Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2013 Sep 11.
Published in final edited form as: Behav Med. 2009 Winter;34(4):145–155. doi: 10.3200/BMED.34.4.145-155

Predictors of Adherence in the Women’s Health Initiative Calcium and Vitamin D Trial

R Brunner 1, J Dunbar-Jacob 2, M S LeBoff 3, I Granek 4, D Bowen 5, L G Snetselaar 6, S A Shumaker 7, J Ockene 8, M Rosal 9, J Wactawski-Wende 10, J Cauley 11, B Cochrane 12, L Tinker 13, R Jackson 14, C Y Wang 15, L Wu 16
PMCID: PMC3770154  NIHMSID: NIHMS84710  PMID: 19064373

Abstract

The authors analyzed data from the Women’s Health Initiative (WHI) Calcium and Vitamin D Supplementation Trial (CaD) to learn more about factors affecting adherence to clinical trial study pills (both active and placebo). Most participants (36,282 postmenopausal women aged 50–79 years) enrolled in CaD 1 year after joining either a hormone trial or the dietary modification trial of WHI. The WHI researchers measured adherence to study pills by weighing the amount of remaining pills at an annual study visit; adherence was primarily defined as taking ≥ 80% of the pills. The authors in this study examined a number of behavioral, demographic, procedural, and treatment variables for association with study pill adherence. They found that relatively simple procedures (ie, phone contact early in the study [4 weeks post randomization] and direct social contact) later in the trial may improve adherence. Also, at baseline, past pill-use experiences, personal supplement use, and relevant symptoms may be predictive of adherence in a supplement trial.

Index Terms: adherence, calcium supplementation, clinical trial, women

In the Women’s Health Initiative (WHI) Calcium and Vitamin D Supplementation Trial (CaD), an intention-to-treat analysis showed that active treatment did not significantly reduce hip fracture, which was the primary outcome measure.1 However, a subanalysis of adherent (at least 80% of study medication taken) participants revealed a significant reduction (29%) in hip fractures; thus, understanding adherence issues is important in reducing such injuries.

Reduced adherence to medications (not always in clinical trials) is associated with increasing age, lower socioeconomic level, smoking, and various indicators of poorer health status.2,3 In several meta-analyses,4–7 improved adherence to medical treatments was associated with social support.8,9 Other meta-analyses have shown that depression (but not anxiety) and hopelessness about one’s medical condition reduce adherence.10–15

In this study, we evaluated factors that have been demonstrated to generally predict higher levels of adherence16 and considered variables specific to intervention trials. To conceptualize disparate measures, we categorized variables as sociodemographic, psychosocial, health status, and procedural. This strategy may aid the systematic evaluation and enhancement of adherence in future trials and clinical interventions.

METHODS

Participants

Detailed descriptions of eligibility criteria, recruitment procedures, and primary findings for the WHI are available elsewhere.17,18 Researchers recruited postmenopausal women (aged 50–79 years) into WHI randomized trials that assessed hormone therapy (HT) or dietary modification (DM) at 40 US centers, and 1 year later invited them to join the CaD trial. Of the 36,282 participants randomized to CaD, we analyzed the 91% who joined initially (not later). Among the participants in the CaD trial, 54% were in HT, 69% percent in DM, and 14% in both. WHI researchers obtained informed consent for each of the trials, using forms approved by local institutional review boards.

Procedure

The trial was a randomized, double-blind, placebo-controlled design19 in which participants took 2 tablets daily of 1,000 mg of calcium carbonate with 400 IU of Vitamin D3 or a matching placebo (study pills provided by Glaxo Smith-Kline [Bensalem, PA]). Until October 1997, only a chewable pill was available; the researchers subsequently offered all participants a swallowable form. WHI researchers assessed adherence annually by weighing remaining pills in conjunction with conducting a structured interview. Pill weighing was not observed by or discussed with participants. Adherence below 80% triggered staff efforts to optimize adherence. To assess for side effects, the researchers implemented “step down” procedures (single daily pill, then even fewer) with monthly re-evaluations; a reduced regimen was considered nonadherence in the present analysis.

WHI researchers collected questionnaires at initial enrollment, 1 year prior to CaD trial randomization (see Table 1). Sociodemographic variables included age, ethnicity, marital status, income, education, and insurance status. Psychosocial variables included depression (measured by the Center for Epidemiological Studies Depression Scale [CES-D]20 and the Diagnostic Interview Schedule21) and personality traits (measured by the Life Orientation Test-Revised22 [optimism], the Cook-Medley Hostility Scale,23 the Ambivalence Over Emotional Expressiveness Questionnaire, and the Emotional Expressiveness Questionnaire24). The Medical Outcomes Study Social Support Questionnaire25 and 4 items taken from the Social Relationships Scale26 measured social support and social strain. The Alameda County Study questionnaire27 measured life events in the prior year. The Rand 36-Item Health Survey (RAND36)28,29 assessed quality of life and general functioning.

TABLE 1.

Key Adherence Predictors in the Women’s Health Initiative (WHI)

Construct Measure Items Subscales Use Reliability and validity Sample item
Sociodemographic variables
 Age, marital status, income, education, insurance status Multiple single item 5 — Wide — What is your age?
Psychosocial variables
 Optimism Life Orientation Test21 Revised (LOT-R) 6 1 Limited α = .75 in WHI; r = −0.42 with depression I’m always hopeful about my future (1 = Strongly disagree to 5 = Strongly agree)
 Depression Center for Epidemiological Studies Depression Scale20 8 0 Wide α = .73; compares well to other clinical measures of depression How often during the past week did you feel sad?
 Hostility Cynicism23 13 1 Limited α = .76; compares well to other trait negativity measures I think most people would like to get ahead
 Social strain Social Relationships Scales26 4 1 Limited No data available; α = .72 in WHI Of the people that are important to you, how many get on your nerves?
 Social support MOS Social Support Questionnaire25 9 4 Wide Some data available from MOS; α = .93 in WHI How often is this available? Someone to love you and make you feel wanted?
 Stressful life events Life Events Scale27 11 0 Wide No info Did your spouse or partner die?
 Quality of Life RAND–3628,29 36 8 Wide Numerous reliability and validity reports How would you rate your current sense of well-being?
Health variables
 Disease factors Self-reported prior diagnosis and family history of outcomes 6 — Wide — Have you ever been told by a health care provider that…?
Symptoms Symptom checklist31 34 No Modified from PEPI No psychometrics Did not occur and how much did they bother you?
Use of personal supplements WHI supplement use (interview) 7 — — — Multivitamin (no minerals), multivitamin (minerals), etc. Dose, unit, pills per week, etc.
 Health habits Single item questions 4 — Wide — Have you smoked at least 100 cigarettes in your lifetime?
Procedural variables
 Pill type Swallowed, chewed, or switched between the two — — — — —
 Clinic identity — 1 — — — —
 Other RCT participation Hormone trial, diet modification trial, or both — — — — —
 4-week follow-up phone contact Yes or No — — — — —
 Type of semiannual contact In clinic or by phone — — — — —
Open in a new tab

Health variables included smoking status, body mass index (BMI), physical activity (eg, walking, housework), and reported breast or colorectal cancer or heart attack in self or immediate family. The WHI researchers assessed physical activity levels with a series of questions with metabolic equivalent time (kcal/kg/wk) scores assigned.30 They calculated osteoporosis risk using age, ethnicity, exercise level, smoking status, hormone usage, family history, prior fractures, and calcium intake, with scores ranging from 0 to 9 (> 4 was high risk). A standard symptom checklist examined physical complaints, particularly gastrointestinal symptoms (eg, bloating or gas, constipation, belly pain).31 A detailed questionnaire measured calcium and other supplement use.

Procedural variables were the pill type (swallowable or chewable), change in pill type, clinic site, parallel participation and adherence in HT and DM, completion of the recommended follow-up call scheduled 4 weeks after randomization, and whether the semiannual (ie, each anniversary date of randomization plus 6 months) contact was a visit or by phone.

Statistical Analysis

Using SAS version 9.0 (SAS, Inc, Cary, NC), we first analyzed participants with nonmissing values for all predictors. To address missing values, we also calculated the inverse selection probability weighted estimator.32,33 We performed multivariate logistic regression analyses to evaluate the likelihood of adherence (≥ 80% per study protocol) to study pills 1 year after randomization (initial change model). We termed the model, which evaluated adherence at 2 years after randomization, the maintenance model. We evaluated the factors in the multivariate logistic regression model simultaneously while adjusting for other covariates in the model. Maximum rescaled R2, C-statistic, and Hosmer-Lemeshow tests evaluated the goodness of fit of the logistic regression models. All p values were 2-tailed. Because of the large number of comparisons, p ≤.01 was considered statistically significant.

We performed the analyses using a hierarchical stage approach that moves from the person to the context. Stage 1 consisted of intrapersonal variables. We added interpersonal variables in stage 2, treatment variables in stage 3, and organizational variables in stage 4. We made adjustments by clinic to account for operational differences among the centers. We examined parameter estimates between stages for robustness and colinearity.

RESULTS

One year after randomization to CaD or placebo, approximately 61% of participants in this WHI clinical trial were adherent (≥ 80% of pills); after 2 years, approximately 63% were adherent. Table 2 presents means and variability of measures tested for associations with adherence.

TABLE 2.

Characteristics by Adherence Status

Characteristic CaD 1st-year adherence ≥ 80%
 CaD 2nd-year adherence ≥ 80%
No (N = 12,611)
 Yes (N = 19,771)
 p No (N = 12,083)
 Yes (N = 20,299)
 p
n % n % n % n %
Sociodemographic factors
 Age group at screening (years)
  50–55 2,207 17.5 2,310 11.7 < .0001 2,028 16.8 2,489 12.3 < .0001
  55–60 3,121 24.7 4,245 21.5 2,892 23.9 4,474 22.0
  60–65 3,008 23.9 5,082 25.7 2,893 23.9 5,197 25.6
  65–70 2,242 17.8 4,492 22.7 2,205 18.2 4,529 22.3
  70+ 2,033 16.1 3,642 18.4 2,065 17.1 3,610 17.8
 Race or ethnicity
  White 99,46 78.9 17,047 86.2 < .0001 9,526 78.8 17,467 86.0 < .0001
  African American 1,579 12.5 1,301 6.6 1,505 12.5 1,375 6.8
  Hispanic 645 5.1 696 3.5 6,40 5.3 701 3.5
  Asian or Pacific Islander 227 1.8 427 2.2 196 1.6 458 2.3
  Unknown 214 1.7 300 1.5 216 1.8 298 1.5
 College or higher
  No 8,007 63.5 12,467 63.1 .3428 7,714 63.8 12,760 62.9 .0735
  Yes 4,511 35.8 7,184 36.3 4,289 35.5 7,406 36.5
 Medical insurance
  No 790 6.3 922 4.7 < .0001 7,47 6.2 965 4.8 < .0001
  Yes 11,682 92.6 18,697 94.6 11,212 92.8 19,167 94.4
 Current health care provider
  No 1,056 8.4 1,464 7.4 .0013 1,009 8.4 1,511 7.4 .0033
  Yes 11,420 90.6 18,134 91.7 10,961 90.7 18,593 91.6
Health variables
 Prior history of CVD
  No 10,547 83.6 16,757 84.8 .0097 10,008 82.8 17,296 85.2 < .0001
  Yes 2,032 16.1 2,977 15.1 2,044 16.9 2,965 14.6
 Prior history of cancer
  No 11,984 95.0 18,822 95.2 .588 11,456 94.8 19,350 95.3 .352
  Yes 496 3.9 804 4.1 500 4.1 800 3.9
 Diabetes at baseline
  No 11,892 94.3 18,613 94.1 .6054 11,342 93.9 19,163 94.4 .0486
  Yes 717 5.7 1,151 5.8 737 6.1 1,131 5.6
 High risk (≥ 4) of osteoporosis
  No 9,460 75.0 14,404 72.9 < .0001 9,052 74.9 14,812 73.0 < .0001
  Yes 3,139 24.9 5,350 27.1 3,017 25.0 5,472 27.0
 Family history of breast cancer or colorectal cancer or MI
  No 4,489 35.6 6,850 34.6 .0615 4,237 35.1 7,102 35.0 .7508
  Yes 7,957 63.1 12,698 64.2 7,681 63.6 12,974 63.9
 Polyps removal prior to AV2
  No 12,201 96.7 19,173 97.0 .2524 11,673 96.6 19,701 97.1 .025
  Yes 410 3.3 598 3.0 410 3.4 598 2.9
 BMI
  < 25 3,239 25.7 5,273 26.7 .0952 3,037 25.1 5,475 27.0 .0004
  25–30 4,518 35.8 7,063 35.7 4,333 35.9 7,248 35.7
   ≥ 30 4,792 38.0 7,326 37.1 4,649 38.5 7,469 36.8
 On any special diet at baseline
  No 7,187 57.0 10,974 55.5 .0008 6,842 56.6 11,319 55.8 .1311
  Yes 5,424 43.0 8,796 44.5 5,241 43.4 8,979 44.2
 Total energy expenditure from physical activities ≥ 11 METs
  No 7,766 61.6 11,812 59.7 < .0001 7,423 61.4 12,155 59.9 < .0001
  Yes 3,840 30.4 6,971 35.3 3,709 30.7 7,102 35.0
 Current smoker vs. never or former smoker
  No 11,311 89.7 18,302 92.6 < .0001 10,882 90.1 18,731 92.3 < .0001
  Yes 1,148 9.1 1,285 6.5 1,053 8.7 1,380 6.8
 Current vs. past or nondrinker
  No 3,490 27.7 5,567 28.2 .4385 3,452 28.6 5,605 27.6 .0496
  Yes 9,011 71.5 14,093 71.3 8,534 70.6 14,570 71.8
 Take calcium supplement at AV1
  No 10,475 83.1 15,616 79.0 < .0001 10,023 83.0 16,068 79.2 < .0001
  Yes 1,970 15.6 3,952 20.0 1,893 15.7 4,029 19.8
Psychosocial variables
 Shortened CES-D/DIS ≥ .06
  No 10,750 85.2 17,564 88.8 < .0001 10,275 85.0 18,039 88.9 < .0001
  Yes 1,442 11.4 1,758 8.9 1,401 11.6 1,799 8.9
 Had ≥ 3 life events
  No 8,724 69.2 14,474 73.2 < .0001 8,318 68.8 14,880 73.3 < .0001
  Yes 3,608 28.6 4,914 24.9 3,486 28.9 5,036 24.8
 Emotion expressiveness ≥ 3
  No 802 6.4 1,397 7.1 .0178 774 6.4 1,425 7.0 .0397
  Yes 11,674 92.6 18,242 92.3 11,187 92.6 18,729 92.3
 Optimism ≥ 19
  No 1,074 8.5 1,474 7.5 .0002 1,071 8.9 1,477 7.3 < .0001
  Yes 11,196 88.8 17,974 90.9 10,693 88.5 18,477 91.0
 Hostility ≥ 8
  No 10,742 85.2 17,297 87.5 < .0001 10,206 84.5 17,833 87.9 < .0001
  Yes 1,331 10.6 1,842 9.3 1,352 11.2 1,821 9.0
 Take any supplement at AV1
  No 4,838 38.4 5,986 30.3 < .0001 4,593 38.0 6,231 30.7 < .0001
  Yes 7,607 60.3 13,582 68.7 7,323 60.6 13,866 68.3
 Take any medication at AV1
  No 2,774 22.0 4,123 20.9 .0122 2,579 21.3 4,318 21.3 .7942
  Yes 9,715 77.0 15,480 78.3 9,378 77.6 15,817 77.9
Interpersonal variables
 Social support ≥ 25
  No 1,188 9.4 1,696 8.6 .0058 1,192 9.9 1,692 8.3 < .0001
  Yes 11,104 88.1 17,686 89.5 10,598 87.7 18,192 89.6
 Social strain ≥ 5
  No 3,247 25.7 5,722 28.9 < .0001 3,027 25.1 5,942 29.3 < .0001
  Yes 9,054 71.8 13,683 69.2 8,759 72.5 13,978 68.9
 Caregiving construct (0, 1 scoring)
  No 7,202 57.1 11,371 57.5 .7135 6,854 56.7 11,719 57.7 .1476
  Yes 5,305 42.1 8,305 42.0 5,130 42.5 8,480 41.8
 Married or with partner vs. divorced, separated, widowed, or single
  No 4,787 38.0 6,942 35.1 < .0001 4,664 38.6 7,065 34.8 < .0001
  Yes 7,760 61.5 12,759 64.5 7,362 60.9 13,157 64.8
Procedural variables
 Had 4-week contact
  No 4,482 48.6 4,741 51.4 < .0001 4,274 46.3 4,949 53.7 < .0001
  Yes 8,129 35.1 15,030 64.9 7,809 33.7 15,350 66.3
 CaD 1st-year pill type
  Chewable 6,611 52.4 8,532 43.2 < .0001 — — — — —
  Mixed 2,358 18.7 2,814 14.2 — — — —
  Swallowable 3,591 28.5 8,425 42.6 — — — —
 CaD 2nd-year pill type
  Chewable — — — — — 5,019 41.5 6,534 32.2 < .0001
  Mixed — — — — 1,867 15.5 2,716 13.4
  Swallowable — — — — 5,053 41.8 11,049 54.4
 DM 1st-year adherence
  Not in DM trial 3,711 29.4 7,136 36.1 < .0001 3,506 29.0 7,341 36.2 < .0001
  DM 1st-year adherence: energy from fat < 25% 6,897 54.7 9,339 47.2 6,607 54.7 9,629 47.4
  DM 1st-year adherence: energy from fat ≥ 25% 2,003 15.9 3,296 16.7 1,970 16.3 3,329 16.4
 HT 1st-year adherence
  Not in HT trial 7,821 62.0 10,022 50.7 < .0001 7,592 62.8 10,251 50.5 < .0001
  HT 1st-year adherence < 80% 857 6.8 801 4.1 815 6.7 843 4.2
  HT 1st-year adherence ≥ 80% 3,933 31.2 8,948 45.3 3,676 30.4 9,205 45.3
 Bloating or gas
  No 4,240 33.6 7,111 36.0 < .0001 4,032 33.4 7,319 36.1 < .0001
  Yes 8,345 66.2 12,627 63.9 8,022 66.4 12,950 63.8
 Constipation
  No 8,219 65.2 13,725 69.4 < .0001 7,844 64.9 14,100 69.5 < .0001
  Yes 4,366 34.6 6,013 30.4 4,210 34.8 6,169 30.4
 Diarrhea
  No 9,377 74.4 14,870 75.2 .0938 8,941 74.0 15,306 75.4 .0071
  Yes 3,208 25.4 4,868 24.6 3,113 25.8 4,963 24.4
 Nausea
  No 11,111 88.1 17,849 90.3 < .0001 10,597 87.7 18,363 90.5 < .0001
  Yes 1,474 11.7 1,889 9.6 1,457 12.1 1,906 9.4
 Increasing appetite
  Missing 26 0.2 33 0.2 < .0001 29 0.2 30 0.1 < .0001
  No 8,136 64.5 13,726 69.4 7,716 63.9 14,146 69.7
  Yes 4,449 35.3 6,012 30.4 4,338 35.9 6,123 30.2
 Decreasing appetite
  No 11,666 92.5 18,473 93.4 .0018 11,131 92.1 19,008 93.6 < .0001
  Yes 919 7.3 1,265 6.4 923 7.6 1,261 6.2
 Heartburn
  No 7,685 60.9 12,701 64.2 < .0001 7,287 60.3 13,099 64.5 < .0001
  Yes 4,900 38.9 7,037 35.6 4,767 39.5 7,170 35.3
 Upset stomach
  No 8,494 67.4 14,288 72.3 < .0001 8,126 67.3 14,656 72.2 < .0001
  Yes 4,091 32.4 5,450 27.6 3,928 32.5 5,613 27.7
Open in a new tab

Note. Percentages may not add to 100% due to round-off error or missing values. AV1 = annual visit 1; AV2 = annual visit 2; BMI = body mass index; CED-D/DIS = Center for Epidemiological Studies Depression Scale; CaD = calcium and vitamin D clinical trial; CVD = cardiovascular disease; DM = dietary modification; HT = hormone treatment; MET = metabolic equivalent; MI = myocardial infarction.

Six variables (see Table 3) were associated with a significantly (p < .001) higher chance of adherence at 1 year, 4 of which were procedural variables. Completion of the semiannual visit in the clinic rather than by phone or mail nearly doubled the chance of adherence. Similarly, 72% of participants completed the recommended check-up phone call 4 weeks after randomization. These participants had a 37% greater chance of 80% CaD adherence at 1 year. Compared with women who were not in the hormone trial (HT; thus in DM and CaD only), 80% adherence to HT pills was associated with a 66% greater chance of adherence to CaD pills. Women taking the swallowable pill in year 1 had a 28% higher probability of 80% adherence compared with women taking the chewable pill (which was the only type available at the time of randomization for 58% of participants) and 51% higher probability of adherence compared with participants switching between the 2 pill forms.

TABLE 3.

CaD adherence at AV2 (1 Year Post-CaD Randomization), Initial Change Model

Effect Logistic regression model of CaD first-year adherence > 80%
p Odds ratio 95% Confidence limits
Sociodemographic variables
 Age group
  50–55 vs. 70–79 < .001 0.82 0.73–0.92
  55–60 vs. 70–79 .438 0.96 0.87–1.06
  60–65 vs. 70–79 .309 1.05 0.95–1.16
  65–70 vs. 70–79 .021 1.12 1.02–1.24
 Race or ethnicity
  African American vs. white < .001 0.62 0.55–0.70
  Hispanic vs. white .052 0.84 0.71–1.00
  Others or unknown vs. white .091 0.86 0.72–1.03
 College or higher vs. high school or less .084 1.06 0.99–1.12
 Had medical insurance
  Yes vs. no .019 1.18 1.03–1.35
 Married or with partner vs. divorced, separated, widowed, or single < .001 1.12 1.06–1.20
 Prior history of CVD
  Yes vs. no .349 0.96 0.89–1.04
 Prior history of cancer
  Yes vs. no .495 1.05 0.91–1.21
 Diabetes
  Yes vs. no .019 1.17 1.03–1.33
 High risk (≥ 4) of osteoporosis
  Yes vs. no .652 0.98 0.92–1.06
 Family history of breast cancer or colorectal cancer or MI
  Yes vs. no .298 1.03 0.97–1.10
 Shortened CES-D/DIS ≥ .06
  Yes vs. no .013 0.88 0.80–0.97
 Had > 3 life events
  Yes vs. no .162 0.95 0.89–1.02
 Emotion expressiveness > 3
  Yes vs. no .031 0.88 0.79–0.99
 Optimism ≥ 19
  Yes vs. no .825 1.01 0.91–1.13
 Hostility ≥ 8
  Yes vs. no .521 0.97 0.88–1.07
 On any special diet at baseline
  Yes vs. no .976 1.00 0.94–1.06
 Total expenditure from physical act ≥ 11
  Yes vs. no .024 1.07 1.01–1.14
 Current smoker vs. never or former smoker
  Yes vs. no < .001 0.77 0.69–0.86
 Current drinker vs. past or nondrinker
  Yes vs. no .06 0.94 0.88–1.00
 Take any supplement at AV1
  Yes vs. no < .001 1.24 1.16–1.33
 Taking any medication at AV1
  Yes vs. no .231 1.05 0.97–1.13
 Take calcium supplement at AV1
  Yes vs. no < .001 1.17 1.08–1.27
Psychosocial variables
 Social support ≥ 25
  Yes vs. no .81 0.99 0.89–1.10
 Social strain ≥ 5
  Yes vs. no .617 0.98 0.92–1.05
 Caregiving burden
  Yes vs. no .552 1.02 0.96–1.08
Treatment factors
 Total number of GI symptoms at baseline
  1 vs. none .003 0.89 0.82–0.96
  2 vs. none < .001 0.78 0.72–0.85
  3 vs. none < .001 0.76 0.68–0.84
 Had 4-week contact
  Yes vs. no < .001 1.37 1.27–1.48
 Contact type 6 months after CaD randomization
  Visit vs. phone or mail < .001 2.09 1.92–2.27
 CaD 1st-year pill type
  Chew vs. swallow < .001 0.72 0.67–0.77
  Mix vs. swallow < .001 0.49 0.45–0.53
 1st-year DM adherence
  Adherent in DM vs. not in DM .004 1.17 1.05–1.31
  Not adherent in DM vs. not in DM .707 1.02 0.93–1.11
 1st-year HT adherence
  < 80% vs. not in HT .06 0.87 0.75–1.01
  ≥ 80% vs. not in HT < .001 1.66 1.53–1.81
 Change clinic from CaD randomization to CaD1
  Yes vs. no .17 0.69 0.40–1.18
Open in a new tab

Note. Max-rescaled R2 = 0.11; AV1 = annual visit 1; C statistics = 0.670; Hosmer-Lemeshow goodnessof-fit test p = .3118. BMI = body mass index; CED-D/DIS = Center for Epidemiological Studies Depression Scale; CaD = calcium and vitamin D clinical trial; CVD = cardiovascular disease; DM = dietary modification; HT = hormone treatment.

Among sociodemographic variables, African American participants had a lower chance of adherence by nearly 40%. Regardless of ethnicity, participants aged 50 to 54 years and 65 to 69 years were less likely than 70- to 79-year-olds to be adherent (by 18% and 12%, respectively). Adherence rates were 12% higher overall if married or living with partner and 18% higher for participants with health insurance.

Among health variables, personal use of a supplement (either any supplement or calcium) was associated with a greater chance of adherence by 24% and 17%, respectively. Increased adherence was more likely in diabetic (17%) or physically active (7%) women. Gastrointestinal symptoms (eg, gas, bellyache, constipation) at baseline were associated with an 11% to 24% greater chance of nonadherence compared with participants not reporting symptoms in this category. Current smokers (23%) and drinkers (6%) were less likely to be adherent than the never or former groups combined.

Among the psychosocial variables reflecting emotional status, only depression was associated (negatively) with adherence (12%).

In the second year of the trial (ie, the maintenance phase), logistical regression results for adherence largely paralleled the results for the first year of the trial (ie, initial change). In year 1 but not in year 2, the following factors were significantly related to lower adherence: the youngest (50–54) compared with the oldest (70–79) age group, elevated depression symptoms, the chewable CaD study pill, and elevated negative emotional expressiveness. In year 2 but not in year 1, the following factors were significantly associated with lower adherence: higher social strain (9%) and higher number of negative life events (10%). In year 1 but not in year 2, a diagnosis of diabetes, higher levels of physical activity, and adherence in the DM trial signaled better adherence.

To address missing data, we calculated the inverse selection probability weighted estimator.34,35 For the CaD first-year adherence, the 2 factors with odds ratios that changed most were ethnicity (Hispanic compared with white dropped from 0.82 [95% confidence interval (CI) = 0.68–0.99] to 0.71 [95% CI = 0.62–0.81]) and age (60–65 compared with 70–79 years, which increased from 1.14 [95% CI = 1.03–1.27] to 1.23 [95% CI = 1.14–1.33]). Because the remaining weighted estimates and their significance levels are fairly close to those from the primary data analysis (ie, excluding participants with missing values), the full results from the weighted estimates are not shown.

Participants assessed their own adherence semiannually as part of the safety interview. We cross-tabulated this subjective assessment with objective pill counts. About 90% of participants who were 80% adherent said they took the pills every day and missed fewer than 10 days in the past month, compared with about 55% of nonadherent participants. Of objectively adherent participants, 51.8% reported taking all pills every day.

COMMENT

Despite a historically perceived ease regarding adherence to pharmacological interventions, data indicate that compliance ranges from 40% to 75%.36 The perception of ease may arise from underappreciating the fact that taking medications is a behavior. After adopting a behavioral approach, as WHI CaD Trial researchers did by increasing participant contact, adherence increased.

Furthermore, being relatively older, white, married or living with partner, and having health insurance was predictive of adherence in this trial and in previous studies.37–39 Our findings reinforced previously documented challenges with adherence among racial minorities and women at a higher health risk (ie, smoking, inactivity, and prior health problems), which indicates a clustering of high-risk health behaviors.40 The finding of better adherence associated with personal supplement use and with hormone trial adherence may have reflected participants’ greater inclination or ability to take pills or adopt positive health habits.

In a smaller trial (n = 107) lasting 6 months of 1,260 mg daily calcium and 1,000 IU vitamin D supplementation from 4 caplets (M age = 76 years, SD = 5.6), 60.7% of the participants were 80% adherent,39 which was nearly identical to the WHI CaD trial rate of 60% of the participants being 80% adherent at year 1. Those researchers found that higher education and income, more alcoholic drinks, and a history of fracture were directly predictive of significantly higher adherence, whereas minority status was indirectly linked to lower adherence through socioeconomic level and no history of hip fracture.

Some findings have suggested that a clinical diagnosis of depression (not self-reported symptoms alone) is necessary to affect adherence,41 but we found lower levels of adherence associated with reporting of depressive symptoms at baseline. As with studies showing social support positively associated with medication adherence,42 we found better adherence with higher social functioning and lower social strain.

Several procedural predictors suggested ways of improving clinical trial adherence. A simple follow-up call made soon after starting intervention significantly improved adherence. Dunbar-Jacob and Schlenk43 found steep declines in adherence at onset of treatment, suggesting that both early and long-term support is important. An in-person visit at the semiannual contact point—compared with contact by phone or mail—was the strongest predictor of 1-year adherence. Adherence rates seemed to increase with more intensive and more direct personal support.44,45 Last, the type of pill (swallowed, not chewed) benefited adherence, which highlights how ease of treatment benefits adherence.

Our results suggest that self-reported adherence plays a role in predicting adherence rates when study conditions exclude pill counts. Asking whether study pills were taken every day widely separated participants who were at least 80% adherent from those who were not. Although forgetting was the most frequently reported cause of poor adherence,46 other reported reasons did not distinguish the adherent and nonadherent groups, so reminders (eg, labeled pill containers and calendars) can play a role in adherence.

A randomized, placebo-controlled trial does not tap the full range of factors affecting adherence. Active involvement in one’s own treatment decisions improves adherence,47 possibly through an increase in self-efficacy,48 and this is limited by strict trial protocols. Also, the use of a placebo may reduce the sense that the treatment has personal health value and thus limit it as an adherence promoter. Nevertheless, the lessons we learned from this analysis may have applications for future studies of nutritional supplements and may be tested in trials and clinical settings where there is an interest in improving adherence to prescribed treatments.

Acknowledgments

The WHI program is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, Department of Health and Human Services.

Contributor Information

Dr. R. Brunner, Department of Family and Community Medicine at the University of Nevada School of Medicine in Reno.

Dr. J. Dunbar-Jacob, University of Pittsburgh School of Nursing in Pittsburgh, PA.

Dr. M. S. LeBoff, Skeletal Health and Osteoporosis Center and Bone Density Unit at Brigham and Women’s Hospital in Boston, MA.

Dr. I. Granek, Department of Preventive Medicine at Stony Brook University in Stony Brook, NY.

Dr. D. Bowen, Department of Social and Behavioral Sciences at the Boston University School of Public Health in Boston, MA.

Dr. L. G. Snetselaar, Department of Epidemiology at the University of Iowa College of Public Health in Iowa City.

Dr. S. A. Shumaker, Department of Public Health Sciences at the Wake Forest University School of Medicine in Winston-Salem, NC.

Dr. J. Ockene, UMMC Division of Preventive and Behavioral Medicine at the University of Massachusetts in Worcester.

Dr. M. Rosal, UMMC Division of Preventive and Behavioral Medicine at the University of Massachusetts in Worcester.

Dr. J. Wactawski-Wende, Department of Social and Preventive Medicine at the State University of New York at Buffalo.

Dr. J. Cauley, Department of Epidemiology at the University of Pittsburgh in Pittsburgh, PA.

Dr. B. Cochrane, Family and Child Nursing Department at the University of Washington in Seattle.

Dr. L. Tinker, Women’s Health Initiative at the Fred Hutchinson Cancer Research Center in Seattle, WA.

Dr. R. Jackson, Division of Endocrinology, Diabetes, and Metabolism at Ohio State University College of Medicine in Columbus.

Dr. C. Y. Wang, Women’s Health Initiative at the Fred Hutchinson Cancer Research Center in Seattle, WA.

Ms. L. Wu, Women’s Health Initiative at the Fred Hutchinson Cancer Research Center in Seattle, WA.

References

  • 1.Jackson RD, LaCroix AZ, Gass M, et al. Calcium plus vitamin D supplementation and the risk of fractures. New Engl J Med. 2006;16(354):669–683. doi: 10.1056/NEJMoa055218. [DOI] [PubMed] [Google Scholar]
  • 2.DiMatteo MR, Giordani PJ, Lepper HS, Croghan TW. Patient adherence and medical treatment outcomes: a meta-analysis. Med Care. 2002;40:794–811. doi: 10.1097/00005650-200209000-00009. [DOI] [PubMed] [Google Scholar]
  • 3.Barr RG, Somers SC, Speizer FE, Camargo CA. Patient factors and medication guideline adherence among older women with asthma. Arch Intern Med. 2002;162:12–26. doi: 10.1001/archinte.162.15.1761. [DOI] [PubMed] [Google Scholar]
  • 4.Yarcheski A, Mahon NE, Yarcheski TJ, Cannella BL. A metaanalysis of predictors of positive health factors. J Nurs Scholarsh. 2004;36:102–108. doi: 10.1111/j.1547-5069.2004.04021.x. [DOI] [PubMed] [Google Scholar]
  • 5.DiMatteo MR. Social support and patient adherence to medical treatment. Health Psychol. 2004;23:207–218. doi: 10.1037/0278-6133.23.2.207. [DOI] [PubMed] [Google Scholar]
  • 6.Lumley J, Oliver SS, Chamberlain C, Oakley L. Interventions for promoting smoking cessation during pregnancy. Cochrane Database Syst Rev. 2004:CD001055. doi: 10.1002/14651858.CD001055.pub3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Gillis AJ. Determinants of a health-promoting lifestyle: an integrative review. J Adv Nurs. 1993;18:345–353. doi: 10.1046/j.1365-2648.1993.18030345.x. [DOI] [PubMed] [Google Scholar]
  • 8.May S, West R. Do social support interventions (“buddy systems”) aid smoking cessation? A review. Tob Control. 2000;9:415–422. doi: 10.1136/tc.9.4.415. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Lewis BA, Marcus BH, Pate RR, Dunn AL. Psychosocial mediators of physical activity behavior among adults and children. Am J Prev Med. 2002;23:26–35. doi: 10.1016/s0749-3797(02)00471-3. [DOI] [PubMed] [Google Scholar]
  • 10.Ciechanowski PS, Katon WJ, Russo JE. Depression and diabetes: impact of depressive symptoms on adherence, function, and costs. Arch Intern Med. 2000;160:3278–3285. doi: 10.1001/archinte.160.21.3278. [DOI] [PubMed] [Google Scholar]
  • 11.Ciechanowski PS, Katon WJ, Russo JE, Hirsch IB. The relationship of depressive symptoms to symptom reporting, self-care and glucose control in diabetes. Gen Hosp Psychiatry. 2003;25:246–252. doi: 10.1016/s0163-8343(03)00055-0. [DOI] [PubMed] [Google Scholar]
  • 12.DiMatteo MR, Lepper HS, Croghan TW. Depression is a risk factor for noncompliance with medical treatment. Arch Intern Med. 2000;160:2101–2107. doi: 10.1001/archinte.160.14.2101. [DOI] [PubMed] [Google Scholar]
  • 13.Zigelstein RC, Fauerbach JA, Stevens SS, Romanelli J, Richter DP, Bush DE. Patients with depression are less likely to follow recommendations to reduce cardiac risk during recovery from a myocardial infarction. Arch Intern Med. 2000;160:1818–1823. doi: 10.1001/archinte.160.12.1818. [DOI] [PubMed] [Google Scholar]
  • 14.Wang PS, Bohn RL, Knight E, Glynn RJ, Mogun H, Avorn J. Noncompliance with antihypertensive medications: the impact of depressive symptoms and psychosocial factors. J Gen Intern Med. 2002;17:504–511. doi: 10.1046/j.1525-1497.2002.00406.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Jokisalo E, Enlund H, Halonen P, Takala J, Kumpusalo E. Factors related to poor control of blood pressure with antihypertensive drug therapy. Blood Press. 2003;12:49–55. [PubMed] [Google Scholar]
  • 16.Dunbar-Jacob J, Mortimer-Stephens MK. Treatment adherence in chronic disease. J Clin Epidemiol. 2001;54:S57–S60. doi: 10.1016/s0895-4356(01)00457-7. [DOI] [PubMed] [Google Scholar]
  • 17.The Women’s Health Initiative Study Group. Design of the women’s health initiative clinical trial and observational study. Control Clin Trials. 1998;19:61–109. doi: 10.1016/s0197-2456(97)00078-0. [DOI] [PubMed] [Google Scholar]
  • 18.Anderson GL, Manson J, Wallace R, et al. Implementation of the women’s health initiative study design. Ann Epidemiol. 2003;13:S5–S17. doi: 10.1016/s1047-2797(03)00043-7. [DOI] [PubMed] [Google Scholar]
  • 19.Jackson RD, LaCroix AZ, Cauley JA, McGowan J. The women’s health initiative calcium–vitamin D trial: overview and baseline characteristics. Ann Epidemiol. 2003;13:S98–S106. doi: 10.1016/s1047-2797(03)00046-2. [DOI] [PubMed] [Google Scholar]
  • 20.Radloff LS. The CES-D scale: a self-report depression scale for research in the general population. App Psychol Meas. 1977;1:385–401. [Google Scholar]
  • 21.Burnam MA, Wells KB, Leake B, Landsverk J. Development of a brief screening instrument for detecting depressive disorders. Med Care. 1988;26:775–789. doi: 10.1097/00005650-198808000-00004. [DOI] [PubMed] [Google Scholar]
  • 22.Scheier MF, Carver CS, Bridges MW. Distinguishing optimism from neuroticism (and trait anxiety, self-mastery, and self-esteem): a re-evaluation of the life orientation test. J Pers Soc Psychol. 1994;67:1063–1078. doi: 10.1037//0022-3514.67.6.1063. [DOI] [PubMed] [Google Scholar]
  • 23.Barefoot J, Dodge K, Peterson B, Dahlstrom W, Williams R. The Cook-Medley hostility scale: item content and ability to predict survival. Psychosom Med. 1989;51:46–57. doi: 10.1097/00006842-198901000-00005. [DOI] [PubMed] [Google Scholar]
  • 24.King LA, Emmons RA. Conflict over emotional expression: psychological and physical correlates. J Pers Soc Psychol. 1990;58:864–877. doi: 10.1037//0022-3514.58.5.864. [DOI] [PubMed] [Google Scholar]
  • 25.Sherbourne CD, Stewart AL. The MOS social support survey. Soc Sci Med. 1991;32:705–714. doi: 10.1016/0277-9536(91)90150-b. [DOI] [PubMed] [Google Scholar]
  • 26.Rook KS. The negative side of social interaction: impact on psychological well-being. J Pers Soc Psych. 1984;46:1097–1108. doi: 10.1037//0022-3514.46.5.1097. [DOI] [PubMed] [Google Scholar]
  • 27.Berkman LF, Syme L. Social networks, host resistance and mortality: a nine-year follow-up study of Alameda County residents. Am J Epidemiol. 1979;109:186–204. doi: 10.1093/oxfordjournals.aje.a112674. [DOI] [PubMed] [Google Scholar]
  • 28.Hays RD, Sherbourne CD, Mazel RM. The RAND 36-Item Health Survey 1.0. Health Econ. 1993;2:217–227. doi: 10.1002/hec.4730020305. [DOI] [PubMed] [Google Scholar]
  • 29.Hays J, Ockene JK, Brunner RL, et al. Effects of estrogen plus progestin on health-related quality of life: results from the women’s health initiative randomized clinical trial. N Engl J Med. 2003;348:1839–1854. doi: 10.1056/NEJMoa030311. [DOI] [PubMed] [Google Scholar]
  • 30.Ainsworth BE, Haskell WL, Leon AS, et al. Compendium of physical activities: classification of energy costs of human physical activities. Med Sci Sport Exerc. 1993;25:71–80. doi: 10.1249/00005768-199301000-00011. [DOI] [PubMed] [Google Scholar]
  • 31.Matthews KA, Wing RR, Kuller LH, Meilahn EN, Plantinga P. Influence of the perimenopause on cardiovascular risk factors and symptoms of middle-aged healthy women. Arch Intern Med. 1994;154:2349–2355. [PubMed] [Google Scholar]
  • 32.Wang CY, Wang S, Zhao LP, Ou ST. Weighted semiparametric estimation in regression analysis with missing covariate data. J Am Stat Assoc. 1997;92:512–525. [Google Scholar]
  • 33.Robins JM, Rotnitzky A, Zhao LP. Estimation of regression coefficients when some regressors are not always observed. J Am Stat Assoc. 1994;89:846–866. [Google Scholar]
  • 34.Schechtman KB, Gordon MO. The effect of poor compliance and treatment side effects on sample size requirements in randomized clinical trials. J Biopharm Stat. 1994;4:223–232. doi: 10.1080/10543409408835085. [DOI] [PubMed] [Google Scholar]
  • 35.Opolka JL, Rascati KL, Brown CM, Gibson PJ. Role of ethnicity in predicting antipsychotic medication adherence. Ann Pharmacother. 2003;37:625–630. doi: 10.1345/aph.1C321. [DOI] [PubMed] [Google Scholar]
  • 36.Ellis S, Shumaker S, Sieber W, Rand C. Adherence to pharmacological interventions: current trends and future directions. Control Clin Trials. 2000;21:S218–S225. doi: 10.1016/s0197-2456(00)00082-9. [DOI] [PubMed] [Google Scholar]
  • 37.Charles H, Good CB, Hanusa BH, Chang CC, Whittle J. Racial differences in adherence to cardiac medications. J Natl Med Assoc. 2003;95:17–27. [PMC free article] [PubMed] [Google Scholar]
  • 38.Van Horn LV, Dolecek TA, Grandits GA, et al. Adherence to dietary recommendations in the special intervention group in the Multiple Risk Factor Intervention Trial. Am J Clin Nutr. 1997;65:S289–S304. doi: 10.1093/ajcn/65.1.289S. [DOI] [PubMed] [Google Scholar]
  • 39.Unson CG, Litt M, Reisine S, Mahoney-Trella P, Sheperd T, Prestwood K. Adherence to calcium/vitamin D and estrogen protocols among diverse older participants enrolled in a clinical trial. Contemp Clin Trials. 2006;27:215–226. doi: 10.1016/j.cct.2006.02.006. [DOI] [PubMed] [Google Scholar]
  • 40.DiMatteo MR, Giordani PJ, Lepper HS, Croghan TW. Patient adherence and medical treatment outcomes: a meta-analysis. Med Care. 2002;40:794–811. doi: 10.1097/00005650-200209000-00009. [DOI] [PubMed] [Google Scholar]
  • 41.Fine LJ, Philogene GS, Gramling R, Coups EJ, Sinha S. Prevalence of multiple chronic disease risk factors. 2001 National Health Interview Survey. Am J Prev Med. 2004;27:18–24. doi: 10.1016/j.amepre.2004.04.017. [DOI] [PubMed] [Google Scholar]
  • 42.Morse EV, Simon PM, Coburn M, Hyslop N, Greenspan D, Balson PM. Determinants of participant compliance within an experimental anti-HIV drug protocol. Soc Med. 1991;32:1161–1167. doi: 10.1016/0277-9536(91)90093-r. [DOI] [PubMed] [Google Scholar]
  • 43.Dunbar-Jacob J, Schlenk E. Patient adherence to treatment regimen. In: Baum A, Revenson TA, Singer JE, editors. Handbook of Health Psychology. Hillsdale, NJ: Lawrence Erlbaum; 2001. pp. 571–580. [Google Scholar]
  • 44.Haynes RB, McKibbon KA, Kanani R. Systematic review of randomized trials of interventions to assist patients to follow prescriptions for medications. Lancet. 1996;348:383–386. doi: 10.1016/s0140-6736(96)01073-2. [DOI] [PubMed] [Google Scholar]
  • 45.Gold DT, Silverman S. Review of adherence to medications for the treatment of osteoporosis. Current Osteoporos Rep. 2006;4:21–27. doi: 10.1007/s11914-006-0011-8. [DOI] [PubMed] [Google Scholar]
  • 46.Dunbar-Jacob J. Understanding the reasons for patients’ noncompliance and how these reasons impact therapeutic regimens and outcomes of care. Paper presented at: Pharmaceutical Care Programs: Their Role in Medication Compliance by EMMO; November 1997; Kansas City, KS. [Google Scholar]
  • 47.Rejeski WJ, Brawley LR, Ettinger W, et al. Compliance to exercise therapy in older participants with knee osteoarthritis: implications for treating disability. Med Sci Sport Exerc. 1997;29:977–985. doi: 10.1097/00005768-199708000-00001. [DOI] [PubMed] [Google Scholar]
  • 48.Culos-Reed SN, Rejeski WJ, McAuley E, Ockene JK, Roter DL. Predictors of adherence to behavior change interventions in the elderly. Control Clin Trials. 2000;21:S200–S205. doi: 10.1016/s0197-2456(00)00079-9. [DOI] [PubMed] [Google Scholar]

RESOURCES