WHO noted that narcolepsy has never before been associated with vaccines.
When there is no prior association, odd diseases occuring in temporal relationship to vaccination are generally felt to be coincidental. Therefore, they are not generally reported to voluntary reporting systems, like the US' VAERS. No data get collected, and it remains unknown whether vaccination increases the incidence of most diseases, particularly rare diseases.
Only when large numbers are vaccinated, the disease is severe, and it becomes obvious within a few weeks, is the link between vaccination and the disease made. Furthermore, the relative risk has had to be high for the link to be identified.
Here are some examples: swine flu vaccine in 1976 caused about 8 times as many Guillain Barre cases within 6-8 weeks as would have been expected (US). Rotavirus vaccine in 1999 caused 22 times the expected incidence of intussusception as expected (US). Swine flu vaccine in 2009-10 caused 9 times as many narcolepsy cases as expected in children (Finland).
We urgently need to establish surveillance systems that analyze diagnoses made at outpatient visits and hospitalizations for at least one year post vaccinations, so that statistically significant links can be identified and assessed for causality. A good system would have the potential to detect diseases that are less severe than those above; it could detect those associated with smaller relative risks; and could detect those with later onset post-vaccination than currently identified (using limited datasets from voluntary reporting).
Such surveillance systems are possible now using data from large HMOs/ group health plans, which CDC has paid to access. But simply looking for correlations between vaccinations and the majority of diseases has never been done... or at least, has never been reported. It would be relatively cheap and easy to generate such correlations.