Using discrete choice experiments to elicit preferences for maternal health care in Ghana Presenting author: David Newlands University of Aberdeen, Scotland,
Post on 02-Jan-2016
222 Views
Preview:
Transcript
Using discrete choice experiments to elicit preferences for maternal health care in Ghana
Presenting author:David NewlandsUniversity of Aberdeen, Scotland, UKd.newlands@abdn.ac.uk
Co-authors:Aba Daniels, University of Ghana Laura Ternent, University of Aberdeen
Discrete choice experiments (DCEs)
• DCEs are a type of stated preference technique for establishing the importance of the different attributes of a good or service
• Now well established in the analysis of health care in developed countries
• Much less commonly used in developing countries – partly because respondents’ self administration of questionnaires requires a literate population
• It is necessary in such settings to provide respondents with additional support, such as visual aids
DCE of maternal care in Ghana
We conducted a DCE to elicit individual preferences for hospital based maternal health care in Ghana, with one sample using the traditional method of self administration and with another sample using visual aids
Study was conducted as part of Immpact (www.immpact-international.org), a multi-country, multi-disciplinary research project on maternal ill health, funded principally by the Gates Foundation
Methods
• Five attributes and appropriate levels were selected to represent aspects of hospital based maternal health care, by consulting expert opinion and from a focus group discussion
Five attributes were:• Care by skilled attendant• Staff attitude• Time taken to get to hospital• Availability of vital equipment• Cost of care
• Attributes and levels were combined into 17 discrete choice sets
Example of a choice set
Choice 1 Hospital A Hospital B
Care by skilled attendants
Trained Midwife Nurse
Staff attitude Unfriendly Friendly
Time taken to get to hospital
90 minutes 120 minutes
Availability of vital equipment
Some All
Cost of care ¢120,000 ¢240,000
Which scenario would you choose? (tick one box only) Prefer A Prefer B
Choice sets
• The choice sets included two tests of rationality and a test of consistency
• The visual aids used to represent the attributes and levels were designed by artists at the University of Ghana
Questionnaire
Questionnaire comprised six sections:
A. Household roster: age, sex, occupation, education, marital status
B. Housing and asset ownershipC. ExpenditureD. Maternal health questionsE. DCE with visual aidsF. Comprehension questions
• Trained enumerators administered the survey (sections A-D, F) and used visual aids to record respondents’ choices (section E)
Sample
• Individuals were selected from three communities in the Cape Coast region of Ghana
• Enumerators entered the village, chose the first house in each village to be interviewed randomly, and chose every other house thereafter until a quota had been reached
• Interviewed either a male or female adult respondent
Respondents’ characteristics
104 men, 102 women
Mean age 33
Marital status: 76% married/cohabiting, 17% never married
Occupation: 24% trading, 22% fishing, 16% skilled self employed, 14% unskilled self employed, 10% skilled employed
Education: 42% no formal education, 15% primary, 31% middle, 8% secondary, 3% post secondary
Results
Logit model (all respondents)
Logit model (men only)
Logit model (women only)
Skilled attendant
-.720***
-0.724*** -0.725***
Attitude -.474*** -0.507*** -0.448***
Time -.002*** -0.004*** -0.0004
Equipment -.813*** -0.887*** -0.748***
Cost -.0015*** -0.002*** -0.0015***
Constant .115*** 0.177*** 0.055
N 3288 1664 1624
Prob(Chi2) 0.0000 0.000 0.000
Psuedo R2 0.1813 0.192 .01743
Attribute Variable name Levels and coding of variables
Care by skilled attendant Skilled attendent Midwife = 0
Nurse = 1
Staff attitude attitude Friendly = 0
Unfriendly = 1
Time taken to get to hospital Time 30 minutes = 30
60 minutes = 60
90 minutes = 90
120 minutes = 120
Availability of vital equipment Equipment All = 0
Some = 1
Cost of care Cost ¢0 = 0
¢60,000 =60
¢120,000 =120
¢240,000 =240
Rationality/consistency checks
% of respondents passing:
Rationality check 1 94
Rationality check 2 95
Consistency check 88
top related