Stakeholder engagement and involvement.

The project was implemented by the National Health Research and Training Institute (NHRTI) in Ndola District (formerly Tropical Diseases Research Centre) and Macha Research Trust (MRT) in Choma District. Consultative meetings with hospital staff were held to understand the characteristics and operational details of the hospitals and laboratories to co-create the residual specimen collection that aligns with routine laboratory procedures. These staff included senior management, laboratory management, laboratory technicians, nurses and other clinical staff, and data staff involved in patient registration, data entry, and management of patient files. The main activities during the planning stage were gathering information on patient characteristics, patient referral patterns, patient registration and flow within the hospital, the number of blood specimens collected each month, and blood specimen and data flow. This information was gathered by abstracting aggregated data from facility registers and observing specimen flow at the hospital.

Characteristics of patients attending the hospitals.

Aggregated characteristics of patients seen at the hospital, such as age, diagnoses, HIV prevalence, and geographic distribution of the health facilities referring the patients, were summarized to understand those seeking care at the hospital and how this population compares to the general population of the district (S2 Table). Data on whether the patient resided within or outside the district or province were limited in the planning stage, so it was not possible to assess if parts of the district were under-represented.

Laboratory specimen flow at health facilities.

After the hospitals were selected, characteristics of the different laboratories were collected, including the types of whole blood specimen tubes received (e.g., presence of anti-coagulants) and tests performed in each laboratory. Based on this, specialized laboratories, such as microbiology, virology, and parasitology, were excluded. For example, as our primary goal was to estimate measles seroprevalence, we excluded specimens from the virology laboratory which only conducted HIV viral load testing and all specimens received from the antiretroviral treatment (ART) clinic to avoid oversampling specimens from people living with HIV, which could bias seroprevalence estimates. Hematology laboratories were selected as the source of residual specimens given the larger number of blood specimens received and the observation that many specimens sent for biochemistry testing were also sent to the hematology laboratory, potentially resulting in duplicate specimens from the same patient.

Specimen characteristics at health facilities.

For planning purposes, we compiled the number of specimens received by month, age group, and ward (inpatient or outpatient) for four months during the preceding year, and then estimated the number received per day. In two facilities this information could be summarized using an electronic database but for the other facility this had to be manually summarized using the paper-based register in the laboratory. In laboratories with a large estimated daily number of specimens received, we systematically sampled to reach a daily target and avoid excluding patients based on when their specimens were received in the laboratory. However, in one facility the number of pediatric specimens received in the laboratory each day was small so all eligible specimens were collected.

Specimen storage after clinical testing.

To develop the protocol, we mapped how the laboratory handled specimens after routine testing was completed, including how the specimens were stored and for how long. Two facilities stored specimens overnight at 2–8°C after testing, so all specimens received in one day were eligible to be processed the following day. One facility stored specimens at room temperature after testing due to limited refrigerator space. To maintain specimen quality at this facility, we restricted specimens to those received on the same day as processing, excluding specimens received between the time of processing and midnight. To assess if these specimens were biased, we purposefully collected “after hours” specimens that would have been excluded because they were collected in the evenings or weekends (depending on the procedure at that facility) during a one-month period for a sensitivity analysis. Additional information was collected during the planning phase on the availability of space and equipment in the laboratory, such as a centrifuge and −20°C freezer, to determine if special arrangements were needed.

Data flow at health facilities.

To assess the feasibility of data abstraction, we reviewed the laboratory registers and consulted data management staff at the facilities who were familiar with the patient record filing system and data flow to the laboratory. Key data points to be collected for residual specimens included age or date of birth, date of specimen collection, sex, and source of the specimen (inpatient or outpatient), all of which were available in the laboratory registers for most patients. We were also interested in the patient’s place of residence (inside or outside of the district and province; compound or village within the district) and history of measles vaccination for children to explore potential biases in the patient population relative to the district’s general population. However, these variables were not readily available at the laboratory. Separate arrangements, such as engaging the existing health facility data staff and reviewing clinical records, were needed to collect deidentified information for residual specimens. We observed that the availability and source of these data may vary by the type of patient and flow of patients at the facility.

Training and initiation.

Following the development of procedures based on formative research conducted at each facility, on-site training for laboratory and data management staff was conducted, and supplies for residual specimen collection and processing (e.g., pipette tips, cryovial tubes and boxes, registers) were distributed to the facilities. The training was also used to refine workflow patterns. The local implementing partners remained on site for approximately one week following training to observe procedures and support facility staff.

Specimen selection and identification of eligible specimens by age.

Eligible specimens were defined as whole blood specimens collected in ethylenediaminetetraacetic acid (EDTA)-containing tubes within the age groups of interest (Table 1), collected on the same or prior day and stored at 2–8°C, had sufficient volume (approximately 0.5 ml of whole blood), and no evidence of severe hemolysis. Specimens which did not meet this criterion were excluded (Table 3). At each facility, laboratory and data staff were engaged in the residual specimen collection and data abstraction activities, in addition to their routine responsibilities (e.g., processing specimens, performing routine clinical and diagnostic testing, registering patients, preparing surveillance reports).

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Residual specimens identified and excluded per day.

https://doi.org/10.1371/journal.pone.0341389.t003

Although the generic flow of residual specimens was similar across facilities (Fig 1), the specific procedures for identifying specimens were influenced by how specimens were stored after testing, the daily number of specimens received in the laboratory, age groups of interest, and the format of specimen metadata in the laboratory (S1 Table). For example, one facility organized specimen tubes by age group to facilitate systematic sampling, which required abstracting age from the electronic laboratory information system as age was not available on the specimen tubes. A second facility primarily relied on paper-based laboratory request forms to identify eligible specimens. Where age was required to organize specimens, either from electronic or paper-based forms, identifying age-eligible individuals and locating or selecting specimens required at least one hour. In the facility where only adult specimens were selected, specimens were first systematically sampled using the specimen tubes, then age eligibility was confirmed using the electronic record system, which reduced the time required for specimen identification by about half. The systematic sampling interval varied based on the anticipated number of specimens received in the hematology laboratory per day based on the historical inventory and the target number of residual specimens to be collected per day. The sampling interval was reassessed periodically and adjusted as needed if the observed number of specimens received differed from the historical data or if the number of residual specimens collected per day needed to be increased or decreased.

Recording the patient’s age on the specimen tube labels would enable technicians to easily organize specimens and substantially reduce the time to identify and select eligible specimens. However, this would require considerable support and buy-in from facility management to implement as a standard practice or to improve adherence if already implemented but not consistently done. Alternatively, specimens could be organized by age groups as routine testing is completed so they are properly arranged for residual processing, or an electronic record system could be used.

Quality control checks for specimens.

Once specimens were selected, they were manually cross-checked against the list of specimens pulled during the prior two weeks to determine if the patient’s sample was recently processed as a residual specimen; if so, the specimen was excluded to reduce biases in the seroprevalence estimate. A two-week window was selected to reduce burden when searching for duplicates and assuming the patient’s serostatus may have changed outside that time frame due to exposure or vaccination. Specimens were also checked to ensure eligibility criteria were met. Excluded specimens were replaced if additional eligible specimens were available.

Most duplicates were from inpatients. Limiting duplicates check to inpatient specimens could improve efficiency. Using an electronic medical record system to assess if the selected patient had multiple specimens sent to the laboratory within the past two weeks would further streamline the process of excluding duplicate specimens.

Specimen processing.

The selected whole blood specimens were labeled with study identification numbers, covering any identifiable information on the tube, centrifuged, aliquoted into two labeled cryovials, placed in freezer boxes, and stored at -80^o C until transported to the central laboratory for testing.

The frequency of specimen transport from the health facility to the laboratory at the research institutes differed based on the distance between the two locations. For facilities located far from the laboratory, specimen transport was combined with trips for other purposes.

Data systems.

There were many challenges identified in the data abstraction process. Data staff needed to access multiple data sources to abstract the requested data for a single specimen. One hospital had an electronic laboratory management system but locating and abstracting linked data for residual specimens using this system was challenging. Therefore, staff needed to manually search records to abstract data from the medical registers or in the individual patient files. At a second hospital, an electronic medical record system was available for outpatients but data for all inpatients and certain data points for outpatients needed to be abstracted from the paper-based ward registers. At the third hospital, all data abstraction was done using the paper-based ward registers, except for the patient’s referring facility. This information was only available at registration; however, where registration took place varied based on how the patient entered the hospital. In some situations, data abstraction staff needed to go to other sections located far apart to access the relevant registers.

Nurses did not report disruptions due to the presence of the data staff manually searching records; however, moving records within the facility could be problematic. The staff made multiple attempts to locate records. For example, if the record had not been filed or was with the medical staff during the first attempt, staff needed to adapt to the timing of ward rounds. Complex filing systems for patients using private insurance also complicated the process of locating records for individuals insured under someone else’s plan. The staff encountered missing records because some patients took their records and others entered the hospital through private arrangements and were not properly registered. On average, locating the patient files and data points took 1–3 days including reattempts, but at times required up to a week. Staff reported that the process of locating records became more challenging as more time passed since collection, so timely communication between the laboratory and data teams was important.

Modifying procedures for recordkeeping, such as requiring clinicians to provide prescription notes so patients do not need to take their medical records to the pharmacy may benefit both clinical care and residual specimen data abstraction.

Data points.

Certain data points were more problematic to abstract than others (S3 Table). Measles vaccination status proved especially challenging, requiring staff to access individual patient files and separate registers. At one facility, few children had vaccination data available because it was not consistently recorded by clinicians. The other facility reported high under-5 card availability for pediatric patients, although clinicians only recorded if the child was fully immunized or up to date; in these circumstances, we assumed the child had received the measles vaccine. Missing data on age were more common for adults where it may simply be recorded as “adult” if the patient did not know their date of birth and was unable to estimate their age. Missing data were also more common for patients in the emergency department, especially if the person was not accompanied by a close relative who knew that information.

Limiting the requested data to those recorded on the laboratory request form would greatly simplify data abstraction, substantially reduce the time required to collect key information, and facilitate implementation of a residual specimen system that is more sustainable. All data abstraction could be completed in the laboratory by the laboratory staff, removing the requirement for a separate data team or the need to access multiple data sources and minimize the time staff had access to identifiable information. In facilities where age is not consistently recorded on laboratory request forms, this would require significant collaboration with the laboratory manager and facility leadership to reinforce that this information must be recorded and to reject specimens missing this information. If there are concerns about the representativeness of the seroprevalence results, periodic “pulse demographic data checks” could be conducted to abstract data on variables not available at the laboratory, such as referring facility, place of residence (e.g., district, village or ward), and vaccination status. These data could be used to explore changes over time in the patient characteristics, which may influence interpretation of the seroprevalence estimates obtained from the residual specimens.

Data staff and resources.

Utilizing data staff familiar with the systems at the facility was critical to facilitate data abstraction, especially given the complexity of locating and accessing registers and clinical records. Having sufficient data staff was important to ensure staff were available to support data abstraction without disrupting routine patient registration.

Other logistical challenges included access to reliable internet to access the electronic medical record system and upload data. Lack of dedicated computers for the study meant that staff had to wait for the workstations to become available to abstract data from the electronic medical record system. Providing dedicated computers and internet bundles to support data abstraction would facilitate data abstraction.

Lessons learned integrating into Routine Hospital Workflows.

Hospitals required functional hematology laboratories, refrigeration capacity for short‑term storage, and electronic laboratory information systems, though reliance on paper‑based records at some sites contributed to delays clinical records took 1–3 days on average to locate. Three facilities successfully integrated residual specimen collection into existing laboratory processes with minimal disruption, and staff were able to complete specimen identification, processing, and metadata abstraction alongside routine duties. At Choma General hospital, an average of 8.8 specimens per day were identified in the hematology laboratory requisition forms (range: 1–22), reflecting a steady supply of residual specimens suitable for serosurveillance. However, approximately 2.6 (27%) of identified specimens were excluded, primarily due to low volume, hemolysis, duplicate samples, or inability to locate the physical specimen indicating the need for improved standardized specimen handling and record‑keeping systems (Table 3). Facilities differed in their daily specimen volumes and data availability, but all were able to identify eligible samples across the target age groups, yielding 2,365 total pediatric samples and 960 adult samples, with missing age data ranging from 3 to 69 specimens per site (Table 1).