Replication Data for Maternal Near Misses and their Determinants Among Women Who Sought Obstetric Care from Fort Portal Regional Referral Hospital, Western Uganda

1.0 Sampling procedures
Fort portal regional referral hospital was purposively sampled. The fact that the study was conducted at only one facility, with interest in only one department (maternity), the sampling process was conducted at a single stage, that is, at and within the aforementioned department. A request to be availed with all clinical records of mothers who received antenatal care and delivered from Fort Portal regional referral hospital between the years 2017 and 2019 was made to relevant authorities at the hospital. Upon reception of those clinical records, that was in upwards of 3000, they were split into batches of 1000, in order to make the sampling process more manageable. Given the large number of patient files, it was feasible to use a systematic random sampling. With this technique, a sampling interval was calculated, so as to determine the skip (K), that will have to be observed, while sampling patient files, in each of the piles of 1000 that was organized. The sampling interval was calculated using the formula; K = N/n, where;
K = Sampling interval
N = Target sample size at facility
n = Sample size
The formula yielded an interval of; K = 3672 / 375 = 9.7 ≈ 10
Following the calculation of the sampling interval, the principal investigator started the sampling process with any of the piles of patient records, in which the first file (on top of the pile) was made the starting point, following which the interval (skip) of 10 was observed and the next file sampled. That process was continued until 203 patients are sampled. The sampled files were each screened for eligibility, and any file found to be illegible was replaced immediately.

2.0 Data Collection
With the justification for the choice of a retrospective chart review design, the study did not have to rely on patient interviews. Therefore, the fact that the study had to rely on secondary patient data, the data collection method of choice for this study had to be medical record abstraction (MRA). With medical record abstraction, direct matching of information in a given patients record, with the entry required on the data abstraction form was made . The data abstraction form was designed in a structured manner that is, with close-ended items that later made it easier to quantify and come up with inferences.

1.0 抽样流程
本研究采用立意抽样法选取福特波特拉区域转诊医院（Fort Portal Regional Referral Hospital）。由于本研究仅在单一场景开展，且仅关注产科一个科室，因此抽样流程仅设置单阶段，即在上述科室内部及周边开展。本研究向医院相关管理部门提交申请，获取2017至2019年间在福特波特拉区域转诊医院接受产前保健并分娩的产妇的全部临床病历。在获取到共计3000余份临床病历后，为便于抽样操作，将其按每1000份为一组进行分组。鉴于患者病历总量庞大，本研究采用系统随机抽样（systematic random sampling）法。采用该方法时，需先计算抽样间隔以确定抽样步长（K），在每一组1000份病历的抽样过程中均需遵循该步长。抽样间隔通过以下公式计算：K = N/n，其中：
K = 抽样间隔（Sampling interval）
N = 机构目标样本量（Target sample size at facility）
n = 实际抽取样本量（Sample size）
经计算，抽样间隔为K=3672/375=9.7≈10。
完成抽样间隔计算后，首席研究员以任意一组病历为起始，以该组病历的堆顶第一份作为抽样起点，按照10的步长依次抽取下一份病历，重复该过程直至抽取203名患者的病历。对每份抽取的病历进行纳入资格筛查，若发现不符合纳入标准则立即更换。

2.0 数据采集
鉴于本研究采用回顾性病历审查设计，无需依赖患者访谈。由于本研究需使用患者的二手数据，因此选取病历数据提取法（Medical Record Abstraction, MRA）作为本研究的数据收集方法。采用病历数据提取法时，需将患者病历中的信息与数据提取表中所需填写的条目进行直接匹配。本研究的数据提取表采用结构化设计，仅包含封闭式条目，便于后续进行量化分析与统计推断。