Validation of administrative data to estimate vaccine impact: Audit of the Fiji hospital admissions electronic database, 2007–2011 & 2014–2015

Effect of ten-valent pneumococcal conjugate vaccine introduction on pneumonia hospital admissions in Fiji: a time-series analysis

“…Regarding our clinical pneumonia outcomes associated with introduction of PCV10, we found a 21% (95% CI 5-35) reduction in severe or very severe pneumonia, 46% (33-56) reduction in hypoxic pneumonia, and 25% (9-38) reduction in radiological pneumonia among children aged 2-23 months, relative to control diseases and using pneumonia data imputed to correct for missing data for case definitions ( figure, table 2, appendix pp [15][16]. Credible intervals for these estimates were wide due to the small numbers of cases (monthly case numbers are shown on appendix p 8) and unexplained variability in the data.…”

“…Firstly, we audited the electronic national hospital admission database between 2007-11 and 2014-15 and found that 89% of all-cause admissions were captured compared with the ward admission registers. 15 We assumed that these missing admissions were not specific to a pneumonia admission but instead related to logistical factors pertaining to any hospital admission. For the clinical pneumonia outcomes, there were missing data for severe or very severe, hypoxic, and radiological pneumonia outcomes due to the patient medical records, chest radiograph, or oxygen saturation data being unavailable for review.…”

“…We quantified the degree of under-reporting of pneumonia cases through an audit of the national hospital admission database and found that 89% of admissions at the tertiary hospitals were captured when compared with the hospital ward registers. 15 Second, the decrease in childhood pneumonia hospital admissions might also be due to other factors unrelated to PCV10, including changes in access to care, admission practice, population changes, and other health interventions. This limitation is inherent to all observational studies and therefore causation cannot be implied generally.…”

Epidemiology of intussusception before and after rotavirus vaccine introduction in Fiji

“…Hospitalisation data are entered into the national hospital admission database by medical coders, using the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, Australian Modification (ICD-10-AM), and stored at the MoHMS Health Information, Research and Analysis Unit (HIRAU). An audit of the national hospital admission database showed it captured 83% of admissions 34 . National population denominators and growth rates were sourced from Fiji Government Bureau of Statistics, 2007 National Census.…”

The impact of the rotavirus vaccine on diarrhoea, five years following national introduction in Fiji

“…Coders assign an ICD-10-AM code for each discharge diagnosis, which are entered, with the other clinical information into the computerised national hospital admission database (PATIS) which generates a patient's unique National Health Number (NHN). The PATIS system has been available at all three tertiary hospitals from 2007 and has been demonstrated in a previous study to be 89% complete [18] . This is an observational study describing all-cause diarrhoea admissions nationwide in all ages at all three public tertiary hospitals (CWMH, Lautoka, Labasa) and rotavirus diarrhoea in < 5-year olds at two hospital-based sentinel sites: the largest tertiary hospital for admissions CWMH, and one secondary public hospital for admissions and outpatients, Savusavu Hospital.…”