Maternal and institutional characteristics associated with the administration of prophylactic antibiotics for caesarean section: a secondary analysis of the World Health Organization Multicountry Survey on Maternal and Newborn Health

Authors

  • N Morisaki,

    Corresponding author
    1. Department of Health Policy, National Center for Child Health and Development, Tokyo, Japan
    2. Department of Paediatrics, Graduate School of Medicine, University of Tokyo, Tokyo, Japan
    • Correspondence: Dr N Morisaki, Division of Health Policy, National Center for Child Health and Development, 10-1-2 Okura, Setagaya-ku, Tokyo 157-8535, Japan. Email morisaki-n@ncchd.go.jp

    Search for more papers by this author
  • T Ganchimeg,

    1. Department of Health Policy, National Center for Child Health and Development, Tokyo, Japan
    Search for more papers by this author
  • E Ota,

    1. Department of Maternal and Child Health, National Center for Child Health and Development, Tokyo, Japan
    Search for more papers by this author
  • JP Vogel,

    1. UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP), Department of Reproductive Health and Research, World Health Organization, Geneva, Switzerland
    2. Faculty of Medicine, Dentistry and Health Sciences, School of Population Health, University of Western Australia, Perth, WA, Australia
    Search for more papers by this author
  • JP Souza,

    1. UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP), Department of Reproductive Health and Research, World Health Organization, Geneva, Switzerland
    Search for more papers by this author
  • R Mori,

    1. Department of Health Policy, National Center for Child Health and Development, Tokyo, Japan
    Search for more papers by this author
  • AM Gülmezoglu,

    1. UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP), Department of Reproductive Health and Research, World Health Organization, Geneva, Switzerland
    Search for more papers by this author
  • on behalf of the WHO Multicountry Survey on Maternal and Newborn Health Research Network


Abstract

Objective

To illustrate the variability in the use of antibiotic prophylaxis for caesarean section, and its effect on the prevention of postoperative infections.

Design

Secondary analysis of a cross-sectional study.

Setting

Twenty-nine countries participating in the World Health Organization Multicountry Survey on Maternal and Newborn Health.

Population

Three hundred and fifty-nine health facilities with the capacity to perform caesarean section.

Methods

Descriptive analysis and effect estimates using multilevel logistic regression.

Main outcome measures

Coverage of antibiotic prophylaxis for caesarean section.

Results

A total of 89 121 caesarean sections were performed in 332 of the 359 facilities included in the survey; 87% under prophylactic antibiotic coverage. Thirty five facilities provided 0–49% coverage and 77 facilities provided 50-89% coverage. Institutional coverage of prophylactic antibiotics varied greatly within most countries, and was related to guideline use and the practice of clinical audits, but not to the size, location of the institution or development index of the country. Mothers with complications, such as HIV infection, anaemia, or pre-eclampsia/eclampsia, were more likely to receive antibiotic prophylaxis. At the same time, mothers undergoing caesarean birth prior to labour and those with indication for scheduled deliveries were also more likely to receive antibiotic prophylaxis, despite their lower risk of infection, compared with mothers undergoing emergency caesarean section.

Conclusions

Coverage of antibiotic prophylaxis for caesarean birth may be related to the perception of the importance of guidelines and clinical audits in the facility. There may also be a tendency to use antibiotics when caesarean section has been scheduled and antibiotic prophylaxis is already included in the routine clinical protocol. This study may act as a signal to re-evaluate institutional practices as a way to identify areas where improvement is possible.

Introduction

Caesarean section is the most important factor known to be associated with postpartum bacterial infections, with a rate of infection reported to be 1–25%,[1-4] which is 5–20 times higher than that of vaginal delivery.[5, 6] There is clear evidence that prophylactic antibiotics for caesarean section reduce the risk of endometritis and other bacterial infections,[7, 8] even in low-risk (before labour and with intact membranes) pregnancies,[7, 9] and the use of universal prophylactic antibiotics has been widely accepted in guidelines for many countries, including the USA[6, 10, 11] and several Asian countries.[12] However, reports showed that there are barriers to changing practices (to provide prophylaxis for all caesarean births),[8, 13-15] with ongoing debate about whether it is possible to identify certain high- or low-risk groups in order to tailor prophylaxis.[6, 13, 15-18]

In a previous study of facility deliveries in 29 countries, we reported that the caesarean section rate was 28.6% and that 13% of these were not administered prophylactic antibiotics.[19] In this secondary analysis, we describe the maternal and institutional characteristics associated with a lack of provision of prophylactic antibiotics for caesarean section, and whether clinical relevance alone is enough to adjust our practices.

Methods

Study population

This is a secondary data analysis of the WHO Multicountry Survey on Maternal and Newborn Health, a cross-sectional survey that was conducted in 359 health facilities in 29 countries in Africa, Asia, Latin America and the Middle East.[19, 20] In brief, a multi-stage cluster sampling method was used to obtain a sample of health facilities in two randomly selected provinces and the capital city of each country. All women who were admitted for delivery or who had severe maternal outcomes during the study period were included in the study. Trained medical staff retrieved demographics and reproductive characteristics, medical conditions during pregnancy, birth outcomes, and complications and interventions received from medical records. Health facility data, including the location and type of health facility, hospital structure and capacity, and availability of essential and comprehensive obstetric and neonatal healthcare resources, were also obtained for each facility.

Data were collected over a period of 2 months from May 2010 to December 2011 in facilities with ≥6000 annual births and over 3 months in facilities with <6000 annual births. In countries in which less than 3000 births were anticipated, the study period was extended to 4 months in all facilities.

We obtained data on 314 055 women who gave birth. Of the 89 149 women who gave birth to infants weighing at least 500 g and at 22 or more completed weeks of gestation by the time of caesarean section, we excluded 27 (0.2%) because of missing prophylactic antibiotic status, and described our data on 89 122 women. For the multivariate analyses, we further excluded 6808 pregnancies (7.6%) because of missing covariates, and based our results on complete case analysis.

Variables and definitions

The use of prophylactic antibiotics was specifically assessed in the questionnaire. Variability among institutional coverage of antibiotic prophylaxis was explored and categorised into three categories according to the distribution: ‘good’ (≥90% coverage), ‘poor’ (50–89% coverage) and ‘very poor’ (<50% coverage).

Other facility-level information, such as size, location and capacity, was provided by the hospital coordinator through a specific, self-explanatory institutional data collection form. Using data available from this form, we created a ‘facility capacity index category’ a proxy for the institution's capacity to provide obstetric care – comprising six areas that reflect the standard of basic services, medical services, emergency obstetric services, laboratory tests, hospital practices and human resources, calculated into a continuous index and categorised as ‘good’, ‘poor’ or ‘very poor’. The Human Development Index (HDI) based on the 2012 ranking[21] was used as a proxy for the different medical and social backgrounds between countries.

We defined bacterial infection as having any of the following complications identified during pregnancy or up to 7 days postpartum: puerperal endometritis, pyelonephritis, systemic infections including sepsis, and other infections needing therapeutic parenteral antibiotics.

To determine maternal risk factors for the lack of use of prophylaxis, as well as for the absence of bacterial infection, we considered the following variables as exposures at the individual level and further categorised them as shown in Table 1: maternal age at delivery; marital status; educational attainment; parity; previous caesarean section; multiplicity of the pregnancy; major fetal congenital malformation; length of gestation; onset of labour; and maternal complications as follows: severe anaemia, defined as haemoglobin < 7 mg/dl; bacterial infections defined as pyelonephritis, sepsis or other systemic infection; HIV or AIDS; malaria or dengue; chronic hypertension; pre-eclampsia or eclampsia; obstetric bleeding (placenta praevia, accreta/increta/percreta placenta, abruptio placenta, ruptured uterus, postpartum haemorrhage or any other obstetric haemorrhage); and other maternal medical conditions (presence of diseases or injuries affecting the heart, lung, liver or kidneys).

Table 1. Maternal characteristics and associations with antibiotic prophylaxis.
 Number of caesarean deliveries (n)Coverage of antibiotic prophylaxis for caesarean section (%)
In all caesarean deliveries (n = 89 121)In institutes with ≥90% prophylaxis coverage (n = 66 271)In institutes with <90% prophylaxis coverage (n = 22 850)
  1. a

    Placenta praevia, accreta/increta/percreta placenta, abruptio placenta, ruptured uterus, postpartum haemorrhage and any other obstetric haemorrhage.

  2. b

    Any chronic or acute injury or disorder affecting the kidneys, heart, lungs or liver.

Total 89121879760
Age (years)
<207145879760
20–3468249859666
≥3513572889760
Marital status
Single8395869667
Married/cohabiting80182879760
Education (years)
06801829661
1–69986849757
7–916275869760
10–1228575899766
>1221161899755
Parity
042347889763
1–237757889759
≥38874819657
Any previous caesarean section 30835889759
Congenital malformation 1077899763
Multiplicity
Singleton86759879760
Twin2284859751
Triplet or more76859747
Anaemia (haemoglobin < 7mg/dl) 2218889779
Obstetric haemorrhagic disorders a 3458879762
Hypertensive disorders
Chronic hypertension776879855
Pre-eclampsia/eclampsia4684899760
Infection
HIV+/AIDS491929874
Malaria/dengue173499832
Onset of labour
Spontaneous labour42782859757
Induction8274889569
No labour37865909763
Preterm delivery 8638889662
Stillbirth 1273829482
Other maternal medical conditions b 1169909755

Statistical analysis

First, we examined the variation within facilities with regard to the coverage of antibiotic prophylaxis, as well as whether this variation was associated with institutional characteristics.

Next, we compared coverage of antibiotic prophylaxis by maternal characteristics, as well as within strata of facilities based on whether they provided above or below 90% coverage of prophylaxis. All subgroups were tested for significant difference between coverage by chi-squared tests.

To determine the effect of maternal characteristics on antibiotic coverage, we constructed multilevel, multivariate logistic regression models, using the maternal characteristics as shown in Table 1, and with three levels [country (level 1), facility (level 2) and individual (level 3)], allowing random effects at higher levels. We took into account differences between countries and institutions by adjusting for country development (HDI categories) and facility capacity (by facility capacity index category), as well as using a multilevel model which allowed evaluation of the differences in the baseline coverage and infection rate between institutions within each country. We repeated this analysis in subgroups of facilities based on whether they provided above or below 90% coverage of antibiotic prophylaxis.

To determine the relationship between maternal characteristics and bacterial infection, we constructed multilevel, multivariate logistic regression models, using the maternal characteristics as shown in Table 1 and with the three levels. We repeated this analysis in subgroups of women based on whether they did or did not receive prophylaxis.

Statistical analysis was conducted using Stata/MP version 12.0 (Stata Corp LP, College Station, TX, USA), and P < 0.05 was considered to be statistically significant.

Results

Of the 359 facilities in 29 countries, three facilities in Uganda replied that they could not administer parenteral antibiotics and 27 facilities, including these three, did not perform any caesarean section. Of the remaining 332 facilities that performed caesarean delivery, coverage of antibiotic prophylaxis varied greatly from 0% to 100% (Figure 1A). From this distribution, coverage was categorised into three categories: 220 facilities as ‘good’ (≥90% coverage), 77 facilities as ‘poor’ (50–89% coverage) and 37 facilities as ‘very poor’ (<50% coverage), 13 of which had 0% coverage. Institutional variability was large within each country, with 13 countries having institutions with coverage both below 50% and above 90% (Figure 1B).

Figure 1.

(A) Distribution of prophylaxis coverage for caesarean section. Analysis of 332 facilities in 29 countries. Facilities are in Japan, Qatar, Argentina, Mexico, Lebanon, Peru, Brazil, Ecuador, Sri Lanka, Jordan, China, Thailand, Mongolia, Occupied Palestinian Territory, Paraguay, Philippines, Vietnam, Nicaragua, India, Cambodia, Kenya, Pakistan, Angola, Nigeria, Nepal, Uganda, Afghanistan, Democratic Republic of Congo and Niger. (B) Distribution of prophylaxis coverage for caesarean section. Analysis of 332 facilities in 29 countries. COD, Democratic Republic of Congo; OPT, Occupied Palestinian Territory.

In Table 2, we describe the characteristics of the facilities, stratified into three categories by coverage. Facilities with higher coverage reported more WHO and local guideline use, with 66% of facilities with ‘good’ coverage using either WHO or local guidelines, compared with 32% of facilities with ‘very poor’ coverage reporting that they used neither.

Table 2. Characteristics of facilities by coverage of prophylactic antibiotics for caesarean section. Analysis of 332 facilities in 29 countries
 TotalInstitutional coverage of prophylaxis P
0–49% (%)50–89% (%)90–100% (%)
  1. WHO, World Health Organization.

  2. a

    Proxy for the medical level of each health facility comprising six categories reflecting the standard of building/basic services, medical services, emergency obstetric services, laboratory tests, hospital practices and human resources.

  3. b

    As of 2012. Very high HDI countries included Japan, Qatar and Argentina. High HDI countries included Mexico, Lebanon, Peru, Brazil, Ecuador and Sri Lanka. Medium HDI countries included Jordan, China, Thailand, Mongolia, Occupied Palestinian Territory, Paraguay, Philippines, Vietnam, Nicaragua, India and Cambodia. Low HDI countries included Kenya, Pakistan, Angola, Nigeria, Nepal, Uganda, Afghanistan, Democratic Republic of Congo and Niger.

Number of facilities 33235 (11)77 (23)220 (66) 
Location
Urban22620 (65)54 (77)152 (74)0.60
Peri-urban446 (19)7 (10)31 (15)
Rural435 (16)9 (13)22 (11)
Average number of maternity beds in use
<4615418 (58)36 (51)100 (51)0.42
47–100757 (23)22 (31)46 (23)
101–190435 (16)6 (8.6)32 (16)
>191301 (3.2)6 (8.6)23 (12)
Maternity exclusive hospital 589 (29)13 (19)36 (18)0.34
Proportion of delivering women receiving care free of charge
<25%14113 (42)23 (33)105 (52)0.05
25–50%331 (3.2)11 (16)21 (11)
50–99%384 (13)10 (15)24 (12)
100%8713 (42)25 (36)49 (25)
Guideline use in facility
Both WHO and local18710 (32)42 (61)135 (66)0.001
Only WHO or local7911 (36)16 (23)52 (23)
None3810 (32)11 (16)17 (8)
Clinical audits in practice 25521 (68)55 (81)179 (88)0.009
Obstetric specialist available on call 26820 (57)60 (78)188 (85)<0.001
Proportion of caesarean sections before labour
<20%11822 (63)30 (39)66(30)0.001
20–60%1298 (23)34 (44)87 (40)
60%855 (14)13 (17)67 (30)
Rate of bacterial infections among caesarean section deliveries
<1%17723 (66)30 (39)124 (57)0.001
1–10%13110 (29)34 (45)87 (40)
>10%222 (5.7)12 (16)8 (3.6)
Facility capacity a
Good684 (10)9 (17)55 (29)0.02
Poor11515 (37)22 (42)78 (41)
Very poor9821 (53)21 (18)56 (30)
Human Development Indexb (HDI) of country
Low12419 (54)35 (44)70 (32)0.13
Medium876 (6.9)16 (21)65 (30)
High968 (23)21 (27)67(30)
Very high252 (8.0)5 (6.5)18 (8.2)

Similarly, the percentages of facilities reporting that they had clinical audits in practice (P = 0.009), or had an obstetric specialist available on site or on call at all hours (P < 0.001), were significantly smaller in facilities with lower coverage. Clinical audits were in practice for 88% of facilities with ‘good’ coverage, compared with 68% in the ‘very poor’ group, and obstetric specialist availability was 85% compared with 57%.

In addition, facilities with ‘very poor’ coverage were more likely to be categorised with a lower facility index (P = 0.002), and had a smaller proportion of caesarean births before labour (P = 0.001), compared with institutes with higher coverage.

The HDI of the country, number of maternal beds, location, proportion of women receiving care free of charge and whether the hospital was a maternity exclusive hospital did not show a significant association with coverage.

Infection rate among caesarean births was 7.8% overall. The rate of infection in mothers who had caesarean births did not decline with coverage. Institutions with an extremely high rate of infection (20–80%) mostly had ‘poor’ coverage of prophylaxis, and 23 (66%) facilities with ‘very poor’ coverage had a low bacterial infection rate of below 1%.

In Table 2, we describe the coverage in various populations. Of the 89 121 caesarean births, 87% were covered with antibiotics. This was 97% in facilities with ‘good’ coverage and 60% in facilities with ‘poor’ or ‘very poor’ coverage.

In facilities with ‘good’ coverage overall, coverage varied only slightly in the range 94–98% when stratified by individual maternal characteristics. This variability was larger (32–82%) in facilities with ‘poor’ or ‘very poor’ coverage, with mothers of triplets or more (coverage 47%), and those with malaria or dengue (32%), having the lowest proportion receiving prophylaxis. Alternatively, mothers with anaemia (79%), HIV or AIDs (74%) and those delivering stillborns (82%) had a larger proportion receiving prophylaxis.

In Tables 3 and 4, we show the adjusted odds ratios (aORs) for the estimated effect of maternal and infant characteristics on the lack of prophylaxis, as well as on bacterial infection, adjusted by country and facility, in addition to other maternal characteristics.

Table 3. Risk factors for the non-use of prophylactic antibiotics for caesarean section stratified by the institutional use of prophylaxis. Analysis of 82 320 caesarean sections in 29 countries
 All caesarean deliveries (n = 82 320)Deliveries in institutions with above 90% coverage (n = 61 476)Deliveries in institutions with below 90% coverage (n = 20 844)
aOR95% CIaOR95% CIaOR95% CI
  1. aOR, adjusted odds ratio; CI, confidence interval.

  2. Multilevel, multivariate logistic regression models were used to obtain aORs: outcome was ‘absence of prophylactic antibiotics’; multilevel analysis was structured as three levels (individual, health facility, country) with random intercepts.

  3. a

    P < 0.05.

  4. b

    P < 0.01.

  5. c

    P < 0.001.

  6. d

    Placenta praevia, accreta/increta/percreta placenta, abruptio placenta, ruptured uterus, postpartum haemorrhage and any other obstetric haemorrhage.

  7. e

    Any chronic or acute injury or disorder affecting the kidneys, heart, lungs or liver.

Age (years)
<201.92(0.96; 1.18)1.05(0.88; 1.26)1.07(0.94; 1.21)
20–341REF1REF1REF
≥351.00(0.92;1.09)0.95(0.83,1.09)1.04(0.93; 1.16)
Marital status
Single0.97(0.88; 1.08)1.06(0.89, 1.25)0.92(0.81; 1.06)
Married/cohabiting1REF1 1REF
Education (years)
01.10(0.96; 1.26)1.19(0.93; 1.51)1.02(0.87; 1.21)
1–61.05(0.94;1.18)0.95(0.78; 1.15)1.07(0.93, 1.22)
7–90.96(0.87; 1.06)1.03(0.88; 1.21)0.90(0.79; 1.02)
10–120.90(0.82; 0.98)a0.99(0.86; 1.13)0.83(0.74; 0.93)c
>121REF1REF1REF
Parity
00.87(0.80; 0.94)b0.78(0.69; 0.88)c0.93(0.84; 1.03)
1–21REF1REF1REF
≥31.01(0.92; 1.12)1.11(0.93; 1.33)0.96(0.85; 1.09)
Any previous caesarean section 0.77(0.71; 0.84)c0.62(0.54; 0.71)c0.88(0.80; 0.97)a
Congenital malformation 0.83(0.62; 1.11)1.01(0.70; 1.55)0.67(0.44; 1.01)
Multiplicity
Singleton1REF1REF1REF
Twin0.76(0.63; 0.93)b0.78(0.53; 1.02)0.77(0.61; 0.98)a
Triplet or more1.48(0.67; 3.23)1.11(0.21; 3.65)1.91(0.63; 5.77)
Anaemia (haemoglobin <7mg/dl) 0.59(0.49; 0.71)c1.05(0.74; 1.48)0.51(0.42; 0.63)c
Obstetric haemorrhagic disorders d 0.98(0.84; 1.15)0.87(0.66; 1.13)1.02(0.83; 1.24)
Hypertensive disorders
Chronic hypertension0.92(0.68; 1.25)0.63(0.33; 1.19)1.04(0.72; 1.50)
Pre-eclampsia/eclampsia0.80(0.69; 0.92)b0.75(0.59; 0.96)a0.79(0.66; 0.94)a
Infection
HIV+/AIDS0.63(0.40; 0.99)a0.90(0.45; 1.77)0.53(0.30; 0.94)a
Malaria/dengue0.93(0.53; 1.64)1.10(0.15; 8.21)0.92(0.51; 1.65)
Onset of labour
Spontaneous labour1REF1REF1REF
Induction0.83(0.75; 0.93)c1.07(0.91; 1.27)0.72(0.63; 0.82)c
No labour0.69(0.64; 0.74)c0.61(0.54; 0.68)c0.74(0.68; 0.81)c
Preterm delivery 1.08(0.98; 1.20)1.22(1.04; 1.42)a0.98(0.85; 1.12)
Stillbirth 1.25(1.01; 1.55)a1.85(1.33; 2.58)c1.01(0.78; 1.32)
Other maternal medical conditions e 0.93(0.64; 1.36)1.39(0.88; 2.18)0.53(0.30; 0.94)a
Table 4. Risk factors for bacterial infection stratified by use of antibiotic prophylaxis for caesarean section. Analysis of 82 320 caesarean sections in 29 countries
 All caesarean sectionsCaesarean sections with prophylaxisCaesarean sections without prophylaxis
aOR95% CIaOR95% CIaOR95% CI
  1. aOR, adjusted odds ratio; CI, confidence interval.

  2. Multilevel, multivariate logistic regression models were used to obtain aORs: outcome was any bacterial infection; multilevel analysis was structured as three levels (individual, health facility, country) with random intercepts.

  3. a

    P < 0.05.

  4. b

    P < 0.01.

  5. c

    P < 0.001.

  6. d

    Placenta praevia, accreta/increta/percreta placenta, abruptio placenta, ruptured uterus, postpartum haemorrhage and any other obstetric haemorrhage.

  7. e

    Any chronic or acute injury or disorder affecting the kidneys, heart, lungs or liver.

Age (years)
<201.16(1.03; 1.28)a1.22(1.09; 1.36)b0.70(0.48; 1.03)
20–341REF1REF1REF
≥351.07(0.98; 1.16)1.06(0.97; 1.16)1.07(0.80; 1.43)
Marital status
Single1.17(1.06; 1.30)b1.15(1.03; 1.28)a1.47(1.03; 2.10)a
Married/cohabiting1REF1REF1REF
Education (years)
01.22(1.05; 1.36)b1.21(1.03; 1.42)a1.32(0.88; 2.00)
1–61.12(0.99; 1.26)1.11(0.98; 1.26)1.22(0.83; 1.78)
7–91.21(1.09; 1.34)c1.18(1.06; 1.32)b1.29(0.92; 1.82)
10–121.05(0.96; 1.14)1.04(0.95; 1.14)1.14(0.85; 1.55)
>121REF1REF1REF
Parity
01.05(0.97; 1.13)1.03(0.95; 1.12)1.01(0.78; 1.30)
1–21REF1REF1REF
≥31.19(1.07; 1.32)c1.25(1.12; 1.39)c0.92(0.67; 1.26)
Any previous caesarean section 0.67(0.61; 0.73)c0.66(0.60; 0.72)c0.69(0.52; 0.90)b
Congenital malformation 6.74(5.72; 7.94)c7.16(6.02; 8.53)c3.11(1.76; 5.47)c
Multiplicitiy
Singleton1REF1REF1REF
Twin1.00(0.86; 1.16)0.91(0.77; 1.06)1.54(0.98; 2.44)
Triplet or more1.58(0.88; 2.85)1.68(0.90; 3.16)1.25(0.21; 7.53)
Anaemia (haemoglobin<7 mg/dl) 1.87(1.63; 2.15)c2.00(1.72; 2.31)c1.22(0.21; 2.10)
Obstetric haemorrhagic disorders d 1.33(1.18; 1.50)c1.47(1.29; 1.67)c1.92(1.28; 2.88)b
Hypertensive disorders
Chronic hypertension1.17(0.93; 1.47)1.16(0.91; 1.47)c0.70(0.27; 1.82)
Pre-eclampsia/eclampsia1.68(1.52; 1.85)c1.71(1.55; 1.90)c1.44(0.99; 2.11)
Infection
HIV+/AIDs1.28(0.92; 1.78)1.26(0.90; 1.78)1.12(0.27; 4.55)
Malaria/dengue1.93(1.09; 3.41)b2.92(1.56; 5.48)b0.23(0.03; 1.86)
Onset of labour
Spontaneous1REF1REF1REF
Induction1.08(0.98; 1.20)1.10(0.98; 1.23)1.30(0.92; 1.84)
No labour0.85(0.79; 0.91)c0.83(0.77; 0.90)c0.92(0.72; 1.16)
Preterm birth 7.43(6.91; 7.98)c8.18(7.57; 8.84)c6.59(5.23; 8.30)c
Stillbirth 7.76(5.60; 10.75)c9.34(6.46; 13.51)c3.79(1.82; 7.86)c
Other maternal conditions e 1.88(1.53; 2.31)c1.87(1.51; 2.31)c2.29(0.93; 5.61)

Maternal age, marital status and education did not show a significant relationship with the non-use of prophylaxis, except for mothers with 10–12 years of education in facilities with low coverage, but single marital status was a significant individual risk factor for infection despite prophylactic status, and a lower age and lower education were also significant risk factors for infection in deliveries receiving prophylaxis.

Nulliparous mothers were significantly more likely to be given prophylaxis than multiparous mothers in facilities with high coverage, but mothers with three or more previous births were significantly more likely to have infection among those who received prophylaxis.

Mothers who had undergone previous caesarean birth were significantly more likely to receive prophylaxis [aOR = 1.30; 95% confidence interval (CI), 1.19–1.41; P < 0.001] in facilities with both high and low coverage, but were significantly less likely to have infection (aOR = 0.67; 95% CI, 0.61–0.73; P < 0.001) despite antibiotic cover. Similarly, caesarean births prior to labour were significantly more likely to receive prophylaxis (aOR = 1.49; 95% CI, 1.35–1.56; P < 0.001), but were significantly less likely to have infection (aOR = 0.85; 95% CI, 0.79–0.91; P < 0.001). Twins were more likely to receive prophylaxis in facilities with low coverage, but we did not find a significant association with multiple pregnancies and infection.

Most maternal complications, i.e. anaemia, obstetric haemorrhagic disorders, chronic hypertension, pre-eclampsia/eclampsia, malaria/dengue and ‘other maternal conditions’, were significantly associated with a higher risk of infection. We did not find an association with HIV. Among these complications, anaemia, HIV and pre-eclampsia/eclampsia were associated with a significantly higher risk of receiving prophylaxis.

Preterm delivery and stillbirth were both highly associated with maternal infection regardless of prophylactic status, but only in facilities with ‘good’ prophylaxis coverage were preterm births and stillbirths given more prophylaxis than term and live births.

Discussion

Main findings

In our study, we found that institutional coverage of antibiotic prophylaxis for caesarean birth varied largely within most countries, with coverage being related more to guideline use and the practice of clinical audits than to size, location or development index of the country.

We also found that, although mothers with certain medical risk factors were more likely to receive antibiotics, those with several characteristics known to be associated with a lower risk of infection were also more likely to receive antibiotic prophylaxis. In our study, mothers with infection, such as HIV, anaemia or pre-eclampsia, were more likely to receive prophylaxis. At the same time, women with multiple pregnancies or previous caesarean births were also more likely to receive prophylaxis.

Interpretation

Multiple pregnancies and previous caesarean birth are themselves indications for ‘scheduled’ caesarean birth. The fact that these factors and scheduled birth itself (caesarean birth before labour or following labour induction) were associated with greater use of prophylaxis raises concern about whether there is a tendency to administer prophylaxis when the caesarean birth is planned, and therefore prophylaxis is part of the routine clinical protocol. Mothers who had spontaneous onset of labour, singleton birth and those who had not previously undergone caesarean section were less likely to receive prophylaxis. This may be related to possible gaps in practice, such as forgetting or skipping the administration of antibiotic prophylaxis, due to time constraints.

This might be more prevalent in facilities in which there is a lack of use of guidelines, compounded by a lack of institutional quality improvement measures when there are no clinical audits to retrospectively examine practices and their consequences. Previous reports have stated that continuous data collection and timely dissemination of the results are important factors catalysing improvements in practice.[22]

Interestingly, we did not find any related study examining maternal and/or institutional characteristics associated with use of antibiotic prophylaxis for caesarean birth. Studies from the USA have reported the increased risk of bacterial infections in socially disadvantaged populations,[23] and a report from Norway studied the variation in institutional guidelines towards the use of prophylaxis,[13] but we could not find any study examining what drives the variations in coverage.

Strengths and limitations

This analysis has several strengths. The WHO Multicountry Survey was conducted in 29 countries, using trained data collectors and a standardised methodology that was refined from our experiences with the previous WHO Global Survey. We obtained data on institutional and individual characteristics of deliveries in 332 facilities that performed caesarean section, which allowed us to investigate not only maternal but also institutional characteristics.

However, there are some limitations. First, as the primary data source was routine medical records, erroneous or absent documentation of prophylaxis and maternal complications in the records could have affected data quality. However, we believe that this bias was minimised as much as possible by training provisions prior to study commencement (building on our experiences in the WHO Global Survey) and data collectors consulting with clinical staff to complement information obtained from the records, where necessary. Missing prophylactic data were minimal.

Second, we could not verify whether infections other than puerperal endometritis occurred before or after birth. However, as the risk of bacterial infections increases substantially after an operation,[6] we based our results on the assumption that bacterial infections were probably postpartum. If antepartum infections constituted a substantial proportion of all captured infections, we would have overestimated the effect of prophylaxis on preventing postpartum infection. However, as data collection was only conducted for the duration of the admission, we acknowledge that the risk of infection post-discharge is not small,[13] and was not captured by this survey or evaluated in our study, which could have led to an underestimation.

Third, we obtained a limited number of variables in the institutional data. Therefore, although we were able to identify several important characteristics related to antibiotic coverage, we may have failed to identify the fundamental common factor.

We also lacked several known factors related to bacterial infection, such as the length of the operation and the qualifications and experience of the surgeon, which could have affected our estimates of the effect of maternal characteristics on bacterial infection. However, surgeon experience is associated with the measured institutional characteristics (larger size, more maternal beds, more frequent caesarean delivery), and a longer operation would be associated with the measured underlying medical factors as well as the skill of the surgeon; therefore, the unmeasured confounding may be small.

Finally, although we produced estimates in Table 3 under the assumption that the effect of each maternal characteristic on the presence of prophylaxis was the same in every facility, this is unlikely. Therefore, our results on factors leading to lower coverage should be considered as an overview of the 332 countries participating in this survey, and it should be taken into account that our findings may not be applicable to each individual facility.

Conclusions

Our study suggests that coverage of antibiotic prophylaxis for caesarean birth may be related to the perception of the importance of guidelines and clinical audits in the facility. Although obstetricians presumably acknowledge the increased risk of infections in many maternal complications when they administer antibiotic prophylaxis, there may also be a tendency to use prophylaxis when caesarean birth has been scheduled and prophylaxis is already included in the routine clinical protocol.

To our knowledge, no other study has assessed the maternal and institutional characteristics associated with coverage of antibiotic prophylaxis for caesarean birth. This study may act as a signal to re-evaluate individual practices in order to identify areas with the possibility of improvement.

Disclosure of interests

We declare that we have no competing interests or conflicts of interest.

Contribution to authorship

NM, RM and AMG initiated the concept. NM and TG designed the study and performed the data analysis. NM wrote the initial manuscript. EO, JPV and JPS provided advice and editing of the manuscript. All authors read and approved the final version of the manuscript.

Details of ethics approval

The UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP) Specialist Panel on Epidemiological Research reviewed and approved the study protocol for technical content. This study was approved by the WHO Ethical Review Committee and the relevant ethical clearance mechanisms in all countries.

Funding

This study was financially supported by the UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction (HRP); World Health Organization (WHO); United States Agency for International Development (USAID); Ministry of Health, Labour and Welfare of Japan; and Gynuity Health Projects. The sponsors had no role in the data collection, analysis or interpretation of the data, the writing of the report or the decision to submit for publication.

Acknowledgements

We would like to acknowledge Emma L. Barber (National Center for Child Health Development, Tokyo, Japan) and Annette Peters (World Health Organization) for editing the manuscript. We wish to thank all members of the WHO Multicountry Survey on Maternal and Newborn Health Research Network, including regional and country co-ordinators, data collection co-ordinators, facility co-ordinators, data collectors and all staff of participating facilities who made the survey possible.

Ancillary