Ghana - Ghana Maternal Health Survey 2007

To achieve the above-mentioned objectives and to obtain an accurate measure of the causes of maternal
mortality at the national level, and for the Reducing Maternal Morbidity and Mortality( R3M) regions (Greater Accra, Ashanti and Eastern regions) and other
regions (Western, Central, Volta, Brong Ahafo, Northern, Upper East and Upper West), 1600 primary sampling
units were selected (half from the R3M regions and half from the other regions) within the 10 administrative
regions of the country, across urban and rural areas. The primary sampling units consisted of wards or subwards
drawn from the 2000 Population Census. This sample size was estimated from information in the 2003 Ghana DHS survey; it was expected that each primary sampling unit would yield, on average, 150 households. GSS and GHS enumerators carried out a complete mapping and listing of the 1600 selected clusters. This first phase of data collection yielded a total of 227,715 households.

A short household questionnaire was administered to identify deaths that occurred in the five years preceding the survey to women age 12-49 in each household listed in the selected cluster. In the second phase of data collection a verbal autopsy questionnaire was administered in all households identified in the first phase as having experienced the death of a woman age 12-49. This yielded a total of 4,203 completed verbal autopsy questionnaires.

In the second phase of fieldwork, 400 clusters were randomly selected from the 1600 clusters identified in the first phase. Households with women age 15-49 were selected from these 400 clusters (half from the R3M regions and half from the other regions) and were stratified by region and urban-rural residence to yield 10,858 completed household interviews and 10,370 individual women's interviews. These households were selected randomly and independently from the households identified in the first phase as having experienced a female death.

Institutional populations (those in hospitals, army barracks, etc.) and households residing in refugee camps were excluded from the GMHS sample.


No deviation of the original sample design was made


A total of 11,579 households were selected for the sample, of which 10,994 were occupied at the time of the survey and 10,858 (or 99 percent) were successfully interviewed. The difference is primarily due to dwellings being vacant or the inhabitants being gone
for an extended period at the time of the survey. In the interviewed households, 10,627 women were identified as eligible for the individual interview (women age 15-49), and interviews were completed for 10,370, or 98 percent. The principal reason for nonresponse among eligible women was the failure to find them at home, despite repeated visits to the household. The refusal rate was low in both urban and rural areas.


Sample weights were calculated for each of the data files. Sample weights for the household data, women data and sister answered data were computed as the inverse of the probability of selection of the household, women and sistered answered data computed at the sampling domain (urban/rural within each region). The weights were adjusted for non-response at the domain level and were then normalized by a constant factor so that the total weighted number of cases equals the total unweighted number of cases.