© International Epidemiological Association 2000
International Journal of Epidemiology 2000;29:542–548
Printed in Great Britain
Community-based prevention of perinatal deaths: lessons from nineteenth-century Sweden T Andersson,a,b U Högberga,b and S Bergströmc
Background Perinatal deaths have been more difficult to prevent than infant deaths in lowincome countries due to its close relation to poor maternal outcome. The aim of the study was to perform a comprehensive population-based analysis of perinatal mortality in a high mortality setting and to determine the impact of midwiferyassisted home deliveries. Method
The study design was a community-based cohort study. In all, 4876 perinatal deaths were recorded among 116 211 newborns in the districts of Sundsvall and Skellefteå in northern Sweden during the years 1831–1899. Relative risks, 95% CI, population attributable proportions and prevented fractions were calculated.
Results
The overall perinatal mortality rate was 42.0 per 1000 births. A previous stillbirth represented one of the most important risk factors (RR = 3.25, 95% CI : 2.97–3.56), with a population attributable proportion of 7%. Two or more previous stillbirths gave an RR of 8.50 (95% CI : 7.58–9.53) and a population attributable proportion of 4%. There was an increased risk of perinatal mortality for mothers over 35 years old, the primiparous and the unmarried, while grandparous women had a higher perinatal mortality that was accounted for completely by a poor history of previous stillbirths and infant deaths among these women. The children of crofters, farmers and workers had higher perinatal mortality, but area had no significant impact. During the years 1881–1890 and 1891–1899, the prevented fractions of midwifery were 15% and 32%, respectively.
Conclusion
Poor reproductive history, particularly previously high perinatal mortality, is associated with high perinatal mortality. Midwifery-assisted at home deliveries successfully reduced perinatal mortality.
Keywords
Nineteenth century, developing countries, epidemiology, history of medicine, perinatal mortality, preventive health services, reproductive medicine
Accepted
8 November 1999
Each year, about eight million perinatal deaths occur, 98% of them in developing countries.1 Perinatal mortality has been more difficult to prevent than infant mortality and has only recently received global attention. Because it is closely linked to maternal outcomes, perinatal mortality can be used as a proxy indicator of maternal mortality and maternal health care status. There are an estimated ten perinatal deaths for each maternal death.2 At least three-quarters of the perinatal deaths that occur a Department of Public Health and Clinical Medicine, Epidemiology, Umeå University, Umeå, Sweden. b Department of Clinical Sciences, Obstetrics and Gynecology, Umeå University, Umeå, Sweden. c Division of International Health (IHCAR), Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden. Reprint requests to: Tobias Andersson, Department of Public Health and Clinical Medicine, Epidemiology, Umeå University, 901 85 Umeå, Sweden. E-mail:
[email protected]
in developing countries are caused by problems that also kill women: obstructed labour, eclampsia, sepsis, and the woman’s nutritional and infection status.3 The majority of perinatal deaths are thus preventable.4 Data regarding perinatal mortality in developing countries derive mainly from hospital-based studies. This fact hampers an assessment of the causes of poor child outcomes because the vast majority of births in rural areas in low-income countries occur at home, attended by relatives or traditional birth attendants.5,6 However, the information on perinatal mortality coming from today’s impoverished countries can be supplemented by examining historical cohorts from periods of high mortality in currently affluent countries. Sweden is one of the few countries offering reliable historical vital statistics for population studies. In developing countries, it may be difficult to refer ‘at risk’ neonates to hospital, and birthing women may have to resort to
542
COMMUNITY-BASED PREVENTION OF PERINATAL DEATHS
domiciliary care, intervention through primary health care and female community health workers.7 The community-based study in Matlab, Bangladesh, showed a perinatal mortality of 75 per 1000. A controlled study found that family planning and maternal health service programmes significantly reduced perinatal mortality, offering support for the inclusion of a strong maternity care component in primary health care strategies.8 In Sweden, in the nineteenth century, the authorities also campaigned for obstetric competence among primary health care doctors and for the assistance of licensed midwives at home deliveries. The competence of the midwives was improved in 1829 when they were licensed to use forceps, sharp hooks and perforators, and in 1881, when by law (The Antiseptic Decree), antiseptic techniques were instituted at midwifery-assisted home deliveries, thus preventing septic as well as non-septic maternal deaths in rural areas.9 The aim of this study was to analyse perinatal outcomes in nineteenth-century Sweden and to evaluate the impact of a community-based intervention.
Subjects and Methods In the seventeenth century, the Swedish clergy created an information system that covered all individuals in their parishes above the age of 6 or 7 years. By the middle of the eighteenth century this registration system covered the entire population. The information system was based upon the catechetical examination registers and was revised annually. Other types of records were also linked to these parish records: in and out migrations, births, baptisms, banns, marriages, deaths, and burials. The Tabellverket (Office of the Registrar General) was founded in 1749 and compiled national statistics. In 1858, the Tabellverket became the Central Bureau of Statistics. Demographic and social data from a large number of parishes in Sweden have been computerized and are available for research at the Demographic Data Base of Umeå University. These statistics have been judged to be reliable.10 During a fire in Sundsvall town in 1888 a number of church books were lost and because of this the reporting of births and deaths in this part of the study area is missing up to 1860. This database was used to examine all live births and stillbirths from 18 parishes in northern Sweden in the years 1831–1899. Altogether, 116 211 births of 32 184 parturients were recorded during the period. The following data were obtained for this study: the mother’s date of birth and marital status, the birth order of the child, each child’s date of birth, previous reproductive outcome (such as previous stillbirths), previous infant deaths (0–11 months), the occupation of the father (foundry proprietor, academic, military, employee, civil servant, farmer, worker, crofter, etc.) domicile, and presence of midwife at birth. Information on midwife’s assistance was not available for all parishes until the second half of the century. Stillbirths were registered in the database. Perinatal mortality was defined as the total number of stillbirths and deaths within 7 days of birth per 1000 live and stillbirths. Infant mortality was defined as all deaths within the first year of life per 1000 live births. Perinatal and infant deaths were calculated using the data of birth and date of death registrations in the database. The determinants for risk factor analysis were categorized as follows: maternal age (,35/.34); parity (I,II–IV/V+); previous
543
stillbirth or not; previous infant death or not; married or unmarried; father’s profession as white-collar worker, farmer, craftsmen, crofter and worker; domicile (the town of Sundsvall, the agricultural parish of Skellefteå, the agricultural parishes of the Sundsvall region [Ljustorp, Hässjö, Tynderö, Sättra, Indal, Tuna, Attmar and Selånger], the old iron foundry communities [Galström, Lögdö, Lagfors], and the new booming industrial sawmill communities [Alnö, Skön, Timrå, Svartvik, Njurunda]). Information on parental education was not considered valid for analysis.11 For the risk factor assessment, surviving children with exposure to risk were compared with the number of surviving children within the group at least risk by relative risks (RR) with 95% CI (Epi-Info 6.04B, 1997) Population attributable proportion (= p(RR – 1)/[1 + p(RR – 1)], where p = the proportion exposed in the population) was calculated when appropriate. Interactions between age, parity and previous reproductive history (stillbirths and/or infant deaths) were analysed by stratification and logistic regression. The impact of intervention was analysed by stratifying background factors for midwifery-assisted home deliveries and other deliveries. The final model for analysis was stratified by time period and the impact of midwifery was adjusted for marital status, reproductive history, age, parity and area. Adjusted rates in the final model were calculated by multiple logistic regression (SPSS for Windows 7.0, 1997). The prevented fraction is defined as the quotient of prevented cases (cases that would have occurred in the absence of exposure minus the cases that did occur) and the cases that did occur.
Results During the years 1831–1899, there were 116 211 newborns in the area of Skellefteå and Sundsvall. Of these, 2662 were stillbirths and 2214 children died within the first 7 days. The perinatal mortality rate (PMR) was 42.0 per 1000 births. Mortality rates were relatively stable during the period 1830–1850. They increased during the 1860s and 1870s, and then decreased again (Figure 1). Due to underreporting and the church books lost in the 1888 fire in Sundsvall town, the PMR varied slightly (Figure 1). As shown in Table 1, advanced maternal age, primiparity and grandmultiparity represented a moderate risk increase. Triplets (RR = 10.55, 95% CI : 7.59–14.66), twins (RR = 3.74, 95% CI : 3.41–4.09), one previous stillbirth (RR = 3.25, 95% CI : 2.97–3.56) and two previous stillbirths (RR = 8.50, 95% CI : 7.58–9.53) had the highest RR. Previous stillbirths, previous infant death and twin births had population attributable proportions between 4% and 11%. Birth out of wedlock represented a certain increase in risk (RR = 1.29, 95% CI : 1.20–1.43), but had a very small population attributable proportion, 3%. Crofters and workers had a moderate but significant risk increase, with a population attributable proportion of 12%. The new sawmill area and the urban and farming areas tended to have an increased risk of perinatal death compared to the old iron-industrial community, although the difference was not significant. Combining age and parity strengthened the risk pattern for primiparity and grandmultiparity, especially in combination with advanced age. Combining previous stillbirth and infant
544
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
Figure 1 Stillbirth rate and perinatal mortality rate per 1000 births, 1831–1899, in Sundsvall and Skellefteå and in the total study area except Sundsvall town
death revealed that the main group at risk were those with two or more previous stillbirths, and that the risk was further strengthened if there had also been a previous infant death. Age, parity and reproductive history interacted. Once reproductive history was taken into account, grandmultiparity lost its relationship to perinatal death, age ,35 and parity V+ (RR = 0.98, 95% CI : 0.87–1.10), age .35 and parity V+ (RR = 1.21, 95% CI : 1.11–1.32), while advanced age was enhanced as a risk factor. Further, the importance of bad reproductive history was enhanced when combined with age and parity (Table 2). The midwifery system was successively implemented during the period studied, with 43.7% of home deliveries being midwife assisted in 1871–1880, increasing to 73.4% during the last decade of the century. Mothers in the town and new sawmill area evidently had better access to midwives than mothers in the farming parishes, 73.6% and 50.8%, respectively, giving a significant decrease in risk of perinatal death in the town and sawmill area (RR = 0.75, 95% CI : 0.66–0.84) and in the farming areas (RR = 0.79, 95% CI : 0.72–0.87). White-collar workers summoned the midwife most frequently, to 77.0% of deliveries, while farmers only summoned her for 60.4% of deliveries. The unmarried had less access and least benefit, (RR = 0.99, 95% CI : 0.82–1.21). In conjunction with the antiseptic decree that was implemented in 1881, midwifery-assisted home delivery had a progressively protective affect on the risk of perinatal death during the last two decades of the study (Table 3). This tendency was enhanced when the model was adjusted for marital status, age and parity, reproductive history, area and father’s occupation for the period 1881–1890
(RR = 0.86, 95% CI : 0.75–0.98) and the period 1881–1899 (RR = 0.71, 95% CI : 0.62–0.82), respectively. Midwifery-assisted home deliveries represented an increased risk for perinatal death during the years 1871–1880, although the adjusted RR increase was not significant (RR = 1.07, 95% CI : 0.90–1.27) (Table 4). The preventive fractions of the intervention for the decades 1881–1890 and 1891–1899 were 15% and 32%, respectively, among those exposed to risk, and 6% and 21%, respectively, among the population (Table 5).
Discussion The most important conclusion from this study is that intervention through midwifery-assisted home deliveries, in combination with antiseptic techniques, were effective in preventing perinatal deaths, with a preventive fraction of 32% during the last decade of the century. The other important finding is that poor obstetric history, particularly at least two previous stillbirths, is directly associated with future poor reproductive outcome. This further underlines the close relationship of perinatal death and poor maternal outcome, supporting perinatal mortality as a proxy indicator of maternal mortality. Class (father’s occupation) and births out of wedlock seem to have been the strongest socioeconomic risk factors for death, while domicile (farming, booming sawmill and urban area) had a non-significant risk increase. Underreporting of vital statistics, stillbirths, and early neonatal deaths is common in low-income countries because of home deliveries and early discharge from puerperal care. The
COMMUNITY-BASED PREVENTION OF PERINATAL DEATHS
545
Table 1 Perinatal death by age, parity, reproductive history, multiple birth, domicile, profession and marital status in Sundsvall and Skellefteå, Sweden, 1831–1899, perinatal mortality, relative risks (95% CI) and population attributable proportion
95% CI
Population attributable proportion %
1.32–1.48
10
1.43
1.33–1.53
10
1.29
1.21–1.39
7
3.25
2.97–3.56
7
8.50
7.58–9.53
4
1.48–1.66
11
3.74
3.41–4.09
7
10.55
7.59–14.66
0
All births
Perinatal deaths
Perinatal mortality
,35
82 264
3126
40.0
1
.34
33 535
1747
52.1
1.40
2–4
54 281
1928
33.5
1
5+
29 460
1472
50.0
1
32 470
1476
46.6
0
111 821
4207
39.1
1
1
3721
455
122.3
2+
669
214
319.9
0
91 340
3418
37.4
1
1+
24 871
1458
58.6
1.57
113 171
4419
39.0
1
2989
436
145.9
51
21
411.8
Relative risk
Age (years)
Parity
Previous stillbirth
Previous infant death
Birth Single Twin Triplet Area Iron foundry
1606
55
34.2
1
Skellefteå
38 261
1672
43.7
1.29
0.97–1.71
9
Sundsvall farming
32 327
1391
43.0
1.21
0.96–1.64
6
Sawmill
30 686
1246
40.6
1.19
0.91–1.55
5
Sundsvall town
13 331
512
38.4
1.13
0.84–1.51
1
Profession White-collar workers
4399
161
36.6
1
Farmers, craftsmen
40 872
1828
44.7
1.22
1.04–1.43
8
Crofters, workers
54 110
2468
45.6
1.25
1.07–1.46
12
Married
94 762
4004
42.2
1
Unmarried
10 615
580
54.6
1.29
1.20–1.43
3
Marital status
results of this study could contribute to gaining a comparable perspective as well as to identifying perinatal risks in high mortality societies. The perinatal mortality of 42.0 in the current study was in the same range as that found in the Matlab survey.12 In the latter study, previous stillbirths represented the most important risk factor for perinatal mortality. Even in high-income countries, obstetric history is of importance in perinatal death.13 The Matlab study found no increase in perinatal death with high parity when age was taken into account.12 Our study showed a risk increase for multiparity, even when stratified by age, while a gradient by increasing age was evident. However, the important finding of our study in this respect was that multiparity as a risk factor was confounded by bad reproductive history. Previous stillbirths were associated with the highest risk for subsequent perinatal death. This finding is consistent with recent findings from Malawi, where previous fetal or neonatal deaths had a population attributable risk of 14%.14 The conclusion should therefore be that multiparity per se is not a risk factor for perinatal death.
The causes of death were not available for the study. However, from 1870 to 1875, 46 of 213 infant deaths in Linköping were due to causes closely related to birth, such as prematurity, convulsions, congenital syphilis, weakness, malformations and apoplectic stroke, whereas 136 deaths were caused by infection.15 The chances of perinatal survival improve with socioeconomic status, as measured by such indicators as parental education and parental occupation.16–18 A relatively low-risk gradient by parental socioeconomic level was observed. This corresponds to the study in rural Bangladesh, where no measures of socioeconomic status could be related to perinatal mortality.12 In our study perinatal mortality was not analysed in relation to parental education. However, reading ability and comprehension have been included in the basic parish education in Sweden since the seventeenth century and marriage was not allowed without an approved certificate of reading ability and preparation for confirmation. The rise in the perinatal mortality rates during the 1860s reflects an increase in general mortality, especially infections,
546
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
Table 2 Perinatal deaths stratified by age and parity, and previous stillbirth and infant death in Sundsvall and Skellefteå, Sweden, 1831–1899. Univariate and multivariate relative risk (95% CI) in a logistic regression model All births
Perinatal deaths
Perinatal mortality
Crude relative risk
95% CI
Adjusted relative risk
95% CI
Age ,35 and parity II–IV
42 841
1383
32.3
1
Age .34 and parity II–IV
11 406
544
47.7
1.50
Age ,35 and parity V+
10 537
456
1.36–1.66
1.47
1.33–1.63
43.3
1.36
1.22–1.51
0.98
Age ,35 and parity I
29 246
0.87–1.10
1287
44.0
1.37
1.28–1.49
1.77
Age .34 parity V+
1.63–1.93
18 911
1014
53.6
1.69
1.56–1.85
1.21
1.11–1.32
3218
189
58.7
1.87
1.60–2.19
2.40
2.05–2.82
Age and parity
Age .34 and parity I Reproductive history No previous stillbirth, no earlier infant death
88 718
3045
34.3
1
Previous stillbirth, no previous infant death
2272
263
115.8
3.37
2.99–3.80
4.50
3.91–5.16
No previous stillbirth, previous infant death
23 103
1162
50.3
1.47
1.37–1.57
1.83
1.69–1.99
Previous stillbirth, previous infant death
1449
192
132.5
3.86
3.37–4.42
5.17
4.38–6.09
Previous stillbirth (2+), no previous infant death
350
110
314.3
9.16
7.81–10.73
15.74
12.46–19.88
Previous stillbirths (2+), previous infant death
319
104
326.0
9.50
8.08–11.16
16.53
12.93–21.13
during these decades, which was partly due to food shortages. Both perinatal mortality and maternal mortality increased in parallel with the general mortality. Epidemics of infectious diseases such as scarlatina and diphtheria were on the increase and contributed to the infant and under 5 mortality in Sundsvall during the 1860s and 1870s.19 The last peak of smallpox epidemics also occurred during this period.20 This study shows that, after 1881, a developed midwifery organization, with licensed midwives assisting at home deliveries, proved to be a successful intervention with a considerable preventive fraction of 15–32% among those exposed to risk. The increased risk before 1881 should be interpreted in the light of the dangers of intervention without knowledge of asepsis. Success was the product of a two-century effort towards improved perinatal care. Johan von Hoorn, the founder of the first midwifery school in Sweden, had stated in 1711 that ‘Of 100 stillbirths, 80 could have been prevented if a competent midwife had been with them. My heart sheds tears of blood every time these innocent souls are lost in death.’ From this point on, the objective was to have home deliveries assisted by a licensed midwife. In the middle of the nineteenth century, the authorities added additional regulations for midwives. It was decided that their duties should not be limited only to childbirth, but should also comprise subsequent care of the infant. Further, the education in basic neonatal care at the midwifery school was improved with an emphasis on warmth, neonatal resuscitation with tactile stimulus for asphyctic children, daily care of the umbilicus, and early breastfeeding.21 Doctors and midwives in nineteenthcentury Sweden campaigned for breastfeeding in areas where cows’ milk feeding was common, and succeeded in reducing infant mortality by 20%.22 Swedish midwifery was firmly established in the rural areas, since midwives were explicitly recruited from the families of farmers. Thus, this intervention
overcame technical constraints and demonstrated good social representation, enabling a successful implementation of obstetric techniques within the specific cultural context.23 By the turn of the nineteenth century, two out of every three deliveries was attended by a licensed midwife (Table 3). Perinatal mortality can be seen as a proxy for maternal health. Effective antenatal health care not only reduces maternal morbidity and mortality, but also saves children at birth. Up to now, community-based data on interventions have not been available, but hospital-based intervention has been successful in reducing perinatal mortality. For example, a Zimbabwean (30.6 per 1000) programme for reducing perinatal mortality included: (1) an educational programme to convince all pregnant women to attend antenatal care at least once, (2) closer monitoring of fetal conditions during labour, and (3) skilful management of dystocia.24 Abnormal presentations, which comprised 4.5% of deliveries in a teaching hospital in Nigeria, has a perinatal mortality of 16–18%. Close and thorough supervision of the patients during the antenatal period and labour to detect abnormal presentations could reduce perinatal mortality.24 A medical audit of perinatal deaths could be successful in improving clinical judgements in maternity care and in the utilization of hospital services, thus reducing perinatal mortality, especially intrapartum fetal mortality.24 Reduced access to obstetric care, however, increases perinatal mortality.24 One way to overcome the logistics of providing this could be maternity waiting shelters, which might improve perinatal outcomes.24 The success of the Swedish intervention should be viewed in the perspective of a well-organized health care system, from national level to parish level. The intervention cannot be transferred directly to developing countries that lack health care resources or in which the majority of the births are attended by traditional, rather than trained, birth attendants. Nevertheless,
COMMUNITY-BASED PREVENTION OF PERINATAL DEATHS
547
Table 3 The effect on perinatal mortality of midwife assistance by marital status, maternal age, parity, reproductive history, profession and domicile in Sundsvall and Skellefteå, during the years 1871–1899, relative risk (95% CI) Births
Midwife attended %
Deaths, non-attended
Deaths, attended
Relative risk
95% CI
64 720
61.6
1333
1589
0.74
0.69–0.80
54 088
63.9
1061
1384
0.74
0.68–0.80
6772
53.5
173
198
0.99
0.82–1.21
Age ,35
46 923
62.7
876
1044
0.71
0.65–0.77
Age ù35
17 772
58.6
456
544
0.84
0.75–0.95
Parity I
30 141
61.7
531
622
0.73
0.65–0.81
Parity II–IV
16 105
57.9
430
459
0.78
0.68–0.88
Parity V+
18 474
64.6
372
508
0.75
0.66–0.85
71–80
25 935
43.7
692
471
0.88
0.78–0.98
81–90
32 966
73.5
411
847
0.74
0.66–0.83
91–99
5819
73.4
230
271
0.43
0.36–0.50
Farmers
20 855
60.4
496
559
0.74
0.66–0.83
Crofters/workers
35 079
63.5
689
897
0.75
0.68–0.83
2744
77.0
27
79
0.87
0.57–1.34
Urban or industrial area
30 775
73.6
391
816
0.75
0.66–0.84
Farming area
33 266
50.8
929
763
0.79
0.72–0.87
0
62 221
61.5
1169
1385
0.74
0.69–0.80
1
2108
63.3
109
142
0.75
0.60–0.95
2+
391
60.4
55
62
0.74
0.55–1.00
0
50 959
61.7
927
1126
0.75
0.69–0.82
1+
13 761
61.1
406
463
0.73
0.64–0.83
Midwife assistance Marital status Married Unmarried Age (years)
Parity
Time period
Profession
White-collar workers Domicile
Previous stillbirth
Previous infant death
Table 4 The impact of midwifery-assisted home delivery in Sundsvall and Skellefteå, during the years 1871–1899, relative risk adjusted for marital status, reproductive history, age, parity and domicile (95% CI) 1871–1880
1881–1890
1891–1899
Relative risk
95% CI
Relative risk
95% CI
Relative risk
95% CI
1.07
0.90–1.27
0.86
0.75–0.98
0.71
0.62–0.82
Midwifery-assisted home deliveries
Table 5 Prevented fraction of midwifery-assisted home deliveries among exposed and in the population, Sundsvall and Skellefteå, 1881–1899, perinatal death per 1000 births Time period
Death rate among exposed
Death rate among unexposed
Cases prevented
Prevented fraction among exposed
1881–1890
41.8
1891–1899
35.2
48.6
61
15%
6%
46.8
244
32%
21%
we believe that the Swedish experience is of interest, especially in terms of what could be achieved by community-based midwifery services. The critical points are: (1) the quality of care at home deliveries, (2) recognition of the need for referral, (3) access to transport, and (4) the quality of care in rural hospitals. Traditional birth attendants can recognize birth asphyxia, and with suitable training they should be able to deal with it.24 The Swedish experience from the late nineteenth century testifies to the importance of the quality of care at the primary level
Prevented fraction in the population
and community-based intervention in successfully reducing perinatal mortality. In this sense, perinatal survival is a suitable index of health.24 Early neonatal mortality is still unacceptably high in developing countries, and it is estimated that the majority of neonatal deaths could be avoided by intervention. To conclude, this study indicates that high perinatal mortality in nineteenth-century Sweden was associated with advanced maternal age, single marital status and low social class. However, the strongest risk factor for perinatal death was previous stillbirths. Multiparity was not a risk factor. In home deliveries,
548
INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
intervention by midwives to prevent perinatal deaths was a successful approach.
Acknowledgements This research was supported by grants from the Swedish Council for Social Science Research, the National Institute of Public Health, and the Medical Faculty, Umeå University, Sweden.
11 Brändström A. “The loveless mothers”: infant mortality in Sweden
during the nineteenth century with special attention to the Parish of Nedertorneå (In Swedish with a summary in English). Umeå Studies in the Humanities Umeå 1984;62. 12 Mostafa G, Foster A, Fauveau V. The influence of socio-biological
factors on perinatal mortality in a rural area of Bangladesh. AsiaPacific Population Journal 1995;10:63–72. 13 Oyen N, Skjaerven R, Irgens LM. Population-based recurrence risk of
sudden infant death syndrome compared with other infant and fetal deaths. Am J Epidemiol 1996;144:300–05.
References 1 World Health Organization. Perinatal Mortality: A Listing of Available
Information. Geneva: World Health Organization, 1996. 2 Campbell O, Koblinsky M, Taylor P. Off to a rapid start: appraising
maternal mortality and services. Int J Gynaecol Obstet 1995;48:S33–52. 3 Commentary: Is perinatal mortality a useful indicator of maternal
mortality? MotherCare Matters. Arlington: MotherCare/Johns Snow 1991;2: p.9. 4 Bai NS, Mathews E, Nair PM, Sabarinathan K, Harikumar C. Perinatal
14 McDermott J, Steketee R, Wirima J. Perinatal mortality in rural
Malawi. Bull World Health Organ 1996;74:165–71. 15 Bengtsson M. The child at risk. Linköping’s study in Arts and Science
(Diss). Linköping 1996. 16 UNICEF. The state of the world’s children, 1995. Oxford: Oxford
University Press, 1995. 17 Forbes JF, Pickering RM. Influence of maternal age, parity and social
class on perinatal mortality in Scotland: 1960–82. J Biosoc Sci 1985;17:339–49.
mortality rate in a south Indian population. J Indian Med Assoc 1991; 89:97–98.
18 Bergström S, Höjer B, Liljestrand J, Tunell R. Perinatal health care
5 Espeut D, Koblinsky M. Methodological issues in perinatal mortality
19 Edvinsson S. Urban Health and Social Class. In Health and Social
research. MotherCare Matters 1997;6:pp.1–18. 6 Fikree FF, Gary RH. Demographic survey of the level and deter-
minants of perinatal mortality in Karachi, Pakistan. Paediatr Perinat Epidemiol 1996;10:86–96. 7 Shah U, Pratinidhi AK, Bhatlawande PV. Perinatal mortality in rural
India: intervention through primary health care. II Neonatal mortality. J Epidemiol Community Health 1984;38:138–42.
with limited resources. London: Macmillan, 1994. Change. Disease, health and public care in the Sundsvall district 1750–1950. Brändström A, TLG. Umeå: Report no 9, Demographic Database, Umeå University 1993. 20 Sköld P. The two faces of small pox: a disease and its prevention in
eighteenth- and nineteenth-century Sweden. University of Umeå: Report from the Demographic Database, 1996;12.
8 Fauveau V, Wojtyniak B, Mostafa G, Sarder AM, Chakraborty J.
21 Cederschiöld FA. Lärobok för barnmorskor. Stockholm 1873. 22 Brändström A. “The loveless mothers”—infant mortality in Sweden
Perinatal mortality in Matlab, Bangladesh: a community-based study. Int J Epidemiol 1990;19:606–12.
during the 19th century with special attention to the parish of Nedertorneå. Umeå: Umeå Studies in the Humanities 62;1984.
9 Högberg U, Wall S, Brostrom G. The impact of early medical
23 Jaffre Y, Prual A. Midwives in Niger: an uncomfortable position
technology on maternal mortality in late 19th century Sweden. Int J Gynaecol Obstet 1986;24:251–61.
between social behaviours and health care constraints. Soc Sci Med 1994;38:1069–73.
10 Högberg U. Maternal mortality in Sweden. Umeå University Medical
24 De Muylder X. Perinatal mortality audit in a Zimbabwean district.
Dissertation. New Series 156. 1985 (ISSN 0346;
Paediatr Perinat Epidemiol 1989;3:284–93.
