_U4

_{(text, p.58-59)}

Sample

•  the group of people in a study _{(text p58)}
•  Sampling: studying a proportion of the population rather than the whole population _{(text p58)}
• threats to sampling _{(Text p59)}

U4

(text, p.19-20)

Population

• the S group or clinical scenario of interest
• may be single patient  or group
  • group has same condition/problem
• as manager > scope may be a wider aspect of delivering or organizing health services
  • nurse led care for ppl with chronic cond'n

U4

(Text, p. 59-61 & glossary & ppt 2-3)

Probability Sampling

(random sampling)

• AKA: random sampling (Txt p47-49, 60, 299) (U4 PP 2-3)
• random selection ensure different units in the population being sampled are equally likely to be selected (text glossary)
• Quantitative studies usually use sampling techniques base on ____ theory.
•  Random selection of Ss from whole target population (U4 PP2)
• two features of ____ ____ (text p. 59-60)

1. researcher (in theory) has access to all the members of a population
2. every member of the pop'n has equal and non-zero chance of being selected

• they cannot have "no chance"

•  balances characteristics of Ss (U4 PP 2-3)
•  Most likely to represent population (U4 PP 2-3)
  •  Not perfect but close (U4 PP 2-3)
•  Can estimate (minimizes) sampling error 
  • diff b/w sample representativeness and pop'ns un/known characteristics  (glossary) (U4 PP 2-3)
• If sample is large enough:
  • differences in outcomes b/w groups are detected statistically
• imitation: Sampling bias (glossary) (U4 PP 2-3)
•  Very expensive and intensive (U4 PP 2-3)
• If sample is too small > CI is wide (less precise)

  ---

• Qualitative research uses non-probability sampling
  • convenience
  • consecutive
  • quota sampling

• Selection of subjects by a non-random method
• Not representative of the target population
• Will  have limited generalizibility due to sampling bias (interpret with caution)
• Convenient and economical

U4

(text, p.60 & pp 3-1)

Random Sampling

see Probability Sampling

U4

(text, p. 60, 83-84)

Sampling Error

• the gap between a sample's representativenes and the population's known or unknown characteristics
• unavoidable b/c no sample can ever totally represent the population
• = chance, or spread of sampling distrobution
• the findings could be a chance occurrence even when there is no true effect.
• Even if there is no true effect, we can observe events that look as though there is an effect, simply because of chance
  • particularly with small numbers

• reduced by probability sampling (U4 PP 2-3)
  • highest chance of sample being represent'v
  • statistical justification for sample size
• Minimized by reducing differences between a samples representativeness and populations known or unknown characteristics 
• ____ ____ generally decreases as sample size increases

• [1b] occurs when sample not representative of population (U4 PP 3-1)
• caused by flaw in research process
  • increasing sample size > increases this bias
• when results are distored by systematic differences in the way Ss are assigned to Tx or Ctrl grp (Glossary)
• Readers look for
  • large enough sample size
  • similar baseline characteristics

• ---

• Seen in cohort analysis
• two biases r/t sampling that affect generalizability are
  • referral filter bias
  • volunteer bias

U4

(text, p. 60)

• the selection that occurs at each stage in the referral process from primary to secondary to tertiary care can generate patient samples that are very different from one another

U4

(text, p.61)

• a bias r/t sampling that affects generalizability 
• people who volunteer to participate in a study may have exposures or outcomes that differ from those of non-volunteers
  • they tend to be healthier

• ensures groups are similar in all respects except exposure to the intervention (text p.106)
• Randomized Controlled Trial (Randomized clinical trial, randomized trial) (RCT): 
  • a study in which individuals are randomly allocated to receive alternative preventive, therapeutic, or diagnostic interventions
  • follow-up to determine the effects of the interventions 
  • (one of the alternatives might be no intervention)
• [3a] AKA: Random Allocation (text p47)
  • all factors that influence outcomes (both known and unknown) are evenly distributed between the groups and therefore, at the end of the study, and differences among the groups can be attributed to the intervention (text p61).
    • sample size & representativeness key
• [3a] used in RCT > groups similar, except for exposure (Text p47)
•  [3a] Was the assignment of patients to tx randomized? (txt p105)
  • ensures that groups are similar in all respects except exposure to the intervention (text p47)
• [1b] Random Allocation to treatment group controlled by investigator (CW p27) (text p47-4)
• all Ss have same chance of being assigned to any of the study groups (text, p.106)
• most statistical tests assume random allocation (CW p27)
• [3a] in RCT Volunteers may not be representative (CW p27)
  • [3a] So data not generalizable (Text p48)

    ---

• methods of randomization
  • computer-generated numbers
  • tables of random numbers
  • coin flipping

    ---

• eliminates selection bias & confounding biases

U4

(text, p. 107, 116) (PP 4-3)

• a part of randomization > prevents from knowing upcoming assignments in advance
• the clinician recruiting patients to the study should be unaware of the treatment group to which the next patient will be allocated
  • if recruiters know th allocation schedule in advance, this may influence their recruitment bhvr.
• [3b] so as not to influence recruitment bhvr 

  ---

• Ideal: sequentially numbered, sealed, opaque envelopes
• calling a central, coordinating office for reach patient ass't
• numbered or coded bottles or containers

U4

(text, p.47, 107-8, 124, 171-2, 187, 198, 121) (ppt 5-1)

• should be long enough to detect outcome of interest (text, p.171)
  • length of f/u period is r/t the latency period of the outcome of interest (text p187)
• [10] 2 Questions: (text p107)
  • [10] how long were Ss f/u? (text p107) (U4 PP 5-1)
    • Long enough > sufficient # of Ss actually experience the outcome of interest (U4 PP 5-1)
  • [10] Where Ss “lost to follow-up” accounted for? (U4 PP 5-1)
    • may underestimate the positive or negative outcomes of disease
  • How many Ss dropped out? (text p107)
    • Suspicion if dropout rate differ bw Tx & Ctrl grp
    • If large number “lost to follow up
      • Validity decreases
      • ‘lost’ Ss may have exp diff outcome
    • Sensitivity analysis
    • Investigators recalculate results using different assumptions about what might have happened to the lost patients
    • More confidence if a sensitivity analysis that tests different assumptions does not overturn the main conclusion
• [10] RCT, follow-up to determine if S experienced a specific outcome (text 47)
  • was f/u sufficiently long and complete?
    • drop-outs > question results validity b/c their exp diff vs other grp
• [10] [2] Longitudinal nature of study (txt p47)
  • [10] Exposure to independent variable precedes development of outcome (text p47)

U4

(text, p. 48-49, 110)

• group differences in outcomes at the end of the study may be because of differences in the groups that existed before the intervention group
• the intervention therefor may appear to have had an effect on an outcome when, in fact, initial differences in the groups influenced the outcome.
• imbalances in baseline characteristics that exist after randomization can be adjusted for, using statistical techniques
  • be confident when results are consistent for analysis done with and without adjustment
• examining characteristics of groups at baseline and comparing them for differences is important
• even if investigators document group differences in baseline characteristics or use statistical techniques to adjust for differences, other factors that were not considered may be responsible for differences in outcome
  

  ---

• differences that cannot be measured would remain unknown

U4

(text, p. 108, 121)(U4 PP 5-1) (glossary)

• patient outcomes are attributed to the group to which the patient was originally assigned, even if the patient dropped out of the study or, for other reasons, did not actually receive the planned intervention (Text, p121)
• Were patients analyzed in the groups to which they were initially randomized?
• In a RCT, all patients are analyzed in the groups to which they were randomized, even if they failed to complete the intervention or received the wrong intervention (Glossary)
• [7] were patients analyzed in the groups to which they were initially randomized (U4 PP 5-1)

  ---

• if omitted from the analysis, we would be left with only the patients who were more likely to be compliant and who had better outcomes
  • the Tx would appear more effective
• the numbers presented in the analysis are close to the numbers initially randomized

U4

(TEXT p69-70, 130-33, 187, 108-9, 116-119, 197, 117)

• keeping Ss, clinicians, outcome assessors, and or analysts unaware of pt allocation (avoid bias)
• prevents study Ss and staff from determining:
  • the group who which Ss have been assigned (this leads to ascertainment bias) 
  • safeguards the sequence after allocation,
  • cannot always be imlimented.
• most often possible to _____ the person measuring the outcome (or analyst)
  • done by coding data, with no id of tx grps
• the people who measure the outcome should NOT be the same as those giving usual care
• authors should specify whether the study was _____ and explicitly state which groups were _____
• [5] In an RCT > investigators can blind or mask: (CW p27)
  • Patients (CW p27)
  • Staff (CW p27)
  • Assessors (CW p27)
• [5] Were patients, clinicians, outcome assessors and data analysts blinded to patient allocation?
• Were patients, clinicians, outcome assessors, and data analysts unaware of patient allocation? (U4 PP 5-2)

  ---

• double blinding: blinding patients, clinicians and outcome assessors.
  • allocation concealment is better indicator of quality

U4

(text, p.109)

• occurs when extra care is given to one group and not the other
  • more common when clinicians not blinded to allocation
• Where Ss in each group treated equally except for the Tx of interest?

U4

(txt p48) (text glossary)

• Tx grp discuss their experiences with Ctrl group
• reduced using 
  • cluster randomization
  • unit of randomization
• study Ss in control group accidentally receive the experimental intervention
  • minimizes potential differences in outcomes between groups

_{U4 - obj A. (p.)}

to assess validity of study results, which includes: 

• The assignment of patients to the intervention verses control group (i.e. Randomization)

_{U4 - obj B. (p.)}

to assess validity of study results, which includes: 

• appropriateness of the randmization process

(allocation concealment)

_{U4 - obj C. (p.)}

to assess validity of study results, which includes: 

• length and completeness of follow-up