We analyzed the data from 853 women with a mean age of 62 (9) years and a mean body mass index of 27.0 (4.2)kg/m2. In Table 1, we present the demographic and clinical parameters included in FRAX as fracture risk factors of these patients.

Twenty percent of the patients had a normal bone mineral density, 55% had osteopenia and 25%, osteoporosis.

Yet excluding age and body mass index, the number of fracture risk factors seems low among the patients included in the study. Twenty-six percent of the patients had no risk of fracture; 35%, one; 30%, two; 6%, three; and 3% had four or five risks of fracture. The high prevalence of secondary osteoporosis (36%) is explained almost entirely by the item “menopause before 45 years”, probably the reason for BMD scanning.

Mean MFR+ was 5.4 (4.8)%. The histogram of the distribution of MFR+ presented in Fig. 1 shows an accumulation of the patients in very low levels of risk of fracture. Mean MFR− was 6.3 (5.5)%; mean HFR+, 1.5 (2.9)%; and HFR−, 2.1 (3.3)%.

Fig. 1.

Major risk of fracture distribution calculated by FRAX including BMD in the algorithm.

(0.08MB).

When BMD was included in the algorithm for the calculation of the risk of fracture, the risk was statistically lower (p<0.001) as shown in Fig. 2, especially in patients with better BMD. As a rule, the inclusion of BMD in the algorithm worsens risk fracture in patients with T-score below −2.

Fig. 2.

Distribution of the absolute risk of fracture by intervals of BMD (T-score) with or without inclusion of BMD in the algorithm. (a) Major risk of fracture. (b) Hip fracture.

(0.2MB).

If we exclude from the analysis the patients who are already in treatment for osteoporosis (210 patients treated, 641 patients untreated), MFR+ becomes even lower: 5.0 (4.6)%. Patients without treatment are younger (61 (9) years versus 64 (8) years) and their frequency of osteoporosis, lower (21% versus 37%), in a statistically significant way.

The generalised use of DXA is limited because it is expensive and time consuming; it is not portable, and it is available only in specialised clinics. The best strategy to avoid waiting lists in Bone Densitometry Units has been the selection of patients at high risk of osteoporosis.

Now, the concept has changed. The target is not the osteoporotic patient but the patient at high risk of fracture. Formerly, for instance, a prior fragility fracture was a classic indication for BMD testing from the point of view of screening for osteoporosis. However, it is not a novelty that, for most clinicians, the diagnosis of a low trauma fracture is an indisputable reason to initiate specific treatment for osteoporosis without a measurement of BMD.

Since FRAX has become available, some uncertainty raises about the necessity of determining BMD in many patients: not only an opportunity but also a challenge has emerged. The real change lies in the acceptation that, not only hard clinical risk factors such as fragility fractures or glucocorticoid intake, but also the combination of several minor risk factors may achieve a level of risk of fracture enough to initiate a fracture prevention treatment without knowing BMD. At the other end of things, many physicians accept that fracture prevention treatments may be not indicated in patients with osteoporosis but low risk of fracture and, then, it would be unnecessary to know their BMD.

In a previous study, we have estimated the fracture risk by FRAX in patients in which the decision to treat or not had been taken along with previous recommendations in real clinical practice and knowing BMD: such as expected or desired, treated patients had more risk factors and their fracture risk was higher. Nevertheless, there was a significant, and probably excessive, overlap between the fracture risks of the two groups.17

Therefore, it seems relevant to define the strategy for the selection of the target population who should undergo BMD testing, avoiding high risk patients, who will be treated without knowing BMD, and low risk patients, who will not be treated independently of their BMD. In fact, we are looking for the patients with a moderate risk of fracture in which BMD may help to make a therapeutic decision. The problem is that, in most countries, it has not been defined what is low, moderate or high risk.

The first step, the one we have intended to address in this study, is to analyze the activity we make in our unit in real clinical practice in relation to the risk of fracture calculated by FRAX. Two main conclusions can be drawn from our results: one, the risk of fracture of the patients sent to our densitometry unit for bone BMD testing is low (71% of the patient had none or one risks or fracture); and two, the introduction of BMD in the algorithm for the calculation of the risk results in a decrease of the risk, except in the patients with BMD below −2 standard deviations of the T-score. These patients with moderate risk and lower BMD are the ones who most benefit from BMD testing but, in our series, there are few of them. It is clear that we need to search a strategy to improve the quality of the referrals to our unit or, in other words, to increase the risk fracture of the patients referred.

The FRIDEX study group, not far geographically from us, has calculated that the use of NOGG thresholds10 applied to FRAX would reduce about to 50% the current number of referrals for BMD testing, thus allowing to focus the activity of bone densitometry units in the patients with moderate fracture risk.18 However, the application of these thresholds, calculated for the United Kingdom, to the Spanish population is not advisable as our risk of fracture is somewhat lower: the United States has been classified as a very high risk country and Spain, a medium risk country. As an example, women aged 70 with a previous fragility fracture have a MFR of 21%, close to the 20% NOGG threshold, if calculated by FRAX for US Caucasian population. The same women have a MFR of 10% when the risk is calculated by FRAX for the Spanish population. The threshold of high risk for us might be perfectly located, then, at 10%; accordingly, an estimate of the threshold for moderate risk could be 5%.

Agreement between risk of fracture with and without the inclusion of BMD in the FRAX algorithm has been analyzed in 180 white women from Tennessee, USA19; it achieved 89.4%. Disagreement occurred in 2 distinct subgroups of patients (10.6% of cases): one comprised older patients with normal T-scores for whom FRAX-scores exceeded the NOF treatment threshold; the other, younger patients with high BMI and low T-scores for whom FRAX-scores did not exceed the threshold. In accordance with the above approach, the latter should not be treated independently of their BMD and, so, BMD testing was not necessary. The first subgroup gives more to think as treatment could be initiated directly without knowing BMD but, in fact, when BMD testing is made, the treatment can be avoided. In our series, 16 out of 29 patients with RFM above 20% (2% of the whole series) and 72 out of 153 above 10% (8%), fall below every threshold when BMD is known and RFM recalculated.

As a limitation, a selection bias exists because the patients were recruited from a Bone Densitometry Unit and not from the general population. Although this is a low risk fracture population, we estimate that the frequency of risk factors of fracture and the prevalence of osteoporosis are higher in the general population of the same age.

Two strategies would be good to clarify the raised issues in future studies: the selection of patients with moderate risk of fracture assessed by FRAX or the inclusion of a large number of subjects from general population. The former seems more operational to analyze the actual value of knowing BMD in these patients since the practice of a bone densitometry is probably not necessary in low and high risk fracture patients. We need more data that allow us to select properly the patients who benefit most from the practice of a BMD testing.

Protección de personas y animales. Los autores declaran que para esta investigación no se han realizado experimentos en seres humanos ni en animales.

Confidencialidad de los datos. Los autores declaran que han seguido los protocolos de su centro de trabajo sobre la publicación de datos de pacientes y que todos los pacientes incluidos en el estudio han recibido información suficiente y han dado su consentimiento informado por escrito para participar en dicho estudio.

Derecho a la privacidad y consentimiento informado. Los autores han obtenido el consentimiento informado de los pacientes y/o sujetos referidos en el artículo. Este documento obra en poder del autor de correspondencia.

The authors declare no conflict of interest.