Verbal Autopsy

Tool Name: Verbal Autopsy

Origin / Source: World Health Organization (1)

Technical area: Maternal Health

To Tool:
http://whqlibdoc.who.int/hq/1995/WHO_FHE_MSM_95.15.pdf (1.2MB)

Basic Description: The verbal autopsy (VA) is a technique whereby surviving family members or other informants are interviewed to elicit the decedent's cause of death, to identify risk factors for the death, and to assess the accessibility and quality of the health care received by the decedent. This review focuses on the use of verbal autopsy interviews to identify women who died from pregnancy-related causes.

Country Applications:Guinea-Bissau, (2) Egypt, (3) Tanzania, Ethiopia, Ghana, (4), (5)

Language(s) Available: English. Investigators have also translated their questionnaires into the local language as required.

Purpose: Results from a verbal autopsy are most useful for

  • promoting understanding of the relative burden of maternal deaths in a population and providing key information for maternal health advocacy by evaluating the magnitude of maternal mortality and other causes of death among women of reproductive age;
  • identifying populations of women at higher risk for maternal death so that health care and social services programs can be designed to help prevent deaths; and
  • assessing the need for health care service quality and improved access.

Technical Scope/Purpose: Depending on the goals of the study and the study population of interest, researchers may use VA's in an attempt to identify all deaths among women of reproductive age (WRA) in a certain hospital, region, or population. (See review of RAMOS methodology.) In various studies, the verbal autopsy questionnaire has also been used to determine

  • all causes of death in adults;
  • deaths due to HIV/AIDS;
  • causes of child mortality;
  • deaths due to maternal; and
  • deaths due to other reproductive health-related causes (such as reproductive tract infections causing pelvic inflammatory disease.)

The questionnaire includes questions about signs and symptoms of illnesses at the time of death, medical history and obstetrical history, various socio-demographic and medical risk factors (such as age, socio-economic status, and distance to health care facility, etc.) and health care seeking behaviors. In addition to such quantitative data, the interviews may also collect qualitative data, such as the knowledge, attitudes, and behaviors of the respondents and deceased regarding health, illness, and health care-seeking behaviors including delays in accessing health care.

Design: The verbal autopsy is a type of questionnaire used in an epidemiological study, generally a retrospective review of deaths. Verbal autopsies are ideally conducted with the person or persons who were with the decedent at the time of their death. Health care personnel who provided health care to the deceased may also be interviewed, but family interviews are the main information source for this type of data collection.

The methods used to identify pregnancy-related deaths are optional, depending on the feasibility in the local setting of identifying deaths among WRA which may have been related to the pregnancy. The duration of the study depends on the number of deaths among WRA that need to be investigated. Death investigation is relatively time-consuming. For example, identification of deaths in WRA by a census among Afghan refugees in Pakistan took several months, and it took one month to train interviewers, conduct VA interviews among families of 66 WRA who died, input and analyze data, and present a report to the participating non-governmental organization (NGO).(6)

Essential partners include local health care providers, NGO's, and the Ministry of Health, etc. Essential staff may include a study coordinator or supervisor or principal investigator, interviewers, interpreters if necessary, data processors and a data analyst. In addition, a team of local stakeholders (such as health care providers, public health social services program providers, and local community leaders) is important in reviewing the study results and making recommendations for improving maternal health.

Method: Quantitative and qualitative

Frequency of Administration: In a large population, VA studies are often only conducted once to determine the scope of the problem of maternal mortality; raise the visibility of this issue among stakeholders, and help set health care priorities.

However, in smaller populations (such as smaller regional or hospital level), this type of study may be conducted again at a later time to evaluate the impact of any health care or social services program changes that developed to address the results of the initial study. Some areas may choose to review deaths periodically, similar to the United Kingdom's long-standing Confidential Enquiry into Maternal Deaths (CEMD) (6).

Key Users of the Information: Health care providers, programmers, and policy makers, communities, and other health researchers.

Key Indicators:

  • Proportional mortality of death among WRA due to pregnancy-related causes.
  • With supplementary data on the number of births and population size the following indicators may be calculated:
  • maternal mortality ratio;
  • maternal mortality rate;
  • lifetime risk of maternal death;
  • and crude mortality rates.

Implementation and Training:

The local study coordinator and/or principal investigator (PI) will need to be familiar with general survey methodology including questionnaire development or adaptation, data collection techniques including interviews and interviewer training, data input, and analyses. Depending on the local research skills available, external epidemiological technical assistance may be required. The first step for the researchers is to review the autopsy questionnaire and customize as necessary.

Training requirements depend on the skills, knowledge, and experience of the interviewers and on the size of the study. If clinical personnel are used to interview families, less training on asking clinical questions will be needed but the interviewers must be directed to ask all questions on the questionnaire and not diagnose cause of death prematurely, thereby possibly assigning cause of death erroneously. Non-clinical interviewers must also be directed to complete the questionnaire, and they may also be taught some questions to probe for further clinical information on a specific sign or symptom (for example, severity of hemorrhage.) For all interviewers, training is essential to

  • ensure an informed consent to the interview is acquired;
  • provide communication skills to appropriately acknowledge grief exhibited by the family;
  • ask questions in a non-leading, non-judgmental manner; and
  • record the answers objectively and legibly.

Skilled clinical personnel should review any available medical records and the completed verbal autopsy questionnaires. If these personnel are not local, they must work with persons familiar with the area and local resource setting to determine the potential barriers to health care access and the means to address those barriers.

Manuals and Guidelines (if applicable):

Campbell O. and Ronsmans C. Verbal Autopsies for Maternal Deaths. Report from the Maternal Health and Safe Motherhood Programme, World Health Organization. London School of Hygiene and Tropical Medicine. WHO/FHE/MSM/95.15.

Data Processing and Analysis: Data can be entered locally into any simple data system including Epi Info. Analyses include calculating indicators (for which data on live births will be needed), proportional mortality of deaths related to pregnancy, and frequencies of risk factors for death (such as age, parity, socioeconomic status, and distance to health care facility). One person, usually the primary investigator, should be responsible for data entry and analyses. If more than one data entry person is required, one person should be responsible for ensuring the accuracy of all data entry and analyses.

Reporting and Dissemination of Results:

It is essential to report back to the people responsible for designing and implementing health care or social service programs, health care providers, and the general community. Preliminary analyses of data can often be conducted in the field such that these results can be presented immediately to local stakeholders. After review of the data, recommendations for programs to prevent maternal deaths can be developed. A final written report should then be sent to the stakeholders. In some cases, results may be presented in public venues so families who were interviewed may attend.

Cost: This study could be done at a local level if the research skills are available locally. If not, at least some donor funding would be required to access the appropriate technical support.

Lessons from Experience: The verbal autopsy technique has been shown to be a valid tool for determining cause of death. (7) This method provides a rich data source on the magnitude of and risk factors for pregnancy-related death and barriers to health care access. (7) (8)

Contact Person:
L. Bartlett, MD, MHSc, Medical Epidemiologist,
Division of Reproductive Health, Centers for Disease Control and Prevention (CDC), Atlanta, Georgia, USA.
LBartlett@cdc.gov

Notes

1. World Health Organization, Maternal Health and Safe Motherhood Programme, 1994. Prepared by Campbell O., Ronsmans C. Verbal Autopsies of Maternal Deaths.

2. ii Hoj L., Stensballe J., Aaby P. Maternal mortality in Guinea-Bissau: the use of verbal autopsy in a multi-ethnic population. International Journal of Epidemiology 1999;28:70-76.

3.Grubb G.S., Fortney J., Saleh S., Gadalla S. et al. A comparison of two cause-of-death classification systems for deaths among women of reproductive age in Menoufia, Egypt. International Journal of Epidemiology 1988;17(2):385-391.

4. Chandramohan D., Maude G.H., Rodriguez L.C., Hayes R.J. Verbal autopsies for adult deaths , their development and validation in a multi-center study. Tropical Medicine and International Health. 1998; 3(6): 436-46.

5. Chandromohan D., Rodrigues LC., Maude GH., Hayes RJ. The validity of verbal autopsies for assessing the causes of institutional maternal death. Studies in Family Panning. 1998; 29(4):414-22.

6. Department of Health, UK. Why Mothers Die. Report on Confidential Enquiries into Maternal Deaths in the United Kingdom, 1994-1996. Department of Health, 1998.

7. Bartlett L., Khan T., Sultana M., Jamieson M., Wilson H. The burden of mortality due to reproductive health-related causes among Afghan refugees in Pakistan, 1999-2000 (abstract). In Conference 2000: Findings on Reproductive Health of Refugees and Displaced Persons: proceedings. Washington, D.C., December 5-6, 2000. Reproductive Health for Refugees Consortium, 2001, pg.18. At: http://www.rhrc.org/conference/index.htm

8. Bartlett L., Khan T., Sultana M., Jamieson M., Wilson H. Results of a reproductive age mortality survey (RAMOS) among Afghan refugee women in Pakistan, 1999--2000. Journal of Pediatric and Perinatal Epidemiology, in press.