Methods for Estimating the Economic and Disease Burden of BC Due to Suboptimal BF

1

Supplementary material

Methods for estimating the economic and disease burden of BC due to suboptimal BF

Fertility and breastfeeding simulation

We modeled a cohort of women were 90% of them have children over their life course, according to estimations of fertility rates by age among Mexican women in 2012 and considered that 10% of women in the cohort were nulliparous[1]. According to the Mexican National Health and Nutrition Survey (ENSANUT 2012) we assumed that only 93.7% of parous women breastfeed at least once[2].

For the duration of breastfeeding baseline scenario, we simulated the months of breastfeeding per child for each parous woman using a geometric distribution truncated between 0-36 months, with the median duration of BF by women´s age group (15-18, 19-24, 25-35, +36) as observed in ENSANUT[2]. We used this information to simulate that 33% of parous women breastfed between 12-36 months per child[2].The lifetime months of breastfeeding per woman were aggregated.There was a lack of information about the correlation between the breastfeeding duration of children from the same mother. Thus we assumed the duration of breastfeeding to be independent for each child for multiparous women.

For the exclusive breastfeeding baseline scenario, we assumed a binomial distribution to simulate that 14% of parous women breastfed all of their infants exclusively for 6 months [2].Women who breastfed, but not exclusively, were considered to do so partially.

For the optimal duration of breastfeeding scenario, we used a truncated geometric distribution per women´s group age, as in the baseline scenario, to simulate that 95% of parous women breastfeed each child between 12 and 36 months. For the exclusive breastfeeding optimal scenario, we assumed that 95% of women breastfeed exclusively the first 6 months of their infant´s life.

Breast cancer incidence and mortality.

As the GLOBOCAN incidence was aggregated into 5-year age groups, we extrapolated linearly to have a specific incidence for ages 15 to 90 years. Breast cancer incidence by type and duration of breastfeeding in parous women, was estimated by multiplying the breast cancer RR of the different breastfeeding practices by the breast cancer incidence in the general population, and considering breastfeeding practices observed in Mexico in 2012.The incidence of BC by accumulated duration or type of BF was calculated with the following formulae [3]:

(1)

(2)

Where i = 15,…,90 years-old and j = months of accumulated BF (>0-6mo, 7-18mo, 19-30mo, 31-54mo, +55mo) if scenario a), or j= type of BF (exclusive or partial) if scenario b). is the GLOBOCAN incidence of BC in Mexican women in 2012. is the current incidence of BC in parous women that do not breastfeed and is the current incidence of BC in parous women that have j months of BF or have j type of BF. is the proportion of parous women that do not breastfeed in 2012, is the proportion of nulliparous women and is the proportion of parous women in 2012 that breastfeed for j months or by type of BF. is the relative risk of BC for nulliparous women versus parous women and is the relative risk of BC for BF j months or by type of BF in comparison to parous women that have never breastfed. The RRs for breast cancer associated with duration of breastfeeding used in the simulations are drawn from estimations reported by The Collaborative Group on Hormonal Factors[4](breastfeeding for < 6 months: RR 0.98; breastfeeding 7-18 months: RR 0.94; breastfeeding 19–30 months: RR 0.89; breastfeeding 31–54 months: RR 0.88; breastfeeding ≥ 55 months: RR 0.73) and the risk of breast cancer in nulliparous women compared to parous women (RR: 1.07) from the same study[4]. The RRs for breast cancer associated to exclusive breastfeeding for 6 months are derived from a recent meta-analyses; the RR for exclusive breastfeeding in parous women was 0.72; and for partial breastfeeding in parous women, it was 0.89 [5].

In the baseline scenario, we assume that women have children and breastfeed before developing breast cancer according to 2012 fertility and breastfeeding rates in México. In the optimally breastfeeding scenario, we assume that 95% of parous women breastfeed before developing breast cancer. Simulated breast cancer cases were assigned to a clinical stage of disease (in situ, I, II, III and IV) at the time of diagnosis and a state of mastectomy (surgery or not), assuming the same distribution as cases observed in Mexican public health institutions [6]. We assumed that women with breast cancer stay in the same clinical stage as diagnosed at baseline. Transition between stages was not modeled, however, lifetime costs for each woman reflects the progression of the disease between clinical stages based on a study conducted by Knaul [7]. Information about cure or death from breast cancer and length of survival at different stages of disease were obtained from a study of breast cancer 5-year survival rates in Mexican women [6]. Death from causes other than breast cancer was simulated before estimating cancer cases to simplify the model. If a woman died before having breast cancer, she was not considered an incident.

Death from other causes than BC

A survival function was estimated with the probability of death from causes other than BC. The probability of death was assessed subtracting deaths from BC [8] from total women´s deaths by age in 2012 [9] and dividing by population of women by age in the same year[9].

Medical costs

Medical costs include consultations, diagnosis, surgical procedures, radiotherapy, chemotherapy, medications, antineoplastic agents, endocrine treatments and follow-up[7,10]. Since palliative care is not reported by IMSS, we considered additional palliative care costs per case/year for BC patients at stage IV[11]. Fixed costs of hospital equipment or infrastructure associated with medical attention are not included. We assumed that allhealthcare costs were covered by the IMSS for employed women with social security benefits and by the FPGC for employed women without social security and unemployed women. We assumed that employed women enrolled in other social security institutions in Mexico (i.e. the 7% covered by ISSSTE, PEMEX, SEDENA, SEMAR) [12] had the same costs as those under the IMSS.We assumed that all women with breast cancer received treatment and estimated costs for five years of treatment for women who survive breast cancer [7] and three years for women with breast cancer in situ (stage 0). For women who died from breast cancer, we estimated medical costs only for the years from diagnosis to death (a maximum of five years).

Non-medical direct costs

Caregivers were randomly assigned to employed or unemployed status, using a binomial distribution [13].

Income and probability of employment

Using a binomial distribution, we randomly assigned each woman with BC to be employed or not, according to their age of BC onset. Four types of employment were considered: salaried and non-salaried employment with or without social security benefits. A monthly average income was assigned to each woman according to her type of employment and age. The minimum wage was assigned to women without employment in the labor market. The probability of employment and average income by women´s age group was estimated from the National Survey of Occupation and Employment second trimester 2013 (ENAOE 2013)[14].

Morbidity costs

The cost of temporary absence from work was considered to be the present value of monthly average income for women, according to age and type of employment at the time of the disease, converted to daily values and multiplied by the number of disability days. For unemployed women, we considered the minimum wage to be the opportunity cost of morbidity. Morbidity costs are estimated during years of survival for women who die from BC or assuming five years of treatment for women who survive BC and three years of treatment for women with BC in-situ (stage 0). We assumed that women with BC at stage IV are permanently disabled to work. Their cost of permanent absence from work is estimated as the present value of an annuity of lost income (by age and type of employment), or annual minimum wage (for unemployed women) during the years the woman lived since detection of BC to death or life expectancy age (whichever comes first).The probability of employment and average income by women’s age group was estimated from the 2013 National Survey of Occupation and Employment second trimester (ENAOE 2013) [14] (see Supplementary Table 1).

Premature mortality costs

We estimated the working years lost at the time of death from breast cancer by considering employment probabilities at the age of disease onset. The cost of mortality is the present value of an annuity of lost income (by age and type of employment), or lost minimum wage (for unemployed women) during the years from death from breast cancer to the life expectancy [15].

Employer costs

According to article 98 of the Social Security law in Mexico, for temporary absence from work the short-term subsidy is 60% of the last daily wage of contribution [16]. For permanent absence from work, article 141 of the Social Security law in Mexico establishes that the disability pension corresponds to 35% of the average of the salaries corresponding to the last five hundred weeks of contribution [16]. For premature death, the life pension for widowers correspond to 90% of the amount that would have corresponded to the insured in the case of disability [16]. Disability and life pensions were estimated as the present value of an annuity of lost income multiplied by the corresponding pension factor at the moment of the disease or death, respectively.

Supplementary Table 1. Parameters used to model the burden of breast cancer attributed to suboptimal breastfeeding.

Variable / Parameter / Description / Probabilistic sensitivity analysis / Sensi-tivity analysis / Source
Demographic characteristics
Demographic composition / 1,115,925 / Number of 15 year-old women in Mexico in 2012 / - / - / [9]
Fertility rates by age group / See figure 1 / Observed number of births in Mexico in 2012 divided by number of women in each age group / - / - / [9]
Life expectancy (years) / 77.4 / Average number of years to be lived by Mexican women in 2012 / - / 65 / [9]
Mortality rates (without BC) / See figure 2 / Total female deaths without breast cancer deaths among the female population by age group in 2012. / - / - / [9]
[8]
Breast cancer epidemiology statistics
BC incidence rate in Mexican women / 15-39 years: 11.7
40-44 years: 58.7
45-49 years: 84.3
50-54 years: 106.5
55-59 years: 118.6
60-64 years: 119.4
65-69 years: 116.6
70-74 years: 108.8
75+ years: 96.6 / Age-standardized rates per 100,000. Estimated from national mortality estimates using modelled survival. / - / Projec-tions of incidence in 2035 / [17]
BC incidence rate separated by BF duration / See figure 3 / Age-standardized rates per 100,000. Separated by BF duration using formulae 1 and 2 / Lognormal distribution for the RR of BC / Using data for RR of BC from
Unar-Munguia, et.al. / [17]
[4]
[5]
BC clinical stages / Stage 0 (in situ): 2.8%
Stage I: 10.5%
Stage II: 38.6%
Stage III: 40.7%
Stage IV: 7.4% / Percentage of detected BC cases by clinical stage. / - / - / [6]
Mastectomy / Stage I: 93.2%
Stage II: 90.1%
Stage III: 42.5%
Stage IV: 57.1% / Percentage of BC cases that received surgery by clinical stage / - / - / [6]
BC survival by clinical stage. / Stage I: 92.1%
Stage II: 90.3 %
Stage III: 63.4%
Stage IV: 47.6%
See figure 4 / Kaplan-Meier survival estimates for BC stages. 5-year follow-up / - / - / [6]
Days of disability per 5 years treatment (for women with BC) / 180 / Number of days in a year a Mexican woman with BC is disabled to work. Approximate to the average doctor visits, chemotherapy sessionsfollow-up visits and laboratory tests required for treating BC in Mexico. / - / - / [18]
Breastfeeding statistics
Duration of any mode of BF by age / 15-18 years: 10.2 (0.42 MAD)
19-24 years: 10.4 (0.42 MAD)
25-35 years: 10.4 (0.41 MAD)
36+ years: 8.0
(0.39 MAD) / Median duration in months per child of any mode of BF in Mexican women 2012 / Normal distribution / - / [2]
Percentage of not BF in parous women / 6.6% / Percentage of women with exclusive BF ≤ 6 mo in 2012 / Beta
distribution / - / [2]
Percentage of exclusive BF / 14.4% / Percentage of women with exclusive BF ≤ 6 mo in 2012 / Beta
Distribution / - / [2]
RR for BC by duration of BF in parous women / BF ≤ 6 mo: 0.98 (95% CI 0.95, 1.01)
BF 7-18 mo: 0.94 (95% CI 0.91, 0.97)
BF 19-30 mo: 0.89 (95% CI 0.84, 0.94)
BF 31-54 mo: 0.88 (95% CI 0.81, 0.94)
BF + 55 mo: 0.73 (95% CI 0.63, 0.82) / Relative Risk for BC in parous women by duration in months of any mode of BF compared to parous women who had never breastfed / Lognormal distribution / Using data from
Unar-Munguia, et.al. / [4]
SRR for BC relative to exclusive BF in parous women / SRR: 0.72 (95% CI 0.58, 0.90) / Summary Relative Risk for BC in parous women with exclusive BF compared to parous women who had never breastfed / Lognormal distribution / - / [5]
SRR for BC relative to any BF in parous women / SRR: 0.89 (95% CI 0.83, 0.95) / Summary Relative Risk for BC in parous women with any mode of BF compared to parous women who had never breastfed / Lognormal distribution / - / [5]
RR for BC for each birth / RR: 1.07 (95% CI 1.05, 1.09) / Relative risk for BC in nulliparous women compared to parous women for each additional birth. / Lognormal distribution / - / [4]
Disability weights
Disability weights for breast cancer / Cancer: diagnosis and primary therapy: 0.294 (95% CI 0.199, 0.411)
Cancer: metastatic
0.484 (95% CI 0.330, 0.643)
Mastectomy:
0.038 (95% CI 0.022, 0.059)
Terminal phase: with medication:
0.508 (95% CI 0.670, 0.348)
Terminal phase: without medication: 0.519 (95% CI 0.683, 0.356 ) / A weight factor used to estimate years lived with disability. It reflects the severity of the disease on a scale from 0 (perfect health) to 1 (equivalent to death). 2010 / Beta distribution / - / [19]
Breast cancer costs
IMSS medical costs of BC per women/year by clinical stage (2015 USD) / Stage 0 (in situ): NA. SPSS cost assumed.
Stage I: $ 7,743
Stage II: $ 10,349
Stage III: $ 15,271
Stage IV: $ 19,557 / Medical costs of BC (diagnostic and treatment per women per year) at the Mexican Institute of Social Security. Costs deflated to 2015 prices and converted to dollars. / Gamma distribution / Increase in cost each year at the rate of inflation / [7]
SPSS medical costs of BC per women/year by clinical stage (2015 USD) / Stage 0 (in situ): $ 2,407
Stage I: $ 13,500
Stage II: $ 13,500
Stage III: $ 18,334
Stage IV: $ 19,267 / Medical costs of BC (diagnostic and treatment per women per year) at Seguro Popular. Deflated to 2015 prices and converted to dollars. / Gamma distribution / Increase in cost each year at the rate of inflation / [10]
Medical costs of palliative care costs of BC per women (2015 USD). / Stage IV $ 141.0 / Medical costs of palliative care deflated to 2015 prices and converted to dollars. / Gamma distribution / Increase in cost each year at the rate of inflation / [11]
Employment statistics
Probability of employment among women ≥ 15 years by age / See figure 5 / Proportion of Mexican women ≥ 15 years employed among women population ≥ 15 years. 2013 / - / - / [14]
Probability of salaried and non-salaried employment among employed women ≥ 15 years by age / See figure 6 / Proportion of Mexican women ≥ 15 years with salaried and non-salaried employment among employed women ≥ 15 years. 2013 / - / - / [14]
Probability of benefits among salaried women ≥ 15 years by age / See figure 7 / Proportion of salaried Mexican women ≥ 15 years with social security benefits among employed women ≥ 15 years. 2013 / - / - / [14]
Probability of employment among carers / 0.196 / Proportion of carers (of women with BC) with employment (salaried or non-salaried). / Beta distribution / - / [13]
Income
Monthly income of women by type of employment and benefits / See figure 8 / Monthly income in 2015 dollars. Women with salaried or non-salaried employment, with or without social security benefits. / - / Annual GDP per capita in dollars 2015
$10,325
(World Bank) / [14]
Disability and pension benefit among employment with social security benefits.
Short term subsidy factor during temporal disability days / 0.60 / “The subsidy in money granted to the insured will be equal to sixty percent of the last daily wage of contribution.” / - / - / [16]
Disability pension factor during permanent disability days / 0.35 / “The amount of the invalidity pension will be equal to a basic amount of thirty-five percent of the average of the salaries corresponding to the last five hundred weeks of contribution prior to the granting of the same.” / - / - / [16]
Life pension (for premature death) factor / 0.90 / “The widow's pension will be equal to ninety percent of the amount that would have corresponded to the insured in the case of disability or that the pensioner had been enjoying” / - / - / [16]

Source: Authors’ elaboration with demographic, epidemiological, breastfeeding and economic information to estimate the microsimulation model.

Supplementary figure 1. Fertility rate among Mexican women in 2012

Source: Authors´ estimation with information from CONAPO 2012

Supplementary figure 2. Mortality rate without breast cancer deaths in Mexican women in 2012

Source: Authors´ estimation with information from CONAPO and SINAIS 2012

Supplementary figure 3. The incidence of breast cancer in 2012 by accumulated months of breastfeeding in Mexican women.

Source: Authors´ estimation with RRs information from Collaborative Group on Hormonal Factors 2002 and Globocan BC incidence 2012.

Supplementary figure 4. Breast cancer survival estimates in Mexican women by stage of disease.

Source: Authors´ estimation with information from Ángeles-Llerenas, et.al. Effect of care-delivery delays on the survival of Mexican women with breast cancer. SPM. 2016

Supplementary figure 5. Probability of employment among Mexican women in 2013.

Source: Authors´ elaboration with information from ENAOE 2013

Supplementary figure 6. Probability of salaried and non-salaried employment among employed Mexican women in 2013.

Source: Authors´ elaboration with information from ENAOE 2013

Supplementary figure 7. Probability of having social security benefits among salaried employed Mexican women in 2013.

Source: Authors´ elaboration with information from ENAOE 2013

Supplementary figure 8. Monthly income (2015 dollars) by age and type of employment. Mexican women.

Source: Authors´ elaboration with information from ENAOE 2013

Supplementary table 2. Sensitivity analyses scenarios of the lifetime potential reduction of the economic and disease burden of breast cancer associated with suboptimal breastfeeding in a cohort of Mexican women a,b. Simulation model#.

Sensitivity analysis scenario / Life-expectancy: simulated age of death from other causes than BC
(95% CI)
(A) / 0% time-discount rate
(95% CI)
(B) / 5% time-discount rate
(95% CI)
(C) / BC incidence increase 1.5% per year until year 2035
(95% CI)
(D) / RR for BC from a dose-response meta-analysis
(95% CI)
(E)
Disease burden
Expected lifetime cases of BC / 3,571***
(3,493 - 3,650) / 3,558***
(3,479, 3,637) / 3,595***
(3,514 - 3,675) / 3,248***
(3,175 - 3,321) / 4,546***
(4,464 - 4,629)
Years lived with disability (YLD) / 4,631***
(3,933 - 5,329) / 5,887***
(5,139 - 6,635) / 5,099***
(4,432 - 5,766) / 6,464***
(5,843 - 7,085) / 5,744***
(5,014 - 6,473)
Years of life lost (YLL) / 13,323***
(12,788 - 13,857) / 16,933***
(16,179 - 17,687) / 9,332***
(8,950 - 9,715) / 23,550***
(23,205 - 23,895) / 20,969***
(15,739-26,200)
DALYS (YLD+YLL) / 17,954***
(17,037 - 18,871) / 22,820***
(21,758 - 23,882) / 14,431***
(13,644 - 15,218) / 30,014***
(29,286 - 30,741) / 26,713***
(21,513-
31,914)
Economic costs (2015 million USD)
Medical direct costs / 69.82***
(64.94 - 74.69) / 223.33***
(209.04 - 237.63) / 39.80***
(37.09 - 42.51) / 369.82***
(366.67 - 372.97) / 67.84***
(61.11-74.56)
Non-medical direct costsd / 4.72***
(4.48 - 4.97) / 15.03***
(14.30 - 15.76) / 2.53***
(2.40 - 2.66) / 24.85***
(24.69 - 25.01) / 5.60***
(5.20-6.00)
Indirect costse / 14.02***
(12.16 - 15.88) / 43.04***
(37.21 - 48.87) / 6.75***
(5.90 - 7.59) / 55.87***
(55.02 - 56.72) / 14.06***
(10.55 - 17.57)
Morbidity costs / 5.56***
(4.81 - 6.31) / 16.47***
(14.29 - 18.64) / 3.05***
(2.68 - 3.42) / 23.45***
(23.08 - 23.82)
Mortality costs (premature death) / 8.46***
(7.31 - 9.61) / 26.57***
(22.79 - 30.36) / 3.70***
(3.20 - 4.19) / 32.42***
(31.91 - 32.93) / 14.65***
(10.95-18.35)
Costs by sector
Employer costsf / 2.65**
(0.45 - 4.85) / 6.32*
(-0.72, 13.37) / 1.26**
(0.27 - 2.26) / 7.50***
(6.48 - 8.52) / 4.13***
(1.89- 6.36)
Healthcare costsg / 69.82***
(64.94 - 74.69) / 223.33***
(209.04 - 237.63) / 39.80***
(37.09 - 42.51) / 369.82***
(366.67 - 372.97) / 67.84***
(61.11- 74.56)
Social costsh / 16.09***
(15.42 - 16.77) / 51.74***
(49.61 - 53.87) / 8.02***
(7.69 - 8.35) / 73.22***
(72.82 - 73.61) / 15.53***
(13.38- 17.68)
Total economic cost / 88.56***
(83.95 - 93.17) / 281.40***
(267.60 - 295.20) / 49.08***
(46.48 - 51.68) / 450.54***
(447.57 - 453.50) / 87.50***
(82.56 - 92.44)

Source: Authors´ elaboration with 1,000 simulations for each scenario. BC, Breast Cancer; BF, Breastfeeding; CI, Confidence Interval; DALYs, Disability-Adjusted Life Years. a.33% vs 95% of women BF duration per child between 12-36 months in Mexico in 2012. b. Multiplying by 11.16 thousand women of 15 years old in 2012 (CONAPO). d. Opportunity costs of the caregivers. e. Morbidity and mortality costs. f. Temporary and permanent subsidies due to morbidity and premature mortality from BC that are paid to employed women with social security benefits, according to the Social Security Law in Mexico. g. Equivalent to direct medical costs h. The cost of caregivers and indirect costs minus the costs faced by employers. *** p<0.001, ** p<0.05.