Three intermediately anchored estimates are presented. The first one starts with the inputs used for the GBD96 based estimate, but with the cause-specific mortality proportions replaced by local estimates. The local cause-specific mortality proportions come from two studies. The rural cause of death study provided the cause-specific mortality proportions for the rural population of Andhra Pradesh. The second one starts with inputs for the COD anchored estimate and replaces the GBD96 disability weights with the APHSV-VAS weights. The third estimate uses APHSV99-Torrance-TTO disability weights instead of the APHSV99-VAS weights. Call this the HSV-Torrance-TTO estimate. So in all we have five estimates mentioned in below table. The epidemiological parameters like incidence, duration and age at onset, remain same as in GBD96. The locally anchored estimates were computed separately for rural and urban areas of the state whereas the GBD96 estimate has been computed for the whole state population. The WDR93 and GBD96 are the two minimally anchored estimates. The three intermediately anchored estimates comprise of the COD anchored, and the two HSV anchored estimates.

The GBD96 estimate of DALY loss per 1000 persons are about 20% lower than the WDR93 estimates. In comparison, the difference between GBD96 and COD anchored estimate is marginal, in the range of 0.6 to 2.5%. For YLL and YLD, we do not readily have the rates from WDR93. Comparing GBD96 with COD anchored estimate, we see that YLD rates remain the same. This is to be expected, since both estimates share the same epidemiological inputs and disability weights. The change in YLL rates is between 0 to 3.7%.

   Top 

Changes in YLL : YLD ratio and Male to Female ratio are minimal between GBD96 and COD anchored estimates. The situation changes completely, however, in case of the HSV-VAS anchored estimates. Recall that the disability weights obtained from the community study in Andhra Pradesh were all higher than the disability weights used in GBD96 study. All projected disability weights based on data from the community survey, are in the range of 0.28 to 0.803 with a median value of 0.474. Compare this with the GBD96 range of disability weights, which range from 0 to 0.85 with a median value of 0.145. As a result the magnitude of HSV anchored disease burden is inflated to nearly twice the COD anchored estimate as well as the GBD96 estimate. The YLL to YLD ratio changes to reflect the higher disability component. Burden due to premature mortality now appears as half of the burden due to disability. The HSV-Torrance-TTO estimates give magnitude of burden intermediate between the COD anchored and HSV-VAS anchored estimates. The reader may recall that the life expectancy at birth in Andhra Pradesh is about 60 years . This is much lower than life expectancy achieved elsewhere in the world. The standard life expectancy used by us is 80 years for males and 82.5 years for females. This standard is based on actual experience in some parts of the world. Considering the intermediate mortality level in Andhra Pradesh, the radical shift in relationship of mortality and disability components suggested by the HSV anchored estimates is puzzling. But before we affirm up our views on the HSV estimates, let's continue with the comparison of different estimates, looked at from additional points of view.

   Top 

Age distribution of DALYs from different estimates has the same pattern here. Changes between WDR93 and GBD96 are substantial. WDR93 estimates put more than 50% of disease burden in the age group 0-4 years and another 20% in the elderly above 60 years. The burden on adolescents and adults (5 to 44 years) was estimated at 15%. This pattern is maintained if we look at the distributions separately for females and males. The GBD96 estimates put the burden on 0-4 year olds at the reduced level of 35%. The burden on adolescents and adults turned out to be higher at 40%. The two HSV anchored estimates, however, put the burden on adolescents and adults at 50%, a situation just opposite the age distribution obtained by the WDR93 estimates.

   Top 

YLL to YLD ratio by age-sex group is mentioned in the above table According to WDR93 estimates, premature mortality is the dominant burden over all age groups. The YLL to YLD ratio range from 1.6 to 5.6. According to the GBD96 and COD anchored estimates, premature mortality is the dominant source of disease burden for all age groups, except in the age group 5-14 years and adult females in age group 15 to 44. In these later age groups, the YLL to YLD ratio range is from 0.5 to 0.6. Premature mortality is the major contributor to disease burden among infants and children in age group 0 to 4 years and the elderly at ages 60+ years. This is consistent with our understanding of the pattern of general mortality, concentrated among the infants and then the elderly. According to the two HSV anchored estimates, disability becomes the dominant source of disease burden at all ages, except the elderly group at 60+ years. Here also, the YLL to YLD ratio is quite low between 1 to 2. Here again we find that the GBD96 and COD anchored estimates are intuitively appealing and are consistent with our knowledge of general mortality pattern. The two HSV anchored estimate produces unacceptably low YLL to YLD ratios. If these estimates were to inform policy, then premature mortality reduction would get a back seat altogether.