After publication of ‘Grootendorst P, Piérard E and Shim M. Life expectancy gains from pharmaceutical drugs: a critical appraisal of the literature. Expert Rev. Pharmacoeconomics Outcomes Res. 9(4), 353–364 (2009)’, hereafter GPS, we became aware of several errors.
In Box 1 on page 362 (‘Estimated reduced form model of life expectancy at birth based on annual US data 1960–2004’), the line ‘Lagged number of new molecular entities approved’ is incorrect. The line should read: ‘Log number of new molecular entities approved.’ The parameter estimates reported in Box 1 are correct.
In Box 2 on page 362 (‘Estimated change in real per capita spending required to increase life expectancy at birth by 1 year, in long-term equilibrium, by type of intervention’), a programming error rendered the estimates incorrect. The corrected estimates appear below. We now also provide estimates of the 95% confidence interval (obtained using the delta method) around our point estimates.
These estimates are discussed in GPS on page 360 and again on 361. On page 360, we compared our estimates to those obtained by Lichtenberg (2004). This comparison is invalid for two reasons. First, the reported estimates are incorrect. Second, Lichtenberg (2004) and GPS measure the longevity gains associated with new drug introductions in different ways, so that the estimates obtained from the respective studies are not directly comparable. Lichtenberg reports the effect of NMEs on the absolute number of life-years gained in the US population, whereas GPS reports the effect of NMEs on life expectancy at birth. Our criticism of Lichtenberg’s reporting of the effect of NMEs on life-years gained on page 358 (near the middle of the second column) of our paper is also invalid as we misinterpreted his methods.
These errors, however, do not change the conclusion reached on page 361; based on our respecification of Lichtenberg’s regression model, it appears that it is more economical to increase life expectancy at birth by spending on public health than it is to spend on drug development.
The editors of Expert Review of Pharmacoeconomics and Outcomes Research and the authors would like to sincerely apologize for any inconvenience or confusion this may have caused the readers.
Box 2. Estimated change in real per capita spending required to increase life expectancy at birth by 1 year, in long-term equilibrium, by type of intervention.
• Develop NME:
US$30.66; 95% confidence interval: $4.40–23.20
• Expenditure on public health:
$5.56; 95% confidence interval: $2.76–8.36
• Expenditures on physicians:
-$80.76; 95% confidence interval: -$194.63–33.10
• Expenditures on hospitals
-$303.86; 95% confidence interval: -$779.98–172.27
Notes
NME: New molecular entity.
Data from Grootendorst P, Piérard E and Shim M.