Main Menu              Health Library      

MAMMOGRAMS and RISK of BREAST CANCER

Abstract:  Mammograms done at intervals of two years or less can significantly reduce the risk of Breast Cancer if they are done regularly for long periods of time.  Risks of Breast Cancer will be reduced about 19% after ten years, 27% after 15 years, and about 35% after 20 years of their use.  Starting mammograms at age 50 will add about 160 Well-Days to the healthy life of an average woman and up to 400 Well-Days for those at high cancer risk, and eventually can reduce risk of Breast Cancer nearly in half.  Added benefits derive from starting mammograms at age 40 or doing them every year, but these added benefits are small.

Although regular mammograms have long been used by millions of women to reduce risk of breast cancer, some researchers occasionally have questioned their value.  Thus a review and brief global analysis of the actual research was done for Life Ahead. This review that follows shows that the potential benefits of regular mammograms in reducing risk of death from breast cancer are quite significant.  A new Life Ahead method may identify their likely benefit more accurately than previously has been possible.

How Mammograms Reduce risk of Breast Cancer:    It is useful to review what mammograms might be expected to accomplish as background for understanding actual research results.  Radiology appears to be able to detect a normally growing tumor of about 15 mm size about 3 years before the tumor will progress to the size likely to be noted by a doctor’s screening.  (This size at detection may be reduced in the near future).  This provides a greater chance of excising a tumor before it becomes a dangerous cancer.  However, a doctor’s usual examination still may detect it in time to prevent this problem. Thus the potential benefit of the mammogram is to eliminate some number of potential dangerous tumors or cancers that otherwise would not be detected by the doctor.  Data cited in the references included in Table M that follows suggest that mammograms will identify about twice the number of early tumors that will be found and successfully treated by usual medical screening. This is a maximum benefit, because as time proceeds more and more small tumors and cancers also will be successfully identified by conventional screening.

Removing an early growth that otherwise would progress will not usually save actual deaths from cancer until many years into the future. The mean survival time for breast cancer from the Seer Cancer Statistics is about 17 years.  See (www.seer.cancer.gov/publications/data). This means that a successful removal of an early cancer may accomplish a 50% chance of saving a death from that cancer 17 years into the future. Recognizing that a mammogram could give a further 3 years lead time before an early cancer might occur, the mean survival time from a detection via mammogram to a 50% likely death from that cancer will be about 20 years. Many cases of Breast Cancer of course are detected and cured.  The real accomplishments of mammograms must be measured by their ability to save actual deaths from Breast Cancer over the long term.

Mammograms Reduce Breast Cancer Risk only Gradually over Time:  Although a most likely time for good benefit in saving deaths via mammogram is 20 years ahead, a few cancers grow much faster than average and will occur sooner. The Seer statistics show that survival from cancer decreases at a near fixed percentage each year, and for a 20 year survival this percentage will be 3.5% per year. This suggests that although it will take 20 years for a tumor found by mammogram to cause a 50% chance of death, 3.5% still might develop to death in a first year, 7% by year 2, 17% by year 5, and 31% by year 10.  This identifies a “maximum” likely benefit from a mammogram assuming no alternate detection was done. But conventional doctor's exams should identify perhaps half of the growths identified via mammogram. This means that a more likely potential for mammograms to reduce cancer deaths is less than these amounts.  The benefit of mammograms will be observed best by carefully tracking the Breast Cancer death rates of regular mammogram users with the Breast Cancer death rates of closely similar populations that do not use them. The difference in death rates of these two groups should be minimal at first, but should gradually increase over years of time as more and more lives are saved by the longer use of mammograms.

Populations of mammogram users and non-users in these research studies were selected by various methods of selecting or randomizing them from an overall population.  Imperfections are inevitable in this selection.  It simply is impossible to select absolutely identical groups of people because those that will be willing to take mammograms always will be somewhat different that those that are not interested in taking them.  But there is a useful way for verifying that selected populations have similar initial breast cancer risks. This is to compare the death rates of the groups in the early or base years of the study.  During this period any benefit of mammogram in reducing reduce breast cancer death rates should be minimal because use of mammograms will be for too short a time to produce any benefit. Thus during early study years the observed Breast Cancer risk for users should be similar to and only slightly lower than the risk of non-users.

Although the Breast Cancer death rates of users and non users should be the same at start of the research, this is not an absolute requirement and is not always achieved.  The real test of mammogram effectiveness will be its ability to develop an increasingly lower death rate for users vs. non-users as time and durations of mammogram use proceed.  Even if there is some difference in the base or initial risk of users and non-users, a benefit of mammograms still should be observable from the change in benefit in user risk vs. the risks of non users. 

The Actual Research:  Table M following summarizes the results of seven major studies of the use of mammograms. These all involved very large populations, typically of 26,000 to 61,000 women.  The table shows risk ratios for differing year periods of use, as 5, 10 and 15 years and notes duration times of some results that approximated these years.  A risk below 1.0 identifies a benefit for mammogram use. All but two results at the 5 year time were close the expected near unity or 0.9 ratio expected, suggesting that the populations being compared did have the similar initial death rates from breast cancer.  At or near the 10 year time an average of all risk ratios was down to 0.84 vs. the 1.08 average at year 5.  And at duration of mammogram use approaching 15 years the average study risk ratio averaged 0.77.  These benefits should improve further as the 20 year time of mean survival is obtained.  Mammograms usually were taken at about 2 year intervals.

The level of benefits expected is often below the error margins of the individual studies. Thus the fact that an average reduction in risks from all of the the 26 different risk comparisons in Table M was achieved increases credibility of the conclusion that mammograms reduce risk of death from breast cancer.  Two comparisons indicate a difference in user and non-user base risk at study start – that of the Malmo group, and the Canadian group for the 40-49 year group. Both of these ratios are based on limited numbers of deaths and have high error margins.  The Canadian value at 7 years for the 50+ year old groups that is based on a larger number of deaths is closer to that expected.  But note that in each of these ‘divergent’ studies the expected reduction in risk ratio was confirmed as the study duration increased. 

Reductions in Risk with Mammograms:  A best present estimate of the success of mammograms in reducing risk over time can be obtained from a regression of the all of the available risk values for use and non-use vs duration of mammogram usage. This approach minimizes any problems due to ‘randomization’ or imbalances of study group risk at study start.  Results of statistical regressions for the reduction in risk per year of use assuming a zero effect of use at study start but with allowance (i.e using of dummy variables) for the initial imbalances in the Malmo and Canadian studies are listed following this discussion.  The average results from all 26 comparisons is a very significant reduction in risk of Breast Cancer by 0.979 per year of mammogram use.  For women starting mammograms at age 50, the risk reduction was 0.975 per year, and for those starting mammograms at age 40 the risk reduction was a slightly lesser 0.982 per year.

These annual reductions in risk are small.  But over many years of use, the all study formula produces cumulative risks for 2 year interval mammograms that are 0.90 (or 10% reduction in risk) by year 5;  0.81 (or 19% reduction) by year 10;  0.725 (or 27.5% reduction by year 15, and 0.65 (or 35% reduction in risk) by year 20.  These are consistent with but slightly better than the values of 0.84 and 0.77 derived from average study risk ratios at the 10 and 15 year times.

Because mammograms target only one cause of premature death their effect in extending healthful life is modest, ranging from about a usual 160 Well-Days for those at average Breast Cancer risk to 400 Well-Days for those at rather high risk. These benefits assume that the alternate to mammograms will be regular visits to the doctor for examinations.  Women that do not schedule such regular doctor visits should benefit much more than this from use of regular mammograms.

Most doctors suggest initiating mammograms at about age 50, but their use will produce some added benefit if started at age 40. Life Ahead now computes likely mammogram benefits for any combination of women's initial risk of breast cancer, frequency and duration of mammogram use, and age at which they are started.  Annual mammograms may provide a 20% advantage over bi-annual mammograms.  Some modest further benefit accrues if they are started at age 40 rather than at age 50.  Half credit is accorded to 3 year mammograms and no credit is given for those at intervals of more than 3 years.  Because a health interested woman should consider regular mammograms, an estimate of their potential Well-Days benefit will be provided by the program if they are not entered as being used.

A Life Ahead Valuation can be Useful:  Conventional health advice can suggest that every woman should do the same thing.  This is type of health advice is adequate for most women.  But Life Ahead can provide a specific valuation of benefit for individuals that previously has not been available.  The Life Ahead formulas can value the benefit of mammograms starting at any age and frequency of use, and for women of any level of cancer risk.  Those having very high risks of cancer almost certainly will benefit significantly from mammograms.  Those older and with very healthful habits and lowest overall risks of cancer may find that a Well-Days benefit for their use is too small to be worth the cost and time required.

Other Considerations can Affect Choice of Mammograms:  These estimates via Life Ahead are useful only for defining a most likely reduction in Breast Cancer risk by use of regular mammograms.  Many other factors become operative in reaching any overall medical conclusion about their use.  There are costs and inconvenience of testing; the problem of many false positives and trauma caused by these; and the effect of the radiology required on other cancer. The differing success of methods of treating those identified with specific types of cancer growths is important to overall decisions on use mammography.  The complexity of these other factors has led occasionally to controversy about their use on overall populations. But a majority opinion still holds that regular continued mammograms provide a highly useful medical approach to the saving of women's lives.  

 Formulas, and Valuation of Mammogram Use in Life Ahead:

(1) For all 26 study comparisons, all starting ages:     

     Log risk ratio = - 0.0214 * years of use.    (0.979 per year of use)

        t of coefficient = 5.78; (5-95% -0.014 to -0.029);  r = 0.77

(2) For 13 comparisons starting at ages 50 and older

     Log risk ratio = - 0.0246 * years of use.   (0.975 per year of use)

       t of coefficient = 5.99; (5-95% -0.016 to -0.033);  r = 0.89

(3) For 13 comparisons, starting ages 40-49:

     Log risk ratio = - 0.0183 * years of use.   (Risk of 0.982 per year of use)

       t of coefficient = 2.77; (5-95% -.0051 to -.0315);  r = 0.66

A t value of about 2 identifies 95% significance. The t values for the coefficient from all 26 research comparisons and that of the 13 comparisons at age 50+ are very high, indicating likelihood much better than 99.9% that mammograms provide benefit. The age 40-49 group shows an a bit lower effect but is at well over 95% significance.

The Life Ahead Model uses incidence of disease at year to compute lifetime risks. The above risks are those measured cumulatively over the years noted, and are not incidence at year. Incidence values at year change at about twice the level of above cumulative risks.  At log incidence ratios of  - 0.0428 * years of use, incidence of breast cancer for use of mammograms at 5, 10 and 15 years will  be about 0.81, 0.65 and 0.53 respectively vs. conventional treatment for the 50+ age group. A summing of the annual values of these rates of incidence produces the cumulative risks to stated year from the Formulas (1) to (3).

A much publicized analysis by Danish researchers some time ago claimed that mammograms have no benefit.  This analysis was based on conjecture about how studies were randomized, and appears to be incorrect.  A memo on flaws in this analysis can be provided on request.

     Table M:  Mammogram Use and Women’s Death Rates from Breast Cancer