The UON methodology has been described in detail by the UON network [8
]. In brief, absolute maternal indications (AMI) are defined as:
• Antepartum haemorrhage due to placenta praevia or abruptio placenta
• Malpresentation (transverse lie, brow presentation etc)
• Ruptured uterus
• Cephalo-pelvic disproportion/obstructed labour based on partograph with action line crossed by the dilation line (for the purpose of this study, this indication was intended to replace all other "indications" like poor progress, dystocia, prolonged labour etc.)
• According to Tanzanian national guidelines and expert opinion, more than two previous Caesarean Section were also included.
These conditions are selected not only because of their life-threatening nature, but also as they require a specific major obstetric intervention which can be verified through health service records. These major obstetric interventions (MOI) were adapted to the Tanzanian situation and included the following:
• Caesarean Section
• Hysterectomy following a caesarean section
• Laparatomy for obstetric interventions
• Destructive operation
• Blood transfusion during pregnancy or delivery.
Blood transfusion was included initially as a major obstetric intervention by the Tanzanian UON team; however, in the course of evaluating the data it became clear that it had not always been reliably recorded. It was therefore decided not to include this in the analysis of results. Therefore, the original UON concept remains as proposed by the UON network and is comparable to studies conducted in other countries. The UON network acknowledges that not all potentially life-saving interventions are included. To keep the concept operational the interventions selected were by necessity specific obstetric interventions such as are reliably recorded in routine registers, theatre ledgers, delivery books, and patient registers [8
The UON network rests on the premise that as an absolute minimum 1–2% of pregnant women shall need a major obstetric intervention to save their lives [9
]. This figure, based upon historical data from England and Wales [10
], is supported by recent UON studies in other countries which have taken thresholds ranging from 1.0% to 1.6% [8
]. Based on a Tanzanian pilot study in Mtwara urban district revealing a value of 2.4% [11
], literature, and expert advice, the Tanzanian UON team decided to set this threshold for Tanzania at 2% of all deliveries.
Study area and study population
The study area chosen was Tanga, a coastal region comprising six districts. Tanga Region has a population of 1,642,015. Three of the six districts took part in the study, namely Lushoto, Muheza and Tanga Municipality. Lushoto and Muheza are rural districts, in which the majority of the population lives in rural areas. Tanga Municipality is an urban district where 80% of the people are living in urban areas. In the study area the populations are 419,970 for Lushoto, 279,423 for Muheza and 243,580 for Tanga Municipality.
The study was conducted in 4 Hospitals; the Regional Hospital in Tanga; the District Designated Hospital in Muheza; the district hospital in Lushoto and in a church-run hospital in Bumbuli, Lushoto District. The methodology was explained to the key maternity staff from the four hospitals together with experts from Muhimbili National Hospital during a two-day workshop in Tanga resulting in the adaptation of the UON questionnaire to the local needs. Furthermore, it was decided to explore whether access to MOI varies according to rural or urban settings. Using the indicator from the 2002 census the distribution of the population was considered against the criterion that up to 10 km from the district main city is urban and >10 km is rural [12
]. According to the above deliberation the population in rural areas is 88% in Lushoto, 89% in Muheza and 18% in Tanga Municipality.
Data collection and analysis
In accordance with the study protocol a questionnaire was filled in for each of the above-mentioned major obstetric interventions carried out, based on the relevant documents (delivery record, antenatal card, theatre book, partograph) and information provided by health workers when needed. The questionnaire was signed by the trained staff member and later by the supervisor. It has to be noted that health workers of the participating maternities were intensively schooled on the importance of partograph use on a routine basis prior to the commencement of the study.
To estimate the UON per expected live birth a crude birth rate of 46 per 1,000 was used as is recommended practice of the Tanzanian Ministry of Health and Social Welfare. The percentage of MOIs per AMI per expected birth was calculated for the districts based on the demographic data from the National Census 2002 [12
]. To estimate the deficit in MOIs as proxy for unmet obstetric need, a threshold of how many interventions are necessary in a given population has to be set. As stated, a threshold of 2% of all deliveries is used for this study.
The study has been conducted over a period of two years between 2000–2002. In Lushoto data was collected over a period of 27 months. All the questionnaires were sent to the regional headquarter in Tanga, where the data entry was undertaken. Altogether, completed questionnaires for 1,260 MOIs were received and analysed in EPI-Info 2000. Out of these, 905 (71%) were carried out for AMIs and were used to assess the unmet obstetric need.
As a measure of quality control the data derived from the UON questionnaires on caesarean section rates were cross-checked with data from the official health information and management system.
Concerning limitations, it has to be noted that when the questionnaire was adapted for the Tanzania context, "blood transfusion" was also considered to be a major obstetric intervention. However, the findings revealed that some women were apparently transfused during pregnancy but did not go on to deliver at that point meaning that certain parts of the questionnaire (maternal outcome/infant outcome) were not filled in. Thus, these questionnaires were not included in the analysis. It could also be the case that blood transfusions were not necessarily perceived as a MOI by health staff which might explain the lack of follow-up.
The comparison of the UON data and service data (Table ) shows that reporting UON data from the districts is almost complete, while it appears that there was considerable under-reporting from Bombo regional hospital in Tanga. Therefore, our results from Tanga municipality may be an underestimate of MOIs and thus an overestimate of the unmet obstetric need.
Comparison of Population-based Caesarean Sections Rates according to the Health Information Management System and Study Data
The study was initiated by the Muhimbili National Hospital and Muhimbili University College of Health Sciences, Dar es Salaam, and implemented in cooperation with the health authorities in Tanga region and the Tanzanian German Programme to Support Health. Ethical approval was obtained from the University College; further approval was obtained from the Regional Authorities in Tanga Region and the Management Teams of the participating hospitals. The information collected in the questionnaires related to the presenting indications and actions taken and did not include any personal data.