Postpartum maternal morbidity requiring hospital admission in Lusaka, Zambia – a descriptive study

Relevant admission and discharge registers at UTH were reviewed for the six-month period July-December 1999. Dependent on the type, timing and severity of a postpartum problem, women may be referred or may self-refer to one of three different units within the hospital: (i) Women with early postpartum complications, defined as problems occurring within 24 hours of delivery, are admitted through the labour ward admission room; (ii) Women with postpartum problems occurring more than 24 hours after delivery are referred or may present themselves to the gynaecology filter clinic, from where they may then be referred on to the emergency admission ward; (iii) Cases of breast abscess are generally admitted though the surgical unit. From any of the units women may be then admitted onto a longer stay gynaecology ward for further management and treatment. Women admitted to hospital irrespective of the length of stay were included in the data capture.

Table 2 outlines the identification process and inclusion criteria that were used in each of the wards. The figures are likely to be an underestimate of the total number of postpartum admissions because as is often found in studies of this nature, diagnosis was frequently poorly recorded in the registers. Only women who could be positively identified as postpartum morbidity admissions were included in the final analysis. Any women admitted to medical wards with non-obstetric conditions – such as malaria – in the later part of the puerperium would have been missed. Using admissions registers to identify postpartum cases also excludes any women who were admitted to hospital prior to the puerperium (for example antenatally, or in labour) and subsequently developed postpartum problems requiring prolonged in patient care.

Cases were crosschecked by name and age between the emergency admission and gynaecology wards and between the short and longer stay surgical wards, to prevent double counting. Cases of breast abscess were also crosschecked between the surgical unit and the gynaecology department to ensure cases had not been referred. Data was entered on Epi. Inf. 6.04 for analysis.

Routine health service statistics for the same six-month study period (July-December 1999) were also collated. There were 19,691 deliveries within UTH and the satellite clinics (5,511 and 14,180 deliveries respectively), of which 1,021 were by caesarean section (an average of 39 per week). Women delivering either at UTH or the clinics were instructed to attend the local satellite clinic for postnatal follow-up visits at week 1 after delivery and again at 6 weeks. Some women were seen at UTH for a postnatal visit at week 1 to follow-up on some complication while those who had a caesarean section were reviewed at 6 weeks. Recorded postnatal check-up at the clinics at 1-week was 40% and 21% at 6 weeks; however, this data was incomplete for some clinics. At UTH, an average of 42 women per week attend the postnatal clinic – most of them after a caesarean section. During the same six-month period, there were 93 maternal deaths at UTH. Of these maternal deaths 8 (8.5%) were known to be due to puerperal sepsis. A much larger number of the women who died had sepsis and stigmata of AIDS (Y Ahmed, personal communication).

Due to the limitations of the routine data sources used in the retrospective review, a small cross sectional study was also conducted using prospective identification of cases in order to verify the findings. Over a 4-week period, from 14 August to 10 September 2000, all early postpartum admissions to the maternity unit at UTH were identified through the labour ward admission register. Obstetric case notes were sought and reviewed. Late postpartum cases (>24 hours and up to 6 weeks after delivery) for the same time-period were also identified and recorded by the same means as described for the retrospective study. On the gynaecology ward, cases were identified through daily review of the ward round books, to identify new admissions, and through consultation with the senior ward sister and the ward clerk.

After crosschecking of data to prevent double counting, 365 maternal postpartum admissions to the hospital were positively identified for the 6-month study period July-December 1999. Cases of retained placenta (n = 55), removed in theatre, were not included in the final analysis as it was not possible to differentiate between clinic referrals and UTH deliveries.

Over the one-month period August-September 2000, 64 admissions to the hospital for postpartum maternal morbidity were identified. The routine hospital procedures differentiate between "early" (up to 24 hours postpartum) and "late" (after 24 hours) postpartum admissions. The former are admitted through the labour ward admissions and the latter through the emergency gynaecology ward. Twenty women (31%) were "early" referrals and 44 (69%) were "late". This categorisation of the data therefore concerns time elapsed between delivery and admission to hospital, and not between time of delivery and onset of the condition.

We have already outlined some of the practical difficulties with using routine hospital data sources such as admission registers. Admission rates estimated from the small prospective part of the study do not suggest, however, that many cases were lost in the identification process in the larger retrospective part. The former identified an average of 11 late-postpartum maternal admissions to hospital per week in the month observed, and the latter, an average of 14 late-postpartum maternal admissions per week over the six months reviewed. In both parts of the study, puerperal sepsis and malaria were identified as leading causes of postpartum morbidity requiring hospital admission. In the prospective identification of postpartum morbidity requiring hospital admission, puerperal sepsis, malaria and hypertensive disease each accounted for the same number of admission (5 in each) but the numbers are too small to draw any conclusions. Some inaccuracy in the classification of certain morbidities may be expected due to the use of admission diagnoses. It is also a limitation in the design of this element of our study that data collection did not extend to the medical wards. It is therefore not possible for us to ascertain whether, or to what extent, there are medical ward admissions of women in the late puerperium with non-obstetric conditions such as malaria.

Puerperal sepsis was the leading cause of postpartum hospital admissions in this population, accounting for 34.8% of all identified postpartum cases in the retrospective part of this study. Other hospital-based studies as well as surveys of women's self-reports of postpartum morbidity report puerperal sepsis as a leading cause of postpartum morbidity in developing countries [5, 6, 15‐18].

Puerperal sepsis cases identified through the retrospective data collection part of the study accounted for more than twice as many cases as the second commonest postpartum morbidity requiring hospital admission, malaria. For the retrospective study the overall rate of puerperal sepsis cases requiring hospital-level care and admission was 0.64% of all supervised deliveries in the public sector services, a rate that falls in between those estimates from earlier hospital-based studies in Niger: 0.22% [5]; and in Nigeria: 1.7% [15]. However, it should again be remembered that in this study, women who delivered in hospital and developed septic complications before discharge would be excluded from this figure. The overall figure can therefore be expected to be higher.

A number of other obstetric-related postpartum morbidities including anaemia, breast abscess, symphisiotomy, puerperal psychosis, after-pains, urinary tract infection, secondary postpartum haemorrhage and retained products of conception were also identified as late-postpartum referrals through the retrospective study. Anaemia in the postpartum period is not an uncommon health problem [1]. Surveys of women's self-reported morbidity frequently cite symptoms in the postpartum period that could be suggestive of or lead to anaemia, including chronic fatigue [19] and excessive bleeding [15, 20, 21].

Malaria and pneumonia together accounted for one-fifth of all the postpartum hospital admissions that we identified. This finding suggests the usefulness of an approach that employs a "time-period" definition to identify cases rather than simply a set of purely obstetrically-related diagnostic categories.

In all, 14.5% of identified postpartum maternal admissions to hospital were due to malaria. While it is widely recognised that the severity and frequency of malaria is greater in pregnant, compared to non-pregnant women, until recently it was generally thought that the importance of pregnancy-related malaria ends with delivery [22] and malaria is rarely mentioned as an important postpartum morbidity in the obstetric literature. Diagne et al.'s study from Senegal [22], however, was one of the first to suggest that the increased susceptibility to malaria in pregnancy persists up to 60 days after delivery. They found that compared to the non-pregnant state, the incidence of episodes of malaria increased significantly during the second and third trimesters of pregnancy and reached a maximum during the first 60 days after delivery.

A number of factors may modify susceptibility to malaria in the postpartum period. The age at which partial immunity to malaria is acquired is critically dependent on transmission intensity [23, 24]. Wide variations are seen in levels of immunity to malaria among Zambian women secondary to geographical and other factors affecting transmission [23]. Many of the malaria cases identified in the study may be the result of recrudescence rather than new infection particularly because the study took place during the transition between dry and wet seasons and study participants were primarily from urban and peri-urban communities. Susceptibility may also be dependent on haematological and nutritional factors as well as HIV status [25].

Pneumonia and Pulmonary TB were important causes of postnatal morbidity in this study and were likely related to HIV/AIDS. The HIV serostatus was rarely available of the index postpartum cases, though unlinked anonymous testing of HIV in the antenatal population in 4 sentinel sites in Lusaka during 1998 showed a high HIV prevalence of 27.4% [26]. HIV positive women are more prone to postpartum infections including urinary tract infections, chest, episiotomy and caesarean section wound infections [27, 28]. Furthermore, in this study, causes of postpartum morbidity included puerperal psychosis, cerebral malaria and HIV related cerebral complications – all of which can be difficult to diagnose with certainty in a malaria and HIV endemic area [29]. Of the 93 cases of maternal mortality during the six-month retrospective study period in 1999, almost a third of the cases were attributed to a presumptive diagnosis of HIV/AIDS and had no other direct or indirect cause of maternal mortality (personal communication, Y Ahmed, 2003).

Review of registers on the labour ward, for both the retrospective and prospective aspects of this study suggest that the majority of referrals in the early-postnatal period (first 24 hours) were for infant rather than maternal indications. Of the early postpartum referrals for maternal indications, retained placenta is the leading reason for referral from clinics to the hospital. This reflects the urban context of the study and the relative ease of transportation that permits a district policy of removal of placenta at hospital level rather than by the practitioner with essential obstetric care skills at the delivery clinic.