Estimating cerebral palsy incidence from prevalence in Tanna Island, Vanuatu: Prevention is more important than rehabilitation

Epidemiological data from low-resource settings are hard-won, making it imperative to extract their full value. In this study, Burgess et al. established the first known prevalence of cerebral palsy (CP) on Tanna Island, Vanuatu.1 However, prevalence is not synonymous with incidence, and in resource-limited settings, this distinction is critical. Prevalence often underrepresents the true scope of a condition because the most severe cases may not survive to be counted. In regions where prospective incidence studies are impractical, prevalence data can still yield plausible incidence estimates if we account for the children who did not survive to be assessed. By combining the data from Burgess et al. with external survival probabilities, we estimated these ‘missing’ cases, demonstrating that the birth incidence of CP in Tanna is roughly twice that of high-income countries (HICs). To translate age-specific prevalence into birth incidence, we treated each reported prevalence age group as a set of survivors from distinct birth cohorts. Since CP mortality is highly dependent on motor severity, we first partitioned the observed cases within each age stratum into two bins—Gross Motor Function Classification System (GMFCS) levels I to II and III to V—based on the marginal distributions reported by Burgess et al. We then approximated the original case counts of these birth cohorts. We divided the observed case counts in each bin by their respective GMFCS-specific survival probabilities, calculated to the midpoint of each age stratum (Table 1). For these survival estimates, we utilized data from Jahan et al.2 and extrapolated 3-year mortality rates into a full hazard schedule, following the methodology described by Day et al.3 These hazards were then applied within a period life table to obtain survival probabilities at the stratum midpoints for each GMFCS group. The incidence at birth estimate—the first for this population—is 3.7 per 1000. This is more than twice that of HICs.4 If CP survival in Tanna is lower than in the Bangladesh proxy data—as is likely given the remote island context—our numerator adjustment for CP-related deaths is too modest. Combining this with the identified concerns about ascertainment raised by Burgess et al. means the true birth incidence is almost certainly higher than 3.7. Burgess et al. concluded from their prevalence data that the burden of CP in Tanna is ‘alarmingly high’ and that prevention must be prioritized. The incidence estimate we derive here shows the true burden is larger than prevalence alone reveals, and the gap between need and capacity is correspondingly wider. Given the magnitude of this problem, public health efforts in Vanuatu must pivot from treatment-focused to prevention-focused. In a setting where villages are accessible only by boat and roads wash out seasonally, expanding rehabilitation capacity is not a realistic solution. Prevention is the only viable path. Not required.