In the absence of cancer registry data, is it sensible to assess incidence using hospital separation records?

Table 5 Sensitivity, specificity, positive predictive value, negative predictive value and agreement (kappa) for ten-year aggregate period 1990–1999.

	All Cancers	Lung	Colorectal	Female Breast	Prostate
Sensitivity	78.58%	77.56%	84.22%	81.60%	73.00%
lower 95% confidence limit (CL)	78.12%	76.27%	83.13%	80.48%	71.85%
upper 95% CL	79.04%	78.78%	85.34%	82.66%	74.12%
Specificity	99.02%	99.88%	99.92%	99.86%	99.66%
lower 95% CL	98.99%	99.87%	99.91%	99.84%	99.64%
upper 95% CL	99.05%	99.89%	99.93%	99.87%	99.68%
Positive predictive value	80.53%	81.82%	87.07%	89.54%	79.35%
lower 95% CL	80.10%	80.60%	85.96%	88.61%	78.26%
upper 95% CL	80.95%	82.97%	88.09%	90.44%	80.51%
Negative predictive value	98.90%	99.85%	99.90%	99.72%	99.52%
lower 95% CL	98.87%	99.84%	99.89%	99.71%	99.50%
upper 95% CL	98.93%	99.86%	99.91%	99.74%	99.54%
Kappa (total)	0.79	0.80	0.86	0.85	0.76
lower 95% CL	0.78	0.79	0.85	0.84	0.75
upper 95% CL	0.79	0.80	0.86	0.86	0.77
Kappa (syear by dyear)	0.91	0.93	0.96	0.95	0.91
lower 95% CL	0.91	0.92	0.95	0.95	0.91
upper 95% CL	0.92	0.94	0.97	0.96	0.92

Notes.
1. Cancer registry and hospital cancer separations ten-year aggregate files merged by study identification number, selecting for cancer type and then for first occurrence of study identification number.
2. For sensitivity, specificity, positive predictive value and negative predictive value, the cancer registry is designated the gold standard.
3. Kappa (total) measures agreement on study identification numbers between registry and separation records.
4. Kappa (syear by dyear) assesses agreement between separation year and diagnosis year for the subset of shared study identification numbers across the ten-year period.
5. Reference population used is the British Columbia midyear population size for both sexes, or single sex with respect to female breast and prostate cancer. See text for discussion of choice of reference population for ten-year period.
6. Source for VIHA population figures from 1990 through 1999: BC Stats 2004: http://www.bcstats.gov.bc.ca/data/pop/pop/estspop.htm.
7. Not shown: measures calculated by individual year. These are available on request.

ISSN: 1475-9276