What do we need to know? Data sources to support evidence-based decisions using health technology assessment in Ghana

This domain includes demographics, vital statistics and burden of disease. We are mindful of the quality and recognise the incentives in operations concerning data collection and analysis even for apparently well-run endeavours such as the Demographic Health Survey (DHS) and the like [23, 24].

Data on the unit costs of health services, interventions and medicines are crucial for economic evaluations and HTA [22]. When HTA has been used to inform decision-making in countries around the world, the costs used in any economic evaluation commonly tend to focus on resources available to a third-party payer. In Ghana, this would reflect the perspective of the National Health Insurance Authority. In our economic evaluation of antihypertensive medicines, the costs for medicines on the essential medicines list [17] and medical services were derived from the NHIS tariff (available on request). The NHIS tariffs are sometimes dated and a cost manual for services is being developed based on the Joint Learning Network toolkit [42]. The medicines tariffs are revised each year and are derived through a combination of surveys, manufacturer’s reference pricing, mark-ups and other considerations. It is a joint process involving the MoH, the Ghana National Drugs Programme, and provider groups. The WHO and Health Action International, in partnership with the MoH, coordinate occasional surveys with a focus on medicines’ accessibility and affordability [43]. The most recent survey was published in 2004 [44] but a survey was conducted with MoH and WHO collaboration in 2018.

In some cases, researchers may be interested in the perspective of the consumer, especially given that only four in ten people are covered by NHIS [12]. Other consumers, particularly if they are not covered by any private health insurance, may be liable for extensive out-of-pocket (OOP) payments [45‐47]. The Demographic and Health Survey provides some information on OOP spending [27]. OOP payments are generally estimated using the National Health Accounts [48], the WHO Global Health Expenditures Database [49] and Government budgets. The GLSS also includes some relevant items [28]. The HDSS centre in Navrongo, together with two centres in Burkina Faso and Vietnam, aims to estimate OOP expenditure in a household survey and develop a set of questions to facilitate OOP disease-specific measurement [50]. The MoH collects information on health expenditure using the National Health Accounts from the World Bank and OECD and plans to supplement this with local data sources in 2019. They will liaise with the Finance Directorates of all agencies under the MoH and likely upload it to the upcoming National Health Observatory [51].

Information on the consumption of health services is vital for economic evaluations. For example, when considering an intervention to reduce a cardiovascular event such as a myocardial infarction, it is important to understand the usual treatment pathway for someone who experiences a myocardial infarction and the resulting resource use and costs. If the intervention also reduces the occurrence of stroke, then it is necessary to account for the pathways and resource implications of someone having a stroke. The main source of routinely collected health data is from largely public (and some private) service providers such as NHIS claims and the DHIMS 2. NHIS claims are recorded – the current proportion of electronic claims is 25% (4th quarter 2019; personal communication, Dsane-Selby) – the rest are paper based [52]. There are annual NHIS reports but there is some delay in making these available online [12]. The NHIS claims data would be an obvious source to explore patterns of use of health services not only at the aggregated level but also, ideally, at the individual level. The DHIMS 2 also has a vast amount of data on health services use [53].

Information on pathways of care, particularly for NCDs, is sparse. There are normative standard treatment guidelines [18] but, in the absence of any monitoring data, it is not known to what extent these are implemented in actual practice. Ghana has an essential medicines list [17, 51] that guides the medicines available for subsidy on the NHIS [54]. The MoH is working with the private sector to integrate their service data into the DHIMS. The private health insurance industry in Ghana has 12 providers and mainly covers corporate entities and less-vulnerable populations [55]. There are claims data held by the private health insurance providers, some of which has been made available for research (personal communication, Hollingworth). These data can be used to examine patterns of health service use and costs.

Health outcomes are often expressed in natural units such as lives saved or a change in some clinical measure such measurement of blood pressure. Economic evaluations that only consider such measures are limited in their usefulness – the relative value of the intervention can only be ascertained within the scope of that health outcome [22]. A more useful measure is one that includes both mortality and morbidity aspects such as disability-adjusted life years (DALY) or quality-adjusted life years (QALY). Each measure requires a valuation of health states (or disability) where one is perfect health and zero is death. Disability-adjusted life years are widely used, especially in LMIC settings, and are based on the Global Burden of Disease (GBD) study linked to years of living with a disability [56].

An alternative generic measure available to researchers and decision-makers in Ghana is the QALY, often used in high-income settings but relatively uncommon in LMICs, particularly SSA [56]. Estimating QALYs relies on the use of a generic or disease-specific multi-attribute instrument to measure health-related quality of life (HRQoL). The most widely used generic measure of this type is the so-called EQ-5D (the descriptive system comprises of five dimensions) developed by the EuroQol Group [57]. Typically, the measurement of changes in HRQoL are reported directly from patients, but the utility (valuation) of these changes are based on public preferences using a choice-based method (commonly time trade-off in the case of EQ-5D). The valuation exercise is usually undertaken separately and the results are often referred to as the population tariff. However, there are no published EQ-5D (three or five level) value sets for Ghana. An orthopaedic hospital in Accra is administering a multi-attribute instrument (SF-36D) to patients before and after surgery [58]. In Ghana, some authors have evaluated HRQoL in studies encompassing university students’ wellbeing (shorter version of the WHO quality of life (WHOQOL-BREF) across four domains) [59], stroke (HRQOLISP-40 instrument across seven domains) [60], asthma (Asthma Quality of Life Questionnaire) [61], and leprosy (disease-specific) short form 20 health survey (six domains) [62]. One study (conference presentation) used the three-level version of EQ-5D in people living with HIV [63]. There is work underway to develop EQ-5D value sets within SSA; sets are now available for Zimbabwe [64] and Ethiopia [65].

Again, there are data on equity, depending on the specific aspect and the pragmatic applicability for HTA purposes; DHS, GLSS, and the Study on Ageing and Adult Health (SAGE) have data on equity. Although not necessarily quantified (or quantifiable), HTA processes can explore context-relevant value judgements that may modify interpretation of a cost-effectiveness analysis. Equity considerations are important in priority-setting decisions. These may include aspects relating to equitable access and the provision of health services to vulnerable populations, plus issues relating to gender, socioeconomic indicators, and tribal and religious affiliations. There may also be geographical aspects to consider, including urban versus rural differences in access and outcomes. There may be a pressing need to address inequities in health outcomes for marginalised groups; to do so, data on the relevant characteristics is required.

There are some equity-relevant data available in Ghana. The sample design of the DHS provides estimates at the national and regional levels as well as for urban and rural areas. Wealth quintiles are available for households in the datasets [27]. The GLSS also provides data on wealth quintiles [28] and the census has data for the socioeconomic status at the district level [26]. Equity aspects of service use can be explored through the NHIS and the DHIMS, especially at district and facility level. The three HDSS sites can provide some data; for example, one project aimed to understand the impact of UHC on the poor and vulnerable. The HDSS at Navrongo was one of two impact of UHC study sites (the other in Vietnam) exploring health services need, access, use, quality of care and reliable measurement of socioeconomic status at the district level [66]. The GBD study collaborators published a study on the healthcare access and quality index based on mortality from causes amenable to personal healthcare in 195 countries and territories, including Ghana [67]. The EQUIST tool from UNICEF (http://​www.​equist.​info/​en/​dashboard) will be linked to the DHMIS (Ghana Health Service) by 2020.

Ghana has several large data sources to support HTA (e.g. DHS, GBD and NHIS) with sufficiently rigorous quality assurance processes. There are few accessible data sources for costs and service use. The NHIS claims data are a potentially rich source of data on these but there are access limitations. The initiative to have facilities submit claims through an electronic platform presents an opportunity to harness such a rich source of data for HTA. Data on services and medicines costs are based entirely on NHIS tariffs, but these need to be regularly reviewed and updated. The 2017 value-added tax exemption on medicines should result in an average decrease of 30% in costs for medicines with the costs of selected medicines decreasing up to 70% [68]. There are almost no data for the domains of quality of life [69] and equity.

To the best of the authors’ knowledge, this is the first comprehensive study to describe the landscape of health system data sources to inform HTA in Ghana and could serve as a blueprint for other SSA countries [70]. The main limitation was to accurately and comprehensively identify all the sources of data for HTA [21]. We have not been able to consider aspects of quality and missing data. As such, we consider this an important starting point for a more complete description and assessment of data suitable for HTA purposes in Ghana.

WHO urges member states to develop and improve the collection of data as noted in the Health Intervention Technology Assessment Resolution [2]. Much of the data described in this paper are derived from government-sponsored information systems and census surveys, where statistics are reported to be systematically exaggerated across Africa [24, 71‐73]. We did not assess the quality of the data sources in this paper but there are promising quality assurance processes to ensure data integrity for some sources [27, 30, 52]. A more comprehensive consideration of quality could be pursued in future studies.

Measures for safeguarding data should be reviewed and improved across Ghana to ensure robust, high-quality evidence capture and protect data against undue political interference [24]. The strength of any economic evaluation lies in the data that populates it, and thus improving the quality of data collected in Ghana, and indeed across Africa, will be a crucial step towards more robust evidence-informed policy decisions. We are currently reviewing the data sources used in selected published economic evaluations from SSA.

In terms of epidemiological data, this important component of HTA is adequately served by data available from the GBD and DHS. However, most other contextually relevant data sources are arguably inadequate for the production of analyses that are able to rely on information most relevant to Ghanaian decision problems. Even though the epidemiological data sources are among the most reliable, there remain substantial gaps. There are several studies exploring the prevalence of NCDs, e.g. hypertension in various Ghanaian populations [74‐77], but, to our knowledge, there are no studies looking at the incidence of NCDs. This would require a large-scale cohort study and it could consider the main NCDs and comorbidities of interest in Ghana, i.e. cardiovascular disease, type 2 diabetes and cancer, and the associated risk factors including hypertension and obesity. With regard to preference-based health-related quality of life measures, Ghanaian policy-makers will need to be aware of the implications of such data to inform the routine use of QALY measures in HTAs. For example, in the overwhelming majority of countries in SSA, value sets are simply not available for the most commonly, and widely understood, generic HRQoL tool, the EQ-5D (five dimensions). The adoption of QALYs is feasible in SSA [65], and awareness raising together with action and support for research efforts at developing value sets for Ghana through peer or South–South channels may be an important means of methods development [7]. This highlights the importance of having a reference case [22] that signals preference (and creates a demand) for certain types of data [78].

It is important to highlight that, for decision-making purposes, there will always be gaps in the available evidence and, when there is evidence, it may be of varying quality [79]. Analysts and decision-makers will therefore be routinely confronted by uncertainty that needs to be acknowledged and managed. Judgement will be needed in deciding whether the data used in any HTA is the ‘best available’ for the evaluation problem being examined, and the implications of any bias introduced when using sub-optimal data sources [80]. This further highlights the appropriate conceptualisation of a functioning (‘institutionalised’) HTA system as one that has both procedural and technical elements [81], is supported politically and has cooperation among key organisations [82]. In relation to the technical elements, a key foundation is establishing a framework (or reference case) for the conduct of economic evaluations in Ghana. Such frameworks not only promote consistency but serve to illuminate the key information required for decision-making purposes, and thus help drive more targeted primary research and data collection [22].

At least initially, the new HTA mechanism in Ghana may have to work with less-than-complete access to good quality, contextually relevant data and evidence, and it will likely be necessary to accept some reliance on internationally produced evidence. Adopting approaches that do not apply hierarchies of evidence in an unthinking manner [83] and emphasising ‘fitness for purpose’ may also assist in helping researchers and decision-makers in Ghana make the best use of available evidence, whether produced domestically or from international sources [84]. For example, if health utilities are required, they could be sourced from the literature and reports from HTA agencies in similar contexts [9]. Likewise, costs should ideally be from Ghanaian sources considering local care pathways but can be supplemented by data from the literature, expert clinical opinion, and costs from similar jurisdictions with appropriate adjustment. As the programme of HTA matures in both the public and private sectors, and especially if its outputs can be used to identify research gaps ideally funded by a dedicated department within the MoH as a start, and potentially a separate agency in the future, the information sources should improve in quality and constraints should lessen over time. In addition, improving existing information systems (and addressing any governance challenges) should be a priority, not least because such action will have broader health system benefits in enabling better monitoring of performance and supporting links between best practice and remuneration. Indeed, the World Health Assembly resolution on HITA [2] rightly encourages member states to strengthen the routine collection of health system data as a necessary step towards achievement of UHC. The key to HTA is to use the best available data while being open about its limitations and the impact on uncertainty.