Rob van Hest Capture-recapture Methods in Surveillance - RePub ...
Rob van Hest Capture-recapture Methods in Surveillance - RePub ...
Rob van Hest Capture-recapture Methods in Surveillance - RePub ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Underreport<strong>in</strong>g of malaria <strong>in</strong> the Netherlands<br />
areas, such as West-Africa. 22 Participation <strong>in</strong> peace-keep<strong>in</strong>g operations or elections, 23,24<br />
the number and nationalities of immigrants and asylum-seekers24-26 or the extent of<br />
certa<strong>in</strong> marg<strong>in</strong>alised groups25,27 can also alter the <strong>in</strong>cidence of imported malaria cases over<br />
a certa<strong>in</strong> period of time, as well as <strong>in</strong>fluence the proportion of the different malaria<br />
parasites. In the Netherlands an <strong>in</strong>crease could be observed <strong>in</strong> the proportion of malaria<br />
caused by P. falciparum. Around 1980 falciparum malaria was responsible for<br />
approximately 40% of all notified malaria cases but 10 years later this had <strong>in</strong>creased to<br />
almost 70%. 21 In the 1990s the proportion of falciparum malaria stabilised around 60%.<br />
Estimates of underreport<strong>in</strong>g are frequently derived from different sett<strong>in</strong>gs. They<br />
can be based upon surveys performed at the national level28 or among small groups. 25,29<br />
The background of the data that are compared with the official notification register may<br />
be different, and can vary between hospital admission data, 3 laboratory-based<br />
<strong>in</strong>formation, 25,28 physicians consult<strong>in</strong>g a Reference Laboratory29 or travellers. 30 The<br />
different registers were sometimes matched at the <strong>in</strong>dividual level, 25 at times <strong>in</strong> a stratified<br />
manner3 or <strong>in</strong> another way. 28 In this study we used a well-described and replicable<br />
method and estimated the completeness of notification of three different malaria registers<br />
through capture-<strong>recapture</strong> analysis.<br />
For the three-sample capture-<strong>recapture</strong> technique to be valid, four assumptions<br />
must hold. 17-20 First, overlap between registers must be established without erroneously<br />
misclassify<strong>in</strong>g people as observed <strong>in</strong> only one, two or all three registers. This can be<br />
achieved when cases can be uniquely identified. We used <strong>in</strong>dividual identifiers for each of<br />
the patients and only two patients could not be identified beyond doubt due to (partially)<br />
miss<strong>in</strong>g identifiers. It is important that only true cases are counted. Ideally both the<br />
positive predictive value and the negative predictive value of the registrations should be<br />
100%. None of our registrations will meet this condition, although <strong>in</strong> the case of malaria<br />
specifically, we assume that the positive predictive value will be high. The large overlap of<br />
the hospital records and the notification data with one or two of the other registrations<br />
also supports this view. When the positive predictive value of registrations is low capture<strong>recapture</strong><br />
analysis will result <strong>in</strong> overestimat<strong>in</strong>g the number of cases.<br />
Second, the registers should preferably, but not necessarily, be ‘<strong>in</strong>dependent’,<br />
mean<strong>in</strong>g that the probability of be<strong>in</strong>g recorded <strong>in</strong> one register is not affected by be<strong>in</strong>g (or<br />
not be<strong>in</strong>g) registered <strong>in</strong> another. Such dependence can result from co-operation between<br />
the agencies that keep the different registrations, exchange of <strong>in</strong>formation or a more or<br />
less predictable flow of patients along various <strong>in</strong>stitutions due to referral. In two-sample<br />
capture-<strong>recapture</strong> methods this assumption is crucial and dependencies can cause underor<br />
overestimation. In the three-sample capture-<strong>recapture</strong> approach pair-wise<br />
dependencies between registers can be handled analytically. In the log-l<strong>in</strong>ear model they<br />
can be identified as <strong>in</strong>teractions <strong>in</strong> the model. S<strong>in</strong>ce we could not rule out pair-wise<br />
dependencies, we decided not to rely on the two-sample capture-<strong>recapture</strong> analysis but<br />
<strong>in</strong>stead to use Fienberg’s method.<br />
Third, the population should be “homogeneous” mean<strong>in</strong>g that the population<br />
under consideration should not be composed of segments that have markedly different<br />
capture-<strong>recapture</strong> probabilities. One way of handl<strong>in</strong>g the homogeneity assumption is to<br />
59