohort: Weakness |
Large sample Long time Not good for disease with long latency Differential loss to FU between groups can bring about bias |
|
|
Case control studies |
* A type of analytical study * Opposite in direction to case control studies * from Outcome to Exposure |
|
|
Case Control: Selection of cases |
All or a sample of source population with outcome of interest * Clear definition of outcome * Prevalent cases may be related to variables associated with survival * Incidence cases may be related to variables associated with development of disease |
|
|
Case Control: Sources of cases |
1. Hospital/Clinic: Easier, but may represent severe cases 2. Population based: Not biased by factors drawing patients to a particular hospital, like availability of a certain treatment or location |
|
|
Case Control: Selection of Controls |
* Select from the same source population as that of cases * Select independent of exposure status * Can be Pop based, Health care facility based or from friends and neighbourhood |
|
|
Odds Ratio |
Result of Case Control analysis 1.Odds that the case was exposed / 2.Odds that the control was exposed. 1 = a. Probability that the case was exposed/ b. Probability that the case was not exposed 2 = a. Probability that the control was exposed/b. Probability that the control was not exposed |
|
|
Odds Ratio: Interpretation |
OR = 1 Exposure not associated with Outcome OR > 1 Exposure is positively associated with outcome OR < 1 Exposure is negatively associated with outcome |
|
|
Case Control: Strengths |
1. Good for rare outcomes 2. Good for long latency diseases 3. Quicker 4. Less expensive 5. Fewer subjects 6. Multiple exposures can be examined |
|
|
Case Control: Weakenesses |
1. Susceptible to recall bias about exposure 2. Selection of an appropriate comparison group is difficult 3. Rates of disease in exposed and unexposed individuals cannot be determined |
|
|
Steps of Case Control studies |
1. Select Cases: those with disease 2. Select Control: those without disease with similar attributes 3. Measure exposure in both groups 4. Calculate 'Exposure odds' for both 5. Find the 'Odds Ratio' by looking at the ratio between both Exposure odds |
|
|
Clinical Trials |
* Experimental arm of epidemiological studies * Investigator ASSIGNS and MANIPULATES the EXPOSURE * Brings findings of basic science research to better prevent, diagnose and treat diseases * Involves people * it is a planned experiment * one or more comparison groups * Prospective study |
|
|
Objectives of Clinical Trials |
New Drug/Treatment/technology/ delivery system/organisation of health care/primary prevention methods/screening programmes/early detection strategies |
|
|
Randomisation |
1. Used in clinical trials to reduce selection bias 2. All subjects stand equal chance to be in any of the groups 3. All groups have similar participants 4. Confounding, Co-inerventions and Bias are minimised |
|
|
Blinding |
* To balance groups during follow up * Levels: Single = Participant Double = Participant and Investigator Triple = Participant, Investigator and Analyst * Reduced Co-interventions: Effects of other therapies, differential treatment by team * Reduced Bias: Participants reporting symptoms differently or Team interpreting reported symptoms differently |
|
|
Phases of Clinical Trials |
Phase 1, Safety and acceptability, up to 50 healthy volunteers Phase 2, Long term safety, dose and schedule, early indications of efficacy, 100 to 500, low risk Phase 3, Effectiveness, 1000 or more, high risk, leads to licensing Phase 4, post - marketing surveillance, Community based |
|
|
Measures of Epidemiological studies |
1. Internal Validity 2. External Validity 4. Accuracy = Validity + Precision
Precise is all results being in the same range or side Valid is all results around the central target, but not in the same side |
|
|
Errors in estimation |
1. Random Errors: unknown or uncontrollable errors of two types a. Sampling error b. Measurement error Minimised by larger sample size and precise measurements
2. Systematic Errors: Major threat, "a process that tends to produce results that depart systematically from true values" |
|
|
Bias |
Threats to validity in epid studies 1. Selection bias 2. Information bias 3. Confuonding |
|
|
Selection bias |
Representativeness of the study population to the target population |
|
|
Types of selection bias |
1. Surveillance mechanisms used to notify of exposure or outcome 2. Screening and diagnosis, prior knowledge of exposure increases risk of higher diagnosis 3. Admission to health care facilities of cases and controls can be biased 4. Selective survival, inclusion of cases that survived who may be less (or more) exposed 5. Non response/Loss to FU of cases and controls who are <or> exposed or at risk |
|
|
Dealing with selection bias at design stage |
Use incident cases Use pop based design Same criteria for Cases and Controls Same procedures, tests and intensity of measurement |
|
|
Dealing with selection bias at Data collection stage |
Minimise non response and loss to fu Keep records of all losses adn have baseline data on them Make sure diagnosis is not affected by exposure status, use blinding. |
|
|
Dealing with selection bias at analysis stage |
1. Compare non-responders with responders on baseline variables where large differences mean selection bias (small difference does not rule out bias)
2. Sensitivity analysis to deduce the direction and magnitude of biases, using the study results and external info. |
|
|
Information Bias |
Do measurements accurately represent the phenomena of interest? Bias results from measurement procedures of variables. |
|
|
Information bias in Cohort studies |
Cohort: collection of info leaning towards a specific outcome due to collection of better outcome data in the exposed than in the non exposed
|
|
|
Information bias in Case Control studies |
Case Control: a. Info leans to specific exposure, b. Recall bias in subjects, better data on exposure in cases than controls |
|
|
Information bias in investigator and subjects |
Investigator: Systematic collection of data supporting the expected results (conscious or unconscious) Prevarication: Systematic distortion of the truth by subjects. |
|
|
Dealing with information bias |
1. Precise operational definitions 2. Detailed measurement of protocols 3. Repeated measurement of key variables 4. Training: certification and re-certification 5. Data audit (of interviewers and data centres) 6. Data cleaning (visually on computer) 7. Re-run of all analysis before publication |
|
|
Effects of Confounding |
* Confusion of effects * Effects of extraneous factors is mistaken for the effect of actual exposure * May simulate an non existent effect * May hide an existing effect * May change the direction of the effect, by affecting both, the exposure and outcome * May increase or decrease the strength of association |
|
|
Dealing with confounding |
Design stage: Restriction Matching by match analysis Randomisation Analysis stage: Stratification: Check if any variable is a confounder Multivariate analysis : Regression analysis |
|
|
Evaluation of association |
Crude association: 1. Is it due to chance, if no 2. Is it due to selection bias, if no 3. Is it due to information bias, if no 4. Is it due to confounding, if no Only then is it a causal association |
|
|
Internal Validity |
To obtain an accurate estimate of disease frequency and effect of exposure on health outcomes in study population If the findings in the study brings out the truth in the study it is called internal valididty. It is to see how independent variables influence the dependent variables. |
|
|
External Validity |
To obtain an estimate that is generalisable to relevant target population. If the truth in the study is generalisable to the trugh in the universe it is called External Validity. There is always a trade off between Internal validity and External validity |
|
|
Limitation |
Bias that cannot be avoided |
|
|
Qualitative research |
Studies social reality from the subjectively interpreted and experienced EMIC perspective Inductive (as opposed to deductive hypothesis testing in quantitative) reasoning. Validity is based on subjective credibility Interpretation of responses is the analysis Requires conversion in abstract inter-cultural catergories |
|
|
When to do Qualitative research? |
To understand the circumstances in which events occur To seek depth of understanding Provide insights into meanings of decisions and actions Need to explore and explain behaviour Unfamiliar subject matter, insufficient research. When suitable vocabulary to communicate with responders is not available To have a holistic view of social phenomenon |
|
|
Methods of data collection in Qualitative research |
In-depth individual interviews Focus group discussions Participant observations |
|
|
Features of Qualitative data |
1. Interpretative and open ended 2. Iterative, than fixed 3. Emergent than pre-structured 4. Partnership between participant and investigator 5. Investigator is the instrument in the research process, 6. learner is the co-interpreter |
|
|
In-depth interviews |
1. Open ended 2. Individual 3. Discover individuals framework of meanings 4. Obtains rich contextual info 5. Avoid investigators assumptions and structures of understanding |
|
|
When to do in-depth interviews |
1. Complex subject 2. Knowledgeable respondent 3. Highly sensitive subject matter 4. Geographically dispersed respondents 5. Peer pressure is an issue 6. Social desirability is a threat |
|
|
Technique of in-depth interview |
1. Follows an interview guide 2. Probes 3. Reflecting on remarks made by informant 4. Collects respondent's perspective and words 5. Level of structure varies. |
|
|
In-depth interview: Advantages |
1. Most in-depth, to understand why of a behaviour 2. Data on how people think and talk: Conceptualisations of behaviour 3. Exact words and language people use amongst themselves. 4. EMIC perspective = insider's persepective |
|
|
In-depth interview: Disadvantages |
1. Based on a few people, not systematic but purposeful or convenience sample 2. Not generalisable 3. Very long, lost of data, takes time to analyse 4. Researchers opinions of what the data means |
|
|
Focus group discussions |
1. 6-8 similar participants 2. Moderator and note taker are extra 3. Flexible interview guide 4. Used when interaction is important 5. Cost and time are issues 6. Idea generation 7. Problem identification 8. Identify local vocabulary/terminology 9. To evaluate messages for an intervention |
|
|
Focus group: Advantages |
Some are more comfortable in groups Natural way to talk about problems Collects info on social normal Can provide lots of data in a limited time |
|
|
Focus group: Disadvantages |
Difficult to access practice of personal or sensitive behaviour Not generalisable because of dominant personalities Transcribing is time consuming Analytic challenge |
|
|
Participant observation |
Qualitative data collection technique Researcher participates in a social event or group to make observations Data is deep and detailed Difficult to systematically collect, hard to take notes, details may be forgotten No defined analytic methods |
|
|
Grounded theory |
Qualitative data analysis method 1. Transcripts to Themes to Text categories 2. Find relations among categories, 3. Build theoretical models 4. Quotes from interviews used as exemplars |
|
|
Content analysis |
Qualitative data analysis method 1. Theoretical framework 2. Set of codes for variables in the theory 3. Apply codes systematically to se of tets 4. Unit-of-analysis-by-variable matrix from the texts and codes 5. Statistical analysis of matrix |
|
|
3 ways to use qualitative methods |
1. A tool to generate ideas for subsequent quantitative study. Qual>Quant>Result 2. To help understand the results of a quantitative study. Quant & Qual> Result 3. The primary data collection method. Sometimes along with quantitative methods. Qual>Results<Quant |
|
|
Analysis in Qualitative methods |
Triangulation, since no one method is adequate to explain all data.
1. Analyst/Theory 2. Methods 3. Data sources |
|
|
How are qualitative methods useful? |
1. Identification of health determinants, underlying attitudes, percepts and behaviour 2. Facilitate understanding of policy, social & legal contexts in which decisions are made 3. Shed light on the success of an intervention 4. Explains social and programmatic impediments to informed choices and use of services |
|
|
Types of data |
1. Qualitative A. Nominal - eg. Names, address B. Ordinal - eg. Stages of a disease condition 2. Quantitative A. Discrete - eg. Family size B. Continuous - eg. Height, Weight |
|
|
Central values of Data |
Mean Median Mode |
|
|
Arithmetic Mean (AM) |
Add all observed values and divide by number of values SUM is denoted by Sigma Xi (the jagged E space Xi) Sample Mean is denoted by X bar (X with a horizontal bar on top) Population Mean is denoted by Mu (the symbol used for micro as in mcg) |
|
|
Median |
The middle value of the distribution, 50% of data will fall either side of median. Useful when there are extreme values In a sequence of 11 values, the sixth value is the median |
|
|
Mode |
Most frequent value The only statistics that can be used on nominal data (many having the same name, colour of car etc) Used to describe an epidemic with respect to time |
|
|
Dispersion and types |
Measure of variability (swimming pool average depth versus extremes) Types: 1. Range 2. Inter quartile range 3. Mean deviation from mean 4. Variance/Standard deviation |
|
|
Range |
The difference between the minimum and maximum values of observation Adv: Quick and easy indicator of distribution Disadv: Influenced by estreme values, since it uses only two data points |
|
|
Inter Quartile Range |
The inteval beweeen the value of the upper quartile (Q3) and the lower quartile (Q1). IQR = Q3 - Q1 Adv: Unaffected by extreme values Disadv: Covers only the middle 50% of observations |
|
|
Mean Deviation |
The average of the absolute (ignoring the sign) deviations of the observations from the arithmetic mean. Adv: Based on all observations Easy to grasp the meaning fo the procedure Disadv: Ignores the sign of the difference of the value of the observation and the arithmetic mean Not widely used because of the availability of a more advantageous measure |
|
|
Standard Deviation |
The square root of the average of the squared deviation of the observations from the arithmetic mean
The square of the SD is called variance
Adv:The SD is the most improtant measure of distribution. SD is in the same unit of measurement as the observation, so it is suitable for further analysis
SD together with arithmetic mean is useful for description of the data |
|
|
Co-efficient of variation (CV) |
To compare the relative variablility in different groups Defn: the coefficient of variation is the SD expressed as a percentage of the AM CV = (SD/AM) * 100 |
|
|
Choice of central/dispersion values |
1. Mean / SD if there are no extreme values 2. Median / IQR if there are extreme values 3. Mode / Range for qualitattive variables/ time distrubution epidemic curve Mean and SD are used the most |
|
|
Define Sampling |
Procedure by which some members fo teh population are selected as representatives of the entire population |
|
|
Define Study Population |
The population to which the study results will be inferred |
|
|
Representativeness |
Time - Seasonality, Day of the week, Time of the day Place - Urban, Rural Person - Age, Sex, Other demographics |
|
|
Why do we sample populations? |
1. To obtain information from a large population 2. To ensure the efficency of a study 3. To obtain more accurate information |
|
|
Sampling terms |
1. Basic Sampling Unit (BSU) - Elementary unit that will be sampled. Eg. People, hospitals, health care workers 2. Sampling frame - List of all sampling units in the population 3. Sampling scheme - Method used to select sampling units from the sampling frame |
|
|
Types of samples |
1. Non-probability sample 2. Probability sample |
|
|
Non-probability sample |
Probability of being selected is unknown Convenience samples - Biased, results show best or worst scenario Subjective samples - Based on knowledge, Time/Resource constraints |
|
|
Probability samples |
Every unit in the population has a known probability of being selectd
Only sampling method that allows to draw valid conclusions about the population |
|
|
Random Sampling |
A type of probability sampling Removes selection bias Allows application of statistical theory |
|
|
Sampling error |
* No sample is a perfect mirror image of the study population * Magnitude of error can be measured in probability samples expressed by standard error of mean, proportion, differences * It is a function of Sample size and variability in measurement. |
|
|
Methods of probability sampling |
1. Simple random sampling 2. Systematic Sampling 3. Stratified sampling 4. Cluster sampling 5. Multistage sampling |
|
|
Simple random sampling |
Principle: Equal chance for all sample units
Procedure: Number each unit and draw randomly
Adv: Simple. Sampling error, if any, is easily measrued
Disadv: Needs complete list of units, does not always achieve best representation |
|
|
Systematic Sampling |
Principle: A unit is drawn every k units, equal chance for each unit Prodecure: Calculate sampling interval (k=Total units/Sample size) Draw random number (< or = to k) for first sample Draw every k units from there on Adv: Ensures representativity, easy to implement Disadv: Dangerous if list has cycles |
|
|
Stratified Sampling |
Principle & Procedure: Classify pop into homogenous groups (Strata), Draw samples from each strata, combine results of all strata Adv: More precise if variable associated with strata All subgroups represented, allowing separate conclusions about each Disadv: Sampling error is difficult to measure Loss of precision if small nubmers sampled in individual strata |
|
|
Cluster sampling |
Principle: Random sample of groups (clusters) of units (first randomly select groups, wards in a town) Secondly, all or proportion of units included from selected clusters
The sampling unit is not a subject but the cluster of subjects. It is assumed that the variability between clusters is minimal and the variability within each cluster is as observed in general population
Adv: Simple, does not need list of all units, Less travel/resources are needed
Disadv: Imprecise if clusters have homogenous units (all upper class in the ward) Sampling error is difficult to measure
|
|
|
Stages of a cluster sample |
Stage 1: * Select the clusters to be included * Compute a cumulative list of the populations ineach unit with a grand total * (I'm stumpted) Divide grand total by number of selected clusters to obtain sampling interval * Chose a random number to identify the first cluster * Use the sampling interval to find the subsequent clusters Stage 2: * In each cluster select a random sample using a sampling frame of subjects or households |
|
|
Multistage sampling |
Principle: Several chained samples, Several statistical units Adv: No complete listing of the population is required, Most feasible approach for large populations Disadv: Several sampling lists, Sampling error is difficult to measure |
|
|
Key sampling issues |
* We sample since we can't study the whole population * Sampling leads to sampling error, but that is measurable * Appropriate sample size ensures precision of the study, like a good design and quality assurance ensuring the validity * Probablilty sampling is the only sampling that lets us the use of statistics as we know them. |
|
|
Steps in estimating Sample Size |
1. Identify the one major study variable 2. Determine the type of estimate of that variable (% or Mean or Ratio etc) 3. Indicate expected frequency of factor of interest (from literature review?) 4. Decide on the desired level of precision of the estimate 5. Decide on acceptable risk (by which the estimate can fall outside its real population value) 6. Adjust for population size 7. Adjust for estimated design effect 8. Adjust for expected response rate |
|
|
Alpha and Confidence Interval |
Alpha is the significance level of a test. It is the probability of rejecting the null hypthesis when it is true (type 1 error) Finding association when there is none
Confidence interval: The probability that an estimate of a population parameter is within certain specified limits of the true value; commonly denoted as 1-Alpha |
|
|
Beta and Power |
Beta: The probability of failing to reject the null hypothesis when it is false (type 2 error) Finding no associaton when there IS association
Power: The probability of correctly rejecting the null hypothesis when it is false; commonly denoted as 1- Beta |
|
|
Precision |
A measure of how close an estimate is to the true vlue of population parameter. It may be expressed in absolute terms or relative to the estimate |
|
|
Info needed to calculate sample size |
1. The desired width of the confidence interval (+ or - 5 units of protein) 2. The level of confidence desired (normally 0.95) 3. The magnitude of population variance (20 grams of protein) Look up example in slides |
|
|
Design effect |
The design effect is 1 (no design effect) in simple random sampling For cluster sampling the design effect is taken as 2. |
|
|
Values needed to find sample size for analytical studies |
* Desired values for the probabilities of Alpha and Beta * The proportion of the baseline (Controls): Exposed for CC studies or Diseased for Cohort studies, based on previous studies * Magnitued of the expected effect (RR or OR): Minimum effect that the investigator considers worthy of finding, based on previous studies * Different formulae depending on design, RQ and type of data |
|
|
The 10% rule of sample size |
* the sample size calculations give only the Minimum needed size * Confounders are not considered, only crude exposure and outcome association is considered * Therefore increase the sample size by 10% for each confounder/ or variable added |
|
|
Selection of study population |
Based on: 1. Representativeness 2. Adequate size 3. Acceptable cost and time |
|
|
Process of selecting Study Population |
Step 1: General population refined by Clinical and Demographic criteria gives target population Step 2: Target population is further refined by geography and temporal (within a specified time) charecteristics to give us the Accessible population Step 3: A subset of the accessible population is the study sample |
|
|
Inclusion and Exclusion criteria |
Inclusion criteria narrow our choise from the general population to accessible population. (Eg. Demographic, Clinical, Geographical and Temporal criteria) Exclusion criteria narrow the choise from the accessible population to give the study sample subset (Eg. Subjects not suited for regular follow up, who have poor quality of data, high risk of adverse effects etc.) |
|
|
Internal validity and External validity in selection of study sample |
The inclusion criteria takes care of the representativeness of the population therefore taking care of the generalisability of the study results which is the External validity The exclusion criteria take care that the study subjects are the right match to bring about the robustness of the study results, therefore contributing to Internal validity. |
|
|
Clinical vs. Community population |
Clinical population if the study sample is patients: Relatively easy and cost effective Community population if the study involves general public: Relatively difficult and expensive |
|
|
Recruitment goals for study sample |
Feasibility must be considered in chosing the accessible population and sampling methods Goals: a. Subjects should adequately represent the target population and b. should be of sufficient size to meet sample size requriements |
|
|
Achieving a representative sample |
Design Phase: chose samping methods well Implementation Phase: a. Avoid errors in applying Inclusion and Exclusion criteria on the target population b. Monitor adherence to the criteria as the study progresses |
|
|
Non responses in a study |
More in observational studies Influences the inernal and external validities of the study Repeat contact attempts Design that avoids discomfort to participants: Incentives, local language questions etc. |
|
|
Steps in selection of study sample |
Step 1: Define Target population, by application of inclusion criteria of demographic and clinical description Step 2: Define Accessible population, by application of inclusion criteria of geographic area and temporality Step 3: Define Subset of population, by application of a parsimonious set of exclusion criteria, by eliminating subjects who are inappropirate or unethical Step 4: Define Sample by using sampling techniques that estimate Sample Size Step 5: Recruitment strategy should be to find a sample large enought to meet the needs of the study and to minimise bias due to non-response or being lost to followup |
|
|
Project management principles |
1. To ensure the defined objectives are met 2. To ensure deliverables are delivered within timeframe and budget at the expected quality standards 3. The end result should give directions for future implications (a better tommorrow) |
|
|
Principles of project management |
1. Resource allocation and management 2. Time management 3. Efficency in process 4. Planning and scheduling activities 5. Monitoring and supervison 6. Reaching the goal with best possible quality standards 7. Communication 8. Data management 9. Finance management 10. Team work and co-ordination |
|
|
Research life cycle under project managment |
Starts from Formulating study objectives > Planning the anaysis > Preparing Data collection instruments > Collecting data > Analysing data > Drawing conclusions. It also includes choosing the design of the study and estimating sample size |
|
|
Indicators for project management |
What will the study generate? Rates, Ratios, Proportions or quantitative variables Identify the information needed to calculate the Indicators, such as outcome variables, Covariates, risk factors, confounders |
|
|
Advantages of an analyisis plan |
Helps to focus on study objectives Dummy tables help to avoid comparisons for which the study was not designed Makes sure only needed data is collected Saves time for publication therefore saves time for disseminisation and policy change |
|
|
Common reasons for study failure |
Badly defined RQ, RH and RO Unrealistic timelines Inappropriate or incompetent staff lacking direction, motivation and training Inadequate monitoring and failure to respond to contingent situations and to carry out mid-course corrections |
|
|
Information collected with data collection tools |
Facts, Knowledge and Jugements Facts: Individual charecterestics, Environmental charecteristics, Behaviour Knowledge: Risk factors, Healty lifestyles Judgements: Opinions and Attitudes |
|
|
Different data collection tools |
1. Abstraction forms - to copy records from other records (medical, personal etc) 2. Structured observation guide - Checklist of items 3. Questionnaire - Self or Interviewer administered (in person, on phone, in computer) |
|
|
Key elements of data collection tools |
1. Clarity of words 2. Balance of phrases 3. Length of sentences 4. Comprehensiveness of responses 5. Constraints of responses 6. Utility of instructions 7. Order of questions 8. Context of questions |
|
|
Four components of a data collection tool |
1. Intro and conclusion: About the presentation, objectives, informed consent and concluding statmetns 2. Instructions for data collectors: Prompts, Skip patterns. Use different font for instructions 3. Identifiers: Exactly identifying the respondent (to be kept seperately), Coded ID numbers 4. Body of the instrument: Open, closed and semi-open items |
|
|
Open questions |
Respondent must generate the answer Advantages: Freedom, stimulates memory, used to generate closed responses later, Useful at hypothesis raising stage Disadvantages: Difficult to code and analyse, May be unfocused or incomplete |
|
|
Open questions with closed answers |
Expressed as an open queston and analysed as a closed-end question. Like an MCQ, but each response has to answered as Yes/No Please refine this description |
|
|
Closed questions |
3 types: Dichotomous options, Multiple options and Quantitative answers |
|
|
Dichotomous Closed questions |
Yes/No, Male/Female Advantages: Forces a clear answer, useful for well framed issues Disadvantages: May oversimplify the issue |
|
|
Mulitple options closed questions |
MCQ, with one or more responses Advantages: More than one choice of answer Disadvantages: Difficlut to choose only one |
|
|
Quantitative Closed questions |
Somehow the answer is quantified as a number, Pain 0 to 10, How many km of walking? Advantage: Creates continuous varialbes Dsiadvantages: May require validation, some questions are difficult to be handled as a continuous variable. |
|
|
Semi-open questions |
Suggested answers with an "other" option Advantage: Leaves door open for an unplanned options Disadvantage: Difficult to analyse |
|
|
Formulating a questionnaire |
1. Short and precise questions 2. Simple everyday language 3. Avoid negatives or double negatives 4. Only one question at a time 5. Be specific 6. Use neutral tone |
|
|
Questionnaire Sorting order |
Simple to complicted, General to specific Casual to intimate Group question on the same topic ID questions at the beginning or at the end In chronological order if a sequence of events is studied If complex questions are abound, introduce simple questions in between Triangulate through multiple questions if the subject is important |
|
|
Data collection tool layout |
Split sections Dont split questions across pages Space out Large fonts Number all Q Vertical format for closed ended questions Standardise coding procedures Use auto-correct procedures (data validation) |
|
|
Finalising a data collection tool |
1. Check and suppress unnecessary questions, Add missing questions 2. Review the instrument: Colleagues, experts, statisitician (for coding) field workers and data entry operators 3. Language: The language of administration to respondent, Transaltion in 3 stages: a. Initial formulation , b. translation and C. Back translation |
|
|
Testing a data tool |
1. Check for clarity, understanding and acceptability 2. Check flow and skip patterns 3. Check pertinence of coding 4. Estimage the time needed to ask all the questions 5. Pilot test with a few volunteers: Similar to target pop, not present in the study |
|
|
Data quality |
A. Reliability (precision): Reproducibility, Repeatability, Stability B. Accuracy (validity): Correctness of a measure |
|
|
Steps of Data collection |
1. Draft a question by question guide, keep improving on the fly 2. Train staff who will collect: Slideshow, Discuss, Clarify, Simulate 3. Initiate collection and ensure quality: Pilot, Supervisor to daily checks, onsite availability, no pressure 4. Review collected data for quality and completeness, 5. Debrief staff to trouble shoot difficulties 6. Validate |
|
|
Review of data |
Team checks before respondent leaves, Supervisor checks before leaving location, Each data collector takes responsibility by signing each filled form, Supervisor counter signs PI Checks as they reach him |
|
|
Checking data |
Completeness Readability Consistency: Do the anwers make sense, do they have internal consistency |
|
|
Validation of data |
Select study participants at random Conduct a second interview Compare results Debrief discrepancies: with Individuals or the whole team as appropriate |
|
|
Data management process |
Define variables Create database and dictionary Enter data and correct errors Create dataset for analysis Backup and archive the dataset
|
|
|
Data management elements |
Data structure Data entry Individual and aggregated databases Mother and daughter databases |
|
|
Database documentation |
Contains: 1. Structure: Name of database, Number of records, etc. 2. Variables: Name, Values, Codes 3. History: Creation and modification records 4. Storage infromation: Media, Location, Backup schedule 5. Additional information |
|
|
Unique identifier |
Numeric Secured by quality assurance practices Has info about subject Each digit represents some ID feature (1&2 village, 3 Street, 4 no in family, 5 person etc.) |
|
|
Entering values |
Interger - specifiy number of digits Numeric - Specify number of decimal places Alpha nummeric - All caps, specify length Dates - specify format |
|
|
Variable names |
Clear: Refer to questionnaire, Understandable acronyms (EXERDLY) Short: No space, 10 charecter limit Consistent: Similar acronym format for related itesm (EXERDLY, EXERMOR, EXEREVE) No duplicates |
|
|
Coding |
Design data collection tools with codes builtin Prefer numerical codes Decide on codes for missing values (9, 99 or 999), Not applicable values (8, 88 or 888) Avoid cumbersome codes Use 1 or 0 as baseline for gradients |
|
|
Data Dictionary |
Explains Variable : Name, Acronym, questionnaire item, values, meaning of each value Useful: when sharing database with others, when revisiting the database after a long time |
|
|
Check before data entry |
Ascertain skip patternsAuto codingCalculations Legal limits of entry Copying data from preceding record |
|
|
Data entry |
Check, comment, clarify Clean data Mark each form when data entry is completed Validate after entry |
|
|
Aggregated databases |
Sorts by certain variable and aggregates by that field count In normal database each record is an observation IN aggregated observation are pooled togther by some variable |
COUNT |
|
Mother and daughter databases |
When info is available at various levels Store info from each level in separate databases Link records accross databases using unique identifier codes. |
|
|
Objectives of data analysis |
Plan the analysis Programme the crude analysis Deal with chance, biases and thrird factors Assess causality Measure clinical/public health impact |
|
|
Sequence of data analysis |
1. Identify the study type 2. Identify the main variables 3. Become familiar with the data 4. Charecterise the study pop 5. Examine exposure/ outcome association 6. Create additional two way tables for other variable or secondary objectives 7. Conduct advance analysis |
|
|
DA: Identify study type |
1. Look at study design and 2. all study documents 3. Review data collection processes and 4. Review data anaysis plan 4. Look at the database 5. Decide on software for analysis |
|
|
Study type and Analysis plan |
1. Descriptive - Acute - Cohort/Surveillance - Incidence 2. Descriptive - Chronic - CS Survey - Prevalence 3. Analytical - Acute - Frequnt - Cohort - Relative risk ratio 4. Analytical - Acute - Rare - CC study - Odds ratio 5. Analytical - Chronic - Frequent - CS Study - Prevalence ratio 6. Analytical - Chronic - Rare - Prevalent CC Study - Prevalence Odds Ratio |
|
|
DA 2. Identify Main variables |
Exposure and outcome Potential biases, confounders Variables for subgroup analysis |
|
|
DA 3. Familiarise with data |
Look up the frequency distribution for each variable Look at all the descriptive data of stuy population Look at the database for review of number of observations to find duplicates or missing values Check if all data fits into approved ranges Check consistency of data. |
|
|
DA 4 Charecterise the study population |
Look at base line charecters of all demographic variable for all groups Look at the frequency distribution of clinical features and health problems |
|
|
DA 5 Examine primary association |
The exposure outcome association, Based on the hypothesis Based on prior knowledge Based on study design |
|
|
DA 6 Additional two way tables |
For the other associations for secondary objectives using other variables |
|
|
DA 7 Conduct advanced analysis |
Look for dose response Stratify the sample and analyse Use Multivariate analysis |
|
|
Tips for Data Analysis |
Be systematic, Don't skip steps Be prepared with empty shells First do the recoding: new groups or starta can be created at this stage for more in depth analysis (eg. ages 20-30, Income:upper etc.) Look at the descriptive data and find associations with main vairables (education-exposure-outcome) Do the analytical analysis last Avoid: Post-hoc analysis trying to find new, unplaned associations with available data. Avoid: Data drenching - Squeezing the data for more findings |
|
|
DA Analytical stage |
Univariate analysis: Frequency of outcome by single descriptive variables Frequency of outcome by other variables
Startified analysis: Frequency of outcome by income, stratified for age and gender. Multivariate analysis: Logistic regression model |
|
|
Ethics where? |
Any study with human participants Risk or not Even for observational studies |
|
|
Evolution of ethical codes |
1947 Nuremberg Code - Risk benefit analysis, competence of researchers and voluntary consent 1964 - Helsinki Declaration - Revised '83, '89, '96, 2000, '08, '13 - Individual rights, informed decisions, investigator's duties, welfare and vulnerability of participants 1978-79 - Belmont report - Ethical principles of Autonomy, Justice, Beneficence. Informed Consent and Ethics committee 1992-93 CIOMS Guidelines - revised 2002 - Adverse drug reaction reporting and saftety of participants, Risk-benefit Balance, Pharmacovigilence 1996 - Int Council on Harmonisation - Good Clinical Practice |
|
|
ICMR Ethical guideines for human participants |
2000 and 2006 - All institutions doing bio-med research to follow in leter and spirit
Other guidelines - Genome policy an genetic research 2000, Indian Good Clinical Practice (GCP) 2001, Amendments to Drugs and cosmetics act 2002, assisted reproductive tech 2005, Stem cell research and bio-banking 2006 |
|
|
Core ethical principles |
Autonomy: Obligation of researcher to respect the decisons made by people concerning their health. Respecting human dignity, not interfering with it. Justice: Obligation to provide all with what they deserve. Treate all equally farily and impartially. Must not impose the research on anyone Beneficence: Positive steps to prevent harm, not only being correct and fair. Sometimes steps to prevent harm to others may put us in a conflicting situation set against their autonomy and justice Non-maleficence - Obligaton to 'first, do no harm'. When harm can't be avoided, minimise it. Wrong to waste resources that could be used for good. |
|
|
Informed consent |
Process of informing the potential particpants about the proposed reasearch in a systematic manner and empower them to take an informed decision to participate in the research study. So that they understand the procedures, risks and benefits, get all questions and concerns answered and take a learned and informed decison to or not to particpate
Repeat several times during the study if necessary Group consent is taken in some situations (tribals) but does not replace a individual consent.
|
|
|
Informed consent document |
It is an appeal or an invitation to participate Written in simple local language With the name of the insitute on top Ends with the signature of the particpant and an independent witness Consists of the following: a. Research description b. Risks c. Benefits d. Alternatives e. Confidentiality f. Compensation g. Contacts h. Voluntary participation and withdrawal |
|
|
Stakeholders in Informed Consent process |
Researcher and institution: Provide infomration, discussion and explanation, ensure comprehension and voluntary decision Participant: Needs to get informed, and make a free and independent conset without yeilding to coercion or force Sponsors, monitors and regulators: Assess the fairness of the consenting procedure and verify the consent documents. |
|
|
Other issues related to IC |
Language: Simple, local language translated (after back translation) Impartial witness: not a part of the study Test understanding of the IC and document it. Apart from written pictorial consent is possible Audio and video consent are mandatory now, in India, for Investigational New Drug (IND) trials |
|
|
Scientific Review of study |
Explores the scientific novelty, rationality and relevence
1. Justification: in context of national priorities 2. Safety, scientific mertis and feasibility: Review of toxicology/ animal studies and and lab data 3. Technology transfer and capacity building at sites.
Looks for soundness of study design: Incl and Excl criteria, Randomisation/ blinding, procedures and followup protocols, sample size calc, End-point assessments and pharmacy plan |
|
|
Regulatory review |
1. Evaluate pre clinical trials data 2. Assess in country regulations for drug/vaccines/product import 3. National regulations on special situations - genetic material, organs, stem cells, reproductive tech etc 4. Intellectual property issues in transfer of samples and data 5. Exchange of visitors/ scientists 6. Foreign funding 7. Research in high-security areas and borders. |
|
|
Range of ethical issues in health research |
1. Competence of researcher/ team 2. Protection of human rights, esp in vulnerable groups 3. Confidentiality and non-discriminatory practices 4. Informed consent and study specific IEC material 5. Mechanism for reproting and managing adverse events 6. Care and support for research participants: Standards, long-term care, Post-trial access to care 7. Reimbursement and compensation 8. Continuing review of progress of the study |
|
|
Role of Institutional Ethics Committe/Board |
1. Does the study have benefit? 2. Are rights protected? 3. Does benefit outweigh risks 4. Will the participants or communities have access to the study findings and the benefits of research? 5. Mechanism of safety, care and support to participants |
|
|
How ethics influence the practice of medicine and research |
1. Growing expectation about accountability: questions about the responsibility of the Govt. & Researchers due to advocacy 2. Universal right to health care 3. Place for self responsibility is fading, blaming the researchers for mishaps 4. Need to include Bio-ethics in medical curriculum Ethics is being addressed and challenged increasingly. The policy makers and researchers should search for solutions with sensitivity and realise the scope for improvement of practices |
|
|
Challenges to Clinical Trials in India |
Rapid expansion in 1st decade of 21st century Then regulatory clamp down due to reforms in 2012-13 Main challenges now are: a. delayed approval b. quality of ethics review c. Problems with import and export of samples d. Deficiency of trained investigators and good centres e. Clause of compensation to participants of clinical trials f. A/V consent for IND trials |
|
|
Review and regulatory bodies |
Scientific review: Insititutional scientific advisory committee/ ICMR Ethics review: Institutional Ethics Committee / National Ethics Committee Regulatory review: MoH Screening committee/ Drug controller general of India/ Genetic Engineeering approval commitee |
|
|
Ethical issues for clinical trials |
1. Mechanism for independent ethical review 2. Mechanisms to ensure protection of human subjects 3. Check for adequate community engagement and support 4. Informed consent 5. Standard of care and post-trial support 6. Use of placebos 7. Confidentiality |
|
|
Issues in trial implementation 1 |
1. IC procedure 2. Strict adherence to Incl and Excl criteria 3. Good lab practices, quality control and assurance 4. Adherence to intervention 5. Follow up 6. Standardisation of protocols in multi-centre trials |
|
|
Issues in trial implementation 2 |
7. Independent monitoring 8. Safety assessment: Report and manage Adverse events 9. Reimbursements, compensation and grievence redressal 10. Trial stoppage rules 11. Documentation archival |
|
|
Impediments to trial participation |
In the participant: 1. No knowledge 2. No access 3. Suspicious or afraid of research 4. Can't afford to participate 5. May not want to go against primary health care provider's wishes In the healh care providers: 1. Lack of awareness of clinical trials 2. Unwilling to lose control of clientele 3. Belief that standard therapy is the best for the patient 4. Concernted abou the added admin burden due to trial |
|
|
Seven steps of a research protocol |
1. Identify topic, RQ and objectives 2. Outline an one-page concept paper 3. Prepare dummy tables 4. Write draft protocol 5. Prepare instruments and annexes 6. Submit for peer review 7. Seek ethics approval |
|
|
Rationale for the one-page concept paper |
Time is precious, for you, for the committees, for everone Brevity forces focus If a concept paper can't be developed, abort the idea and save time |
|
|
Outline of One-page bullet style concept paper |
1. Background and justification 2. Objectives 3. Methods 4. Expected benefits 5. Key references 6. Budget |
|
|
CP Background and Justification |
Importance of the problem The known about the problem The unkmown about the problem (lacunae) |
|
|
CP Objectives |
2-3 objectives, Specific to general, Primary and secondary |
|
|
CP Methods |
One point per bullet: * Study design * Study pop * Operational definitions * Sampling procedures * Sample size * Data collection * Analysis plan * Human participant protection |
|
|
CP Expected Benifits |
What action will be taken following results Future research agenda |
|
|
CP Key references |
Not more than five As per std guidelines of icmje.org |
|
|
CP Budget |
4-5 lines No detailed justification Divided into Salaries/ per-diem, trave, equipment, supplies and miscl. |
|
|
ICMR calls CP as Pre-proposal format |
Title in 25 words Intro in 250 words Novelty in 100 words Applicability in 100 words Description: Methods, Feasibility, Outcome and Budget in 700 Words |
|
|
ICMR's Short-term studentship (STS) for medical UGs format |
Title 25 words Intro 300 words Objectives 100 words Methodology 800 words Implications 100 words References 300 words |
|
|
Draft Protocol |
Uses the concept pater as summary for outline Does not exceed 2000 words Intro is <20% of total Five to ten key references |
|
|
Methods section in protocol |
1. Study design 2. Description of interventions (for experimental studies only) 3. Study population 4. Operational definitions 5. Sampling procedures 6. Sample size 7. Data collection 8. Analysis plan 9. Project implementation plan (Quality assurance practices) 10. Human subject protection |
|
|
Protocol: Study design |
Explains how objectives lead to indicators and to study design Describes the type of study (experimental/cohort/case control/ Cross sectional) Describes logistical arrangements (prospective/retorspective) |
|
|
Protocol: Description of interventions |
Describes intervention or treatment Who? What? When? How? |
|
|
Protocol: Study population |
Use time, place, persons Inclusion and exclusion criteria Not to be confused with study sample Explain how study population is suitable to address the objectives |
|
|
Protocol: Operational definitions |
Spells out and justifies: Key exposures and key outcomes Clarity and specificity is essential in defining them References, if applicable |
|
|
Protocol: Sampling procedure |
1. Describes and justifies the type of sample used (random, systematic, cluster etc) 2. The procedure to collect the sample in practical terms 3. Explains randomisations procedure, when applicable 4. Refrences |
|
|
Protocol: Sample size |
Details all parametres used to arrive at sample size Explains the formulae and software used References |
|
|
Protocol: Data collection |
Lists all data to be collected How will the data be collected: Instruments and methods Who will collect data: Training and background of staff |
|
|
Protocol: Data analysis |
Data entry Software used Recoding stage: new groups based on existing variables Descriptive stage: Prevalance and Incidence Analytical stage: Types of analysis (univariage, stratified, Mulitvariate) |
|
|
Protocol: Project implementation |
Address the sequential steps of the processes: A. Data collection, entry, analysis and reporting B. Roles and responsibilities of various investigators C. Project governance procedures D. Co-ordination of project activities E. Project timeline Explain the quality assurance practices to be followed in each of the above steps |
|
|
Protocol: Human subject protection |
Explains the steps that will lead to: A. Minimisation of risks and confidentiality B. Maximisation of benefits C. Compensations (without undue incentive) D. Informed consent E. Approval procedures (ethics committee) |
|
|
Protocol: Data collection instruments |
List in full: Questionnaire, abstraction forms, Structured observation guide, Interview/ FGD guide etc. |
|
|
Protocol: Annexes |
Will contain: Standard Operating Procedures (SOP) Training framework for field workers Participant recruitment material Adverse event reporting and management forms Informed consent forms Study management forms |
|
|
Protocol: Finalising |
Send for peer review to colleagues and experts Review by ethics committee Archive all drafts, to see the way the protocol draft has changed at various points www.equator-network.org has templates of protocol for different study types. |
|
|
Scope of health research |
1. Get new information 2. Verify available Information 3. Explain cause and effect 4. Test new methods/drugs 5. Evaluation of ongoing programmes |
|
|
Research question to objectives |
1. Use scientific/epidemiological terms for objectives 2. One verb per objective 3. Sort objectives as primary and secondary 4. For descriptive studies use 'to estimate' 5. For analytical studies use 'to determine' |
|
|
Dimentions of research |
Theoretical - Applied Preventive - Theraputic Bench based - Bed side Exploratory - Confirmatory Implementational - Translational |
|
|
A good hypothesis |
Simple: One exposure and one outcome Specific: Clear definition of participants and variables State in advance: In writing Focued on primary objective |
|
|
Categories of research questions |
1. Descriptive questions - Only observation 2. Analytical questions - Involves comparisions, Interventions to tesh a hypthesis |
|
|
Practical answers through health research |
1. At an individual level: a. healthy behaviour, b. Prevention, c. Early diagnosis, d. Proper treatment, e. Rehabilitation 2. At the community level: a. Improve health behaviour and priorities, b. Prevention and control programmes, c.Support to the affected persons, d. Stigma reduction |
|
|
Uncertainity to research question |
1. Frame the problem in specific clinical/public health terms 2. Focus on one issue 3. Use plain everyday language 4. One operational verb (more, only if needed) 5. Should cover the question, the answer and the planned action 6. Stated only as a question. |
|
|
Considetations in planning health research |
Adequate justification for use of resources A clear research question Standard case definitions, unambigous outcomes Sample - representative and of adequate size |
|
|
The 'So What' test of a study |
FINER Feasibility Interesting Novel Ethical Relevant |
|
|
Research Hypothesis |
* It is a more specific version of the research question. with details on Sample, Exposure, Outcome and Statistics * Descriptive studies do not need a hypthesis, only for analytical studies * Research questions that need a research hypothesis are those that use the comparitive words. (leads to, causes, compared to, < or >, associated with, related to, similar to) |
|
|
Stages of research |
Planning - Methods, Team, Reviews Data collection Meaningful conclusions Appropriage decisoins leading to actions Help in reduction of suffering |
|
|
Breadth and depth of inquiry |
* Humans - Healthy, at risk, diseased, dead * Environment and society - Housing, Social practices etc. * Health care delivery - Infrastructure and delivery of health care |
|
|
What is a research question |
An uncertainity that the researcher wants to resolve by making measurements in the study population. Begins with an uncertainity and narrows down to a concrete, researchable issue It is about what the researcher wants to know, not about how he will get to know. |
|
|
2 main challenges |
1. Confounders: Affect variable and outcome Reduced by proper design and stratification 2. Effect modifiers: Affects only outcome due to extraneous factors influencing it. Disturbs the Variable-Outcome relationship Reduced by proper knowledge |
|
|
Important methods in research |
Pilot Participants Data collection instruments Measurement tools Statistical analysis Quality control |
|
|
Life cycle of research |
Data needs --> Research question --> Objectives --> Plan the analysis --> Prepare data collection instruments --> Collect data --> Analyse data --> Draw conclusions --> Recommendations --> Inform stake holders |
|
|
Focus of health research |
Improve population health Predict Illness Prevent disease Effectively reduce morbidity and mortality Interventions at various levels for disease prevention and control |
|
|
Major study designs |
Quantitative - Qualitative Observational - Experimental Prospective - Retrospective |
|
|
Sources of research questions |
1. Mastering what's published 2. Being alert to new ideas 3. Cultivating a healthy skepticism 4. Try to apply new tech to old problems 5. Keep imagination up 6. Choose a guide or mentor |
|
|
How to conceive a research question |
Review state-of-the-art information Raise a question Decide on the its worthiness by peer review Define measurable exposure and outcomes Sharpen the initial question Refine the question by specifying details |
|
|
Errors in research |
1. Random errors: Human, Chance and unknown errors Reduce by precise measurements and increasing sample size 2. Systematic errors or Bias: Due to fautlty tools, measurements, procedures etc. Reduce by improving study design |
|