1) Weight, BMI and health behaviours amongst women joining the FHRPC, 1989–2018
At date of censor (30/04/19), the FHRPC database contained 15,102 subjects of which 10,920 were eligible for inclusion in the analyses (Fig. 1). Average age at FHRPC entry was consistent across the periods and 35.9% of women in the most recent period were ≤ 35 years (Table 1). Median BC risk was 33.3% (IQR 20.0-33.3) meaning that women had a median 1 in 3 lifetime chance of developing BC. The proportion of women from the two most deprived quintiles increased over time as did the proportion from non-white ethnicity groups.
Table 1: Baseline characteristics for full valid cohort and by period of FHRPC entry
|
Full cohort
|
1989-1998
|
1999-2008
|
2009-2018
|
Number of participants
|
10,920*
|
3,041
|
3,443
|
4,364
|
Age at FHRPC entry1
Missing
|
39.6 (9.2)
n=0
|
39.2 (8.9)
|
39.2 (9.0)
|
40.2 (9.4)
|
Age ≤35 years2
|
4040 (37.0%)
|
1174 (38.6%)
|
1276 (37.1%)
|
1566 (35.9%)
|
Estimated lifetime BC risk (%)3, 4
|
33.3 (20.0-33.3)
|
25.0 (16.7-33.3)
|
33.3 (25.0-33.3)
|
30.0 (23.3-33.3)
|
IMD quintile2
1 (most deprived)
2
3
4
5 (least deprived)
Missing
|
2125 (20.5%)
1856 (17.9%)
1772 (17.1%)
2101 (20.3%)
2490 (24.1%)
n=576
|
483 (17.6%)
447 (16.3%)
473 (17.2%)
622 (22.6%)
722 (26.3%)
n=294
|
675 (20.4%)
602 (18.2%)
585 (17.7%)
625 (18.9%)
931 (22.0%)
n=141
|
953 (22.6%)
797 (18.9%)
703 (16.6%)
841 (19.9%)
931 (22.0%)
n=139
|
IMD rank (lower = more deprived)3
Missing
|
17754 (8133-26077)
n=576
|
19304 (9812-26621)
n=294
|
17674 (8129-26227)
n=141
|
16441 (7529-25675)
n=139
|
Ethnicity
White
Asian or Asian British
Black, Black British, Caribbean or African
Mixed or multiple ethnic groups
Other ethnic group
Missing
|
6443 (94.1)
173 (2.5)
61 (0.9)
75 (1.1)
92 (1.3)
n=4076
|
2388 (98.5)
15 (0.6)
7 (0.3)
8 (0.3)
6 (0.2)
n=617
|
2290 (95.3)
38 (1.6)
21 (0.9)
18 (0.7)
36 (1.5)
n=1040
|
1732 (87.4)
119 (6.0)
33 (1.7)
48 (2.4)
50 (2.5)
n=2382
|
Height (m)1
Missing
|
1.64 (0.07)
n=2,596
|
1.63 (0.07)
n=190
|
1.64 (0.07)
n=669
|
1.65 (0.07)
n=1,706
|
Weight (kg)3
Missing
|
64.4 (57.6-74.4)
n=3,656
|
64.8 (57.2-69.9)
n=414
|
68.4 (58.1-75.3)
n=1,281
|
70.7 (59.0-78.9)
n=1,923
|
BMI (kg/m2)3
Missing
|
24.0 (21.6-27.6)
n=3,688
|
23.4 (21.5-26.3)
n=420
|
24.3 (21.9-27.9)
n=1,290
|
24.7 (22.0-28.9)
n=1,939
|
BMI categories2:
Underweight (<18.5 kg/m2)
Healthy weight (18.5-24.9 kg/m2)
Overweight (25-29.9 kg/m2)
Obese (≥30.0 kg/m2)
Missing
|
165 (2.3%)
4080 (56.4%)
1851 (25.6%)
1136 (15.7%)
n=3,688
|
59 (2.3%)
1664 (63.5%)
621 (23.7%)
277 (10.6%)
n=420
|
39 (1.8%)
1181 (54.9%)
577 (26.8%)
356 (16.5%)
n=1,290
|
66 (2.7%)
1214 (50.1%)
647 (26.7%)
498 (20.5%)
n=1,939
|
Alcohol categories2:
Non-drinker
Low risk (≤14 units/wk)
Hazardous (14.1-35 units/wk)
Harmful (≥35 units/wk)
Missing
|
3034 (37.7%)
3572 (44.4%)
1249 (15.5%)
184 (2.3%)
n=2,881
|
864 (30.7%)
1384 (49.2%)
501 (17.8%)
66 (2.3%)
n= 226
|
1113 (42.8%)
984 (37.8%)
431 (16.6%)
73 (2.8%)
n=842
|
1039 (40.2%)
1187 (45.9%)
313 (12.1%)
45 (1.7%)
n=1780
|
Alcohol units per week (excluding categories of non-drinkers and missing)3
|
9.0 (4.0-16.0)
|
9.0 (5.0-16.0)
|
10.0 (6.0-18.0)
|
8.0 (4.0-14.0)
|
Smoking status2:
Non-smoker
Former smoker
Current smoker
Missing
|
3611 (53.0%)
1470 (21.6%)
1732 (25.4%)
n=4,107
|
1405 (52.4%)
379 (14.1%)
899 (33.5%)
n=358
|
997 (51.0%)
478 (24.4%)
481 (24.6%)
n=1487
|
1190 (55.6%)
607 (28.4%)
344 (16.1%)
n=2223
|
Pack-years (former and current smokers only)3
Missing
|
7.5 (2.5-15.0)
n=69
|
9.8 (3.5-18.0)
n=27
|
7.5 (2.5-15.0)
n=18
|
5.2 (2.3-12.4)
n=24
|
1 mean (SD), 2 n (%), 3 median (IQR: 25th and 75th percentiles), 4 based on a modification of the Claus tables until 2003 and the Tyrer-Cuzick model thereafter
*Total is greater than the sum of joining periods as n=25 women joined 1987-88, and n=47 joined in the first part of 2019
Height, weight and BMI are higher in the more recent entry periods (Table 1). The prevalence of obesity has nearly doubled with an increase from 10.6% in the 1989-98 period to 20.5% in the 2009-18 period. Alcohol intake above the UK recommended maximum of 14 units per week reduced from 20.1% in the first period to 13.8% in the most recent period, and current smoking from 33.5–16.1%.
LOESS curves for BMI at FHRPC entry (Figs. 2a-c) show the increase in BMI with age of FHRPC entry up to around age 53 after which it levels off and shows a slight decrease in older women (Fig. 2a), and an increase over time with some levelling off and a slight decrease from around 2010 (Fig. 2b). There was no association with birth year (cohort, Fig. 2c). Linear analysis in Table 2 revealed a statistically significant relationship for all three variables. Women who were older when they joined the FHRPC had a higher BMI than younger women (0.027 BMI units/y, 95% CI 0.017–0.037), BMI increased by 0.049 units for each year of entry (95% CI 0.039–0.059), and there was a higher BMI in women who were born earlier (0.022 BMI units/y, 95% CI 0.014–0.031).
Table 2
Output from linear partial least squares analysis for age, period and cohort effects
Variables
|
BMI (kg/m2 per year)
Coefficient (95%CI)
|
Alcohol (average weekly units per year)
Coefficient (95%CI)
|
Smoking (average weekly cigarettes per year)
Coefficient (95%CI)
|
Age
|
0.027 (0.017 to 0.037)
|
-0.459 (-0.576 to -0.343)
|
0.623 (0.476 to 0.783)
|
Year of FHRPC entry (Period)
|
0.049 (0.039 to 0.059)
|
-0.751 (-0.867 to -0.635)
|
-0.783 (-0.873 to -0.693)
|
Birth year (Cohort)
|
0.022 (0.014 to 0.031)
|
-0.191 (-0.258 to -0.123)
|
-1.021 (-1.165 to -0.875)
|
Alcohol intake at entry is associated with age and year of entry (Fig. 2d-f). Intake remains steady until around age 45 then decreases with an overall decrease of 0.459 (95% CI -0.576 to -0.343) weekly units per year of age. There was an increase in alcohol intake to around 1996 then a steady decrease, with an overall decrease of 0.751 (95% CI -0.867 to -0.635) weekly units per year over the analysis period. Linear analysis also showed a decrease in alcohol consumption in women with a later birth year (-0.191, 95% CI -0.258 to -0.123).
Weekly cigarette smoking at entry is also associated with all three variables of age, and year of entry, and birth year (Fig. 2g-i). Number of weekly cigarettes smoked increased to around age 45 then decreased with an overall increase over the study period of 0.623 (95% CI 0.476–0.783) weekly cigarettes per year of age. There are decreases in weekly cigarettes smoked with both more recent year of entry (-0.783/year, 95% CI -0.873 to -0.693), and more recent birth year (-1.021/year, 95% CI -1.165 to -0.875).
2a) Changes in weight and BMI after joining the FHRPC
There was a 55.3% response rate to the FHRisk questionnaire with n=3,283 valid responses (Figure 1). Compared with the rest of the eligible FHRPC population, FHRisk responders were marginally older and had higher BC risk (Supplementary Table 2). They were also more affluent, more likely to be of white ethnicity, had a lower median BMI at FHRPC entry and a lower proportion had obesity, were more likely to be a non-drinker but had a higher weekly alcohol intake amongst those who consumed alcohol.
After FHRPC entry, women’s weight and BMI increased (Table 3). Amongst women returning a valid FHRisk questionnaire, median weight at FHRPC entry (mean age 40.2 ±8.7 years) was 63.5 (IQR 57.2-72.6) kg and 38.1% had overweight or obesity, which had increased to 67.6 (60.3-77.6) kg and 51.8% by mean age 51.0 (±10.5) years.
Twenty-eight percent of women who were a healthy BMI at FHRPC entry had increased to overweight or obese by FHRisk completion, and 30% of those with overweight at FHRPC entry had developed obesity by FHRisk completion. However, less than 12% of those with overweight or obesity at FHRPC entry had shifted into a lower category by FHRisk completion (Supplementary Table 3a). Women gained 4.5 (IQR 0.0-12.5) % weight between FHRPC entry and FHRisk completion and weight gain per year was 0.25 (0.00-0.68) kg (Supplementary Table 4).
Analysis using ANCOVAs with multiple imputation (Table 4, with complete case analysis in Supplementary Table 5) revealed no significant differences in weight change after FHRPC entry between the three joining periods, after adjusting for time between Clinic entry and FHRisk completion.
2b) Changes in weight and BMI during adulthood from age 20
Over 30% of women with a healthy BMI at age 20 had shifted into the overweight or obese categories by the time of FHRPC entry (Supplementary Table 3b). Only 18% of women who were overweight at age 20 had shifted down into the healthy weight category by the time of FHRPC entry. The uplift in BMI category between age 20 and FHRisk completion was even more pronounced. Forty-seven percent of women with a healthy BMI at age 20 had developed overweight or obesity by the time of the FHRisk study, and 58% of those with overweight had developed obesity (Supplementary Table 3c). Women gained median 9.7 (IQR 1.4-20.6) % weight during the 20.1 (±8.5) years from age 20 and FHRPC entry, or 0.28 (0.04-0.60) kg per year (Supplementary Table 4). This had increased to 15.8 (IQR 6.2-28.7) % weight by the time of FHRisk. Analysis using ANCOVAs with multiple imputation (Table 4, with complete case analysis in Supplementary Table 5) adjusted for duration between timepoints revealed that the most recent joining period had gained more weight between age 20 and joining the FHRPC than the previous two periods (9.5 [8.3-10.6] kg for 2009-2018 period vs. 8.0 [7.4-8.7] kg [p=0.03] and 6.0 [5.5-6.6] kg [p<0.001] for 1999-2008 and 1989-1998 periods respectively). There were no significant differences in patterns of weight change between the three FHRPC entry periods between the other time points: age 20 or FHRPC entry to FHRisk completion.
Data from participants with BMI at all three time points (n=2,075, Supplementary Table 6) shows that prevalence of obesity trebled between age 20 and FHRPC entry (mean age 39.6 [SD 8.8], 3.6% to 12.1%) and increased again to FHRisk completion (mean age 51.4 [SD 10.4], 20.2%).
3) Changes in alcohol consumption after joining the FHRPC
Between FHRPC entry and completion of FHRisk there was a small increase in median weekly alcohol. The proportion of non-drinkers reduced from 42.5% to 24.6% and the number of women reporting drinking within recommended levels increased from 39.8% to 61.0% (Table 3). The proportion of women drinking at hazardous or harmful levels (>14 units per week) also reduced between joining FHRPC and completion of FHRisk, however, 14.4% of FHRisk responders still consumed alcohol above the recommended maximum level.