SEARCH

SEARCH BY CITATION

Keywords:

  • BONE FRACTURES;
  • BONE MINERAL DENSITY;
  • DEPRESSION;
  • OSTEOPOROSIS;
  • TRICYCLIC ANTIDEPRESSANTS

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

Because studies of the association between tricyclic antidepressant (TCA) treatment and risk of fracture have shown inconsistent findings, we sought to assess whether people who take TCAs are at increased risk of fracture. Relevant studies published by June 2012 were identified through database searches of Scopus, MEDLINE, EMBASE, PsycINFO, ISI Web of Science, and WorldCat Dissertations and Theses from their inception, and manual searching of reference lists. Only original studies that examined the association between TCA treatment and risk of fracture were included. Two investigators independently conducted literature searches, study selection, study appraisal, and data abstraction using a standardized protocol. Disagreements were resolved by consensus. Twelve studies met inclusion criteria. Because of the heterogeneity of these studies, random-effects models were used to pool estimates of effect. Overall, TCA use was associated with significantly increased fracture risk (relative risk [RR], 1.45; 95% confidence interval [CI], 1.31–1.60; p < 0.001). Increased fracture risk associated with TCA use was also observed in studies that adjusted for bone mineral density (RR, 1.54; 95% CI, 1.24–1.90; p < 0.001) or depression (RR, 1.49; 95% CI, 1.28–1.67; p < 0.001). Strength of association with TCA exposure duration ≥6 weeks (RR, 1.13; 95% CI, 1.00–1.28) was substantially weaker than association with TCA exposure duration <6 weeks (RR, 2.40; 95% CI, 1.41–4.08). Prior TCA exposure had no significant effect on fracture risk (RR, 1.04; 95% CI, 0.86–1.26; p = 0.70). After accounting for publication bias, we found the overall association between TCA use and fracture risk to be slightly weaker (RR, 1.36; 95% CI, 1.24–1.50) but still significant (p < 0.001). Findings of this meta-analysis indicate that treatment with TCAs may convey an increased risk of fracture, independent of depression and bone mineral density. © 2013 American Society for Bone and Mineral Research.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

Osteoporotic fractures have become a major public health problem worldwide, especially with the global growth in aging populations. By the year 2050, the incidence of hip fracture is expected to double worldwide.1 Fractures often cause severe disability and excess mortality, especially in elderly persons.2 Osteoporotic fractures have been and will continue to be contributors to increased global social and economic burdens.

Depression is common, affecting 26% of women and 18% of men in the US population.3 It is most often treated with antidepressants in a primary care setting. Antidepressants are one of the most commonly prescribed drugs in the Western world. Tricyclic antidepressants (TCAs) and selective serotonin reuptake inhibitors (SSRIs) are two of the most widely prescribed classes of antidepressants. Our previous meta-analyses have demonstrated that depression is associated with lower bone density4 and higher fracture risk,5 and that SSRI treatment is associated with increased fracture risk.6 Treatment with TCAs may also contribute to increased fracture risk because TCAs are highly anticholinergic, with adverse effects that include sedation and postural instability.7, 8 A number of epidemiologic studies have examined the association of TCA treatment and osteoporotic fractures9–20; however, these results are controversial. Some studies have found an association between TCA treatment and increased fracture risk,10, 13–17, 19 whereas other studies failed to find a significant association.9, 11, 12 The association between TCA use and fracture risk reportedly differs depending on dose,11, 13, 16, 18, 19 exposure duration,11, 13, 14, 17 and time of exposure.14–16 Although these inconsistencies have been explored in several narrative reviews,21–23 no meta-analyses or high-quality systematic reviews have been reported.

The aims of this study were to quantitatively assess all qualified studies that have examined the effect of TCAs on fracture risk and to gather more accurate and precise information about this effect. We hypothesized that TCA treatment would be associated with an increased risk of fractures. In addition, we examined whether the effect varied by sex, study design, anatomical site of fracture, dose, exposure duration, geographic location, or adjustment for bone mineral density (BMD) or depression.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement and the Meta-Analysis of Observational Studies in Epidemiology (MOOSE) guidelines were followed.24, 25 The objectives, primary outcomes, search strategy, inclusion criteria, and methods for study selection, data extraction, and data synthesis of this meta-analysis were defined in advance in a protocol. Data fields were predefined, and sensitivity analysis and subgroup analysis were also prespecified in the protocol.

Search strategy and data sources

A comprehensive literature search of MEDLINE (from 1946) using OVID, without language restrictions, was conducted using the following search terms: tricyclic antidepressants, antidepressant, amitriptyline, nortriptyline, protriptyline, imipramine, desipramine, doxepin, trimipramine, fractures, osteoporosis, osteopenia, bone density, and bone (Supporting Information). Using the same strategy, we also searched EMBASE (from 1988), PsycINFO (from 1806), ISI Web of Science (from 1975), and SCOPUS (from 1960). The last literature search was conducted on May 29, 2012. With Ovid AutoAlert, the MEDLINE literature search was automatically updated to August 18, 2012. We also searched WorldCat Dissertations and Theses, the conference abstracts of the American Society for Bone and Mineral Research, and the proceedings of the International Osteoporosis Foundation World Conference on Osteoporosis from 2000 to 2011. Medical librarians were consulted during the literature search. Two investigators (QW and WQ) independently manually examined reference lists from the original studies9–20 and from the review articles21–23 obtained during the electronic search.

Study selection

In the initial screening stage, simple relevance criteria were used for study selection: (1) human participants, (2) use of TCA as exposure, and (3) BMD and/or fracture as outcome. Each title and each abstract of articles obtained through the electronic search were independently reviewed by two investigators (QW and WQ), who excluded only those citations they both deemed obviously irrelevant. Citations with disagreement were included for full review.

In the second stage of study selection, the full content of each article obtained during the previous stage was reviewed and evaluated. We included cohort and case-control studies that reported data on individuals with a fracture or BMD who were either exposed to TCAs or not exposed to antidepressants. No related randomized controlled trials were found. Cross-sectional studies were excluded. Using predetermined selection criteria and assessment methods, two investigators (QW and WQ) independently evaluated the full content of each English-language article. Articles in other languages were reviewed and evaluated by multilingual investigators using the same criteria and assessment methods. Areas of uncertainty or disagreement were resolved by consensus. We included studies that reported the hazard ratio (HR), relative risk (RR), or odds ratio (OR) of fracture or change in BMD associated with TCA use.

Study appraisal

We assessed the methodological quality of included studies using the Newcastle-Ottawa Scale.26 As recommended by the MOOSE study group,24 the quality scores were not used as weights in the analyses; instead, they were used in the sensitivity analysis by excluding articles with a score of <7.0 (low quality).

Data abstraction

The data abstraction form was adapted and modified from our previous meta-analysis.6 All data were independently abstracted by two investigators (QW and WQ) using the standard data abstraction form and protocol. No major discrepancies or disagreements arose between abstractors; minor discrepancies or errors were resolved by discussion and by rechecking the original reports. The following information from each study was extracted: study characteristics (name of first author, publishing year, journal, country where study was conducted, study setting and design, inclusion and exclusion criteria, total number of participants, incident case and duration of follow-up in cohort studies, and number of cases and controls in case-control studies); participants' characteristics (age, sex, and race, if available); ascertainment of outcomes (fracture and bone loss) and corresponding regions; assessment of exposure (TCA use); analysis strategy; and risk estimates (adjusted RR, OR, and HR and 95% confidence intervals [CIs]).

For overall pooled analysis, when multiple estimates were presented in the original reports, we chose to use the estimate from the TCA level closest to one defined daily dose; the estimate that adjusted for the largest number of confounders; the estimate from osteoporotic fracture; and the estimate from current TCA users, if applicable. Other estimates may be used in the corresponding subgroup analyses. We did not contact the authors because no further information was needed.

Statistical analysis

The confounder-adjusted RR was used as a measure of the association between TCA use and fracture risk. The OR was used as a surrogate measure of the corresponding RR in case-control studies; because the absolute risk of fracture is low, the ORs approximated the RRs. For studies that reported outcomes stratified by subgroups, the effect size across subgroups in each individual report was estimated with meta-analysis. For stabilization of the variance and normalization of the distributions, HRs or RRs were transformed into their natural logarithms27 before the data were pooled. The variance of the natural logarithm of the HR or RR was derived from corresponding 95% CIs provided in the original studies. To calculate the overall estimate, we weighted each report by the reciprocal of its variance.

To estimate the robustness of our findings, we conducted multiple prespecified sensitivity analyses. The effects of TCAs were examined by the fracture definition, adjustment of important confounders, and quality score. We conducted one post hoc sensitivity analysis by including studies only with elderly participants (aged ≥65 years). To assess heterogeneity, we used the Cochran Q statistic28 and the Higgins I2 index.29 In light of existing heterogeneity, we present herein the overall effect obtained with the random-effects model.30

We conducted a number of prespecified subgroup analyses to assess whether the effect of TCAs on fracture risk was modified by demographic or clinical variables. These variables included sex, study design, study location, TCA dose, TCA exposure duration, TCA exposure time, and whether the study controlled for depression or BMD. We were not able to conduct subgroup analysis for race/ethnicity because most studies did not specify race/ethnicity and others used multiethnic populations. We did not perform multivariate meta-regression analysis because of the limited number of studies that qualified for this meta-analysis and the unavailability of certain key variables, such as sex and race/ethnicity, in the original reports.

We examined the potential for publication bias by constructing a funnel plot, in which log RRs were plotted against their standard errors.31 In addition, the rank correlation test of Begg and Mazumdar32 was used to test for significant publication bias. Furthermore, we used the trim-and-fill method to estimate and adjust for the potential effects that unpublished studies might have had on the measured outcome. The trim-and-fill method is a funnel-plot-based method of identifying and adjusting for publication bias in a meta-analysis.33 This method requires no assumptions about the mechanisms that lead to publication bias. Although it may underestimate the true effect size when no publication bias exists,34 the trim-and-fill method is recommended for routine use in meta-analyses.35

For the data analyses, we used Stata 10.0 (StataCorp, College Station, TX, USA) statistical software. Statistical significance was set at a p value of 0.05 or less.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

Study selection flow is illustrated in Fig. 1. After removing duplicate publications, we identified a total of 1539 potential articles. On the initial screening of titles and abstracts by two investigators (QW and WQ), 41 full-text articles were retrieved and assessed for eligibility. Agreement between the two investigators was modest at this initial stage (κ = 0.66). In the second stage of reviewing full-text articles, 12 studies with fracture data9–20 and one study with BMD data36 met the inclusion criteria. Agreement between the two investigators was good at this second stage (κ = 0.90). The study with BMD data was finally dropped because meta-analysis could not be conducted on only one study. The remaining studies (n = 12) included five cohort studies and seven case-control studies. No related randomized controlled trials were found. Reports on all 12 studies were published in English.

thumbnail image

Figure 1. Study selection for meta-analysis. BMD = bone mineral density; TCAs = tricyclic antidepressants.

Download figure to PowerPoint

The characteristics of the study participants and the designs of the cohort studies are summarized in Table 1. Of the five cohort studies, four were conducted on persons aged ≥65 years. The mean follow-up period ranged from 4.1 years to 10 years. The potential confounding effects of age and sex (if applicable) were controlled for in all five cohort studies, whereas BMD was controlled for in only three cohort studies. All seven case-control studies14–20 were conducted on mixed-sex groups; six of the seven were conducted outside the United States. Matching and/or adjustments were performed for various potential confounders (Table 2).

Table 1. Characteristics of Five Cohort Studies on the Association of Tricyclic Antidepressants With Fracture Risk
Author, publication year, locationStudy populationExposure assessmentOutcomes assessmentOutcomesMean follow-up (years)Variables controlled
  1. US = United States; TCA = tricyclic antidepressant; BMD = bone mineral density; MrOS = Osteoporotic Fractures in Men Study; IADL = instrumental activities of daily living; m-MMSE = modified version of the Mini-Mental State Examination; PPI = proton pump inhibitors; GDS = Geriatric Depression Scale; UK = United Kingdom; DM = diabetes mellitus; HTN = hypertension; PD = Parkinson disease; OCD = obsessive-compulsive disorder; NSAIDs = nonsteroidal anti-inflammatory drugs.

Ensrud and colleagues9

(2003); US
8127 white women aged ≥65 years in Study of Osteoporotic Fractures; 316 used TCAInterview, with verification of TCA use from containersPhone interview or postcard; confirmed by radiographic reportNonspine fracture4.4Age, health status, use of ≥1 medication, walking for exercise, functional impairment, fall in previous year, cognitive function, weight change, gait speed, inability to rise from chair, femoral neck BMD
Hip fracture4.8

Lewis and colleagues10

(2007); US
5876 men aged ≥65 years in MrOS; 1.7% were TCA usersInterview by trained staffTriannual questionnaires by mail or telephone, verified by physician adjudication of medical records and radiograph reportsNonspine fracture4.1Age, BMD

Ziere and colleagues11

(2008); Netherlands
7983 men and women aged ≥55 years in Rotterdam Study; 25 were current TCA usersPharmacy; dispensing recordsReported by general practitioners through a computerized system or check of medical recordsNonvertebral fracture8.4Age, sex, depression during follow-up period, disability category, lower-limb disability

Diem and colleagues12

(2011); US
8412 women aged ≥70 years in Study of Osteoporotic Fractures; 340 were TCA usersInterview, with verification of TCA use from containersQuestionnaires by postcard or telephone every 4 months; verified by radiographic reportsNonspine fracture, first hip fracture, and wrist fracture10Age, health status, IADL, ability to rise from chair, m-MMSE, smoking, alcohol use, estrogen use, bisphosphonate use, benzodiazepine use, thiazide use, PPI use, oral steroid use, weight, GDS score, walks for exercise, history of prior fracture, total-hip BMD and history of falls in previous year

Coupland and colleagues13

(2011); UK
60,746 patients aged ≥65 years; 34.64% were TCA usersPrescription recordsPrimary care computer records and death certificatesFractures (limb, ribs, skull, vertebrae, pelvis)5Age, sex, depression, deprivation, smoking status, comorbidities (ischemic heart disease, DM, HTN, stroke, cancer, dementia, epilepsy or seizures, PD, hypothyroidism, OCD) and use of other drugs (eg, statins, NSAIDs, antipsychotics, lithium, aspirin, anti-HTN drugs, anticonvulsants, hypnotics, anxiolytics)
Table 2. Characteristics of Seven Case-Control Studies on the Association of Tricyclic Antidepressants With Risk of Fracture
Author, publication year, locationFracture casesControlsCase assessmentExposure assessmentControlled variables
  1. US = United States; TCA = tricyclic antidepressant; HTN = hypertension; UK = United Kingdom; GPRD = General Practice Research Database; PHARMO RLS = PHARMO Record Linkage System; PD = Parkinson disease; HRT = hormone replacement therapy; DM = diabetes mellitus; NSAID = nonsteroidal anti-inflammatory drug; DMARD = disease-modifying antirheumatic drug; IBD = inflammatory bowel disease.

Ray and colleagues14

(1987); US
1021 patients aged ≥65 years with hip fractures among Medicaid enrollees5606 randomly selected Medicaid enrollees aged ≥65 years with no hip fractureMedicaid diagnosis codeMedicaid prescriptionSex, race, age, index year, and home status, diagnosis of dementia

Ray and colleagues15

(1991); Canada
4501 persons aged ≥65 years with a fracture of the proximal femur with first hospitalization for hip fracture; 3.2% were TCA users24,041 hospitalized patients selected by random sampling; 1.8% were TCA usersHospital discharge diagnosisMedical records of prescriptionAge, sex, calendar year, nursing home residence on index date, and for hospitalization and use of specific medications (narcotic analgesics, anti-HTN, and other cardiovascular drugs) in year preceding index date

Liu and colleagues16

(1998); Canada
8239 men and women aged ≥66 years with a hip fracture in an Ontario acute care hospital; 953 were TCA users41,195 individuals in Registered Person Database; 3,197 were TCA usersClinical diagnosisPrescription claimAge, sex, comorbidity (eg, depression, dementia, osteoporosis), previous drug exposure (eg, sedative, tranquilizer, cardiac drug, anti-PD agent, thyroid-replacement drug, anticonvulsant, insulin, glucocorticoid, estrogen, etidronate)

Hubbard and colleagues17

(2003); UK
16,341 men and women with mean (SD) age of 79 (12) years with a hip fracture or fractured neck of femur in GPRD; 17.8% were TCA users29,889 patients in GPRD; 11.9% were TCA usersClinical diagnosisComputerized recordings of prescriptionAge, sex, general practice, duration of available GPRD data, history of falls, and history of prescriptions for hypnotics and antipsychotics

Vestergaard and colleagues18

(2006); Denmark
124,655 men and women with mean (SD) age of 43.4 (27.4) years with a fracture; 4774 were TCA users373,962 residents in Civil Registration System; 8948 were TCA usersRadiographic and clinical diagnosisPrescription history in databaseAge; sex; psychiatric comorbidity (eg, manic depression, schizophrenia, alcoholism, eating disorder); medication use (eg, anxiolytic, sedative, neuroleptic, corticosteroid, antiepileptic, cardiovascular agent, lithium); hospital stay; prior fracture; income; working, educational, and residential status; Charlson index

van den Brand and colleagues19

(2009); Netherlands
6763 men and women aged ≥18 years with hip or femur fracture in Dutch PHARMO RLS; 256 were TCA users26,341 hospitalized patients; 591 were TCA usersHospital record for a first fracture of the hip or femurReviewing prescription informationAge; sex; geographical region; other antidepressant; use of benzodiazepine, antipsychotic, lithium, anti-PD agent, anticonvulsant, oral-inhaled corticosteroid, bronchodilator, HRT, antiarrhythmic, thiazide diuretic, β-blocker, opioids, anticonvulsants, DM drug, ≥2 dispensings of an NSAID, DMARDs, metoclopramide hydrochloride; history of malignant neoplasm, mental disorder, cerebrovascular disease, obstructive airway disease, or IBD

Verdel and colleagues20

(2010); Netherlands
16,717 men and women aged ≥18 years with a fracture in PHARMO RLS61,517 patients in PHARMO RLSRecorded clinical diagnosis in databaseReviewing prescription informationAge; sex; geographical area; calendar time; cancer; cardiovascular disease; cerebrovascular disease; IBD; mental disorder; obstructive airway disease; use of antidiabetic, antiepileptic, anti-PD drug, antipsychotic, benzodiazepine, β-blocker, DMARD, HRT, NSAID, oral glucocorticoid, opioid

The Cochran Q statistic (p < 0.001) and the Higgins I2 index (77.1%) indicated that there was heterogeneity among the 12 studies. Figure 2 shows the RR (95% CI) of fractures associated with TCA use in each study and overall. Compared with patients who had taken no TCAs, those who had taken a TCA had an overall RR of 1.45 (95% CI, 1.31–1.60; p < 0.001).

thumbnail image

Figure 2. Risk of fracture associated with tricyclic antidepressants use by individual study and by all studies combined. CI = confidence interval.

Download figure to PowerPoint

The estimated fracture risk changed little when studies with different criteria were included (Table 3). For example, the overall RR increased slightly to 1.48 when osteoporotic fracture was included as the only outcome measure. When the analysis was confined to the 10 studies that adjusted for important fracture risk factors such as age, sex, major comorbidity, and medications known to increase fracture risk, the overall RR decreased slightly to 1.42. After the exclusion of three studies that included persons younger than 65 years, the overall RR varied little. Finally, when the analysis was confined to the nine studies for which the quality score was at least 7, the overall RR decreased slightly to 1.39.

Table 3. Relative Risk of Fracture Associated With Use of Tricyclic Antidepressants According to Different Exclusion Criteria
Studies includedStudies (n)Relative risk (95% CI)p
  • CI = confidence interval.

  • a

    Excludes Coupland and colleagues13 (2011).

  • b

    Excludes Lewis and colleagues10 (2007) and Ziere and colleagues11 (2008).

  • c

    Excludes Ray and colleagues15 (1991), Liu and colleagues16 (1998), and Vestergaard and colleagues18 (2006).

  • d

    Excludes Vestergaard and colleagues18 (2006), Ziere and colleagues11 (2008), and Verdel and colleagues20 (2010).

All studies121.45 (1.31–1.60)<0.001
Studies that used osteoporotic fractures as outcomea111.48 (1.31–1.67)<0.001
Studies that controlled for important fracture risk factorsb101.42 (1.29–1.57)<0.001
Studies with participants aged ≥65 years onlyc91.46 (1.30–1.65)<0.001
Studies with methodological quality score ≥7d91.39 (1.25–1.55)<0.01

Table 4 summarizes the pooled estimates of RR associated with TCA use in subgroups of studies according to study design, anatomical site of fracture, sex, TCA dose, duration of TCA exposure, exposure time, adjustment for BMD, adjustment for depression, and geographic location. The association between fracture risk and TCA use increased significantly in all subgroups except the group that formerly used TCAs. In particular, the fracture risk was substantially higher when the duration of TCA exposure was <6 weeks compared with a duration of ≥6 weeks. Additionally, the increased fracture risk was significant for current TCA use, but not for a history of TCA use. The fracture risk associated with TCA was higher in studies with case-control design, hip/femur fracture, male sex, higher TCA dose, adjustment for depression, adjustment for BMD, and conducted outside the United States. However, these differences were not statistically significant, as indicated by their overlapping 95% CIs.

Table 4. Relative Risk of Fracture Associated With Use of Tricyclic Antidepressants in Subgroups Defined by Characteristics of Study Design, Anatomical Site of Fracture, Sex, Dose, Duration of Exposure, Time of Exposure, Confounder Adjustment, and Study Location
Studies includedStudies (n)Relative risk (95% CI)p
  1. CI = confidence interval; DDD = defined daily dose; BMD = bone mineral density.

Study design
 Cohort51.28 (1.14–1.46)<0.001
 Case-control71.52 (1.31–1.76)<0.001
Anatomical site of fracture
 Nonspine/nonvertebral41.39 (1.08–1.77)0.009
 Hip/femur71.52 (1.28–1.81)<0.001
Sex
 Women71.43 (1.22–1.67)<0.001
 Men32.16 (1.47–3.16)<0.001
Dose
 ≤0.5 DDDs41.20 (1.08–1.34)<0.001
 >0.5/≤1.0 DDDs21.52 (1.20–1.94)<0.001
 >1.0 DDDs41.47 (1.28–1.69)<0.001
Exposure duration
 <6 weeks32.40 (1.41–4.08)0.001
 ≥6 weeks41.13 (1.00–1.28)0.04
Exposure time
 Former41.04 (0.86–1.26)0.70
 Current41.67 (1.53–1.81)<0.001
Adjusted for BMD
 Yes51.54 (1.24–1.90)<0.001
 No71.39 (1.25–1.55)<0.001
Adjusted for depression
 Yes71.49 (1.28–1.67)<0.001
 No51.45 (1.19–1.75)<0.001
Geographic location
 International81.44 (1.29–1.60)<0.001
 United States41.54 (1.13–2.09)<0.007

Publication bias was suspected in this meta-analysis, as indicated by the funnel plot (Fig. 3) and the Egger test (t = 2.48; p = 0.03). However, the overall effect size was still significant (RR, 1.36; 95% CI, 1.24–1.50; p < 0.001) after trim-and-fill correction for missing data.

thumbnail image

Figure 3. Funnel plot of log relative risk versus standard error of the log relative risk.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

We conducted a comprehensive meta-analysis to examine the association between TCA treatment and risk of fracture based on all available qualified studies that were conducted in a wide range of geographic regions and populations. From this meta-analysis, we found that the relative risk of fracture was increased by 45% in TCA users compared to that in non-TCA users. This moderate, but clinically relevant, increase in fracture risk was consistent in multiple sensitivity analyses (Table 3) and in subgroup analyses (Table 4). After adjustment for possible publication bias, the increased risk associated with TCA treatment (RR, 1.36; 95% CI, 1.24–1.50; p < 0.001) remained moderate and clinically important, although slightly weaker. These findings strongly suggest that taking TCAs is associated with an increased risk of fractures.

The underlying mechanism for the association between TCA use and fractures remains unclear. Both bone loss and a higher propensity to fall contribute to increased fracture risk. Antidepressant use has been associated with an increased risk of falls in elderly patients,8, 37 which may explain their association with fractures. At the early stage of TCA treatment, patients have decreased blood pressure,38 which may increase the risk of orthostatism and the risk of falling. TCA use can increase the patient's heart rate,39 which may influence cardiac output and decrease blood flow to the central nervous system, thus leading to confusion and delirium. In addition, TCA treatment has been reported to increase body sway and decrease postural balance.40 All these adverse effects have the potential to increase propensity for falls and thus to increase fracture risk.

Several findings in our meta-analysis also strongly suggest that the effect of TCAs is to increase the propensity for falling rather than to cause bone loss: (1) adjusting for BMD did not decrease the estimates, which suggests that TCA use may exert an increased risk of fracture independent of BMD reduction; (2) the increased risk of fracture was associated with current TCA use but not with former TCA use, and the acute effect suggests that the effect of TCAs is to increase falls instead of to affect the skeleton itself; and (3) the estimate for hip fracture was greater than that in other anatomical sites, and hip fracture is relatively more influenced by falls. Therefore, TCA treatment appeared to increase fracture risk mainly by increasing the risk of falling. Patients taking TCAs should be targeted for fall prevention education to prevent fractures.

Our previous meta-analyses indicated that depression is associated with fracture and decreased BMD.4, 5 Confounding by depression may contribute to the increased risk of fracture in TCA users. However, our current findings indicate that TCAs may exert an increased risk of fracture independent of depression, because we observed a similar elevated risk of fracture associated with TCA treatment in the subgroup that was adjusted for depression. However, in these original reports, various methods were used to assess depression, and severity and chronicity of prior depression may not be taken into account in risk estimates, therefore confounding by indication cannot be excluded. In addition, TCA treatment may decrease current symptoms of depression, yet prior depression may still elevate risk of fracture.

We observed some variations in the association of TCA treatment with fracture risk in subgroups of this study. Risk of fracture associated with TCA use in case-control studies was higher than estimated in cohort studies. Estimates in case-control studies may be more likely to be influenced by depression than are estimates in cohort studies, because TCA users in the case-control studies were from clinics where depression might be more prevalent and severe than in TCA users in cohort studies. This possibility is supported by our previous meta-analysis demonstrating that depression is a risk factor of fracture and bone loss.5 Interestingly, the risk estimate was less robust in women than in men. All female participants were older than 65 years. Aging and menopause substantially affect BMD reduction and fracture risk, and both may also play a role in overall survival due to comorbidities. Therefore, survival bias may explain the difference in risk estimates between men and women.

In addition, we observed that the increased risk of fracture with a shorter duration of TCA treatment (<6 weeks) is much higher than that with extended treatment (≥6 weeks). The decreased effect over time may be due to tachyphylaxis.41 With long-term administration of TCAs, adaptive changes could occur. An adjustment to the cardiovascular effect of TCAs38, 39 may explain the decrease in fracture risk with long-term TCA treatment.

TCAs have increasingly been replaced by antidepressants with an improved safety and adverse-effect profile, such as the SSRIs. TCAs, which were the first choice for pharmacologic treatment of clinical depression for decades, are still considered highly effective antidepressants. TCAs are probably more successful for treating melancholic depression than are other antidepressant drug classes.42 In addition, lower-dose TCAs have been widely used in clinics to treat migraine and tension-type headaches,43 irritable bowel syndrome,44 and symptoms of painful diabetic neuropathy.45 In this meta-analysis, we found that even lower-dose TCAs (<0.5 defined daily dose) are associated with an increased risk of fracture, which suggests that physicians who prescribe low-dose TCAs should consider and prevent that increased risk of fracture, especially for elderly patients or those already at high risk. To our knowledge, our study is the most comprehensive meta-analysis to date to investigate the association between treatment with TCAs and risk of fracture. Robustness of the association was ensured by testing various assumptions. In addition, subgroup analyses were completed to examine the effects of study design, anatomical site, sex, TCA dose, duration of exposure, TCA exposure time, adjustment for depression or BMD, and study location.

However, our study does have several limitations. Because of the limited number of studies that met inclusion criteria (n = 12), we could not perform a multivariate meta-regression analysis to further investigate the sources of heterogeneity presented in this meta-analysis. Nonetheless, as described in Table 4, we performed a number of subgroup analyses with available key variables. Heterogeneity may be partially explained by subgroup differences in study design, anatomical site, sex, exposure duration, and study location. Another limitation is that we could not assess risks of falling in this study because few individual studies accounted for falls. Nonetheless, falls remain an important potential etiology in the association between TCA treatment and fracture risk. In addition, other medications may have the potential to affect bone strength, such as glucocorticoids46, 47 and anticonvulsants,48 which were reported to be associated with risk of fracture. The use of these medications was not reported in most studies; therefore, the possibility that the association was confounded by other medications cannot be ruled out. Finally, a few individual studies in this meta-analysis used claims data, which may lack information on nutrients, physical functioning, cognition, or other potential confounders. However, it is unlikely that these confounders would alter the magnitude or validity of pooled estimates in this meta-analysis. Additional studies with information on these potential confounders are warranted to further examine the association between TCA treatment and fracture risk.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

Our findings indicate that the use of TCAs is associated with an increased risk of fracture. TCAs may exert an increased risk of fracture independent of depression and BMD. The increased risk associated with TCAs is moderate but may have a substantial clinical impact. Physicians prescribing even low-dose TCAs should consider the increased risk of fracture, especially for older persons already at a risk. Fracture risk should be monitored in patients taking TCAs, especially during the initial period of TCA treatment.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

We thank Ann M Farrell and Kay E Wellik for their consultation on the literature search strategy.

Authors' roles: Study concept (QW, MDC, KAF); Study design (QW, MDC); Literature search (QW, WQ); Study selection (QW, WQ, MDC, KAF); Study appraisal (QW, WQ); Data abstraction (QW, WQ); Data analysis (QW, JGH); Interpretation of data (QW, JGH); Drafting of the manuscript (QW); Critical revision of the manuscript (QW, WQ, MDC, JGH, KAF); Approval of the final version of manuscript (QW, WQ, MDC, JGH, KAF). QW had full access to the data in the study and takes responsibility for the integrity and accuracy of the data analysis.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information
  • 1
    Gehlbach SH, Avrunin JS, Puleo E. Trends in hospital care for hip fractures. Osteoporos Int. 2007 May; 18(5):58591.
  • 2
    Melton LJ3rd. Cooper C. Magnitude and impact of osteoporosis and fractures In: Marcus R, Feldman D, Kelsey J, editors. Osteoporosis. San Diego: Academic Press; 2001: p. 55767.
  • 3
    Shim RS, Baltrus P, Ye J, Rust G. Prevalence, treatment, and control of depressive symptoms in the United States: results from the National Health and Nutrition Examination Survey (NHANES), 2005–2008. J Am Board Fam Med. 2011 Jan-Feb; 24(1):338.
  • 4
    Wu Q, Magnus JH, Liu J, Bencaz AF, Hentz JG. Depression and low bone mineral density: a meta-analysis of epidemiologic studies. Osteoporos Int. 2009 Aug; 20(8):130920.
  • 5
    Wu Q, Liu J, Gallegos-Orozco JF, Hentz JG. Depression, fracture risk, and bone loss: a meta-analysis of cohort studies. Osteoporos Int. 2010 Oct; 21(10):162735.
  • 6
    Wu Q, Bencaz AF, Hentz JG, Crowell MD. Selective serotonin reuptake inhibitor treatment and risk of fractures: a meta-analysis of cohort and case-control studies. Osteoporos Int. 2012 Jan; 23(1):36575.
  • 7
    Ray WA. Psychotropic drugs and injuries among the elderly: a review. J Clin Psychopharmacol. 1992 Dec; 12(6):38696.
  • 8
    Thapa PB, Gideon P, Cost TW, Milam AB, Ray WA. Antidepressants and the risk of falls among nursing home residents. N Engl J Med. 1998 Sep 24; 339(13):87582.
  • 9
    Ensrud KE, Blackwell T, Mangione CM, Bowman PJ, Bauer DC, Schwartz A, Hanlon JT, Nevitt MC, Whooley MA. Study of Osteoporotic Fractures Research Group. Central nervous system active medications and risk for fractures in older women. Arch Intern Med. 2003 Apr 28; 163(8):94957.
  • 10
    Lewis CE, Ewing SK, Taylor BC, Shikany JM, Fink HA, Ensrud KE, Barrett-Connor E, Cummings SR, Orwoll E. Osteoporotic Fractures in Men (MrOS) Study Research Group. Predictors of non-spine fracture in elderly men: the MrOS study. J Bone Miner Res. 2007 Feb; 22(2):2119.
  • 11
    Ziere G, Dieleman JP, van der Cammen TJ, Hofman A, Pols HA, Stricker BH. Selective serotonin reuptake inhibiting antidepressants are associated with an increased risk of nonvertebral fractures. J Clin Psychopharmacol. 2008 Aug; 28(4):4117.
  • 12
    Diem SJ, Blackwell TL, Stone KL, Cauley JA, Hillier TA, Haney EM, Ensrud KE. Study of Osteoporotic Fractures Research Group. Use of antidepressant medications and risk of fracture in older women. Calcif Tissue Int. 2011 Jun; 88(6):47684.
  • 13
    Coupland C, Dhiman P, Morriss R, Arthur A, Barton G, Hippisley-Cox J. Antidepressant use and risk of adverse outcomes in older people: population based cohort study. BMJ. 2011 Aug 2; 343:d4551.
  • 14
    Ray WA, Griffin MR, Schaffner W, Baugh DK, Melton LJ3rd. Psychotropic drug use and the risk of hip fracture. N Engl J Med. 1987 Feb 12; 316(7):3639.
  • 15
    Ray WA, Griffin MR, Malcolm E. Cyclic antidepressants and the risk of hip fracture. Arch Intern Med. 1991 Apr; 151(4):7546.
  • 16
    Liu B, Anderson G, Mittmann N, To T, Axcell T, Shear N. Use of selective serotonin-reuptake inhibitors or tricyclic antidepressants and risk of hip fractures in elderly people. Lancet. 1998; May 2; 351(9112):13037.
  • 17
    Hubbard R, Farrington P, Smith C, Smeeth L, Tattersfield A. Exposure to tricyclic and selective serotonin reuptake inhibitor antidepressants and the risk of hip fracture. Am J Epidemiol. 2003 Jul 1; 158(1):7784.
  • 18
    Vestergaard P, Rejnmark L, Mosekilde L. Anxiolytics, sedatives, antidepressants, neuroleptics and the risk of fracture. Osteoporos Int. 2006;17(6):80716.
  • 19
    van den Brand MW, Pouwels S, Samson MM, van Staa TP, Thio B, Cooper C, Leufkens HG, Egberts AC, Verhaar HJ, de Vries F. Use of anti-depressants and the risk of fracture of the hip or femur. Osteoporos Int. 2009 Oct; 20(10):170513. Erratum in: Osteoporos Int. 2009 Oct; 20(10) 1715.
  • 20
    Verdel BM, Souverein PC, Egberts TC, van Staa TP, Leufkens HG, de Vries F. Use of antidepressant drugs and risk of osteoporotic and non-osteoporotic fractures. Bone. 2010 Sep; 47(3):6049.
  • 21
    Vestergaard P. Fracture risks of antidepressants. Expert Rev Neurother. 2009 Jan; 9(1):13741.
  • 22
    Ginzburg R, Rosero E. Risk of fractures with selective serotonin-reuptake inhibitors or tricyclic antidepressants. Ann Pharmacother. 2009 Jan; 43(1):98103.
  • 23
    Rizzoli R, Cooper C, Reginster JY, Abrahamsen B, Adachi JD, Brandi ML, Bruyère O, Compston J, Ducy P, Ferrari S, Harvey NC, Kanis JA, Karsenty G, Laslop A, Rabenda V, Vestergaard P. Antidepressant medications and osteoporosis. Bone. 2012 Sep; 51(3):60613.
  • 24
    Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, Moher D, Becker BJ, Sipe TA, Thacker SB. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. Meta-analysis of observational studies in epidemiology: a proposal for reporting. JAMA. 2000 Apr 19; 283(15):200812.
  • 25
    Moher D, Liberati A, Tetzlaff J, Altman DG. PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med. 2009 Aug 18; 151(4):2649, W64.
  • 26
    Wells GA, Shea B, O'Connell D, Peterson J, Welch V, Losos M, Tugwell P. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomized studies in meta analyses [Internet]. Ottawa, ON, Canada: Ottawa Hospital Research Institute. 2011. [cited 2012 Nov 14]. Available from: http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp.
  • 27
    Walter SD, Cook RJ. A comparison of several point estimators of the odds ratio in a single 2 × 2 contingency table. Biometrics. 1991 Sep; 47(3):795811.
  • 28
    Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002; Jun 15; 21(11):153958.
  • 29
    Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003 Sep 6; 327(7414):55760.
  • 30
    Furukawa TA, Guyatt GH, Griffith LE. Can we individualize the ‘number needed to treat’?: an empirical study of summary effect measures in meta-analyses. Int J Epidemiol. 2002 Feb; 31(1):726.
  • 31
    Sterne JA, Gavaghan D, Egger M. Publication and related bias in meta-analysis: power of statistical tests and prevalence in the literature. J Clin Epidemiol. 2000 Nov; 53(11):111929.
  • 32
    Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics. 1994 Dec; 50(4):1088101.
  • 33
    Duval S, Tweedie R. Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics. 2000; Jun; 56(2):45563.
  • 34
    Peters JL, Sutton AJ, Jones DR, Abrams KR, Rushton L. Performance of the trim and fill method in the presence of publication bias and between-study heterogeneity. Stat Med. 2007 Nov 10; 26(25):454462.
  • 35
    Ruzni N, Idris N. Performance of the trim and fill method in adjusting for the publication bias in meta-analysis of continuous data. Am J Appl Sci. 2012;9(9):15127.
  • 36
    Diem SJ, Blackwell TL, Stone KL, Yaffe K, Haney EM, Bliziotes MM, Ensrud KE. Use of antidepressants and rates of hip bone loss in older women: the study of osteoporotic fractures. Arch Intern Med. 2007 Jun 25; 167(12):12405.
  • 37
    Darowski A, Chambers SA, Chambers DJ. Antidepressants and falls in the elderly. Drugs Aging. 2009;26(5):38194.
  • 38
    Christensen P, Thomsen HY, Pedersen OL, Gram LF, Kragh-Sorensen P. Orthostatic side effects of clomipramine and citalopram during treatment for depression. Psychopharmacology (Berl). 1985;86(4):3835.
  • 39
    Rodriguez de la Torre B, Dreher J, Malevany I, Bagli M, Kolbinger M, Omran H, Luderitz B, Rao ML. Serum levels and cardiovascular effects of tricyclic antidepressants and selective serotonin reuptake inhibitors in depressed patients. Ther Drug Monit. 2001 Aug; 23(4):43540.
  • 40
    Pollock BG. Adverse reactions of antidepressants in elderly patients. J Clin Psychiatry. 1999; 60 Suppl 20:48.
  • 41
    Katz G. Tachyphylaxis/tolerance to antidepressive medications: a review. Isr J Psychiatry Relat Sci. 2011;48(2):12935.
  • 42
    Mitchell PB, Mitchell MS. The management of depression. Part 2. The place of the new antidepressants. Aust Fam Physician. 1994 Sep; 23(9): 1771-3, 177681.
  • 43
    Jackson JL, Shimeall W, Sessums L, Dezee KJ, Becher D, Diemer M, Berbano E, O'Malley PG. Tricyclic antidepressants and headaches: systematic review and meta-analysis. BMJ. 2010 Oct 20; 341:c5222.
  • 44
    Rahimi R, Nikfar S, Rezaie A, Abdollahi M. Efficacy of tricyclic antidepressants in irritable bowel syndrome: a meta-analysis. World J Gastroenterol. 2009 Apr 7; 15(13):15483.
  • 45
    Wong MC, Chung JW, Wong TK. Effects of treatments for symptoms of painful diabetic neuropathy: systematic review. BMJ. 2007 Jul 14; 335(7610):87.
  • 46
    Romas E. Bone loss in inflammatory arthritis: mechanisms and therapeutic approaches with bisphosphonates. Best Pract Res Clin Rheumatol. 2005 Dec; 19(6):106579.
  • 47
    Harpavat M, Keljo DJ, Regueiro MD. Metabolic bone disease in inflammatory bowel disease. J Clin Gastroenterol. 2004 Mar; 38(3):21824.
  • 48
    Kinjo M, Setoguchi S, Schneeweiss S, Solomon DH. Bone mineral density in subjects using central nervous system-active medications. Am J Med. 2005 Dec; 118(12):1414.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Conclusion
  8. Disclosures
  9. Acknowledgements
  10. References
  11. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
jbmr_1813_sm_SuppInfo.doc22KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.