Endometritis occurs as a result of contamination of the endometrial cavity with vaginal organisms during labour and delivery. Invasion of the myometrium may occur and the highly vascular environment of the postpartum genital tract predisposes to the development of septicaemia. This can be rapid and overwhelming and ‘puerperal fever’ was and, as described above, in developing countries still is, an important cause of maternal death. Infection can impede the natural process of uterine involution and lead to severe postpartum haemorrhage. Therefore, if not identified and managed early, high rates of morbidity and even mortality can occur.

A standard definition of endometritis or uterine infection was not found. The causal organisms are usually normal inhabitants of the genital tract (Pastorek & Sanders 1991) and gaining a sample for culture without contamination is problematic, so positive culture is not a prerequisite for diagnosis. In the past, midwifery rules and codes of practice regulated care by defining the range of temperature rise which would initiate referral to a medical practitioner. For example, the Central Midwives Board Rules of 1962 stated that if a woman had ‘a rise of temperature above 99.4° F, on three successive days, or a rise of temperature to 100.4° F, a registered medical practitioner must be summoned’ (p. 60). With subsequent revisions to these rules and codes over the last four decades, there is no longer a defined range for what constitutes pyrexia. A transient elevation in maternal temperature is not uncommon during the postnatal period, for example among women who experience breast engorgement on or about the third postpartum day (RCM 2002), although the aetiology of this is unclear. The genital tract is a primary source of infection at this time, although other sites of infection must be considered (Calhoun & Brost 1995).

The World Health Organization (WHO) identifies the symptoms of uterine infection as a temperature >38° C and any of the following: feeling very weak, abdominal tenderness, foul-smelling lochia, profuse lochia, uterus not well contracted, lower abdominal pain or history of heavy vaginal bleeding (WHO 2006).

The threshold for treating as endometritis a pyrexial postpartum woman in whom alternative sources of infection have been excluded is likely to vary both individually and institutionally. In one study (Parrot et al 1989), irrespective of cause being established, antibiotics were prescribed for all women with a temperature of 37.8° C for more than 24 hours after a caesarean section. NICE guidelines state that sepsis may be suspected in the presence of two or more of these signs and symptoms: fever >38.5° C on one occasion or fever of 38° C taken 4 hours apart; chills; abdominal tenderness and no other recognised source of infection; uterine subinvolution; offensive or heavy lochia or tachycardia (NICE 2006). The level of evidence for this recommendation is a best practice point based on the experience of the Guideline Development Group.

In practice, however, endometritis is effectively the default diagnosis for any febrile illness without other cause in the immediate postpartum period, especially within the first 24 hours.

Lack of a standard definition makes it difficult to ascertain accurate rates of occurrence and the estimate will be affected by the rigour with which alternative sources of infection are sought in pyrexial postpartum women, and by the level of ascertainment of infection after discharge from hospital.

In a random sample of women delivered vaginally at an Iowa hospital over a 14 -year period, the rate of endometritis was 1.6% (Ely et al 1996), and a study from Israel of the outcome of 75,947 term and preterm singleton deliveries between 1980 and 1997 found an endometritis rate of 0.17% following vaginal delivery and 2.63% following caesarean section (Chaim et al 2000). With retrospective reviews, however, it is difficult to be sure that follow-up was complete as women with endometritis may not be readmitted or may be admitted to a different unit. Using routine health insurer data, Yokoe et al (2001) estimated rates of endometritis at 0.2% following vaginal delivery and 0.8% after caesarean section during the first 30 days postpartum. As with other postpartum infections, estimates of occurrence of endometritis are now influenced by the use of prophylactic antibiotics, particularly following caesarean section.

Findings from the BLiPP study (Alexander et al 1997), described in more detail below, suggest that approximately 2% of women are admitted to hospital with complications associated with abnormal vaginal bleeding and/or uterine infection within 12 weeks of delivery.

The most important risk factor for endometritis is caesarean delivery. Smaill & Hofmeyr (2000) reviewed 66 trials of antibiotic prophylaxis in caesarean section. Endometritis was consistently classified across trials as pyrexia in the absence of another cause. In the control groups, the overall incidence of endometritis in women undergoing elective caesarean section was 6.4% and 28.6% in women who had an emergency caesarean. The use of antibiotic prophylaxis for both planned and unplanned caesarean had a significant protective effect against endometritis, reducing the rate of endometritis by two-thirds.

Other factors reported to predispose to endometritis after vaginal deliveries, as presented in the background to a protocol for a Cochrane systematic review of treatment regimes, include: bacterial vaginosis; genital cultures positive for anaerobic gram-negative bacilli; prolonged rupture of the membranes; the presence of meconium during labour; low infant birth weight and multiple vaginal examinations (French & Smaill 2004). No association between postnatal endometritis and the use of water for pain relief or delivery has been found (Robertson et al 1998).

Textbooks of midwifery and obstetrics suggest that the postpartum vaginal loss should no longer be red after the first 7 days and lochia should diminish within 4–8 weeks of delivery (Abbott et al 1997, Howie 1986). This definition of normal postnatal vaginal loss formed the basis for practice, and any deviation from it would be defined as abnormal bleeding. However, the evidence base for these assumptions is not described and evidence from more recent studies suggests that the normal duration and character of vaginal loss postpartum may be much more varied (Marchant et al 1999, Oppenheimer et al 1986, Visness et al 1997).

Oppenheimer et al (1986), in a prospective cohort study, recruited 617 women who delivered a live infant during a 10-week period and interviewed them within 48 hours of delivery. Each was asked to complete a diary sheet for up to 60 days and record one of three categories (lochia rubra, serosa and alba) which best described the colour of their lochia. Women were also given a fourth choice to describe when vaginal loss had ceased or normal pre-pregnancy discharge returned. Diary sheets were returned when no loss had been recorded for 7 days, except for recurrence of menses; only 236 women (38%) completed and returned diary sheets. The median duration of lochia rubra was 4 days and for lochia serosa, 22 days; 87 (36%) women did not experience lochia alba, as their loss ceased with lochia serosa. No mean duration of lochia alba was given for the remaining 149 women. Results confirmed clinical impressions that lochia persists for longer than is generally anticipated; 15% of women still had lochia serosa at 6 weeks postpartum, and 4% at 60 days. The researchers concluded that prolonged vaginal loss was neither unusual or abnormal, although some caution should be applied to these findings because of the low response rate.

Visness et al (1997) carried out a prospective study of breastfeeding women in Manila, The Philippines, to examine their experience of postnatal vaginal loss and compare findings with the duration and stages of lochia, as described in a commonly used obstetric text book in the USA. The study was undertaken as part of a randomised controlled trial of the effectiveness of the lactational amenorrhoea method of contraception. All women who chose this method of contraception, had given birth vaginally and had previously breastfed a child for a minimum of 12 months were invited to take part; 477 consented (the total number of eligible women was not given). Follow-up was for 1 year from the day of delivery. The women were given calendars to record daily, starting from the day of delivery, whether they had experienced vaginal bleeding or spotting. The median duration of lochia was 27 days (range 5–90 days) and did not vary by age, parity, infant characteristics or breastfeeding frequency. A quarter of the women reported that blood loss stopped and then recommenced. Lochia lasted for longer than the period of 2 weeks defined in the obstetric textbook and the return of menses was rare before 8 weeks.

It would be difficult to replicate this study in the UK, as few women continue breastfeeding beyond 6 months (Bolling et al 2007) and those that do are likely to have sociodemographic characteristics which differ from the general childbearing population. However, that lochial flow persisted up to and beyond the sixth postnatal week was confirmed by a UK study – the Blood Loss in the Postnatal Period (BLiPP) study (Marchant et al 1999). BLiPP was a three-part research study to investigate the hypothesis that ‘routine abdominal palpation of uterine fundal height in postnatal women from 24 hours after delivery until the midwife discharges the woman from her care, fails to predict abnormal uterine bleeding or uterine infection’ (Marchant & Alexander 1996, p. 402). One aim of the first part of the study was to describe the range of normal vaginal loss from 24 hours until 3 months after delivery.

To collect data for this, a prospective survey of 524 women from two health districts in the south of England was carried out between 1995 and 1996. Women were asked to complete two questionnaires and two diaries at intervals up to 16 weeks post delivery. Information on demographic and delivery characteristics were obtained from birth registers for all eligible women (i.e. women who delivered in the local area and who could speak/read/write English). The first questionnaire was completed from 48 hours to 5 days after delivery, depending on whether study recruitment had taken place on the postnatal ward or if the woman had been contacted following discharge. Women were asked about their experience of vaginal loss immediately following delivery and about the loss experienced on the day the questionnaire was completed. They were offered a range of descriptions to help them answer questions on the colour and amount of vaginal loss; 350 (67%) women responded. Between days 2 and 10, women completed the first diary (318 (61%) responded), followed by the second diary completed on the 14th, 21st and 28th postnatal days (284 (54%) responded). Women were asked to record details about various aspects of their postnatal health, such as psychological state and infant feeding, as well as the amount and duration of vaginal loss. The second questionnaire, sent 3 months after delivery, included questions on the duration and description of vaginal loss and about any problems associated with prolonged or excessive vaginal loss after the first 28 days up to completion of the questionnaire; 324 (62%) women responded.

Vaginal loss was more varied in the amount, colour and duration than described in commonly used midwifery textbooks, a similar finding to that of Visness et al (1997). The duration of vaginal loss ranged from 2 to 86 days after delivery, with a mean of 24 days. Duration was not associated with parity and 6% of women reported vaginal loss from 6 weeks up to 12 weeks postpartum. One interesting finding was that seven (4%) of the 175 primiparae who returned the first questionnaire were unaware that they would experience blood loss after childbirth. Of the 324 women who returned the questionnaire, 64 (20%) had been worried about their vaginal loss between 28 days and 3 months following the birth (Marchant et al 2002). The most common problems reported were passage of clots and heavy or moderate loss. The researchers concluded that findings could be used to inform women and health professionals of the colour, amount and duration of vaginal loss for the first 3 months following childbirth.

Secondary PPH has traditionally been defined as a severe blood loss occurring after the first 24 hours of delivery and up to 6 weeks postpartum. Alternative definitions of secondary PPH vary in the time since birth (Alexander et al 2003) or in not defining any interval as being postpartum (Schuurmans et al 2000, WHO 2003). NICE postnatal guidelines have defined primary and secondary PPH as excessive vaginal blood loss but there is no precise definition of what constitutes a severe or excessive blood loss (NICE 2006). That bleeding is ‘severe’ will sometimes be self-evident, but is mostly subjective at the point where advice is sought. In clinical practice the passage of ‘large’ blood clots is often used as an indication of potential problems. There is no evidence base to show if the size or number of clots passed is indicative of potential or actual morbidity. In the absence of relevant evidence the decision to take further action will continue to be subjective.

Because the definition of ‘severe blood loss’ is imprecise, the incidence of secondary PPH is difficult to ascertain, as reports may only refer to the most adverse outcomes. Prospective studies were not found and, as with endometritis, retrospective hospital-based reviews are likely to be underestimates as patients managed in the community and those admitted to other units will not be included. In one such case-note review in Hong Kong, the secondary PPH cases identified formed 0.38% of all deliveries, but the authors gave no details on how complete ascertainment of cases was likely to have been (Fung et al 1996). The rate of secondary PPH cases over a 2-year period between 1978 and 1980 in Detroit was 1.5%, but again no assessment can be made of case ascertainment for this study (Lee et al 1981). Rome (1975) estimated the rate of occurrence in his series to be 1.3%. As reported under endometritis, approximately 2% of women are admitted to hospital with complications associated with abnormal vaginal bleeding and/or uterine infection within 12 weeks of delivery (Alexander et al 1997).

The International Federation of Gynecology and Obstetrics reported the most common causes of secondary PPH to be subinvolution of the placental site, infection and retained products of conception (RPOC) (ACOG 1990). Retained products of conception may provide the focus for infection or their presence may mechanically inhibit retraction of the uterus. RPOC were found in 17.9% of 106 women with secondary PPH in Rome’s (1975) series.

It is likely, however, that excessive bleeding will not always occur with retained products of conception and an ultrasound study supports this (Tekay & Joupila 1993). In some cases of secondary PPH, the primary problem may be failure of the normal process of involution without either infection or RPOC.

Khong & Khong (1993) undertook a case-note and histological review of women with postpartum haemorrhage occurring between 24 hours and 6 months following delivery. The purpose of the study was to look for evidence of abnormal placentation in women who had secondary postpartum haemorrhage. In the non-pregnant uterus, the arteries are small in diameter and spiral. In the process of forming the placental bed in pregnancy, these vessels become much larger and more flaccid. Failure of this normal placentation has been linked with fetal growth retardation, pre-eclampsia and other pregnancy conditions (Khong et al 1986, Pijnenborg et al 1991). The authors’ aim was to determine evidence of failure of the normal process whereby these vessels collapse and thrombose prior to regeneration of the normal non-pregnant pattern. They called this subinvolution of the placental bed.

The study population was defined by histological material received by the pathology laboratory at one Australian hospital. Therefore, women who were managed conservatively or had surgery but from whom no material was obtained were excluded. Of the 169 cases identified over a 5-year period between 1986 and 1991, 109 (64.5%) presented within 6 weeks of delivery and 54 (35.5%) between 6 weeks and 6 months postpartum. Including cases up to 6 months later in the analysis means the findings could relate to a subsequent pregnancy, and the following results relate to the cases presenting within 6 weeks. In 32% (n = 35) of the 109 cases presenting before 6 weeks, retained placental tissue was identified. Evidence of endometritis was adjudged to be present in only 4.6% (n = 5). The authors classified 20.2% (n = 22) as showing ‘subinvolution of placental bed vessels’, the remaining 43.1% (n = 47) specimens being ‘non-diagnostic material’, normal endometrium, decidua or showed ‘normal’ placental involution.