Continuing to flog the dead horse of post-scarcity fantasy, I thought I’d bring my day job to bear on the task, and test the cost-effectiveness of a cleric-based public health measure to reduce infant mortality in a developing (medieval) nation.

Introduction

Infant mortality was a significant public health problem in the medieval era, and in the absence of explicit evidence to the contrary it is reasonable to assume that it is also a significant cause of morbidity and mortality in medieval fantasy settings. Reduction of infant mortality leads to increased wealth as families devote resources to tasks other than childbirth, and also to reduced family sizes, a significant element of economic growth in most developing nations. Furthermore, control over fertility is considered a significant element of women’s emancipation, and reduced infant mortality reduces family size.

As a public health task the reduction of infant mortality is not particularly challenging, but ultimately relies on access to advanced medical care for the small minority of mothers for whom drastic complications arise. Such medical care is not available in many developing nations, but in medieval fantasy settings it is easily provided by divine spell-casters, through the wide range of magical healing technology available. Until recently, it was believed that this technology was too rare and expensive to be used for non-adventuring tasks. In this report we investigate the cost-effectiveness of devoting divine magic to averting infant mortality, under two different intervention models, and show that even under the extremely inequitable economic conditions of a classic medieval fantasy setting, this intervention is cheap, cost-effective, and likely to lead to significant economic gains at very low cost.

Methods

A simple decision model was developed for a medieval fantasy setting under the assumption that its mortality profile was approximately similar to that of Afghanistan. The model was tested for a small community of 2000, but consideration given to its extension beyond this small community. Two intervention models were tested:

  • The Clerical Attendance model: in which clerics attend every birth at the point where complications ensue, and use either of the cure light wounds, cure moderate wounds, and Remove Disease spells to intervene and prevent infant mortality
  • The Potion Distribution model: Because medieval fantasy settings have very poor transport networks, an alternative model based on distributing potions to skilled birth-attendants was considered

Both models were compared to a control model in which skilled birth-attendants were the only healthcare available to the population. Under the Clerical Attendance Model, it is assumed that these women can call a cleric when a woman begins to experience difficulties in labour, and relative risks of infant mortality were assumed on the basis that clerical intervention would improve childbirth outcomes but would sometimes come too late. Under the Potion Distribution Model, the skilled attendant would apply the potion when it was judged necessary, eliminating the need for a cleric to be present and significantly improving outcomes.

The population of the medieval fantasy setting was assumed to have a demographic profile approximately equivalent to modern day Afghanistan:

  • High birth rate: 37.5 per 1000
  • High infant mortality: 134 per 1000 live births

Population was assumed to be 30 million where overall population figures were required. For a hamlet of 2000 people, this leads to the following outcomes:

  • 75 births
  • 10.0275 infant deaths

Infant mortality was modeled on the assumption that women fall into 3 risk categories, with different probabilities of complications in each category. Where complications occur they were assumed to always lead to mortality under the control case (skilled birth attendant only). The ratio of risk groups was:

  • Low risk: 50 births, risk of complications 1.75 %
  • Medium risk: 22 births, risk of complications 30%
  • High Risk: 3 births, risk of complications 85%

This produces 10.025 deaths from 75 births, so is closely similar to the expected number of deaths. The interventions were expected to experience similar rates of complications (used for calculating costs) but reduced death rates. For the Clerical Attendance model, relative risks of death were:

  • Low risk: 0 risk of complications (RR=0)
  • Medium risk: 0.33
  • High risk: 0.25

That is, medium risk women had 1/3 the chance of dying of complications under this intervention, and high risk women 1/4 the risk.

For the Potion Distribution model, deaths in all 3 groups were assumed to be eliminated completely.

Costs for the both models were calculated on the assumption that when complications occurred the following spells were necessary:

  • Low risk: Cure Light Wounds
  • Medium Risk: Cure Moderate Wounds
  • High Risk: Cure Light Wounds, Cure Moderate Wounds, Remove Disease

Spells were cast at a cost of 50gp per level; potions were generated at the costs given in the Dungeon Master’s Guide.

Clerical load was also calculated for the Clerical Attendance model; that is, the number of clerics per 1000 required to support this model on the assumption that a cleric works no more than 200 days a year and sees one case per day.

Quality-adjusted life years (QALYs) saved were calculated assuming life expectancy in the medieval fantasy world was equal to that of Afghanistan (44 years) and outcomes expressed as incremental cost effectiveness ratios (ICERs), that is, the additional cost per QALY. Costs were in gold pieces, on the assumption that a basic medieval fantasy job (Maid) earns 36 gps per year (see Table 4-1, DMG). The wages of the skilled birth attendant were assumed to be 100Gps per year, i.e. approximately 3 x that of a maid in the era.

Sensitivity analysis was not conducted, because this is a blog.

Result

In one year, the 2000-population hamlet could expect to experience 10.025 deaths. Under the two interventions, expected deaths are as follows:

  • Clerical Attendance model: 2.8
  • Potion Distribution Model: 0

That is, at least 7.5 lives were saved per 2000 population. QALYs for the base case and interventions are:

  • Birth Attendant Only: 2878.4
  • Clerical Attendance: 3198.5
  • Potion Distribution: 3322.5

And costs were:

  • Birth Attendant Only: 100 Gps
  • Clerical Attendance: 1978.8 Gps
  • Potion Distribution: 4828.75 Gps

Giving ICERs for the two interventions of:

  • Clerical Attendance: 6.2 Gps / QALY
  • Potion Distribution: 10.9 Gps/ QALY

Both ICERs are significantly less than the annual income of the person saved (36 Gps). The cost per birth was:

  • Clerical Attendance: 26.4
  • Potion Distribution: 64.4

Thus, childbirth could be managed with improved safety at less than the cost of a cure light wounds spell, or less than a year’s wages for a lower-class job in this world; childbirth could be rendered completely safe for less than the cost of 2 such spells. The total income for this community in one year is at least 72,000 GPs, so even the more expensive program could be paid for through a tax of no more than 10%.  Under such a tax system the cleric offering the services would be expected to pay at least 400 Gps tax, and this income could be easily diverted into a partial subsidization scheme for the poorest members of the community.

Note also that under the Potion Distribution model all child deaths are averted. Given that this would probably lead to a reduction in parity of, on average, 3 children per woman, this would lead to an increase in productivity of probably 2.25 years per woman, which gives an income of slightly more than the cost of the scheme under a free market model.

Clerical Load

With approximately 10 complications per 75 births, i.e. 10 complications per 2000 population, we expect that there would be 400 complications per 80000 individuals. A single 7th level Cleric could cover these 400 complications, so we expect not to need more than 1 such cleric per 80000. Under the Potion Distribution model, we need only 2.55 complications in the high risk group per 2000, or 200 complications per 157000 individuals. So we would need a single 7th level cleric per 157000 individuals, making one Remove Disease per day. This cleric would lose 6000 xp per year, so would need to adventure for the remainder of the year; or, for a more reasonable human resources regime, we could allow 1 7th level cleric per 80000 individuals on the assumption that one was adventuring at any time. To allow for death during adventuring, we should assume one cleric per 55,000 individuals. In a population the size of Afghanistan, we would require 545 clerics of this level or higher.

Reduction in Service Load

Given that reduced infant mortality leads to reduced birth rates and lower levels of parity, we would expect a rapid reduction in the number of births per year, and a concomitant reduction in costs and clerical load. Over the long term, we should expect the total cost under both schemes to drop rapidly.

Conclusion

Both schemes proposed here are highly cost-effective, being less than the income gained from the lives saved over their entire life course. Divine intervention to reduce infant mortality is an extremely effective public health intervention that simultaneously reduces personal suffering, death rates, and poverty and has significant demographic and economic effects. It can be paid for easily through a low rate of taxation and the cost reduces over time. In every sense, it is a model public health intervention. Policy-makers, hereditary kings and infernal dictators are advised to adopt this policy as soon as practicable in order to guarantee that they are Universally Loved. Churches of healing that are not already doing this gratis should hang their heads in shame. Paladins everywhere should hang their heads in shame anyway. Fantasy authors should ask themselves if they considered this cost-effectiveness analysis before they wrote their bubblegum-world stories, or if they were just being lazy.

It’s worth considering the extent of poverty in the lower classes of these worlds. A Cure Light Wounds spell costs 50gps, but the average maid earns 36 gps. Being a maid in the medieval era is not exactly the lowest class of job one can expect; it’s not tanning, bone-picking or any of the other taboo jobs, and many women aspire to this sort of work. What Paladin isn’t shagging his maid[1]? What maid isn’t lovin’ it[2]? The WHO defines “catastrophic health expense” as any health care event that costs more than 40% of your annual household consumption. Assuming no savings, cure light wounds costs a maid 140% of her annual consumption. Compare and contrast: in modern Japan a trip to hospital for a broken arm will cost me a maximum of 210,000 Yen without insurance, and if I work full time at Lawson (a pretty low-paid job) I earn 1400000 yen a year. Lower-class people in D&D are poor. But through clerics working together in an organized system, they can eliminate one of the most tragic and significant health problems facing our “advanced” world, at less than (or at least, little more than) the cost of a year’s wages for a reasonably low-class member of society. Even when these clerics are extremely rare, they can still do it. This should serve as a strong hint at the fact that even with scarce sources of magic, medieval fantasy settings should become rich very fast.

fn1: What do you mean “none of them”? Are you suggesting paladins are gay? That’s … blasphemy!

fn2: I’ve read George RR Martin[3], you can’t fool me

fn3: Actually I haven’t, but I’ve watched the TV series