Appendix

The developmental origins hypothesis or Barker Theory has received recent attention in the media and proposes that extended periods of nutritional restriction may be associated with changes in the body that result in poorer health outcomes later in life.

According to this theory, foetuses adapt themselves rather beautifully to hostile environments and this enables their survival. However, the very same changes could also lead to problems later in life such as coronary heart disease, hypertension and type 2 diabetes – diseases common amongst Muslims.

Research in this area includes determining health outcomes late in life and relating them to foetal occurrences. Almond and Mazumder (2011) were the firs to investigate potential links between Ramadhaan and poorer health outcomes. They go as far as purporting that fasting in Ramadhaan in very early in pregnancy results in fewer male births and higher instances of disabilities in Ugandan and Iraqi populations. Rather scary claims.

We have found that, based on the data presented, no claims can be that the supposed correlations are solely attributable to fasting. Various complex environmental factors have been shown to be key, none of which are accounted for and that we detail below.

The authors contend that they cannot distinguish those who do fast and those who did not during Ramadhaan (a pretty essential factor) nor do they claim that all the women included amongst those who fasted Ramadhaan are likely to be Muslim(!). Additionally, they apply a blanket assumption that all Muslims fast whilst pregnant.

This is often not true, and perhaps more so for their Ugandan sample which will comprise of a significant number of Indian immigrant Ismailis whose numbers currently total around 15,000 in Uganda. Ismailis believe the only compulsory fast is when a new moon occurs on Friday – about twice per year, and they do not see fasting in Ramadhaan as obligatory.

If we begin with the findings of reduced rates of male births, there is considerable country-to-country variation across the globe. Confounding factors include the age of the mother, with very young or older mothers lowering and raising the rate of male births respectively. Particular birth years and different races can produce noticeably varying sex ratios.

The number of male babies born in the U.S. is dropping and has been on a steady decline for the last three decades. The decline in male births is equivalent to 135,000 fewer white males in the last 30 years in the U.S. and 127,000 fewer males in Japan[1] from 1970 to 2002, suggesting that environmental factors are one explanation for these trends.

A similar shift in the sex ratio at birth has been found in numerous other countries: many Latin American populations, Finland, Norway, Wales and the Netherlands, several Arctic, cities in Italy and among fish-eating women in the Great Lakes region.

The highest sex ratio decline in the world is in the Aamjiwnaang First Nations in Canada. This small area has only 46 males born to every 132 females. The reservation is surrounded on three sides by petrochemical, polymer, and chemical industrial plants, it has a creek running through it, plagued with high levels of mercury and PCBs and the highest toxin levels in the air in all of the Canada.

Highly interestingly, this nature reserve is on the border with Michigan and is only about an hour away from where most Muslims in Michigan live; Detroit. The two communities are linked by water: the Aamjiwnaang is on the shores of the St. Clair River which runs into Lake St. Clair of the Great Lakes (where fish eating women have also been found to have reduced male births) and becomes the Detroit River. The reduction of males in this area has evidence of effects that are wholly separate to Ramadhaan fasting.

In addition, stress effects other than from fasting have also been shown to impact the number of males born. For example, throughout the entire U.S., the chances of miscarrying a male foetus increased after the September 11, 2001, attacks compared to the months before and after[2]. Such national sources of stress and trauma most certainly affect Muslims in Uganda and in Iraq.

Iraq’s history is littered with trauma. Participants in the Almond and Mazumder (2011) study were born in the years 1958-1977. Sadly, this entire period has events that may well impact results.

1958 was the year of the 14 July Revolution that ended the Hashemite dynasty, establishing the Republic of Iraq. When `Abd al-Karīm Qāsim seized power, unrest and opposition continued until he was overthrown and shot during the 1963 Ramadan Revolution. Around 5,000 Iraqis were killed in the fighting from February 8–10, 1963.

There was also the presence of British forces in support of Kuwaiti independence from Iraq up until 1971. The First Kurdish–Iraqi War lasted from 1961 until 1970, resulting in the deaths of  some 105,000 people. The Second Kurdish Iraqi War was only two years later and lasted until 1975.

Additionally, there was the 1971 Iraq poison grain disaster, the largest mercury poisoning disaster when it occurred, with worst cases resulting in blindness or death caused by central nervous system failure. 6,530 patients were admitted to hospital with poisoning, and 459 deaths were reported.

These huge events must have played a shaping role in the lives of all involved and to ignore the effects of trauma in utero, during childhood and adulthood, as a result of such stresses and to attribute disability rates solely to fasting when pregnant is quite a leap.

Muslims in Uganda, along with other minorities, have long been marginalised and identified as needing “special protective and/or corrective measures to be able to attain a state of ‘normalcy’ in society.”[3]

Most notable in Ugandan history is Amin’s reign of terror where 300,000[4] Ugandans were murdered and 90,00 Indians and Pakistanis were given 90 days to leave Uganda, resulting in chaos on Uganda’s economy. Even after Amin’s overthrow in 1979, Muslims were heavily discriminated against and hundreds killed.

Uganda is one of the world’s poorest countries and Ugandan Muslims had been denied access to all public sectors until 1962, forcing them to open schools of their own as a reaction to overcome discrimination. It is only since the establishment of such schools that literacy rates have risen sharply.

The presence of war with neighbouring countries, internal conflict with rapid and violent changes of government and a lengthy civil war since independence from Britain 1962 have also been factors of instability.

Muslims in Uganda and Iraq are undoubtedly subject to the stress effects of national grief and trauma. These factors are probably a whole host of other socio-economic factors are far more likely to play a part in the health of pregnant women and their children. We have examples of such effects in other countries, for example, Arab women in Israel have worse birth outcomes than Jewish ones.[5] The same could be said of Ugandan Muslim women compared to non-Muslim women.

A causal link between fasting and adverse effects at any stage in life cannot be based upon the Ugandan and Iraqi samples used by Almond and Muzumder (2011) nor their findings. In terms of the Michigan sample, aside from suspected environmental factors, their theory doesn’t fit reality.

If we take Arab Americans over the age of 25 in Michigan (the same sample) we find that that they have lower mortality rates from cancer and chronic lower respiratory disease compared to both whites and blacks. Female Arab Americans had lower mortality rates from heart disease, cancer, stroke and diabetes than whites and blacks.

Mortality rates do not equate incident rates in a population but they are valuable indicators of the prevalence of the related causes of death. Mortality rates are compelling indicators of health whereas reliance on consensus data for disability with a lack of clear parameters for determining disability when applied cross-culturally is open to challenge.

In fact, the data on disability from Iraq and Uganda cannot be directly compared to such difference in measurement and the collected data itself is questionable as only 1.5% of Iraqis and about 5% of Ugandans reported disability compared to 12%  percent in the US suggesting a significant missing proportion. Were the missing data incorporated the picture could be entirely different.

A similar study, by van Ewijk, R. (2011) claims a confirmation of Almond and Muzumder’s (2011) findings[6] based on data collected in Indonesia by the RAND project. Indirect measures of health and serious disease were used, such as chest pain upon movement and time taken for wounds and cuts to heal. These are indicators of serious conditions but are not sufficient as indices in their own.

Furthermore, some of his results appear puzzling. van Ewijk, R. (2011) found the effects on adulthood to be related to Ramadan falling about halfway in the pregnancy whereas Almond and Mazumder (2011) report the effects for exposure early in pregnancy. Van Ewijk also reports evidence of effects of hypertension in younger people but not in older participants, turning the long-term damaging factor of the theory on its head.

In fact, the results could even be interpreted to suggest that foetal adjustments during Ramadhaan have positive outcomes in adulthood on some measures. They could equally be stretched to indicate that conception post Ramadhaan actually improves the health of the child born, perhaps due to improved maternal health and well being thanks to a Ramadhaan pre-cursor that prepares the mother’s body for childbearing.

A new study by van Ewijk et al. (2013) in Indonesia[7] suggests that adult Muslims who had been in utero during Ramadan are slightly thinner and shorter, being on average 8mm shorter. I comment no further on their straw clutching. Others are also not convinced with the limitations in their work[8].

This team of economists have one study alone that is of relevance to us. In their assessment of 7 year olds they report to have found Ramadhaan fasting in early pregnancy to have detrimental effects on Key Stage 1 attainment. As with the studies above, the results are unreliable and better explained by existing common sense theories. They fatally fail to have controlled for English not being the first language of these children.

The Department for Education report a gap between pupils whose first language is English[9]. This most certainly a factor in performance of Pakistani and Bangladeshi children as the sample selected are from areas in Britain where Muslim concentrations are very high. This would automatically include a greater proportion of children that do not have English as a first language, and those that maintain their mothers tongues more closely.

Interestingly, they compare Pakistani and Bangladeshi children to Black Caribbean children. Data is available for Black Africans but they exclude them from analyses, perhaps as a large proportion of them will be Muslim (though an number of Black Caribbeans will also be Muslim). Black Caribbeans will not be disadvantaged by not having English at home yet Black Africans may well do.

The disadvantage of not having English as the first language is particularly prominent in Key Stage 1 rather than Key Stage 2, when children are younger and have had less exposure to English and schooling in general. The authors chose to study Key Stage 1.

The study only controls for the known differences in attainment of pupils eligible for free school meals, leaving aside not only the language the child speaks at home but also the identification of Special Education Needs (SEN). Note that the authors did have access to both these sets of data.

Pakistani and Bangladeshi children are over-represented[10] in those with SEN. All children from all backgrounds with no identified SEN outperform pupils with SEN. Socio-economic disadvantage and gender have stronger associations than ethnicity with prevalence of SEN but even after controlling for these Pakistani and Bangladeshi children specifically are over-represented amongst those with SEN.

Note also that Pakistanis and Bangladeshis are most likely to live in the most deprived areas[11] and that significant variations between minorities justify a separate focus on both deprivation and population concentration.

Furthermore, there is very little transparency in how they actually used KS1 scores and particularly the use of PLASC data (Pupil Level Annual School Census data) which were only available for 2 years of the 9 years studied. They do not explain what they did for remaining years or what they did with that data.

Finally, no account is made of classification of mixed race children in the Black Caribbean group where it is most prevalent, having done so for the Asians. When they did exclude mixed race Caribbeans the results were “less precise” but sure enough, they chose not to publish those.

We need to bear in mind that this isn’t a body of evidence. It is one team of economists, not health professionals, churning out a few studies using similar methodology with the same ideology, reporting evidence, by hook or by nook, to drive policies in targeting pregnant women not to fast. It must be balanced, first with the large body of evidence indicating fasting is safe (see appendix ii for the 40+ papers we took) and we await publications that are in progress that challenge Almond and van Ewijk et al.

One such critique has been published. Brown’s (2011) critique[12] pulls apart the theoretical foundation of Almond’s studies. This particular paper concentrates on a one of Almond’s cornerstone pieces on the 1918 U.S. influenza pandemic and they systematically show that without having controlled for socioeconomic factors Almond’s results are seriously flawed. When they are factored in the reported effects disappear altogether or are greatly diminished.

Brown highlighted a need for controlling household fixed effects and this has not been done but in one of their studies. This single study found exposure to Ramadan fasting in utero in Indonesia results in 4.5% fewer hours worked and a 3.2% increase in the probability of being self-employed. They interpret these as negative wealth effects. Not all would agree that they are. They also find lower cognitive test scores and math scores in children aged 7-15.

The authors make reference to “the possibility that behavioral changes related to schooling inputs may be one possible channel through which the tests score effects are taking place, apart from the direct effects on one’s cognitive ability from fasting, it is not clear if the Ramadan effects are indeed driven solely by religious fasting (as medial theory predicts) or by some other factor not directly related to religiosity…the framework suggests that the above predictions may be biased by parental characteristics. When parents make decisions for children, parental characteristics may be important. Those who invest more in unexposed children’s schooling may be those who also encourage their children to be involved in skilled occupations and who encourage hard work, leading to more labor supply.”

They certainly have something there. Suryadarma (2011) has found that the large gap in education between Muslims and non-Muslims in Indonesia is entirely removed when scholastic ability and parental education are controlled for. The college gap remains and is likely to be caused by differences in the labour market for Muslims and non-Muslims.[13]

There are massive differences in earnings post-college between Muslim males and is non-Muslim males: 50.2% vs 128.5%. Non-Muslims also have a higher employment rate after college-level that does not exist for college-level educated Muslims. Thus possibly the best explanation as to why Muslims do not continue nor aspire within the secular education system are that factors that are shaped by education and employment, factors that are entirely aside from pre-natal associations.

Finally, there are basically two large education systems operating in Indonesia; a secular system and an Islamic system madrasah system. In the secular system, religion is one of among many taught subjects whereas Islam is the foundation of the curriculum in madrasahs. The children in the madrasah system are unaccounted for.

The general pattern of criticism in this small collection of papers by Almond and van Ewijk et al are that various glaring influences on Muslims within the selected populations are ignored and inaccuracies in the sampling and measurements leave the work incomplete and of no practical use.

These studies have provided a spring board for further investigation at best and we await the unfolding of a flurry of better quality research in this interesting area. To conclude on long term effects of Ramadan fasting, which overlaps with pregnancy in three of every four births; resulting in about 1 million Muslims alive today having being in utero during Ramadan, we state that it is currently spurious and irresponsible to give rise to the idea that the vast majority of Muslims are susceptible to adverse outcomes due to their mothers having fasted Ramadhaan.

[1] Davis, D. Environmental Health Perspectives, April, 9 2007. News release, University of Pittsburgh Schools of the Health Sciences.

[2] Bruckner et al. (2001), “Male Fetal Loss in the U.S. Following the Terrorist Attacks of September 11, 2001” BMC Public Health,

[3]Baker, (2001), “Uganda: The Marginalization of Minorities” MRG International Report

[4]In 1977 Amnesty International estimated 300,000 dead

[5] Torche and Shwed, (2013) “The Hidden Costs of War” Mellon Biennial Conference at Columbia University

[6] van Ewijk, R. (2011). “Long-term health effects on the next generation of Ramadan fasting during pregnancy.” J Health Econ 30(6): 1246-1260.

[7] van Ewijk R. J G., et al., (2013) Associations of Prenatal Exposure to Ramadan With Small Stature and Thinness in Adulthood. Results From a Large Indonesian Population-Based Study. Am J Epidemiol. 2013;177(8):729-736.

[8]Susser, E. & Ananth, C. V. (2013) Invited Commentary: Is Prenatal Fasting During Ramadan Related to Adult Health Outcomes? A Novel and Important Question for Epidemiology American Journal of Epidemiology

[9] Key Stage 1 Attainment by Pupil Characteristics, in England 2009/10 Department for Education

[10] Lindsay, G. eta l., (2006) Special Educational Needs and Ethnicity: Issues of Over- and 

[11] Dorsett, R. (1998) Ethnic minorities in the inner city, The Policy Press in association with the Foundation ISBN 1 86134 130

[12] Brown, R. (2011) The 1918 U.S. Inuenza Pandemic as a Natural Experiment, Revisited http://dupri.duke.edu/pdfs/ryanbrownpaper.pdf

[13] Suryadarma, D. (2011) What explains the Muslim Disparity in Education Attainment: Explanations from Indonesia