Planning for the future of the Australian midwifery workforce: the Midwifery Futures workforce model

Identifying supply

The number of Australian midwives in December 2022 was sourced from the National Health Workforce Data Tool (Table 1).¹⁰ The proportion of midwives working in midwifery and providing clinical care was 79% in 2019,³ with the same proportion identified in the Midwifery Futures survey of 2023.¹¹ The average hours worked per week per midwife was 20 h in 2019.³ The current full-time equivalent (FTE) number of midwives was calculated as midwife headcount × proportion working in a clinical role × (average hours worked per week/38 h per week) (Table 2).

Identifying demand

National live birth rate numbers were sourced from the United Nations World Population Prospects 2022 data set¹² (Table 1). A pregnancy multiplier rate was estimated by adding the proportion of stillbirths,¹³ abortions,¹⁴ and miscarriages¹⁵ to live births. Doing so recognised that midwives also provide care for people experiencing stillbirth, abortion, and miscarriage. For the past 5 years, 2019–2023, the number of pregnancies has averaged 1.65 times the number of live births, and we assumed this would remain constant over time (Table 2).

Demand for midwifery services (workload) was taken as the number of pregnancies in Australia each calendar year per FTE midwife (Table 2). The average workload per FTE midwife in Australia between 2018 and 2022 was calculated as 40 pregnancies per year. This workload reflects the existing levels of casemix complexity in the Australian maternity care system, and it was assumed that this casemix and, therefore, workload would remain constant across all future years in the model.

Modelling future midwife numbers

Each subsequent year’s projection of the number of midwives providing clinical care was calculated by adding inflows to the previous year’s number and removing outflows (Fig. 1). Inflows were the number of new Australian graduates and midwives immigrating to Australia. Graduate numbers were supplied by the Nursing and Midwifery Board of Australia. The default number of graduates per year used in the model was the annual average in years 2019–2023.

Fig. 1.

Modelling inflow and outflow variables.

Outflows were the number of deaths, retirements, and voluntary attrition. Deaths were calculated as the headcount in each age category multiplied by the age-specific mortality rate per 1000 people for that age category (Table 2). All midwives over the statutory retirement age of 67 were assumed to have retired. As the data were reported in age category blocks of 10 years, it was assumed that 80% of those aged 65–74 years would retire.

Reliable data on immigration and voluntary attrition could not be located. Instead, the net average change in the midwife headcount was calculated to estimate the net effect of both immigration and voluntary attrition after accounting for the other known changes (Table 2). Using data from the preceding 5 years (2017–2022) showed the net average change was +1.9%, indicating that immigration exceeded voluntary attrition over the past 5 years. The model used this default rate of net change.

Scenario testing

Base model assumptions are provided in Table 3. Adjusting key variables in the model allowed permitted examination of the expected results under different scenarios.

A survey of midwives on the Nursing and Midwifery Board of Australia register was carried out specifically for Midwifery Futures¹¹ and asked midwives about their future intentions. Of the 2428 respondents, 887 (37%) said that they were currently considering leaving midwifery altogether. However, not every midwife who considers leaving the profession will do so. Actual attrition rates are lower, even when accounting for retirement and death. Previously, age-disaggregated 5-year attrition rates for Australian midwives between 2013 and 2018 were estimated to be 7%.⁴ The average attrition rate for midwives in the State of the World’s Midwifery 2014 data set was 8.2%.¹⁸

Based on this limited evidence, the effect of a 10% attrition rate on workload was modelled. Graduate numbers and weekly working hours were independently adjusted to determine the changes required to maintain the current number of FTE midwives at this attrition rate.

Forecast midwife numbers in Australia 2023–2030

Using the default assumptions for each of the key variables (Table 3), the estimated number of midwives for the years 2024–2030 is shown in Fig. 2, both by headcount and FTE. Required FTE indicates the number of FTE midwives required to achieve a workload of 40 pregnancies per FTE midwife per year, assuming the birth rate remains the same as in 2022. The numbers for the years 2017–2023 represent the actual headcount and FTE for each year.

Fig. 2.

Modelled number of midwives for Australia, 2023–2030.

In this model, the number of midwives is projected to increase. This model assumes graduate numbers will continue at their current annual rate of 1305, and inflows will continue to exceed outflows, with a net change (immigration–attrition) of 1.9% annually. The increase means that by 2030 the average workload per FTE midwife per year could fall from 40 to 34 pregnancies. Alternately, fewer graduates and immigrants would need to be recruited.

Modelling scenarios

The lack of reliable data regarding attrition did not allow projection based on historic data. Instead, we modelled the impact of a 10% attrition rate.

The impact of a 10% attrition rate would be rapid and devastating (Fig. 3). By 2030, a deficit of 3943 FTE midwives would occur, with a reduction in headcount below 2022 numbers of 9496 midwives. Assuming pregnancy rates remain the same, the average workload for midwives would rise each year. By 2030, a 50% increase in workload to 66 pregnancies per FTE midwife would be required.

Fig. 3.

Effect of 10% annual attrition rate on Australian midwife numbers, 2023–2030.

To fill the gap without additional immigration, it would be necessary to produce more graduates (Fig. 4). This presents a significant challenge, as 4100 new graduates would have been required in 2023 – a threefold increase on the current number of new midwifery graduates. The number of graduates required is especially high in the first year, as the model assumes all midwives aged over 67 years will leave the workforce, which history tells us is unlikely. Once these retirees have been replaced, required graduate numbers fall then level out at twice the current number.

Fig. 4.

Number of Australian midwifery graduates required to replace a 10% attrition rate, 2023–2030.

Alternatively, it would be possible to fill the gap if remaining midwives worked more hours per week in midwifery. By 2030, the average midwifery hours worked per week per midwife in this scenario would need to rise to 33 h (Fig. 5).

Fig. 5.

Average hours per week required to manage a 10% attrition rate, 2023–2030.

Strengths and limitations

A strength of this model was the assumption that all women would receive pregnancy care in a continuity of carer model. Continuity of midwifery care is recognised as the optimal maternity care model, with better perinatal and maternal outcomes²⁵ and reduced healthcare costs.²⁶ Using this approach illustrated the size of the midwifery workforce required and how it is possible to meet, indeed, to exceed, current demand.

We encountered a lack of data for voluntary attrition and for immigration. The difference between outflows and inflows must be small, as the total number of midwives registered each year has not changed significantly between 2017 and 2022 (Fig. 2). Reliable data for the number of new graduates joining the profession from Australian training institutions each year were available, and we could estimate the expected number of retirements and the very small number of deaths in service. Consequently, the static totals for these years may disguise a level of underlying ‘churn’, with midwives leaving the profession being replaced by a similar number of new registrations from overseas. The true scale of this churn remains unknown.

Our model assumed that the workforce was adequately meeting demand in 2022 and the years prior. If this was not the case, then the predicted FTE requirements may under- or over-represent the number of midwives. Our model is unable to account for changing workforce requirements relating to changes in the skill mix of midwives, casemix complexity, and factors that affect clinical productivity. Changes in the proportions of women receiving care in different models of care might alter demand for midwifery services, and we were unable to account for this in the model. Like the Health Workforce Australia model, we focused on modelling the clinical workforce. We acknowledge that for the midwifery profession to remain robust, it is vital that midwives also have roles in all the places where decisions about the profession is made, in education, policy, and research. While we have performed modelling for each state and territory (provided in the Supplementary material Fig S1), regional and rural areas are typically most affected by workforce shortages. Our model does not include this analysis as we did not have access to detailed information about the place of practice for midwives.

Recommendations

We recommend a robust modelling tool be developed, including measures of both the clinical and non-clinical midwifery workforce, with reports on workforce predictions according to state and territory and according to levels of rurality. Policy decisions should be informed by best practice modelling and include considerations of the effect of suggested changes on the nursing workforce.

Universities, health services, and policy makers working together to increase the number of midwifery students can provide a future safety net against attrition rates. Efforts to retain midwives and promote a small increase in working hours would benefit from increased positions in midwifery continuity of carer roles, and a focus on the provision of workplace flexibility, adequate working conditions, and appropriate remuneration.