Appropriateness of laboratory expenditure for primary health care facilities across South Africa

Background Primary health care (PHC) services have been prioritised from a cost-containment perspective. To manage expenditure, facility managers use the Laboratory Handbook that indicates the Essential Laboratory List (ELL) tests. Aim The aim of this study was to analyse PHC laboratory expenditure to assess the impact of the ELL in South Africa. Setting We reported ELL compliance at the national, provincial and health district levels. Methods A retrospective cross-sectional study was used to analyse data for the 2019 calendar year. The unique tariff code descriptions were used to develop a lookup table to identify ELL compliant testing. Researchers analysed data for the human immunodeficiency virus (HIV) conditional grant tests and by facility for the bottom two districts. Results There were 356 497 tests (1.3%) that were not ELL compliant that equated to an expenditure of $2.4 million. Essential Laboratory List compliance ranged from 97.9% to 99.2% for clinics, community healthcare centres and community day centres. The provincial ELL compliance ranged from 97.6% for the Western Cape to 99.9% for the Mpumalanga province. The average cost per ELL test was $7.92. At the district level, ELL compliance ranged from 93.4% for Central Karoo to 100% for Ehlanzeni. Conclusions High levels of ELL compliance have been demonstrated from the national to the health district level, demonstrating the value of the ELL. Contribution This study provides data for quality improvement initiatives at primary care facilities.


Introduction
Public health expenditure in South Africa doubled from R107 billion in 2010-2011 to R222bn at the end of the 2019-2020 financial period. 1 Annual growth in health expenditure averaged around 8.5% between 2012-2013 and 2015-2016; however, the downturn in the economic growth in real terms declined by 1.8% per annum. 1 Coinciding with the downturn in the economy, the uninsured population (i.e. those without medical aid) has also grown substantially between 2012-2013 and 2019-2020 eroding the per capita real trend. 1 In order to mitigate for the decreasing budget and expanding demands, the health sector has instituted a number of contingency measures to reduce health expenditure, such as: (1) control of personnel and medicine costs, (2) protection of 'non-negotiable' budget items, (3) saving on nonessential items, (4) reduction in capital spending on buildings and (5) the prioritisation of primary health care (PHC) and chronic medicine dispensing and the distribution at additional and alternate sites to reduce queues and improve access with a focus on health outcomes. 1 Of particular relevance to the focus of the current research is that the average growth in expenditure for laboratory services was 3.4% between 2012-2013 and 2019-2020, which is above the real growth in the budget. 1 Primary health care services have been prioritised both from a cost-containment perspective and improving health outcomes of the population by improving the quality of healthcare services. The Ideal Clinic Realisation and Maintenance (ICRM) programme aims to improve the effectiveness and responsiveness of the health system specifically at the PHC level, which was launched by the Director General of the National Department of Health (NDOH) in 2013. 2 The ICRM programme provides the minimum inputs and processes required from PHC facilities to Background: Primary health care (PHC) services have been prioritised from a cost-containment perspective. To manage expenditure, facility managers use the Laboratory Handbook that indicates the Essential Laboratory List (ELL) tests.
http://www.phcfm.org Open Access deliver a desired output ( Figure 1). An Ideal Clinic is defined as: [A] clinic with good infrastructure, adequate staff, adequate medicine and supplies, good administrative processes and sufficient bulk supplies, that uses applicable clinical policies, protocols, guidelines as well as partner and stakeholder support, to ensure the provision of quality health services to the community. 3 (p. 4) The ICRM framework consists of 10 components as follows: (1) administration, (2) integrated chronic disease and/or clinical services management, (3) medicines, supplies and laboratory services, (4) human resources for health, (5) support services, (6) infrastructure, (7) health information management systems, (8) (7) facility controls rapid test kit performances by running one negative and one positive control on a weekly basis. 3 To manage laboratory expenditure, facility managers are required to use the PHC Laboratory Handbook that indicates the Essential Laboratory List (ELL) tests. 4 Furthermore, a dedicated laboratory request form has been introduced based on the ELL. The PHC facility manager is now responsible for developing control measures for the rational utilisation of laboratory expenditure in line with evidence-based guidelines. 5 Laboratory demand management aims to improve the requisition of the appropriate laboratory test as well as result in the reductions in public health expenditures without affecting clinical outcomes. 6 It is incumbent for the health authority to develop a broad-based demand management strategy that defines what is appropriate test based on the clinical services offered. 6 This may involve standardising the repertoire of tests that may be requested by level of care. 6 A local study stated that laboratory services will have to formulate strategies to address both under-and overutilisation of laboratory tests and ensure that the proper use of clinical laboratory testing contributes to improved patient care. 7 This will require laboratories to monitor test usage for cost-effectiveness and appropriateness. 7 Some studies have used rule-based systems or algorithms to implement laboratory demand management that is more suited to higher levels of care. 8,9,10 Other laboratory demand management strategies include education, the redesign of laboratory request forms, use of computerised physician order entry (CPOE) and implanting reimbursement models. 11 In a South African setting, the development of an ELL list that is implemented using a standardised PHC laboratory request appeared to be the most optimal demand management strategy.
The NHLS operates a platform of over 268 diagnostic laboratories across South Africa serving 80% of the population. 12 The NHLS operates a platform of 226 laboratories across South Africa, ranging from highly sophisticated centralised, high volume and academic to low test volume distant rural laboratories. 12 The core function of the NHLS is to provide cost-effective and efficient health laboratory services in the public healthcare sector. 12

Aims and objectives
This study aims to analyse the laboratory expenditure across 52 health districts to assess ELL compliance. A secondary aim was to assess whether the ideal clinic initiatives introduced have resulted in the widespread adoption of the ELL across South Africa.

Study design
A retrospective cost analysis was conducted for laboratory expenditure data for PHC services across South Africa for the 2019 calendar year.

Study setting and population
Data were extracted for all PHC health facilities across 52 health districts. We used the district health information system (DHIS) organisational hierarchy to include expenditure for PHC clinics, community health centres (CHCs) and community day centres (CDCs). No sampling was performed as all PHC facilities with laboratory expenditure data were included. A comprehensive list of account numbers was obtained from the NHLS Finance department. We used this list to identify account numbers associated with PHC facilities.

Data collection and preparation
We received three separate data extracts: (1) PHC billing accounts expenditure data for all PHC account numbersroutine tests requested on the PHC request form (N1) and allocated to the default billable account number (e.g. 32ZKOK000005 for the Kokosi Clinic), (2) Xpert MTB/RIF conditional grant expenditure identified by the prefix 'ZGXP' and (3) comprehensive care, management and treatment (CCMT) of HIV programme conditional grant expenditure identified by the prefix antiretorviral expenditure (ZARV). Data were received as a password-protected file and personally collected from the corporate data warehouse (CDW).
Each data extract included the following variables: (1) billable account number (e.g. 16ZJEF000001 for PHC facilities in the Kouga municipality), (2) customer name, (3) facility code, (4) facility description, (5) year, (6) month, (7) tariff code, (8) tariff description, (9) test volumes and (10) expenditure. The expenditure for a laboratory test is itemised as tariff codes; for example, tariff code 2210 denotes the haemoglobin test. 13 The tariff code and tariff description are assigned within the laboratory information system (LIS) for each test. There is both a one-to-one and one-to-many relation between a tariff code and a test. For example, the alanine transaminase (ALT) test has a single tariff code (2685) when compared with the lipogram panel, which has multiple tariff codes for each analyte, for example, total cholesterol (2855), triglyceride (3440), high-density lipoprotein (HDL) (2865) and the calculated low-density lipoprotein (2860). A panel is a series of tests performed on one specimen that is usually related to a single condition or disease. 14 Each tariff code is associated with a state price that determines expenditure, that is, test volume multiplied by the tariff code state price. SAS 9.4 (Cary, NC, USA) was used to append these three expenditure datasets, ensuring that variables were in the same data type format (number or string) with the same header row. The appended SAS dataset was used to commence with data preparation. The lookup table approach was used as it allows the physical representation of attributes, that is, variables in a relational database. 16 They can be used to store information for one or more related attributes. 16 The lookup table aimed to identify laboratory expenditure that was ELL compliant.
For each tariff code, the ELL status, test description, discipline and test group variables were coded by a team of experts. For tests not in the ELL, data were categorised as 'Not an ELL' test compared with 'ELL test'. The test group identified panels where more than one tariff code was reported for a test (e.g. lipogram).
A structured query language (SQL) procedure was used to create a left outer join between the appended laboratory expenditure dataset, tariff code and facility lookup tables, with the tariff code and facility description as primary keys. All non-PHC expenditure data was excluded using the DHIS OU_Type variable, that is, researchers included clinics, CHC and CDC. This resulted in the creation of a new SAS dataset for PHC laboratory expenditure indicating ELL compliance for each row of data.
A series of data quality checks were conducted before commencing with the data analysis. These included: (1) ensuring that all tariff codes had an ELL status populated, (2) identifying that the top 10 tariff codes based on expenditure match national NHLS volume and revenue reports (data not shown), (3) ensuring that no expenditure data were reported for hospitals and (4)

Ethical considerations
Ethical approval was obtained from the Human Research and Ethics Committee (HREC) (Medical) at the University of Witwatersrand (M200545). Approval to extract expenditure data was obtained from the NHLS Academic Affairs, Research and Quality Assurance (AARQA) Department. The expenditure data were extracted by the CDW, which is the national repository of all public section laboratory data generated by NHLS laboratories. No patient identifiers were collected and therefore informed consent was not required.     (Figure 3). Only two districts reported ELL compliant laboratory expenditure of less than 94.8% (Namakwa and Central Karoo).

Health facility analysis for the two worst health districts (Namakwa and Central Karoo)
Across the 28 facilities in the Namakwa District, the median compliance rate to ELL was 95.6% (interquartile range [IQR]: 92.7% -97%). The lowest compliance was 89.7% from a single facility. Twenty-five percent (n = 7) of facilities reported an ELL compliance rate of ≤ 93%. There was no statistical difference in the ELL compliance rate between these two districts (p > 0.05). Across the 10 facilities in Central Karoo District, the median ELL compliance rate was 93.5% (IQR: 92.7% -94.3%). The lowest compliance was 90.9% at a single facility.
The spatial analysis at the health facility level in the Namakwa and Central Karoo health districts revealed no clusters of poor ELL compliance (Figure 4). There were five health facilities with ELL compliance between 89.7% and 91.9% (red circles).

Discussion
The objective of this study was to assess compliance of laboratory expenditure across 52 health districts to a PHC ELL in South Africa. Primary health care services have been exposed to the ideal clinic initiative that aims to improve the utilisation of minimum inputs and processes to deliver a desired health output. 5 The development of an ELL is a demand management approach that achieves cost containment, standardisation of practices and better health outcomes for the population by improving the quality of healthcare services. 5  The high level of ELL compliant laboratory expenditure (> 98%) is a promising finding and is indicative of the successful implementation of a combination of demand management interventions. Furthermore, the finding from this study is an improvement from the previous 90% reported for the NHI pilot districts. 13 This improvement indicates that these interventions have continued to yield good levels of ELL compliance many years later. The combination of the ELL and a standardised laboratory request form have made it possible to extend work for these pilot districts to all PHC across South Africa to around 3800 facilities.
This indicates that these interventions have been implemented across PHC facilities with similar levels of ELL compliance despite the variation in operating hours, package of services offered, size of the facility and the percentage of healthcare offered by doctors between clinics and community healthcare centres. Furthermore, across nine provinces, ELL compliance was over 97%. This is a positive finding that indicates that the interventions have been cascaded across all PHC facilities in South Africa and is indicative of the sustainable intervention of laboratory expenditure demand management. It should be noticed that Free State, Limpopo, Mpumalanga, Northern Cape, has good compliance probably because of low testing carried out on the population.
In developing countries, patient records are collected on pen and paper systems used by clinical staff and then entered later into an information system by data entry staff either locally or remotely. 18 The absence of a medical record system specifically for PHC services makes it difficult to develop more advanced methods of demand management such as rule-based systems and/or algorithms. 8,9,10 Even with the ideal clinic system, aggregate reporting is based on the use of standardised clinical stationery and registers that collate data using systems such as Tier.Net, DHIS and Health Patient Registration System (HPRS). 19,20,21 In the absence of a more advanced data system, it would be difficult to implement more advanced demand management strategies. There have been demand management approaches deployed at hospitals in South Africa where rule-based algorithms have been deployed on the LIS. 10,22,23,24 Some of these approaches could be deployed to PHC services. However, it must be observed that a local study reported that these approaches do not appear to be effective with limited impact on clinician test requesting pattern. 24 In addition, with such high levels of ELL compliance, the rule-based algorithm would have a much smaller impact at PHC facilities because of the presence of a limited test repertoire.  26 This ranged from 54.6% to 92.2% for the Eastern Cape and Western Cape provinces, respectively. 26 The Mpumalanga province reported a VL coverage of 74.0%. 26 This finding contradicts the 72.2% contribution of HIV viral load testing to expenditure in this province. The 2017 National Antenatal Sentinel HIV Survey in South Africa reported that the overall HIV prevalence was highest in KwaZulu-Natal (41.1%) followed by the Mpumalanga province (37.3%). 27 Despite reporting similar HIV prevalence, less than 50% of ELL compliant expenditure was for HIV viral load testing in the KwaZulu-Natal province. This indicates that a higher HIV burden is not able to explain the higher average test cost in the Mpumalanga province. Perhaps, antiretroviral treatment (ART) enrolment data could explain the increased average test cost noticed.
The approaches reported in this study should be developed as an interactive dashboard to make it possible for district managers to quickly flag poor ELL compliance. Similar approaches have demonstrated the value of interactive dashboards to improve turn-around-time (TAT) performance. 28,29 In addition, HIV and TB cohort monitoring and evaluation dashboards have been used to report key indicators from the national to the health facility level (e.g. virological suppression). 30,31,32 For the ELL dashboard development, the updated lookup table should be used as it incorporated recent evidence-based recommendations to increase the test repertoire for PHC services. This would make it possible for all levels of healthcare management to assess compliant laboratory expenditure.

Limitations
This study only assessed ELL compliance. Without the clinical data, it is not possible to determine the appropriateness of the diagnostic testing.

Conclusion
High levels of ELL compliance have been demonstrated from the national to the health district level. Similarly, across the three categories of PHC services, we observed high levels of ELL compliance. Our findings have demonstrated the value of a national ELL-aligned standardised laboratory request form for PHC services. We identified the need for better monitoring for poor-performing facilities. One of the next steps would be to determine a benchmark percentage for ELL compliance, which would make it easier to identify outliers. In addition, this would make it possible to create routine exception reports for district managers. There is a need to develop an integrated dashboard to rapidly identify poor-performing facilities for targeted interventions and quality improvement programmes.