Overestimated lead times in cancer screening has led to substantial underestimation of overdiagnosis.

BACKGROUND: Published lead time estimates in breast cancer screening vary from 1 to 7 years and the percentages of overdiagnosis vary from 0 to 75%. The differences are usually explained as random variations. We study how much can be explained by using different definitions and methods. METHODS: We estimated the clinically relevant lead time based on the observed incidence reduction after attending the last screening round in the Norwegian mammography screening programme. We compared this estimate with estimates based on models that do not take overdiagnosis into account (model-based lead times), for varying levels of overdiagnosis. Finally, we calculated overdiagnosis adjusted for clinical and model-based lead times and compared results. RESULTS: Clinical lead time was about one year based on the reduction in incidence in women previously offered screening. When overdiagnosed tumours were included, the estimates increased to 4-9 years, depending on the age at which screening begins and the level of overdiagnosis. Including all breast cancers detected in women long after the end of the screening programme dilutes the level of overdiagnosis by a factor of 2-3. CONCLUSION: When overdiagnosis is not taken into account, lead time is substantially overestimated. Overdiagnosis adjusted for model-based lead time is a function tending to zero, with no simple interpretation. Furthermore, the estimates are not in general comparable, because they depend on both the duration of screening and duration of follow-up. In contrast, overdiagnosis adjusted for clinically relevant tumours is a point estimate (and interpreted as percentage), which we find is the most reasonable method.

[When will mammography screening reduce breast cancer mortality?]. / Når vil mammografiscreening gi redusert brystkreftdødelighet?

BACKGROUND: Norwegian authorities have decided to start a mass mammography screening programme. One third of the population has been included in a pilot study from 1996. The Cancer Registry of Norway maintains that there will be at least a 30% reduction in mortality, but that an effect may not be detected until the ten-year follow-up. METHODS: We assume that the effect of mammography screening is constantly increasing over a ten-year period before maximum effect is reached. We also assume that the effect of screening in the age group 50-74 years is 80% of maximum effect. We simulate Norwegian breast cancer mortality rates under the assumption that mammography screening reduces breast cancer mortality in the age group 50-69 years with 15% and 30% effects, respectively. We also simulate 30% and 50% effect in the pilot study. RESULTS: If the effect in the Norwegian population is 30%, one may expect to see a significant decline after five years; however, if the effect is only 15%, one has to wait for a longer time. If the effect is 50%, as the Cancer Registry of Norway has argued, one should see a significant effect in the pilot study after six years. CONCLUSIONS: We think it a contradiction to argue that mammography screening reduces breast cancer mortality by 30%, but that one has to wait ten years to observe an effect on national mortality rates. We suggest that the breast cancer mortality rate in the pilot study is estimated. We also argue that observed reductions of less than 10% in Sweden, Finland and England are strong evidence that the effect of mammography today is far less than 30%.

[Norwegian breast cancer mortality rates and validity of Swedish mammographic studies]. / Norske tall for brystkreftdødelighet og validitet i svenske mammografistudier.

BACKGROUND: The claim that screening for breast cancer with mammography reduces breast cancer mortality is mainly based on the results from the Swedish two-county trial (WE study), where the effect was reported to be 30% for the age group 50-69 years. The two-county trial has recently been criticised for inadequate randomisation and for not following the study protocol. METHODS: We do some simple calculations to study whether the WE study is robust for an alternative statistical analysis. We use stage-specific breast cancer mortality in the Norwegian population as the baseline mortality rate in Sweden. Then we study the expected reduction in overall breast cancer mortality in the WE study while we vary the mortality rate in stage 1 and the stage distribution. RESULTS: We show that a 30% reduction in overall mortality rate is in conflict with observed decline in mortality in stage 1 and the expected stage migration. One either has to decrease mortality in stage 1, or increase the reduction of tumours with distant metastases, or both, to much higher levels than those reported in Sweden to get a 30% reduction in overall mortality of breast cancer. CONCLUSIONS: Our study adds further evidence to the proposal that the WE study is biased and not valid.

Cancer incidence for Swedish twins studied by means of bivariate frailty models.

Cancer incidence rates for Swedish twins born between 1928 and 1965 and who both were alive at age 30 are studied by means of bivariate frailty models. Altogether, 7,280 fraternal (DZ) and 4,699 identical (MZ) twin pairs were followed up through December 31, 1995, for cancer status. The association between cancer incidence rates was statistically greater among the MZ than among the DZ pairs and stronger between women than between men; however, the magnitude of this association is relatively small and decreases over time. The relative decrease in dependency (association) is most easily detected using shared frailty models but may also be demonstrated, at least for women, using correlated frailty models. We also demonstrate that estimates of the correlation coefficient are similar when using any correlated frailty models derived from the power variance family but that these estimates disagree regarding the age at which the dependence is most important. The relative importance of dependence across age may sometimes be more interesting than the correlation coefficient itself. The latter may usually be estimated using alternative methods. Furthermore, when estimating correlation coefficients close to the boundary of the parameter space, simulation studies indicate that the correlated inverse Gaussian frailty model is more robust than the gamma frailty model.

[Is the benefit of organized mass screening for cervix cancer and breast cancer in Norway scientifically justified?]. / Er nytten av organiserte masseundersøkelser for livmorhalskreft og brystkreft i Norge vitenskapelig bevist?

BACKGROUND: It has been argued that mass screening may reduce mortality from cervical cancer and breast cancer in Norway by 50% and 39%, respectively. However, some authors doubt that mass screening for breast cancer is justifiable. This article discusses whether mass screening for cervical cancer in Norway is more effective than opportunistic screening. MATERIAL AND METHODS: The scientific methods used to justify organised mass screening for cervical cancer are reviewed and discussed, as are the randomised trials used to justify screening by mammography. The author is especially critical of the use of descriptive statistics to justify screening, on the assumption that there is a basic constant incidence or mortality rate. RESULTS: The incidence rates of pre-malignant conditions and cervical cancer in Norway are presented. There is no essential increase in the number of pre-malignant conditions after three years of organised mass screening compared to the previous three-year of unorganized screening. However, the number of pre-malignant conditions did increase from 1987 to 1992 due to a national recommendation for testing every three years. By contrast, incidence rates for cervical cancer remained almost constant in the 1987-97 period. INTERPRETATION: There is no scientific proof of organised mass screening for cervical cancer being preferable to unorganized screening. The author warns against only comparing observed rates with historic rates and rates in other countries; alternatively, rates among those who participate in the screening might be compared in a simultaneous analysis to rates among those who do not respond to the invitation. The author also suggests estimating correlation coefficients between individual test results to identify high-risk individuals.

Survival of glaucoma patients.

PURPOSE: To investigate the survival of patients with capsular or simple glaucoma compared with that of the common population, with particular attention to the impact of sex and use of acetazolamide (Diamox). METHODS: The 30 year survival of 1147 patients with capsular or simple glaucoma who were finally hospitalized at the Eye Department, Rikshospitalet, Oslo, from 1961 to 1970, are analysed, using log rank tests. The time varying impacts of sex and acetazolamide on survival are also studied using a regression model. RESULTS: There was a significant increased mortality for patients with acetazolamide, and for men also those not using it. The observed mortality for men was initially lower than the average Norwegian population, but later the mortality increased more rapidly in the glaucoma group. This may be explained by a selection of the healthiest patients to Rikshospitalet, and actually indicates that the excess mortality is even higher than calculated here. CONCLUSION: The analysis of data indicated increased mortality for glaucoma patients when the disease had lasted for some time. This was especially pronounced for men using acetazolamide. A similar study from a period when acetazolamide was not in common use and an analysis of causes of death is also asked for.

Adjusting and comparing survival curves by means of an additive risk model.

Survival curves may be adjusted for covariates using Aalen's additive risk model. Survival curves may be compared by taking the ratio of two adjusted survival curves; the ratio is denoted the generalized relative survival rate. Adjusting both survival curves for all but one of a common set of covariates gives the partial relative survival rate, which measures the covariate-specific contribution to the generalized relative survival rate. The generalized and partial relative survival rates have interpretations similar to the traditional relative survival rates frequently used in cancer epidemiology. In fact, the traditional relative survival rate can be generalized to a regression context using the additive risk model. This population-adjusted relative survival rate is an alternative and useful method for removing confounding effects of age, cohorts, and sex. The authors use a data set of malignant melanoma patients diagnosed from 1965 to 1974 in Norway. The 25-year survival of 1967 individuals is studied.

Long-term survival of breast cancer in Norway by age and clinical stage.

In this paper 25-year survival for 8802 Norwegian women with breast cancer diagnosed during the period 1965 to 1974 is studied. It is suggested that some of the contradictory reports in the literature of the prognostic effect of age and clinical stage on long-term survival may be caused by interactions and time varying effects of covariates. When using a linear non-parametric regression model that allows the covariates to vary over time, age and clinical stage are found to be significant long-term prognostic factors. A significantly higher excess mortality for women less than 35 years at diagnosis disappeared after 8 years, while for those above 55 years an important effect of age on the long-term survival, especially for those with regional cancer, was seen. The effect of clinical stage on survival varies strongly over time, and was significant between 15 and 20 years.

Frailty modelling for the excess hazard.

Long-term excess hazards for cancer survival sometimes tend to zero or become negative even though we expect them to be positive. This may be explained by selection at diagnosis; individuals with certain cancers may have an increased risk of dying of other diseases in general. Then comparing with population mortality rates is not correct. Alternatively, we may have a continuous selection of the most robust individuals after diagnosis. When there are unobserved heterogeneity, and those with highest risk of dying of cancer also have the highest risk of dying of other diseases, this will cause selection after diagnosis. This may be modelled by multivariate frailty variables, and a corrected excess hazard may be estimated. In two examples, these corrected excess hazards give a better estimate when comparing to the cause-specific cancer mortality. Actually, this study questions the usefulness of long-term excess hazard rates.

[Breast cancer--incidence, mortality and stage shifting in Norway]. / Brystkref--insidens, mortalitet og stadiemigrasjon i Norge.

The age-standardized incidence rate of breast cancer has increased by 50% over the period 1965-94. There has been a much lesser increase in the corresponding age-standardized mortality rate because of better treatment and stage shifting. Stage shifting means that the proportion of individuals with a given clinical stage changes over time. The proportion of individuals diagnosed as clinical stage 1 was seen to increase from about 50% to 60% in the time period mentioned above. The proportion classified as stage 2 at the time of diagnosis is constant at 30%. The relative numbers of individuals diagnosed as stage 3 or 4 were reduced from 10% to 3% and from 10% to 5%, respectively. After correction for confounding effect of age and the clinical stage, the age-standardized 3-year relative survival rate increased from 90% to 95% and from 67% to 85% for stages 1 and 2, respectively. The impact of advancing breast cancer diagnosis independent of the screening programme, is discussed. Finally, the evaluation of screening programmes using shift migration models and simulations is discussed.

A proportional regression model for 20 year survival of colon cancer in Norway.

Twenty year survival of all Norwegians with colon cancer registered in a period of 10 years is estimated by both relative survival rates, and with a proportional regression model for the excess intensity. Male colon cancer patients have a significant positive excess mortality at least 20 years after diagnosis, while the excess mortality for females is about zero after 10 years. Stratified analyses for men indicate non-proportionality throughout the follow-up period, and when this information is included in the regression model, there are significant effects of age between 60 and 70 years and for pelvic cancer. The use of proportional regression models is also discussed when excess intensities are close to zero or negative.