Psoriasis, fracture risk and bone mineral density: the HUNT Study, Norway

Summary

Background

An association between psoriasis and osteoporosis has been reported.

Objectives

To investigate, in a large prospective population‐based Norwegian study, whether psoriasis is associated with increased risk of forearm or hip fracture; to investigate the cross‐sectional association between psoriasis and bone mineral density (BMD) T‐score in a subpopulation.

Methods

Hospital‐derived fracture data from Nord‐Trøndelag County (1995–2013) were linked to psoriasis information, BMD measurements and lifestyle factors from the third survey of the Nord‐Trøndelag Health Study 2006–08 (HUNT3); socioeconomic data from the National Education Database; and use of medication from the Norwegian Prescription Database.

Results

Among 48 194 participants in HUNT3, we found no increased risk of forearm or hip fracture in 2804 patients with self‐reported psoriasis [overall age‐ and sex‐adjusted hazard ratio 1·03, 95% confidence interval (CI) 0·82–1·31]. No clear association was found between psoriasis and mean BMD T‐score; overall age‐ and sex‐adjusted differences in total hip, femoral neck and lumbar spine BMD T‐scores were 0·02 (95% CI −0·11 to 0·14), 0·05 (95% CI −0·06 to 0·17) and 0·07 (95% CI −0·09 to 0·24), respectively. No clear association was found between psoriasis and prevalent osteoporosis in either total hip, femoral neck or lumbar spine; overall age‐ and sex‐adjusted odds ratio was 0·77 (95% CI 0·54–1·10). Associations did not change substantially after adjustment for education, smoking, systemic steroid use and body mass index.

Conclusions

We found no association between psoriasis and risk of fracture. The study did not indicate reduced BMD T‐score or higher prevalence of osteoporosis among patients with psoriasis.

Psoriasis is a common, chronic inflammatory skin disease with possible systemic inflammatory impact.1,2 The disease is associated with other chronic inflammatory diseases, including psoriatic arthritis and Crohn disease.2 An association between psoriasis and osteoporosis has been reported,3,4 and chronic inflammation has been put forward as a potential underlying biological link between psoriasis and several comorbidities, including osteoporosis.5,6,7 However, previous studies of the association between psoriasis and bone mineral density (BMD) are few and mainly small, with cross‐sectional designs and limited information with regard to medication use and other potential confounding factors,8,9,10,11,12,13,14 and the results are inconsistent. Fractures are considered the main clinical consequence of osteoporosis and are associated with a high morbidity, mortality and economic burden, with 43 000 deaths in the European Union attributable to fracture in 2010.15 In this prospective, comprehensive, population‐based Norwegian study, we assessed whether psoriasis was associated with increased risk of forearm or hip fracture. We also investigated the cross‐sectional association between psoriasis and BMD‐T score, as well as the association between psoriasis and osteoporosis.

Materials and methods

We used the unique 11‐digit identification number of all Norwegian citizens to link data from the third Nord‐Trøndelag Health Study (HUNT3; 2006–08) to hospital‐derived information on forearm and hip fractures in Nord‐Trøndelag County (1995–2013);16,17,18 the Norwegian Prescription Database (NorPD), established in 2004;19,20 and the National Education Database (NUDB), which provides Norwegian education statistics since 1970.21

Study population

HUNT is an ongoing population‐based study, which has collected data in three surveys: HUNT1 (1984–86), HUNT2 (1995–97) and HUNT3 (2006–08). The surveys include clinical measurements, blood sampling and questionnaires on general health measures and a broad range of self‐reported diseases and symptoms. In each survey, all inhabitants in Nord‐Trøndelag County, aged 20 years or older, were invited to participate.17 The present study included participants of HUNT3, in which 93 860 adults were invited and 50 805 (54·1%) participated. Among participants in HUNT3 born after 1 January 1921 (n = 50 329), we excluded a total of 2135 people owing to missing information on psoriasis, education, smoking or body mass index (BMI), leaving 48 194 participants (26 270 women; 21 924 men) for the present study.

Exposure: psoriasis

Psoriasis was defined by an affirmative answer to the question ‘Have you had or do you have psoriasis?’ included in a general health‐related questionnaire sent to all participants. Of 2928 (missing n = 22) identified subjects, 2082 (71%) reported age of onset of psoriasis. Additional psoriasis‐related information was obtained from a separate psoriasis questionnaire given to individuals with self‐reported psoriasis and completed by 2094 patients (71%). The psoriasis question has previously been validated in the HUNT3 population.22

As a proxy for psoriasis severity, we linked information on self‐reported psoriasis to data on previously dispensed systemic treatment of psoriasis from the NorPD. This national database contains nearly complete information of dispensed prescribed drugs, classified according to the Anatomical Therapeutic Chemical (ATC) classification system, to all noninstitutionalized Norwegian citizens.19 This method for severity classification has been previously validated, and the results indicated a sensitivity of 98%.23 The following drugs were defined as antipsoriatic: geralen (ATC‐code D05BA03), acitretin (D05BB02), fumaric acid (D05BX51), methotrexate (L04AX03 and L01BA01, excluding infusion), efalizumab (L04AA21), etanercept (L04AB01), adalimumab (L04AB04) and ciclosporin (L04AD01). The NorPD includes all prescribed and purchased biologic psoriasis agents available in Norway within the actual time period, except infliximab (L04A B02) as this drug is administrated as a hospital infusion. As no dermatological clinic provides infliximab treatment in Nord‐Trøndelag, the patients with psoriasis eligible for this treatment were admitted to the Department of Dermatology at St Olavs Hospital in Trondheim. Here, 23 patients were treated with infliximab from August 2001 to February 2007, and all had previously been treated with methotrexate.24 We therefore consider it likely that these patients were captured by our definition of moderate‐to‐severe psoriasis by earlier prescription of conventional systemic therapies.

Outcome: forearm and hip fractures

Data on first forearm or hip fracture [International Classification of Diseases (ICD)‐9813 and 820, ICD‐10 S52 and S72] was obtained from The Nord‐Trøndelag Hospital Trust. It provides prospectively recorded and subsequently validated information on forearm and hip fractures in patients older than 16 years who were treated or followed up from 1995 to 2013 at the only two hospitals in Nord‐Trøndelag County. A fracture was defined when (i) the ICD code was accompanied by a medical record confirming forearm or hip fracture; or (ii) a fracture was diagnosed by X‐ray. Details about the classification and validity of this fracture information have been published previously.18,25

Outcome: bone mineral density and osteoporosis

Among HUNT3 participants born after 1 January 1921 and living in one of the five largest municipalities, 14 247 participants were eligible for dual‐energy X‐ray absorptiometry (DXA) BMD measurements, and 11 772 subjects (82·6%) participated. In total, 938 participants were excluded because of invalid BMD measurements (e.g. owing to prosthesis or osteosynthesis materials) or missing information on psoriasis question, education, smoking or BMI, leaving 10 784 participants with valid BMD measurements for the analyses. The sample consisted of a 10% random sample of the total HUNT cohort, as well as selected groups according to sex, age and a broad spectrum of self‐reported lung symptoms. To obtain estimates representative of the total HUNT3 population, we applied probability weights in the statistical analyses to weight participants with BMD measurements to the total HUNT3 study population. BMD was measured and T‐score calculated at the total hip, femoral neck and lumbar spine by Lunar Prodigy Advance (Scanex Medical Systems AS, Strømmen, Norway). Regular phantom calibration of the densitometer was performed according to the existing densitometry procedures and quality‐assessment guidelines at HUNT. According to guidelines from The International Society for Clinical Densitometry 6th Adult Position Development Conference, lumbar spine BMD T‐score was calculated as the mean of the BMD T‐score in L1–L4, excluding vertebrae likely to introduce areas with high BMD not related to bone composition, for example degenerative changes, compression of vertebrae or atherosclerosis of the aorta, as indicated by a > 1·0 BMD T‐score difference from an adjacent vertebra.26 For hip measurements, the lowest of either left or right total hip and femoral neck BMD T‐score was used.26 BMD was dichotomized for each site by a T‐score ≤ −2·5, the cut‐off for defining osteoporosis according to the World Health Organization (WHO).27 From NorPD we included prescription of antiosteoporotic drugs between January 2004 and date of participation in HUNT3, as an alternative criterion for osteoporosis. The following drugs were defined as antiosteoporotic drugs: bisphosphonates (ATC code M05BA and M05BB), denosumab (M05BX04), parathyroid hormone (H05AA03), teriparatide (H05AA02) and raloxifene (G03XC01).

Covariates

Information on highest level of education (≤ 9 years, 10–12 years, ≥ 13 years), was collected by linkage to the NUDB.21 From NorPD we obtained information on dispensing of prescribed systemic steroid treatment (ATC code H02) between January 2004 and date of participation in HUNT3. All participants in HUNT3 were asked about smoking history (categorized as current, former or never smokers) and weekly physical activity. Frequency, duration and intensity were combined to form a physical activity index, which was further categorized into four groups. Details of the index have been published previously.28 Weight was measured with the participants wearing light clothes and no shoes, and BMI was calculated by dividing body weight with the square of body height, expressed in kg m⁻². Waist circumference was measured after maximal expiration at the height of the umbilicus by using a nonstretchable measuring tape.16 High‐sensitivity C‐reactive protein (CRP) was measured in 50 053 HUNT3 participants (99%) and categorized in quintiles.29

Statistical analysis

All statistical analyses were conducted using Stata for Windows (version 13.1; StataCorp, College Station, TX, U.S.A.). The association between psoriasis and a first‐time forearm or hip fracture was estimated in Cox regression models, where people were followed up from the date of participation in HUNT3 until a first‐time forearm or hip fracture, death, emigration or end of follow‐up (31 December 2012), whichever occurred first. In total, 1831 patients were excluded from this analysis owing to a hospital record of first forearm or hip fracture between 1995 and participation in HUNT3. Additionally, we assessed the association between duration of psoriasis and first‐time forearm or hip fracture, using 20 years’ duration (the median duration of psoriasis in this population was 21 years) to separate between shorter and longer exposure to psoriasis. To investigate the association according to severity of psoriasis, we defined self‐reported psoriasis with dispensed systemic treatment as moderate/severe psoriasis, self‐reported psoriasis without dispensed systemic treatment as mild psoriasis and patients with no self‐reported psoriasis as reference group. For this categorization, we only considered drugs dispensed between 2004 and maximum 6 months later than the patient's participation date in HUNT3. For Cox regression analysis the proportional hazard assumption was evaluated by the estimation and plotting of Schoenfeld residuals and the proportional hazards assumption tests in Stata. Owing to insufficient statistical power, no separate analyses of forearm and hip fractures were undertaken.

For the subgroup analysis of psoriasis and BMD T‐score, statistical analyses were performed using the survey tabulate command and weighting the BMD samples using probability weights that denote the inverse of the probability that the patient was included because of the sampling design. This method means that our point estimates will not systematically differ from those observed if all participants in HUNT3 had BMD measurement T‐scores, whereas the calculation of confidence intervals (CIs) correctly accounts for the number of patients that actually participated in the BMD measurement programme. The association between psoriasis and mean BMD measured at each site was estimated using linear regression analyses, and the association between psoriasis and osteoporosis was examined using logistic regression analyses.

Analyses were performed with both sexes combined, as well as stratified by sex. We found no evidence of interaction between psoriasis and age (< 55/≥ 55 years). Possible confounding factors were identified by a priori knowledge, and in three separate models adjustments were made for education, smoking, systemic steroid use and BMI (continuous). According to a recent Cochrane review, osteoporosis is not considered to be a systemic side‐effect of topical steroid use in adults, and adjustment was not made for this use.30 As an alternative measure of adiposity, we also adjusted for waist circumference in a separate model. In sensitivity analyses, we adjusted for high‐sensitivity CRP and physical activity. In another sensitivity analyses, we excluded 330 participants with self‐reported psoriatic arthritis.

Ethics

The study was approved by the Regional Committee for Medical and Health Research Ethics in Mid‐Norway. All participants in HUNT3 have provided signed written, informed consent.

Results

Characteristics of the HUNT3 population are given in Table 1. In general, participants with self‐reported psoriasis were slightly older, had a higher BMI, reported more tobacco use and had slightly less education than participants without self‐reported psoriasis. According to sample design, participants with BMD measurements were weighted to obtain a representative sample of the HUNT3 population, and general characteristics did not differ substantially from the total HUNT3 population.

We found no increased risk of forearm or hip fracture according to self‐reported psoriasis in the HUNT3 population (Table 2). The overall age‐ and sex‐adjusted hazard ratio (HR) was 1·03 (95% CI 0·82–1·31), with HRs being 1·24 (95% CI 0·95–1·60) in women and 0·55 (95% CI 0·30–1·00) in men. The associations did not change substantially after adjustment for education, smoking, systemic steroid use and BMI. Additional adjustment for waist circumference, high‐sensitivity CRP and physical activity did not change the results (data not shown). Subgroup analyses excluding participants with self‐reported psoriatic arthritis yielded essentially similar results overall, but sex‐stratified analyses suggested some evidence of a reduced fracture risk associated with psoriasis in men [age, education, systemic steroid, smoking and BMI‐adjusted HR 0·39 (95% CI 0·19–0·83)], but not in women (P‐value for interaction with sex = 0·001).

We found no evidence that more (overall age, sex, education, systemic steroid use, smoking and BMI‐adjusted HR 1·14 (95% CI 0·79–1·64) or less (overall age, sex, education, systemic steroid use, smoking and BMI‐adjusted HR 1·15, 95% CI 0·74–1·77) than a 20‐year duration of psoriasis was associated with increased fracture risk compared with people without psoriasis. When we considered systemic treatment of psoriasis as an indicator for disease severity, we found no clear association between fracture risk and moderate/severe psoriasis (overall age, sex, education, systemic steroid, smoking and BMI‐adjusted HR 1·68, 95% CI 0·90–3·14) or mild psoriasis (overall age, sex, education, systemic steroid, smoking and BMI‐adjusted HR 0·98, 95% CI 0·76–1·26).

No association was found between psoriasis and total hip, femoral neck or lumbar spine BMD T‐score (Table 3). The overall age‐ and sex‐adjusted differences in total hip, femoral neck or lumbar spine BMD T‐score according to self‐reported psoriasis were 0·02 (95% CI −0·11 to 0·14), 0·05 (95% CI −0·06 to 0·17) and 0·07 (95% CI −0·09 to 0·24), respectively. Furthermore, we found no association between psoriasis and osteoporosis in lumbar spine, total hip or femoral neck (Table 4). The overall age‐ and sex‐adjusted odds ratio (OR) was 0·77 (95% CI 0·54–1·10), with the OR being 0·91 (95% CI 0·63–1·31) in women and 0·59 (95% CI 0·27–1·29) in men. The associations did not change substantially after additional adjustment for education, smoking, systemic steroid use and BMI. Additional adjustment for waist circumference, high‐sensitivity CRP and physical activity did not change the results. Subgroup analyses excluding participants with self‐reported psoriatic arthritis, and analyses including participants with prescribed antiosteoporotic medication as an alternative criterion for osteoporosis, yielded similar results. When we considered systemic treatment of psoriasis as an indicator of disease severity, we found no clear association between psoriasis severity and osteoporosis among the subpopulation of HUNT3 participants with DXA BMD measurements [the overall fully adjusted OR for osteoporosis was 0·21 (95% CI 0·04–1·02) among patients with moderate/severe psoriasis, and 0·85 (95% CI 0·58–1·24) for mild psoriasis, compared with patients with no psoriasis].

Discussion

In this population‐based study, we found no clear association between psoriasis and increased risk of forearm or hip fracture, reduced BMD T‐score or higher prevalence of osteoporosis.

Our findings are in contrast with several other studies. A Taiwanese case–control study within a large insurance database from Taiwan reported 65% increased odds of osteoporosis among patients with psoriasis,4 and a large cross‐sectional study from Israel utilizing a large healthcare provider database indicated an 86% increased prevalence of osteoporosis among men with psoriasis (but not among women).3 Furthermore, a hospital‐based cross‐sectional study from Brazil suggested a higher prevalence of X‐ray‐verified vertebral fractures among patients with psoriasis compared with the general Brazilian population.31 However, the results of our study are supported by a recent prospective study from Germany, where a high‐resolution computed tomography scan of the forearm was performed to evaluate quantitative and qualitative changes of bone in patients with psoriasis. No differences in bone architecture could be shown in patients with psoriasis compared with healthy controls.32 Interestingly, a Swedish study reported a 0·7 higher hip BMD T‐score in 35 postmenopausal women with ultraviolet B (UVB)‐treated psoriasis compared with controls, although this association may reflect specific effects of the UVB treatment.

A major strength of our study is the population‐based design, which provided information less prone to bias than diagnostic information collected primarily for insurance or health‐providing purposes. Furthermore, the large sample size in our study contributed to more precise estimates, and linkage to individually validated hospital data provided information on fractures, a hard end point when it comes to clinical implications of osteoporosis. The prospective registration of fractures enabled assessment of the temporal direction of the association, which is not possible in cross‐sectional studies. Also, BMD measurements in a subsample of more than 10 000 patients enabled analyses of both mean BMD T‐score and assessment of osteoporosis defined in accordance with the WHO definition. Finally, the broad range of information in the HUNT study enabled us to examine potential confounding factors such as education level, smoking, systemic steroid use and obesity. Nonetheless, as in all observational studies, the results should be interpreted with caution owing to the possibility of unknown or unmeasured confounders.

As in similar studies, a limitation of our study is the use of self‐reported psoriasis not diagnosed by a physician or a dermatologist. In 2010 we performed a validation study to clarify the validity of the psoriasis question used in HUNT3, and found a positive predictive value of 78% (95% CI 69–85).22 The majority of the false‐positive cases had a history of psoriasis that could not be verified by a dermatologist or previous skin biopsy (e.g. guttate psoriasis), and our study indicated self‐report of psoriasis to be a valid instrument for studies of psoriasis in HUNT3. A potential limitation is the rather high number of nonparticipants in the bone densitometry analysis. This was mainly owing to sickness among the technical staff performing the densitometry, and can therefore be considered to be missing completely at random and not to induce bias. Furthermore, the 50 805 participants in the HUNT3 survey represented 54% of the overall number invited. However, a nonparticipation study indicated that the HUNT3 survey rather underestimated the prevalence of several chronic diseases.33 Regarding the potential impact of low participation on the observed associations, it seems unlikely that psoriasis played any major role in the decision of whether those invited would participate or not; consequently, the risk of bias in the comparison between the two groups is considered small. It should be acknowledged that because we included all participants with psoriasis in a general population, a large proportion of them had mild psoriasis. Therefore, we cannot exclude the possibility that certain psoriasis subgroups, for example severe psoriasis, early‐onset disease or psoriatic arthritis, may be associated with osteoporosis and increased fracture risk. After excluding participants with psoriatic arthritis, we found some evidence that psoriasis could be associated with a reduced fracture risk in men. However, despite the fact that the CI supports that there is a protective association, this finding was based on a low number of fractures among men with psoriasis, and we found no corresponding association in women. Consequently, this finding should be interpreted cautiously.

In this large population‐based prospective study we found no association between psoriasis and risk of fracture. In addition, the study did not indicate reduced BMD T‐score or a higher prevalence of osteoporosis among patients with psoriasis. Consequently, our findings suggest no need for increased awareness of osteoporosis and fracture risk among people with psoriasis in a general population.

Acknowledgments

The Nord‐Trøndelag Health Study (HUNT Study) is a collaboration between HUNT Research Centre (Faculty of Medicine, Norwegian University of Science and Technology), Nord‐Trøndelag County Council, Central Norway Health Authority and the Norwegian Institute of Public Health. We want to thank clinicians and other employees at Nord‐Trøndelag Hospital Trust for their support and for contributing to data collection in this research project.

References

Author notes

Funding sources This study was funded by a Liaison Committee between the Central Norway Regional Health Authority (RHA) and the Norwegian University of Science and Technology (NTNU). The funder had no involvement in study design, data collection, data analysis, manuscript preparation and/or publication decisions.

Conflicts of interest None declared.

Plain language summary available online