Reducing measles risk in Turkey through social integration of Syrian refugees

Turkey hosts almost 3.5M refugees and has to face a humanitarian emergency of unprecedented levels. We use mobile phone data to map the mobility patterns of both Turkish and Syrian refugees, and use these patterns to build data-driven computational models for quantifying the risk of epidemics spreading for measles – a disease having a satisfactory immunization coverage in Turkey but not in Syria, due to the recent civil war – while accounting for hypothetical policies to integrate the refugees with the Turkish population. Our results provide quantitative evidence that policies to enhance social integration between refugees and the hosting population would reduce the transmission potential of measles by almost 50%, preventing the onset of widespread large epidemics in the country. Our results suggest that social segregation does not hamper but rather boosts potential outbreaks of measles to a greater extent in Syrian refugees but also in Turkish citizens, although to a lesser extent. This is due to the fact that the high immunization coverage of Turkish citizens can shield Syrian refugees from getting exposed to the infection and this in turn reduces potential sources of infection and spillover of cases among Turkish citizens as well, in a virtuous cycle reminiscent of herd immunity.

💡 Research Summary

Turkey currently hosts more than 3.5 million Syrian refugees, a population that arrived from a country where routine immunisation programmes have collapsed during the civil war. The authors set out to quantify how the risk of a measles resurgence in Turkey depends on the degree of social mixing between refugees and the host population. Using anonymised Call Detail Records (CDRs) from a national mobile‑phone operator, they reconstructed a multilayer mobility network at the prefecture (district) level: one layer represents Turkish citizens, the other Syrian refugees, and the strength of inter‑layer links is controlled by a “social‑integration parameter” α (α = 0 % = complete segregation, α = 100 % = full integration).

Immunity levels were estimated separately. For Turkish citizens, vaccination coverage for the measles‑containing MMR vaccine among birth cohorts 2006‑2016 exceeds 96 %, implying that only about 3.8 % are susceptible. For refugees, the authors inferred the age‑specific susceptibility from the 2017 Syrian measles outbreak (effective reproduction number Re ≈ 1.32) and from the age distribution of refugees in Turkey, arriving at a susceptible fraction of roughly 9.9 % (95 % CI ≈ 8‑12 %). Thus, while the host population is near the 95 % herd‑immunity threshold, the refugee cohort sits just below it.

A spatially explicit SEIR model was then coupled to the mobility matrices and the immunity profiles. The basic reproduction number R₀ for measles was explored in the realistic range 12‑18. For each combination of R₀, refugee susceptibility, and α, the effective reproduction number Re (the average secondary cases in a partially immune population) was computed, and stochastic simulations were run to estimate outbreak probability, final size, and geographic spread.

Key findings:

1. Integration dramatically lowers transmission potential. When α increases from 0 % to 70 %–80 %, Re falls by almost 50 %, often dropping below the epidemic threshold (Re < 1). For example, with R₀ = 15 and 9.8 % refugee susceptibility, the probability of an outbreak larger than 20 cases falls from 100 % (α = 0 %) to below 10 % (α ≈ 70 %).
2. Segregation amplifies risk for both groups. In the fully segregated scenario, large outbreaks are confined to the refugee community but can still generate thousands of cases, and spill‑over to Turkish citizens becomes non‑negligible when even a modest fraction of refugee contacts involve hosts.
3. Spatial diffusion is curtailed by mixing. Under high segregation, simulated epidemics can reach more than 300 of Turkey’s 1 021 prefectures; with ≥70 % integration, most scenarios predict spread limited to fewer than ten prefectures, typically near the Syrian border rather than the major urban centres.
4. Policy implications. Complete homogeneous mixing is unrealistic, but measures that raise α—such as dispersing refugee housing across neighborhoods, encouraging joint use of schools, health centres, and public spaces, and fostering community‑building activities—can harness the high immunity of Turkish citizens as a “buffer” for the less‑immune refugee cohort, creating a virtuous herd‑immunity effect.

The study therefore provides robust, data‑driven evidence that social integration is not merely a humanitarian goal but a concrete public‑health strategy. By reducing segregation, Turkey can halve the measles transmission potential, prevent large‑scale outbreaks, and protect both refugees and the host population.