Secondary Logo

Journal Logo

Letters

Re: Estimating the Population-level Effectiveness of Vaccination Programs in the Netherlands

Weisenborn, Toon; Nederveen, Aart J.; Stalpers, Lukas J.A.

Author Information
doi: 10.1097/EDE.0000000000001158
  • Free

To the Editor:

In 2018, van Wijhe et al1 published in this journal an analysis of historical mortality data from the Netherlands for four infectious diseases included in the Dutch State Vaccination Program that started in 1953. They found that the ratio between observed and expected mortality reached near 0% within 10 birth cohorts for each vaccine-preventable disease, implying a near-100% effectiveness of the vaccination program. We wish to describe an alternative approach to these data, resulting in a different conclusion.

Van Wijhe et al estimated the effectiveness of the vaccination program by comparing the observed disease-specific childhood mortality between 1953 and 1975 with an expected mortality without vaccination. They estimated the expected mortality by fitting mortality data of the pre-vaccination era to an exponentially declining function. Their key assumption is that each disease-specific mortality rate is a time-independent fraction of the all-cause childhood mortality. The publicly available data from CBS-Statistics Netherlands show that all-cause childhood mortality declined roughly exponentially from 1900 until 1950, excluding the periods of the two World Wars.2 A similar pattern of exponential decline in all-cause mortality was found in the G7 countries.3 However, as earlier found by Wolleswinkel-van den Bosch et al,4 the Dutch mortality data also show that the mortality due to all infectious and parasitic diseases together declined linearly between 1900 and 1950.4 The case can hence be made to favor linear decline for fitting and extrapolation.

Van Wijhe et al performed exponential fits on the mortality data and expressed mortality as average years of life lost under the age of 20 per live birth. Linear fits, however, have better R2 values than exponential fits for the four diseases shown in the Figure. Linear extrapolation leads to the conclusion that vaccination averted virtually no deaths from diphtheria and pertussis, as the lines intersect the axis around 1953. For tetanus, the intersection lies within the vaccination period but still a substantial difference exists between the linear and the exponential fit. The trend in the data for poliomyelitis is neither linear nor exponential, with a sharp drop in 1957 when vaccination started.

FIGURE.
FIGURE.:
Childhood mortality burden per live birth, The Netherlands, 1903–1975, for the first four infectious diseases in the Dutch State Vaccination Program. Black line = observed mortality; blue dashed line = exponential fit; red dashed line = linear fit. R 2 values were calculated for the pre-vaccination periods 1903–1913, 1919–1933, and 1948–1952 (diphtheria), 1903–1953 (pertussis), 1920–1956 (poliomyelitis), and 1903–1917 and 1925–1953 (tetanus). Gray area indicates the vaccination period. Figure is available in color online.

In conclusion, we agree that van Wijhe et al have provided one coherent interpretation of mortality decline due to the four infectious diseases. Here, we argue that an alternative interpretation exists that is at least equally defendable. Linear modeling would lead to the conclusion that van Wijhe et al overestimated the effectiveness of the Dutch State Vaccination Program.

Toon Weisenborn
Physicist em. Nordhorn, Germany

Aart J. Nederveen
Department of Radiology and
Nuclear Medicine
Amsterdam University Medical Centers
University of Amsterdam
Amsterdam, The Netherlandsa
[email protected]

J.A. Stalpers
Department of Radiation Oncology
Amsterdam University Medical Centers
University of Amsterdam
Amsterdam, The Netherlands.

REFERENCES

1. van Wijhe M, McDonald SA, de Melker HE, Postma MJ, Wallinga J. Estimating the population-level effectiveness of vaccination programs in the Netherlands. Epidemiology. 2018;29:215–223.
2. CBS-Statistics Netherlands. Available at: https://opendata.cbs.nl/statline/#/CBS/nl/dataset/37556/table?ts=1544215896296. Accessed July 31, 2019.
3. Tuljapurkar S, Li N, Boe C. A universal pattern of mortality decline in the G7 countries. Nature. 2000;405:789–792.
4. Wolleswinkel-van den Bosch JH, Looman CW, Van Poppel FW, Mackenbach JP. Cause-specific mortality trends in The Netherlands, 1875-1992: a formal analysis of the epidemiologic transition. Int J Epidemiol. 1997;26:772–781.
Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.