Building a research registry for studying birth complications and outcomes in six Palestinian governmental hospitals

A prospective birth cohort study is being conducted including six governmental hospitals in Palestine: Rafidea hospital (Nablus), Palestine Medical Complex (Ramallah) and Hebron Hospital (Hebron) in the West Bank and Shifa Hospital (Gaza city), Shada Al Aqsa (Deir Ballah) and Al Helal Emirati Hospital (Rafah) in Gaza. The birth cohort study also includes an intervention study aiming to prevent perineal tears and will be described elsewhere. The project development phase started in 2013 and the initiative was launched in February 2015. Data collection started on the 1st March 2015 and was scheduled to end on the 30th April 2017.

All governmental hospitals were selected based on their high workload, geographical location and are all teaching sites for physicians, midwives and nurses. Table 1 presents some characteristics of the participating hospitals.

In summer 2014, the first draft of the CRF for documentation of relevant events during labour and delivery including complications and perineal tears in English was made. The contents of the CRF were based on previous research published from Palestine [21, 22] Norway [18, 19] and the World Health Organization [23, 24] (Additional file 1: presents a summary of the case registration form (CRF) and Additional file 2: The Case Registration Form (CRF) for Palestinian Perineum and Childbirth Study).

Data was collected on all birth events in the six hospitals and registered prospectively by the research teams, mainly by midwives. For this particular study, we used data collected between March 1st and December 31st, 2015. For operational reasons, data registration was restricted to labour and delivery wards; the registration of elective cesarean section was excluded. The reason was simply that women for elective cesarean section are usually admitted to the postpartum or gynecological ward, and are normally cared for by a separate team of midwives and nurses; not included in the initial training for this pilot. All midwives providing intrapartum care for women registered information on all births using the CRF, and the two midwives in each research team entered the data from the CRFs to the DHIS2 system on daily basis or as close as possible to the date of CRF completion. After the completion of data entry, the CRFs were locked up in cupboard prepared in each labor ward beforehand. Only research teams (two midwives) have access to the cupboard. All data were de-identified at point of data entry. Each CRF was given a serial number before data entry that was used to identify the entered CRF. The data was entered into an electronic system that is secured by individual passwords for research team members and investigators. The study was evaluated by the Regional Committee for Medical & Health Research Ethics, Section C, South East Norway, which found the Research Project to be outside the remit of the Act on Medical and Health Research 2008. The project can therefore be implemented without its approval, as long it is in done in accordance with common rules for health care services in Palestine and Norway regarding e.g. privacy.

DHIS2, created by department of global infrastructure at the University of Oslo, was used to collect data. It is worth noting that Palestine is the first country in the Arab World to use DHIS 2 in hospitals, but the system is for the time being also piloted for use in primary health care centers in the West Bank [25].

All healthcare providers working at the labour and delivery wards in the six hospitals were made aware of how to use the CRF form. The CRF form was initiated once the woman was admitted to labour and delivery ward and the data continued to be completed as labour process progressed. Each day the research team collects the completed forms, checks for completeness, enters them into the DHIS2 software, and keeps all forms in folders categorized according to serial numbers for each month.

To ensure quality (i.e. completeness and reliability) of the collected data, we undertook several measures. In addition to piloting the CRF and training on data registration and data entry, we also monitored the completeness of data collection and entry according to the hospitals’ registries of births and assessed the reliability of data registration and entry in three West Bank hospitals.

Monitoring the completeness of data collection and entry in six hospitals:

The data collection was monitored for completeness by various methods. Weekly field visits were conducted to monitor the process of data registration and identify any difficulties in registration. During field visits, the completed CRFs were reviewed to ensure absence of general and systematic errors in the form and completeness of the variables.

Additionally, a monthly auditing scheme was established to ensure data was collected on all eligible women. Each month, the numbers of registered CRFs in the hospital were cross-checked against the hospital birth register in order to ensure the completeness of data registration. Table 2 shows a summary of the monthly indicators monitored each month in the six hospitals. In order to observe completeness, errors and missing, all births entered in March to July 2015 were investigated for 22 variables using SPSS software. Feedback was given to midwives and research teams accordingly.

Based on our follow up in August 2015, 22 variables were selected to be audited for reliability. The selected variables are presented in Table 3.

A subsample of 80 birth events were audited in each of the three West Bank hospitals between March and December 2015, i.e. a total of 240 birth events. The 80 CRFs from each of the three hospitals were randomly selected. Given the difficulties accessing medical records in the Gaza hospitals, and the fact that Gaza has not yet electronic medical record system; the three hospitals in Gaza were not included in the reliability study. Thus, a subsample comprising of 240 CRFs were randomly selected from the three hospitals in the West Bank between March and December 2015 to be audited. Each birth event was audited for the selected 22 variables (a total of 1760 registrations in each hospital). The 22 variables were registered on a separate form based on information from three sources; the CRF, the DHIS2, and the electronic medical record in the hospital. The audit was conducted over two periods; 30 birth events (660 variables) from March to July, 2015 and 50 birth events (1100 variables) from August to December, 2015. Between the two periods, we communicated the results of the first patch to research teams and midwives in hospitals; particularly the missing and errors in data registration and entry, and encouraged them to improve the processes.

The collected and entered data for each variable was compared against the data documented from the MoH electronic medical record for completeness, similarities and errors. The variables were coded into four categories; correct (variable details are matched between the CRF, DHIS2 and MoH), error (variable details on the CRF or DHIS2 do not match the MoH system), missing (variable details are not documented on the CRF, or DHIS2) and not available (variable details are not available in the MoH system to be crosschecked). However, we compared details of these variables between the CRF and DHIS2. The variables; maternal education, weight, height, and number of antenatal visits were not available in the MoH system. The data was analyzed by IBM SPSS Statistics for Windows (version 21.0, Chicago, IL, USA). The unit of analysis was the variables.

A total of 34,482 deliveries occurred between March 1st and December 31st 2015 in the six participating hospitals (Table 2). There were 30,511 (88.5%) eligible birth events to be registered for this study (Table 2). Of all eligible birth events, there were 29,054 (95.2%) completed CRFs. The registration of data was complete in two hospitals (Hebron with (100%) and Aqsa with 99.9%), while the registration of data was less complete at the PMC (96.8%), Hilal Emirati (98.7%), Shifa (97.2%) and Rafidia (72.1%) hospitals (Table 2). Rafidia hospital consistently has the highest proportion of missing birth events registration every month during the data collection period (Fig. 2) (Fig. 1).

In general, as shown in Tables 4 and 5, the rate of the correct (matched between the three sources of information) data registration and entry was high in the three hospitals during the two periods of audit. During the first period of audit, the highest rate of errors in data registration and DHIS was observed at both Rafidia 47 (7.1%), and at the PMC 51 (7.7%), respectively. While the highest rate of missing in data registration and DHIS was observed at the PMC 54 (8.2%) and Rafidia 56 (8.5%), respectively (Table 4). During the second period of audit, the highest rate of errors was observed at the PMC in both data registration 93 (8.5%) and entry 104 (9.5%). While the highest rate of missing was observed at Rafidia in both data registration 127 (11.5%) and DHIS 60 (5.5%) (Table 5). Figure 2 summarizes the rates of errors and missing variables in the registered and entered data for the two periods among the participating hospitals in the West Bank.

As shown in Tables 6 and 7, in both Rafidia and the PMC hospitals, errors in data registration (CRF) and entry (DHIS2) were most frequent for the following variables; age, date of last menstrual period, gestational age, history during current pregnancy, previous medical history, cervical dilation upon admission, and duration of the first stage of labour. The highest percentage of errors in data registration (CRF) and entry (DHIS2) was observed at Rafidia and the PMC was for maternal age (11/29; 37.9% and 11/30; 36.7%) and (8/26; 30.8% and 13/30; 43.3%), respectively. While at Hebron the highest percentage of errors in data registration (CRF) was observed for oxytocin augmentation (7/30; 23.3%) and in the data entry (DHIS2) for current history (5/30; 16.7%).

Missing of registration and entry data were, however, more frequent in the following variables; history of current pregnancy, education, maternal weight, maternal height, oxytocin augmentation and augmentation duration as shown in the proportions (N) in Tables 6 and 7. The highest rate of missing in data registration (CRF) in Rafidia was observed for maternal education (29/50; 58%), at the PMC data on augmentation duration (9/30; 30%), and at Hebron hospital data on last menstrual period (6/50; 12%) Table 6. The highest rate of missing in data entry (DHIS2) was observed at Rafidia for maternal height (8/30; 26.7%), at the PMC for the variable on oxytocin augmentation and duration (8/30; 26.7%), and at Hebron hospital for the variable of the last menstrual period (4/30; 13.3%) Table 7.

As shown in Fig. 2, improvements were observed given decreasing rates of missing variables in data registration and data entry in the three hospitals between the two periods of audit, except for Rafidia hospital (i.e., the rate of missing variables in the CRFs increased; 7.8% to 11.5%). No improvement was observed in the rate of errors of variable registration and entry in the three hospitals, except for Alia hospital in Hebron (i.e., the rate of errors in data entry was slightly decreased; 6.1% to 5.8%).

Information on education, the number of antenatal care visits, maternal weight and maternal height, were not available in the MoH electronic information system, and thus, the comparison for completeness and quality was conducted between the CRF and the DHIS.

A total of 16 selected cases (eight from Rafidia, four from the PMC, and four from Hebron) could not be retrieved from the Ministry electronic information system to do the audit due to errors or insufficient registration of an identification number (i.e., ID or file number).