Estimating rheumatic fever incidence in New Zealand using multiple data sources

Abstract
SUMMARY: Rheumatic fever (RF) is an important public health problem in New Zealand (NZ). There are three sources of RF surveillance data, all with major limitations that prevent NZ generating accurate epidemiological information. We aimed to estimate the likely RF incidence using multiple surveillance data sources. National RF hospitalization and notification data were obtained, covering the periods 1988–2011 and 1997–2011, respectively. Data were also obtained from four regional registers: Wellington, Waikato, Hawke's Bay and Rotorua. Coded patient identifiers were used to calculate the proportion of individuals who could be matched between datasets. Capture–recapture analyses were used to calculate the likely number of true RF cases for the period 1997–2011. A range of scenarios were used to correct for likely dataset incompleteness. The estimated sensitivity of each data source was calculated. Patients who were male, Māori or Pacific, aged 5–15 years and met the Jones criteria, were most likely to be matched between national datasets. All registers appeared incomplete. An average of 113 new initial cases occurred annually. Sensitivity was estimated at 80% for the hospitalization dataset and 60% for the notification dataset. There is a clear need to develop a high-quality RF surveillance system, such as a national register. Such a system could link important data sources to provide effective, comprehensive national surveillance to support both strategy-focused and control-focused activities, helping reduce the incidence and impact of this disease. It is important to remind clinicians that RF cases do occur outside the well-characterized high-risk group.