Osteoporosis International

, 19:1379

Systematic review of trends in prophylaxis of corticosteroid-induced osteoporosis: the need for standard audit guidelines

Authors

    • Department of Pharmacotherapy and Pharmaceutical CareUniversity of Groningen
    • Pharmacy DepartmentAntonius Hospital Sneek
  • M. Naunton
    • School of ScienceCharles Darwin University
  • E. N. van Roon
    • Department of Pharmacotherapy and Pharmaceutical CareUniversity of Groningen
    • Pharmacy DepartmentTjongerschans Hospital
    • Pharmacy DepartmentMedical Centre Leeuwarden
  • G. A. W. Bruyn
    • Department of RheumatologyMedical Centre Leeuwarden
  • J. R. B. J. Brouwers
    • Department of Pharmacotherapy and Pharmaceutical CareUniversity of Groningen
    • Pharmacy DepartmentTjongerschans Hospital
    • Pharmacy DepartmentMedical Centre Leeuwarden
Review

DOI: 10.1007/s00198-008-0598-y

Cite this article as:
Duyvendak, M., Naunton, M., van Roon, E.N. et al. Osteoporos Int (2008) 19: 1379. doi:10.1007/s00198-008-0598-y

Abstract

Summary

Corticosteroid-induced osteoporosis (CIOP) is currently undertreated. Systematic review of the literature revealed that the percentage of patients treated adequately is dependent on study quality. Therefore, it remains unknown whether adherence to the guidelines is really so poor. Five major quality criteria provide the standard for future studies on this scope.

Introduction

It has recently been stated that the degree of prophylaxis of corticosteroid-induced osteoporosis (CIOP) is low and effort should be put into determining reasons for non-prescribing of preventive agents. The aim of this study was to identify: how many studies adequately audit the prevalent guideline; the longitudinal trends in prevention of CIOP; which patient groups appear to be most undertreated; and which intervention strategies are effective.

Methods

We performed a comprehensive search of MEDLINE and systematically recorded the outcomes and quality of published studies, using five major criteria.

Results

Twenty-four studies were included in the analysis. The quality of the included studies was poor (31%) or moderate (37%). There was a longitudinal increase in quality of the studies and percentage of prevention. Multivariable linear regression showed that the quality of the study was the only independent predictor of the prevention rate reported in the study.

Conclusions

The results show undertreatment of CIOP might be due to insufficient quality of the studies rather than poor practice or failure to recognise the right patients. Future interventions should comply with five major quality criteria, and a multifaceted approach is required in order to make an impact on the underprescribing of CIOP prophylaxis.

Keywords

BisphosphonatesCorticosteroidGuideline adherenceOsteoporosisProphylaxisStudy quality

Introduction

Corticosteroids (CS) are widely used for an extensive number of indications. They are used primarily for their anti-inflammatory and immunosuppressive effects in the treatment of many conditions including chronic obstructive airways disease, rheumatic disorders, inflammatory bowel disease and haematological disorders. Two cross-sectional studies in England showed that 0.5–0.9% of the population were taking oral CS for periods longer than 3 months [1, 2]. In the elderly >55 years this increased to 1.4%.

Corticosteroids have many side effects, including adrenal suppression, susceptibility to infection and increased insulin tolerance. Moreover, long-term use is associated with bone fragility and an increased risk of fracture, predominantly in the elderly and patients using high doses [3]. Today, corticosteroid-induced osteoporosis (CIOP) is probably the most common secondary type of osteoporosis [4]. It is estimated that as many as 50% of patients requiring long-term (>3 months) CS will ultimately suffer fractures [4, 5].

Yood stated that the degree of prevention of CIOP is low and that research aimed at determining the degree of prophylaxis should no longer be performed [6]. Yood’s recommendations were to put more effort in determining the reasons for non-prescribing and selection of effective intervention strategies.

This paper systematically reviews studies conducted on the use of preventive medication in CS users with the aim of identifying: 1) how many studies adequately audit the prevalent guideline in that region; 2) what are the longitudinal trends in prevention of CIOP; 3) which patient groups appear to be most undertreated; and 4) which intervention strategies appear to be the most effective. Analysis of studies published in this domain may lead to a more effective intervention strategy.

Methods

To identify studies investigating the frequency of osteoporosis prevention in CS users a comprehensive MEDLINE search was conducted using the following combination of MESH-terms: “corticosteroids or glucocorticoids” and “osteoporosis” and “prevention or prophylaxis”. The search was limited to English language papers from 1990 to January 2007.

Papers describing original studies on prevention of CIOP in patients >18 years were selected independently by two of the authors (MD, MN). Additional references were identified from bibliographies from the selected studies.

Studies were included only if the primary endpoint of the study was the percentage of patients receiving preventive medication for CIOP. Studies were excluded from the review if no (retrievable) guideline was referenced in the published audit.

Studies were also excluded if the minimum length of CS use for inclusion of patients was not stated in the methods of the article or was shorter than 3 months.

Studies were included and excluded after consensus was reached between three authors (MD, MN, EvR). A few studies have determined results of specific subgroups. Subgroup analyses were included separately if the methodology of a subgroup analysis met the inclusion criteria of our review. Data on study outcomes were extracted using a standardized data record form. Data extraction was performed separately for the included subgroups.

The following outcomes were recorded: the percentage of patients receiving bisphosphonates, percentage of patients receiving any “prescription only” prophylaxis (any treatment (any Tx)) excl. Calcium/Vitamin D (Ca/Vit D)), percentage of patients receiving any prophylaxis (any Tx incl. Ca/ Vit D)), and percentage of patients with a bone mineral density (BMD) measurement.

To assess study quality the following parameters were selected and recorded:
  1. 1.

    Was the length of CS use for patient inclusion according to the guideline used in the study?

     
  2. 2.

    Was the minimum CS dose for patient inclusion according to the guideline used in the study?

     
  3. 3.

    Were the prophylactic treatments selected as endpoints according to the guideline used in the study?

     
  4. 4.

    Were the BMD cut-off values of the guideline used for patient inclusion and exclusion?

     
  5. 5.

    Was the referenced guideline prevalent at time of patient inclusion? For example, weren’t patients included before the referenced guideline was published?

     
In order to assess the scientific quality of the studies three authors (MD, MN, EvR) independently judged the included studies using a standardised scoring-system. A guideline overview chart was constructed to assist in the assessment (Table 1). For each of the aforementioned quality parameters, a maximum of 1 point could be scored. The total score for each study was calculated (range: 0–5). A quality score of 0–2 was judged as poor, 3 as moderate, and 4–5 as good. Consensus was reached between the aforementioned authors in case of initial disagreement.
Table 1

Overview of guidelines on CIOP

Guideline

Guideline type

Screening inclusion CS use

independent risk factors for treatment indication

>6 months

>3 months

all dosages

>5 mg/day

>7.5 mg/day

fracture

BMD T<-2.5

BMD T<-1.5

BMD T<-1

>15 mg/day

>65 years

Hypogonadism

UK 95 [44]

Consensus

+

 

+

 

 

+

+

 

 

+

 

Men: Testos

ACR 96 [45]

Consensus

+

 

+

  

+

  

+

  

Men: Testos

UK 98 [72]

Consensus

+

   

+

+

 

+

 

+

+

HRT

NOS 98 [73]

Consensus

+

   

+

+

 

+

 

+

+

HRT

NOS 99 [74]

Consensus

+

   

+

+

 

+

 

+

+

HRT

SAfr 00 [75]

Consensus

+

   

+

  

+

   

HRT

ACR 01 [46]

Consensus

 

+

 

+

    

+

  

HRT

Tas 01 [22]

Consensus

 

+

  

+

+

 

+

 

+

 

HRT

Green 02 [76]

Consensus

 

+

  

+

+

 

+

   

HRT

NOS 02 [77]

EBM

 

+

+

  

+

 

+

  

+

 
 

Treatment 1st choice

Treatment 2nd choice

Calcium/Vit D

Guideline

HRT

BIS

Calcitriol

Calcitonin

Raloxifen

Fluoride

HRT

BIS

Calcitriol

Calcitonin

Fluoride

All

Dependent on diet/ sunlight

UK 95

PMP only

      

+

+

+

   

ACR 96

+

  

fractures

   

PMP only

 

+

 

+

 

UK 98

 

+

+

+

 

+

      

+

NOS 98

 

+

      

+

+

 

+

 

NOS 99

PMP only

+

+

         

+

SAfr 00

 

+

+

         

+

ACR 01

 

+

       

only prevalent CS-users

 

+

 

Tas 01

 

+

  

+

   

+

  

+

 

Green 02

PMP only

+

  

+

      

+

 

NOS 02

PMP only

+

+

+

        

+

Abbreviations: CS: Corticosteroid; BMD: Bone Mineral Density; Testos: Testosterone; HRT: Hormone replacement therapy; PMP: Postmenopausal women; BIS: Bisphosphonates; Vit D: Vitamin D; UK: United Kingdom; ACR: American college of rheumatology; NOS: National osteoporosis society; SAfr: South Africa; Tas: Tasmania, Australia; Green: Greenwich, UK; EBM: evidence based medicine

For the included studies we determined the year of data collection. If this was not stated we assumed data were collected in the year of publication. Scatter plots of the different outcomes were constructed, grouping the individual studies by year.

Statistics

All analyses were performed using SPSS version 12 (SPSS, Chicago, IL United States).

We used a composite endpoint for the studies: ‘overall treatment’: this was either any Tx (excl. Ca/Vit D) or any Tx (incl. Ca/Vit D), depending on which one was reported. In cases where both were reported any Tx (incl. Ca/Vit D) was used. We performed multivariable linear regression analyses to determine which factors contributed significantly to the composite endpoint ‘overall treatment’ for the studies. The variables that we assessed for the model were: number of subjects, mean/median age of population, % female, mean/median dose of CS (prednisolone equivalents in mg), mean/median length of therapy (months), the study quality score, and year of data collection. Where data was missing for the variables mean/median age of population, % female, and mean/median dose of CS (prednisolone equivalents in mg), the mean value of the known data was used for imputation. A variable was included in the model if univariate regression analysis resulted in a p−value <0.05. A p−value <0.05 was considered significant.

Results

Overview of retrieved studies

The initial MEDLINE search retrieved 198 articles of which 36 were selected for full paper analysis. We retrieved 19 studies investigating the degree of prevention of CIOP and another 20 studies from references contained within these 19 studies. Of these 39 studies, 24 studies [1, 729] (Fig. 1) and eight subgroup analyses [7, 12, 14, 18, 20, 22, 25] met the inclusion criteria. 15 studies were excluded [2, 3043]. Table 2 describes the characteristics, study population, and osteoporosis prophylaxis of the 24 studies identified. The average pharmacological prophylaxis rate was 31 ± 25% (range 1–86) for bisphosphonates, 41 ± 23% (range 7–86) for any antiresorptive treatment (excl. Ca/Vit D) and 54 ± 30% (range 11–93) for any treatment (incl. Ca/Vit D).
https://static-content.springer.com/image/art%3A10.1007%2Fs00198-008-0598-y/MediaObjects/198_2008_598_Fig1_HTML.gif
Fig. 1

Flowchart of article selection for review

Table 2

Study characteristics and results

Study (year of publication, publication type)

Number of Subjects (% women)

Setting Population

Mean/Median CS dose (mg/day)

Minimal and mean/median Length of CS-use

Prophylaxis

Guideline

Methods

Comments

Walsh (1996) [1]

303 (65)

General practice patients+ hospital outpatient clinic, Notinghamshire, UK (1995)

Mean = 8

>3 months, 3 years (0.3–37) (n = 149)

Any Tx (excl Ca/Vit D) = 9%

UK, 1995 [44]

Retrospective review of 65,786 patients from computerised GP records. Paper review of n = 149

 
 

Median = 6.8

HRT = 8%

Calcium = 2%

Calcium+Vit D = 2%

Vitamin D = <1%

Bisphosphonates = 1%

Bell (1997) [7]

100 (NS)

Hospital outpatients taking at least 1g of prednisolone in the previous 6 months, UK

9.8 (±6.1)

>6 months, NS

Patients with diagnosis of osteoporosis (n = 27):

UK, 1995 [44]

retrospective review of 4816 case notes of patients admitted to medical clinic

Data of 4 pts not specified in article. 53% of pts where diagnosis of osteoporosis was not considered were not reviewed.

Any Tx (excl Ca/Vit D) = 81% = 22% of total.

Etidronate = 57% = 16% of total

Buckley (1999) [8]

147 (58)

Hospital outpatients, academic and associated veteran hospital, Virginia, USA, (Aug 1997-Feb 1998)

10 (±6)

>3 months, 1–2 years

Any Tx (excl Ca/Vit D) = 21%

ACR, 1996 [45]

>5 mg per day identified through pharmacy records and then interviewed by phone

28% of patients refused cooperation.

Bisphosphonates = 7%

Any Tx incl. ca/vit D = 40%, but vit D use is 45%. Any Tx couldn’t be reconstructed from original data.

HRT = 40% (in PMP women, 12% of total)

Calcium = 29%

Vitamin D/multivitamin = 45%

Hougardy (2000) [17]

NS (NS)

Academic hospital inpatients Hobart, Australia (Sep 1999-Jan 2000)

NS

>6 months, NS

>7.5mg/day, >6 months:

UK NOS, 1998 [73]

Retrospective review of medical records of all patients using inhaled or oral CS

 

Any Tx (incl. Ca/Vit D) = 65%

BMD = 40%

Rothberg (2000) [24]

1614 (54)

Community patients, South Africa (1998)

NS

>3 months, NS

Any Tx (incl. Ca/Vit D) = 11%

South Africa, 2000 [75]

Retrospective review of patients pharmaceutical and clinical claims data

Both oral and inhaled CS, dose not specified.

 

Bisphosphonates = 3%

Calcium±Vit D = 6%

HRT (women>40 years) = 2%

Testosterone (men) = 0.4%

Elliott (2000) [12]

72 (0)

Community veteran patients from hospital pharmacy database, Wisconsin, USA

12.5 (7.5–37.5)

>6 months, 30 (6–74) months

Bisphosphonates = 10%

ACR, 1996 [45]

Retrospective review

 

(Apr-Oct 1997)

Calcium±Vit D = 69%

BMD = 60%

Low T-score (-1): (n = 32)

Any Tx (excl. Ca/Vit D) = 38%

Bisphosphonates = 22%

Naganathan (2000, letter) [21]

69 (NS)

Rheumatology clinic outpatients, Australia (Apr-Oct 1998)

7 (5–75)

≥6 months, 58 (6–445) months

Any Tx (incl. Ca/Vit D) = 18%

UK, 1998 [72]

Prospective cross-sectional cohort study investigating the effect of steroids on fractures

BMD = 100% because of study protocol.

Osiri (2000) [23]

365 (59)

Academic hospital, outpatient clinic (Jun 1993-Apr 1999)

11.4 (±6.7)

≥3 months, 63 (±70) months

Any Tx (incl. Ca/Vit D) = 29%

ACR, 1996 [45]

Retrospective review of medical records

 

Bisphosphonates = 10%

Calcium = 25%

Vit D = 13%

HRT = 6%

Calcitonin = 1%

Smith G (2001, letter) [26]

15 (NS)

Neurology department

26 (10–43)

≥6 months, NS

Bisphosphonates = 73%

UK NOS, 1999 [74]

Retrospective review over 5 years of Myasthenia gravis patients

Patients included before guidelines came out. Cumulative dose used

Myasthenia gravis ambulatory patients

In first 6 months of CS-use

Lewis (2001) [18]

18, 13 need prophylaxis independent of BMD (NS)

Myasthenia gravis ambulatory patients, Neurology Department,

Mean 16.6 (7.5–40)

≥6 months, NS

Any Tx (incl. Ca/Vit D) = 89%

UK, 1998 [72]

Retrospective audit of MG patients receiving > 7.5mg/day

 

Cardiff, Wales, UK (Jul-Aug 1999)

Bisphosphonates = 61%

Calcium = 17%

Vit D = 11%

Prophylaxis required:

Any Tx (incl. Ca/Vit D) = 85%

Bisphosphonates = 54%

Mudano (2001) [20]

2378 (63)

Members managed care organisation, USA, (Jan 1995-Mar 1998)

Mean = 15.2 (±13.7)

≥3 months, 1.4 ± 1.3 year

Any Tx (excl Ca/Vit D) = 21%

ACR, 1996 [45]

Retrospective review of medical and pharmacy claims database of 3 million patients

9% prior fracture, 13% of all women already on HRT before CS. >80years excluded, no minimum daily dose.

Bisphosphonates = 8%

BMD = 9%

Women > 50:

Any Tx (excl Ca/Vit D) = 41%,

Bisphosphonates = 16%

HRT = 23%

BMD = 16%

Ettinger (2001) [14]

8807 (60)

Health plan members, Northern California, USA

2–3g/yr (32%)

≥12 months, NS

Any Tx (excl Ca/Vit D) = 8.8%

ACR, 1996 [45]

Retrospective review of pharmacy health plan database of 232,782 members using CS

Exclusion of oncology patients. Minimum of 2 g cumulative dose within 12 months, excluded all pts already on preventive medication(n = 416)

(Jan 1997-Dec 1999)

3–4 g/yr =(21%)

Bisphosphonates = 7.4%

>4g/yr (47%)

BMD = 7.0%

Women >65:

Any Tx (excl Ca/Vit D) = 16.3%

Hart (2002) [16]

92 (61)

Hospital general Inpatients, London, UK, (Jan-Sep 1999)

13.6

≥6 months, 7 years

Any Tx (excl. Ca/Vit D) = 41%

UK NOS, 1998 [73]

Prospective study. Review of medical records + Patient interview + questionaire

51 patients qualified for prophylaxis according to guidelines

Any Tx (incl. Ca/Vit D) = 61%

6% had ceased bisphosphonate use

HRT = 14%

Bisphosphonates = 27%

Calcium±Vit D = 20%

Erb (2002) [13]

235 (ca. 1:2; male:female)

Rheumatology patients, West Midlands, UK

7.5–14.9 (86%)

≥6 months, NS

Any Tx (incl. Ca/Vit D) = 86%

UK NOS, 1998 [73]

Audit of 10 rheumatology units during 2 week period (1766 pts)

12% of pts had suffered fracture.

>15 (13%)

Bisphosphonates = 47%

According to guideline: 7% over-treated, 30% under-treated

3 pts unknown

Calcium = 22%

 

Calcium+Vit D = 31%

HRT = 7.2%

Calcitriol = 0.9%

Testosterone = 0.4%

BMD = 43%

Chattopadhyay (2002, letter) [9]

60 (62)

Hospital inpatients, Northern Wales, UK (May - July 2001)

NS

≥6 months, NS

Any Tx (excl Ca/Vit D) = 43%

UK NOS, 1998 [73]

Prospective review of prescription and medical records of 308 patients

 

Bisphosphonates = 40%

HRT = 3.3%

Solomon (2002) [27]

106 (NS)

Academic rheumatology outpatients, Massachusetts, USA (Jan 1998- Feb 2000)

NS, >7.5 mg/day

≥6 months, NS

>7.5 mg >6 months:

ACR, 2001 [46]

Review of medical records and medication lists of 623 patients with diagnosis code of RA in Jan-Feb 1999

Dose determined on first visit of each quarter, no real use checked

Any Tx (excl. Ca/Vit D) = 45%

Dolan (2004) [11]

201 (59%; n = 92)

General Practices, London, UK (2001)

Median 2–5 mg/day (n = 92)

≥3 months, 2–5 years (n = 92)

Overall:

Local [76]

Cross-sectional intervention study + Survey in 92 patients not previously treated

All dosages included. DXA cutt-off: T<-2,5. Only 0.12% of population used CS

Any Tx (excl. Ca/Vit D) = 43%

Intervention (n = 21; BMD<-2.5):

Any Tx (excl. Ca/Vit D) = 48%

Naunton (2004) [22]

120 (63)

Academic hospital general inpatients, Hobart, Australia (2002)

7.5 (3–100)

≥3 months, 60 (3–372) months

Any Tx (incl. Ca/Vit D) = 57%

Local [22]

Patient interview and medical records review. Educational intervention program

All dosages included

Bisphosphonates = 24%

Calcitriol = 15%

Calcium 19%

Vit D = 11%

Raloxifen = 1%

HRT = 14%

Testosterone = 2%

>7.5 mg/day:

Any Tx (incl. Ca/Vit D) = 51%

Walker-Bone (2004) [29]

124 (NS)

Rheumatology outpatients, Portsmouth, UK (Mar - Aug 2002)

NS,

≥6 months, NS

Patients at high risk:

UK, 1998 [72]

Retrospective review of 1290 NHS trust hospital outpatients of which 189 (15%) were using CS. Case notes were reviewed

Patients at high risk because of dose >15mg or other risk factors, e.g., fracture, low BMD, early menopause, >65 years. 7% of pts refused treatment

 

35% of pts used ≥7.5 mg/day

Any Tx (excl. Ca/Vit D) = 61%

 

Bisphosphonates = 56%

HRT = 6%

Curtis (2005) [10]

NS (NS)

Members managed care organisation, USA, (2001–2003)

NS

≥3 months, NS

Women > 50:

ACR, 2001 [46]

Retrospective review of medical and pharmacy claims database of 3 million patients

Ca/Vit D only in survey responders (38%). Dose limit not clear

Any Tx (excl. Ca/Vit D) = 62%

Bisphosphonates = 31%

BMD = 42%

Feldstein (2005) [15]

3031 (60)

Members Health Maintenance Organization, USA (Jan 2000 - Dec 2001)

20(±16.5),>5 mg/day

≥3 months, NS

Any Tx (excl. Ca/Vit D) = 38% (women 57%, men 9%)

ACR, 2001 [46]

Retrospective review of electronic administrative clinical database with 450,000 members

9% had osteoporosis diagnosis, 9% previous fracture

 

Bisphosphonates + SERMS + Calcitonin = 15% (women 18%, men 9%)

Treatment only determined within 6 months of start of CS

HRT = 46% (women, total = 28%)

BMD = 10% (women 13%; men 5%)

Lozsadi (2006) [19]

48 (46)

General Neurology practice, UK (2003)

44 (7.5–100)

≥3 months, NS

Any Tx (incl. Ca/Vit D) = 92%

UK NOS, 1998 [73]

3 years post intervention audit, Prospective review of pharmacy discharge records and outpatient prescriptions of prednisolone and medical records

No routine access to DXA-scans. Determination of average daily dose not clear

Any Tx (excl. Ca/Vit D) = 86%

Bisphosphonates = 86%

Calcium+Vit D = 6%

Saag (2006) [25]

NS (NS)

Members form private insurance or medicare risk plan, USA

      

(Jan 1996 - Dec 2001)

NS, ≥7.5 mg/day

≥6 months, NS

Women ≥50:

ACR, 2001 [46]

Retrospective review of pharmaceutical and medical claims database for 7 million individuals form 20 US states. 3125 patients included in total study

BMD reimbursement after 1997. New users only (no CS 12 months before index date). Determination of average daily dose not clear: 2 claims of oral CS use within 6 months for inclusion. BMD from 12 months prior to start CS. Prophylaxis only if started within 12 months after start CS

Any Tx (excl. Ca/Vit D) = 39%

Bisphosphonates + SERMS + Calcitonin = 15%.

Solomon (2006) [28]

45 (NS)

Academic Hospital Rheumatology Department outpatients, Boston, USA (2003–2004)

NS

≥3 months, NS

CS ≥ 5 mg/day ≥3 months:

ACR, 2001 [46]

Randomized selection of RA patient retrospective review of medical records. Inclusion during 2 months window after educational + case-finding intervention

All dosages. 5% of pts had osteoporosis or fracture

Any Tx (excl. Ca/Vit D) = 40%

Drug use data from rheumatologist notes

BMD = 51%

BMD or Any Tx = 64%

Abbreviations: CS: corticosteroid; Any Tx: Any prophylaxis for CIOP; NS: Not specified in article; Ca: calcium; Vit D: vitamin D; HRT: hormone replacement therapy; PMP: postmenopausal; RA: rheumatoid arthritis; DXA: dual energy X-ray absorptiometry for bone mineral density (BMD) Measurement; SERMS: selective estrogen receptor modulators

Figure 2a provides an overview of the percentage of studies with the various quality scores. The median quality score of the studies was 3 (range 0–5). 10 (31%) studies were of “poor quality” (0–2), 12 (37%) were of “moderate quality” (3) and 10 (31%) were of “good quality” (4–5). Only 3 (9.4%) studies scored the maximum of 5 quality points.
https://static-content.springer.com/image/art%3A10.1007%2Fs00198-008-0598-y/MediaObjects/198_2008_598_Fig2_HTML.gif
Fig. 2

a Overview of quality of studies (n = 32). b Longitudinal patterns of quality of studies

Table 3 provides an overview of the various quality parameters for each study. Failure to use BMD cut-off values for patient in- and exclusion was the most frequently non-compliant parameter with guideline recommendations, with only 9 (28%) studies complying. Secondly, the presented pharmacological endpoint in 16 studies (50%) was not entirely matching the guidelines 1st and 2nd drugs of choice. For example, eight studies presented the endpoints including calcium and vitamin D, and “any treatment” (excl. Ca/Vit D) couldn’t be recalculated from the presented data [12, 13, 17, 2024].
Table 3

Overview of compliance to quality parameters per study

Study

A

B

C

D

E

Total quality score

Elliott [12]

Yes

Yes

Yes

Yes

Yes

5

T-score <-1

Lewis [26] Treatment independent of BMD

Yes

Yes

Yes

Yes

Yes

5

Walker-Bone [29]

Yes

Yes

Yes

Yes

Yes

5

Dolan [11]

Yes

No

Yes

Yes

Yes

4

BMD<-2.5

Hart [16]

Yes

No

Yes

Yes

Yes

4

Lewis [18]

Yes

Yes

Yes

No

Yes

4

Saag [25]

Yes

Yes

Yes

No

Yes

4

Women >50

Smith G [26]

Yes

Yes

No

Yes

Yes

4

Solomon 2002 [27]

Yes

Yes

Yes

No

Yes

4

Solomon 2006 [28]

Yes

Yes

Yes

No

Yes

4

Bell [7] Diagnosis of osteoporosis

Yes

No

Yes

Yes

No

3

Curtis [10]

Yes

No

Yes

No

Yes

3

Dolan [11]

Yes

No

Yes

No

Yes

3

Elliott [12]

Yes

Yes

No

No

Yes

3

Erb [13]

Yes

Yes

No

Yes

No

3

Ettinger [14]

Yes

Yes

No

No

Yes

3

Ettinger [14]

Yes

Yes

No

No

Yes

3

Women >65

Feldstein [15]

No

Yes

Yes

No

Yes

3

Hougardy [17]

Yes

Yes

No

No

Yes

3

Lozsadi [19]

No

Yes

Yes

No

Yes

3

Naunton [22]

Yes

Yes

No

No

Yes

3

>7.5 mg/day

Saag [25]

Yes

Yes

No

No

Yes

3

Bell [7]

Yes

No

No

Yes

No

2

Buckley [8]

No

Yes

No

No

Yes

2

Chattopadhyay [9]

No

No

Yes

No

Yes

2

Naganathan [21]

Yes

No

No

No

Yes

2

Naunton [22]

Yes

No

No

No

Yes

2

Mudano [20]

No

Yes

No

No

No

1

Mudano [20]

No

Yes

No

No

No

1

Women >50

Osiri [23]

No

Yes

No

No

No

1

Walsh [1]

No

No

Yes

No

No

1

Rothberg [24]

No

No

No

No

No

0

Total scored Yes

23

21

16

9

24

 

A Was the length of CS use for patient inclusion according to guideline?

B Was the minimum dosage for patient inclusion according to guideline?

C Were the prophylactic treatments selected as endpoint according to guideline?

D Were the BMD cut-off values of the guideline used for patient in- and exclusion?

E Was the referenced guideline prevalent at time of patient inclusion?

Figure 2b shows the longitudinal trend in the study quality with univariate linear regression analysis demonstrating that this increasing trend was significant (p = 0.04).

Longitudinal patterns of prophylaxis

Figure 3a specifically addresses the longitudinal trends in the use of bisphosphonates. There has been a significant (p < 0.01) increase in the use of bisphosphonates from 1995 to 2003 from approximately 1% to 57%. Figure 3b shows a similar significant trend from approximately 25% to 67% (p = 0.03) for the composite endpoint ‘overall treatment’.
https://static-content.springer.com/image/art%3A10.1007%2Fs00198-008-0598-y/MediaObjects/198_2008_598_Fig3_HTML.gif
Fig. 3

a Longitudinal patterns of bisphosphonate use for CIOP. b Longitudinal patterns of any prophylaxis for CIOP

Predictors of study outcome

Figure 4 shows there is a significant (p < 0.01) relationship between the study quality and the number of patients receiving ‘overall treatment’. In univariate linear regression the parameters mean dose (p < 0.05), number of subjects (p < 0.05), year of data collection (p < 0.05) and study quality score (p < 0.01) showed a significant relationship with the aforementioned endpoint. Imputation of missing data had no influence on the outcome of the analyses.
https://static-content.springer.com/image/art%3A10.1007%2Fs00198-008-0598-y/MediaObjects/198_2008_598_Fig4_HTML.gif
Fig. 4

Relation between study quality and any prophylaxis for CIOP

Multivariable linear regression using all significant variables from the univariate analysis shows that study quality score is the only significant (p < 0.01) positive predictor of the prevention rate reported in the study (i.e., increase study quality, increase % of prevention). There was a trend towards significance for the number of subjects as a negative predictor (i.e., increase number of subjects, decrease % of prevention). The mean CS dose was missing in 57% of the cases and when the average dose of 14.9 mg/day was used for imputation in the multivariable analysis, the study quality was still a significant positive predictor (p = 0.02), together with mean dose (p < 0.05), whereas number of subjects was a negative predictor (p = 0.02).

Factors influencing prophylaxis

Table 4 shows the factors positively associated with prophylaxis of CIOP as mentioned in the included studies. Female sex, older age, previous fracture and rheumatology patients are the factors most often associated with higher prevention rates. Female sex was a positive predictor in all studies except for the study from Ettinger et al. [14]. This study stratified the multivariable analysis by sex because females were receiving significantly more prophylactic agents compared to men. Therefore, female sex was not an outcome in their regression analysis.
Table 4

Factors influencing prophylaxis for CIOP from multivariable logistic regression analysis

study

Higher age

White race

Higher daily dose

Higher cumulative dose

BMD known

Female sex

PMP

Longer duration of CS use

>1 comorbidity

Fracture

>3 months > 5 mg/day

Rheum patients

Buckley [8]

 

+

   

+

     

+

Mudano [20]

     

+

  

+

  

+

Ettinger [14]

+

  

+ (g/year)

     

+

  

Solomon (2002) [27]

     

+

+

 

+

   

Curtis [10]

   

+

 

+ (>50 year)

  

+

+

  

Saag [25]

+

  

+

+

+

     

+

Solomon (2006) [28]

+

    

+

   

+

+

 

Osiri [23]

+

    

+

 

+

 

+

 

+

Abbreviations: BMD: Bone Mineral Density; PMP: Postmenopausal; Rheum: Rheumatology; + represents more prophylaxis than reference group.

Effectiveness of interventions

Two of the included studies in this review studied the effects of a specific intervention on CIOP-prophylaxis. In the study by Naunton et al. [22], a comprehensive educational intervention was undertaken, aimed at increasing the use of pharmacological prevention for osteoporosis in patients with prolonged use of CS. General practitioners, pharmacists, and patients were educated. There was a fourfold increase in bisphosphonate use (in hospitalised patients) from 6% before the intervention to 24% following the intervention.

More recently, Lozsadi et al. performed a post-intervention audit on a small group (n = 48) of neurology patients who were using high dose (mean dose 44 mg/day) oral CS (>7.5 mg, >3 months) [19]. The intervention comprised of a discussion with medical staff about the prevalent guideline and results of a pre-intervention study (2000) which showed that 40% of patients were being treated with bisphosphonates (first line therapy). It should be noted that although the intervention recommended risedronate or alendronate, in the first post-intervention audit (2001), etidronate use surprisingly increased to 67% (36% pre-intervention). The overall use of any prophylaxis was very high (92%) in the 2nd post-intervention audit (2003) and the use of bisphosphonates was also very high (86%; 82% alendronate). The authors in this paper state that access to DXA was not routinely available, and it may be possible that some patients treated with a bisphosphonate did not require it, although given the high mean dose this is unlikely.

Discussion

In 1995 the UK was the first to publish guidelines for the management of CIOP [44]. Other countries, including the USA [45, 46], Australia [47], Belgium [48], and The Netherlands [49, 50], have since then developed similar guidelines. Many studies have published results suggesting undertreatment of the prevention of CIOP. The first studies in the UK (1995 [40], 1996 [1]) demonstrated that as few as 1% of patients on CS were receiving effective treatments such as bisphosphonates. The poor prevention was probably due to lack of regulatory approval for any agent specifically for use in CIOP, lack of reimbursement, lack of dissemination of guidelines, lack of BMD-testing, concerns of side-effects for newly-registered medication, and a lack of studies demonstrating efficacy of bisphosphonates to reduce fracture risk at that time. In addition, only surrogate markers for fracture risk (BMD measurements) were available at that time demonstrating that etidronate might be effective in the prevention and treatment of CIOP [5155]. In 1998, Saag et al. published the first randomised trial demonstrating that alendronate was effective in increasing BMD in patients with CIOP [56]. This was followed by Reid et al. in 2000, showing that risedronate reduced vertebral fractures in CIOP patients [57]. Similar results were produced for alendronate and reported by Adachi et al. in 2001 [58]. Alendronate and risedronate were only approved by the FDA and in Europe for prevention of CIOP in 1999 and 2000 respectively [59, 60]. Collectively, these events must explain why patients treated with long-term CS have appeared undertreated with bisphosphonates.

Study quality

We retrieved 39 studies determining the prophylaxis rate for CIOP, of which 24 used a published guideline for their audit and included patients using long-term CS. The median quality score of these studies was 3 (range 0–5) and only 9.4% of the included studies used all five major criteria for prophylaxis decision making regarding the in- and exclusion criteria of their patients. This means that in only 9.4% of studies all included patients needed prophylaxis according to the guideline. In 72% of studies BMD data were not available or not taken into account, although most CIOP guidelines use BMD cut-off values for clinical decision making. This is also stated in the article by Curtis et al. [10], showing that absolute DXA-rates remain low, although a temporal improvement was present.

The results of our review show there is a significant relationship between the study quality and the reported percentage of patients receiving CIOP-prophylaxis. This raises the question of whether prevention rates are really so low. Although year of study was not a significant factor in our multivariable analysis, longitudinal trend analysis did show an improvement in CIOP prophylaxis, which coincides with an improvement in study quality. The study by Curtis et al. [10] showed that with the rise of DXA testing rates the prophylaxis-rate also improved, although a direct relation between these parameters was not presented. Our univariate regression analysis (data not shown) did not show a significant relation between the percentage of patients receiving DXA and the prophylaxis-rate.

In order to improve study quality and provide an audit standard, we suggest taking the following points into account:
  • Studies should determine the degree of prevention using treatments mentioned in the prevalent guideline excluding Ca/Vit D, since calcium and/or vitamin D monotherapy are not considered effective options for the prophylaxis of CIOP in any of the guidelines (Table 1). Multiple studies suffer from the fact that it is not clear what percentage of the patients receive calcium and/or vitamin D.

  • There is no clear guidance on the risk of inhaled CS; therefore studies should not include patients on monotherapy inhaled CS therapy until further evidence on the risk is available. De Vries et al., using a large population database, have recently shown that inhaled CS is not an independent risk factor for fracture [61]. Furthermore, efficacy of preventive treatment has not been clearly established. Hence, the degree of CIOP prevention is not as low as some report. For example, in the study by Rothberg et al. [24] the authors included subjects on monotherapy inhaled CS. However, monotherapy inhaled CS are not included in most CIOP guidelines, but were mentioned as risk factor for the development of CIOP in the South African general osteoporosis guideline [62] and Australian guidelines [47].

  • Studies should also determine what percentage of patients refuse treatment or become non-adherent to therapy. In the study from Walker—Bone et al. [29] it was shown that 7% of patients refused treatment. Finally, the studies by Hart et al. and Smith et al. showed that 6% and 10%, respectively, of patients had ceased preventive therapy, although the duration of use was not specified [16, 41]. In the other studies reviewed, the outcomes are not reported consistently: either the percentage of patients that are offered treatment by their prescriber or the percentage of patients that actually use it. Adherence to osteoporosis treatment has been shown to be a problem in the management of postmenopausal osteoporosis with studies reporting persistence of bisphosphonate therapy after one year as low as 24–45% [63, 64]. Also, in some countries (e.g., Australia, but not in the UK or The Netherlands) reimbursement might play a role in the initial prescribing and persistence with CIOP prevention/treatment [22].

  • CIOP guidelines generally consist of multiple decision making steps for the clinician to consider in the management of the patient. Thus, in order to determine the degree of prophylaxis according to the guideline of the population at risk, preferably, all the risk factors needed to make a clinical decision should be considered. For example, the study by Walker—Bone et al. [29] presented the results of adherence to the guideline using the guideline-flowchart. The results actually describe the clinical pathway followed. This way it is very easy to grasp the critical steps in the decision making process that need to be targeted for an intervention study. We acknowledge this requires a lot of data gathering, but it enhances the quality of the study. The study by Elliott et al. in male veterans shows the influence of the selection of an endpoint on the outcomes of the study [12]. Of the 32 men with low bone mass only 38% were treated (22% used bisphosphonates). Nineteen of the 32 patients with DXA-documented low bone mass had contraindications to testosterone treatment. Testosterone was measured in 7 of the 13 men who did not have contraindications to therapy. Only one man was hypogonadal and was treated with testosterone. This illustrates eloquently that although treatment with bisphosphonates was low, adherence to the guideline [45] was actually higher than reported (18/32 = 56%).

  • Studies should exclude patients using CS intermittently. In the study by Bell et al. for example, the cumulative dose within a certain timeframe was determined, instead of the average daily dose [7]. Other studies have also included patients with only multiple courses of CS per year [41, 43]. This may lead to inclusion of patients using a high dose in a short period of time (e.g., <3 months or <6 months, depending on the guideline) [65] and who may not strictly qualify for CIOP prophylaxis according to the guideline. The study by Hougardy et al. demonstrated, quite logically, that the degree of prevention is highly dependent on the inclusion criteria [17]. For example, in their study of all CS users only 21% used any treatment. However, in those patients using CS >7.5 mg/day for >6 months 65% used any treatment. This clearly demonstrates that study methods, more than doctor’s compliance to the guideline, determine the outcome of the study.

Longitudinal patterns

Because of the heterogeneity of the studies, we cannot precisely assess the longitudinal patterns of prophylaxis. However, our results clearly show a trend towards a greater degree of osteoporosis prophylaxis in CS users in more recent years. It needs to be taken into account that the guidelines have also changed. A few recent studies conducted in the USA have determined longitudinal patterns in their population [10, 25, 27, 28, 42]. Therefore, longitudinal patterns of CIOP prophylaxis from other areas still need to be assessed. In the study by Curtis et al. the authors didn’t adjust for any change in guidelines which occurred during the study period [10]. The American College of Rheumatology (ACR) guideline recommendations on minimal length of CS use changed from six to three months between 1996 and 2001 (Table 1). Since the study included all patients using CS for more then 90 days, patients in the 1995–1998 group that may not have complied with guideline recommendations and thoughtfully received no treatment by physicians might have been included. Therefore, prophylaxis rates could appear lower than they actually were according to guidelines.

Factors influencing prophylaxis

It is clear that the most dominant factor for receiving prophylaxis is female sex. Men appear systematically undertreated, which has also been shown in studies examining primary osteoporosis [66]. Clearly, health professionals now need to focus more attention on men. However, greater public awareness directed at men will also be required. Other groups that need more attention include younger patients and patients without fractures that need prophylaxis according to guidelines. Prescribers may be reluctant to initiate CIOP prophylaxis in younger patients, particularly females because bisphosphonates have a long half-life and are contraindicated in pregnancy. Alternatively, younger patients may not be willing to accept that they are at risk for fractures and refuse to initiate preventative therapy. Moreover, the absolute fracture risk in younger patients is lower and therefore the threshold for treatment is higher in some guidelines. For example in the Dutch guidelines the threshold for treatment in premenopausal women and men <70 years of age is 15 mg/day of prednisolone equivalents [49, 50]. In addition, the increasing reports of osteonecrosis of the jaw [67, 68] might also hinder prescribers to initiate treatment in younger patients who may require future dental treatment.

Intervention studies should preferably be carried out in the general population where undertreatment appears more common compared to rheumatology patients.

Intervention strategies

This review shows that the number of studies on the efficacy of different intervention strategies is low, since only 2 studies could be included. Two studies were excluded, which had, moreover, no effect on the prescribing of anti-resorptive agents [37, 42].

The two included studies were multifaceted, the main component being education of general practitioners, pharmacy staff and patients [22] or a small group of specialists [19], respectively. Both studies used a historic control group. Therefore it remains unclear what the real effects of these interventions were. Moreover, the increase of etidronate use in the study by Lozsadi et al. [19] instead of the recommended alendronate or risedronate raises the question as to whether the increase was a result of the intervention or of some other factor(s), such as promotion of etidronate by the pharmaceutical industry at that time or changes in drug formularies favouring etidronate [69].

It is our opinion that intervention strategies must be multifaceted [70] and directed at educating prescribers of CS, informing patients when prescribing long-term therapy, and regularly informing patients at the moment of dispensing about the effects of CS on bone. We advocate that this can only be effectively done in a regional setting. National and regional health-care structures have to be used to compose a multidisciplinary team of professionals leading the interventions on the local level [71]. Furthermore, prescribers can be logistically supported by pharmacists, who make CIOP part of their pharmaceutical care program. Alerting/reminding prescribers for the need to consider CIOP prophylaxis after dispensing CS >3 months if the patient is not using adequate prophylaxis, might be effective. This is the scope of our ongoing research. In addition, reminders could be built in to physicians prescribing programs to alert them to consider osteoporosis prophylaxis in patients using long-term CS.

Conclusion

This review has identified that many studies which conduct audits to assess the prophylaxis rate of CIOP have moderate (37%) to poor (31%) quality. The key finding is that the quality of the study is an independent predictor for the degree of prevention for CIOP reported by the study.

It is apparent that patients treated with CS are not receiving appropriate prophylaxis. However, there has been noticeable improvement from the earlier studies, which were conducted in the mid-1990s in the prevention of CIOP, particularly if specific interventions have been undertaken.

Future (intervention) studies that assess CIOP prophylaxis should aim to recruit only patients who require prophylaxis according to the prevalent guidelines. Furthermore, these studies should state clearly which part(s) of the decision making steps, as stated in the prevalent guideline at the time of study, are assessed for adherence.

Many studies/audits report an undertreatment of CIOP and our results show that this might be due to an insufficient quality of the studies/audits performed rather than poor practice or failure to recognise those patients who require prophylaxis, which we could not determine from our study. Future interventions should comply with five major quality criteria; a multifaceted approach involving health providers who care for corticosteroid users, public education, increased access to DXA, is required in order to make an impact on the underprescribing of CIOP prophylaxis.

Acknowledgement

We thank Roy Stewart (statistician, MSc) for the assistance on the statistical analyses.

Conflicts of interest

Tjongerschans Hospital has received an unrestricted educational grant from Merck Sharp & Dohme.

Copyright information

© International Osteoporosis Foundation and National Osteoporosis Foundation 2008