Article Text

Download PDFPDF

Access to bone densitometry increases general practitioners’ prescribing for osteoporosis in steroid treated patients
  1. A L Dolan,
  2. E Koshy,
  3. M Waker,
  4. C M Goble
  1. Department of Rheumatology, Queen Elizabeth Hospital, Greenwich, London, SE18, UK
  1. Correspondence to:
    Dr A L Dolan
    Department of Rheumatology, Queen Elizabeth Hospital, Stadium Road, London SE18, UK;


Background: Availability of access to bone densitometry in the UK varies widely and there are concerns as to appropriate prescribing. Studies suggest inadequate use of osteoporosis prophylaxis in steroid users, despite recent guidelines.

Objective: To examine in a case-control study whether access to bone densitometry affects GPs’ osteoporosis prescribing in high risk steroid users.

Method: 10 general practices were included, five from primary care trusts (PCTs) with access to bone densitometry and five with limited access. Patients receiving prednisolone for >3 months were identified by database search. Patients receiving no prophylaxis other than calcium and vitamin D (Ca/D) were subsequently included. Appropriate patients in five practices were offered DXA scan (cases) and review. Patients in practices without access to scans (controls) were reviewed. GPs’ opinions leading to treatment were sought by structured questionnaire.

Results: 132 (0.12%) patients were receiving prednisolone for ⩾3 months, but no osteoporosis prophylaxis other than Ca/D. Pre-study prophylaxis ranged from 18 to 36%. Of 48 patients scanned, 21 (44%) were abnormal and 18 (38%) received new treatment. 13/44 (30%) controls received new treatment. 10/21 (48%) with abnormal scans started a bisphosphonate, compared with 7/44 (16%) controls (RR = 3, p = 0.004). No difference in risk factors for fracture was found in treated and untreated controls.

Conclusions: GPs were three times more likely to start potent osteoporosis treatment after abnormal scans than GPs relying on clinical information. In practice, risk factors were not adequately assessed. Database searches may identify patients needing osteoporosis prophylaxis; however, DXA enables more appropriate patient treatment.

  • osteoporosis
  • bone densitometry
  • corticosteroids
  • decision making
  • general practitioners
  • BMD, bone mineral density
  • Ca/D, calcium and vitamin D preparations
  • DXA, dual energy x ray absorptiometry
  • GP, general practitioner
  • HRT, hormone replacement therapy
  • PCT, primary care trust

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Access to bone densitometry services in the UK varies widely, with a consequent diversity of prescribing patterns. Although there is now a range of therapeutic interventions for osteoporosis, concerns have been raised as to the appropriateness of prescribing. Bone densitometry is an important way of identifying the need for treatment in patients with risk factors. It is recognised to affect women’s decisions to take hormone replacement therapy (HRT) to prevent fractures,1,2 especially if bone mineral density (BMD) is low. The extent to which availability of bone densitometry alters general practitioners’ (GPs) decision making on osteoporosis prescribing for a high risk group receiving steroids has not been studied.

Patients taking steroids are at increased risk of osteoporotic fracture, with 30% of patients taking steroids for five years or more experiencing low impact fractures3 and those with rheumatoid arthritis having a 4–5 times increased risk of fracture.4 Fracture risk increases with cumulative dose, but there is probably no safe minimum steroid dose, with those receiving as little as 2.5 mg having an increased risk of vertebral fracture.3 A number of recent guidelines on osteoporosis prophylaxis in steroid users have been published,5,6 and yet many doctors do not connect the use of steroids with the need to prevent steroid induced osteoporosis.7–9 The proliferation of guidelines for GPs means that many go unheeded.10 It is unclear whether we should be treating all steroid users with osteoporosis prophylaxis. However, it is more cost effective, as demonstrated in post-menopausal osteoporosis, to direct treatment with bone densitometry towards high risk groups rather than to treat indiscriminately.11 Similarly, a recent study of risedronate in prevention of hip fractures has shown that selection of patients according to risk factors alone does not identify patients who will benefit from treatment.12

This parallel group, interventional cross sectional study aims at assessing whether a bone density dual energy x ray absorptiometry (DXA) scan alters the decision to treat, in patients already taking steroids, in 10 similar GP practices, with and without easy access to the service. Identification of steroid users in order to perform this study might in itself alter prescribing habits. The first part of this study was, therefore, a survey of current steroid users within each practice and identification of current levels of osteoporosis prophylaxis.


Ten GP practices were identified in neighbouring primary care trusts (PCTs) in representative, similar semiurban areas of southeast London in 2001. Practices had responded to an open invitation to local practices to take part. Five practices in the PCT of Greenwich and Bexley had direct GP access to bone densitometry; whereas the five in the neighbouring Bromley PCT had limited access, reflecting differential service provision. Ethics committee approval was granted.

A parallel group, interventional case-control design was chosen. Patients, aged 35–80 inclusive, taking steroids in the previous six months at all doses, were identified from a search of practice databases. Only patients taking oral prednisolone preparations for ⩾3 months were included in the study. Patients taking only inhaled or intramuscular steroids, or other corticosteroid preparations, were excluded. Demographic data were collected on the identified patients, together with information about their current steroid dose and length of use, current osteoporosis prophylaxis, underlying diagnosis and previous fracture history, and other current drugs. All patients completed a questionnaire which examined their other risk factors for osteoporosis and falls, and this information was available to the GPs at their follow up appointments.

Intervention group

All patients in Greenwich and Bexley PCT study practices who were not receiving any current osteoporosis prophylaxis (antiresorptive agent) other than calcium and vitamin D preparations (Ca/D) received a letter from their GP offering a DXA scan. These formed the intervention group. Those patients who had not had densitometry in the past year had a DXA scan (Hologic QDR 4500c at spine and hip). After densitometry, they were asked to see their GP for a review of their treatment. The local bone densitometry service gave a standardised clinical report where a T score of ⩽−2.5 or z ⩽−1 was taken as abnormal, based on the WHO definitions. This may be an underestimate of fracture risk for steroid users, where a criterion of −1.5 may be used.13 However, GPs had attended a workshop on steroid osteoporosis and the protocol. They received copies of steroid guidelines and were aware that the intervention threshold in the guideline was higher. This is a pragmatic study looking at their response to reports.


Those patients in Bromley PCT, who were similarly identified, formed the control group. These patients did not have DXA scans, but completed the questionnaire and were invited for a review of treatment by their GP. For ethical reasons, after the project was complete, GPs were given the opportunity to refer these patients for DXA scans.

GPs completed a structured questionnaire on factors leading to their prescribing decisions and recorded osteoporosis treatment given (table 3). Before the study all GPs had attended a three hour workshop on osteoporosis and were provided with the “2000 Royal College of Physicians guidelines on osteoporosis management”. The GPs were also guided through the inclusion and exclusion criteria for the study, so that all of the practices were following the same methods of identification over the same time frame.

Statistical analysis was performed using SPSS for Windows (SPSS Chicago, IL). Descriptive statistics were obtained and data tested for normality. A χ2 test was applied to questionnaire data for differences between cases and controls.


Two hundred and one patients were identified as taking prednisolone for ⩾3 months. This constituted between 0.1% and 0.35% of the total practice populations (mean 0.12%). One hundred and thirty two patients were identified as taking prednisolone for ⩾3 months but no osteoporosis prophylaxis other than Ca/D (1000 mg calcium and 800 IU vitamin D). Use of antiresorptive agents (bisphosphonates, HRT, or selective oestrogen receptor modulators) in appropriate patients before the study ranged between 18% and 46% (mean 34%) according to practice.

Ninety two of 132 patients (70%) eligible for further inclusion in the study completed the subsequent assessment. Of the 92 patients (44 (48%) female, 48 (52%) male) assessed, 30 (33%) had received Ca/D previously. Mean age was 66.7 (range 38–81). The median time for steroid use was 2 to ⩽5 years (range 3 months to more than 20 years) and median dose 2.5–5 mg. Eighteen (20%) had previously had a low impact fracture. Failure to attend was 32% (23/71) in the intervention group and 39% (17/44) in controls. Offer of a DXA scan rather than just “doctor assessment” did not seem to affect participation. Three in the intervention group were considered too unwell to attend for a DXA scan by their doctors. Forty eight were fully assessed in the intervention group (68% of those eligible) and 44 assessed in controls (72% of those eligible). Table 1 shows the demographics of the patients from the two areas. Control patients were older and there was more pre-assessment use of Ca/D.

Table 1

Demographic details of intervention and controls groups

In the intervention group 21 (44%) had abnormal DXA scans, defined as T<−2.5 or z<−1. Thirteen of 21 (62%) with abnormal scans received any new treatment, compared with 5/27 (19%) of those with normal scans, p<0.004. Bisphosphonates were chosen for 10/21 (48%) of those with abnormal scans compared with just 1/27 (4%) if the scan was normal (RR = 12.8, p<0.0001) (table 2). Of those with abnormal scans who did not receive treatment, three were considered too ill to do so by their GP and two did not have treatment initiated by their GP “because they were under the care of a specialist clinic”. Those with abnormal DXA scans were more likely to be receiving a prednisolone dose of 7.5 mg or more (RR = 2.14). There was no difference in duration of steroid use or prevalence of previous fracture in those with normal or abnormal scans.

Table 2

Treatment and scan outcome

In controls, where GPs assessed patients without a DXA scan, 13/44 (30%) got new treatments. This was not significantly different from the intervention group considered as a whole (38%). However, only 7/44 (16%) controls received bisphosphonates (table 2). There was no difference in the presence of risk factors in the treated and non-treated controls (sex, steroids >1 year, prednisolone >7.5 mg, age, or previous fracture). Steroid dose ⩾7.5 mg is an indication for bone protection in the majority of guidelines yet only 23% of those taking high dose steroid received treatment. The result of the database search and interview is that 29/44 (66%) of the non-scanned DXA group are now receiving some treatment (previous Ca/D and new treatment), compared with 20/44 (45%) before; an increase of 20%.

The likelihood of receiving potent treatment, such as a bisphosphonate, was three times greater in the DXA scanned intervention group, where treatment with a bisphosphonate was started in 48% of patients with abnormal scans compared with only 16% of controls (p = 0.004).

Treatment chosen by GPs consisted of Ca/D (1 g calcium+800 IU vitamin D) and bisphosphonates, including daily or weekly alendronate, risedronate, or cyclic etidronate. None chose to start either HRT or raloxifene.

GPs’ decision making processes were assessed by a structured questionnaire (table 3). In 22/48 (46%) cases GPs felt the DXA scan dictated treatment, in the intervention group, despite 21 having abnormal scan reports. There may be some error in questionnaire response because two control group GPs responded positively to this, yet did not have DXA scans available. This is often a weakness in opinion based questionnaires. Risk factors were felt to influence decision making for 26/48 (54%) in the intervention group and 35/44 (80%) controls, with a trend towards their greater importance for those not relying on DXA scans (p<0.06).

Table 3

Outcome of structured questionnaire. GPs were asked: “To what extent did these factors influence your decision on appropriate bone protective treatment?” Results show number who “agreed” or “strongly agreed” with the question posed. All questions asked are shown

GPs felt current treatment was enough in only 2/27 where the DXA scan was abnormal, whereas all felt treatment was adequate in the presence of a normal scan. When no DXA scan was available 17/44 (38%) felt current treatment was adequate. Eleven per cent of all patients declined offered treatment, independent of DXA scan. This may reflect fear of side effects in this elderly group who are taking many drugs or a lack of understanding of risk-benefit.

There was a tendency for patients to express a preference for treatment in the DXA scan group, although numbers in each group were small. Not surprisingly few GPs treating patients in the scanned group felt the need for more tests. However, GPs expressed a need for more tests in only 36% of the control group, suggesting they are content to base decision making on risk factors; this response did not correlate with steroid dose. GPs expressed a desire to treat to prevent further problems in 72% of all cases, but their action did not seem influenced by medicolegal concerns.


This study demonstrates the value of bone densitometry in leading to appropriate treatment. Patients with an abnormal DXA scan were three times more likely to be given potent treatment than those from the area where GPs were making a clinical decision without the benefit of a DXA scan.

The incidence of steroid use in this population, 0.12%, is less than the 0.5% reported by Walsh et al in a UK community population9 and 0.72% in an Icelandic population.7 This may in part reflect the age restrictions of the group assessed. Most patients in this study had been receiving steroids for two years or more. Length of steroid use is often prolonged, with 20% of a UK population who received a steroid prescription still taking them six months later.14 Before the intervention, adequate bone protective treatment was underprescribed (18–35%) despite local and national guidelines. A similar deficit was identified in previous studies.7,8 Database search and GP education have some value, increasing treatment levels in the control group by 20%. Bisphosphonate use increased by 16%, on a background practice use of 34% before selection of the study group. Yet selection of patients by BMD leads to a much higher and more appropriate use of potent treatments.

A number of guidelines on the management of steroid induced osteoporosis have been produced in recent years in the UK5 and USA,6 and in this study GPs were issued with locally adapted guidelines on “Management of steroid induced osteoporosis”. Risk factors such as previous fracture are independent indicators of future fracture risk in steroid induced or postmenopausal osteoporosis. Previous fracture is a key indicator of the need to treat because a distal radial fracture may predate more significant low trauma fractures, and 20% of patients with vertebral fractures have a fracture again within a year.15 The extent of the steroid induced fracture risk is dependent on dose and length of use.3 Consistent with this, we found the DXA scan was more likely to be abnormal in those with higher prednisolone dose, with a 2.14 times likelihood of prednisolone being >7.5 mg in those with abnormal DXA scans. Age is also an independent risk factor for fracture that cannot be accounted for purely by age related change in BMD, reflected in the T score.16 This may indicate an increased tendency to fall. Despite guideline recommendations, when GPs relied on clinical factors alone to decide on treatment there was no apparent difference in the incidence of risk factors, (such as duration of treatment, prednisolone dose, previous fracture, age, or previous calcium use) in those who started new treatment compared with those who did not. This suggests that, in practice, these factors were not adequately taken into account. GPs’ own perception was that risk factors did dictate treatment in those for whom scans were not available compared with the scanned group (p<0.06) (table 3). Since this study was undertaken, new CIOP guidelines have been published by the Royal College of Physicians, UK (Dec 2002). These are less reliant on DXA, as patients are recommended for treatment on clinical grounds alone—that is, age over 65 and previous fracture. Our control group demonstrates that prescribing on the basis of risk factors alone does not result in the use of such potent treatments as when the GP is led by additional data from a DXA scan. This may be a shortcoming in achieving the aims of the new guidelines. Increased DXA scanner provision might encourage prescription tuned to individual patients.

In the DXA scanned group those with an abnormal scan were over three times more likely to receive new treatment than if the DXA scan result was normal (62% v 19%); GPs are acting on the current DXA scan findings. This is understandable but may not be wholly appropriate, because bone density will fall with steroid use, particularly in the first six months3 and preventive studies on patients starting bisphosphonates show the protective advantage of prophylactic.17,18 Guidelines at the time recommended that all patients receiving high dose steroids need treatment. Yet an intervention was started in only 20% of patients with a normal DXA scan receiving prednisolone of >7.5 mg, despite this in itself being an indication for osteoporosis prophylaxis in most guidelines.

The results of this study emphasise the difficulty in conducting preventive strategies in a busy GP surgery and that uptake of guideline information is inconsistent, as reported elsewhere.10 We have shown that case finding by database search is an effective strategy for identifying patients, leading to more receiving bone protective treatment. However, bone densitometry to identify abnormal results leads to more patients receiving relevant treatment. In the light of a good evidence base that prevention of steroid induced osteoporosis is beneficial, it is imperative that practices adopt strategies that lead to appropriate patient care.


Thanks are due to the GPs and their teams who contacted patients and responded to our questionnaire on management, including Dr Caroline Fraser, Dr Shanti Mendonca, Dr Tina Challacombe, Dr Ellen Wright, Dr James Muir, Dr Matthews, Dr Nicola Payne, Dr N Ratnarajan, Dr Grace Whitfield, Dr Virginia Todd.