Uses of private health provider data for governance in low-income and middle-income countries: results from a scoping review

Abstract

Abstract

This work grew from an interest in understanding how private data are used for health system governance in low-income and middle-income countries (LMICs).

Objective

We conducted a scoping review to understand how the public sector collects routine data from the private health sector and uses it for governance purposes. The private health sector was defined to include both formal and informal, for-profit or non-profit, actors delivering healthcare services.

Findings

We identified 4014 individual English language studies published between 2010 and 2021. We reviewed titles and abstracts of all, with 50% reviewed by two authors to ensure a common application of inclusion criteria. 89 studies were selected for review in full; following this, 26 articles were included in the final selection as they directly report on the use of routine private sector data for governance in LMICs. Only English language studies were included, limiting the scope of possible conclusions.

Results

Data were most commonly collected by the Ministry of Health or a subministerial office, with extraction from District Health Information System 2 specifically cited for three studies. 16 studies collected data on infrastructure and distribution, 15 on service delivery, 12 on health financing, 7 on pharmaceuticals and other consumables, 4 on health workforce, 4 on quality of care and 4 on epidemic surveillance.

Conclusion

The studies identified provide examples of the public sector’s capacity to collect and use data routinely collected from the private sector to perform essential governance functions. The paucity of studies identified is an indication that more attention is needed to ensure that this key area of health system governance is undertaken and that lessons learnt are shared. This review provides insights to understanding private sector health data collection and use for governance in LMICs, and for guiding activities to assess and improve this according to country context and capacity.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This study is an objective and systematic scoping review of an important and until-now understudied low-income and middle-income country health systems.

• The methodology was comprehensive enough to capture a large number of articles and the study question was sufficiently focused to allow informed findings on a specific topic.

• The same narrow focus was also a limitation: restricting ourselves to private sector data use for governance found a limited number of studies, had our focus widened to private sector data existence, regardless of use for governance, we would have had a larger basis for generalisable findings.

• The study was limited by only including English-language publications.

• Our findings provide examples of how private sector data is and can be used for health system governance, however, it is quite likely that many examples of routine data collection and use for oversight and governance exist which have not been the subject of academic study and so are not captured in this review.

Background

Health systems in almost all low-income and middle-income countries (LMICs) are ‘mixed’, meaning that they include both public and private actors.^{1 2} These sectors can compete or collaborate in terms of resources and services.3,8 Effective health system governance requires the accurate collection and use of data from both sectors to monitor and help improve its overall performance.⁹

Proper data management on both public and private sector activities can lead to improved policies, efficiency and readiness during epidemic outbreaks.^{10 11} However, gathering and integrating data, especially from private health facilities outside government reporting structures, presents challenges.

As a result, health authorities are not always able to pinpoint where private health providers are, or what is the nature, cost and quality of the services they deliver.¹²

This study aims to explore the type of private health data collected by the public sector in LMICs and to identify potential challenges as well as illustrations of how the public sector uses private health data for better health system governance in those countries.¹³

Methods

The authors conducted a scoping review to determine what published research had drawn from private health sector data routinely collected by the public sector in LMICs.

Definitions

We intentionally used a broad definition of ‘the private health sector’ to include both formal and informal healthcare providers, for-profit or non-profit. Private health providers range from Patent and Proprietary Medical Vendors in Nigeria or Registered Medical Practitioners in India, registered drug vendors and pharmacies, to private doctor consulting rooms and clinics, all the way to world-class tertiary hospitals.14,17 As this study is focused on service provision, it has excluded manufacturers, importers, wholesalers, laboratories and blood banks. There were no exclusion criteria related to the type of public entities collecting data from the private health sector.

This study used the definition of ‘governance’ provided in the WHO’s Health System Performance Assessment (HSPA) framework: ‘ensuring (that) strategic policy frameworks exist and are combined with effective oversight, coalition-building, regulation, attention to system design and accountability’.¹⁸ Within that framework, ‘information and intelligence’ is a subfunction of governance, linked to the other key functions of the health system: provision of health services, financing and resource generation.¹⁹

We used the World Bank definition of LMICs as per the fiscal year 2021, which is based on gross national income per capita.

Search method and screening process

The scoping review included published literature from January 2010 to November 2021 following the framework described by Munn et al.²⁰ Researchers used Embase, PubMed, Web of Science and World Wide Science to identify articles. Articles were included if they mentioned at least one keyword from each of three categories: ‘private health sector’, ‘data’ and ‘governance’. Each of those categories referred to an extensive list of related terms (table 1). Articles were included if any of the terms were found. Exclusion criteria were then applied regarding articles published in a language other than English were excluded, as well as articles that did not relate to LMICs as defined by the World Bank.

Table 1. Search terms.

1/Health Private SectorPrivate health*Private health carePrivate healthcarePrivate health insurancePrivate health sectorPrivate health centerPrivate health facilit*Health private sectorPrivate careHospital*Private polyclinic* Private clinic*Clinic*Private dispensar*Dispensar* pharmacyChemist shopDrug vendorDrug sellerMedicine Vendor

2/DataData*InformationIndicatorsMeasurementStatisticsRecordsMonitoringDHISMIS“data collection”[Mesh] data [tiab]

3/GovernanceGouvernanceGovernanceGovernmentPolicyManagementAdministrationInstitution*Public health insuranceContract* paymentReimburs* accredit* oversite regulat*

The search results and review steps are shown in the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) chart in figure 1. Search results were exported to a citation manager where duplicates were removed. The study team uploaded all remaining articles (N=4014) to a publicly available platform to facilitate screening for titles and abstracts (http://rayyan.ai/reviews). Two reviewers looked at 2007 (50%) of the articles to ensure that there was an alignment in the screening process. Conflicts were resolved after a discussion. For the remaining 50%, all the articles where one of the reviewers was uncertain about the decision were flagged for secondary review. The screening narrowed the articles included to those which described not simply the production or identification of routine private sector health data, but its collection and use for governance purposes, either by formal government agencies or by other public entities.

Figure 1. PRISMA chart. LMICs, low-income and middle-income countries; PRISMA, Preferred Reporting Items for Systematic reviews and Meta-Analyses.

Two members of the research team reviewed 89 full-text articles: 43 articles were excluded after a full review, 17 because they were not related to routine data collected by national authorities, 17 because they did not focus on the private sector, 5 because they did not cover health providers, 2 because they did not focus on LMICs and 2 because they were not published in English.

Data extraction

Data were subsequently extracted out of the 46 remaining articles. To ensure a rigorous and uniform approach to extracting relevant data, two reviewers developed a standardised data extraction template. Sections included the type and frequency of data collected, the collecting entity, the type of actors data were collected from, the purpose of data collection, as well as insights on data quality, data analysis methods, challenges or recommendations. After further analysis, only 26 were included in the scoping review because they addressed direct data collection of data from private health providers by national authorities. The remaining 19 other articles were identified and reviewed for insights on the collection and use of private health sector data by the public sector but excluded because they did not relate to specific cases of direct data collection and use. All the studies were extracted by one reviewer and verified by a second to ensure completeness and consistency.

Patient and public involvement

Neither patients nor the public were involved in this study.

Results

Among the 26 articles included, 1 used data from a lower-income country, 13 from lower-middle-income countries and 12 from upper-middle-income countries. Using the regions defined by WHO, nine were based in African Region, seven were in the Region of the Americas, five were in the Easter Mediterranean Region, two were in the European Region, two werevin the Western Pacific Region and one was in the South-East Asia Region.

Different data sources for different data uses

11 of the included studies used series data, collected every month or more frequently, indicating for some countries the existence of routinely updated databases of pharmaceutical markets21,24; insurance claims,²⁵ human resources,^{26 27} health facility information,^{25 28} infectious disease reporting²⁹ and quality of care.^{25 30}

Private health sector data were most commonly reported as being collected by the Ministry of Health or a subministerial office, with extraction from District Health Information System 2 specifically cited for three studies from Kenya and Tanzania.^{27 28 31} Other public entities involved in the collection of private health sector data were the National Health Insurance^{25 32}; the Pharmacy Council or Food and Drug Administration^{24 33}; the Medical Regulatory Agency and the Central Medical Stores who worked in collaboration with IQVIA, an international commercial health data company.²¹

In some contexts, the collection of private health sector data appeared to take place as part of national or regional pilot projects supported by international development organisations. In Afghanistan for example, private health sector data collection efforts were supported through the Afghan Private Hospital Association in the context of a public–private partnership (PPP) aiming at contracting private health facilities to deliver essential health services.³⁴ In addition, some of the private data were collected by public actors through routine inspections.

In 18 out of 26 papers, the studies integrated public and private data.21,2325 The remaining nine papers used either national supply data or data extracted only from private sector providers.

Studies on vertical programmes, HIV and Malaria, stood out for their use of widely implemented electronic data systems tracking logistics, drug stocks, and patient treatment and outcomes.^{21 23 28 31}

Types of data collected

Within the 26 studies identified which analysed aspects of private healthcare delivery in LMICs using government-collected data, most of them reported the collection of several data types: 16 collected data on private infrastructure and distribution, 15 on private service delivery, 12 on health financing, 7 on private pharmaceuticals and other consumables, 4 on health workforce, 4 on quality of care and 4 on epidemic surveillance data. Details are shown in online supplemental table. Data on private ‘infrastructure distribution’, ‘pharmaceuticals and consumables’ and ‘health workforce’ relate to the ‘resource mobilisation’ function of the health system. Data related to the ‘provision of health services’ function of the health system included ‘types and volumes of services provided’ and ‘quality of care’—which is an overarching goal of the health system that is particularly linked to the provision of health services. The ‘health financing’ data category relates to the eponymous function of the health system. ‘Epidemiologic surveillance’ data were added by the research team as it was not explicitly included in the WHO HSPA Framework.

Infrastructure distribution

The most common type of private health sector data collected was infrastructure distribution. This includes unique identification data for private facilities, as well as basic information on location, ownership status and level of care provided. This is sometimes referred to as a Master Facility List.^{31 37} The different country examples of data use of private infrastructure distribution demonstrated how governments could use this data to better plan and allocate resources to increase access to and equitable coverage of healthcare services.

With the exception of a study in Nigeria, all studies that mentioned the collection of private health infrastructure distribution data also included other types of data. In Iraq, private infrastructure distribution information combined with demographic data and medical history was used to determine the effect of place of birth on the risk of primary postpartum haemorrhage.⁴¹ In Belize and Iran, data were collected on private services delivered at the facility level and so we inferred that data on infrastructure distribution were available to public health authorities, even though not explicitly mentioned.^{13 25}

Types and volumes of services provided

Service delivery data included data on the types and volumes of private services provided at the facility level. The studies on service delivery demonstrated the challenges of collecting private providers’ data to generate market intelligence needed for multiple governance functions. A study in Kenya compared performance on HIV data reporting by facility ownership type and found that public facilities showed better timeliness and completeness of reporting than private ones.³¹ A possible reason to explain that difference was a stronger support from vertical programmes in public facilities. Similar findings in Kenya illustrated the underperformance of private facilities compared with public ones on the reporting of malaria indicators.²⁸

The studies on service delivery data also highlighted its use in facility and systems-level planning and assessment. In the case of India, researchers combined service delivery with infrastructure distribution data from the private sector to calculate travel time to surgical facilities and better understand supply-side information on essential obstetric and gynaecological services.³⁸ In China, service delivery data were used to assess private hospital expansion between 2006 and 2015, showing an increasing gap in average size between public and private health facilities.⁴⁴ In Malaysia, inpatient utilisation records allowed to determine that expansion in private hospitals has not led to performance improvements in public hospitals.⁴⁵

Brazil and Mexico, extensive service delivery data collection systems including patient data allowed for elaborate disease-specific analysis. In Mexico, patient data from public and private hospitals were used to estimate the burden of Hodgkin lymphoma as well as to compared outcomes of patients with acute ischaemic stroke in public and private settings.^{39 42}

Health financing

Health financing information disaggregated by public and private sector was often available in the identified studies; when combined with other types of data it allowed for a range of possible governance uses. In Morocco, health financing data included health expenditure for both the public and the private sector as well as expenses by type of service and insurance profile.³² It showed that the consumption of healthcare by the population covered by basic health insurance had almost doubled between 2009 and 2014, with a sharper increase in the public sector compared with the private sector. In Turkey, health financing data combined with infrastructure and service delivery data allowed the government to compare differences between the private and the public sector capacity and service delivery at the regional level.⁴³ In Belize, health financing data combined with service delivery data made it possible to assess differences in technical efficiency between public and private facilities.²⁵ In Brazil, health financing data were available through the national health system (Sistema Único de Saúde). Combined with patient-level data, it allowed research on health service utilisation by public and private health facilities and economic evaluation analysis on pneumococcal disease and sepsis.^{30 36}

Pharmaceuticals and other consumables

The type of routine data collected on private pharmaceuticals and other consumables included location of pharmacies, prices and quantity sold in the private sector, demographic information on buyers using private pharmacies, national production and importation volumes. These data enabled governments to assess access and affordability of essential medicines in the private market. Several pilot projects were tested in African countries to ensure access to quality and affordable drugs to the population through public and private channels and avoid stock-outs. In Zambia, an innovation was tested to use market data on malaria medicine available in private pharma facilities to inform evidence-based policy.²¹ The Accredited Drug Dispensing Outlet model was tested as a PPP in Tanzania to assess access to essential medicines in private drug shops. It included the collection of information not only on drug sales and prices but also on their quality.³³ In Tanzania, a PPP pilot project pooled medicine tenders in the Dodoma region, involving the collection of similar private data types.⁴⁶ In Namibia, an integrated pharmaceutical management information system cross-referenced four data bases to improve the timeliness and reliability of data regarding patient decision-making as part of the national antiretroviral therapy programme.²³

Comprehensive collection of data on private pharmaceuticals enables governments to conduct more sophisticated policy analysis. Evidence of a data collection system on the pharmaceutical sector was also found in Iran, in what appeared to be a more systematic manner, allowing for a study tracking on medicine consumption sales at national level²⁴ and another one to identify indicators of fraud and abuse in general physicians’ drug claims.¹³ In Colombia, a study was able to show that after drug price control was implemented, price inflation decreased sharply while pharmaceutical expenditure almost doubled.²² Possible explanations included an increased affordability of drugs, as well as enhanced marketing efforts by pharmaceutical companies to maintain their profits.

Health workforce

Basic health workforce data collection was reported in Turkey, Afghanistan and India, including the number and type of health professionals by public or private facility which allowed for a better understanding of the availability and mobilisation of Human Resources for Health (HRH).^{26 34 38} A more elaborate human resource information system was found in Tanzania, where two information systems were rolled out to capture information on public and private HRH such as employees’ personal and professional details, training and employment history and salary scale.²⁷ While the establishment and utilisation of those systems took time and effort, it allowed health council management teams to realise their HRH personnel needs and allocate resources based on demand and expertise.

Quality of care

Two studies related to health service quality of care data while two others involved medication quality data. Quality of care data were systematically used in conjunction with other databases. In Turkey, user satisfaction data were cross-referenced with province-level data on health resource availability and utilisation to assess the impact of family medicine introduction on primary healthcare consultations.²⁶ In Brazil, medical malpractice data were extracted from jurisprudence records and allowed for a public/private comparison of healthcare outcomes for neurosurgery procedures.⁴⁰ The Tanzania drug supply pilot projects upmentioned included quality data such as drug customer complaints, drug order fulfilment rate and medication delivery time.^{33 46}

Epidemic surveillance

In addition to the data types described above, the scoping review also identified cases of routine epidemiological data collected from the private sector by public entities. Those included the reporting of mandatory disease surveillance information, as well as other specific diseases in the context of vertical programmes. The studies underscored the challenges faced by the public sector in collecting epidemiological data from private health facilities to monitor diseases. In Kenya, two studies related to data reporting of malaria and HIV respectively, both found higher reporting rates in the public sector than in the private sector.^{28 31} A study in Nigeria focused specifically on assessing the level of compliance of private facilities with the disease surveillance system found that 40% of facilities were compliant.²⁹ In Afghanistan, the level of reporting of private sector facilities to the disease early warning system was considered an indicator of government stewardship.³⁸

The sources of data used in studies are summarised in table 2.

Table 2. Scoping review extraction table.

Authors/year	Study title	Methodology	Study objective/purpose	Setting	Study period	Public entity using data
Coghlan et al 201821	A new approach to gathering pharmaceutical market data to support policy implementation and access to medicines: as demonstrated by malaria medicines in Zambia.	Case study	Report on an innovation to provide regular, national-level data on the total pharmaceuticals market in Zambia	Zambia	2009–2014	Zambia Medicines Regulatory Agency
Bowser et al 201325	A preliminary assessment of financial stability, efficiency, health systems and health outcomes using performance-based contracts in Belize	Difference-in-difference	Assess the difference in technical efficiency between private and public facilities	Belize	2006–2009	National Health Insurance
Yassine et al 202032	Assessment of the medical expenditure of the basic health insurance in Morocco	Cross-sectional analysis	Evaluate the healthcare consumption for the population insured under the Basic Health Insurance	Morocco	2009–2014	National Moroccan Health Insurance Agency
Rutta et al 201533	Accrediting retail drug shops to strengthen Tanzania’s public health system: an ADDO case study.	Case study	Assess the scalability, transferability and sustainability of the Tanzania accredited drug dispensing outlet (ADDO) model	Tanzania	May-Aug, 2014	Pharmacy Council
Ishijima et al 201527	Challenges and opportunities for effective adoption of HRH information systems in developing countries: national rollout of HRHIS and TIIS in Tanzania.	Case study	Describe the development and rollout of two information systems to capture information from both the public and private sectors.	Tanzania	2009–2014	Ministry of Health
Githinj et al 201728	Completeness of malaria indicator data reporting via the District Health Information Software 2 in Kenya, 2011–2015.	Case study	Assess the completeness of malaria indicators data reported by health facilities in Kenya through DHIS2	Kenya	Jan 2011-Dec 2015	Ministry of Health
Makinde and Odimegwu 202029	Compliance with disease surveillance and notification by private health providers in South-West Nigeria.	Cross-sectional study	Investigate the level of compliance with disease surveillance reporting and the factors that affect their participation	Nigeria, South-West	Nov 2016-Nov 2017	Federal Ministry of Health, state government officers
Makinde et al 201837	Distribution of health facilities in Nigeria: Implications and options for UHC.	Secondary data analysis	Review the geographical and sectoral distribution of health facilities in Nigeria	Nigeria	2011–2013	Federal Ministry of Health
Sartori et al 201336	Estimating Health Service Utilization (HSU) for treatment of pneumococcal disease (PD): The case of Brazil	Economic evaluation	Describe the process of developing estimates of HSU for PD in introduction of pneumococcal conjugate vaccine.	Brazil	2004	Ministry of Health
Hone et al 201726	Effect of primary health care reforms in Turkey on health service utilization and user satisfaction	Longitudinal study, regression	Assess the effect of family medicine on health service utilisation and user satisfaction.	Turkey	2002–2013	Ministry of Health, Directorate General of Health Research
Quintano Neira et al 201830	Epidemiology of sepsis in Brazil: Incidence, lethality, costs, and other indicators for Brazilian Unified Health System hospitalizations from 2006 to 2015.	Logistic regression analysis	Assess trends in the incidence, lethality, costs and other indicators of sepsis for Brazilian Unified Health System hospitalizations	Brazil	2006–2015	Brazilian Unified Health System (SUS, Sistema Único de Saúde)
Ensor et al 202038	Factors influencing use of essential surgical services in North-East India: a cross-sectional study of obstetric and gynaecological surgery	Cross-sectional study	Understand demand-side factors influence OB/GYN procedures in the underserved and remote setting of North-East India.	North-East India	2015–2016	Health Management Information System, district Census
Cross et al 201747	Government stewardship of the for-profit private health sector in Afghanistan	Review, Key Informant Interviews (KII)	Examine the progress made by the MoPH towards more effective stewardship.	Afghanistan	NA	Ministry of Health
Gesicho et al 202031	Health Facility Ownership Type and Performance on HIV Indicator Data Reporting in Kenya	Retrospective observational study	Evaluate the relationship between facility ownership type and performance on HIV indicator data reporting.	Kenya	2011–2018	District Health Information Software 2 (DHIS2)
Prada et al 201822	Higher pharmaceutical public expenditure after direct price control: improved access or induced demand? The Colombian case	Descriptive analysis	Investigate the effects of the Colombian policy efforts to control expenditure by controlling prices	Colombia	2011–2015	SISMED (Drug Price Information System)
Rivas-Vera et al 201939	Hodgkin lymphoma (HL) - Burden of the disease in Mexico. Construction of proxy measure with data of the National Health System	Health burden analysis	Estimate the burden of HL in Mexico	Mexico	2016	General directorate of health information, National Institute of Geography and Statistics
Mabirizi et al 201823	Implementing an Integrated Pharmaceutical MIS for Antiretrovirals and Other Medicines: Lessons From Namibia	Programme eval (review and KII interviews)	Describe the pharmaceutical MIS used to provide data for Namibia’s national antiretroviral therapy (ART) programme.	Namibia	2016–2019	Ministry of Health and Social Services
Joudaki et al 201613	Improving Fraud and Abuse Detection in General Physician Claims: A Data Mining Study	Data mining approach and lit. review	Identify the indicators of healthcare fraud and abuse in general physicians’ drug prescription claims.	Iran, Lorestan province	2011	Social security organisation
Wiedenmayer et al 201946	Jazia prime vendor system- a public-private partnership to improve medicine availability in Tanzania: from pilot to scale	Delphi study	Complement the national supply chain in case of stock-outs with a simplified, procurement procedure.	Tanzania, Dodoma region	2014–2018	Dodoma regional administration and local government
de Macêdo Filho et al 202040	Malpractice and socioeconomic aspects in neurosurgery: a developing-country reality	Regression analysis	Describe medical malpractice in neurosurgery procedures and how they culminate in unfavourable outcomes	Brazil	Jan 2008 and Feb 2020	Brazilian Hospital Information System, Unified Health System (SUS, Sistema Único de Saúde)
Raof 201441	Management of primary postpartum Hemorrhage inside Erbil city: Iraq	Retrospective study	Determine the effect of place of birth on the risk of primary postpartum haemorrhage	Iraq, Erbil province	2012–2014	Directorate of Health in Erbil
Deng et al 201844	Private hospital expansion in China: a global perspective	Review, descriptive analysis	Assess the private hospital development in China during 2005–2016 from a global perspective	China	2005–2016	Ministry of Health
Ruiz-Sandoval et al 201842	Public and Private Hospital Care Disparities of Ischemic Stroke in Mexico: Results from the Primer Registro Mexicano de Isquemia Cerebral (PREMIER) Study	Retrospective study	Identify the impact on short- and middle-term outcomes of patients with acute ischaemic stroke among public and private Mexican medical care.	Mexico	Jan 2005 - Jun 2006	Primer Registro Mexicano de Isquemia Cerebral
Nwagbara and Rasiah 201545	Rethinking health care commercialization: evidence from Malaysia	Descriptive and cross-sectional study	Determine whether an expansion in private hospitals has led to performance improvements in public hospitals	Malaysia	2006–2010	Ministry of Health
Aksan et al 201043	The change in capacity and service delivery at public and private hospitals in Turkey: A closer look at regional differences	Retrospective study	Evaluate the change in capacity and service delivery at public and private hospitals in Turkey.	Turkey	2001–2006	Statistical Almanacs of Inpatient Services
Kebriaeezadeh et al 201324	Trend analysis of the pharmaceutical market in Iran; 1997–2010; policy implications for developing countries	Descriptive and cross-sectional study	Analyse the Iranian pharmaceutical market	Iran	1997–2010	Iranian Ministry of Health, Food and Drug Organization

ADDOAccredited Drug Dispensing OutletOB/GYNobstetric and gynaecology

Table 3 presents country examples identified through the scoping review, illustrating how routine data from the private health sector can be used for each of the governance functions and subfunctions of the health system.

Table 3. Governance examples.

Governance of whole system	Data types	Select country example
Set policy and vision Define a strategic direction for the health system as well as policies, laws and guidelines to achieve it.	Service delivery	Turkey faced substantial regional inequalities both in access to healthcare and key health indicators. In 2004, the Turkish government launched a programme to incentivise investment in private hospitals as a strategy to reduce health disparities.43 The Ministry conducted a study to assess changes service delivery capacity of both public and private hospitals from 2001 to 2006 using service delivery data on public and private hospitals in 81 provinces from the Statistical Almanacs of Inpatient Services. Private ownership of hospital increased from 22.1% in 2001 to 29.4% in 2006. During the same period, the increase in public hospital capacity was very small. The study confirmed the Ministry’s policy objective to increase private investment.
Include stakeholder voiceRequires both a fora and government capacity to initiate and sustain processes to engage stakeholder groups.	Resource generation/service delivery	The government of Namibia implemented an integrated pharmaceutical management information system in order to improve availability of antiretrovirals (ARVs) for people living with HIV.23 The system consisted of four different interlinked data collection and management tools. Its implementation required significant levels of stakeholder engagement between the Ministry of Health (MoH) and service delivery points, including capacity building. Due to those efforts, the government was able to identify ‘quick adopters’—that is, pharmacies that were interested in transitioning from a paper-based system to an electronic one which enabled an incremental roll-out. In 2018, 90% of transactional commodity and patient dispensing data from more than 85% of all ART sites were collected.
Generate information and intelligence Collect, analyse and use data, information and intelligence of and for the health system.	Service delivery	WHO supported the Kenya MOH to assess reporting comprehensiveness for 19 malaria indicators.48 The study extracted data from the DHIS-2 online platform between 2011 and 2015 from 6235 public and 3153 private facilities. Overall, the study’s results showed significant improvement in the completeness of malaria indicators since the adoption of the DHIS-2 system in 2011. Although private facilities are essential contributors to malaria care, they had lower reporting rates than their public counterparts. The study suggested that higher public investments in infrastructure and human resources dedicated to DHIS-2 reporting system as well as higher demand for public data reporting to meet donor requirements and justify resource needs explained the reporting differential.
Legislate and regulate	Resource generation	Since 2003, the for-profit private health sector was growing fast in Afghanistan and remained largely unregulated.47 The MOH launched an initiative in 2008 to increase oversight and regulation of the private sector and to better leverage its public and private resources to achieve its national health goals. Based on a WHO study, the Afghan government took several actions to strengthen its oversight of the private sector by (1) establishing a public–private dialogue forum; (2) setting a minimum quality of care standards for private hospitals; (3) issuing public–private partnership and procurement regulation and (4) establishing a committee to review complaints from the for-profit private sector and allow for decision enforcement and sanctions.
Resource generation-Plan and allocate resources	Infrastructure and medical equipment	The Nigerian MOH conducted a secondary analysis of its facility register to assess the geographical and sectorial distribution of healthcare facilities and to discuss its implications for achieving UHC strategies.37 The analysis showed (1) one-third (33%) of all healthcare facilities are privately owned; (2) three-fourths (76%) are private secondary facilities with a high concentration of specialised healthcare workers in the private sector and (3) private facilities are concentrated in the southern part of the country. The analysis supported the case for better integration of the private sector in the Universal Health Coverage (UHC) strategy to address choice and quality concerns of a category of patients, especially in the southern part of the country.
Resource generation-Plan and allocate resources	Human resource	The Tanzanian MOH developed two platforms to capture information from the public and private sectors: the Human Resource for Health Information System (HRHIS) and the Training Institution Information System (TIIS).27 The national rollout of these platforms took 6 years and faced multiple challenges. The private sector was initially reluctant to share its human resource data. However, after some advocacy, a few private facilities agreed to participate in the system as pilots. The HRHIS captured reliable information on healthcare workers as well as their level of training. HRHIS data coverage was estimated to be 94% in the public sector and 83% in the private sector, showing substantial participation by both sectors. The Council Health Management Teams were able to have a realistic picture of the current HR supply and expertise across the public and private sector and was to allocate personnel to health facilities accordingly.
Resource generation-Plan and allocate resources	Pharm-ceuticals and consumables	In Zambia, the Medicine Regulatory Agency and Central Medical Stores created an innovative programme to collect market data on medicine imports and sales through a unique national platform used by both the public and the private sectors.21 IQVIA, an international commercial health data company, the Medicine for Malaria Venture and TESS Development Advisors, assisted the Zambian counterparts to design the system as well as analyse the data. While the Zambia malaria market is mostly dominated by the public sector, the private sector represented between 2% and 29% of imports market value and 4% and 27% of import units sold, depending on years. The data collected allowed Zambian authorities to better monitor policy compliance with both public and private stakeholders.
Service delivery-Deliver public health	Disease surveillance	The Nigeria MOH adopted an Integrated Disease Surveillance and Response strategy in 2005 which required both public and private facilities to participate in routine reporting for 41 priority diseases and conditions.29 The MOH estimated that the private sector managed 33% of facilities and 60% of healthcare contacts. A cross-sectional study conducted among 507 private health facilities in South-West Nigeria between 2016 and 2017 showed that only 40% of facilities routinely reported diseases to health authorities, underscoring the poor participation of the private sector to the disease surveillance system and providing insights on how ensure data reporting tools are widely accessible to and that private healthcare workers are trained disease surveillance regulation and data reporting.
Service delivery-Assure quality of services	Infrastructure service statics, health finance	Brazil has a robust health information system with several databases publicly available online. The Hospital Information System (SIHSUS - Sistema de Informaçōes Hospitalares do Sistema Único de Saúde) presents demographic information, hospital lengths of stay, costs, diagnosis and patient health outcome for all hospital encounters performed under its unified health system (UHS).30 The National Registry of Health Facilities provides information on facility size, type and resources, updated monthly. Researchers used those databases to assess trends in the incidence, mortality and costs of sepsis for Brazilian UHS hospitalisations from 2006 to 2015. The data revealed that the private sector had lower levels of sepsis mortality, shorter length of stay and lower average daily hospitalisation costs than the public sector. Smaller hospitals were also found to be more efficient than larger ones. The availability of comprehensive facility and patient-level data on service delivery allowed the MOH to conduct detailed performance analysis of the ‘specialised care’ across different types of facilities, as well as the identification of best practices to improve specialised care delivery.
Service delivery-Improve quality of services	Service statics consumer info	Turkey started a major reform to improve access to primary healthcare (PHC) in 2003.26 In 2005, the family-medicine (FM) model was introduced, expanding the range of maternal and child health services that were provided free of charge in both public and private facilities. By 2011, the entire population was covered by this programme. A study assessed the effect of the FM model on service utilisation and user satisfaction using data collected monthly from primary, secondary and tertiary-level facilities consolidated at the province level and centralised by the MOH. The MOH was able to conclude that the introduction of the FM model was associated with an increase in PHC utilisation, as well as modest shift in utilisation from secondary to primary care. Unfortunately, while the study distinguishes between public primary and secondary healthcare, it grouped all private healthcare into one category.
Health financing-Purchase goods and services	National Health Insurance (NHI)	Belize implemented an NHI programme using both payment-for-performance and contracts with public and private primary care providers to incentivise facilities to achieve specific health outcomes among the population.25 The NHI conducted a difference-in-difference study to compare general practitioner productivity per hour and rational use of health resources and consumables between public and private facilities. Overall, the study showed that health districts with contracts through the NHI programmes had better health outcomes. In addition, the private sector achieved better technical efficiency with higher levels of patients seen per hour as well as images and laboratory tests ordered per patient. However, the private sector also showed a lower level of compliance with clinical protocols, lower patient satisfaction and overprescribing. As a result, the public sector received a higher percentage of monthly efficiency payments.

DHIS-2District Health Information System 2

Discussion

Findings from the scoping review indicate significant differences in routine data collected on the private health sector by countries and in how that data are used for system governance by the public sector. This study found differences in the completeness and accuracy of data collected and used, the type of data (core vs services vs financial), which providers are included in data collection, and the range of data collected from providers. Some countries displayed sophisticated systems of routine private health sector data collection and use by the public sector. Most did not. The most complete data collection systems were found in Brazil and Mexico which collected data on patient demographic characteristics, cost of care, duration of stay, health outcomes and even quality of care through a number of different national data sources.^{30 36 40}

Comprehensive data collection systems were also identified when a national insurance scheme was in place, such as in Morocco where information was available on infrastructure distribution, health financing as well as services delivered at facility level.³² Compiling different data sources at regional level, India had the same information in addition to healthcare workforce.³⁸ That was also the case in Belize. However, in other countries, the routine collection of data from private facilities by public entities appeared to be much more reliant on vertical programmes and pilot projects.

As important as what the study found is what it did not find: the majority of LMIC countries did not appear in studies in data collection and use by government. While many studies have shown that private sector data frequently exists,^{2 21} our search found very few studies which could confirm or illustrate collection and use of routine private health sector data by the public sector. We are unable to say if this lacunae in the literature is due to the absence of routine data collection and usage or the challenges in documenting and studying governance, but our study suggests there is a real gap, either in practice or study and documentation.

Limitations

This study has several limitations. Foremost among them is the narrow focus on the collection and use of private sector data for governance. The findings of the scoping review allowed the research team to see multiple examples of where governmental systems have collected private sector data but have not used them for governance purposes. Regarding data collection, there is an inherent selection bias in this, as the team was unable to know where data may exist and not be shared with researchers; or where data may be available, but no researchers have thus far made use of the data for studies published in peer-reviewed journals identified in this review. While the results can say where data exist, the absence of studies identified cannot be interpreted as indicating the absence of data; rather, as noted above, we know that relevant private sector data exists but are unable to know if it is put to use to carry out governance functions or not. We chose to focus on articles that included both collection and use of private health sector data routinely collected by public entities. If data were used for governance purposes by the public entities but the articles did not reflect that we could not add it to our scope, which might have been limiting. Similarly, when data were collected in an ad hoc basis, as is the case of punctual surveys, for instance, the studies were excluded from our scope.

In addition, while included studies did indeed involve the collection of private sector data by the public sector for governance purposes, the resulting analyses were often conducted by researchers and not by those public entities themselves. This scoping review hence gives a sense of the potential uses of private sector data for different governance functions but does not necessarily reflect how private sector data is effectively used for governance.

The COVID-19 pandemic was not a specific focus of our data collection method. None of the articles included were linked to that specific topic, which could be explained by the fact that we only included articles published up to 2021. An interesting angle for further analysis would be to assess whether the pandemic has changed how the public sector collects and uses routine data from the private sector—however, this was not included in our scope.

This scoping review was limited to publications in English, and so omits large LMIC regions. If countries in non-English-speaking Asia, Francophone Africa or Latin America had documented literature on the collection and use of private sector data by the public sector that was not in English, it would not have been captured in our analysis. For this reason, our results can only be used to describe some of the activities happening in the English-speaking or English-language researched, regions of the world.

Conclusion

First and foremost, it is important to note that the narrow focus of our review looks only at routine data collected for, or used for, regulation of the private sector by the public sector. A great deal more data exists regarding the private health sector in LMICs than is included in our study; what we sought to elucidate was what, if any, of that data is put to use for the improvement of overall health system functioning. This study question is important because the private sector represents a substantial percentage of essential health service delivery in nearly all LMICs. Despite this, the experience of the authors, individually during our many years of work, and through our professional roles studying and advising on private sector governance for the World Bank, USAID, WHO and others, lead us to believe that the availability and use of data on private sector infrastructure, staffing, services, quality and costs are limited in most LMIC ministries of health. If true, the implications are significant: limiting government capacity for resource allocation planning, emergency preparedness and assuring the ongoing delivery of affordable and quality health services. This all depends on access to and use of timely and accurate data. Proper health system governance cannot function with only a fraction of service data in hand—only representing the public sector. This was evident most recently during the COVID-19 pandemic, when lack of information on private sector staff availability and infrastructure capacity—on continuous positive airway pressure machines or oxygen provision, for example—hampered national responses in many countries.

The goal of this study has been to examine the type of data governments collect on the private health sector and to understand how private health sector data can be used to strengthen the public sector’s capacity to govern a mixed health system. The scoping review has provided evidence on the diverse range of routine data collected on the private health sector by the public sector as well as some of the challenges involved in that process. Evidence from countries such as Brazil and China demonstrates that investments in strengthening government capacity to collect and use this data enable health ministries to perform sophisticated analysis that strengthens its ability to perform essential governance functions needed to improve a mixed health system’s overall performance. Given the importance of this subject, the paucity of studies identified shows that more attention is needed to ensure that this key area of health system governance is undertaken and that lessons learnt are widely shared.

Data availability statement

Data are available on reasonable request.