Home » DHS-7 Questionnaire Review - Archived » Health Systems » Community questionnaire for availability of health services
|Community questionnaire for availability of health services [message #1812]
||Thu, 03 April 2014 17:05
Registered: December 2013
Suggestion type: Additions|
1. What is the information needed?
Community availability of health services
2. What questions will elicit this information?
See attached file for an example of how these data were collected in Bangladesh for health. Community-level information was also collected in Colombia (1986), Indonesia (1994), and Benin (2001). I have attached the community questionnaires for these surveys (extracted from the respective final reports) along with this posting. Some of these surveys also collection information on availability of schools and sanitation in the community.
Indicate whether or not the questions have ever been fielded or validated.
3. How will the resulting information be used?
This information can be used to investigate health service use patterns, i.e. child immunizations, family planning, antenatal care, curative care visits for both men and women, etc.
Demand for health care is often influenced by the availability and accessibility of health services. That is, in addition to cultural norms, social network influence, and individual-level knowledge and preferences, individuals make the decision to seek care for themselves or family members after weighing the benefits against monetary/financial costs and opportunity costs (time, distance, etc.). Information on distance, travel time, transportation cost, and price of service is currently not collected in the DHS.
In the past, absent true community-level data, researchers have created "community" aggregate measures. Usually the sampling cluster is chosen as the level of aggregation for these community variables. However, since those aggregate measures are subject to the responses of sampled individuals, these measures may be biased. In his 2006 paper, Kravdal showed through simulations that these biases are generally small (<4%) if the level of aggregation has a "large enough" number of people (>=25) and the intra-cluster correlation (ICC) is "large enough" (>=0.2). In practice, the cluster sizes in the DHS varies by country. Some countries have clusters that are on average >=25 people, and others have clusters that are on average much smaller than 25 people. The best way to create these aggregate measures is also disputed. Some researchers use all observations within a cluster for aggregation, while others favor creating the measures without the index case. Either way, these measures are unstable when the clusters are very small, say 3 or 4 observations in the cluster. Asking these questions of community leaders (or several community leaders, for that matter) removes some of the bias associated with measures derived from the sample.
Finally, in the DHS, the questions on barriers to accessing care is asked only of those who are not using the services. Further, the reasons for not accessing services pertain to those services related to maternal and child health, leaving out men's care-seeking patterns. Although some have used the proportion of women reporting "cost too much" as the reason for not delivering their youngest child in a health facility as a proxy measure of "cost as barrier to accessing care," there are serious methodological issues with creating proxy measures for "barrier to care" based on the responses of non-users. For instance, since "cost too much" and "too far" are subjective measures, we do not know what, on average, is an acceptable cost for which people are willing to pay or distance for which people are willing to travel for care. This has policy implications as well, i.e., when planning where to build new health posts, when setting user fees, or when developing community programs to incentivize people to get their children fully vaccinated.
4. What is the priority of suggested additions compared with what is already in the questionnaires?
These questions will be asked of community leaders, rather than sampled women or men, so it should not affect the consistency of responses for the usual respondents, which is a concern listed in the Power Point presentation accompanying the guidelines for suggesting changes. As such, it should be a medium-high priority item.
5. If suggesting more than one addition, what is the priority among the suggested additions?
6. Should the additional data be collected in all countries, or only in selected types of countries (e.g., countries with a particular type of program, countries with prevalence of a particular infection >5% or 10%)?
These data should be collected in all countries.
Kravdal Ø. A simulation-based assessment of the bias produced when using averages from small DHS clusters as contextual variables in multilevel models. Demogr Res. 2006;15:120. doi:10.4054/DemRes.2006.15.1.
[Updated on: Thu, 03 April 2014 17:07]
Report message to a moderator
Current Time: Fri Jul 3 22:43:30 Eastern Daylight Time 2020