Echo of a Tropical Fever: How Dengue Returns as Paralysis Seven Months Later ✎

Original: "Increasing Trends of Guillain-Barré Syndrome (GBS) and Dengue in Hong Kong"
arXiv:1703.04238 · 2017-03-13 · CC BY-NC-SA 4.0 · independent analysis · 3 min
In Hong Kong, waves of imported dengue and a rare paralysis move in step with climate cycles, and the link between them emerges with a seven-month delay.

Hong Kong sits at a crossroads: tropical diseases arrive with travelers, while the local climate follows global cycles. In this city, from 2000 to 2016, two afflictions grew in different ways: dengue fever surged in sharp waves, while Guillain–Barré syndrome (acute paralysis, often after infections) crept upward slowly but steadily. Doctors have long noticed that individual patients sometimes develop paralysis after a tropical fever. But can this link be seen at the city level?

Researchers collected monthly case data for both diseases from open sources at the local health center, added meteorological indicators from the Hong Kong Observatory, and the global climate index MEI from the U.S. National Oceanic and Atmospheric Administration. MEI reflects El Niño oscillations, the phenomenon that pushes warm water across the Pacific and alters weather worldwide. First, they calculated correlations between diseases and weather, then shifted the series by various lags to find the maximum cross-correlation. For the dengue–MEI link, they built a Poisson regression with lags from 0 to 11 months. Finally, they applied wavelet analysis with the Morlet wavelet to see how the synchrony of fluctuations changed over time rather than being constant.

Wavelet analysis works like a musical score: it shows in which years two melodies played in unison and when they diverged. Ordinary correlation gives a single number for the whole period, but the wavelet draws a map of coincidences over time.

The results resembled an echo in the mountains: the cry of dengue returns as paralysis seven months later. Dengue and MEI are more strongly linked: the index explains more than 12% of dengue variability, and the maximum correlation occurs at a lag of 3–5 months. Between dengue and Guillain–Barré, no direct link is apparent, but with a seven-month shift, the correlation reaches 0.28. Wavelet analysis showed that from 2005 to 2015, dengue and Guillain–Barré fluctuations periodically resonated, especially in years of major dengue outbreaks (2002, 2007, 2010, 2014). Their seasonal peaks are opposite: paralysis is more common in February–March, fever in August–September.

Guillain–Barré has a weak but statistically significant link with weather: colder and less evaporation means more cases (correlation coefficient about −0.28). Dengue, in contrast, follows the global climate rhythm.

Previously, the sequence “dengue, then paralysis” was known only from clinical descriptions of individual patients. Now it is clear that the link does not always appear but flares up in certain periods that coincide with major dengue waves. This changes the perspective: Guillain–Barré can be seen not as random background but as an echo of tropical epidemics reaching the subtropics with a delay. For health services, this means that MEI and imported dengue data could serve as an early warning for neurology departments: expect a rise in paralysis a few months after a climate signal.

Hong Kong, like a sensitive barometer, captures the pulse of distant tropics. The echo that returns after seven months can now be not only heard but also anticipated.

Guillain–Barré syndrome is named after two French neurologists who described it in 1916, while dengue fever has been known in China since ancient times as "water fever" because joint pain feels like bones breaking.

In Greg Bear's novel "Blood Music," an infection acts as a conductor, making the body's cells sing in unison. Here, dengue and Guillain–Barré also resonate, but not directly; rather, through a shared climatic rhythm, like two parts in one score.

E[dengue_{t+τ} | MEI_t] = exp(α + β MEI_t + ε_{t+τ})
The logarithm of the expected number of dengue cases depends linearly on the MEI index with a time shift, allowing estimation of how well the climate index predicts a fever surge.
ψ(t) = π^{-1/4} exp[-i(2π f_0)t] exp(-t^2/2)
A mother wavelet that extracts oscillations of different durations in a signal while preserving time information, unlike the standard Fourier transform.
W_{ψ,x}(a, κ) = ∫ x(t) ψ^*_{a,κ}(t) dt
Convolution of the data with a stretched and shifted wavelet, producing a map of oscillation power across periods and time.
Concepts
Multivariate ENSO Index parkinsons disease epilepsy