SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References

Objective To estimate maternal mortality in two samples of a population in northern Tanzania.

Setting Rural communities and antenatal clinics, Mbulu and Hanang districts, Arusha region, Tanzania.

Population From a household survey 2043 men and women aged 15–60, and from an antenatal clinic survey 4172 women aged 15–59.

Method The indirect sisterhood method.

Main outcome measures The risk of maternal deaths per 100,000 live births (maternal mortality ratio), and the lifetime risk of a maternal death.

Results The risk of a maternal death per 100,000 live births was 362 (95% CI 269–456) and 444 (95% CI 371–517) for the household and antenatal clinic surveys, respectively. The lifetime risk of maternal death was 1 in 38 and 1 in 31, respectively, for the two surveys. A significantly lower risk of maternal death was observed for the respondents attending antenatal clinics closer to the hospital than for those attending clinics further away: 325 (95% CI 237–413) compared with 561 (95% CI 446–677) per 100,000 live births. Lifetime risk of maternal death was 1 in 42 and 1 in 25, respectively.

Conclusions The risk of maternal death per 100,000 live births in this area were comparatively high, but in our survey substantially lower than in previous surveys in Tanzania. Increasing distance from the antenatal clinics to the hospital was associated with higher maternal mortality. There was no significant difference between results based on household and antenatal clinic data, suggesting that accessible health facility data using the sisterhood method may provide a basis for local assessment of maternal mortality in developing countries.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References

Since the Safe Motherhood Initiative was launched in 19871, the issue of maternal mortality has been given renewed attention. According to the International Classification of Diseases2 (ICD-10) issued by the World Health Organisation, a maternal death is defined as the death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the duration and the site of the pregnancy, from any cause related to or aggravated by the pregnancy or its management, but not from accidental or incidental causes. The level of maternal death may be defined as the number of maternal deaths during a given time period per 100,000 live births in the same period. Ectopic pregnancies, abortions and stillbirths are not included in the denominator. This quantity is often referred to as the maternal mortality ratio. Because the number of pregnancies per woman is high in many developing countries, the lifetime risk of death in connection with a pregnancy is also an important measure. The lifetime risk is the probability of dying of maternal causes cumulated across the reproductive years (35 years).

Worldwide, new approaches for estimating maternal mortality levels have been sought, owing to the lack of sufficient national statistics in many countries. In 1996 a report was published by the World Health Organisation, using new models for estimation based on the former statistics3.

Based on the new figures, the annual number of maternal deaths globally is estimated to be approximately 585,000. In Eastern Africa the estimate is 97,000 death of 1060 per 100,000 live births. In Tanzania the estimate for the risk of maternal death is 770 per 100,000 live births, corresponding to approximately 8700 maternal deaths annually.

In the official Tanzania Ministry of Health report for 1997, the risk of maternal death was estimated to be 200–400 per 100,000 live births4. The Demographic and Health Surveys estimated a risk of maternal death of 529 per 100,000 live births5. A largescale epidemiologi-cal survey in Tanzania, known as The Adult Morbidity and Mortality Project from 19976, has found the risk of maternal death per 100,000 live births to be 306 (Hai), 733 (Dar es Salaam) and 977 (Morogoro rural). A recent study from Bagamoyo district in Tanzania found a risk of maternal death of 961 per 100,000 live births7.

A maternal death is a catastrophe not only for the mother, but also for her children and family, and the society as a whole. Because of the scarcity of information on maternal mortality in the study area, and the discrepancy between the official figures and the results from various other studies, our objective was to assess more accurately the level of maternal mortality in a rural area of northern Tanzania, and to compare the levels found in a household survey with the levels found in a health facility survey. We used the indirect sisterhood method developed by Grahamet al.8. The sisterhood method is an indirect retrospective estimation of maternal mortality by which respondents are asked about their sisters'survival or death in relation to pregnancy. The survey was done as part of a larger survey on reproductive health and maternal mortality in the study area.

METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References

The study was carried out in 1995 and 1996 in the catchment area of Haydom Lutheran Hospital, a church run institution, in a rural area in northern Tanzania, Arusha region. We chose two administrative divisions, Dongob-esh and Basotu, from Mbulu and Hanang districts, respectively. The two divisions had a total of 12 wards comprised of 42 villages. Based on the 1988 census and an annual growth rate of 3.8% for Arusha Region9, the 1995 estimates of the total population in these two divisions and of women aged 15–44 were 143,000 and 28,000, respectively10.

All four main language groups in East Africa are represented in the study area. The Iraqw (Cushitic) and the Datoga (Nilotic) are the two largest groups. They live mainly by subsistence farming and pastoralism.

During the study period there was one hospital, one health centre, 14 dispensaries and 16 stations with monthly mother and child health services within the two divisions. These were both governmental and running 12 of the 32 antenatal clinics in the area.

The indirect sisterhood method for assessment of maternal mortality was applied for two populations: women attending antenatal clinics in the area, and men and women from a household survey.

Eight field assistants (four men and four women) participated in the household survey, and five field assistants (all women) were trained for the work in the antenatal clinics. Six of the field assistants had some type of education beyond standard seven, while seven had standard seven. They were trained in interviewing techniques over a period of several weeks. The questions asked in the interviews were translated into Swahili, and then translated back into English again, in order to confirm the meaning and wording. The field assistants were all fluent in the local tribal languages, which were used during the interview if necessary.

The first group consisted of 4179 women aged 15–59 who were interviewed consecutively at their first antenatal visit at the 12 antenatal clinics of Haydom hospital, from February 1995 until March 1996. Only residents of Basotu and Dongobesh divisions were included in the study. The distances from the hospital to the clinics were 7–45 kilometres by road. One clinic (distance 75 kilometres) was reached by air because of the inaccessible roads.

The second group was part of a larger household survey. Within the two divisions, seven sub-villages from four villages were chosen. Logistic considerations, and the wish to include the two main ethnic groups of the area (Iraqw and Datoga), were the main reasons for choosing these specific sub-villages. All households were visited in these sub-villages. From April 1995 to September 1996 2049 men and women aged 15–60, were interviewed. The distances from Haydom hospital to the sub-villages were 18–31 kilometres in mean cross-sectional distance. Most people travelled by bicycle, ox-cart or on foot, as there was no public transport available in these villages.

In each household, every person living there by right was registered. Using this roster, we marked the eligible respondents (men and women aged 15–60) for the sisterhood interview. Respondents by proxy were accepted if the proxy was a close relative who knew the person well. The interviewers ensured the privacy of the respondent during the interview.

Of the 3090 aged 15–60 in the household survey, 277 were not interviewed since only one of the siblings in each family was eligible; 2049 (72.8%) responded to the questionnaire. Among those who did not respond, one was mentally retarded, two refused, 110 were not present at the time of interview and for 651 the reason was not clear. Of 2049 respondents, six of unknown age were excluded from the survey, leaving 2043 respondents for responses by proxy. Ages were missing for seven respondents in the antenatal clinic survey. Therefore there were 4172 respondents remaining for analysis.

According to the procedure for the indirect sisterhood method (Graham et al.8), we asked the following four basic questions:

  • 1
    How many sisters, born of the same mother, have you ever had, who have reached the age of 15 years or older?
  • 2
    Of those sisters, born of the same mother, who have reached the age of 15 or more, how many are still alive?
  • 3
    Of those sisters, born of the same mother, who have reached the age of 15 or more, how many have died?
  • 4
    Of those sisters who have died, and who are born of the same mother, how many died during pregnancy, delivery or within six weeks after delivery?

The interviewer checked that the sum of questions two and three was equal to the total in question one. Inaccurate reporting of age presented a problem: many did not know their exact age, so misreporting occurred in all age groups. Due to the extended family system, we emphasised that the sisters must be born to the same mother.

In the antenatal survey date of interview, maternal age and name of antenatal clinic were recorded. In the household survey date of interview, sex, age, ethnic affiliation, religion, clan and household identification were registered as background information. This enabled us to link the sisterhood survey to the larger household survey with information on other factors such as literacy and socioeconomic status.

Analyses were done on the antenatal clinic sample and the household sample separately. Within the antenatal clinic sample we stratified on distance from Hay-dom hospital to the clinics. Within the household sample, we stratified by sex, literacy, distance from ownership of land, ownership of cows, ethnic affiliation and villages with and without antenatal clinic services from Haydom hospital.

The main outcome measures were the risk of maternal death per 100,000 live births, and a woman's lifetime risk of dying from maternal causes. Another outcome measure derived from the sisterhood questions was the proportion of sisters dying from a maternal cause in relation to the total number of sisters dying. The adjustment factors referred to in the method were derived from model fertility and mortality distributions in developing countries as defined by Graham et al.8. These factors were used to convert the number of sisters being reported to sister units of risk exposure to maternal deaths.

The total fertility rate reflects the number of live births a woman would bear if she were to experience the existing age-specific fertility rates of a given period. In our study the total fertility rate was calculated from data collected in the household survey (Table 1).

Table 1.  Age specific fertility rates and total fertility rate from a household survey in Hanang and Mbulu districts, Tanzania, 1995–96.
Age groupMidpoint of age groupNo. of births last 12 monthsNo. of womenBirths/woman*Fertility rates per 1000 womenTotal fertility rate
  1. * Number of births last twelve months/ number of women, for each age group.

  2. (Births/woman) × 1000.

  3. Sum of births/ women × 5 (for the 5 years included in an age group).

15–1917.592900.0331.03 
20–2422.5933030.31306.93 
25–2927.51062700.39392.59 
30–3432.5652050.32317.07 
35–3937.5271350.20200.00 
40–4442.5171060.16160.38 
45–4947.57960.0772.92 
TOTAL 32414051.481480.927.40

The study was approved by the National Committee for Research Ethics in Medicine in Norway and the Commission for Science and Technology in Tanzania. Local leaders on all levels gave consent to the study, and the local people were informed through village meetings. Each respondent gave individual consent and was informed about the freedom to answer, the freedom to exit the study at any time, and that continued services would be given in case of exit from the study.

Data analysis

Data entry was done twice for detection and subsequent correction of entry errors, using the Epi-Info version 5.0 and 6.0 programs. SPSS for Windows (version 9.0) was used for statistical analysis. The 95% confidence intervals were calculated for the maternal mortality ratio using the method described by Hanley et al.11.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References

The total fertility rate estimated from the household survey was 7.40 (Table 1). Using this figure in the further calculations, we found the risk of maternal death per 100,000 live births estimated from the household survey and the antenatal clinic survey (ages 15–49) to be 362 (95% CI 269–456) (Table 2), and 444 (95% CI 371–517) (Table 3), respectively. This corresponded to a lifetime risk of dying of maternal causes of 1 in 38 and 1 in 31, respectively. When analysing for ages 15–60, we found the risk of maternal death per 100,000 live births estimated from the household survey and the antenatal clinic survey (ages 15–59) to be 388 (95% CI 302–475) (Table 2), and 445 (95% CI 373–518) (Table 3), respectively. This corresponded to a lifetime risk of dying of maternal causes of 1 in 35 and 1 in 31, respectively. The proportion of sisters dying from maternal causes estimated for ages 15–60 was 22.5% in the household survey and 43.3% in the antenatal clinic survey. The lifetime risk of dying of maternal causes in the household survey was highest in the respondent age group 30–34: the risk was 1 in 26. In the antenatal clinic survey a similar lifetime risk of dying of maternal causes of 1 in 26 was found in both the 20–24 and the 25–29 respondent age groups.

Table 2.  Maternal mortality estimates for Hanang and Mbulu districts, Tanzania, 1995–96, using the sisterhood method in a household survey among men and women 15–60 years. Values are given as n, unless otherwise indicated. Col = column; TFR = total fertility rate for study area; MMR = risk of maternal death per 100,000 live births.
  Col 3 Col 4 Sisters ≥ 15 years     
Col 1 Age group of respondentCol 2 No. of respondentsTotalDeadCol 5 Maternal deathsCol 6 Adjustment factors*Sister-units of exposure (Col 3 × Col 6)Lifetime risk of dying of maternal causes (Col 5/Col 7)Proportion of dead sisters dying of maternal causes (Col 5/Col 4)
  1. *Adjustment factor defined by model fertility and mortality distributions in developing countries as described by Graham et al.8.

  2. After adjustment by a factor (total number of sisters from 25–60)/(total number of respondents 25–60), which gives a factor of 2.73. Original reported number in col.3 age 15–19 was 250.

  3. After adjustment by a factor of 2.73. Original reported number was 937.

  4. §Lifetime risk of maternal death from reports of respondents ages 15–49 = 58/2188 = 0.0265 (1 in 38) Obstetric risk per 100,000 live births (MMR): l–[(l–Lifetime risk)1/TFR] = l–[(l–0.0265)1/7.40] = 0.00362 (362/100,000 live births).

  5. Lifetime risk of maternal death from reports of respondents ages 15–60 = 77/2716 = 0.0284 (1 in 35) Obstetric risk per 100,000 live births (MMR): l–[(l–Lifetime risk)1/TFR] = l–[(l–0.0284)1/7.40] = 0.00388 (388/100,000 live births).

15–19201549610.107590.01690.1667
20–2441111222360.2062310.02600.2609
25–29384108449130.3433720.03500.2653
30–3430486147170.5034330.03930.3617
35–392145974190.6643960.02270.2195
40–441614183540.8023350.01190.1143
45–491524023880.9003620.02210.2105
15–491827503323958 21880.0265§0.2427
50–549824753100.9582370.04230.1887
55–59822204090.9862170.04150.2250
≥ 6036741101.000740.00000.0000
TOTAL2043557434377 27160.02840.2245
Table 3.  Maternal mortality estimates for Hanang and Mbulu districts, Tanzania, 1995–96, using the sisterhood method in an antenatal clinic survey among women 15–59 years. Values are given as n, unless otherwise indicated. Col = column; TFR = total fertility rate for study area; MMR = risk of maternal death per 100,000 live births.
  Col 3 Col 4 Sisters ≥ 15 years     
Col 1 Age group of respondentCol 2 No. of respondentsTotalDeadCol 5 Maternal deathsCol 6 Adjustment factors*Sister-units of exposure (Col 3 × Col 6)Lifetime risk of dying of maternal causes (Col 5/Col 7)Proportion of dead sisters dying of maternal causes (Col 5/Col 4)
  1. * Adjustment factor defined by model fertility and mortality distributions in developing countries as described by Graham et al.8.

  2. After adjustment by a factor found by taking the (total number of sisters from 25–60)/ (total number of respondents 25–60), which gives a factor of 2.98. Original reported number in col.3 age 15–19 was 497.

  3. After adjustment by a factor of 2.98. Original reported number was 3596.

  4. §Lifetime risk of maternal death from reports of respondents ages 15–49 = 142/4388 = 0.0324 (1 in 31) Obstetric risk per 100,000 live births (MMR): l–[(l–Lifetime risk)1/TFR] = 1–[(1–0.0324)1/740] = 0.00444 (444/100,000 live births).

  5. Lifetime risk of maternal death from reports of respondents ages 15–59 = 143/4401 = 0.0325 (1 in 31) Obstetric risk per 100,000 live births (MMR): l–[(l–Lifetime risk)1/TFR] = 1–[(1–0.0325)1/740] = 0.00445 (445/100,000 live births).

15–192808341330.107890.03370.2308
20–241465436677350.2068990.03890.4545
25–291200347384460.34311910.03860.5476
30–34730219382340.50311030.03080.4146
35–39361111347140.6647390.01890.2979
40–441073642290.8022920.03080.4091
45–492683410.900750.01340.2500
15–49416912,426329142 43880.0324§0.4316
50–5425110.95850.20881.0000
55–5918000.98680.00000.0000
TOTAL417212,439330143 44010.03250.4333

In the antenatal clinic survey the risk of a maternal death per 100,000 live births (ages 15–49) for respondents from Haydom hospital was 325 (95% CI 237–413), and the lifetime risk of dying of maternal causes was 1 in 42. For those attending clinics situated more than eight kilometres from the hospital, the risk of maternal death per 100,000 live births was 561 (95% CI 446–677) (lifetime risk 1 in 25) (Table 4).

Table 4.  Lifetime risk of dying of maternal causes, maternal mortality ratios with 95% confidence intervals, and time reference years estimated by the sisterhood method in an antenatal clinic survey and a household survey in Hanang and Mbulu districts, Tanzania, 1995–96. Values are given as n, unless otherwise indicated. ANC = antenatal clinic; TFR = total fertility rate for study area; MMR = risk of maternal death per 100,000 live births.
 No. of respondentsNo. of maternal deathsLifetime risk of maternal deathMMR* (95% CF)Time reference (years)
  1. *MMR was found by the formula8: 1–[(Probability of survival)1/TFR] = 1–[(1–Lifetime risk)1/TFR].

  2. The CI of the MMR was calculated by11: Let Q = lifetime risk, then let B = sister-units of exposure. The CI for Q is: QUpper= Q +{1.96 × sum [(Q × (1-Q)/B]} and QLower= Q−{l.96 × sum [(Q × (1-Q)/B]}. The QUpper and the QLower are consequently applied to the formula of MMR given above to obtain the MMRUpper and the MMRLower.

  3. The time reference period was calculated by using the formula8: T = sumBi/Ti/sumBi where Bi is sister-units of exposure and Ti is the number of years prior to data collection for each age group, using the mid-point of each age group.

Total antenatal clinic survey     
  Age group 15–60 years41721431 in 31445 (373–518)9.4
  Age group 15–49 years41691421 in 31444 (371–517)9.4
Respondents attending     
  ANC < 8 kms from Haydom2117521 in 42325 (237–413)9.2
  ANC > 8 kms from Haydom2052901 in 25561 (446–677)9.5
Total household survey     
  Age group 15–60 years2043771 in 35388 (302–475)17.2
  Age group 15–49 years1827581 in 38362 (269–456)11.4
Household     
  Respondents living < 28 kms from Haydom987281 in 40339 (214–464)11.3
  Respondents living ≥ 28 kms from Haydom840301 in 35387 (249–526)11.4
  Sub-village with ANC services from Haydom976261 in 48286 (177–397)11.7
  Sub-village without ANC services from Haydom851321 in 28496 (325–668)11.3
  Males722281 in 30459 (289–628)11.4
  Females1105301 in 45303 (195–412)11.3
  Ownership of land ≥ 3 acres756211 in 43315 (181–451)11.4
  Ownership of land > 3 acres1062371 in 34401 (273–530)11.4

The risk of maternal death per 100,000 live births and the lifetime risk of dying of maternal causes were estimated from the household survey by stratification by different variables within the ages of 15–49 (Table 4). The difference was most noticeable in the comparison of the sub-villages which had antenatal clinic services from Haydom hospital, and those that did not have these services; the risk of maternal death per 100,000 live births being 286 (95% CI 177–397) compared with 496 (95% CI 325–668), respectively. The lifetime risk was 1 in 48 for those with antenatal clinic services and 1 in 28 for those who did not have these services. There was a difference, although not significant, between male and female respondents, the lifetime risk being 1 in 30 based on male respondents and 1 in 45 based on female respondents. The risk of maternal death per 100,000 live births was 459 (95% CI 289–628) based on male respondents as compared with 303 (95% CI 195–412) based on female respondents. The results for the stratification by distance from the sub-village to Haydom hospital and ownership of land did not show any significant differences in the risks (Table 4). Literacy, ownership of cows and ethnic affiliation did not show any significant difference either (data not shown). The values of the time reference years T vary between 9.2 to 17.2 years for the different analyses, reflecting the retrospective nature of the method (Table 4).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References

The total fertility rate of 7.40 found in our household survey was considerably higher than the total fertility rate found in the Demographic and Health Survey in Tanzania in 1996, which was 6.34 for women aged 15–49 in rural areas5, and also higher than the official total fertility rate of 6.5 from the Ministry of Health based on the 1988 population census4. The population in the study area is relatively stable, and adheres to a way of living that has remained largely unchanged over many years. Thus, for the study area, we think that the total fertility rate measured during this survey may be representative for the whole time period reflected in the sisterhood survey.

The risk of maternal death per 100,000 live births of 362 (household survey) and 444 (antenatal clinic survey) in our study is somewhat higher than in a study using the sisterhood method in a rural area in Kwimba district in north-western Tanzania12,13. This study, showed a risk of maternal death per 100,000 live births of 288 (based on Kazaura, ages 15–4912) or 297 (based on Walraven, ages 15–60+13). However, our study shows lower figures than most direct estimates of maternal mortality in Tanzania, which have showed the risk of maternal death per 100,000 live births in the ranges from 306 (Hai district) to 977 (Morogoro rural)6, and 961 per 100,000 live births in Bagamoyo district7. The prevalence of HIV among women attending the antenatal clinics of Haydom hospital was very low, being 2/733 (0.3%) in 1996 (unpublished hospital data) and 2/467 (0.4%)14 in 1998. Among blood donors at Haydom hospital the prevalence was 14/694 (2.0%) in 199814. The hospital also has emergency obstetric services and a 24-hour ambulance service which gives priority to delivering women. These different factors may be possible explanations for the lower levels of maternal mortality seen in the study area as compared with most other estimates from Tanzania.

The sisterhood method may overestimate maternal mortality since it may register all pregnancy-related deaths, including those due to accidental or incidental causes. However, it may also underestimate the level because early pregnancies, abortions and ectopic pregnancies are not likely to be registered15,16.

Other studies using the sisterhood method in rural areas in Sub-Saharan Africa show the risk of maternal death per 100,000 live births and the lifetime risk of dying of maternal causes ranging from 316 (1 in 57) in Zimbabwe to 1549 (1 in 12) in Zambia17–20. The results of our study are comparable to those from Zimbabwe17 and Mozambique (risk of maternal death per 100,000 live births of 410; lifetime risk of dying of maternal causes, 1 in 38)18, but lower than the levels found in Kenya (risk of maternal death per 100,000 live births of 599; lifetime risk of dying of maternal causes, 1 in 30)19 and in Zambia20.

The proportion of sisters dying of maternal causes estimated in the household survey (22.5%) was similar to the 25 to 33% ratio estimated by Royston and Lopez21 for developing countries, and similar to the result found in the Demographic and Health Survey for Tanzania (27.4%)5. However, the proportion of sisters dying of maternal causes in the antenatal clinic survey (43.3%) was considerably higher than their estimates. It was closer to that found in Kenya (48%) among a pastoral population19. This may indicate that mortality for other nonmaternal causes was relatively low in this group.

There was no significant difference between the risk of maternal death per 100,000 live births and the lifetime risk of maternal death found in the antenatal clinic survey and those found in the household survey. This has also been shown in a study from Nicaragua using the sisterhood method. In that study, a comparison between a household based survey and a health facility based survey showed virtually identical results for the risk of maternal death per 100,000 live births and the lifetime risk of dying of maternal causes22.

The significant difference in the risk of maternal death per 100,000 live births seen when stratifying by distance to the hospital in the antenatal clinic survey, reflects how distance to the health facility and transport are important factors when dealing with maternal deaths. Even though the results reflect the respondents' place of living, and not that of the sisters who actually died, they may indicate the importance of distance. The observation that obstetric coverage declines with increasing distance, and that a greater distance and a lack of transport may contribute to an increase of maternal complications and deaths, has been observed in many studies from similar settings in sub-Saharan Africa23–26.

In the household study, respondents from the sub-villages with antenatal services from Haydom hospital showed a lower risk of maternal death per 100,000 live births and a lower lifetime risk of dying of maternal causes than respondents who did not have access to these services in their sub-villages. However, since the conclusive. The time period reflected in these analyses was approximately 11 years (Table 4), thus it is not only the present antenatal situation that is relevant. Although recent reports have indicated the lack of effect of antenatal services in reducing maternal mortality27,28, our study suggested that such services may make a difference. A study from India by Bloom et al.29 showed that there was a positive association between the level of care obtained during pregnancy and the use of safe delivery care. Thus, our study may suggest that access to antenatal care, in an area where emergency obstetric services are available, may be associated with a reduction of maternal mortality.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References

In summary, we have demonstrated a risk of maternal death per 100,000 live births of 362 (95% CI 269–456, household) and 444 (95% CI 371–517, antenatal). This is lower than most other estimates from various studies in Tanzania4–7, but higher than the one previous study using the indirect sisterhood method in Mwanza12,13. We also showed how the distance from the antenatal clinic to the hospital affected the maternal mortality levels.

Our study showed that when using the sisterhood method, there was no significant difference between results from a household survey and from an antenatal clinic survey, thereby giving a reliable estimate of the level of maternal mortality in the study area. By including the indirect sisterhood method in the antenatal clinic routines, it may be possible to obtain a rapid assessment of maternal mortality levels at a low cost. Even though the sisterhood method is a retrospective method, by using the age-group of 20–25 years it may also be possible to establish a sentinel surveillance system based on data from attendees at the antenatal clinics. Such a system could be used to monitor secular change in maternal mortality, and also the impact of preventive efforts and health system changes on maternal mortality in different areas and countries.

Acknowledgements

We are grateful to the Norwegian Research Council and The Centre for International Health at the University of Bergen, Norway, for funding. We thank the Regional Development Director and Regional Medical Officer (Arusha Region) and the District Commissioners, District Medical Officers, Maternal and Child Health co-ordinators and local division, ward and village administrators in Hanang and Mbulu districts for their assistance. We also gratefully acknowledge the invaluable assistance of the doctor in charge at Haydom Lutheran Hospital, Ole Halgrim Evjen Olsen, and of Matron Selina Sanka and the staff of the hospital, especially the staff at the Maternal and Child Health clinics. We are indebted to the field assistants for their commitment and hard work. Also, we thank the Diocese of Mbulu, of the Evangelical Lutheran Church of Tanzania, for their support. Most of all, we thank the respondents for their willingness to participate in this study.

References

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. References
  • 1
    Mahler H. The safe motherhood initiative: a call to action. Lancet 1987; 1: 668670.
  • 2
    World Health Organization. ICD-10: International Statistical Classification of Diseases and Health Related Problems. Vol 1. (10th revision). Geneva : WHO, 1992.
  • 3
    World Health Organization UNICEF. Revised 1990 Estimates of Maternal Mortality. A New Approach by WHO, UNICEF. Report No. WHO/ FRH/ MSM/ 96.11, UNICEF/ PLN/ 96.1. Geneva : WHO, UNICEF, 1996.
  • 4
    Ministry of Health of United Republic of Tanzania. Health Statistics Abstract 1997. Dar es Salaam : Health Information, Research and Statistics Section, Planning Department, Ministry of Health, 1997.
  • 5
    Tanzania Bureau of Statistics and Macro International Inc. Tanzania Demographic and Health Survey 1996. Dar es Salaam , Tanzania , and Calverton , Maryland : Bureau of Statistics, Planning Commission, Macro International, 1997.
  • 6
    Ministry of Health of United Republic of Tanzania. Policy Implications of Adult Morbidity and Mortality. End of Phase 1 Report. Dar es Salaam: Ministry of Health, 1997.
  • 7
    Macleod J, Rhode R. Retrospective follow-up of maternal deaths and their associated risk factors in a rural district of Tanzania. Trop Med Int Health 1998; 3: 130137.
  • 8
    Graham W, Brass W, Snow RW. Estimating maternal mortality: the sisterhood method. Stud Fam Plann 1989; 20: 125135.
  • 9
    Tanzania Bureau of Statistics. Tanzania Sensa 1988–1988 Population Census: Preliminary Report. Dar es Salaam : Bureau of Statistics, Ministry of Finance, Economic Affairs and Planning, 1988.
  • 10
    Tanzania Bureau of Statistics. Tanzania Census 1988 Population Census regional profile—Arusha. Dar es Salaam : President's Office— Planning Commission, Bureau of Statistics, 1991.
  • 11
    Hanley JA, Hagen CA, Shiferaw T. Confidence intervals and sample-size calculations for the sisterhood method of estimating maternal mortality. Stud Fam Plann 1996; 27: 221227.
  • 12
    Kazaura RM. Maternal mortality estimation by sisterhood method [dissertation]. London : London School of Hygiene and Tropical Medicine, 1991.
  • 13
    Walraven GEL, Mkanje RJB, van Roosmalen J, van Dongen PWJ, Dolmans WMV. Assessment of maternal mortality in Tanzania. Br J Obstet Gynaecol 1994; 101: 414417.
  • 14
    Haydom Lutheran Hospital. Annual Report 1998. Haydom , Mbulu district , Tanzania , 1998.
  • 15
    AbouZahr C, Wardlaw T, Stanton C, Hill K. Maternal mortality. World Health Stat Q 1996; 49: 7787.
  • 16
    Shahidullah M. The sisterhood method of estimating maternal mortality: the Matlab experience. Stud Fam Plann 1995; 26: 101106.
  • 17
    Oosterhuis JWA. Estimating maternal mortality by sisterhood method in rural Zimbabwe. Trop Doc 1993; 23: 6768.
  • 18
    Cutts FT, Dos Santos C, Novoa A, David P, Macassa G, Soares AC. Child and maternal mortality during a period of conflict in Beira City, Mozambique. Int J Epidemiol 1996; 25: 349356.
  • 19
    Mace R, Sear R. Maternal mortality in a Kenyan pastoralist population. Int J Gynecol Obstet 1996; 54: 137141.
  • 20
    Le Bacq F, Rietsema A. High maternal mortality levels and additional risk from poor accessibility in two districts of Northern Province, Zambia. Int J Epidemiol 1997; 26: 357363.
  • 21
    Royston E, Lopez AD. On the assessment of maternal mortality. World Health Stat Q 1987; 40: 214224.
  • 22
    Danel I, Graham W, Stupp P, Castillo P. Applying the sisterhood method for estimating maternal mortality to a health facility-based sample: a comparison with results from a household-based sample. Int J Epidemiol 1996; 25: 10171022.
  • 23
    Urrio TF. Maternal deaths at Songea Regional Hospital, southern Tanzania. East Afr Med J 1991; 68: 8186.
  • 24
    de Groot ANJA, Slort W, van Roosmalen J. Assessment of the risk approach to maternity care in a district hospital in rural Tanzania. Int J Gynecol Obstet 1993; 40: 3337.
  • 25
    Urassa E, Massawe S, Lindmark G, Nystrøm L. Operational factors affecting maternal mortality in Tanzania. Health Pol Plann 1997; 12: 5057.
  • 26
    Jahn A, Kowalewski M, Kimatta SS. Obstetric care in southern Tanzania: does it reach those in need Trap Med Int Health 1998; 3: 926932.
  • 27
    Greenwood AM, Greenwood BM, Bradley AK et al. A prospective survey of the outcome of pregnancy in a rural area of the Gambia. Bull World Health Organ 1987; 65: 635643.
  • 28
    Moller B, Lushino O, Meirik O, Gebre-Medhin M, Lindmark G. A study of antenatal care at village level in rural Tanzania. Int J Gynecol Obstet 1989; 30: 123131.
  • 29
    Bloom SS, Lippeveld T, Wypij D. Does antenatal care make a difference to safe delivery? A study in urban Uttar Pradesh, India. Health Pol Plann 1999; 14: 3848.