Abstract

High consumption of coffee has been suggested to reduce the risk of some late-onset diseases and death but also to contribute to the development of osteoporotic fractures. Results of previous fracture studies have been inconsistent, and a comprehensive study is needed. The longitudinal population-based Swedish Mammography Cohort, including 61,433 women born in 1914–1948, was followed up from 1987 through 2008. Coffee consumption was assessed with repeated food frequency questionnaires. During follow-up, 14,738 women experienced fracture of any type, and 3,871 had a hip fracture. In a subcohort (n = 5,022), bone density was measured and osteoporosis determined (n = 1,012). After multivariable adjustment, there was no evidence of a higher rate of any fracture (hazard ratio per 200 mL coffee = 0.99; 95% confidence interval: 0.98, 1.00) or hip fracture (hazard ratio per 200 mL coffee = 0.97, 95% confidence interval: 0.95, 1.00) with increasing coffee consumption. A high coffee intake (≥4 cups daily) versus a low intake (<1 cup daily) was associated with a 2%–4% lower bone density, depending on site (P < 0.001), but the odds ratio for osteoporosis was only 1.28 (95% confidence interval: 0.88, 1.87). Thus, high coffee consumption was associated with a small reduction in bone density that did not translate into an increased risk of fracture.

Coffee is widely consumed, especially in the United States and Europe. Health benefits associated with coffee consumption, including reduced risk of death, have recently been recognized (1). Negative effects have also been suggested, such as an increased risk of hip fractures (2). These fractures constitute a large and growing problem worldwide in both women and men, with a profound impact on quality of life (3) and mortality rate (4). In the causation of osteoporosis and fractures, several dietary factors have been implicated, including the potential negative effects of caffeine-containing beverages (5). High intakes of coffee and caffeine have been associated with increased risk of fractures in some observational studies of women (6–11), including our own (12), but not in others (13–19). In recent in vitro studies, it has been suggested that caffeine has direct or indirect deleterious effects on osteoblasts (20, 21). Caffeine might contribute to loss of bone by increased urinary calcium excretion (22) and a decreased intestinal absorption efficiency of calcium (23). In addition, some authors have noted that the association between high caffeine intake and low bone mineral density (BMD) is demonstrated only in women with a concomitant low calcium intake (24, 25).

Because the conceivable effect of high coffee consumption on fracture risk and bone density is probably modest, large, long-term studies are needed. Sweden is one of the world's leading nations in terms of both coffee consumption per capita and osteoporotic fracture incidence (26, 27). Furthermore, Sweden is well suited for population-based studies because of its renowned quality of official health registers and traceability of persons. The aim of the present study was to extend our first analysis, which was based on a subset of the Swedish Mammography Cohort (12), to determine whether high coffee consumption is associated with risk of fractures. We have now used the whole cohort, a better case-ascertainment method, repeated assessment of coffee intake for better precision, and a longer follow-up period with more fracture and hip fracture cases. In addition, the association between coffee consumption and osteoporosis was investigated in the Swedish Mammography Cohort Clinical (SMCC).

MATERIALS AND METHODS

Study population

The Swedish Mammography Cohort was initiated in 1987–1990 (28). At that time, all women (n = 90,303) who lived in the counties of Uppsala and Västmanland in central Sweden and had been born between 1914 and 1948 received an invitation by mail to participate in a mammography screening program. A food frequency questionnaire (FFQ) covering diet and lifestyle was enclosed with the invitation. The questionnaire was completed by 66,651 women (74%). A second, expanded questionnaire was sent in 1997 to all those who were still living in the study area. The response rate to the follow-up questionnaire was 70%. In the present fracture study, data for 61,433 participants at baseline and 38,984 at follow-up in 1997 were available for analysis.

The SMCC is a randomly selected subsample of the Swedish Mammography Cohort sample (28) of women living in the city of Uppsala. These women were invited to respond to a third questionnaire together with a clinical investigation that included dual-emission x-ray absorptiometry measurements between November 2003 and October 2009. A total of 5,022 women participated in the SMCC (participation rate 65%). A flow chart describing the study sample is included (Figure 1). The regional ethics committees at Uppsala University, Uppsala, Sweden, and Karolinska Institutet, Stockholm, Sweden, approved the studies.

Figure 1.

A flow chart describing the study samples in the Swedish Mammography Cohort (counties of Uppsala and Västmanland, Sweden, 1987–2008) and the 3 food frequency questionnaires (food frequency questionnaire 1, food frequency questionnaire 2, and food frequency questionnaire 3). Excluded were those with an erroneous personal identification number, a questionnaire that was not dated, erroneous dates of moving out of the study area or death, implausible energy intakes (±3 standard deviations from the mean value of the log-transformed reported energy intake), or a cancer diagnosis (except for nonmelanoma skin cancer and only before the baseline questionnaire).

Figure 1.

A flow chart describing the study samples in the Swedish Mammography Cohort (counties of Uppsala and Västmanland, Sweden, 1987–2008) and the 3 food frequency questionnaires (food frequency questionnaire 1, food frequency questionnaire 2, and food frequency questionnaire 3). Excluded were those with an erroneous personal identification number, a questionnaire that was not dated, erroneous dates of moving out of the study area or death, implausible energy intakes (±3 standard deviations from the mean value of the log-transformed reported energy intake), or a cancer diagnosis (except for nonmelanoma skin cancer and only before the baseline questionnaire).

Dietary assessment

The 3 self-administered FFQs have been described previously (12, 28, 29). Briefly, intakes of nutrients were estimated by multiplying the nutrient content (30) of age-specific portion sizes by the consumption frequency of each food item. Nutrient intakes were adjusted for total energy intake (1,700 kcal, which was the average intake in the study population) with the residual method (31).

Coffee consumption was assessed in all 3 FFQs. In the baseline FFQ, the participants were asked how often, on average, during the previous 6 months they had consumed coffee, black tea, and other foods and beverages according to 8 predefined categories. In the second and third FFQs, the participants were asked open questions on how many cups of coffee they had been drinking per day or week during the previous year. According to a validation study, the correlation coefficient between coffee consumption in the first FFQ and the mean of four 1-week weighed food records was 0.6 (Alicja Wolk, Karolinska Institutet, unpublished data). Coffee consumption was categorized as <1, 1, 2–3, or ≥4 cups daily. One cup of coffee is estimated to contain on average 177 mL. We also calculated an approximate continuous variable for coffee by consecutive integers, each corresponding to a 200-mL increase of coffee intake. It should be noted that the consumption of decaffeinated coffee and tea was very low in Sweden (32).

Outcomes

Our main outcomes, clinical fractures of any type (International Classification of Diseases, 10th Revision, diagnosis codes S12, S22, S32, S42, S52, S62, S72, S82, or S92) and hip fractures (International Classification of Diseases, 10th Revision, codes S720, S721, or S722), were identified from the Swedish National Patient Registry (33) and from local hospital registers from the date of cohort entry in 1987–1990 through December 31, 2008. Individual matching of fractures to the study participants enabled complete fracture identification. Re-admissions from a previous fracture event were separated from incident fracture admissions by a previously validated and accurate method (34). Fractures caused by high-energy trauma (approximately 1% of all fractures) were retained as outcomes in the analysis because the association between this type of fracture and BMD has a strength similar to that between low-energy trauma fractures and BMD (35). Pathological fractures caused by malignant disease were not considered as an outcome.

Total body, proximal femur, and lumbar spine BMD (in g/cm2) measured by dual-emission x-ray absorptiometry (Lunar Prodigy, GE Healthcare Lunar, Madison, Wisconsin) and osteoporosis were regarded as secondary outcomes in the SMCC subcohort. Osteoporosis was defined as a BMD 2.5 standard deviations or more below the mean of a young adult reference range (36). The precision error for BMD, based on triple measurements in 15 subjects, was 0.8%–1.5% depending on site. The long-term precision coefficient of variation was less than 1%. We also studied potential associations between coffee consumption and fall frequency (at least 1 fall or at least 2 falls during the previous year) as reported in the SMCC questionnaire.

Comorbidity and lifestyle information

The Charlson weighted comorbidity index (37, 38) was calculated on the basis of diagnostic codes from the Swedish National Patient Registry, which includes diagnoses from 1963 onward. Lifestyle information was obtained from the questionnaires. All questionnaires included weight and height, parity, living conditions (categorized as living alone; yes or no), and educational level (<9 years, 9–12 years, >12 years, and other education, such as vocational). The second questionnaire and the third (SMCC) questionnaire also included information on postmenopausal estrogen therapy (ever use), cortisone use, supplementation with calcium and vitamin D, smoking habits (current, former, or never), and leisure-time physical activity during the past year (with 5 predefined levels, from 1 hour per week to >5 hours per week). In comparison with activity records and accelerometer data, physical activity data collected in the questionnaire from 1997 are valid (39).

Determination of serum 25-hydroxyvitamin D

Serum 25-hydroxyvitamin D levels were measured in samples from 5,000 of the women in SMCC through the use of high-pressure liquid chromatography interfaced by atmospheric pressure chemical ionization and tandem mass spectrometry (Vitas AS, Oslo, Norway). The interassay coefficients of variation were 9.67% and 10.15% at 36.94 nmol/L and 48.08 nmol/L, respectively. We used the sum of 25-hydroxyvitamin D3 and 25-hydroxyvitamin D2 as an estimate of total 25-hydroxyvitamin D in serum.

Statistical analyses

We used survival analysis for the study of associations between coffee consumption and fracture risk. Time at risk for each participant was calculated from the date of the baseline questionnaire until the dates of first fracture at any site and first hip fracture, date of death, date of emigration, or the end of the study period (December 31, 2008), whichever came first. Time to second fracture (any or hip) was also analyzed. Age-adjusted failure curves set at age 70 years illustrate fracture incidence for each category of coffee intake. Cox proportional hazards models were used to estimate age- and multivariable-adjusted hazard ratios with 95% confidence intervals. The multivariable model included continuous variables of age, body mass index, height, total energy intake, consumption of alcohol, and dietary intakes of calcium, vitamin D, retinol, protein, phosphorous, and potassium; as well as calcium supplementation (yes or no), vitamin D supplementation (yes or no), tea consumption (number of cups per day), educational level (<9, 9–12, >12 years, or other), physical activity (5 categories), smoking status (never, former, or current), previous fractures before the study period (yes or no), Charlson comorbidity index (continuous, 1–16), living condition (living alone or not), cortisone use (yes or no), and hormone replacement therapy (yes or no).

Individuals with missing values for coffee consumption (1%) were considered nonconsumers and were categorized as consuming less than 1 cup of coffee per day (i.e., the reference category). We performed multiple imputations by using the Markov chain Monte Carlo method to construct baseline values for several potentially important covariates (smoking status, physical activity, supplementation with calcium and vitamin D, cortisone use, and hormone replacement therapy) that were not assessed at baseline. The values of these covariates were imputed for the women who did not participate in the 1997 questionnaire survey because of death or other causes (36%).

The proportionality assumption was verified with log-log plots and formally by use of Shoenfeld residuals. There was a tendency toward nonproportionality for the association between coffee consumption and hip fracture. However, sensitivity analysis by logistic regression showed estimates similar to the Cox models, indicating no large violation of the proportionality assumption (results not shown). Coffee consumption and covariates assessed at more than 1 point in time, including the Charlson comorbidity index, were treated as time-updated variables to account for changes during follow-up.

Logistic regression was used to assess the associations between coffee consumption and the risk of falls and the risk of osteoporosis in the SMCC. We used a multivariable model (described above) modified by the addition of 2 variables: year of birth (replaced age) and current use of bisphoshphonates. We further calculated adjusted means of BMD for categories of coffee consumption by using a general linear model. In the SMCC, coffee consumption and other dietary variables in the model were based on cumulative averages (31) from the 3 dietary questionnaires. Information on the remaining variables was collected from the most recent questionnaire.

We investigated potential effect modification by calcium on our estimates by calculating the relative excess risk that was due to interaction (40). Low calcium intake, defined as an intake less than 700 mg per day, has been found to be associated with an increased rate of fractures in our cohort (28). Moreover, we investigated whether a low total serum 25-hydroxyvitamin D concentration (<50 nmol/L) according to the Institute of Medicine guidelines (41) modified the association between coffee intake and BMD. These analyses were also adjusted for season (January–March, April–June, July–September, and October–December) to take into account the seasonal variation in serum 25-hydroxyvitamin D concentrations. All statistical analyses were performed in Stata, version 11, software (StataCorp LP, College Station, Texas).

RESULTS

Coffee and fracture risk

Descriptive characteristics of the cohorts are listed in Table 1 and in Appendix Table 1. About 60% of the participants reported a daily consumption of 2–3 cups of coffee. During a median of 19.4 years of follow-up and a total of 1,051,278 person-years, 14,738 women (24% of the cohort) experienced a first fracture at any site. Included in the total number of first fractures were 3,871 fractures of the hip. Age-adjusted incidence proportions of fractures (any and hip fractures) during follow-up by each coffee consumption category are displayed in Figure 2. After multivariable adjustment, there was no association between increasing coffee consumption and rate of any fracture (hazard ratio (HR) = 0.99; 95% confidence interval (CI): 0.98, 1.00) or hip fracture (HR = 0.97; 95% CI: 0.95, 1.00) per 200 mL of coffee (Table 2). Women consuming 4 or more cups per day had hazard ratios of 0.96 (95% CI: 0.90, 1.02) for any type of fracture and 0.88 (95% CI: 0.78, 1.00) for hip fracture, as compared with nonconsumers of coffee (<1 cup/day). We also examined associations with an even higher level of coffee consumption (based on the second FFQ). Compared with an intake level of less than 1 cup/day, consumption of 8 or more cups/day was not associated with a higher rate of any fracture (HR = 1.20; 95% CI: 0.82, 1.75) or of hip fracture (HR = 0.95; 95% CI: 0.80, 1.14).

Table 1.

Baseline Characteristics of Women in the Swedish Mammography Cohort, Counties of Uppsala and Västmanland, Sweden, 1987–1990

  No. of Cups of Coffee Per Day
 
<1
 
1
 
2–3
 
≥4
 
No. Mean (SD) No. Mean (SD) No. Mean (SD) No. Mean (SD) 
Number of women 5,534 9.0  8,272 13.5  36,557 59.5  11,070 18.0  
Age at entry, years   54.0 (10.0)   56.7 (10.1)   54.2 (9.7)   49.8 (8.1) 
Body mass index at entrya   24.9 (5.5)   25.0 (4.2)   24.8 (4.1)   24.7 (4.6) 
Average intake per dayb             
 Energy, kcal   1,506 (488)   1,513 (457)   1,589 (445)   1,659 (506) 
 Calcium, mg   889 (294)   901 (259)   911 (246)   926 (265) 
 Supplemental calcium, mgc   296 (447)   270 (360)   252 (349)   248 (348) 
 Vitamin D, μg   4.24 (1.52)   4.39 (1.41)   4.43 (1.34)   4.45 (1.45) 
 Alcohol, g   2.01 (3.60)   2.32 (3.55)   2.56 (3.40)   2.90 (3.76) 
Median intake per dayd             
 Coffee, g   0 (69)   177 (28)   442 (38)   768 (60) 
 Tea, g   181 (555)   222 (239)   34 (222)   17 (91) 
Lifestyle factors and medications             
 Leisure-time physical activity levele             
  1 (lowest) 754 22.1  792 20.7  3,094 18.8  2,251 20.7  
  2 773 22.7  908 23.7  3,942 23.9  2,527 23.2  
  3 1,075 31.6  1,281 33.5  5,667 34.4  3,582 32.9  
  4 405 11.9  451 11.8  2,001 12.1  1,236 11.4  
  5 (highest) 399 11.7  392 10.3  1,784 10.8  1,295 11.9  
 Smoking statuse             
  Current 636 16.6  654 15.5  3,546 19.7  3,995 33.2  
  Former 842 21.9  1,014 24.1  4,217 23.4  2,655 22.0  
  Never 2,360 61.5  2,546 60.4  10,282 57.0  5,396 44.8  
 Two or more Charlson comorbidities 182 3.3  248 3.0  720 2.0  167 1.5  
 Educational level <9 years 4,197 77.9  6,594 81.3  28,891 80.4  8,674 79.6  
 Fracture before baseline 271 4.9  445 5.4  1,722 4.7  415 3.8  
 Hormone replacement therapy use, evere 1,817 44.9  2,085 48.5  8,441 45.9  5,243 42.8  
 Cortisone usee 349 13.0  381 12.9  1,369 10.7  851 9.9  
 Calcium supplement usec,e 317 7.8  338 7.9  1,232 6.7  683 5.6  
 Marital status: single 1,577 28.8  2,375 29.0  8,035 22.2  2,445 22.3  
 Nulliparity 770 13.9  1,090 13.2  3,913 10.7  925 8.4  
  No. of Cups of Coffee Per Day
 
<1
 
1
 
2–3
 
≥4
 
No. Mean (SD) No. Mean (SD) No. Mean (SD) No. Mean (SD) 
Number of women 5,534 9.0  8,272 13.5  36,557 59.5  11,070 18.0  
Age at entry, years   54.0 (10.0)   56.7 (10.1)   54.2 (9.7)   49.8 (8.1) 
Body mass index at entrya   24.9 (5.5)   25.0 (4.2)   24.8 (4.1)   24.7 (4.6) 
Average intake per dayb             
 Energy, kcal   1,506 (488)   1,513 (457)   1,589 (445)   1,659 (506) 
 Calcium, mg   889 (294)   901 (259)   911 (246)   926 (265) 
 Supplemental calcium, mgc   296 (447)   270 (360)   252 (349)   248 (348) 
 Vitamin D, μg   4.24 (1.52)   4.39 (1.41)   4.43 (1.34)   4.45 (1.45) 
 Alcohol, g   2.01 (3.60)   2.32 (3.55)   2.56 (3.40)   2.90 (3.76) 
Median intake per dayd             
 Coffee, g   0 (69)   177 (28)   442 (38)   768 (60) 
 Tea, g   181 (555)   222 (239)   34 (222)   17 (91) 
Lifestyle factors and medications             
 Leisure-time physical activity levele             
  1 (lowest) 754 22.1  792 20.7  3,094 18.8  2,251 20.7  
  2 773 22.7  908 23.7  3,942 23.9  2,527 23.2  
  3 1,075 31.6  1,281 33.5  5,667 34.4  3,582 32.9  
  4 405 11.9  451 11.8  2,001 12.1  1,236 11.4  
  5 (highest) 399 11.7  392 10.3  1,784 10.8  1,295 11.9  
 Smoking statuse             
  Current 636 16.6  654 15.5  3,546 19.7  3,995 33.2  
  Former 842 21.9  1,014 24.1  4,217 23.4  2,655 22.0  
  Never 2,360 61.5  2,546 60.4  10,282 57.0  5,396 44.8  
 Two or more Charlson comorbidities 182 3.3  248 3.0  720 2.0  167 1.5  
 Educational level <9 years 4,197 77.9  6,594 81.3  28,891 80.4  8,674 79.6  
 Fracture before baseline 271 4.9  445 5.4  1,722 4.7  415 3.8  
 Hormone replacement therapy use, evere 1,817 44.9  2,085 48.5  8,441 45.9  5,243 42.8  
 Cortisone usee 349 13.0  381 12.9  1,369 10.7  851 9.9  
 Calcium supplement usec,e 317 7.8  338 7.9  1,232 6.7  683 5.6  
 Marital status: single 1,577 28.8  2,375 29.0  8,035 22.2  2,445 22.3  
 Nulliparity 770 13.9  1,090 13.2  3,913 10.7  925 8.4  

Abbreviation: SD, standard deviation.

a Weight (kg)/height (m)2.

b Energy-adjusted average nutrient data were estimated with data from the baseline questionnaire and the 1997 questionnaire.

c 10,055 subjects answered the question on calcium supplement use.

d Median intake per day (interquartile range)—estimated with data from the baseline questionnaire and the 1997 questionnaire.

e Information available from only the 1997 questionnaire.

Table 2.

Coffee Consumption and Rate of Fracture of Any Type and of the Hip Among 61,433 Women in the Swedish Mammography Cohort, Counties of Uppsala and Västmanland, Sweden, 1987–2008

Exposure, Cups of Coffee/dayb No. of Fractures Person-years at Risk Rate per 1,000 Person-years 95% CI Adjusted for Age
 
Adjusteda
 
HR 95% CI HR 95% CI 
Any fracture         
 <1 1,522 98,841 15.4 14.6, 16.2 1.00  1.00  
 1 1,975 129,595 15.2 14.6, 15.9 0.98 0.91, 1.04 0.96 0.90, 1.03 
 2–3 7,817 575,105 13.6 13.3, 13.9 0.94 0.89, 0.99 0.95 0.89, 1.00 
 ≥4 3,424 247,738 13.8 13.4, 14.3 0.92 0.87, 0.98 0.96 0.90, 1.02 
Hip fracture         
 <1 448 106,180 4.2 3.8, 4.6 1.00  1.00  
 1 561 140,590 4.0 3.7, 4.3 0.89 0.79, 1.01 0.87 0.76, 0.98 
 2–3 2,106 621,160 3.4 3.2, 3.5 0.89 0.81, 0.99 0.90 0.81, 1.00 
 ≥4 756 263,680 2.9 2.7, 3.1 0.80 0.71, 0.90 0.88 0.78, 1.00 
Exposure, Cups of Coffee/dayb No. of Fractures Person-years at Risk Rate per 1,000 Person-years 95% CI Adjusted for Age
 
Adjusteda
 
HR 95% CI HR 95% CI 
Any fracture         
 <1 1,522 98,841 15.4 14.6, 16.2 1.00  1.00  
 1 1,975 129,595 15.2 14.6, 15.9 0.98 0.91, 1.04 0.96 0.90, 1.03 
 2–3 7,817 575,105 13.6 13.3, 13.9 0.94 0.89, 0.99 0.95 0.89, 1.00 
 ≥4 3,424 247,738 13.8 13.4, 14.3 0.92 0.87, 0.98 0.96 0.90, 1.02 
Hip fracture         
 <1 448 106,180 4.2 3.8, 4.6 1.00  1.00  
 1 561 140,590 4.0 3.7, 4.3 0.89 0.79, 1.01 0.87 0.76, 0.98 
 2–3 2,106 621,160 3.4 3.2, 3.5 0.89 0.81, 0.99 0.90 0.81, 1.00 
 ≥4 756 263,680 2.9 2.7, 3.1 0.80 0.71, 0.90 0.88 0.78, 1.00 

Abbreviations: CI, confidence interval; HR, hazard ratio.

a Covariates included were age, body mass index, height, total energy intake, and dietary intakes of calcium, vitamin D, retinol, protein, phosphorous, potassium, and alcohol (all continuous); as well as vitamin D supplementation, tea consumption (number of cups per day), educational level, physical activity level, smoking status, previous fracture, Charlson comorbidity index, living condition (living alone or not), nulliparity, cortisone use, and hormone replacement therapy.

b Coffee consumption was treated as a time-updated variable based on the baseline and 1997 food frequency questionnaires.

Figure 2.

Age-adjusted incidences of any fracture (A) and hip fracture (B) in the Swedish Mammography Cohort (counties of Uppsala and Västmanland, Sweden, 1987–2008) in relation to follow-up time displayed as Kaplan-Meier failure curves for the 4 categories of coffee consumption (<1, 1, 2–3, or ≥4 cups per day). Numbers at risk during follow-up are presented in Appendix Table 2.

Figure 2.

Age-adjusted incidences of any fracture (A) and hip fracture (B) in the Swedish Mammography Cohort (counties of Uppsala and Västmanland, Sweden, 1987–2008) in relation to follow-up time displayed as Kaplan-Meier failure curves for the 4 categories of coffee consumption (<1, 1, 2–3, or ≥4 cups per day). Numbers at risk during follow-up are presented in Appendix Table 2.

We found no evidence of an interaction between coffee consumption and calcium intake on the risk of fracture. The relative excess risk due to the interaction between coffee consumption and calcium intake was 0.030 (95% CI: −0.103, 0.164) for any fracture and −0.040 (95% CI: −0.312, 0.232) for hip fracture. Coffee consumption was not associated with an increased rate of 2 fractures of any type (4,971 fractures; HR = 1.00; 95% CI: 0.97, 1.02) or of the hip (1,355 hip fractures; HR = 0.96; 95% CI: 0.91, 1.00).

Relation between coffee consumption and BMD

In the SMCC (n = 5,022), 20% (n = 1,012) of subjects were diagnosed as having osteoporosis. The adjusted odds ratio for osteoporosis per 200 mL of coffee per day was 1.06 (95% CI: 0.98, 1.15). When high consumers (≥4 cups) were compared with nonconsumers, the multivariable-adjusted odds ratio for osteoporosis was 1.28 (95% CI: 0.88, 1.87) (Table 3).

Table 3.

Coffee Consumption and Odds Ratio of Osteoporosis Among 5,022 Women in the Swedish Mammography Cohort Clinical, Counties of Uppsala and Västmanland, Sweden, 2008–2009

Exposure, Cups of Coffee/Dayb No. of Women
 
Women With Osteoporosis
 
Adjusted for Age
 
Adjusteda
 
No. No. OR 95% CI OR 95% CI 
<1 375 7.5 72 19.2 1.00  1.00  
881 17.5 160 18.2 0.83 0.60, 1.14 0.95 0.67, 1.34 
2–3 3,009 59.9 631 21.0 1.10 0.83, 1.45 1.23 0.89, 1.70 
≥4 757 15.1 149 19.7 1.17 0.85, 1.62 1.28 0.88, 1.87 
Exposure, Cups of Coffee/Dayb No. of Women
 
Women With Osteoporosis
 
Adjusted for Age
 
Adjusteda
 
No. No. OR 95% CI OR 95% CI 
<1 375 7.5 72 19.2 1.00  1.00  
881 17.5 160 18.2 0.83 0.60, 1.14 0.95 0.67, 1.34 
2–3 3,009 59.9 631 21.0 1.10 0.83, 1.45 1.23 0.89, 1.70 
≥4 757 15.1 149 19.7 1.17 0.85, 1.62 1.28 0.88, 1.87 

Abbreviations: CI, confidence interval; OR, odds ratio.

a Odds ratios were adjusted for year of birth, body mass index, height, total energy intake, and dietary intakes of calcium, vitamin D, retinol, protein, phosphorous, potassium, and alcohol (all continuous); as well as vitamin D supplementation, tea consumption (number of cups per day), educational level, physical activity level, smoking status, previous fracture, Charlson comorbidity index, living condition (living alone or not), nulliparity, cortisone use, hormone replacement therapy, and use of bisphoshphonates.

b Coffee consumption was based on the cumulative average from all 3 food frequency questionnaires.

A high level of coffee consumption was associated with modestly lower BMD at all measured sites (Figure 3). Women consuming 4 or more cups versus less than 1 cup per day had an average of 4% (P < 0.001) lower BMD of the lumbar spine (difference = 0.048 g/cm2; 95% CI: 0.047, 0.050) and 2% (P < 0.001) lower BMD of both the proximal femur (difference = 0.022 g/cm2; 95% CI: 0.021, 0.023) and of the total body (difference = 0.018 g/cm2; 95% CI: 0.017, 0.019). These differences were smaller among women with serum 25-hydroxyvitamin D levels lower than 50 nmol/L and larger among women with levels of 50 nmol/L or higher. Thus, the differences in BMD between women consuming 4 or more cups and those consuming less than 1 cup were 0.020 g/cm2 (95% CI: 0.016, 0.024) at the lumbar spine, 0.005 g/cm2 (95% CI: 0.002, 0.007) at the proximal femur, and 0.007 g/cm2 (95% CI: 0.005, 0.008) for the total body for women with lower vitamin D levels. Among women with higher vitamin D levels, the corresponding differences in BMD were 0.066 g/cm2 (95% CI: 0.064, 0.068) at the lumbar spine, 0.032 g/cm2 (95% CI: 0.031, 0.033) at the proximal femur, and 0.024 g/cm2 (95% CI: 0.023, 0.025) for the total body.

Figure 3.

Adjusted mean values of bone mineral density at the proximal femur, at the lumbar spine (L1–L4), and for the total body in relation to coffee consumption (<1, 1, 2–3, or ≥4 cups per day) in the Swedish Mammography Cohort, counties of Uppsala and Västmanland, Sweden, 1987–2008. The error bars indicate 95% confidence intervals. The mean values at each site were adjusted for year of birth; body mass index and height at time of bone mineral density measurement; total energy intake; dietary intakes of calcium, vitamin D, retinol, protein, phosphorous, potassium, and alcohol; calcium supplementation; vitamin D supplementation; tea consumption (number of cups per day); educational level (<9, 9–12, >12 years, or other); physical activity (5 categories); smoking status (never, former, or current); previous fractures before the study period; Charlson comorbidity index; living condition (living alone or not); nulliparity; cortisone use; hormone replacement therapy; and use of bisphosphonates.

Figure 3.

Adjusted mean values of bone mineral density at the proximal femur, at the lumbar spine (L1–L4), and for the total body in relation to coffee consumption (<1, 1, 2–3, or ≥4 cups per day) in the Swedish Mammography Cohort, counties of Uppsala and Västmanland, Sweden, 1987–2008. The error bars indicate 95% confidence intervals. The mean values at each site were adjusted for year of birth; body mass index and height at time of bone mineral density measurement; total energy intake; dietary intakes of calcium, vitamin D, retinol, protein, phosphorous, potassium, and alcohol; calcium supplementation; vitamin D supplementation; tea consumption (number of cups per day); educational level (<9, 9–12, >12 years, or other); physical activity (5 categories); smoking status (never, former, or current); previous fractures before the study period; Charlson comorbidity index; living condition (living alone or not); nulliparity; cortisone use; hormone replacement therapy; and use of bisphosphonates.

Coffee and low-energy falls

A large proportion (90%, n = 4,521) of the women in the SMCC answered questions on their history of low-energy falls. Of the 90%, 18% (n = 811) had experienced at least 1 fall and 8% (n = 382) at least 2 falls during the previous year. When women in the highest coffee consumption category (≥4 cups) were compared with those in the lowest category (<1 cup), the adjusted odds ratio was 0.98 (95% CI: 0.68, 1.41) for at least 1 fall in the previous year and 1.27 (95% CI: 0.75, 2.15) for at least 2 falls in the previous year.

DISCUSSION

In this large, prospective cohort study of middle-aged and older Swedish women, we did not observe an association between coffee consumption and fracture risk. Previous studies on consumption of coffee or intake of caffeine and fracture risk in women have shown discrepant results. Some demonstrate increased risk of fractures with high intakes (6–11), whereas others were unable to demonstrate such an association (13–19). In contrast to our present results, we have previously observed an increased rate of osteoporotic fractures in women consuming at least 4 cups of coffee per day in a subset of the present cohort (12). Limitations of our previous analysis, some of which were shared with other studies, were a shorter maximum follow-up time (13 vs. 21 years), a lower number of fracture cases, and, most importantly, incomplete case ascertainment (because we were able to identify only fractures that occurred in the county in which the study was performed) (12). In the present study, we have complete case ascertainment by matching to nationwide inpatient and outpatient registers that include a large number of fractures. Furthermore, we previously were not able to adequately control for lifestyle habits and comorbidity, which could potentially influence the association, although adjustments in our present study did not confer major changes on the estimates.

In our prior analysis (12), we found that the higher risk of fracture with high coffee intake was concentrated in women with a modest calcium intake. This effect modification was not confirmed in the present analysis. In the SMCC we estimated the associations by vitamin D status. Unexpectedly, we found a somewhat stronger association between coffee intake and BMD among women with vitamin D values higher than 50 nmol/L. A low vitamin D level might reflect frailty (42), and in the subgroup with low vitamin D values, the association between coffee consumption and bone density might have been overshadowed by factors related to frailty that more strongly influence BMD.

In the majority of studies investigating associations between coffee/caffeine consumption or intake and BMD in women, no association could be demonstrated (16, 43–48), whereas such associations have been observed in some studies (25, 49, 50). Several explanations are possible for the absence of association in these studies, including small study size (16, 45–47), low consumption of coffee (46, 47), or lack of separate assessment of tea and coffee consumption (46, 48). The small differences in BMD associated with coffee consumption in the present study do not seem to impact the risk of osteoporosis or incident fractures. Importantly, earlier research has shown that the influence of BMD, versus accidental fall–related factors (51), on fracture rate is more modest in the oldest-old than in young-old individuals. Theoretically, the small decrease in BMD associated with high coffee consumption might be counteracted by a reduced likelihood of hypotension (52, 53) and comorbidities (1, 54), leading to a lower propensity for injurious falls. The net effect could therefore be no substantial excess risk of fracture. No association between coffee consumption and fall risk was observed in our study, but self-reported falls might not accurately reflect actual injurious falls (55–57).

Strengths and limitations

Several strengths of the study deserve consideration. Because of the high incidence of fractures in our setting, together with a long follow-up, we were able to identify a large number of incident fractures with high accuracy and complete coverage. We were also able to study BMD as a secondary outcome in a large subcohort. The use of repeated FFQs is likely to have reduced measurement error in exposure information. A generally high consumption level in combination with considerable range in the consumption of coffee in this cohort constitutes an additional major strength. Finally, tea consumption is low in this cohort, which might be another advantage in that, for example, the flavonoid compounds in tea could have estrogen-like positive effects on bone (58).

The study has some weaknesses that should be addressed. Although conclusions about causality cannot be drawn, the prospective design limits the potential for participants to provide selective information because of recall bias or changes in dietary habits due to the fracture event. It might not be possible to generalize the results to women of different ethnic origins or to men. The self-administered questionnaires did not include reference to the size of a cup of coffee, which could have led to misclassification because the volume of a cup can range from 150 to 250 mL (or even larger). Nevertheless, a validation study indicated that the reported coffee consumption seems to be a reasonable estimate of the exposure. Information on caffeinated soft drinks was not available. Consumption of soft drinks (any kind) was low, with 1.4% of the women consuming 1 can of soda or more each day. We did not exclude fractures that were due to high trauma because there are indications of a comparable increased risk of both low- and high-trauma fracture with decreasing bone density in the elderly (35) and because cause-of-injury coding can be inaccurate (59). However, we were able to analyze time to a second fracture, which theoretically would be a better measure of osteoporotic fracture risk than a single fracture, but no increased risk of fracture was observed.

Although we were able to adjust our associations for several potential covariates, including comorbidities, there still could be residual confounding. Information on smoking and physical activity was available from only the second questionnaire, and data for these variables were imputed for the baseline investigation. However, imputation is less likely to introduce bias than is performance of complete-subject analysis (60), and our sensitivity analyses without time-updated covariates did not change our conclusions. Finally, a genetic variant of the vitamin D receptor might render carriers more susceptible to the effects of caffeine on bone (61). Genetically determined differences in the metabolism of caffeine might also be of relevance in this context (62), but genotyping was not done in the present study.

The lack of increased risk of fractures in our study, even among women consuming high amounts of coffee, is important. This ends a long-running debate about coffee as a potential risk factor for osteoporotic fractures, at least in women.

ACKNOWLEDGMENTS

Author affiliations: Department of Surgical Sciences, Section of Orthopedics, Uppsala University, Uppsala, Sweden (Helena Hallström, Liisa Byberg, Eva Warensjö Lemming, Karl Michaëlsson); Risk and Benefit Assessment Department, National Food Agency, Uppsala, Sweden (Helena Hallström, Anders Glynn); and National Institute of Environmental Medicine, Division of Nutritional Epidemiology, Karolinska Institute, Stockholm, Sweden (Alicja Wolk).

The Swedish Research Council financed the study.

Helena Hallström presented the study at the European Calcified Tissue Society annual meeting, May 19–23, 2012, in Stockholm, Sweden.

Conflict of interest: none declared.

REFERENCES

1
Freedman
ND
Park
Y
Abnet
CC
, et al.  . 
Association of coffee drinking with total and cause-specific mortality
N Engl J Med
 , 
2012
, vol. 
366
 
20
(pg. 
1891
-
1904
)
2
Cumming
RG
Nevitt
MC
Cummings
SR
Epidemiology of hip fractures
Epidemiol Rev
 , 
1997
, vol. 
19
 
2
(pg. 
244
-
257
)
3
Riggs
BL
Melton
LJ
The worldwide problem of osteoporosis: insights afforded by epidemiology
Bone
 , 
1995
, vol. 
17
 
5 suppl
(pg. 
505S
-
511S
)
4
Center
JR
Nguyen
TV
Schneider
D
, et al.  . 
Mortality after all major types of osteoporotic fracture in men and women: an observational study
Lancet
 , 
1999
, vol. 
353
 
9156
(pg. 
878
-
882
)
5
Samelson
EJ
Hannan
MT
Epidemiology of osteoporosis
Curr Rheumatol Rep
 , 
2006
, vol. 
8
 
1
(pg. 
76
-
83
)
6
Kiel
DP
Felson
DT
Hannan
MT
, et al.  . 
Caffeine and the risk of hip fracture: the Framingham Study
Am J Epidemiol
 , 
1990
, vol. 
132
 
4
(pg. 
675
-
684
)
7
Hansen
SA
Folsom
AR
Kushi
LH
, et al.  . 
Association of fractures with caffeine and alcohol in postmenopausal women: the Iowa Women's Health Study
Public Health Nutr
 , 
2000
, vol. 
3
 
3
(pg. 
253
-
261
)
8
Hernandez-Avila
M
Colditz
GA
Stampfer
MJ
, et al.  . 
Caffeine, moderate alcohol intake, and risk of fractures of the hip and forearm in middle-aged women
Am J Clin Nutr
 , 
1991
, vol. 
54
 
1
(pg. 
157
-
163
)
9
Jokinen
H
Pulkkinen
P
Korpelainen
J
, et al.  . 
Risk factors for cervical and trochanteric hip fractures in elderly women: a population-based 10-year follow-up study
Calcif Tissue Int
 , 
2010
, vol. 
87
 
1
(pg. 
44
-
51
)
10
Meyer
HE
Pedersen
JI
Løken
EB
, et al.  . 
Dietary factors and the incidence of hip fracture in middle-aged Norwegians. A prospective study
Am J Epidemiol
 , 
1997
, vol. 
145
 
2
(pg. 
117
-
123
)
11
Nieves
JW
Grisso
JA
Kelsey
JL
A case-control study of hip fracture: evaluation of selected dietary variables and teenage physical activity
Osteoporos Int
 , 
1992
, vol. 
2
 
3
(pg. 
122
-
127
)
12
Hallström
H
Wolk
A
Glynn
A
, et al.  . 
Coffee, tea and caffeine consumption in relation to osteoporotic fracture risk in a cohort of Swedish women
Osteoporos Int
 , 
2006
, vol. 
17
 
7
(pg. 
1055
-
1064
)
13
Albrand
G
Munoz
F
Sornay-Rendu
E
, et al.  . 
Independent predictors of all osteoporosis-related fractures in healthy postmenopausal women: the OFELY Study
Bone
 , 
2003
, vol. 
32
 
1
(pg. 
78
-
85
)
14
Cummings
SR
Nevitt
MC
Browner
WS
, et al.  . 
Risk factors for hip fracture in white women. Study of Osteoporotic Fractures Research Group
N Engl J Med
 , 
1995
, vol. 
332
 
12
(pg. 
767
-
773
)
15
Fujiwara
S
Kasagi
F
Yamada
M
, et al.  . 
Risk factors for hip fracture in a Japanese cohort
J Bone Miner Res
 , 
1997
, vol. 
12
 
7
(pg. 
998
-
1004
)
16
Hansen
MA
Overgaard
K
Riis
BJ
, et al.  . 
Potential risk factors for development of postmenopausal osteoporosis—examined over a 12-year period
Osteoporos Int
 , 
1991
, vol. 
1
 
2
(pg. 
95
-
102
)
17
Holbrook
TL
Barrett-Connor
E
Wingard
DL
Dietary calcium and risk of hip fracture: 14-year prospective population study
Lancet
 , 
1988
, vol. 
2
 
8619
(pg. 
1046
-
1049
)
18
Huopio
J
Kröger
H
Honkanen
R
, et al.  . 
Risk factors for perimenopausal fractures: a prospective study
Osteoporos Int
 , 
2000
, vol. 
11
 
3
(pg. 
219
-
227
)
19
van Geel
AC
Geusens
PP
Nagtzaam
IF
, et al.  . 
Timing and risk factors for clinical fractures among postmenopausal women: a 5-year prospective study
BMC Med
 , 
2006
, vol. 
4
 pg. 
24
 
20
Lu
PZ
Lai
CY
Chan
WH
Caffeine induces cell death via activation of apoptotic signal and inactivation of survival signal in human osteoblasts
Int J Mol Sci
 , 
2008
, vol. 
9
 
5
(pg. 
698
-
718
)
21
Tsuang
YH
Sun
JS
Chen
LT
, et al.  . 
Direct effects of caffeine on osteoblastic cells metabolism: the possible causal effect of caffeine on the formation of osteoporosis
J Orthop Surg Res
 , 
2006
, vol. 
1
 pg. 
7
 
22
Massey
LK
Opryszek
AA
No effects of adaptation to dietary caffeine on calcium excretion in young women
Nutr Res
 , 
1990
, vol. 
10
 
7
(pg. 
741
-
747
)
23
Barger-Lux
MJ
Heaney
RP
Caffeine and the calcium economy revisited
Osteoporos Int
 , 
1995
, vol. 
5
 
2
(pg. 
97
-
102
)
24
Barrett-Connor
E
Chang
JC
Edelstein
SL
Coffee-associated osteoporosis offset by daily milk consumption. The Rancho Bernardo Study
JAMA
 , 
1994
, vol. 
271
 
4
(pg. 
280
-
283
)
25
Harris
SS
Dawson-Hughes
B
Caffeine and bone loss in healthy postmenopausal women
Am J Clin Nutr
 , 
1994
, vol. 
60
 
4
(pg. 
573
-
578
)
26
Andersson
HC
Hallström
H
Kihlman
BA
Intake of caffeine and other methylxanthines during pregnancy and risk for adverse effects in pregnant women and their foetuses
 , 
2004
Copenhagen
Nordic Council of Ministers
 
27
Kanis
JA
Oden
A
McCloskey
EV
, et al.  . 
A systematic review of hip fracture incidence and probability of fracture worldwide
Osteoporos Int
 , 
2012
, vol. 
23
 
9
(pg. 
2239
-
2256
)
28
Warensjö
E
Byberg
L
Melhus
H
, et al.  . 
Dietary calcium intake and risk of fracture and osteoporosis: prospective longitudinal cohort study
BMJ
 , 
2011
, vol. 
342
 pg. 
d1473
 
29
Larsson
SC
Virtamo
J
Wolk
A
Fish consumption and risk of stroke in Swedish women
Am J Clin Nutr
 , 
2011
, vol. 
93
 
3
(pg. 
487
-
493
)
30
Bergström
L
Hagman
U
Ericsson
H
, et al.  . 
The food composition database KOST: the National Food Administration's information system for nutritive values of food (Swedish)
Vår Föda
 , 
1991
, vol. 
43
 
8
(pg. 
439
-
447
)
31
Willett
WC
Howe
GR
Kushi
LH
Adjustment for total energy intake in epidemiologic studies
Am J Clin Nutr
 , 
1997
, vol. 
65
 
4 suppl
(pg. 
1220S
-
1228S
)
32
Michels
KB
Holmberg
L
Bergkvist
L
, et al.  . 
Coffee, tea, and caffeine consumption and breast cancer incidence in a cohort of Swedish women
Ann Epidemiol
 , 
2002
, vol. 
12
 
1
(pg. 
21
-
26
)
33
Calltorp
J
Adami
HO
Astrom
H
, et al.  . 
Country profile: Sweden
Lancet
 , 
1996
, vol. 
347
 
9001
(pg. 
587
-
594
)
34
Gedeborg
R
Engquist
H
Berglund
L
, et al.  . 
Identification of incident injuries in hospital discharge registers
Epidemiology
 , 
2008
, vol. 
19
 
6
(pg. 
860
-
867
)
35
Mackey
DC
Lui
LY
Cawthon
PM
, et al.  . 
High-trauma fractures and low bone mineral density in older women and men
JAMA
 , 
2007
, vol. 
298
 
20
(pg. 
2381
-
2388
)
36
Kanis
JA
Assessment of fracture risk and its application to screening for postmenopausal osteoporosis: synopsis of a WHO report. WHO Study Group
Osteoporos Int
 , 
1994
, vol. 
4
 
6
(pg. 
368
-
381
)
37
Charlson
ME
Pompei
P
Ales
KL
, et al.  . 
A new method of classifying prognostic comorbidity in longitudinal studies: development and validation
J Chronic Dis
 , 
1987
, vol. 
40
 
5
(pg. 
373
-
383
)
38
Quan
H
Sundararajan
V
Halfon
P
, et al.  . 
Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data
Med Care
 , 
2005
, vol. 
43
 
11
(pg. 
1130
-
1139
)
39
Orsini
N
Bellocco
R
Bottai
M
, et al.  . 
Validity of self-reported total physical activity questionnaire among older women
Eur J Epidemiol
 , 
2008
, vol. 
23
 
10
(pg. 
661
-
667
)
40
Andersson
T
Alfredsson
L
Kallberg
H
, et al.  . 
Calculating measures of biological interaction
Eur J Epidemiol
 , 
2005
, vol. 
20
 
7
(pg. 
575
-
579
)
41
Ross
AC
Manson
JE
Abrams
SA
, et al.  . 
The 2011 report on dietary reference intakes for calcium and vitamin D from the Institute of Medicine: what clinicians need to know
J Clin Endocrinol Metab
 , 
2011
, vol. 
96
 
1
(pg. 
53
-
58
)
42
Ensrud
KE
Ewing
SK
Fredman
L
, et al.  . 
Circulating 25-hydroxyvitamin D levels and frailty status in older women
J Clin Endocrinol Metab
 , 
2010
, vol. 
95
 
12
(pg. 
5266
-
5273
)
43
Hannan
MT
Felson
DT
Dawson-Hughes
B
, et al.  . 
Risk factors for longitudinal bone loss in elderly men and women: the Framingham Osteoporosis Study
J Bone Miner Res
 , 
2000
, vol. 
15
 
4
(pg. 
710
-
720
)
44
Holm
K
Dan
A
Wilbur
J
, et al.  . 
A longitudinal study of bone density in midlife women
Health Care Women Int
 , 
2002
, vol. 
23
 
6-7
(pg. 
678
-
691
)
45
Lloyd
T
Johnson-Rollings
N
Eggli
DF
, et al.  . 
Bone status among postmenopausal women with different habitual caffeine intakes: a longitudinal investigation
J Am Coll Nutr
 , 
2000
, vol. 
19
 
2
(pg. 
256
-
261
)
46
Lloyd
T
Rollings
NJ
Kieselhorst
K
, et al.  . 
Dietary caffeine intake is not correlated with adolescent bone gain
J Am Coll Nutr
 , 
1998
, vol. 
17
 
5
(pg. 
454
-
457
)
47
Packard
PT
Recker
RR
Caffeine does not affect the rate of gain in spine bone in young women
Osteoporos Int
 , 
1996
, vol. 
6
 
2
(pg. 
149
-
152
)
48
Reid
IR
Ames
RW
Evans
MC
, et al.  . 
Determinants of the rate of bone loss in normal postmenopausal women
J Clin Endocrinol Metab
 , 
1994
, vol. 
79
 
4
(pg. 
950
-
954
)
49
Korpelainen
R
Korpelainen
J
Heikkinen
J
, et al.  . 
Lifestyle factors are associated with osteoporosis in lean women but not in normal and overweight women: a population-based cohort study of 1222 women
Osteoporos Int
 , 
2003
, vol. 
14
 
1
(pg. 
34
-
43
)
50
Wetmore
CM
Ichikawa
L
LaCroix
AZ
, et al.  . 
Association between caffeine intake and bone mass among young women: potential effect modification by depot medroxyprogesterone acetate use
Osteoporos Int
 , 
2008
, vol. 
19
 
4
(pg. 
519
-
527
)
51
Khani
BR
Ye
W
Terry
P
, et al.  . 
Reproducibility and validity of major dietary patterns among Swedish women assessed with a food-frequency questionnaire
J Nutr
 , 
2004
, vol. 
134
 
6
(pg. 
1541
-
1545
)
52
Mesas
AE
Leon-Munoz
LM
Rodriguez-Artalejo
F
, et al.  . 
The effect of coffee on blood pressure and cardiovascular disease in hypertensive individuals: a systematic review and meta-analysis
Am J Clin Nutr
 , 
2011
, vol. 
94
 
4
(pg. 
1113
-
1126
)
53
Rakic
V
Beilin
LJ
Burke
V
Effect of coffee and tea drinking on postprandial hypotension in older men and women
Clin Exp Pharmacol Physiol
 , 
1996
, vol. 
23
 
6-7
(pg. 
559
-
563
)
54
Muley
A
Muley
P
Shah
M
Coffee to reduce risk of type 2 diabetes? A systematic review
Curr Diabetes Rev
 , 
2012
, vol. 
8
 
3
(pg. 
162
-
168
)
55
Barber
J
Muller
S
Whitehurst
T
, et al.  . 
Measuring morbidity: self-report or health care records?
Fam Pract
 , 
2010
, vol. 
27
 
1
(pg. 
25
-
30
)
56
Kanten
DN
Mulrow
CD
Gerety
MB
, et al.  . 
Falls: an examination of three reporting methods in nursing homes
J Am Geriatr Soc
 , 
1993
, vol. 
41
 
6
(pg. 
662
-
666
)
57
Robinovitch
SN
Feldman
F
Yang
Y
, et al.  . 
Video capture of the circumstances of falls in elderly people residing in long-term care: an observational study
Lancet
 , 
2013
, vol. 
381
 
9860
(pg. 
47
-
54
)
58
Wu
CH
Yang
YC
Yao
WJ
, et al.  . 
Epidemiological evidence of increased bone mineral density in habitual tea drinkers
Arch Intern Med
 , 
2002
, vol. 
162
 
9
(pg. 
1001
-
1006
)
59
Bergström
MF
Byberg
L
Melhus
H
, et al.  . 
Extent and consequences of misclassified injury diagnoses in a national hospital discharge registry
Inj Prev
 , 
2011
, vol. 
17
 
2
(pg. 
108
-
113
)
60
van der Heijden
G
Donders
A
Stijnen
T
, et al.  . 
Imputation of missing values is superior to complete case analysis and the missing-indicator method in multivariable diagnostic research: a clinical example
J Clin Epidemiol
 , 
2006
, vol. 
59
 
10
(pg. 
1102
-
1109
)
61
Rapuri
PB
Gallagher
JC
Nawaz
Z
Caffeine decreases vitamin D receptor protein expression and 1,25(OH)2D3 stimulated alkaline phosphatase activity in human osteoblast cells
J Steroid Biochem Mol Biol
 , 
2007
, vol. 
103
 
3-5
(pg. 
368
-
371
)
62
Hallström
H
Melhus
H
Glynn
A
, et al.  . 
Coffee consumption and CYP1A2 genotype in relation to bone mineral density of the proximal femur in elderly men and women: a cohort study
Nutr Metab (Lond)
 , 
2010
, vol. 
7
 pg. 
12
 

Appendix

45

Appendix Table 1.

Characteristics of Women in the Swedish Mammography Cohort Clinical, Counties of Uppsala and Västmanland, Sweden, 1987–2009

  No. of Cups of Coffee Per Day
 
<1
 
1
 
2–3
 
≥4
 
No. Mean (SD) No. Mean (SD) No. Mean (SD) No. Mean (SD) 
No. of women 375 7.5  881 17.5  3,009 59.9  757 15.1  
Age at DXA investigation, years   67.3 (7.0)   68.6 (7.2)   67.7 (6.7)   66.1 (6.0) 
Body mass index (cumulative average of all questionnaires)a   24.2 (3.7)   24.6 (3.6)   24.6 (3.5)   24.8 (3.5) 
Average intake per dayb             
 Energy, kcal   1,692 (406)   1,649 (386)   1,718 (386)   1,782 (398) 
 Calcium, mg   971 (228)   987 (198)   1,019 (199)   1,049 (212) 
 Supplemental calcium, mgc   389 (177)   374 (182)   359 (357)   391 (173) 
 Total vitamin D, µg   5.51 (1.89)   5.58 (1.92)   5.50 (1.67)   5.32 (1.59) 
 Alcohol, g   4.55 (5.75)   5.73 (5.04)   5.29 (4.56)   4.92 (4.44) 
Median intake per dayd             
 Coffee, g   67 (130)   282 (83)   501 (160)   812 (148) 
 Tea, g   441 (443)   238 (286)   109 (216)   30 (132) 
Lifestyle factors and medications             
 Leisure-time physical activity level             
  1 (lowest) 62 19.9  124 17.1  406 16.5  124 20.2  
  2 57 18.3  141 19.5  511 20.7  133 21.6  
  3 111 35.7  247 34.1  840 34.1  204 33.2  
  4 51 16.4  109 15.1  380 15.4  71 11.5  
  5 (highest) 30 9.7  103 14.2  330 13.4  83 13.5  
 Smoking status             
  Current 21 5.6  38 4.3  256 8.5  140 18.5  
  Former 94 25.1  294 33.4  1,029 34.2  307 40.6  
  Never 260 69.3  549 62.3  1,724 57.3  310 41.0  
 Two or more Charlson comorbidities 1.6  1.0  29 1.0  1.1  
 Educational level ≤9 years 148 39.5  461 52.3  1,651 54.9  445 58.8  
 Fracture before DXA scan 65 17.3  159 18.1  533 17.7  142 18.8  
 Bisphosphonate use 2.4  15 1.7  47 1.6  11 1.5  
 Marital status: single 80 21.3  210 23.8  577 19.2  140 18.5  
 Nulliparity 53 14.1  102 11.6  326 10.8  68 9.0  
  Hormone replacement therapy use, ever 229 61.1  570 64.7  1,858 61.8  437 57.7  
 Cortisone use 44 11.7  100 11.4  283 9.4  81 10.7  
 Vitamin D supplement use 78 20.8  161 18.3  439 14.6  95 12.6  
 Calcium supplement use 57 15.2  116 13.2  357 11.9  80 10.6  
  No. of Cups of Coffee Per Day
 
<1
 
1
 
2–3
 
≥4
 
No. Mean (SD) No. Mean (SD) No. Mean (SD) No. Mean (SD) 
No. of women 375 7.5  881 17.5  3,009 59.9  757 15.1  
Age at DXA investigation, years   67.3 (7.0)   68.6 (7.2)   67.7 (6.7)   66.1 (6.0) 
Body mass index (cumulative average of all questionnaires)a   24.2 (3.7)   24.6 (3.6)   24.6 (3.5)   24.8 (3.5) 
Average intake per dayb             
 Energy, kcal   1,692 (406)   1,649 (386)   1,718 (386)   1,782 (398) 
 Calcium, mg   971 (228)   987 (198)   1,019 (199)   1,049 (212) 
 Supplemental calcium, mgc   389 (177)   374 (182)   359 (357)   391 (173) 
 Total vitamin D, µg   5.51 (1.89)   5.58 (1.92)   5.50 (1.67)   5.32 (1.59) 
 Alcohol, g   4.55 (5.75)   5.73 (5.04)   5.29 (4.56)   4.92 (4.44) 
Median intake per dayd             
 Coffee, g   67 (130)   282 (83)   501 (160)   812 (148) 
 Tea, g   441 (443)   238 (286)   109 (216)   30 (132) 
Lifestyle factors and medications             
 Leisure-time physical activity level             
  1 (lowest) 62 19.9  124 17.1  406 16.5  124 20.2  
  2 57 18.3  141 19.5  511 20.7  133 21.6  
  3 111 35.7  247 34.1  840 34.1  204 33.2  
  4 51 16.4  109 15.1  380 15.4  71 11.5  
  5 (highest) 30 9.7  103 14.2  330 13.4  83 13.5  
 Smoking status             
  Current 21 5.6  38 4.3  256 8.5  140 18.5  
  Former 94 25.1  294 33.4  1,029 34.2  307 40.6  
  Never 260 69.3  549 62.3  1,724 57.3  310 41.0  
 Two or more Charlson comorbidities 1.6  1.0  29 1.0  1.1  
 Educational level ≤9 years 148 39.5  461 52.3  1,651 54.9  445 58.8  
 Fracture before DXA scan 65 17.3  159 18.1  533 17.7  142 18.8  
 Bisphosphonate use 2.4  15 1.7  47 1.6  11 1.5  
 Marital status: single 80 21.3  210 23.8  577 19.2  140 18.5  
 Nulliparity 53 14.1  102 11.6  326 10.8  68 9.0  
  Hormone replacement therapy use, ever 229 61.1  570 64.7  1,858 61.8  437 57.7  
 Cortisone use 44 11.7  100 11.4  283 9.4  81 10.7  
 Vitamin D supplement use 78 20.8  161 18.3  439 14.6  95 12.6  
 Calcium supplement use 57 15.2  116 13.2  357 11.9  80 10.6  

Abbreviations: DXA, dual-emission x-ray absorptiometry; SD, standard deviation.

a Weight (kg)/height (m)2.

b Intake per day refers to the energy-adjusted cumulative average intake.

c Supplemental calcium was used by 610 women.

d Median intake per day (interquartile range) estimated with data from all 3 food frequency questionnaires.

Appendix Table 2.

Numbers at Risk for Any Fracture and Hip Fracture, Corresponding to the Kaplan-Meier Failure Curves in Figure 2, in the Swedish Mammography Cohort, Counties of Uppsala and Västmanland, Sweden, 1987–2008

Cups of Coffee/Day Years of Follow-up
 
10 15 20 
Any fracture      
 <1 5,534 5,194 5,300 4,492 2,507 
 1 8,272 7,700 6,292 5,260 2,928 
 2–3 36,557 34,716 27,089 23,221 13,175 
 ≥4 11,070 10,605 14,788 13,036 7,438 
Hip fracture      
 <1 5,534 5,329 5,528 5,125 3,066 
 1 8,272 7,951 6,884 6,029 3,640 
 2–3 36,557 35,599 29,653 26,385 16,069 
 ≥4 11,070 10,851 14,686 14,807 9,049 
Cups of Coffee/Day Years of Follow-up
 
10 15 20 
Any fracture      
 <1 5,534 5,194 5,300 4,492 2,507 
 1 8,272 7,700 6,292 5,260 2,928 
 2–3 36,557 34,716 27,089 23,221 13,175 
 ≥4 11,070 10,605 14,788 13,036 7,438 
Hip fracture      
 <1 5,534 5,329 5,528 5,125 3,066 
 1 8,272 7,951 6,884 6,029 3,640 
 2–3 36,557 35,599 29,653 26,385 16,069 
 ≥4 11,070 10,851 14,686 14,807 9,049 

Author notes

Abbreviations: BMD, bone mineral density; CI, confidence interval; FFQ, food frequency questionnaire; HR, hazard ratio; SMCC, Swedish Mammography Cohort Clinical—subcohort of the Swedish Mammography Cohort.