Multiple Baseline Design

1
Multiple Baseline Design

• Change occurred
• A result of Rx.
• Sig change defined by ?.

2
Design implementation

• Select units for potential inclusion in the
  research (defined criteria )
• Randomly select from these for study
• Establish baseline and ensure its stability
• Randomly select unit for first intervention
• Implement intervention strategies in that unit
  until desired effect achieved.
• Randomly select next unit and apply same
  intervention

3
Possible variations in MBL

• Same behaviours different units of study
• Similar interventions for same behaviours in
  different settings or units.
• Interventions to alter alcohol harm, tobacco
  ,diabetes , physical activity , asthma
  management, in different units such as
  individuals communities ,emergency departments
  ,churches, schools ,states, country, research
  products
• Changing intervention research with indigenous
  communities

4
Published behavioural research indigenous health

• 1987-88

1997-98
5
Some variations in MBL

• Same behaviours different settings
• Same setting, different behaviours
• Modification of cervical , mammogram screening in
  same community/individuals. In ED changing
  providers adherence to best evidence management
  of asthma and then intervene with another
  clinical behaviour. Potential contamination
  between like behaviours.
• The same funding agencies attempt to change
  research outcomes such as publications across
  topic areas

6
Focus of alcohol-related behavioural research
over time
Same search terms. Paper on focus Expected a
shift to interventions Intervene in alcohol
research then another area
7
Limitations of Design

• Fewer study units limits generalisability
• Measures must be suitable for repeated use.
• Definition of a stable baseline
• How far apart must interventions be staggered
• Intervention effective by chance in selected
  units
• Depends upon successful , abrupt temporal
  relationship between intervention and measure

8
Strengths

• Each unit acts as own control
• All settings get intervention if it is effective
• Feasible to use with individuals, small and large
  units
• More intervention research and knowledge
• Can examine additive/deleting intervention
  components.
• Research costs reduce if data routinely collected
  .
• Appropriate and acceptable statistical analysis
  exist
• Consistent with decision making process used by
  wide range of influential groups. Policy makers
  in Police,education,health, utility suppliers

9
The practioner/scientist question of research
designs

• Dont ask what intervention question a research
  design can answer but what credible research
  design can be used to effectively evaluate an
  intervention which might help our community