Reporting of rapid diagnostic testing in Kenya's health system, 2018-2020: triangulation of routine and survey data

doi:10.21203/rs.3.rs-2103027/v1

Download PDF

Research Article

Reporting of rapid diagnostic testing in Kenya's health system, 2018-2020: triangulation of routine and survey data

https://doi.org/10.21203/rs.3.rs-2103027/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 30 Mar, 2023

Read the published version in BMC Health Services Research →

You are reading this latest preprint version

Background

Understanding access to and availability of rapid diagnostic tests (RDTs) is essential for attaining universal health care and reducing health inequalities. Although routine data helps measure RDT coverage and health access gaps, a significant proportion of healthcare facilities fail to report their monthly diagnostic test data to routine health systems, which affects routine data quality. This study sought to understand whether non-reporting by facilities is due to a lack of diagnostic and service provision capacity by triangulating routine and survey data in Kenya.

Methods

Monthly facility-level data on ten RDTs (malaria, HIV, syphilis, blood transfusion, anaemia, diabetes, pregnancy, urinary tract infections, kidney disease, and meningitis) was sourced from Kenya's health information system (2018–2020), and 2018 harmonised health facility assessment. After harmonization and linking the two data sources, we compared reporting in the routine system among facilities with i) diagnostic capacity only and ii) both confirmed diagnostic capacity and service provision. We also assessed trends in reporting for facilities without diagnostic capacity. Analyses were conducted at the national level, disaggregated by RDT, facility level and ownership.

Results

Twenty-one per cent (2821) of all facilities expected to report routine diagnostic data in Kenya were included in the triangulation. Majority (86%) were primary-level facilities under public ownership (70%). Overall, survey response rates across facilities were high (> 70%). Malaria and HIV had the highest response rate (> 96%) as well as the broadest coverage in diagnostic capacity across facilities (> 76%). Reporting among facilities with diagnostic capacity varied by test, with HIV and malaria having the lowest reporting rates, 58% and 52%, respectively, while the rest ranged between 69% and 85%. Among facilities with both service provision and diagnostic capacity, reporting ranged between 52% and 83% across common tests. Public and secondary facilities had the highest reporting rates across all tests. A small proportion of health facilities without diagnostic capacity submitted testing reports in 2018, most of which were primary facilities.

Conclusion

Non-reporting in routine health systems is not always due to a lack of capacity. Further analyses are required to inform other drivers of non-reporting to ensure reliable routine health data.

rapid diagnostic test

routine data

health facility survey

triangulation

Kenya

Diagnostic testing is vital to improving the standard of care in any health care system. However, diagnostics account for the biggest disparity in the continuum of care for different health conditions [1, 2]. Access to accurate and timely diagnostics minimises reliance on clinical symptoms for diagnosis. It supports more targeted treatments that result in better disease management and effective use of limited resources, particularly in low- and middle-income countries (LMICs) [1, 3, 4]. Therefore, delays in diagnosis caused by limited access or inadequate capacity could undermine efforts made globally to achieve the sustainable development goals (SDGs) on health and lowering disease morbidity and mortality [5, 6].

About half of the world's population has limited access to diagnostics [1]. However, if this gap was narrowed by 10% across select priority conditions, it is estimated that approximately 1.1 million premature deaths would be averted in LMICs [1]. This emphasises the potential impact of improved access to diagnostics, particularly in primary health care, in accelerating universal health coverage (UHC) [1, 2, 7]. Further, across LMICs, heterogeneities in access to diagnostic tests exist between countries and across facility levels [5, 8]. Although primary healthcare facilities are the first point of care, they have the largest diagnostic capacity deficit [9]. Only about 19% of the population in LMICs have access to basic diagnostic testing, excluding malaria and HIV. For instance, in 2004–2018, ten LMICs (including Kenya), overall test availability for basic primary care was 19%, while advanced primary care and hospitals were 49% and 68%, respectively [3].

Rapid Diagnostic Tests (RDTs), which are point-of-care diagnostics, provide results with shorter turnaround times than conventional laboratory-based testing and do not require sophisticated equipment [10]. These tests are beneficial for filling gaps in diagnostics in LMICs [6, 9]. Improvements in diagnostic support over the years have been concentrated on key programmatic areas like HIV/AIDS, TB, and malaria in LMICs [3, 11]. However, this only covers a portion of the broad diagnostic spectrum; thus, there are still substantial availability and quality gaps, even for diseases with a high priority for public health [3, 5, 12].To address this problem across countries, since 2018, World Health Organisation (WHO) has published an annual essential diagnostics list (EDL), a suite of recommended diagnostics that should be available at point-of-care, laboratories and across all health tiers to increase timely and life-saving diagnoses [7]. The Lancet Commission further suggests the development of an evidenced-based integrated and tiered network and a national EDL across countries to inform their national diagnostics strategy [1].

A comprehensive understanding of access to, and availability of RDTs in LMICs, remains a crucial priority in advancing the UHC agenda [7] and reducing accessibility gaps. Routine data [13–15] have been widely used in monitoring progress in health service provision, including laboratory testing across health systems [3]. For example, test volumes of RDTs across health facilities can be compared with the population needing diagnostic testing to determine unmet needs, ensuring efficient use and allocation of resources in the community. More importantly, to improve the utility of routine health data and deal with some of its known limitations, triangulation of facility data with health facility surveys is a valuable data validation approach [3, 16]. Sample health facility surveys offer alternative sources of information in understanding facility capacities beyond facilities' routine data from the health management information System (HMIS). Such surveys assess the readiness of healthcare facilities to provide services [17], for example, the most recent (2018) Kenya Health Facility Assessment (KHFA), a nationally representative facility assessment that evaluated various aspects of service and diagnostic availability through in-person facility visits [18]. Such surveys provide a rich set of cross-sectional data for a sample of facilities; however, they are conducted infrequently and only focus on select health services.

Routine diagnostic testing data reported monthly by facilities through the HMIS based on the District Health Information Systems 2 (DHIS2) platform is more ubiquitously available. Since 2011, the platform has provided routine data on tests done at a facility level every month in Kenya. Despite the availability of routine data across the country for an entire year, data quality has been a concern, primarily due to non-reporting and incomplete entries across health facilities [13, 15, 19–23]. Dealing with these challenges requires an understanding of the causes driving non-reporting. Here we triangulated between routine data reported through a HMIS and a cross-sectional health facility assessment survey to assess whether a lack of diagnostic capacity drove non-reporting in Kenya's DHIS2 at the facility level for ten RDTs between 2018 and 2020.

Country context

Kenya is in the East Africa region with a total population of 47.6 million, based on its 2019 census, and an inter-censual growth rate of 2.2% [24]. The average population density was 82 per square kilometre (sq. km) and was highly variable at the sub-national (county) level ranging from less than 20 people per sq. km to over 500 people per sq.km. Population distribution has major consequences on health policy and management decisions relating to allocating resources, distribution of health facilities, and diagnostic availability. Since 2013, Kenya's health sector has operated under a decentralised governance structure where functions of administration and sub-national health care provision decisions were devolved to 47 county governments. However, the management of tertiary health facilities, regulatory processes, and health policy direction was retained at the national level.

Health provision in Kenya is structured into six tiers of increasingly complex service delivery, namely community-level services (level 1), primary health care consisting of dispensaries, clinics (level 2) and health centres (level 3), sub-county hospitals and medium-sized private hospitals (level 4), primary referral facilities (level 5), and tertiary facilities (level 6) [25]. Health facilities are owned and managed by the government, non-governmental institutions, faith-based organisations or private-for-profit entities. Medical laboratories in Kenya also fall under similar ownership categories and may be located either within a health facility or as stand-alone facilities, both of which are regulated by the Kenya Medical Laboratory Technicians and Technologists Board (KMLTTB) [13]. The diagnostic tests available in Kenya range from simple RDTs to complex laboratory tests, the availability of which is variable and may correspond to the ownership and level of the health facility.

Methodology Overview

Figure 1 summarises the overall approach applied to achieve the objective. Data on ten RDTs from DHIS2 for the years 2018-2020 and the 2018 Kenya health facility survey were assembled at the facility level. The variables within the survey were then harmonised and merged with the DHIS2 data. Descriptive analysis was used to summarise the overall diagnostic capacity and service provision at a national level for all ten RDTs based on the cross-sectional survey, the gold standard. The analysis was disaggregated by RDT, facility level and ownership at the national level. We hypothesised that all facilities reported to provide service and/or had diagnostic capacity ought to have submitted a monthly DHIS2 diagnostic report. Consequently, among the facilities with confirmed capacity for i) testing, ii) or both testing and service provision, we computed the proportion of facilities that submitted a report in DHIS2. Data reported in DHIS2 in 2018 was also validated against diagnostic capacity in KHFA.

Figure 1

Data assembly

RDTs: Laboratory testing data for health facilities is submitted to DHIS2 using the reporting tool MoH 706, which encompasses 91 diagnostic tests available in Kenya. Guided by the most recent WHO EDL [26] details on assay format for different tests and local clinician knowledge, we identified 19 RDTs from the MoH 706 form. We define an RDT as a point-of-care test that is quick and easy to perform, especially when an already processed sample is available. These RDTs were then compared with tests included in KHFA to obtain a list of RDTs for the triangulation. We included an RDT in the triangulation list if it was reported through DHIS2 and on either diagnostic capacity or both diagnostic capacity and service provision were evaluated in the KHFA.

Based on the KMLTTB checklist supplemented with local lab technician\technologist knowledge and informed by literature, we assigned the facility levels at which each RDT was expected and classified tests as common or uncommon [13]. The triangulation list comprised ten RDTs: malaria, HIV, blood grouping, urine chemistry, Treponema Pallidum Hemagglutination (TPHA), Venereal Disease Research Laboratory (VDRL), Human Chorionic Gonadotropin (HCG-pregnancy), HB estimation, blood sugar (random blood sugar) and Cryptococcal Antigen (CRAG) tests (Table 1). Nine RDTs that were excluded (oral glucose tolerance test, rheumatoid factor, brucella, H. pylori, hepatitis A, B and C tests, community health worker (CHW) malaria RDT and anti streptolysin o titer test) were reported through DHIS2 but not assessed in the KHFA. CHW malaria RDT lacked details on diagnostic capacity from the KHFA, hence excluded.

Table 1: Rapid diagnostic tests reported in DHIS2, whose availability was assessed in the 2018 KHFA survey

RDT Test	Health outcome	Common	Level of reporting	EDL classification
HIV	HIV	Y	All levels	Sexually transmitted infections
Venereal disease research laboratory (VDRL)	Syphilis	Y		Sexually transmitted infections
Blood grouping	Blood Transfusion	Y		Haematology
HB estimation	Anaemia	Y		Haematology
Blood Sugar	Diabetes	Y		Clinical chemistry
HCG (Human chorionic gonadotropin)	Pregnancy	Y
Urine Chemistry	UTI, Kidney Disease	Y
Cryptococcal Antigen (CRAG) test*	Cryptococcal Meningitis	Y
Malaria RDT	Malaria	Y	Level 2-3	bacteriology, mycology and parasitology
Treponema Pallidum Hemagglutination (TPHA)	Syphilis	N	Level 4-6	Sexually transmitted infections

* Mostly available at facilities offering antiretroviral therapy (secondary facilities), UTI. - Urinary tract infections

Diagnostic health facility database: We aimed to assemble a geocoded database of health facilities of all service providers expected to submit diagnostics data to DHIS2. To achieve this, we harmonised Kenya's two main lists of health facilities, the Kenya master health facility list (KMHFL) and DHIS2 (Moturi et al., unpublished observation). We retained all facilities expected to routinely carry out diagnostic testing, including details on facility level and ownership, MFL code and DHIS2 identifier. Health facilities excluded were mainly medical stores and depots, specialist clinics, treatment centres, and rehabilitation centres where diagnostic testing is not routinely carried out.

We accessed DHIS2 through MoH authorised login [27] on 25 November 2021. From DHIS2, we assembled data on diagnostic testing, which is reported within the MoH 706 tool. These data were extracted monthly at the facility level for 2018-2020. We excluded DHIS2 data submitted after December 2020 from our analysis because KHFA was conducted in 2018, thus would result in a substantial temporal mismatch. We included data from 2020 to understand reporting biases attributable to the COVID-19 pandemic [15]. Data sourced from DHIS2 was merged to the health facility database using the unique DHIS2 identifier. The final database contained a list of all health facilities showing the number of monthly RDT tests submitted by facilities.

Kenya harmonised health facility assessment survey: The KHFA conducted from November to December 2018 is a comprehensive survey of health facility capacity across themes of service availability and readiness, quality of care and management. The KHFA used the KMHFL as its sampling frame, comprising 10,535 health facilities. The selection of facilities was stratified across counties, facility type, managing authority and settings (rural or urban), resulting in 2,980 facilities randomly selected. All facilities at levels 5 and 6 and all public level 4 hospitals were included in the sample.

The survey entailed in-person facility visits and interviews with the most knowledgeable staff in each service area. Availability of diagnostic items was determined for the day of the survey and stockouts for the preceding 3 months of the survey period. Relevant information on diagnostics and service provision was extracted for each facility's responses in the following questionnaires: laboratory and blood transfusion, emergency, Maternal and Child Health (MCH) and family planning, maternity and newborn, and HIV service area. Relevant questions from these questionnaires on test and service availability are summarised in Additional file 1 (Tables 1 and 2). Data were collected electronically on Android phones for 2,927 (98%) out of the target 2,980 facilities.

Data harmonization

Utilising the MFL code that uniquely identifies facilities, the KHFA dataset [18] was linked to the DHIS2 health facility database. A total of 2,821 facilities were matched successfully, excluding 106 facilities, of which 100 health facilities (95 private-for-profit and 5 public health facilities) in KHFA were not registered in DHIS2, and thus data on testing could not be obtained. The other six were specialist facilities (ENT and dermatology) and were not expected to conduct diagnostic tests of interest routinely. Private-for-profit health facilities might not be enrolled in DHIS2 since enrolment requires them to submit monthly reports to DHIS2, which might increase workload. On the other hand, public facilities may have experienced delays in registration.

Questions in the KHFA on test and service availability (Additional file 1, Table 1) were duplicated across different departments in a facility or disaggregated for select populations (e.g., HIV testing for TB patients, adolescents, children etc.). This phenomenon was further compounded by fragmented data collection during the survey, resulting in variable completeness levels and occasionally conflicting information in related questions [18]. Consequently, for purposes of triangulation, we opted to consider the broader availability of a service or test within the facility. We use the term diagnostic capacity to refer to overall RDT availability (valid and expired test kits) within a facility. In contrast, service provision refers to a facility's general capacity to screen for, diagnose and/or treat a given disease. We thus referenced and merged related questions to obtain final responses for diagnostic capacity and service provision. Any confirmation of test availability or service provision in pertinent questions was considered definitive (Additional file 1, Table 1 & 2). After harmonisation, facilities with no response in either category were treated as missing entries and excluded in the consequent analysis.

Triangulation

We independently analysed the harmonised responses for diagnostic capacity and service provision for each RDT with reports from DHIS2 for 2018-2020 and 2018. The analysis over three years controlled for disruptions that might have contributed to non-reporting, such as stock-outs or improved reporting in DHIS2 in the post-survey period. We carried out triangulation using two separate scenarios; a comparison of reporting in DHIS2 (2018-2020) among facilities with i) diagnostic capacity only and ii) both confirmed diagnostic capacity and service provision based on KHFA. In the second scenario, total RDTs with diagnostic capacity and service provision data were lower since not all facilities reporting services reported had the corresponding RDTs.

As a first step, we undertook a descriptive analysis of response rates in KHFA for diagnostic availability and service provision questions for each RDT. For each of the two scenarios, among the health facilities with established capacity, we computed the proportion that submitted reports (>=1 test) in DHIS2. We also assessed whether any facilities with no diagnostic capacity as per KHFA submitted any report in DHIS2 in 2018, the survey year. All analysis was conducted per RDT at the national level and disaggregated by facility level and ownership using R software environment (Version 4.2) and StataCorp. 2021 [Stata Statistical Software: Release 17. College Station, TX: StataCorp LP].

The total number of health facilities used for triangulation was 2,821, geographically distributed across all 47 counties (Fig. 2). This sample represented 21% of all health facilities expected to submit diagnostic reports in the DHIS2 in Kenya. The majority (70%) were public facilities owned by either the government, non-governmental or faith-based organisations, while private-for-profit was 30% (845). The primary health facilities were 2,412 (86%), while secondary and tertiary facilities were 409 (14%).

Figure 2

The number of health facilities responding to diagnostic capacity and service provision-related questions in the KHFA survey varied by test and by facility (Table 2). Responses related to service provision were available for five RDTs (Table 2). Overall, responses to service provision were above 86% (2,427 facilities) for 6 out of 7 of the services of interest. Responses were highest (100%) for HIV testing for ANC, malaria and HIV diagnosis, while the lowest response rate at 71% was for diabetes diagnosis (Table 2). On the other hand, facility diagnostic capacity was reported for 10 RDTs. For eight of the ten RDTs, over 79% of facilities reported their diagnostic status. Similar to service provision, malaria and HIV RDTs had the highest response rate of over 96% representing 2,700 out of 2821 facilities forming the sample. The status of CRAG and blood grouping tests were the least reported, with less than 45% of facilities responding (Table 2).

Table 2

Response rate to service provision and test availability questions after harmonisation of variables(N = 2821).
RDT	Services Available	Services provision (%)	Test availability (%)
Syphilis	Syphilis Diagnosis	86.4	79
	Syphilis ANC Testing	86	79
Blood Sugar	Diabetes Diagnosis	71	91
HB Estimation	Anaemia Diagnosis	92	79
Malaria	Malaria Diagnosis	100	97
HIV	HIV Diagnosis	99.8	96
	HIV ANC Testing	100	96
Urine Chemistry	N/A	N/A	91
HCG			91
CRAG			44
Blood Grouping			34

Across all common tests, most facilities with confirmed testing capacity submitted reports of testing in DHIS2 (Fig. 3). Further, the proportion of facilities with testing capacity that made a report varied by the test with HIV and malaria having the lowest proportions reporting at 58% and 52%, respectively, while the highest proportions in reporting were for CRAG and blood grouping tests at 85%. The reporting of the other five common tests ranged between 69% and 79%. On the other hand, TPHA (an uncommon test) was not reported in most facilities (71%) despite having reported capacity (Fig. 3).

Figure 3

Heterogeneity was observed in reporting across ownership and facility levels among the facilities with confirmed diagnostic capacity (Fig. 4). The private and primary care health facilities had poorer reporting than public and secondary health facilities. Across all tests, primary facilities with capacity have less than 81% making a report in DHIS2, which is even lower for private facilities with less than 62% reporting (Fig. 4). Comparably, secondary and public facilities each had over 80% of facilities with capacity, reporting to DHIS2 in 6 out of 10 tests. When further disaggregated by ownership, patterns of reporting by primary and secondary facilities reveal similar trends (Additional file 1, Figs. 3 & 4).

The availability of some RDTs among facilities with diagnostic capacity corresponded to the expected reporting trends when analysed by facility ownership status or level. In the case of tests for priority programmes, HIV and malaria, reporting in DHIS2 showed poorer reporting in private (40% and 49%) compared to public facilities (56% and 65%), respectively (Fig. 4, bottom panel). Conversely, uncommon tests such as TPHA, with the lowest reporting rates in all facilities (Fig. 3), and CRAG both have significantly better reporting in higher-level facilities (52% and 91%, respectively) where they are expected to be available (Fig. 4, top panel) compared to primary level facilities (23% and 68%).

Figure 4

Responses to service provision questions in the survey showed that despite a facility offering a service, it did not always have a corresponding RDT; most visible differences were observed in TPHA (Fig. 5, left panel). Facilities with both the service and the corresponding RDT were highest for HIV, with up to 2,435 (97% of facilities providing HIV services) facilities having both (Fig. 5, left panel). The range of facilities having both service provision and diagnostic capacity across other tests was between 1,482 (89%) for blood sugar and 2,309 (96%) for HIV ANC in 6 tests, with Hb estimation having the least at only 724 (35%). Nationally, among health facilities with both confirmed service provision and diagnostic capacity as per the survey, the proportions of facilities submitting test reports to DHIS2 ranged from 52–83% across common tests. Notably, HIV and Malaria RDTs had substantially low reporting rates in DHIS2 (58% and 52%) despite recording high diagnostic capacity and the largest sample size in the KHFA (93% and 76%), respectively. Similar to previous trends, the lowest reporting was associated with TPHA.

Figure 5

When data on diagnostic capacity from the survey were compared to reports in DHIS2 for 2018, there were minor differences in the information provided by the two sources. Albeit a small proportion, several health facilities without the diagnostic capacity for specific tests, according to the survey, reported conducting the test in DHIS2 in 2018. This varied across RDTs, with the most significant inconsistency exhibited in HB estimation, with 19% of facilities without recorded testing capacity in the survey reporting in DHIS2 (Fig. 6).

Figure 6

Routine health data is essential for quantifying health care utilisation, estimating the reach of interventions in the community and monitoring progress toward national and global targets such as SDGs [28]. However, data quality concerns, primarily due to the non-reporting of health facilities, have continued to persist, impacting the accurate assessment of the performance of a country's health system [15, 19-21, 23]. These concerns become even more significant when attempting to estimate the current supply and demand for point-of-care testing to guarantee an adequate diagnostics supply. Consequently, to address quality concerns in routine data reported through DHIS2, we sought to establish whether the non-reporting of health facilities in DHIS2 is due to a lack of capacity triangulating diagnostic test reports made in DHIS2 with health capacity information from a comprehensive facility survey, KHFA.

Our findings show that most facilities that offer both diagnostic services and have requisite RDTs reported testing on the DHIS2 platform. However, about 17-48% of the facilities did not report, and yet they could conduct the common tests (Figure 5). More importantly, despite routine tests, HIV and malaria RDTs had 42-48% of facilities with RDT kits available at the time of the survey failed to report testing in DHIS2 over three years. When the analysis was restricted to data from 2018 to ensure congruence in temporal comparison, a similar pattern was observed, with 38% and 43% of facilities failing to report for HIV and malaria, respectively. Applying the same temporal restriction (2018), across the other RDTs, facilities failing to report despite having RDT kits were heterogeneous, ranging from 2% (CRAG) to 55% (TPHA). Further, reporting is driven by level, and facility ownership as primary health facilities and private-for-profit owned facilities account for the largest share of poor reporting despite having capacity consistent with findings in other studies [29, 30].

Although survey data is constrained since it only provides a cross-sectional picture, its high degree of accuracy can be used to understand better non-reporting problems related to the RDTs in DHIS2 through triangulation. The survey data have shown that RDT kits are routinely available within health facilities but may not necessarily be captured in the routine health information system reporting. Therefore, relying on routine data only to inform the availability of tests may underestimate the actual diagnostic capacity among health facilities, emphasising the need for validation and triangulation with other data sources. Our results are consistent with a study in which no health facility reported diagnostic testing for malaria in DHIS2, despite 90% of health facilities having the diagnostic capacity and 40% confirming malaria-positive cases [31].

The historical investment [3, 32] in high-priority diseases such as HIV and malaria was apparent as most facilities (>76%) reported having these RDT kits (Additional file 1, Figure 1). For such priority programmes, routine data presents an invaluable source for surveillance, facilitating accurate assessments of current disease prevalence to determine intervention coverage. Therefore, it would be expected that the reporting pattern in DHIS2, particularly for these two tests, would correspondingly be as high as the testing capacity; however, a high non-reporting rate was observed. Multiple programmes are keen to track HIV and malaria in Kenya and may have introduced parallel reporting channels within DHIS2. Thus, the lack of integration of monthly reporting tools and the elimination of old reporting tools after updates to remove redundancies within DHIS2 could contribute to poor reporting within the MoH 706 tool [33]. Further, technical efforts such as Electronic Medical Record Systems (EMRs) can potentially improve facilities' reporting performance [34, 35]. However, the lack of interoperability between EMRs and DHIS2 to allow seamless data transmission remains challenging [29]. Additionally, testing and reporting within EMR and DHIS2 are subject to human behaviour [36, 37]. The reliance on syndromic diagnosis may result in the under-utilisation of diagnostic tests even where they are already available [38]. Therefore, there is a need to address challenges such as parallel reporting channels and interoperability between systems to obtain reliable diagnostic metrics in pursuing robust metrics to track the SDGs and the UHC agenda.

Compared to the secondary health facilities, primary facilities had poor reporting in the DHIS2 for facilities with capacity. This might be ascribed to the fact that not all primary facilities have access to the DHIS2 system; instead, some rely on the resources available at the sub-county level. Further, the facility records data on manual forms that are later aggregated to the platform, which delays validation checks to ensure data quality [29]. The manual entry may have contributed to the instances where facilities with no testing kits submitted a report of testing in DHIS2, which was also largely prevalent in primary health facilities (Figure 6, Panel 2). In some cases, the sub-county health records officer (HRIO) tasked with data entry may be overwhelmed by the number of facilities reported to contribute to delays and errors [29, 39, 40]. This and other potential drivers may potentially account for some portion of the primary facilities submitting reports to DHIS2, yet they had no diagnostic capacity (Figure 6, Panel 2) [29, 39, 40]. The possibility of erroneous reports driven by data entry errors or pressure to meet performance targets or the quality of KHFA requires further examination.

Among the health facilities, private facilities had poorer reporting rates relative to the public facilities even when further disaggregated by levels (Additional file 1, Figure 3, Panel 1). This is directly linked to the diagnostic capacity that is skewed towards the public facilities. For all RDTs, 63%-84% of facilities in the public sector had diagnostic capability compared to only 16-37% in the private sector (Additional file 1, Figure 2). Private health facilities might be hesitant to adopt HMIS if the benefits gained do not supersede their existing systems contributing to poor reporting in DHIS2 [37]. Ensuring access to essential RDTs cannot be met solely through the public sector. However, the inequality in capacity between public and private facilities evident from the survey data highlights an existing capacity gap that needs to be filled to increase the availability of RDTs in the community. Kenya's Vision 2030 set forth the agenda to expand public-private partnerships in healthcare service provision to ensure adequate health services for the growing population [41]. In addition, a better understanding of the sub-national level diagnostic landscape will better tailor and guide public-private partnerships where needed.

DHIS2 reporting for RDTs, particularly malaria and HIV, is likely to have been impacted by stockouts in 2018. These two RDTs had the highest stockout rates, 31% and 13%, respectively, out of the six RDTs where the stock was evaluated (Additional file 1, Table 3). Over half of the facilities with stockout (51% for malaria and HIV) in 2018 did not submit reports to DHIS2. Dealing with systemic stockout issues, such as supply chains, through simple approaches like SMS reminders [42] might reduce the problem. Resultantly, RDTs will be available when needed by patients and may increase the likelihood of reporting utilization.

To optimise the role of RDTs in supporting clinical decisions and filling the diagnostic gap in LMICS, an assessment of the availability of tests is critical. Kenya has a substantial availability of RDTs (hence diagnostic capacity) across the health system. However, the availability varied between RDT type and facility. Across all the RDTs, diagnostic capacity was high (at least twice) in primary facilities (over 67%) when compared to the secondary facilities (less than 33% except CRAG). These results show that primary public facilities could be the largest consumer of RDTs attributable to better health accessibility (affordable, available, spatially accessible) as most citizens' first point of contact.

Where the health sector faces enormous human resource gaps compounded by funding and infrastructure limitations (8), RDTs are a sustainable solution. However, comparing RDT volumes consumed across the health system versus those needing diagnostic testing is necessary to determine if these RDTs satisfy the community's needs. Consequently, in the context of limited resources in LMICs, such an analysis would align disease burden, healthcare utilisation, and inequities in diagnostic access for targeted resources for RDTs. Therefore, focusing investment on RDTs may offer the most significant opportunity to bridge the health system diagnostic gap as they are easily implemented in rural and primary health care contexts while requiring minimal technical training to perform and interpret the results [8, 10]. Support by government and donors in training and through entrenching the use of data in decision-making would further incentivise reporting and strengthen routine health systems.

In Kenya, several other variables besides a lack of capacity for health facilities have contributed to non-reporting. First, zero reports (no testing) in DHIS2 are usually converted to missing values by the DHIS2 system [13, 19]. This makes it impossible to distinguish between zero reports and non-reporting. An assumption is thus made that any missing value in the dataset was not reported for that period which inflates non-reporting rates. Second, it is also possible that non-reporting in DHIS2 for some tests may have been triggered by clinicians' preference for alternative tests despite KHFA demonstrating confirmed service availability. For instance, a full hemogram is a frequently utilized test panel that provides a wide range of haematological parameters, including Hb levels that haemoglubinometers would have rapidly offered.

Further, with the introduction of Kenya's devolved health structure [39], an eventual increase in the number of sub-counties (from 185 to 316) may have resulted in constrained Health Information System (HIS) resources [43, 44]. Moreover, poor induction of health officers during the transition period, particularly on the new system and reporting, could cause some of the data quality issues currently experienced with the DHIS2 system [44, 45]. Finally, other elements contributing to reduced reporting and require further investigation include health workers' strikes [22, 46], low motivation for health workers, poor logistical infrastructure [8, 15, 29] and limited human resources.

Limitations

In DHIS2, it is impossible to distinguish between zero and missing reports as both are recorded as blanks that may inflate non-reporting. Therefore, in this work, we assumed a test report to have been submitted if at least 1 test was conducted. Triangulation included DHIS2 data for 2020, which may have lower reporting due to disruption during the COVID pandemic; however, the 2018/2019 patterns did not differ from those of 2020. Our sample size for triangulation excluded 106 facilities from the cross-sectional survey as they were not registered on DHIS2. Hence the platform does not capture all functional facilities that are expected to report data. Delay in reporting where reports were submitted after the data was downloaded from DHIS2 may have affected the analysis. Survey questions on service availability for malaria and diabetes were framed as a combination of offering diagnosis and/or treatment. We assumed an affirmative response included diagnosis based on diabetes management requiring frequent blood sugar checks and wide availability of malaria RDTs due to its priority status. Several RDTs had a small sample of facilities that responded during data collection in KHFA; for example, blood grouping had only 34% of facilities responding. Therefore, the triangulation was informed by an inadequate sample size for several RDTs.

Data gaps in routine systems have significant implications in informing diagnostics resource allocation. This study shows that non-reporting in routine health systems is not always due to a lack of capacity. Therefore, further interrogations and analysis are required to inform other drivers of non-reporting and appropriate actions to ensure high data quality and reliable routine health systems.

CRAG Cryptococcal Antigen

DHIS2 District Health Information Systems version 2

EDL Essential Diagnostics List

EMRs Electronic Medical Record System

HCG Human Chorionic Gonadotropin

HIRO Health Records Information Officer

HMIS Health Management Information System

HIV Human Immunodeficiency Virus

KHFA Kenya Harmonised Health Facility Assessment

KMHFL Kenya Master Health Facility List

LMIC Low- and Middle-Income Country

MFL Master Facility List

RDT Rapid Diagnostic Test

UHC Universal Health Coverage

SARA Service Availability and Readiness Assessments

SDG Sustainable Development Goals

TPHA Treponema Pallidum Hemagglutination

VDRL Venereal Disease Research Laboratory

Availability of data and material:

The datasets supporting the conclusions of this article are available online with authorized access provided by the Kenya Ministry of Health; DHIS2 portal

(https://hiskenya.org/dhis-web-commons/security/login.action) and the national master health facility portal (http://kmhfl.health.go.ke/#/home). The Ministry of Health provided the KHFA dataset. The datasets used and/or analyzed during the current study are available to others from these sources through the ministry of Health Kenya.

Acknowledgements:

We are grateful for support from the Ministry of Health for providing the KHFA dataset. We would also like to acknowledge Dr. Michuki Maina for clarifying issues on diagnosis, Gacheri Rael, Waweru Simon and Edna Salat for their insights regarding reporting of lab diagnostic data in Kenya, and George Mbevi who helped in the extraction of the DHIS2 laboratory data.

Ethics approval and consent to participate

This study used secondary aggregated data only, all publicly available to registered users from online data repositories. No individual patient-level data was used in this publication.

Consent for publication

Not applicable - The manuscript does not contain data from any individual person.

Competing interests

The authors declare that they have no competing interests.

Funding:

This work was funded by the Bill and Melinda Gates Foundation (Investment ID: INV-033803). EAO is further supported by Wellcome Trust Intermediate (#201866) and Senior Fellowships (#224272), which also supported BNR. PMM is further supported by the Royal Society Newton International Fellowship (NIF/R1/201418). AKM also benefited from grant support from the Wellcome Trust (#212176). All authors are grateful for the support of the Wellcome Trust to the Kenya Major Overseas Programme (#203077).

Authors' contributions

AKM: Data curation, Formal Analysis, Investigation, Methodology, Software, Validation, Visualisation, Writing – original draft, Writing – review & editing; BNR: Data curation, Formal Analysis, Investigation, Methodology, Software, Validation, Visualisation, Writing – original draft, Writing – review & editing; FB: Investigation, Validation, Writing – review & editing ,EAO: Conceptualisation, Data curation, Funding acquisition, Resources, Validation, Writing – review & editing; PMM: Conceptualisation, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. All authors contributed to the final manuscript

Fleming KA, Horton S, Wilson ML, Atun R, DeStigter K, Flanigan J, Sayed S, Adam P, Aguilar B, Andronikou S. The Lancet Commission on diagnostics: Transforming access to diagnostics. The Lancet. 2021;398(10315):1997–2050.
Catharina B, Madhukar P. :Diagnostic Gaps in Global Health. https://www.thinkglobalhealth.org/article/diagnostic-gaps-global-health. Accessed August 2022.
Yadav H, Shah D, Sayed S, Horton S, Schroeder LF. Availability of essential diagnostics in ten low-income and middle-income countries: results from national health facility surveys. The Lancet Global Health. 2021;9(11):e1553–60.
Carter JY, Lema OE, Wangai MW, Munafu CG, Rees PH, Nyamongo JA. Laboratory testing improves diagnosis and treatment outcomes in primary health care facilities. Afr J Lab Med. 2012;1(1):1–6.
Wilson ML, Fleming KA, Kuti MA, Looi LM, Lago N, Ru K. Access to pathology and laboratory medicine services: a crucial gap. The Lancet. 2018;391(10133):1927–38.
Sayed S, Cherniak W, Lawler M, Tan SY, El Sadr W, Wolf N, Silkensen S, Brand N, Looi LM, Pai SA. Improving pathology and laboratory medicine in low-income and middle-income countries: roadmap to solutions. The Lancet. 2018;391(10133):1939–52.
WHO.: Primary health care on the road to universal health coverage: 2019 global monitoring report. In.; 2019. https://www.who.int/publications/i/item/9789240029040. Accessed 25 August 2022.
Petti CA, Polage CR, Quinn TC, Ronald AR, Sande MA. Laboratory medicine in Africa: a barrier to effective health care. Clin Infect Dis. 2006;42(3):377–82.
Leslie HH, Spiegelman D, Zhou X, Kruk ME. Service readiness of health facilities in Bangladesh, Haiti, Kenya, Malawi, Namibia, Nepal, Rwanda, Senegal, Uganda and the United Republic of Tanzania. Bull World Health Organ. 2017;95(11):738.
Pardo JRP. Rapid Diagnostics and Biomarkers for Antimicrobial Stewardship. In: Antimicrobial Stewardship. edn.: Elsevier; 2017. pp. 69–84.
Economic crisis has not hit health funding: Bill Gates-India News, Firstpost https://www.firstpost.com/india/economic-crisis-has-not-hit-health-funding-bill-gates-328340.html. Accessed 25 August 2022.
Chamas C, Barbeitas MM, Correa M, Kameda K, de Oliveira ACD, Villarinho L. Innovation in diagnostics: addressing gaps in low-and middle-income countries. Bull World Health Organ. 2022;100(8):467.
Bahati F, McKnight J, Swaleh F, Malaba R, Karimi L, Ramadhan M, Kiptim PK, Okiro EA, English M. Reporting of diagnostic and laboratory tests by general hospitals as an indication of access to diagnostic laboratory services in Kenya. PLoS ONE. 2022;17(4):e0266667.
Gesicho M, Babic A. Designing a Dashboard for HIV-data Reporting Performance by Facilities: Case Study of Kenya. Stud Health Technol Inform. 2022;295:238–41.
Wambua S, Malla L, Mbevi G, Kandiah J, Nwosu AP, Tuti T, Paton C, Wambu B, English M, Okiro EA. Quantifying the indirect impact of COVID-19 pandemic on utilisation of outpatient and immunisation services in Kenya: a longitudinal study using interrupted time series analysis. BMJ Open. 2022;12(3):e055815.
Shumbej T, Menu S, Gebru T, Girum T, Bekele F, Solomon A, Mesfin D, Jemal A. Essential in-vitro laboratory diagnostic services provision in accordance with the WHO standards in Guragae zone primary health care unit level, South Ethiopia. Trop Dis Travel Med Vaccines. 2020;6:4.
Service availability and readiness assessment (SARA) https://www.who.int/data/data-collection-tools/service-availability-and-readiness-assessment-(sara). Accessed May 2022.
MOH: Kenya Harmonised Health Facility Assessment (KHFA) 2018/2019 Main Report. Nairobi. In.; 2019. https://www.health.go.ke/resources/downloads/. Accessed on May 2022.
Maina I, Wanjala P, Soti D, Kipruto H, Droti B, Boerma T. Using health-facility data to assess subnational coverage of maternal and child health indicators, Kenya. Bull World Health Organ. 2017;95(10):683–94.
Maina JK, Macharia PM, Ouma PO, Snow RW, Okiro EA. Coverage of routine reporting on malaria parasitological testing in Kenya, 2015–2016. Glob Health Action. 2017;10(1):1413266.
Gesicho MB, Were MC, Babic A. Data cleaning process for HIV-indicator data extracted from DHIS2 national reporting system: a case study of Kenya. BMC Med Inform Decis Mak. 2020;20(1):293.
Amouzou A, Maiga A, Faye CM, Chakwera S, Melesse DY, Mutua MK, Thiam S, Abdoulaye IB, Afagbedzi SK, Ag Iknane A, et al: Health service utilisation during the COVID-19 pandemic in sub-Saharan Africa in 2020: a multicountry empirical assessment with a focus on maternal, newborn and child health services. BMJ Glob Health 2022, 7(5).
Odhiambo JN, Sartorius B. Mapping of anaemia prevalence among pregnant women in Kenya (2016–2019). BMC Pregnancy Childbirth. 2020;20(1):711.
Kenya National Bureau of Statistics (KNBS). 2019 Kenya population and housing census reports. https://www.knbs.or.ke/?p=5732. Accessed August 2022.
Kenya. Health Sector Strategic and Investment Plan (KHSSP) 2014–2018. Transforming health: Accelerating attainment of universal health coverage. Nairobi, Kenya https://www.health.go.ke/wp-content/uploads/2016/03/KHSSP-BOOK.pdf. Accessed November 2021.
WHO: Second WHO model list of essential in vitro diagnostics. World Health Organization. In.; 2019. https://www.who.int/medical_devices/publications/Standalone_document_v8.pdf. Accessed October 2021.
Kenya Health Information System(KHIS) [https://hiskenya.org/dhis-web-commons/security/login.action]. Accessed 25 November 2021.
Alegana VA, Okiro EA, Snow RW. Routine data for malaria morbidity estimation in Africa: challenges and prospects. BMC Med. 2020;18(1):121.
Gesicho MB, Babic A. Identifying barriers and facilitators in HIV-indicator reporting for different health facility performances: A qualitative case study. PLoS ONE. 2021;16(2):e0247525.
Gesicho MB, Babic A, Were MC. Health Facility Ownership Type and Performance on HIV Indicator Data Reporting in Kenya. Stud Health Technol Inform. 2020;270:1301–2.
Githinji S, Oyando R, Malinga J, Ejersa W, Soti D, Rono J, Snow RW, Buff AM, Noor AM. Completeness of malaria indicator data reporting via the District Health Information Software 2 in Kenya, 2011–2015. Malar J. 2017;16(1):344.
WHO: Progress report on HIV, viral hepatitis and sexually transmitted infections 2019: accountability for the global health sector strategies, 2016–2021. In.: World Health Organization; 2019. https://apps.who.int/iris/bitstream/handle/10665/324797/WHO-CDS-HIV-19.7-eng.pdf. Accessed 30 August 2022.
Okello G, Gerrets R, Zakayo S, Molyneux S, Jones C. "Every day they keep adding new tools but they don't take any away": Producing indicators for intermittent preventive treatment for malaria in pregnancy (IPTp) from routine data in Kenya. PLoS ONE. 2018;13(1):e0189699.
Kariuki JM, Manders EJ, Richards J, Oluoch T, Kimanga D, Wanyee S, Kwach JO, Santas X. Automating indicator data reporting from health facility EMR to a national aggregate data system in Kenya: An Interoperability field-test using OpenMRS and DHIS2. Online J Public Health Inform. 2016;8(2):e188.
Ngugi PN, Gesicho MB, Babic A, Were MC. Assessment of HIV Data Reporting Performance by Facilities During EMR Systems Implementations in Kenya. Stud Health Technol Inform. 2020;272:167–70.
Bernardi R, Sarker S, Sahay S. The role of affordances in the deinstitutionalization of a dysfunctional health management information system in Kenya: An identity work perspective. MIS Quarterly: Management Information Systems. 2019;43(4):1177–200.
Macharia J, Maroa C: Health management information systems (HMIS) implementation characteristics that influence the quality of healthcare in private hospitals in kenya. In: 2014 IST-Africa Conference and Exhibition, IST-Africa 2014: 2014; 2014.
English M, Esamai F, Wasunna A, Were F, Ogutu B, Wamae A, Snow RW, Peshu N. Assessment of inpatient paediatric care in first referral level hospitals in 13 districts in Kenya. The Lancet. 2004;363(9425):1948–53.
Hagel C, Paton C, Mbevi G, English M. Clinical Information Network information systems interest g: Data for tracking SDGs: challenges in capturing neonatal data from hospitals in Kenya. BMJ Glob Health. 2020;5(3):e002108.
Kuyo RO, Muiruri L, Njuguna S. Organizational Factors Influencing the Adoption of the District Health Information System 2 in Uasin Gishu County, Kenya. Int J Med Res Health Sci. 2018;7(10):48–57.
2030 GKV: Government of the Republic of Kenya, Ministry of Planning and National Development and the National Economic and Social Council (NESC), Office of the President, Nairobi,. In.; 2007.
Githinji S, Kigen S, Memusi D, Nyandigisi A, Mbithi AM, Wamari A, Muturi AN, Jagoe G, Barrington J, Snow RW, et al. Reducing stockouts of life saving malaria commodities using mobile phone text-messaging: SMS for life study in Kenya. PLoS ONE. 2013;8(1):e54066.
Karuri J, Waiganjo P, Orwa D: Perceived impact of devolved health services on implementation of DHIS2 in Kenya: A qualitative study. In: 2018 IST-Africa Week Conference, IST-Africa 2018: 2018; 2018.
Manya A, Sahay S, Braa J, Shisia B: Understanding the effects of decentralization on health information systems in developing countries: A case of devolution in Kenya. In: 2018 IST-Africa Week Conference, IST-Africa 2018: 2018; 2018.
Njeru I, Kareko D, Kisangau N, Langat D, Liku N, Owiso G, Dolan S, Rabinowitz P, Macharia D, Ekechi C, et al. Use of technology for public health surveillance reporting: opportunities, challenges and lessons learnt from Kenya. BMC Public Health. 2020;20(1):1101.
Bandali S, Thomas C, Wamalwa P, Mahendra S, Kaimenyi P, Warfa O, Fulton N. Strengthening the "P" in Maternal and Perinatal Death Surveillance and Response in Bungoma county, Kenya: implications for scale-up. BMC Health Serv Res. 2019;19(1):611.

No competing interests reported.

AdditionalFile1.docx
Title: Reporting of rapid diagnostic testing in Kenya's health system Description: Additional File 1 contains 4 figures and 3 tables detailing the harmonisation process used in KHFA survey, stockout responses and additional figures of reporting patterns and diagnostic capacity.

Download PDF

Journal Publication

published 30 Mar, 2023

Read the published version in BMC Health Services Research →

Editorial decision: Major revision
21 Feb, 2023
Reviews received at journal
19 Feb, 2023
Reviewers agreed at journal
08 Feb, 2023
Reviews received at journal
22 Jan, 2023
Reviewers agreed at journal
06 Jan, 2023
Reviewers invited by journal
05 Jan, 2023
Editor assigned by journal
03 Jan, 2023
Editor invited by journal
27 Nov, 2022
Submission checks completed at journal
27 Nov, 2022
First submitted to journal
26 Sep, 2022

You are reading this latest preprint version

Reporting of rapid diagnostic testing in Kenya's health system, 2018-2020: triangulation of routine and survey data

Status:

Journal Publication

Version 1

Abstract

Background

Methods

Results

Conclusion

Figures

Background

Methods

Results

Discussion

Conclusion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1