Painful conditions are common among persons with Alzheimer disease (AD) and other forms of dementia, as the point prevalence of pain among persons with dementia is approximately 50%.3 Many studies indicate that the prevalence of pain increases with age,1 and similarly, the prevalence of dementia disorders increases with age.5 Due to problems in speech and communication, pain in persons with dementia is frequently difficult to diagnose and assess. Signs of pain may be distorted by uncommon symptoms, and pain may produce behavioral and psychological symptoms of dementia (BPSD), eg, depression and agitation.
Treatment with opioids may be necessitated by more severe pain. Previous studies report that annual prevalence of opioid use was 28% among community-dwelling persons with dementia,11 and the 6-month prevalence was 7% among those who had just received diagnoses of AD.9 Some studies show that community-dwelling persons with dementia are more likely to use opioids than persons without dementia,11 whereas some studies report higher use of strong opioids but not mild opioids.9 Approximately one-third of opioid users with AD were long-term users in a previous study,8 which is in line with studies on frequency of pain in this population.10 However, older persons are particularly vulnerable to adverse effects of opioid use due to aging-related changes in pharmacokinetics and pharmacodynamics.28 As a consequence of these changes, the response to conventional opioid doses may be increased, which, in turn, could result in excessive central nervous system (CNS) adverse effects such as higher level of sedation.16 In addition, older persons are at increased risk of delirium and other adverse cognitive effects of opioids.4,15 Central nervous system polypharmacy is particularly common among older opioid users,7 which may be detrimental due to increased risk of cognitive decline and falls.21,31
Persons with cognitive impairment are at increased risk of falling.6 Age is the most important risk factor for AD and for several other cognitive disorders causing dementia.5 Alzheimer disease causes about two-thirds of all dementia cases. Persons with AD have 2-fold increased risk of hip fracture and higher mortality after hip fracture compared with age- and sex-matched comparison persons without AD.27 As persons with AD frequently use opioids, it is important to evaluate whether opioid use increases risk of hip fractures among persons with AD. To the best of our knowledge, no previous study has assessed risk of hip fracture associated with incident opioid use and duration of use among persons with dementia/AD.
The objective of this study was to investigate whether incident opioid use is associated with risk of hip fractures among community-dwelling persons with AD. The risk was also estimated in terms of duration of use and opioid strength.
2.1. Study cohort
This study was based on the MEDALZ (Medication use and Alzheimer's disease) cohort that includes all 70,718 community-dwelling persons who were diagnosed with AD between 2005 and 2011 in Finland.26 They were identified from Special Reimbursement register as described in detail before,26 with diagnoses according to the NINCDS-ADRDA14 and DSM-IV criteria.
Several nationwide registers have been used when linking data for the MEDALZ cohort; Prescription register (years 1995-2015), Special Reimbursement register (1972-2015), Hospital Discharge register (1972-2015), and socioeconomic data since 1970 and causes of death 2005 to 2015 from Statistics Finland. Each resident is assigned with a unique personal identification number, and all registers are linkable with this number. Drug use was collected from the Prescription register that includes information on all reimbursed drug dispensings. Prescription register data are categorized according to Anatomical Therapeutic Chemical (ATC) classification system codes30 and includes purchased amount as defined daily doses, which is the assumed average maintenance dose per day for a drug used for its main indication in adults. Drugs used during stays in hospitals and public nursing homes are not recorded in the register. Hospital Discharge register includes all hospital care periods with corresponding discharge diagnoses. Special Reimbursement register includes data on entitlements for higher reimbursement of drugs due to chronic diseases.
Opioids were defined according to ATC class N02A and are available only as prescription drugs. Opioids were further categorized as weak opioids (codeine and tramadol), buprenorphine, and strong opioids (morphine, hydromorphone, oxycodone, fentanyl, dextropropoxyphene, and pentazocine). Duration of opioid use was modelled with the PRE2DUP method.24 It is based on sliding averages of defined daily doses and individual drug use patterns. Each ATC code for each person is modelled separately and by taking into account on hospitalizations, stockpiling of drugs, variation in purchase events, and changing dose. For the main analyses of this study, continuous use of “any opioid” was then constructed by combining overlapping drug use periods of all specific drug substances. For subanalyses according to opioid strength, weak and strong opioid uses were formed by combining overlapping use periods within these classes to retrieve time on “weak opioids” and “strong opioids.”
Hip fractures were based on diagnoses recorded in the Hospital Discharge register. According to ICD-10 codes, hip fracture was defined as S72.0-S72.2. Corresponding ICD-8 and ICD-9 codes were used when defining persons with a previous hip fracture before AD diagnosis. Only the first diagnosis was considered for each person.
2.5. Study setting
For this study, persons diagnosed with AD in 2010 to 2011 (N = 23,100) were included (before these years, codeine–paracetamol products were inconsistently reimbursed). Exclusion criteria for this study were prevalent opioid use and long hospitalization (>50% of the washout period and those having >90 days hospital/institutional stay at the end of the washout period) during a 1-year washout period before AD diagnoses (Fig. 1). We excluded persons using opioids and those having a long hospital stay during the period, as drug exposure cannot be accurately defined for hospitalized persons. Exclusions based on previous opioid use were conducted to avoid prevalent user bias (new user design). Persons who had a previous hip fracture (since 1972 until diagnosis of AD) were excluded from all analyses. In addition, persons hospitalized/institutionalized for the entire follow-up period were excluded.
An exposure-matched cohort including opioid users with 1:1 matched nonuser comparison persons was formed after the exclusions were made for both users and nonusers. At the initiation date of opioid use, a nonuser was matched for each user based on age (±2 years), sex, and time since AD diagnoses (±90 days) with incidence density sampling. These matching criteria were chosen based on having the most impact on risk of opioid initiation and hip fracture. Time since AD diagnoses was used as a proxy for severity of illness. After excluding 2 users without a matched nonuser, the matched cohort included 4570 users and 4570 matched nonusers. The follow-up started on the date of opioid initiation for users, and this matching date was set as an index date for nonusers.
Covariates associated with opioid use8,9 and risk of hip fracture27 were chosen based on previous studies, and are listed in Table 1. A propensity score representing the probability of opioid use given the measured confounders was derived with logistic regression model to balance potential confounding factors between opioid users and nonusers. Inverse probability of treatment (IPT) weighting based on propensity score was used, and balance of covariate distributions before and after IPT weighting was conducted with standardized differences, values >10% considered as meaningful differences. All covariates in Table 1 were included in the propensity score and described in further detail in supplementary Table 1 (available at http://links.lww.com/PAIN/A668).
2.7. Statistical analyses
Statistical analyses were performed using SAS statistical software, version 9.4 (SAS Institute, Inc, Cary, NC). Register maintainers retrieved data from the registers and submitted deidentified data to the research team. According to Finnish legislation, no ethics committee approval was required, and participants were not contacted in any way.
Users and nonusers were followed up until the first of the following events occurred; hip fracture (outcome), >90 days hospitalization/institutionalization period, death, or the end of study follow-up (December 31, 2015). For users, the follow-up also ended on discontinuation of use and for nonusers, initiation of opioid use.
Cox proportional hazard models were used, and proportional hazard assumption was ascertained by Kaplan–Meier graphs. The models took into account on the matched design (own strata were used for each matching group). The main analyses compared opioid users with nonusers. For analyses on duration of use on the risk of hip fracture, time-varying exposure was modeled using categorical time-dependent variable with classes for ≤60 days, 61 to 180 days, 181 to 365 days, and over 365 days of exposure. For analyses of opioid strength, users were stratified according to opioid strength at initiation of use and were censored if/when person switched or initiated concomitant use with multiple opioid strengths. The analyses were also censored after 500 days of follow-up due to sparsity of data. In these analyses, those 35 users initiating with multiple opioid strengths were excluded.
In addition, “intention-to-treat” (ITT) analyses were conducted to assess the effect of informative censoring, ie, drug use is discontinued due to adverse effects that would lead to the hip fracture. In ITT analyses, opioid users were considered as users for 180 days since the initiation of use, and their follow-up was not censored if they discontinued the use or were hospitalized (due to other reasons than hip fracture). Thus, the follow-up ended on hip fracture, death, the end of study follow-up, or after 180 days whichever occurred first. Corresponding “as treated” analyses were conducted by restricting the follow-up to the first 180 days. All analyses were conducted unadjusted and IPT-weighted.
Age-adjusted incidence rates were calculated as events per 100 person-years for opioid users and nonusers. Attributable risk was calculated with incidence rates with the following formula: [Incidence (exposed) − Incidence (unexposed)].
Due to the exposure-matched design, opioid users (N = 4750) and nonusers (N = 4750) were similar in sex (66.7% of women in both groups), age (mean age 83.0 years, SD 6.9 in users and 83.0 years, SD 6.8 in nonusers), and time since AD diagnoses (median time in days 745, interquartile range [IQR] 314-1248 in users and 745, IQR 315-1247 in nonusers). However, opioid users were more likely to have comorbid conditions, have previous fracture or cancer, and use paracetamol and psychotropic drugs (Table 1). Inverse probability of treatment weighting balanced these differences between the groups.
Median follow-up time was longer in nonusers (681 days, IQR 245-1038) than in users (141 days, IQR 21-126). Among opioid users, 76 hip fractures occurred during the use, whereas 200 hip fractures occurred among nonusers. Age-adjusted incidence rate of hip fractures was 3.47 (95% CI 2.62-4.33) during opioid use and 1.94 (95% CI 1.65-2.22) during nonuse of opioids (Table 2). Attributable risk of hip fractures in opioid use was 1.53 per 100 person-years, ie, hip fractures that can be attributed to opioid exposure. Opioid use was associated with an increased risk of hip fracture (IPT-weighted hazard ratio 1.96, 95% CI 1.27-3.02). The risk was observed during the first 2 months of use (IPT-weighted hazard ratio 2.37, 95% CI 1.04-5.41), and after that, the risk was not significant, although point estimates suggested an attenuated, but increased risk. Intention-to-treat and as-treated analyses for the first 180 days resulted in similar findings as the main analyses.
Among opioid initiators, 4715 users initiated with one opioid class. Of these initiations, mild opioids were most frequent class (45.6%, N = 2149), followed by buprenorphine (39.7%, N = 1870; of which, 99.3% were transdermal formulation) and strong opioids (14.8%, N = 696). Mild opioid initiations consisted mostly on codeine combinations (N = 1732, 80.6%), and the remaining 19.4% (N = 417) used tramadol. The majority of strong opioid users initiated with oxycodone (N = 460, 66%), followed by fentanyl (N = 217, 31%). The risk of hip fracture increased by the increasing opioid strength, with buprenorphine and strong opioid use being associated with hip fractures while weak opioids were not. However, the confidence intervals were wide and partially overlapping between the categories.
We found an increased risk of hip fractures associated with opioid use in persons with AD. The risk was elevated during the first 2 months of use and nonsignificant in longer duration of use. In addition, a suggestion of dose–response relationship in opioid strength was observed.
Our results are in line with previous studies and meta-analyses on the association between opioid use and risk of hip fracture.17,25 Our risk estimate is somewhat higher than in previous studies. As previous studies have not assessed risk of hip fracture associated with incident opioid use and duration of use in persons with dementia or AD, it is possible that opioid initiation has a larger impact on fracture risk in this group. However, we did not compare the risks to general older population. One factor may be related to short follow-up times among opioid users, due to discontinuation of use. Thus, the risk estimate is mainly driven by short-term use, and we found that the risk was highest in the beginning of the use. Previous studies have also reported that the risk is highest at the initiation and attenuated and disappears in longer duration.13
There may be several underlying mechanisms explaining the association between opioid use and increased risk of hip fracture. Central nervous system effects such as dizziness and sedation are common adverse effects of opioid therapy, especially during the initiation phase, and these CNS effects may result in an increased risk of fall-related fractures.20 It is likely that decreased attention and psychomotor functioning among persons with AD affect hip fracture risk, and opioids can further increase this risk by inducing cognitive dysfunction and even delirium.4 The use of opioids is often associated with polypharmacy,7 which represents additional risk of drug-induced confusion and falls because several commonly used medications are recognized as fall-risk–increasing drugs.15,21,29
Pain and especially severe pain may induce or be an underlying cause of BPSD in persons with AD3 and consequently, increase the risk of falling. Due to the challenges in identification and diagnosing pain among persons with AD,3 opioids might also be used for chronic pain, such as neuropathic pain, for which other pharmacotherapeutic options could be more suitable. In case of severe BPSD, if there is a risk for the AD person to harm himself/herself or other persons, it is possible that opioids are prescribed to exclude pain as a potential cause of these serious symptoms. As we did not have data on pain or BPSD, it is possible that the association between opioid use and hip fractures is somewhat confounded by these underlying indications.
It has been suggested that chronic use of opioids has negative effect on bone metabolism and it is associated with the development of osteoporosis.2 This is due to opioids having both direct and indirect effects on bone metabolism; they interfere directly with osteoblast activity and cause hypogonadism by inhibition of hypothalamic-pituitary-adrenal axis. However, the process of opioid-induced bone loss takes time, and the risk of hip fracture was highest in the beginning of opioid use. Thus, our results are more likely related to the CNS adverse effects and risk of falls than reduced bone density due to opioids. The risk was not significant in longer duration of opioid use, which may indicate patients' adaptation to opioid therapy. This is also in line with reports stating that tolerance to sedative effects of opioids develops after the initiation phase, although data on older persons or persons with cognitive impairment are scarce.32
A suggestion of dose–response relationship in the strength of opioids used and risk of hip fracture was found. Weak opioid use was not associated with the risk, whereas buprenorphine and strong opioid uses were associated with the risk. Increasing risk with increasing opioid strength provides additional evidence for the association. Some previous studies have reported similar findings on higher opioid dose.19 We did not assess doses in morphine equivalents because our study population mainly initiated with one opioid and because there was not much variation in doses nor high dose use. Buprenorphine was the most commonly initiated opioid in our study, followed by codeine combinations. The majority of buprenorphine initiations were transdermal patches (99%). Transdermal buprenorphine should be reserved for nonacute pain, as achieving peak plasma concentrations takes days.12 Its use in long term for nonmalignant pain is far more common compared with strong opioids in persons with AD.8 As previously described,8 choice of opioid in persons with AD is driven by the dose form with transdermal patches frequently favored over oral ones. This may explain the relatively frequent use of buprenorphine and fentanyl in our study population, as these 2 opioids are available as transdermal patches in Finland.
4.1. Strengths and limitations
Strength of our study includes a large, nationwide cohort of community-dwelling persons with AD and thus, resulting in high generalizability to community-dwelling persons with AD. Due to inconsistent reimbursement status of codeine in earlier years, the study cohort was restricted to persons diagnosed in 2010 to 2011. However, these years likely represent recent drug use patterns better. The analyses were restricted to first hip fracture, as the previous study showed high validity of Finnish register-based diagnoses of hip fractures for the first event.22 We also used a validated method for modelling drug use based on purchases recorded in the Prescription register data.23,24
We used new user design,18 ie, excluded prevalent opioid users, to avoid survival and selection biases that are related to prevalent users who tolerate the drug effects. As age, sex, and AD are major factors impacting on the risk of hip fracture, a nonuser was matched for each opioid user by these factors, at the initiation date of use. Time since AD diagnoses was considered as a proxy for the progression of the disease, and no other data on cognition decline or severity of the disease were available. Inverse probability of treatment weighting with propensity score aimed to balance other differences between opioid users and nonusers. As in all observational studies, residual confounding may still exist. Indication of opioid use and severity and duration of pain were not available in the registers. In addition, we did not have data on many lifestyle-related risk factors for hip fracture, such as nutrition, alcohol consumption, smoking, and body mass index, nor balance or vision problems. However, these were likely, at least partially, captured by the comorbidities used as adjustments.
Opioid use was associated with an increased risk of hip fractures among persons with AD. The risk was evident in the initiation of treatment and attenuated after 2 months of use. As older persons with AD are at increased risk of injurious falls, there are considerable challenges in treatment of moderate or severe pain. Whether the risk of injurious falls can be avoided or reduced by slow titration of opioid doses in the beginning of treatment needs further research.
Conflict of interest statement
H. Taipale, J. Tiihonen, and A. Tanskanen have participated in research projects funded by Janssen and Eli Lilly with grants paid to the institution where they were employed. A. Tanskanen is a member of Janssen advisory board. J. Tiihonen reports personal fees from the Finnish Medicines Agency (Fimea), European Medicines Agency (EMA), Eli Lilly, Janssen-Cilag, Lundbeck, and Otsuka; and has received grants from the Stanley Foundation and Sigrid Jusélius Foundation. M. Koponen has received personal research grant from Oy H. Lundbeck Ab foundation outside the submitted work. The remaining authors have no conflicts of interest to declare.
Data were retrieved from the registers by the register maintainers, and deidentified register data were submitted to the research team. Participants were not contacted in any way. According to Finnish legislation, no ethics committee approval is required in these circumstances.
Author contributions: All authors contributed to design of the study, data collection, data interpretation, and critically revised and approved the manuscript. H. Taipale, M. Koponen, and A. Hamina designed statistical analyses. H. Taipale had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis (acts as guarantor).
Supplemental digital content
Supplemental digital content associated with this article can be found online at http://links.lww.com/PAIN/A668.
. Abdulla A, Adams N, Bone M, Elliott AM, Gaffin J, Jones D, Knaggs R, Martin D, Sampson L, Schofield P; Geriatric Society British. Guidance on the management of pain in older people. Age Ageing 2013;42(suppl 1):i1–57.
. Coluzzi F, Mattia C, Raffa RR, Pergolizzi J. The unsolved case of bone-impairing analgesics: the endocrine effects of opioids on bone metabolism. Ther Clin Risk Manag 2015;11:515.
. Corbett A, Husebo B, Malcangio M, Staniland A, Cohen-Mansfield J, Aarsland D, Ballard C. Assessment and treatment of pain in people with dementia
. Nat Rev Neurol 2012;8:264–74.
. Dhingra L, Ahmed E, Shin J, Scharaga E, Magun M. Cognitive effects and sedation. Pain Med 2015;16:S37–43.
. Fargo K; Alzheimer's Association Report. 2014 Alzheimers disease facts and figures. Alzheimer's Dement 2014;10:e47–92.
. Friedman SM, Mendelson DA. Epidemiology of fragility fractures. Clin Geriatr Med 2014;30:175–81.
. Gerlach LB, Olfson M, Kales HC, Maust DT. Opioids and other central nervous system–active polypharmacy in older adults in the United States. J Am Geriatr Soc 2017;65:2052–6.
. Hamina A, Taipale H, Tanskanen A, Tolppanen AM, Karttunen N, Pylkkänen L, Tiihonen J, Hartikainen S. Long-term use of opioids for non-malignant pain among community-dwelling persons with and without Alzheimer's disease in Finland. PAIN 2017;158:252–60.
. Hamina A, Taipale H, Tanskanen A, Tolppanen AM, Tiihonen J, Hartikainen S. Differences in analgesic use in community-dwelling persons with and without Alzheimer's disease. Eur J Pain 2017;21:658–67.
. Hunt LJ, Covinsky KE, Yaffe K, Stephens CE, Miao Y, Boscardin J, Smith AK. Pain in community-dwelling older adults with dementia
: results from the National Health and Aging Trends Study. J Am Geriatr Soc 2015;63:1503–11.
. Jensen-Dahm C, Gasse C, Astrup A, Mortensen PB, Waldemar G. Frequent use of opioids in patients with dementia
and nursing home residents: a study of the entire elderly population of Denmark. Alzheimers Dement 2014;11:691–699.
. Kress HG. Clinical update on the pharmacology, efficacy and safety of transdermal buprenorphine. Eur J Pain 2009;13:219–30.
. Li L, Setoguchi S, Cabral H, Jick S. Opioid
use for noncancer pain and risk of fracture in adults: a nested case-control study using the general practice research database. Am J Epidemiol 2013;178:559–69.
. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer's disease: report of the NINCDS-ADRDA work group under the auspices of department of health and human services task force on Alzheimer's disease. Neurology 1984;34:939–44.
. Moore AR, O'Keeffe ST. Drug-induced cognitive impairment in the elderly. Drugs Aging 1999;15:15–28.
. van Ojik AL, Jansen PAF, Brouwers JRBJ, van Roon EN. Treatment of chronic pain in older people. Drugs Aging 2012;29:615–25.
. Ping F, Wang Y, Wang J, Chen J, Zhang W, Zhi H, Liu Y. Opioids increase hip fracture
risk: a meta-analysis. J Bone Miner Metab 2016;35:289–97.
. Ray WA. Evaluating medication effects outside of clinical trials: new-user designs. Am J Epidemiol 2003;158:915–20.
. Saunders KW, Dunn KM, Merrill JO, Sullivan M, Weisner C, Braden JB, Psaty BM, Von Korff M. Relationship of opioid
use and dosage levels to fractures in older chronic pain patients. J Gen Intern Med 2010;25:310–15.
. Seppala LJ, van de Glind EMM, Daams JG, Ploegmakers KJ, de Vries M, Wermelink AM, van der Velde N. Fall-risk-increasing drugs: a systematic review and meta-analysis: III. Others. J Am Med Dir Assoc 2018;19:372.e1–372.e8.
. Seppala LJ, Wermelink AM, de Vries M, Ploegmakers KJ, van de Glind EM, Daams JG, van der Velde N. Fall-risk-increasing drugs: a systematic review and meta-analysis: II. Psychotropics. J Am Med Dir Assoc 2018;19:372.e1–372.e8.
. Sund R, Nurmi-Lüthje I, Lüthje P, Tanninen S, Narinen A, Keskimäki I. Comparing properties of audit data and routinely collected register data in case of performance assessment of hip fracture
treatment in Finland. Methods Inf Med 2007;46:558–66.
. Tanskanen A, Taipale H, Koponen M, Tolppanen AM, Hartikainen S, Ahonen R, Tiihonen J. Drug exposure in register-based research—an expert-opinion based evaluation of methods. PLoS One 2017;12:e0184070.
. Tanskanen A, Taipale H, Koponen M, Tolppanen AM, Hartikainen S, Ahonen R, Tiihonen J. From prescription drug purchases to drug use periods—a second generation method (PRE2DUP). BMC Med Inform Decis Mak 2015;15:21.
. Teng Z, Zhu Y, Wu F, Zhu Y, Zhang X, Zhang C, Wang S, Zhang L. Opioids contribute to fracture risk: a meta-analysis of 8 cohort studies. PLoS One 2015;10:1–13.
. Tolppanen AM, Taipale H, Koponen M, Lavikainen P, Tanskanen A, Tiihonen J, Hartikainen S. Cohort profile: the Finnish Medication and Alzheimer's disease (MEDALZ) study. BMJ Open 2016;6:e012100.
. Tolppanen AM, Taipale H, Tanskanen A, Tiihonen J, Hartikainen S. Comparison of predictors of hip fracture
and mortality after hip fracture
in community-dwellers with and without Alzheimer's disease—exposure-matched cohort study. BMC Geriatr 2016;16:204.
. Trifirò G, Spina E. Age-related changes in pharmacodynamics: focus on drugs acting on central nervous and cardiovascular systems. Curr Drug Metab 2011;12:611–20.
. de Vries M, Seppala LJ, Daams JG, van de Glind EMM, Masud T, van der Velde N. Fall-risk-increasing drugs: a systematic review and meta-analysis: I. Cardiovascular drugs. J Am Med Dir Assoc 2018;19:371.e1–371.e9.
. WHO Collaborating Centre for Drug Statistics Methodology, 2018. The anatomical therapeutic chemical classification system. Available at (http://www.whocc.no/atc_ddd_index/)
. Accessed May 22, 2018.
. Wright RM, Roumani YF, Boudreau R, Newman AB, Ruby CM, Studenski SA, Shorr RI, Bauer DC, Simonsick EM, Hilmer SN, Hanlon JT. Impact of central nervous system (CNS) medication use on cognition decline in community dwelling older adults: findings from the health, aging and body composition study. J Am Geriatr Soc 2009;57:243–50.
. Young-McCaughan S, Miaskowski C. Definition of and mechanism for opioid
-induced sedation. Pain Manag Nurs 2001;2:84–97.