1.gif (1892 bytes)

Brief Reports

Indian Pediatrics 2003; 40:766-771 

Use of Verbal Autopsy by Health Workers in Under-Five Children


Baridalyne Nongkynrih, K. Anand, S.K. Kapoor

From Comprehensive Rural Health Service Project, Ballabgarh, Haryana and Center for Community Medicine, All India Institute of Medical Sciences, New Delhi 110 029, India.

Correspondence to: Dr. K. Anand, Assistant Professor, Comprehensive Rural Health Service Project, Ballabgarh, Haryana 121 004, India. E-mail: [email protected]

Manuscript received: August 21, 2002; Initial review completed: September 30, 2002; Revision accepted: February 26, 2003.

Abstract:

We conducted this study to ascertain whether health workers can routinely administer verbal autopsy (VA) to determine the cause of death in their area. The health workers were trained to administer verbal autopsy tool on the deaths occurring in the population under them. All the verbal autopsies of under-five deaths between January 2000 to December 2001 were reviewed by a pediatrician. There were 262 deaths of children under five years in this period, and 71% of them were infants, out of which 34% were neonatal deaths. The health workers reported PEM, fever, pneumonia, and diarrhea as the leading causes of death, based on the existing system of obtaining information provided by the family members. Verbal autopsy forms reviewed by a pediatrician also showed that apart from fever, the first three causes of death were the same i.e., PEM, diarrhea and pneumonia. This study shows that health workers can be trained to use the verbal autopsy to ascertain the cause of death.

Key words: Health-workers, Mortality, Under-five children, Verbal autopsy

Under-five mortality is an important indicator of the quality of health care. The Under Five Mortality Rate has reduced from 109.3 to 94.9 per 1000 live births. Similarly, the Neonatal mortality rate for India has declined from 48.6 to 43.4 per 1000 live births and IMR has declined from 78.5 to 70 per 1000 live birth(1). However, the national goal of decreasing the IMR to 60 still remains to be achieved.

In order to bring about further reduction in these mortality rates, there is a need to study the causes of death in this age group. The method of finding out the causes of death is not uniform throughout India. In rural areas, till 1997, there was a system of lay reporting (Survey of Causes of Death-Rural), which was based on a sample of the country. It is now incorporated with the Sample Registration System (SRS), which gives the causes of death of a sample of the population. However, SRS covers only less than 1% of all deaths, and the methodology used for arriving at the cause of death needs improvement(2).

As most of the deaths in the rural areas occur at home and a medical certification by a qualified practitioner is not possible, there is a need for verbal autopsy. A verbal autopsy is an investigation of the chain of events or circumstances preceding death through interview of the mother or any other person who was looking after the patient, focusing on easily recognizable signs and symptoms. Verbal autopsies have been validated and used for ascertaining the cause of death in many countries. In 1992, recognizing the need for uniform and valid criteria for diagnosing common causes of death, Bang et al.(3) proposed a set of criteria for the cause of death among neonates and those aged 1-59 months.

So far, it has been seen that in almost all studies, the verbal autopsy has been administered by research workers or experts to find out the feasibility and practicality of its use in various studies. In addition, in such studies, all deaths taking place in the study sample are likely to be covered; unlike in a routine system, where all deaths may not be registered and investigated. However, if one has to use the VA as an instrument for ascertaining causes of death on a nationwide scale, it should be possible for health care providers at all levels to be able to use this tool, and should be incorporated along with the routine health systems.

In 1993, a verbal autopsy tool was developed for use in Comprehensive Rural Health Services Project Ballabgarh(4). It was validated in two stages, first by a doctor by using hospital deaths. Secondly, validation was done in the community for deaths occurring in the practice area. This tool had high sensitivity and specificity in both neonatal and post neonatal deaths(5). Hence, since 1999, it was decided that the health workers themselves should use the VA for all under five deaths in this area. This is being carried out as a part of their routine activity and covers all the under-five deaths in the villages under the project area. We report the results of this activity in this paper.

Methods

The Intensive Field Practice Area of CRHSP Ballabgarh includes 28 villages spreading about 20 kilometers. More details of this project are available elsewhere(4). A verbal autopsy is conducted for every under five death in the study area, using a structured interview schedule which has been prepared separately for 0-28 days, and 29 days -5 years. The verbal autopsy tool for 29 days-5 years consists of initial screening questions. This is followed by six separate modules, which are to be filled according to the symptoms, e.g., fever with rash, respiratory symptoms, diarrhea, convulsions, not feeding, and fever of sudden onset. Each symptom would then lead to the diagnosis of common causes of death in this age group. All the male health workers were trained to fill this verbal autopsy form. We used male health workers only as registration of vital events is a part of their job responsibility. The health workers were also asked to note the cause of death according to the symptoms/events preceding death as told to them by the relatives. All the deaths were verified by the medical officers of the PHC. These forms were reviewed by a faculty (SKK) who is trained in pediatrics and public health, to arrive at a diagnosis. In this study, we have reviewed the deaths of children under five years that took place between January 2000 and December 2001 in all the 28 villages. The verbal autopsy forms of under-five deaths were retrieved and reviewed by a senior consultant, who ascertained the cause of death according to the symptoms recorded. In addition to the cause of death, information about age, sex, month of death and village were also obtained.

Results

During this two-year period (January 2000 to December 2001) there were 262 deaths of children under five years. A majority of the deaths (69%) occurred between May and September, with a maximum number in the month of July (17%). Of the 262 under-five deaths, 186 (71%) deaths were seen in infants less than one year of age. The age distribution of these deaths (Table I) shows that out of all the infant deaths, 34% (63/186) were neonates. Of the neonatal deaths, 62% (39/63) were early neonatal deaths. The proportion of males was more in the neonatal age group, while the reverse is seen in the post-neonatal period. Almost 60% of infant deaths in the age group 29 days to one year occurred in females.

TABLE I

Age and Gender Distribution of Under-Five Deaths
Age group 
Male (%) 
Female (%) 
Total (%) 
0-6 days 
22 (56.4) 
17 (43.6) 
39 (14.9)
7- 28 days 
13 (54.2) 
11(45.8) 
24 (9.2)
29 days -364 days 
51 (41.5) 
72 (58.5) 
123 (46.9)
365 days - 5 years 
28 (37.0) 
48 (63.0) 
76 (29.0) 
Total 
114 (43.5) 
148 (56.5) 
262 (100)

 

Out of the 262 deaths, verbal autopsy forms were available for only 161 (61.5%). Out of these 161 forms, 33 (20.5%) had to be excluded from the analysis since they were incomplete or incorrectly filled. Therefore a total of 128 forms were finally included in the analysis of which 45 were in the neonatal period and 83 in the post-neonatal period. As multiple causes were ascribed to a death, there were 173 causes of death derived from 128 VA forms.

TABLE II

Comparison of Ranking of Causes of Death of Under-Five Deaths 
as Reported by Health Workers and as Diagnosed from VA forms
			 	

Cause of death as reported by Health Workers

Cause of death by verbal autopsy
(n=128)$
 

For all deaths
(n = 262)

For deaths whose verbal autopsy forms were available (n=128)*

1. PEM (17%)
PEM(17.9%) 
PEM(18%)
2. Fever (12.7%) 
Fever (10.9%) 
Diarrhea (16.8%)
3. Pneumonia (11.9%) 
Pneumonia (10.9%) 
Pneumonia (8.1%)
4 .Diarrhea(11.5%) 
Diarrhea (10.2%) 
Birth asphyxia (8.1%)
5. Prematurity (5.0%) 
Congenital  abnormalities (7.8%) 
Prematurity (7.5%)
6. Congenital abnormalities (3.8%)
Prematurity (7.l%) 
Septicemia (6.4%)
7. Others(26.5%) 
Others (28.9%) 
Others (28.7%)+
8. Unknown (11.6%) 
Unknown (6.3%) 
Unknown (6.4%)
*As VA forms were available for 128 deaths, the second column of this table shows 
the causes of deaths reported by health workers for this group.
$There were 173 causes of death from 128 deaths since there were multiple causes;
 hence denominator for these percentages is 173.
+Others include: accidents and injuries, precipitate delivery, cyanotic heart disease, 
anemia, infantile hemiplegia, intussusception, umbilical hemorrhage, respiratory distress 
syndrome, cerebral palsy, aspiration, tuberculosis, Sudden Infant Death Syndrome.

 

Table II compares the causes of death as recorded by the health workers and as ascertained by the verbal autopsy. The first column shows the ranking of causes of death as reported by the health workers for all the 262 deaths. According to the health workers, the major cause was PEM. The second column shows the ranking of causes of death separately for the 128 children whose verbal autopsy forms were available to see whether the causes of death by the workers differ in those cases with and without verbal autopsy forms. The causes of death remained the same in both categories. By doing this analysis we can assume that, there is no difference between the ‘causes of death of those children with and without the verbal autopsy forms’. The cause of death as ascertained by the verbal autopsy is shown in the third column of Table II. The causes of death were assigned as the immediate or direct causes, and contributory causes of death. Of the multiple causes of death, the most common underlying cause was found to be PEM associated with either pneumonia or diarrhea as contributory causes. Birth asphyxia also contributed to a significant proportion of deaths. In both cases, prematurity was an important cause of death, next only to diarrhea and pneumonia. The workers gave fever as a cause of death, but its rank goes down by the verbal autopsy as some specific cause gets ascribed to it. Conditions like birth asphyxia and septicemia, being more difficult and technical, such diagnoses are not expected to be made by health workers. We can also assume that there is no difference between the major causes of death as diagnosed by the trained health workers and those by health personnel, provided that the verbal autopsy form is completely and correctly filled. Here we also see that the number of unknown causes go down from 11% to 6.4%.

TABLE III

Causes of Death in the Neonatal and Post Neonatal Period According to VA
                       	

0-28 days (n = 50)
 

29 days -5 years (n = 123)
 

1.Birth asphyxia (18%) 
1. PEM (25.2%)
2. Prematurity (16%) 
2.Diarrhea (22.7%)
3. Low birth weight/small for date (14%)
3.Pneumonia (11.4%)
4. Septicemia (10%) 
4.Septicemia (5.7%)
5. Congenital abnormalities (8%)
5. Others (34.9%)
6. Others (34%)
 

 

Table III depicts the causes of death in the neonatal and post neonatal period separately. Here the commonest cause of death in neonatal period was birth asphyxia, and PEM was the leading cause of death in the post-neonatal period.

Discussion

This study has attempted to find out whether the verbal autopsy tool can be used by health workers to find out the cause of death in the rural area. The diagnosis was mainly symptom-based according to the information collected by verbal autopsy. One limitation of this study is that the verbal autopsy forms for all the deaths were not available. Only 61.5% were available out of which only 80% could be analyzed due to reasons stated above. Since the cause of death by lay reporting was not different among those whose forms were available and those whose forms were not, we therefore believe that the results of this study are representative of all the under-five deaths in this area. This however highlights the need for regularizing the verbal autopsy method in a routine system. Training and supervision can overcome this.

In our study, we found that prematurity was an important cause of death, similar to the other Indian studies(6,7). However, we found that birth asphyxia was the leading cause of death, which points to the quality of intra- natal care. The different results observed may be due to the different study areas.

We have seen that the verbal autopsy can be used for two purposes. Firstly, it can, help to give information for the health planners to prioritize health services based on the mortality pattern of an area. Secondly, health workers can also get a feedback of the information that they have collected; and not merely a meaningless piece of information that they collect and forget. In other words, the VA can be used for collecting information for action at the local level. In fact this exercise was a part of an overall strategy to reduce IMR in the field practice area. The results of this exercise were also shared with the workers. However, as was noted in the study, there was a problem of not getting the records. This is a usual problem faced in any routine activity. This can be improved only if the health workers realize that finding about the causes of death in their own area is going to help them in providing better health care. We also noted the need for regular training of the workers, emphasizing the need to fill the verbal autopsy forms properly, get accurate information thus making it easier to arrive at a definite cause of death.

Based on the discussion, we are moving ahead in two fronts both aimed at simplification. First, we are trying to shorten the tool. While our forms were mainly devised based on theoretical disease constructs, we are now reviewing it to see which questions are essential to make a diagnosis. Secondly, we are trying to harmonize the questions/symptoms with the Integrated Management of Childhood Illness (IMCI) guidelines. The workers have already been trained in these guidelines, so they will find it easier to understand and ask the questions. Even the diagnostic criteria could be based on these guidelines.

In conclusion, the use of the verbal autopsy tool by health workers to find out the cause of death is feasible. It can also provide information for local action by health authorities to reduce the under-five mortality rate, according to area-specific causes of death.

Contributors: KA and SKK were involved in concept, design, training of health workers for the study. NB acquired the data, analyzed it and drafted the article. The manuscript was critically reviewed by KA and SKK and approved by all.

Funding: None.

Competing interests: None stated.

 

Key Messages


• Health workers can be trained to ascertain the cause of under five death by using the verbal autopsy in the rural areas.

• This provides information for local action by health authorities to reduce under-five mortality rate.

 

 

 References


 

1. Ministry of Health and Family Welfare, Annual Report 1999-2000, Government of India, New Delhi, 2000.

2. Mahapatra P, Chalapati Rao PV. Cause of death reporting systems in India: a performance analysis. Natl Med J India. 2001; 14: 154-162.

3. Bang AT, Bang RA. Diagnosis of causes of childhood deaths in developing countries by verbal autopsy: suggested criteria. Bull WHO 1992; 70: 499-507.

4. Anand K, Kant S, Kumar G, Kapoor SK. Development is not essential to reduce infant mortality rate in India: experience from the Ballabgarh project. J Epidemiol Community Health 2000; 54: 247-253.

5. Longjam UD.A study of causes of death in children under five years using verbal autopsy as a diagnostic tool. (Thesis, MD Community Medicine). All India Institute of Medical Sciences, New Delhi; 1995.

6. Awasthi S, Pande VK. Cause specific mortality in under fives in the urban slums of Lucknow, North India. J Trop Pediatr 1998; 44: 358-362.

7. Srivastava SP, Kumar A, Ojha AK. Verbal autopsy determined causes of neonatal deaths. Indian Pediatr 2001; 38: 1022-1024.

Home

Past Issue

About IP

About IAP

Feedback

Links

 Author Info.

  Subscription