Workforce requirements for comprehensive ischaemic stroke care in a developing country: the case of Saudi Arabia

A wide body of evidence exists on methods for workforce supply estimation, covering both the supply element and the demand for services. Some such methods include regression-based models, simulation models, Markov chains, and system dynamics. Epidemiological modelling is commonly used to determine demand for health care services, considering population demographics, health status, and the types of service required [12, 13]. Existing supply has typically been estimated via national databases, online sources (such as the indicators available from the OECD), or tailored surveys in specific countries [12]. The future supply of healthcare professionals, meanwhile, is often modelled via a stock-and-flow approach, incorporating the training and employment of new healthcare specialists, retirement, immigration, and emigration, and taking into account practical changes such as flexible working hours and improvements in hospital efficiency [14]. Our approach used some elements of these methods (such as epidemiological modelling to determine future demand for stroke care services); however, we did not seek to model future staff supply via the methods described above. Instead, we combined the estimates of future demand with those of current staff availability and used international standards for stroke care as the basis for calculating the additional resources required.

A five-step process was followed to quantify the workforce gap for ischaemic stroke. First, comprehensive data collection was performed to determine the current workforce in the Kingdom for ischaemic stroke management and the existence of specialist stroke units. Data were collected for each region and supplemented with interviews with officials from the Saudi Ministry of Health. Second, an epidemiological model was created to predict population changes over the next decade, which was combined with local stroke incidence data to estimate the number of strokes occurring each year. Third, a model of future care was developed, reflecting changing patient care options and the availability of new therapies for specific groups of stroke patients, and based around three facility types. Fourth, a review of stroke care guidelines and recommendations from around the world was undertaken to inform the staff-to-patient ratio assumptions for the full range of specialists required within each of the three care facility types. Fifth, these elements were combined with future expectations of the use of novel stroke interventions to estimate staff requirements (and their costs) in stroke units and rehabilitation facilities over 10 years.

An extensive search of multiple sources was undertaken to ascertain the current workforce for stroke care within Saudi Arabia. Workforce data were collected from the Ministry of Health’s statistical yearbook from 2016 [15]; this included all physicians by specialisation, nurses, and all allied healthcare staff, with the exception of dietitians and psychologists, by geographical region. The data collection was supplemented via a survey of all hospitals providing stroke services within Saudi Arabia and was co-ordinated by one of the authors (FA-S) as part of the Ministry of Health’s Stroke Vision 2030 programme. Regional directors in the Kingdom were contacted by email via the Ministry of Health, with the survey cascaded down to individual hospitals to determine staff provision dedicated to stroke care. Thus, hospital-level data on staff in each of the specialties was recorded and scaled up to the national level. The data collected were validated by the authors to confirm accuracy.

To accurately determine staffing needs in the Kingdom over the next decade, robust estimates of stroke incidence were required to reflect population and demographic changes. Epidemiological modelling was therefore undertaken using local population, mortality, and stroke incidence data to predict the number of first strokes occurring each year over the next decade.

Data from the Saudi General Authority for Statistics were obtained to form the basis of the current population, split by age category and gender and available for both native Saudis and non-Saudis [8], allowing the populations to be modelled separately to reflect the effect of immigration and emigration. Mortality data were applied to the baseline population data to model deaths occurring over a 10-year period [8]. The movement of non-Saudis in and out of the country (many arrive to work but return to their home countries upon retirement) was also factored into the analysis. Thus, year-by-year population estimates were obtained by modelling births, deaths, and the gradual ageing of the population.

These data were then combined with Saudi age- and gender-specific rates of first stroke to estimate the number of first strokes occurring per year, of which 85% were assumed to be ischaemic [16, 17]. In each year, the population was stratified into age groups and multiplied by the stroke incidence rates for the corresponding age bands to estimate the number of new strokes occurring. The age- and gender-specific stroke risks were assumed to be constant over the horizon of the analysis, and therefore the growth and ageing of the underlying population drove the changes in predicted number of strokes over time. Recurrent strokes were excluded due to a lack of robust data on their incidence.

We then developed a generic structure for patient management, using the information obtained from the research and interviews described previously, and a framework was created according to expected future treatment approaches. In Saudi Arabia, only around 5% of stroke patients are admitted to acute stroke units located in comprehensive stroke centres with specialist staff. Thus, most patients do not have access to treatments such as intravenous tissue plasminogen activator (IV-tPA) and mechanical thrombectomy (MT). The proposed care model was based around increased adoption of IV-tPA and MT, drawing on international evidence regarding patient eligibility and applying a gradual uptake of these interventions [17‐21].

Evidence suggests that around 85% of all strokes are ischaemic [22], and of these, 40% are large-vessel occlusions (60% are other-vessel occlusions) [17]. Amongst patients with LVO, the model assumes 5% use of MT in year 1 (either as a single intervention or in combination with IV-tPA), rising to 50% by year 10. All remaining LVO patients, plus all patients with OVO, were assumed to receive IV-tPA alone or non-thrombolytic therapy. Across all LVOs and OVOs, the model assumed that the proportion of patients treated with MT would increase from 2% in year 1 to 20% in year 10.

To reflect this shift in treatment mix, the model focused on three types of stroke facility, each requiring a different mix of specialist staff: hyper-acute stroke units (HASUs), acute stroke units (ASUs), and inpatient rehabilitation centres (Fig. 1). Under this care model, an increasing proportion of ischaemic stroke patients would be admitted to a HASU, regardless of stroke severity and the treatment to be administered, and would subsequently be discharged to one of the following locations: ASU, home, or inpatient rehabilitation. Our approach considered a phased introduction of HASUs, ASUs, and inpatient rehabilitation services over time, with 11% of patients assumed to be managed in a HASU in year 1, rising to 70% by year 10 with the establishment of new services. In the absence of sufficient capacity to allow all patients to be admitted to a HASU, all remaining patients were assumed to be managed using existing stroke care services.

Alternative models of care exist and depend on local resources and the feasibility of re-organising existing services. However, for the analyses presented here, we have adopted this approach to future stroke care. Under this model, we assumed that 50.5% of patients would be discharged directly to an ASU, 16.6% to inpatient rehabilitation (either directly from the HASU, as indicated by the dotted line in Fig. 1, or following a spell in the ASU), and 37.7% to home, based on interviews with stroke experts in Saudi Arabia.

The fourth phase of the study involved a review of international stroke care recommendations to identify best practices regarding workforce requirements for HASUs, ASUs, and inpatient rehabilitation. The review considered studies from countries with more developed and co-ordinated stroke care services and which had published detailed workforce requirements, and we adopted these guidelines as ‘gold-standard’ approaches for the Saudi healthcare system to adopt. The national stroke clinical guidelines issues by the Royal College of Physicians in the United Kingdom and recommendations from the Canadian Stroke Network were used as the basis for staffing ratios for care in HASUs and ASUs [5, 6]. Both documents provide detailed recommendations regarding the establishment of new stroke units and guidance on staff requirements. Requirements for stroke rehabilitation units were based upon standards from the Australasian Faculty of Rehabilitation Medicine [7], which provide staff-to-patient ratios for inpatient stroke rehabilitation care. Although implementation of such recommendations and guidelines in countries with less developed systems is not straightforward, this approach provides a ‘best-practice’ target for evaluating requirements in Saudi Arabia and reflects the Ministry of Health’s willingness to adopt successful strategies. Such an approach provided a framework upon which future staff requirements could be estimated.

A detailed breakdown of the recommended staff needs from these countries is provided in Additional files 1, 2, and 3.

The staffing ratios described were used as the baseline assumptions for requirements in Saudi Arabia. However, differences in the countries’ models of care, combined with the geographic spread of the population in Saudi Arabia, meant that adjustments were necessary to ensure feasibility. The international recommendations were therefore supplemented by consultations with local stroke experts (including radiologists and rehabilitation staff) to develop Saudi-specific staffing ratios. These discussions allowed us to derive local staff-to-patient ratios that covered the full range of staff, with the exception of nurses, whose resources were assumed to be re-organised and re-distributed according to the new care model.

We used the stroke incidence projections and data on population density to model the expected number of ischaemic strokes per year in each of the Kingdom’s 13 regions, allowing a more realistic assessment of local needs to be captured and scaled up to a national level. Staff requirements were then determined by calculating the expected number of beds needed in the three care settings. Bed requirements for HASUs and ASUs in each region were calculated based on an assumed occupancy of 50 cases per bed per year. For inpatient rehabilitation, bed needs were based on 16.6% of patients requiring inpatient rehabilitation after discharge from the HASU; length of stay in the rehabilitation unit was based on a mean of 25 days, using data from Australia [23, 24].

The number of interventional neuroradiologists (INRs) required for the HASUs in each region was calculated based on the number of ischaemic strokes and an acceptable number of LVO stroke cases per INR per year. Given the learning curve involved in the use of MT, we assumed that the number of procedures per year undertaken by each INR would increase from 15 in year 1 to 63 by year 10. One full-time equivalent (FTE) stroke neurologist was assumed to be required to cover five beds within a HASU, and based on discussions with a Saudi rehabilitation expert, the number of FTE physicians specialising in physical medicine and rehabilitation within HASUs and ASUs was assumed to be 0.084 per five beds (equivalent to 0.5 FTE per 30 beds).

The final staffing ratios (per five beds) applied for the model are shown in Table 1, covering the various specialist staff types needed in each care setting. All of the numbers given in the table were taken directly from the stroke guidelines issued in the United Kingdom and in Canada, the details of which have been previously published [5, 6]. Nursing staff are considered outside the scope of the study since sufficient numbers were assumed to be available via re-organisation of existing staff.

It was assumed that the new HASUs, ASUs, and rehabilitation units could be created within existing hospitals via reconfiguration of current infrastructure. Therefore, the economic analysis focused around the costs of new staff requirements. Salary data from the Saudi Ministry of Health were obtained for each staff type [25], to which 40% overheads were added (Additional file 4).

Using the staffing requirements described in the previous section, the total costs of the new care model were estimated by applying the staff unit costs to the individual resource components. In order to focus on the cost of staff provision, we have excluded costs related to the acquisition of new equipment (for example, bi-plane angiography suites).